{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 191.20458891013385, "eval_steps": 500, "global_step": 300000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.006373486297004461, "grad_norm": 9.50260066986084, "learning_rate": 6e-08, "loss": 0.8141, "step": 10 }, { "epoch": 0.012746972594008922, "grad_norm": 11.740574836730957, "learning_rate": 1.2666666666666666e-07, "loss": 0.8284, "step": 20 }, { "epoch": 0.019120458891013385, "grad_norm": 16.99808692932129, "learning_rate": 1.9333333333333334e-07, "loss": 0.8016, "step": 30 }, { "epoch": 0.025493945188017845, "grad_norm": 11.063021659851074, "learning_rate": 2.6e-07, "loss": 0.8329, "step": 40 }, { "epoch": 0.03186743148502231, "grad_norm": 18.429140090942383, "learning_rate": 3.266666666666667e-07, "loss": 0.9546, "step": 50 }, { "epoch": 0.03824091778202677, "grad_norm": 10.585908889770508, "learning_rate": 3.933333333333333e-07, "loss": 0.8555, "step": 60 }, { "epoch": 0.04461440407903123, "grad_norm": 13.518843650817871, "learning_rate": 4.6e-07, "loss": 0.8202, "step": 70 }, { "epoch": 0.05098789037603569, "grad_norm": 8.653311729431152, "learning_rate": 5.266666666666667e-07, "loss": 0.6907, "step": 80 }, { "epoch": 0.05736137667304015, "grad_norm": 9.918243408203125, "learning_rate": 5.933333333333333e-07, "loss": 0.5854, "step": 90 }, { "epoch": 0.06373486297004462, "grad_norm": 6.905433654785156, "learning_rate": 6.6e-07, "loss": 0.5492, "step": 100 }, { "epoch": 0.07010834926704908, "grad_norm": 3.717879056930542, "learning_rate": 7.266666666666668e-07, "loss": 0.4975, "step": 110 }, { "epoch": 0.07648183556405354, "grad_norm": 4.2783684730529785, "learning_rate": 7.933333333333334e-07, "loss": 0.3872, "step": 120 }, { "epoch": 0.082855321861058, "grad_norm": 2.390763998031616, "learning_rate": 8.6e-07, "loss": 0.2653, "step": 130 }, { "epoch": 0.08922880815806246, "grad_norm": 2.8818657398223877, "learning_rate": 9.266666666666667e-07, "loss": 0.2991, "step": 140 }, { "epoch": 0.09560229445506692, "grad_norm": 2.502659797668457, "learning_rate": 9.933333333333335e-07, "loss": 0.2696, "step": 150 }, { "epoch": 0.10197578075207138, "grad_norm": 1.6304689645767212, "learning_rate": 1.06e-06, "loss": 0.2705, "step": 160 }, { "epoch": 0.10834926704907584, "grad_norm": 1.7764955759048462, "learning_rate": 1.1266666666666667e-06, "loss": 0.1726, "step": 170 }, { "epoch": 0.1147227533460803, "grad_norm": 2.821688652038574, "learning_rate": 1.1933333333333335e-06, "loss": 0.2019, "step": 180 }, { "epoch": 0.12109623964308477, "grad_norm": 2.787238836288452, "learning_rate": 1.26e-06, "loss": 0.2024, "step": 190 }, { "epoch": 0.12746972594008923, "grad_norm": 2.0788207054138184, "learning_rate": 1.3266666666666667e-06, "loss": 0.2077, "step": 200 }, { "epoch": 0.1338432122370937, "grad_norm": 2.0954220294952393, "learning_rate": 1.3933333333333335e-06, "loss": 0.1764, "step": 210 }, { "epoch": 0.14021669853409816, "grad_norm": 1.7847673892974854, "learning_rate": 1.46e-06, "loss": 0.1948, "step": 220 }, { "epoch": 0.14659018483110262, "grad_norm": 1.7281008958816528, "learning_rate": 1.5266666666666667e-06, "loss": 0.1823, "step": 230 }, { "epoch": 0.15296367112810708, "grad_norm": 2.379450798034668, "learning_rate": 1.5933333333333335e-06, "loss": 0.1772, "step": 240 }, { "epoch": 0.15933715742511154, "grad_norm": 2.32367205619812, "learning_rate": 1.6600000000000002e-06, "loss": 0.1728, "step": 250 }, { "epoch": 0.165710643722116, "grad_norm": 2.5996227264404297, "learning_rate": 1.7266666666666667e-06, "loss": 0.172, "step": 260 }, { "epoch": 0.17208413001912046, "grad_norm": 2.1081321239471436, "learning_rate": 1.7933333333333332e-06, "loss": 0.1574, "step": 270 }, { "epoch": 0.17845761631612492, "grad_norm": 3.374861478805542, "learning_rate": 1.86e-06, "loss": 0.1547, "step": 280 }, { "epoch": 0.18483110261312938, "grad_norm": 2.300109624862671, "learning_rate": 1.9266666666666667e-06, "loss": 0.1748, "step": 290 }, { "epoch": 0.19120458891013384, "grad_norm": 2.2415449619293213, "learning_rate": 1.9933333333333334e-06, "loss": 0.1773, "step": 300 }, { "epoch": 0.1975780752071383, "grad_norm": 2.5365710258483887, "learning_rate": 2.06e-06, "loss": 0.1671, "step": 310 }, { "epoch": 0.20395156150414276, "grad_norm": 2.430039167404175, "learning_rate": 2.126666666666667e-06, "loss": 0.1341, "step": 320 }, { "epoch": 0.21032504780114722, "grad_norm": 2.3952574729919434, "learning_rate": 2.1933333333333332e-06, "loss": 0.1666, "step": 330 }, { "epoch": 0.21669853409815168, "grad_norm": 2.762911796569824, "learning_rate": 2.26e-06, "loss": 0.1726, "step": 340 }, { "epoch": 0.22307202039515614, "grad_norm": 2.1954283714294434, "learning_rate": 2.326666666666667e-06, "loss": 0.1511, "step": 350 }, { "epoch": 0.2294455066921606, "grad_norm": 2.528956890106201, "learning_rate": 2.3933333333333334e-06, "loss": 0.151, "step": 360 }, { "epoch": 0.23581899298916506, "grad_norm": 2.1332144737243652, "learning_rate": 2.46e-06, "loss": 0.158, "step": 370 }, { "epoch": 0.24219247928616955, "grad_norm": 1.6394823789596558, "learning_rate": 2.526666666666667e-06, "loss": 0.1562, "step": 380 }, { "epoch": 0.248565965583174, "grad_norm": 1.5594468116760254, "learning_rate": 2.593333333333333e-06, "loss": 0.1625, "step": 390 }, { "epoch": 0.25493945188017847, "grad_norm": 2.135986089706421, "learning_rate": 2.66e-06, "loss": 0.1656, "step": 400 }, { "epoch": 0.26131293817718293, "grad_norm": 1.1881332397460938, "learning_rate": 2.726666666666667e-06, "loss": 0.1632, "step": 410 }, { "epoch": 0.2676864244741874, "grad_norm": 2.092585802078247, "learning_rate": 2.7933333333333334e-06, "loss": 0.1415, "step": 420 }, { "epoch": 0.27405991077119185, "grad_norm": 1.8596680164337158, "learning_rate": 2.86e-06, "loss": 0.1776, "step": 430 }, { "epoch": 0.2804333970681963, "grad_norm": 2.1945323944091797, "learning_rate": 2.926666666666667e-06, "loss": 0.1672, "step": 440 }, { "epoch": 0.28680688336520077, "grad_norm": 0.947393536567688, "learning_rate": 2.9933333333333336e-06, "loss": 0.1229, "step": 450 }, { "epoch": 0.29318036966220523, "grad_norm": 3.158277750015259, "learning_rate": 3.06e-06, "loss": 0.138, "step": 460 }, { "epoch": 0.2995538559592097, "grad_norm": 1.4556961059570312, "learning_rate": 3.1266666666666667e-06, "loss": 0.1444, "step": 470 }, { "epoch": 0.30592734225621415, "grad_norm": 1.6013380289077759, "learning_rate": 3.1933333333333334e-06, "loss": 0.1618, "step": 480 }, { "epoch": 0.3123008285532186, "grad_norm": 1.6442179679870605, "learning_rate": 3.2599999999999997e-06, "loss": 0.1677, "step": 490 }, { "epoch": 0.3186743148502231, "grad_norm": 3.0498273372650146, "learning_rate": 3.326666666666667e-06, "loss": 0.1414, "step": 500 }, { "epoch": 0.32504780114722753, "grad_norm": 1.5951189994812012, "learning_rate": 3.3933333333333336e-06, "loss": 0.1558, "step": 510 }, { "epoch": 0.331421287444232, "grad_norm": 2.0183534622192383, "learning_rate": 3.46e-06, "loss": 0.1464, "step": 520 }, { "epoch": 0.33779477374123645, "grad_norm": 2.1769559383392334, "learning_rate": 3.526666666666667e-06, "loss": 0.1649, "step": 530 }, { "epoch": 0.3441682600382409, "grad_norm": 2.2509684562683105, "learning_rate": 3.5933333333333334e-06, "loss": 0.1371, "step": 540 }, { "epoch": 0.3505417463352454, "grad_norm": 1.3416584730148315, "learning_rate": 3.66e-06, "loss": 0.1477, "step": 550 }, { "epoch": 0.35691523263224983, "grad_norm": 1.9828554391860962, "learning_rate": 3.7266666666666664e-06, "loss": 0.1613, "step": 560 }, { "epoch": 0.3632887189292543, "grad_norm": 1.1496295928955078, "learning_rate": 3.7933333333333336e-06, "loss": 0.1274, "step": 570 }, { "epoch": 0.36966220522625876, "grad_norm": 1.9544023275375366, "learning_rate": 3.86e-06, "loss": 0.1302, "step": 580 }, { "epoch": 0.3760356915232632, "grad_norm": 1.5555181503295898, "learning_rate": 3.926666666666667e-06, "loss": 0.1251, "step": 590 }, { "epoch": 0.3824091778202677, "grad_norm": 1.9086092710494995, "learning_rate": 3.993333333333334e-06, "loss": 0.1565, "step": 600 }, { "epoch": 0.38878266411727214, "grad_norm": 2.619568109512329, "learning_rate": 4.06e-06, "loss": 0.1337, "step": 610 }, { "epoch": 0.3951561504142766, "grad_norm": 1.56656813621521, "learning_rate": 4.126666666666667e-06, "loss": 0.1366, "step": 620 }, { "epoch": 0.40152963671128106, "grad_norm": 2.150005340576172, "learning_rate": 4.1933333333333336e-06, "loss": 0.1543, "step": 630 }, { "epoch": 0.4079031230082855, "grad_norm": 1.9915505647659302, "learning_rate": 4.26e-06, "loss": 0.1369, "step": 640 }, { "epoch": 0.41427660930529, "grad_norm": 1.766042709350586, "learning_rate": 4.326666666666667e-06, "loss": 0.1255, "step": 650 }, { "epoch": 0.42065009560229444, "grad_norm": 2.2898106575012207, "learning_rate": 4.393333333333333e-06, "loss": 0.1621, "step": 660 }, { "epoch": 0.4270235818992989, "grad_norm": 2.1378886699676514, "learning_rate": 4.4600000000000005e-06, "loss": 0.1411, "step": 670 }, { "epoch": 0.43339706819630336, "grad_norm": 3.2848191261291504, "learning_rate": 4.526666666666667e-06, "loss": 0.1481, "step": 680 }, { "epoch": 0.4397705544933078, "grad_norm": 1.6556960344314575, "learning_rate": 4.593333333333333e-06, "loss": 0.1452, "step": 690 }, { "epoch": 0.4461440407903123, "grad_norm": 2.0749261379241943, "learning_rate": 4.66e-06, "loss": 0.1507, "step": 700 }, { "epoch": 0.45251752708731674, "grad_norm": 2.629366397857666, "learning_rate": 4.726666666666667e-06, "loss": 0.1446, "step": 710 }, { "epoch": 0.4588910133843212, "grad_norm": 1.869321584701538, "learning_rate": 4.793333333333334e-06, "loss": 0.1481, "step": 720 }, { "epoch": 0.46526449968132566, "grad_norm": 1.557539939880371, "learning_rate": 4.86e-06, "loss": 0.1342, "step": 730 }, { "epoch": 0.4716379859783301, "grad_norm": 1.2169694900512695, "learning_rate": 4.926666666666667e-06, "loss": 0.113, "step": 740 }, { "epoch": 0.4780114722753346, "grad_norm": 1.5026923418045044, "learning_rate": 4.9933333333333335e-06, "loss": 0.1396, "step": 750 }, { "epoch": 0.4843849585723391, "grad_norm": 1.6789538860321045, "learning_rate": 5.06e-06, "loss": 0.1273, "step": 760 }, { "epoch": 0.49075844486934356, "grad_norm": 1.173251986503601, "learning_rate": 5.126666666666667e-06, "loss": 0.1433, "step": 770 }, { "epoch": 0.497131931166348, "grad_norm": 2.4430761337280273, "learning_rate": 5.193333333333333e-06, "loss": 0.1631, "step": 780 }, { "epoch": 0.5035054174633524, "grad_norm": 1.9218428134918213, "learning_rate": 5.2600000000000005e-06, "loss": 0.1375, "step": 790 }, { "epoch": 0.5098789037603569, "grad_norm": 1.734273910522461, "learning_rate": 5.326666666666667e-06, "loss": 0.1482, "step": 800 }, { "epoch": 0.5162523900573613, "grad_norm": 1.686924695968628, "learning_rate": 5.393333333333334e-06, "loss": 0.1378, "step": 810 }, { "epoch": 0.5226258763543659, "grad_norm": 1.936724066734314, "learning_rate": 5.46e-06, "loss": 0.1444, "step": 820 }, { "epoch": 0.5289993626513703, "grad_norm": 1.8305740356445312, "learning_rate": 5.5266666666666666e-06, "loss": 0.156, "step": 830 }, { "epoch": 0.5353728489483748, "grad_norm": 1.2125890254974365, "learning_rate": 5.593333333333334e-06, "loss": 0.1254, "step": 840 }, { "epoch": 0.5417463352453792, "grad_norm": 1.9790780544281006, "learning_rate": 5.66e-06, "loss": 0.1397, "step": 850 }, { "epoch": 0.5481198215423837, "grad_norm": 1.9570019245147705, "learning_rate": 5.726666666666667e-06, "loss": 0.1421, "step": 860 }, { "epoch": 0.5544933078393881, "grad_norm": 1.190658450126648, "learning_rate": 5.793333333333334e-06, "loss": 0.1328, "step": 870 }, { "epoch": 0.5608667941363926, "grad_norm": 2.010284423828125, "learning_rate": 5.86e-06, "loss": 0.1241, "step": 880 }, { "epoch": 0.567240280433397, "grad_norm": 1.7606861591339111, "learning_rate": 5.926666666666667e-06, "loss": 0.132, "step": 890 }, { "epoch": 0.5736137667304015, "grad_norm": 0.8547699451446533, "learning_rate": 5.993333333333333e-06, "loss": 0.1231, "step": 900 }, { "epoch": 0.579987253027406, "grad_norm": 1.2310320138931274, "learning_rate": 6.0600000000000004e-06, "loss": 0.1247, "step": 910 }, { "epoch": 0.5863607393244105, "grad_norm": 1.368887186050415, "learning_rate": 6.126666666666667e-06, "loss": 0.1196, "step": 920 }, { "epoch": 0.5927342256214149, "grad_norm": 1.5329285860061646, "learning_rate": 6.193333333333334e-06, "loss": 0.1284, "step": 930 }, { "epoch": 0.5991077119184194, "grad_norm": 1.4317708015441895, "learning_rate": 6.26e-06, "loss": 0.1322, "step": 940 }, { "epoch": 0.6054811982154238, "grad_norm": 1.296454668045044, "learning_rate": 6.3266666666666665e-06, "loss": 0.1133, "step": 950 }, { "epoch": 0.6118546845124283, "grad_norm": 1.8704478740692139, "learning_rate": 6.393333333333333e-06, "loss": 0.131, "step": 960 }, { "epoch": 0.6182281708094327, "grad_norm": 2.585780382156372, "learning_rate": 6.460000000000001e-06, "loss": 0.1445, "step": 970 }, { "epoch": 0.6246016571064372, "grad_norm": 1.1671991348266602, "learning_rate": 6.526666666666667e-06, "loss": 0.1135, "step": 980 }, { "epoch": 0.6309751434034416, "grad_norm": 1.8688873052597046, "learning_rate": 6.5933333333333335e-06, "loss": 0.1204, "step": 990 }, { "epoch": 0.6373486297004461, "grad_norm": 2.5886378288269043, "learning_rate": 6.660000000000001e-06, "loss": 0.135, "step": 1000 }, { "epoch": 0.6437221159974506, "grad_norm": 2.5049116611480713, "learning_rate": 6.726666666666667e-06, "loss": 0.1513, "step": 1010 }, { "epoch": 0.6500956022944551, "grad_norm": 1.8935885429382324, "learning_rate": 6.793333333333333e-06, "loss": 0.1421, "step": 1020 }, { "epoch": 0.6564690885914596, "grad_norm": 1.5030502080917358, "learning_rate": 6.8599999999999995e-06, "loss": 0.11, "step": 1030 }, { "epoch": 0.662842574888464, "grad_norm": 1.3139643669128418, "learning_rate": 6.9266666666666675e-06, "loss": 0.1206, "step": 1040 }, { "epoch": 0.6692160611854685, "grad_norm": 1.3345052003860474, "learning_rate": 6.993333333333334e-06, "loss": 0.1156, "step": 1050 }, { "epoch": 0.6755895474824729, "grad_norm": 1.3748908042907715, "learning_rate": 7.06e-06, "loss": 0.1339, "step": 1060 }, { "epoch": 0.6819630337794774, "grad_norm": 1.4622594118118286, "learning_rate": 7.126666666666667e-06, "loss": 0.1259, "step": 1070 }, { "epoch": 0.6883365200764818, "grad_norm": 1.5618497133255005, "learning_rate": 7.193333333333334e-06, "loss": 0.1482, "step": 1080 }, { "epoch": 0.6947100063734863, "grad_norm": 1.174689769744873, "learning_rate": 7.26e-06, "loss": 0.1198, "step": 1090 }, { "epoch": 0.7010834926704907, "grad_norm": 1.6729302406311035, "learning_rate": 7.326666666666666e-06, "loss": 0.1109, "step": 1100 }, { "epoch": 0.7074569789674953, "grad_norm": 1.5087976455688477, "learning_rate": 7.393333333333334e-06, "loss": 0.1275, "step": 1110 }, { "epoch": 0.7138304652644997, "grad_norm": 1.3268859386444092, "learning_rate": 7.4600000000000006e-06, "loss": 0.1097, "step": 1120 }, { "epoch": 0.7202039515615042, "grad_norm": 1.0452150106430054, "learning_rate": 7.526666666666667e-06, "loss": 0.1268, "step": 1130 }, { "epoch": 0.7265774378585086, "grad_norm": 1.128790020942688, "learning_rate": 7.593333333333334e-06, "loss": 0.148, "step": 1140 }, { "epoch": 0.7329509241555131, "grad_norm": 2.143430233001709, "learning_rate": 7.660000000000001e-06, "loss": 0.1335, "step": 1150 }, { "epoch": 0.7393244104525175, "grad_norm": 2.210578203201294, "learning_rate": 7.726666666666667e-06, "loss": 0.1194, "step": 1160 }, { "epoch": 0.745697896749522, "grad_norm": 1.5607821941375732, "learning_rate": 7.793333333333334e-06, "loss": 0.1258, "step": 1170 }, { "epoch": 0.7520713830465264, "grad_norm": 1.99208402633667, "learning_rate": 7.860000000000001e-06, "loss": 0.1021, "step": 1180 }, { "epoch": 0.758444869343531, "grad_norm": 1.4818758964538574, "learning_rate": 7.926666666666666e-06, "loss": 0.1228, "step": 1190 }, { "epoch": 0.7648183556405354, "grad_norm": 2.065819501876831, "learning_rate": 7.993333333333334e-06, "loss": 0.1289, "step": 1200 }, { "epoch": 0.7711918419375399, "grad_norm": 2.015576124191284, "learning_rate": 8.06e-06, "loss": 0.137, "step": 1210 }, { "epoch": 0.7775653282345443, "grad_norm": 1.9455817937850952, "learning_rate": 8.126666666666668e-06, "loss": 0.1242, "step": 1220 }, { "epoch": 0.7839388145315488, "grad_norm": 1.1709002256393433, "learning_rate": 8.193333333333333e-06, "loss": 0.1233, "step": 1230 }, { "epoch": 0.7903123008285532, "grad_norm": 1.6992369890213013, "learning_rate": 8.26e-06, "loss": 0.1131, "step": 1240 }, { "epoch": 0.7966857871255577, "grad_norm": 1.7543067932128906, "learning_rate": 8.326666666666668e-06, "loss": 0.1021, "step": 1250 }, { "epoch": 0.8030592734225621, "grad_norm": 1.9983257055282593, "learning_rate": 8.393333333333333e-06, "loss": 0.1187, "step": 1260 }, { "epoch": 0.8094327597195666, "grad_norm": 1.5104361772537231, "learning_rate": 8.46e-06, "loss": 0.1122, "step": 1270 }, { "epoch": 0.815806246016571, "grad_norm": 1.4873898029327393, "learning_rate": 8.526666666666667e-06, "loss": 0.1176, "step": 1280 }, { "epoch": 0.8221797323135756, "grad_norm": 2.229330062866211, "learning_rate": 8.593333333333335e-06, "loss": 0.1139, "step": 1290 }, { "epoch": 0.82855321861058, "grad_norm": 1.1514588594436646, "learning_rate": 8.66e-06, "loss": 0.127, "step": 1300 }, { "epoch": 0.8349267049075845, "grad_norm": 1.0396504402160645, "learning_rate": 8.726666666666667e-06, "loss": 0.1243, "step": 1310 }, { "epoch": 0.8413001912045889, "grad_norm": 1.17670738697052, "learning_rate": 8.793333333333334e-06, "loss": 0.1207, "step": 1320 }, { "epoch": 0.8476736775015934, "grad_norm": 1.0504316091537476, "learning_rate": 8.86e-06, "loss": 0.1064, "step": 1330 }, { "epoch": 0.8540471637985978, "grad_norm": 1.0945193767547607, "learning_rate": 8.926666666666667e-06, "loss": 0.1051, "step": 1340 }, { "epoch": 0.8604206500956023, "grad_norm": 0.6964119672775269, "learning_rate": 8.993333333333334e-06, "loss": 0.1262, "step": 1350 }, { "epoch": 0.8667941363926067, "grad_norm": 2.0203912258148193, "learning_rate": 9.06e-06, "loss": 0.1269, "step": 1360 }, { "epoch": 0.8731676226896112, "grad_norm": 2.3909099102020264, "learning_rate": 9.126666666666667e-06, "loss": 0.1114, "step": 1370 }, { "epoch": 0.8795411089866156, "grad_norm": 1.132116436958313, "learning_rate": 9.193333333333334e-06, "loss": 0.1202, "step": 1380 }, { "epoch": 0.8859145952836202, "grad_norm": 1.6137852668762207, "learning_rate": 9.260000000000001e-06, "loss": 0.1088, "step": 1390 }, { "epoch": 0.8922880815806246, "grad_norm": 2.0166521072387695, "learning_rate": 9.326666666666667e-06, "loss": 0.1014, "step": 1400 }, { "epoch": 0.8986615678776291, "grad_norm": 1.089432954788208, "learning_rate": 9.393333333333334e-06, "loss": 0.1186, "step": 1410 }, { "epoch": 0.9050350541746335, "grad_norm": 1.5007840394973755, "learning_rate": 9.460000000000001e-06, "loss": 0.1165, "step": 1420 }, { "epoch": 0.911408540471638, "grad_norm": 1.334079384803772, "learning_rate": 9.526666666666666e-06, "loss": 0.0929, "step": 1430 }, { "epoch": 0.9177820267686424, "grad_norm": 2.3709166049957275, "learning_rate": 9.593333333333334e-06, "loss": 0.1198, "step": 1440 }, { "epoch": 0.9241555130656469, "grad_norm": 1.677902102470398, "learning_rate": 9.66e-06, "loss": 0.1239, "step": 1450 }, { "epoch": 0.9305289993626513, "grad_norm": 1.6223987340927124, "learning_rate": 9.726666666666668e-06, "loss": 0.1049, "step": 1460 }, { "epoch": 0.9369024856596558, "grad_norm": 1.6598254442214966, "learning_rate": 9.793333333333333e-06, "loss": 0.1128, "step": 1470 }, { "epoch": 0.9432759719566602, "grad_norm": 2.6494898796081543, "learning_rate": 9.86e-06, "loss": 0.1157, "step": 1480 }, { "epoch": 0.9496494582536648, "grad_norm": 1.5539672374725342, "learning_rate": 9.926666666666668e-06, "loss": 0.1048, "step": 1490 }, { "epoch": 0.9560229445506692, "grad_norm": 2.537384271621704, "learning_rate": 9.993333333333333e-06, "loss": 0.1218, "step": 1500 }, { "epoch": 0.9623964308476737, "grad_norm": 1.366798996925354, "learning_rate": 1.006e-05, "loss": 0.1126, "step": 1510 }, { "epoch": 0.9687699171446782, "grad_norm": 1.446845293045044, "learning_rate": 1.0126666666666667e-05, "loss": 0.1201, "step": 1520 }, { "epoch": 0.9751434034416826, "grad_norm": 1.1611850261688232, "learning_rate": 1.0193333333333335e-05, "loss": 0.1075, "step": 1530 }, { "epoch": 0.9815168897386871, "grad_norm": 1.2011315822601318, "learning_rate": 1.026e-05, "loss": 0.1019, "step": 1540 }, { "epoch": 0.9878903760356915, "grad_norm": 2.2322311401367188, "learning_rate": 1.0326666666666667e-05, "loss": 0.1226, "step": 1550 }, { "epoch": 0.994263862332696, "grad_norm": 0.8722662329673767, "learning_rate": 1.0393333333333334e-05, "loss": 0.113, "step": 1560 }, { "epoch": 1.0006373486297004, "grad_norm": 1.3424410820007324, "learning_rate": 1.046e-05, "loss": 0.0845, "step": 1570 }, { "epoch": 1.0070108349267048, "grad_norm": 1.2302227020263672, "learning_rate": 1.0526666666666667e-05, "loss": 0.0988, "step": 1580 }, { "epoch": 1.0133843212237095, "grad_norm": 1.23185396194458, "learning_rate": 1.0593333333333334e-05, "loss": 0.1074, "step": 1590 }, { "epoch": 1.0197578075207139, "grad_norm": 1.995958685874939, "learning_rate": 1.0660000000000001e-05, "loss": 0.1182, "step": 1600 }, { "epoch": 1.0261312938177183, "grad_norm": 0.5706260800361633, "learning_rate": 1.0726666666666667e-05, "loss": 0.1015, "step": 1610 }, { "epoch": 1.0325047801147227, "grad_norm": 1.0292627811431885, "learning_rate": 1.0793333333333334e-05, "loss": 0.1, "step": 1620 }, { "epoch": 1.0388782664117273, "grad_norm": 1.7296640872955322, "learning_rate": 1.0860000000000001e-05, "loss": 0.1326, "step": 1630 }, { "epoch": 1.0452517527087317, "grad_norm": 1.43869948387146, "learning_rate": 1.0926666666666667e-05, "loss": 0.1292, "step": 1640 }, { "epoch": 1.0516252390057361, "grad_norm": 1.0628490447998047, "learning_rate": 1.0993333333333334e-05, "loss": 0.1034, "step": 1650 }, { "epoch": 1.0579987253027405, "grad_norm": 1.5121562480926514, "learning_rate": 1.106e-05, "loss": 0.1076, "step": 1660 }, { "epoch": 1.0643722115997452, "grad_norm": 1.4332911968231201, "learning_rate": 1.1126666666666668e-05, "loss": 0.0942, "step": 1670 }, { "epoch": 1.0707456978967496, "grad_norm": 1.0733493566513062, "learning_rate": 1.1193333333333333e-05, "loss": 0.103, "step": 1680 }, { "epoch": 1.077119184193754, "grad_norm": 1.2125245332717896, "learning_rate": 1.126e-05, "loss": 0.1027, "step": 1690 }, { "epoch": 1.0834926704907584, "grad_norm": 1.374131202697754, "learning_rate": 1.1326666666666668e-05, "loss": 0.1063, "step": 1700 }, { "epoch": 1.089866156787763, "grad_norm": 1.331972360610962, "learning_rate": 1.1393333333333333e-05, "loss": 0.1032, "step": 1710 }, { "epoch": 1.0962396430847674, "grad_norm": 1.8142362833023071, "learning_rate": 1.146e-05, "loss": 0.1147, "step": 1720 }, { "epoch": 1.1026131293817718, "grad_norm": 1.474809169769287, "learning_rate": 1.1526666666666668e-05, "loss": 0.1012, "step": 1730 }, { "epoch": 1.1089866156787762, "grad_norm": 1.0845487117767334, "learning_rate": 1.1593333333333333e-05, "loss": 0.1036, "step": 1740 }, { "epoch": 1.1153601019757808, "grad_norm": 1.7612701654434204, "learning_rate": 1.166e-05, "loss": 0.1063, "step": 1750 }, { "epoch": 1.1217335882727852, "grad_norm": 0.7480101585388184, "learning_rate": 1.1726666666666667e-05, "loss": 0.1016, "step": 1760 }, { "epoch": 1.1281070745697896, "grad_norm": 0.9497768878936768, "learning_rate": 1.1793333333333334e-05, "loss": 0.0917, "step": 1770 }, { "epoch": 1.134480560866794, "grad_norm": 1.1517006158828735, "learning_rate": 1.186e-05, "loss": 0.1166, "step": 1780 }, { "epoch": 1.1408540471637987, "grad_norm": 1.1664561033248901, "learning_rate": 1.1926666666666667e-05, "loss": 0.0917, "step": 1790 }, { "epoch": 1.147227533460803, "grad_norm": 2.021282434463501, "learning_rate": 1.1993333333333334e-05, "loss": 0.1115, "step": 1800 }, { "epoch": 1.1536010197578075, "grad_norm": 1.5622092485427856, "learning_rate": 1.206e-05, "loss": 0.1136, "step": 1810 }, { "epoch": 1.159974506054812, "grad_norm": 1.4180028438568115, "learning_rate": 1.2126666666666667e-05, "loss": 0.1024, "step": 1820 }, { "epoch": 1.1663479923518165, "grad_norm": 2.3993589878082275, "learning_rate": 1.2193333333333334e-05, "loss": 0.1037, "step": 1830 }, { "epoch": 1.172721478648821, "grad_norm": 1.6957080364227295, "learning_rate": 1.2260000000000001e-05, "loss": 0.0915, "step": 1840 }, { "epoch": 1.1790949649458253, "grad_norm": 1.1840903759002686, "learning_rate": 1.2326666666666667e-05, "loss": 0.1119, "step": 1850 }, { "epoch": 1.1854684512428297, "grad_norm": 1.1787384748458862, "learning_rate": 1.2393333333333334e-05, "loss": 0.1034, "step": 1860 }, { "epoch": 1.1918419375398344, "grad_norm": 1.8163771629333496, "learning_rate": 1.2460000000000001e-05, "loss": 0.1029, "step": 1870 }, { "epoch": 1.1982154238368388, "grad_norm": 1.1066467761993408, "learning_rate": 1.2526666666666666e-05, "loss": 0.086, "step": 1880 }, { "epoch": 1.2045889101338432, "grad_norm": 1.2255226373672485, "learning_rate": 1.2593333333333335e-05, "loss": 0.1021, "step": 1890 }, { "epoch": 1.2109623964308476, "grad_norm": 1.2998727560043335, "learning_rate": 1.2659999999999999e-05, "loss": 0.0968, "step": 1900 }, { "epoch": 1.2173358827278522, "grad_norm": 1.1388373374938965, "learning_rate": 1.2726666666666668e-05, "loss": 0.1145, "step": 1910 }, { "epoch": 1.2237093690248566, "grad_norm": 2.0356638431549072, "learning_rate": 1.2793333333333335e-05, "loss": 0.1188, "step": 1920 }, { "epoch": 1.230082855321861, "grad_norm": 1.5201404094696045, "learning_rate": 1.286e-05, "loss": 0.0939, "step": 1930 }, { "epoch": 1.2364563416188656, "grad_norm": 2.302424907684326, "learning_rate": 1.2926666666666668e-05, "loss": 0.1095, "step": 1940 }, { "epoch": 1.24282982791587, "grad_norm": 1.1126807928085327, "learning_rate": 1.2993333333333335e-05, "loss": 0.0864, "step": 1950 }, { "epoch": 1.2492033142128744, "grad_norm": 1.4584306478500366, "learning_rate": 1.306e-05, "loss": 0.0968, "step": 1960 }, { "epoch": 1.2555768005098789, "grad_norm": 1.5292973518371582, "learning_rate": 1.3126666666666667e-05, "loss": 0.0891, "step": 1970 }, { "epoch": 1.2619502868068833, "grad_norm": 1.7257049083709717, "learning_rate": 1.3193333333333335e-05, "loss": 0.1062, "step": 1980 }, { "epoch": 1.2683237731038879, "grad_norm": 0.7163188457489014, "learning_rate": 1.326e-05, "loss": 0.1001, "step": 1990 }, { "epoch": 1.2746972594008923, "grad_norm": 0.9911763072013855, "learning_rate": 1.3326666666666667e-05, "loss": 0.1065, "step": 2000 }, { "epoch": 1.2810707456978967, "grad_norm": 1.2415435314178467, "learning_rate": 1.3393333333333333e-05, "loss": 0.104, "step": 2010 }, { "epoch": 1.2874442319949013, "grad_norm": 1.0172119140625, "learning_rate": 1.346e-05, "loss": 0.0911, "step": 2020 }, { "epoch": 1.2938177182919057, "grad_norm": 1.5576133728027344, "learning_rate": 1.3526666666666669e-05, "loss": 0.0859, "step": 2030 }, { "epoch": 1.3001912045889101, "grad_norm": 1.4438010454177856, "learning_rate": 1.3593333333333332e-05, "loss": 0.1007, "step": 2040 }, { "epoch": 1.3065646908859145, "grad_norm": 1.0720605850219727, "learning_rate": 1.3660000000000001e-05, "loss": 0.0864, "step": 2050 }, { "epoch": 1.312938177182919, "grad_norm": 1.5696890354156494, "learning_rate": 1.3726666666666669e-05, "loss": 0.0873, "step": 2060 }, { "epoch": 1.3193116634799236, "grad_norm": 1.2315691709518433, "learning_rate": 1.3793333333333332e-05, "loss": 0.1094, "step": 2070 }, { "epoch": 1.325685149776928, "grad_norm": 1.1183645725250244, "learning_rate": 1.3860000000000001e-05, "loss": 0.0857, "step": 2080 }, { "epoch": 1.3320586360739324, "grad_norm": 1.6838148832321167, "learning_rate": 1.3926666666666668e-05, "loss": 0.1003, "step": 2090 }, { "epoch": 1.338432122370937, "grad_norm": 1.1183300018310547, "learning_rate": 1.3993333333333334e-05, "loss": 0.0831, "step": 2100 }, { "epoch": 1.3448056086679414, "grad_norm": 1.8366953134536743, "learning_rate": 1.4060000000000001e-05, "loss": 0.111, "step": 2110 }, { "epoch": 1.3511790949649458, "grad_norm": 1.9242589473724365, "learning_rate": 1.4126666666666668e-05, "loss": 0.1073, "step": 2120 }, { "epoch": 1.3575525812619502, "grad_norm": 1.864620327949524, "learning_rate": 1.4193333333333334e-05, "loss": 0.0901, "step": 2130 }, { "epoch": 1.3639260675589546, "grad_norm": 1.115146279335022, "learning_rate": 1.426e-05, "loss": 0.0825, "step": 2140 }, { "epoch": 1.3702995538559593, "grad_norm": 1.6277066469192505, "learning_rate": 1.4326666666666666e-05, "loss": 0.1161, "step": 2150 }, { "epoch": 1.3766730401529637, "grad_norm": 1.6365883350372314, "learning_rate": 1.4393333333333333e-05, "loss": 0.0848, "step": 2160 }, { "epoch": 1.383046526449968, "grad_norm": 1.2011946439743042, "learning_rate": 1.4460000000000002e-05, "loss": 0.0996, "step": 2170 }, { "epoch": 1.3894200127469727, "grad_norm": 1.8417644500732422, "learning_rate": 1.4526666666666666e-05, "loss": 0.1142, "step": 2180 }, { "epoch": 1.395793499043977, "grad_norm": 1.2948949337005615, "learning_rate": 1.4593333333333333e-05, "loss": 0.1095, "step": 2190 }, { "epoch": 1.4021669853409815, "grad_norm": 1.2903964519500732, "learning_rate": 1.4660000000000002e-05, "loss": 0.0864, "step": 2200 }, { "epoch": 1.408540471637986, "grad_norm": 1.643641710281372, "learning_rate": 1.4726666666666666e-05, "loss": 0.094, "step": 2210 }, { "epoch": 1.4149139579349903, "grad_norm": 1.1676900386810303, "learning_rate": 1.4793333333333335e-05, "loss": 0.0972, "step": 2220 }, { "epoch": 1.421287444231995, "grad_norm": 0.9590888023376465, "learning_rate": 1.4860000000000002e-05, "loss": 0.0826, "step": 2230 }, { "epoch": 1.4276609305289993, "grad_norm": 1.325158953666687, "learning_rate": 1.4926666666666667e-05, "loss": 0.1018, "step": 2240 }, { "epoch": 1.4340344168260037, "grad_norm": 0.6889841556549072, "learning_rate": 1.4993333333333334e-05, "loss": 0.0875, "step": 2250 }, { "epoch": 1.4404079031230084, "grad_norm": 1.7010300159454346, "learning_rate": 1.5060000000000001e-05, "loss": 0.088, "step": 2260 }, { "epoch": 1.4467813894200128, "grad_norm": 0.9164239168167114, "learning_rate": 1.5126666666666667e-05, "loss": 0.0886, "step": 2270 }, { "epoch": 1.4531548757170172, "grad_norm": 1.0260887145996094, "learning_rate": 1.5193333333333334e-05, "loss": 0.0828, "step": 2280 }, { "epoch": 1.4595283620140216, "grad_norm": 1.2509478330612183, "learning_rate": 1.5260000000000003e-05, "loss": 0.1153, "step": 2290 }, { "epoch": 1.4659018483110262, "grad_norm": 1.040419578552246, "learning_rate": 1.5326666666666667e-05, "loss": 0.0992, "step": 2300 }, { "epoch": 1.4722753346080306, "grad_norm": 1.8140629529953003, "learning_rate": 1.5393333333333334e-05, "loss": 0.0923, "step": 2310 }, { "epoch": 1.478648820905035, "grad_norm": 1.4857326745986938, "learning_rate": 1.546e-05, "loss": 0.0777, "step": 2320 }, { "epoch": 1.4850223072020396, "grad_norm": 1.146477460861206, "learning_rate": 1.5526666666666668e-05, "loss": 0.1113, "step": 2330 }, { "epoch": 1.491395793499044, "grad_norm": 1.3774482011795044, "learning_rate": 1.5593333333333335e-05, "loss": 0.0964, "step": 2340 }, { "epoch": 1.4977692797960485, "grad_norm": 0.9920369982719421, "learning_rate": 1.566e-05, "loss": 0.0965, "step": 2350 }, { "epoch": 1.5041427660930529, "grad_norm": 0.6378938555717468, "learning_rate": 1.5726666666666666e-05, "loss": 0.0765, "step": 2360 }, { "epoch": 1.5105162523900573, "grad_norm": 0.9574321508407593, "learning_rate": 1.5793333333333333e-05, "loss": 0.1047, "step": 2370 }, { "epoch": 1.5168897386870617, "grad_norm": 1.6412683725357056, "learning_rate": 1.586e-05, "loss": 0.083, "step": 2380 }, { "epoch": 1.5232632249840663, "grad_norm": 1.1442257165908813, "learning_rate": 1.5926666666666668e-05, "loss": 0.1049, "step": 2390 }, { "epoch": 1.5296367112810707, "grad_norm": 1.5450184345245361, "learning_rate": 1.5993333333333335e-05, "loss": 0.0973, "step": 2400 }, { "epoch": 1.5360101975780753, "grad_norm": 1.226181149482727, "learning_rate": 1.606e-05, "loss": 0.093, "step": 2410 }, { "epoch": 1.5423836838750797, "grad_norm": 0.7279766798019409, "learning_rate": 1.612666666666667e-05, "loss": 0.1147, "step": 2420 }, { "epoch": 1.5487571701720841, "grad_norm": 0.9621545076370239, "learning_rate": 1.6193333333333336e-05, "loss": 0.0895, "step": 2430 }, { "epoch": 1.5551306564690885, "grad_norm": 2.1044633388519287, "learning_rate": 1.626e-05, "loss": 0.1017, "step": 2440 }, { "epoch": 1.561504142766093, "grad_norm": 0.9241976737976074, "learning_rate": 1.6326666666666667e-05, "loss": 0.0952, "step": 2450 }, { "epoch": 1.5678776290630974, "grad_norm": 1.1970354318618774, "learning_rate": 1.639333333333333e-05, "loss": 0.0748, "step": 2460 }, { "epoch": 1.574251115360102, "grad_norm": 0.6390184760093689, "learning_rate": 1.646e-05, "loss": 0.0833, "step": 2470 }, { "epoch": 1.5806246016571064, "grad_norm": 1.1987965106964111, "learning_rate": 1.652666666666667e-05, "loss": 0.0919, "step": 2480 }, { "epoch": 1.586998087954111, "grad_norm": 1.0885887145996094, "learning_rate": 1.6593333333333333e-05, "loss": 0.0812, "step": 2490 }, { "epoch": 1.5933715742511154, "grad_norm": 0.7863695025444031, "learning_rate": 1.666e-05, "loss": 0.0861, "step": 2500 }, { "epoch": 1.5997450605481198, "grad_norm": 1.272106409072876, "learning_rate": 1.6726666666666667e-05, "loss": 0.0728, "step": 2510 }, { "epoch": 1.6061185468451242, "grad_norm": 0.9821702837944031, "learning_rate": 1.6793333333333334e-05, "loss": 0.0795, "step": 2520 }, { "epoch": 1.6124920331421286, "grad_norm": 0.9293336868286133, "learning_rate": 1.686e-05, "loss": 0.0723, "step": 2530 }, { "epoch": 1.6188655194391333, "grad_norm": 2.0479531288146973, "learning_rate": 1.692666666666667e-05, "loss": 0.0938, "step": 2540 }, { "epoch": 1.6252390057361377, "grad_norm": 1.4463709592819214, "learning_rate": 1.6993333333333332e-05, "loss": 0.0931, "step": 2550 }, { "epoch": 1.6316124920331423, "grad_norm": 1.106081485748291, "learning_rate": 1.706e-05, "loss": 0.084, "step": 2560 }, { "epoch": 1.6379859783301467, "grad_norm": 1.3168553113937378, "learning_rate": 1.712666666666667e-05, "loss": 0.0861, "step": 2570 }, { "epoch": 1.644359464627151, "grad_norm": 1.2882040739059448, "learning_rate": 1.7193333333333334e-05, "loss": 0.0885, "step": 2580 }, { "epoch": 1.6507329509241555, "grad_norm": 1.623707890510559, "learning_rate": 1.726e-05, "loss": 0.087, "step": 2590 }, { "epoch": 1.65710643722116, "grad_norm": 1.4060157537460327, "learning_rate": 1.7326666666666668e-05, "loss": 0.1084, "step": 2600 }, { "epoch": 1.6634799235181643, "grad_norm": 1.3172701597213745, "learning_rate": 1.7393333333333335e-05, "loss": 0.0881, "step": 2610 }, { "epoch": 1.669853409815169, "grad_norm": 1.099621295928955, "learning_rate": 1.7460000000000002e-05, "loss": 0.072, "step": 2620 }, { "epoch": 1.6762268961121733, "grad_norm": 1.0552417039871216, "learning_rate": 1.7526666666666666e-05, "loss": 0.1015, "step": 2630 }, { "epoch": 1.682600382409178, "grad_norm": 1.4513311386108398, "learning_rate": 1.7593333333333333e-05, "loss": 0.0931, "step": 2640 }, { "epoch": 1.6889738687061824, "grad_norm": 0.7885988354682922, "learning_rate": 1.766e-05, "loss": 0.0897, "step": 2650 }, { "epoch": 1.6953473550031868, "grad_norm": 0.7710033059120178, "learning_rate": 1.7726666666666667e-05, "loss": 0.0821, "step": 2660 }, { "epoch": 1.7017208413001912, "grad_norm": 1.3100814819335938, "learning_rate": 1.7793333333333335e-05, "loss": 0.1115, "step": 2670 }, { "epoch": 1.7080943275971956, "grad_norm": 1.3566712141036987, "learning_rate": 1.7860000000000002e-05, "loss": 0.1096, "step": 2680 }, { "epoch": 1.7144678138942, "grad_norm": 0.9596314430236816, "learning_rate": 1.7926666666666666e-05, "loss": 0.0943, "step": 2690 }, { "epoch": 1.7208413001912046, "grad_norm": 1.5328532457351685, "learning_rate": 1.7993333333333333e-05, "loss": 0.0847, "step": 2700 }, { "epoch": 1.727214786488209, "grad_norm": 1.2858617305755615, "learning_rate": 1.8060000000000003e-05, "loss": 0.098, "step": 2710 }, { "epoch": 1.7335882727852137, "grad_norm": 1.620842456817627, "learning_rate": 1.8126666666666667e-05, "loss": 0.1128, "step": 2720 }, { "epoch": 1.739961759082218, "grad_norm": 1.5014934539794922, "learning_rate": 1.8193333333333334e-05, "loss": 0.0933, "step": 2730 }, { "epoch": 1.7463352453792225, "grad_norm": 1.610737919807434, "learning_rate": 1.826e-05, "loss": 0.0838, "step": 2740 }, { "epoch": 1.7527087316762269, "grad_norm": 1.0811647176742554, "learning_rate": 1.832666666666667e-05, "loss": 0.0785, "step": 2750 }, { "epoch": 1.7590822179732313, "grad_norm": 1.580681562423706, "learning_rate": 1.8393333333333336e-05, "loss": 0.0839, "step": 2760 }, { "epoch": 1.7654557042702357, "grad_norm": 1.1532455682754517, "learning_rate": 1.846e-05, "loss": 0.0958, "step": 2770 }, { "epoch": 1.7718291905672403, "grad_norm": 1.2616368532180786, "learning_rate": 1.8526666666666667e-05, "loss": 0.0991, "step": 2780 }, { "epoch": 1.7782026768642447, "grad_norm": 0.7279543876647949, "learning_rate": 1.8593333333333334e-05, "loss": 0.0807, "step": 2790 }, { "epoch": 1.7845761631612493, "grad_norm": 1.0883718729019165, "learning_rate": 1.866e-05, "loss": 0.0908, "step": 2800 }, { "epoch": 1.7909496494582537, "grad_norm": 0.8580917716026306, "learning_rate": 1.8726666666666668e-05, "loss": 0.099, "step": 2810 }, { "epoch": 1.7973231357552581, "grad_norm": 1.086074948310852, "learning_rate": 1.8793333333333335e-05, "loss": 0.1012, "step": 2820 }, { "epoch": 1.8036966220522626, "grad_norm": 0.8041730523109436, "learning_rate": 1.886e-05, "loss": 0.0895, "step": 2830 }, { "epoch": 1.810070108349267, "grad_norm": 0.9425352811813354, "learning_rate": 1.8926666666666666e-05, "loss": 0.0876, "step": 2840 }, { "epoch": 1.8164435946462714, "grad_norm": 0.9173586964607239, "learning_rate": 1.8993333333333337e-05, "loss": 0.0871, "step": 2850 }, { "epoch": 1.822817080943276, "grad_norm": 1.0493013858795166, "learning_rate": 1.906e-05, "loss": 0.0835, "step": 2860 }, { "epoch": 1.8291905672402804, "grad_norm": 1.195335030555725, "learning_rate": 1.9126666666666668e-05, "loss": 0.0834, "step": 2870 }, { "epoch": 1.835564053537285, "grad_norm": 1.6238062381744385, "learning_rate": 1.9193333333333335e-05, "loss": 0.0706, "step": 2880 }, { "epoch": 1.8419375398342894, "grad_norm": 1.0624531507492065, "learning_rate": 1.9260000000000002e-05, "loss": 0.0985, "step": 2890 }, { "epoch": 1.8483110261312938, "grad_norm": 1.3885860443115234, "learning_rate": 1.932666666666667e-05, "loss": 0.0773, "step": 2900 }, { "epoch": 1.8546845124282982, "grad_norm": 1.0686782598495483, "learning_rate": 1.9393333333333336e-05, "loss": 0.0889, "step": 2910 }, { "epoch": 1.8610579987253026, "grad_norm": 0.9558495879173279, "learning_rate": 1.946e-05, "loss": 0.0956, "step": 2920 }, { "epoch": 1.867431485022307, "grad_norm": 1.1304773092269897, "learning_rate": 1.9526666666666667e-05, "loss": 0.0631, "step": 2930 }, { "epoch": 1.8738049713193117, "grad_norm": 1.3271628618240356, "learning_rate": 1.9593333333333334e-05, "loss": 0.0731, "step": 2940 }, { "epoch": 1.880178457616316, "grad_norm": 0.8929280638694763, "learning_rate": 1.966e-05, "loss": 0.0736, "step": 2950 }, { "epoch": 1.8865519439133207, "grad_norm": 0.6617772579193115, "learning_rate": 1.972666666666667e-05, "loss": 0.0883, "step": 2960 }, { "epoch": 1.892925430210325, "grad_norm": 1.1443796157836914, "learning_rate": 1.9793333333333332e-05, "loss": 0.0795, "step": 2970 }, { "epoch": 1.8992989165073295, "grad_norm": 1.21043860912323, "learning_rate": 1.986e-05, "loss": 0.0801, "step": 2980 }, { "epoch": 1.905672402804334, "grad_norm": 1.102117896080017, "learning_rate": 1.992666666666667e-05, "loss": 0.0873, "step": 2990 }, { "epoch": 1.9120458891013383, "grad_norm": 1.3731114864349365, "learning_rate": 1.9993333333333334e-05, "loss": 0.0952, "step": 3000 }, { "epoch": 1.9184193753983427, "grad_norm": 1.1329739093780518, "learning_rate": 2.006e-05, "loss": 0.0706, "step": 3010 }, { "epoch": 1.9247928616953474, "grad_norm": 0.5612623691558838, "learning_rate": 2.0126666666666668e-05, "loss": 0.0698, "step": 3020 }, { "epoch": 1.9311663479923518, "grad_norm": 1.3422327041625977, "learning_rate": 2.0193333333333332e-05, "loss": 0.0916, "step": 3030 }, { "epoch": 1.9375398342893564, "grad_norm": 1.1616122722625732, "learning_rate": 2.0260000000000003e-05, "loss": 0.0706, "step": 3040 }, { "epoch": 1.9439133205863608, "grad_norm": 0.9410479068756104, "learning_rate": 2.032666666666667e-05, "loss": 0.0767, "step": 3050 }, { "epoch": 1.9502868068833652, "grad_norm": 1.0274105072021484, "learning_rate": 2.0393333333333333e-05, "loss": 0.0841, "step": 3060 }, { "epoch": 1.9566602931803696, "grad_norm": 0.9492813348770142, "learning_rate": 2.046e-05, "loss": 0.0653, "step": 3070 }, { "epoch": 1.963033779477374, "grad_norm": 1.442031741142273, "learning_rate": 2.0526666666666668e-05, "loss": 0.0876, "step": 3080 }, { "epoch": 1.9694072657743786, "grad_norm": 1.257333755493164, "learning_rate": 2.0593333333333335e-05, "loss": 0.1017, "step": 3090 }, { "epoch": 1.975780752071383, "grad_norm": 1.355202317237854, "learning_rate": 2.0660000000000002e-05, "loss": 0.0953, "step": 3100 }, { "epoch": 1.9821542383683877, "grad_norm": 1.1883312463760376, "learning_rate": 2.0726666666666666e-05, "loss": 0.0733, "step": 3110 }, { "epoch": 1.988527724665392, "grad_norm": 0.946861207485199, "learning_rate": 2.0793333333333333e-05, "loss": 0.0594, "step": 3120 }, { "epoch": 1.9949012109623965, "grad_norm": 1.412591814994812, "learning_rate": 2.086e-05, "loss": 0.0817, "step": 3130 }, { "epoch": 2.001274697259401, "grad_norm": 1.3307621479034424, "learning_rate": 2.0926666666666667e-05, "loss": 0.0782, "step": 3140 }, { "epoch": 2.0076481835564053, "grad_norm": 1.5508772134780884, "learning_rate": 2.0993333333333334e-05, "loss": 0.0769, "step": 3150 }, { "epoch": 2.0140216698534097, "grad_norm": 1.8503458499908447, "learning_rate": 2.106e-05, "loss": 0.0853, "step": 3160 }, { "epoch": 2.020395156150414, "grad_norm": 1.7811330556869507, "learning_rate": 2.1126666666666665e-05, "loss": 0.0721, "step": 3170 }, { "epoch": 2.026768642447419, "grad_norm": 1.1020867824554443, "learning_rate": 2.1193333333333336e-05, "loss": 0.0847, "step": 3180 }, { "epoch": 2.0331421287444233, "grad_norm": 1.0944749116897583, "learning_rate": 2.1260000000000003e-05, "loss": 0.0853, "step": 3190 }, { "epoch": 2.0395156150414278, "grad_norm": 1.286003828048706, "learning_rate": 2.1326666666666667e-05, "loss": 0.0749, "step": 3200 }, { "epoch": 2.045889101338432, "grad_norm": 0.9382231831550598, "learning_rate": 2.1393333333333334e-05, "loss": 0.0794, "step": 3210 }, { "epoch": 2.0522625876354366, "grad_norm": 1.6390314102172852, "learning_rate": 2.146e-05, "loss": 0.1117, "step": 3220 }, { "epoch": 2.058636073932441, "grad_norm": 0.8031259775161743, "learning_rate": 2.152666666666667e-05, "loss": 0.0954, "step": 3230 }, { "epoch": 2.0650095602294454, "grad_norm": 1.2209105491638184, "learning_rate": 2.1593333333333336e-05, "loss": 0.0822, "step": 3240 }, { "epoch": 2.07138304652645, "grad_norm": 1.52094566822052, "learning_rate": 2.166e-05, "loss": 0.0915, "step": 3250 }, { "epoch": 2.0777565328234546, "grad_norm": 1.176224708557129, "learning_rate": 2.1726666666666666e-05, "loss": 0.0861, "step": 3260 }, { "epoch": 2.084130019120459, "grad_norm": 0.6373295783996582, "learning_rate": 2.1793333333333334e-05, "loss": 0.0731, "step": 3270 }, { "epoch": 2.0905035054174634, "grad_norm": 1.152594804763794, "learning_rate": 2.186e-05, "loss": 0.0755, "step": 3280 }, { "epoch": 2.096876991714468, "grad_norm": 1.1399019956588745, "learning_rate": 2.1926666666666668e-05, "loss": 0.084, "step": 3290 }, { "epoch": 2.1032504780114722, "grad_norm": 1.0591745376586914, "learning_rate": 2.1993333333333335e-05, "loss": 0.0769, "step": 3300 }, { "epoch": 2.1096239643084767, "grad_norm": 0.8013867139816284, "learning_rate": 2.206e-05, "loss": 0.0816, "step": 3310 }, { "epoch": 2.115997450605481, "grad_norm": 1.0244139432907104, "learning_rate": 2.212666666666667e-05, "loss": 0.0906, "step": 3320 }, { "epoch": 2.1223709369024855, "grad_norm": 1.5283808708190918, "learning_rate": 2.2193333333333337e-05, "loss": 0.0755, "step": 3330 }, { "epoch": 2.1287444231994903, "grad_norm": 1.2056852579116821, "learning_rate": 2.226e-05, "loss": 0.0719, "step": 3340 }, { "epoch": 2.1351179094964947, "grad_norm": 0.6835068464279175, "learning_rate": 2.2326666666666667e-05, "loss": 0.079, "step": 3350 }, { "epoch": 2.141491395793499, "grad_norm": 0.9059255719184875, "learning_rate": 2.2393333333333335e-05, "loss": 0.0686, "step": 3360 }, { "epoch": 2.1478648820905035, "grad_norm": 1.4286351203918457, "learning_rate": 2.2460000000000002e-05, "loss": 0.0956, "step": 3370 }, { "epoch": 2.154238368387508, "grad_norm": 1.0044485330581665, "learning_rate": 2.252666666666667e-05, "loss": 0.1064, "step": 3380 }, { "epoch": 2.1606118546845123, "grad_norm": 1.326513409614563, "learning_rate": 2.2593333333333336e-05, "loss": 0.076, "step": 3390 }, { "epoch": 2.1669853409815167, "grad_norm": 1.187397837638855, "learning_rate": 2.266e-05, "loss": 0.0863, "step": 3400 }, { "epoch": 2.173358827278521, "grad_norm": 1.5882831811904907, "learning_rate": 2.2726666666666667e-05, "loss": 0.0848, "step": 3410 }, { "epoch": 2.179732313575526, "grad_norm": 0.9371598362922668, "learning_rate": 2.2793333333333334e-05, "loss": 0.0665, "step": 3420 }, { "epoch": 2.1861057998725304, "grad_norm": 0.8871974349021912, "learning_rate": 2.286e-05, "loss": 0.0729, "step": 3430 }, { "epoch": 2.192479286169535, "grad_norm": 1.0369257926940918, "learning_rate": 2.292666666666667e-05, "loss": 0.0834, "step": 3440 }, { "epoch": 2.198852772466539, "grad_norm": 1.0512542724609375, "learning_rate": 2.2993333333333332e-05, "loss": 0.0655, "step": 3450 }, { "epoch": 2.2052262587635436, "grad_norm": 1.6292001008987427, "learning_rate": 2.306e-05, "loss": 0.0735, "step": 3460 }, { "epoch": 2.211599745060548, "grad_norm": 1.6295171976089478, "learning_rate": 2.312666666666667e-05, "loss": 0.0862, "step": 3470 }, { "epoch": 2.2179732313575524, "grad_norm": 1.0787677764892578, "learning_rate": 2.3193333333333334e-05, "loss": 0.0896, "step": 3480 }, { "epoch": 2.224346717654557, "grad_norm": 1.106917142868042, "learning_rate": 2.326e-05, "loss": 0.0891, "step": 3490 }, { "epoch": 2.2307202039515617, "grad_norm": 0.6044514775276184, "learning_rate": 2.3326666666666668e-05, "loss": 0.0665, "step": 3500 }, { "epoch": 2.237093690248566, "grad_norm": 1.183518886566162, "learning_rate": 2.3393333333333335e-05, "loss": 0.0681, "step": 3510 }, { "epoch": 2.2434671765455705, "grad_norm": 1.143831729888916, "learning_rate": 2.3460000000000002e-05, "loss": 0.0831, "step": 3520 }, { "epoch": 2.249840662842575, "grad_norm": 1.1012294292449951, "learning_rate": 2.352666666666667e-05, "loss": 0.079, "step": 3530 }, { "epoch": 2.2562141491395793, "grad_norm": 0.9378037452697754, "learning_rate": 2.3593333333333333e-05, "loss": 0.0695, "step": 3540 }, { "epoch": 2.2625876354365837, "grad_norm": 0.7672236561775208, "learning_rate": 2.366e-05, "loss": 0.0677, "step": 3550 }, { "epoch": 2.268961121733588, "grad_norm": 0.7897099256515503, "learning_rate": 2.3726666666666668e-05, "loss": 0.0631, "step": 3560 }, { "epoch": 2.275334608030593, "grad_norm": 1.0121351480484009, "learning_rate": 2.3793333333333335e-05, "loss": 0.0847, "step": 3570 }, { "epoch": 2.2817080943275974, "grad_norm": 1.0353792905807495, "learning_rate": 2.3860000000000002e-05, "loss": 0.0957, "step": 3580 }, { "epoch": 2.2880815806246018, "grad_norm": 0.9366838335990906, "learning_rate": 2.3926666666666666e-05, "loss": 0.089, "step": 3590 }, { "epoch": 2.294455066921606, "grad_norm": 1.0918177366256714, "learning_rate": 2.3993333333333333e-05, "loss": 0.0732, "step": 3600 }, { "epoch": 2.3008285532186106, "grad_norm": 0.9088542461395264, "learning_rate": 2.4060000000000003e-05, "loss": 0.0696, "step": 3610 }, { "epoch": 2.307202039515615, "grad_norm": 0.7565183043479919, "learning_rate": 2.4126666666666667e-05, "loss": 0.0745, "step": 3620 }, { "epoch": 2.3135755258126194, "grad_norm": 0.739345133304596, "learning_rate": 2.4193333333333334e-05, "loss": 0.0619, "step": 3630 }, { "epoch": 2.319949012109624, "grad_norm": 0.9682939648628235, "learning_rate": 2.426e-05, "loss": 0.0828, "step": 3640 }, { "epoch": 2.326322498406628, "grad_norm": 1.3012282848358154, "learning_rate": 2.432666666666667e-05, "loss": 0.0697, "step": 3650 }, { "epoch": 2.332695984703633, "grad_norm": 1.3370519876480103, "learning_rate": 2.4393333333333336e-05, "loss": 0.0693, "step": 3660 }, { "epoch": 2.3390694710006374, "grad_norm": 1.290480136871338, "learning_rate": 2.4460000000000003e-05, "loss": 0.0806, "step": 3670 }, { "epoch": 2.345442957297642, "grad_norm": 1.7388986349105835, "learning_rate": 2.4526666666666667e-05, "loss": 0.0871, "step": 3680 }, { "epoch": 2.3518164435946463, "grad_norm": 1.2476669549942017, "learning_rate": 2.4593333333333334e-05, "loss": 0.0777, "step": 3690 }, { "epoch": 2.3581899298916507, "grad_norm": 1.3203452825546265, "learning_rate": 2.466e-05, "loss": 0.0826, "step": 3700 }, { "epoch": 2.364563416188655, "grad_norm": 0.8643007278442383, "learning_rate": 2.4726666666666668e-05, "loss": 0.0802, "step": 3710 }, { "epoch": 2.3709369024856595, "grad_norm": 1.2604588270187378, "learning_rate": 2.4793333333333335e-05, "loss": 0.0928, "step": 3720 }, { "epoch": 2.3773103887826643, "grad_norm": 1.1747511625289917, "learning_rate": 2.486e-05, "loss": 0.074, "step": 3730 }, { "epoch": 2.3836838750796687, "grad_norm": 1.140848994255066, "learning_rate": 2.4926666666666666e-05, "loss": 0.0805, "step": 3740 }, { "epoch": 2.390057361376673, "grad_norm": 0.9357328414916992, "learning_rate": 2.4993333333333337e-05, "loss": 0.0816, "step": 3750 }, { "epoch": 2.3964308476736775, "grad_norm": 0.9001865983009338, "learning_rate": 2.506e-05, "loss": 0.0637, "step": 3760 }, { "epoch": 2.402804333970682, "grad_norm": 0.8584229946136475, "learning_rate": 2.512666666666667e-05, "loss": 0.0862, "step": 3770 }, { "epoch": 2.4091778202676863, "grad_norm": 1.199390172958374, "learning_rate": 2.5193333333333335e-05, "loss": 0.089, "step": 3780 }, { "epoch": 2.4155513065646907, "grad_norm": 1.5466145277023315, "learning_rate": 2.526e-05, "loss": 0.0731, "step": 3790 }, { "epoch": 2.421924792861695, "grad_norm": 0.89247727394104, "learning_rate": 2.5326666666666666e-05, "loss": 0.0951, "step": 3800 }, { "epoch": 2.4282982791587, "grad_norm": 0.83308345079422, "learning_rate": 2.5393333333333336e-05, "loss": 0.0605, "step": 3810 }, { "epoch": 2.4346717654557044, "grad_norm": 1.4955074787139893, "learning_rate": 2.546e-05, "loss": 0.0738, "step": 3820 }, { "epoch": 2.441045251752709, "grad_norm": 0.7073909640312195, "learning_rate": 2.5526666666666664e-05, "loss": 0.0712, "step": 3830 }, { "epoch": 2.447418738049713, "grad_norm": 1.0291156768798828, "learning_rate": 2.5593333333333334e-05, "loss": 0.0624, "step": 3840 }, { "epoch": 2.4537922243467176, "grad_norm": 1.3625471591949463, "learning_rate": 2.566e-05, "loss": 0.1009, "step": 3850 }, { "epoch": 2.460165710643722, "grad_norm": 1.9747023582458496, "learning_rate": 2.5726666666666665e-05, "loss": 0.0915, "step": 3860 }, { "epoch": 2.4665391969407264, "grad_norm": 1.3161057233810425, "learning_rate": 2.5793333333333336e-05, "loss": 0.0742, "step": 3870 }, { "epoch": 2.4729126832377313, "grad_norm": 0.5482892394065857, "learning_rate": 2.586e-05, "loss": 0.0676, "step": 3880 }, { "epoch": 2.4792861695347357, "grad_norm": 0.6921859383583069, "learning_rate": 2.5926666666666667e-05, "loss": 0.0746, "step": 3890 }, { "epoch": 2.48565965583174, "grad_norm": 1.1666264533996582, "learning_rate": 2.5993333333333337e-05, "loss": 0.0915, "step": 3900 }, { "epoch": 2.4920331421287445, "grad_norm": 0.7556964159011841, "learning_rate": 2.606e-05, "loss": 0.0715, "step": 3910 }, { "epoch": 2.498406628425749, "grad_norm": 0.768217921257019, "learning_rate": 2.6126666666666665e-05, "loss": 0.0741, "step": 3920 }, { "epoch": 2.5047801147227533, "grad_norm": 0.9767629504203796, "learning_rate": 2.6193333333333336e-05, "loss": 0.0756, "step": 3930 }, { "epoch": 2.5111536010197577, "grad_norm": 1.4233574867248535, "learning_rate": 2.6260000000000003e-05, "loss": 0.0894, "step": 3940 }, { "epoch": 2.517527087316762, "grad_norm": 1.6541444063186646, "learning_rate": 2.6326666666666666e-05, "loss": 0.0844, "step": 3950 }, { "epoch": 2.5239005736137665, "grad_norm": 1.1772117614746094, "learning_rate": 2.6393333333333337e-05, "loss": 0.0785, "step": 3960 }, { "epoch": 2.530274059910771, "grad_norm": 1.4708929061889648, "learning_rate": 2.646e-05, "loss": 0.0936, "step": 3970 }, { "epoch": 2.5366475462077758, "grad_norm": 0.8775594830513, "learning_rate": 2.6526666666666668e-05, "loss": 0.0816, "step": 3980 }, { "epoch": 2.54302103250478, "grad_norm": 0.8465993404388428, "learning_rate": 2.6593333333333335e-05, "loss": 0.071, "step": 3990 }, { "epoch": 2.5493945188017846, "grad_norm": 0.7804232239723206, "learning_rate": 2.6660000000000002e-05, "loss": 0.066, "step": 4000 }, { "epoch": 2.555768005098789, "grad_norm": 0.7311035990715027, "learning_rate": 2.6726666666666666e-05, "loss": 0.0809, "step": 4010 }, { "epoch": 2.5621414913957934, "grad_norm": 1.1837273836135864, "learning_rate": 2.6793333333333337e-05, "loss": 0.0825, "step": 4020 }, { "epoch": 2.568514977692798, "grad_norm": 2.203735589981079, "learning_rate": 2.686e-05, "loss": 0.0705, "step": 4030 }, { "epoch": 2.5748884639898026, "grad_norm": 0.9898331165313721, "learning_rate": 2.6926666666666667e-05, "loss": 0.0634, "step": 4040 }, { "epoch": 2.581261950286807, "grad_norm": 1.2964098453521729, "learning_rate": 2.6993333333333338e-05, "loss": 0.0735, "step": 4050 }, { "epoch": 2.5876354365838115, "grad_norm": 1.2006776332855225, "learning_rate": 2.7060000000000002e-05, "loss": 0.0716, "step": 4060 }, { "epoch": 2.594008922880816, "grad_norm": 1.2477272748947144, "learning_rate": 2.7126666666666666e-05, "loss": 0.0559, "step": 4070 }, { "epoch": 2.6003824091778203, "grad_norm": 0.6044740080833435, "learning_rate": 2.7193333333333336e-05, "loss": 0.0813, "step": 4080 }, { "epoch": 2.6067558954748247, "grad_norm": 0.7347907423973083, "learning_rate": 2.7260000000000003e-05, "loss": 0.0765, "step": 4090 }, { "epoch": 2.613129381771829, "grad_norm": 0.7471100091934204, "learning_rate": 2.7326666666666667e-05, "loss": 0.06, "step": 4100 }, { "epoch": 2.6195028680688335, "grad_norm": 1.0894196033477783, "learning_rate": 2.739333333333333e-05, "loss": 0.0755, "step": 4110 }, { "epoch": 2.625876354365838, "grad_norm": 0.8457331657409668, "learning_rate": 2.746e-05, "loss": 0.0706, "step": 4120 }, { "epoch": 2.6322498406628427, "grad_norm": 2.0056066513061523, "learning_rate": 2.752666666666667e-05, "loss": 0.0814, "step": 4130 }, { "epoch": 2.638623326959847, "grad_norm": 1.1211191415786743, "learning_rate": 2.7593333333333332e-05, "loss": 0.0741, "step": 4140 }, { "epoch": 2.6449968132568515, "grad_norm": 1.2720351219177246, "learning_rate": 2.7660000000000003e-05, "loss": 0.0698, "step": 4150 }, { "epoch": 2.651370299553856, "grad_norm": 1.2304461002349854, "learning_rate": 2.7726666666666667e-05, "loss": 0.0737, "step": 4160 }, { "epoch": 2.6577437858508604, "grad_norm": 1.1421581506729126, "learning_rate": 2.7793333333333334e-05, "loss": 0.0642, "step": 4170 }, { "epoch": 2.6641172721478648, "grad_norm": 1.1228699684143066, "learning_rate": 2.7860000000000004e-05, "loss": 0.0668, "step": 4180 }, { "epoch": 2.6704907584448696, "grad_norm": 0.8477457165718079, "learning_rate": 2.7926666666666668e-05, "loss": 0.0578, "step": 4190 }, { "epoch": 2.676864244741874, "grad_norm": 0.6787257790565491, "learning_rate": 2.7993333333333332e-05, "loss": 0.0798, "step": 4200 }, { "epoch": 2.6832377310388784, "grad_norm": 0.6480321884155273, "learning_rate": 2.8060000000000002e-05, "loss": 0.0713, "step": 4210 }, { "epoch": 2.689611217335883, "grad_norm": 0.8902840614318848, "learning_rate": 2.8126666666666666e-05, "loss": 0.0874, "step": 4220 }, { "epoch": 2.6959847036328872, "grad_norm": 0.9456228017807007, "learning_rate": 2.8193333333333333e-05, "loss": 0.0844, "step": 4230 }, { "epoch": 2.7023581899298916, "grad_norm": 0.6799092888832092, "learning_rate": 2.8260000000000004e-05, "loss": 0.0867, "step": 4240 }, { "epoch": 2.708731676226896, "grad_norm": 0.6477775573730469, "learning_rate": 2.8326666666666668e-05, "loss": 0.0819, "step": 4250 }, { "epoch": 2.7151051625239004, "grad_norm": 0.9661011099815369, "learning_rate": 2.839333333333333e-05, "loss": 0.0833, "step": 4260 }, { "epoch": 2.721478648820905, "grad_norm": 0.9668269753456116, "learning_rate": 2.8460000000000002e-05, "loss": 0.0801, "step": 4270 }, { "epoch": 2.7278521351179092, "grad_norm": 1.0115479230880737, "learning_rate": 2.852666666666667e-05, "loss": 0.0655, "step": 4280 }, { "epoch": 2.734225621414914, "grad_norm": 1.2372345924377441, "learning_rate": 2.8593333333333333e-05, "loss": 0.0733, "step": 4290 }, { "epoch": 2.7405991077119185, "grad_norm": 1.1563453674316406, "learning_rate": 2.8660000000000003e-05, "loss": 0.0682, "step": 4300 }, { "epoch": 2.746972594008923, "grad_norm": 1.0354273319244385, "learning_rate": 2.8726666666666667e-05, "loss": 0.0683, "step": 4310 }, { "epoch": 2.7533460803059273, "grad_norm": 0.8020623922348022, "learning_rate": 2.8793333333333334e-05, "loss": 0.0845, "step": 4320 }, { "epoch": 2.7597195666029317, "grad_norm": 0.9466817378997803, "learning_rate": 2.8860000000000005e-05, "loss": 0.082, "step": 4330 }, { "epoch": 2.766093052899936, "grad_norm": 1.0749906301498413, "learning_rate": 2.892666666666667e-05, "loss": 0.0688, "step": 4340 }, { "epoch": 2.772466539196941, "grad_norm": 0.8603000044822693, "learning_rate": 2.8993333333333332e-05, "loss": 0.0611, "step": 4350 }, { "epoch": 2.7788400254939454, "grad_norm": 1.47003972530365, "learning_rate": 2.9060000000000003e-05, "loss": 0.0677, "step": 4360 }, { "epoch": 2.78521351179095, "grad_norm": 0.6875565052032471, "learning_rate": 2.912666666666667e-05, "loss": 0.0693, "step": 4370 }, { "epoch": 2.791586998087954, "grad_norm": 1.0347020626068115, "learning_rate": 2.9193333333333334e-05, "loss": 0.0718, "step": 4380 }, { "epoch": 2.7979604843849586, "grad_norm": 0.49027177691459656, "learning_rate": 2.9260000000000004e-05, "loss": 0.0617, "step": 4390 }, { "epoch": 2.804333970681963, "grad_norm": 0.6546693444252014, "learning_rate": 2.9326666666666668e-05, "loss": 0.0777, "step": 4400 }, { "epoch": 2.8107074569789674, "grad_norm": 0.9329594373703003, "learning_rate": 2.9393333333333335e-05, "loss": 0.0737, "step": 4410 }, { "epoch": 2.817080943275972, "grad_norm": 1.2554725408554077, "learning_rate": 2.946e-05, "loss": 0.0938, "step": 4420 }, { "epoch": 2.823454429572976, "grad_norm": 0.8347150087356567, "learning_rate": 2.952666666666667e-05, "loss": 0.0726, "step": 4430 }, { "epoch": 2.8298279158699806, "grad_norm": 1.3713853359222412, "learning_rate": 2.9593333333333333e-05, "loss": 0.0747, "step": 4440 }, { "epoch": 2.8362014021669855, "grad_norm": 1.1022520065307617, "learning_rate": 2.9659999999999997e-05, "loss": 0.0737, "step": 4450 }, { "epoch": 2.84257488846399, "grad_norm": 0.5521169900894165, "learning_rate": 2.9726666666666668e-05, "loss": 0.0592, "step": 4460 }, { "epoch": 2.8489483747609943, "grad_norm": 1.3383432626724243, "learning_rate": 2.9793333333333335e-05, "loss": 0.0631, "step": 4470 }, { "epoch": 2.8553218610579987, "grad_norm": 0.5947378873825073, "learning_rate": 2.986e-05, "loss": 0.0632, "step": 4480 }, { "epoch": 2.861695347355003, "grad_norm": 0.8243908882141113, "learning_rate": 2.992666666666667e-05, "loss": 0.0661, "step": 4490 }, { "epoch": 2.8680688336520075, "grad_norm": 0.5694260597229004, "learning_rate": 2.9993333333333333e-05, "loss": 0.0749, "step": 4500 }, { "epoch": 2.8744423199490123, "grad_norm": 1.345126748085022, "learning_rate": 3.006e-05, "loss": 0.1045, "step": 4510 }, { "epoch": 2.8808158062460167, "grad_norm": 0.8360680937767029, "learning_rate": 3.012666666666667e-05, "loss": 0.0756, "step": 4520 }, { "epoch": 2.887189292543021, "grad_norm": 0.7815035581588745, "learning_rate": 3.0193333333333335e-05, "loss": 0.07, "step": 4530 }, { "epoch": 2.8935627788400256, "grad_norm": 0.9285995960235596, "learning_rate": 3.0259999999999998e-05, "loss": 0.0913, "step": 4540 }, { "epoch": 2.89993626513703, "grad_norm": 1.153631567955017, "learning_rate": 3.032666666666667e-05, "loss": 0.083, "step": 4550 }, { "epoch": 2.9063097514340344, "grad_norm": 0.7436297535896301, "learning_rate": 3.0393333333333336e-05, "loss": 0.0617, "step": 4560 }, { "epoch": 2.9126832377310388, "grad_norm": 0.9512943029403687, "learning_rate": 3.046e-05, "loss": 0.0872, "step": 4570 }, { "epoch": 2.919056724028043, "grad_norm": 1.808171033859253, "learning_rate": 3.052666666666667e-05, "loss": 0.0686, "step": 4580 }, { "epoch": 2.9254302103250476, "grad_norm": 0.7790931463241577, "learning_rate": 3.059333333333334e-05, "loss": 0.0871, "step": 4590 }, { "epoch": 2.9318036966220524, "grad_norm": 0.6268696188926697, "learning_rate": 3.066e-05, "loss": 0.0682, "step": 4600 }, { "epoch": 2.938177182919057, "grad_norm": 0.8465157747268677, "learning_rate": 3.072666666666667e-05, "loss": 0.0772, "step": 4610 }, { "epoch": 2.9445506692160612, "grad_norm": 0.7893638014793396, "learning_rate": 3.0793333333333336e-05, "loss": 0.0576, "step": 4620 }, { "epoch": 2.9509241555130656, "grad_norm": 1.395041823387146, "learning_rate": 3.086e-05, "loss": 0.0766, "step": 4630 }, { "epoch": 2.95729764181007, "grad_norm": 0.8314698338508606, "learning_rate": 3.092666666666667e-05, "loss": 0.0791, "step": 4640 }, { "epoch": 2.9636711281070744, "grad_norm": 0.7972301244735718, "learning_rate": 3.0993333333333334e-05, "loss": 0.0679, "step": 4650 }, { "epoch": 2.9700446144040793, "grad_norm": 1.237327218055725, "learning_rate": 3.106e-05, "loss": 0.093, "step": 4660 }, { "epoch": 2.9764181007010837, "grad_norm": 0.9242610931396484, "learning_rate": 3.112666666666667e-05, "loss": 0.066, "step": 4670 }, { "epoch": 2.982791586998088, "grad_norm": 1.5865044593811035, "learning_rate": 3.119333333333334e-05, "loss": 0.0794, "step": 4680 }, { "epoch": 2.9891650732950925, "grad_norm": 1.0787328481674194, "learning_rate": 3.126e-05, "loss": 0.0823, "step": 4690 }, { "epoch": 2.995538559592097, "grad_norm": 1.1212153434753418, "learning_rate": 3.132666666666667e-05, "loss": 0.0738, "step": 4700 }, { "epoch": 3.0019120458891013, "grad_norm": 0.6594878435134888, "learning_rate": 3.1393333333333337e-05, "loss": 0.0717, "step": 4710 }, { "epoch": 3.0082855321861057, "grad_norm": 1.0209071636199951, "learning_rate": 3.146e-05, "loss": 0.0613, "step": 4720 }, { "epoch": 3.01465901848311, "grad_norm": 1.0822089910507202, "learning_rate": 3.1526666666666664e-05, "loss": 0.083, "step": 4730 }, { "epoch": 3.0210325047801145, "grad_norm": 0.9222205281257629, "learning_rate": 3.1593333333333335e-05, "loss": 0.0692, "step": 4740 }, { "epoch": 3.0274059910771194, "grad_norm": 0.7965155243873596, "learning_rate": 3.166e-05, "loss": 0.0638, "step": 4750 }, { "epoch": 3.033779477374124, "grad_norm": 0.8105457425117493, "learning_rate": 3.172666666666667e-05, "loss": 0.0752, "step": 4760 }, { "epoch": 3.040152963671128, "grad_norm": 0.623116135597229, "learning_rate": 3.179333333333333e-05, "loss": 0.0722, "step": 4770 }, { "epoch": 3.0465264499681326, "grad_norm": 0.9211954474449158, "learning_rate": 3.186e-05, "loss": 0.0743, "step": 4780 }, { "epoch": 3.052899936265137, "grad_norm": 1.0942955017089844, "learning_rate": 3.192666666666667e-05, "loss": 0.0619, "step": 4790 }, { "epoch": 3.0592734225621414, "grad_norm": 1.1815531253814697, "learning_rate": 3.199333333333334e-05, "loss": 0.0725, "step": 4800 }, { "epoch": 3.065646908859146, "grad_norm": 1.0571568012237549, "learning_rate": 3.206e-05, "loss": 0.0707, "step": 4810 }, { "epoch": 3.07202039515615, "grad_norm": 1.062965750694275, "learning_rate": 3.2126666666666665e-05, "loss": 0.0796, "step": 4820 }, { "epoch": 3.078393881453155, "grad_norm": 1.058998942375183, "learning_rate": 3.2193333333333336e-05, "loss": 0.0742, "step": 4830 }, { "epoch": 3.0847673677501595, "grad_norm": 0.47686275839805603, "learning_rate": 3.226e-05, "loss": 0.0592, "step": 4840 }, { "epoch": 3.091140854047164, "grad_norm": 0.9305811524391174, "learning_rate": 3.232666666666666e-05, "loss": 0.062, "step": 4850 }, { "epoch": 3.0975143403441683, "grad_norm": 0.6117062568664551, "learning_rate": 3.2393333333333334e-05, "loss": 0.0625, "step": 4860 }, { "epoch": 3.1038878266411727, "grad_norm": 0.8363755941390991, "learning_rate": 3.2460000000000004e-05, "loss": 0.054, "step": 4870 }, { "epoch": 3.110261312938177, "grad_norm": 2.193182945251465, "learning_rate": 3.252666666666667e-05, "loss": 0.0827, "step": 4880 }, { "epoch": 3.1166347992351815, "grad_norm": 0.8609194755554199, "learning_rate": 3.259333333333334e-05, "loss": 0.0881, "step": 4890 }, { "epoch": 3.123008285532186, "grad_norm": 0.7422274947166443, "learning_rate": 3.266e-05, "loss": 0.0608, "step": 4900 }, { "epoch": 3.1293817718291908, "grad_norm": 1.7015436887741089, "learning_rate": 3.2726666666666666e-05, "loss": 0.07, "step": 4910 }, { "epoch": 3.135755258126195, "grad_norm": 0.9272896647453308, "learning_rate": 3.279333333333334e-05, "loss": 0.0746, "step": 4920 }, { "epoch": 3.1421287444231996, "grad_norm": 1.6632956266403198, "learning_rate": 3.286e-05, "loss": 0.0769, "step": 4930 }, { "epoch": 3.148502230720204, "grad_norm": 0.9707281589508057, "learning_rate": 3.2926666666666664e-05, "loss": 0.0664, "step": 4940 }, { "epoch": 3.1548757170172084, "grad_norm": 0.7668961882591248, "learning_rate": 3.2993333333333335e-05, "loss": 0.066, "step": 4950 }, { "epoch": 3.1612492033142128, "grad_norm": 1.1414968967437744, "learning_rate": 3.3060000000000005e-05, "loss": 0.0668, "step": 4960 }, { "epoch": 3.167622689611217, "grad_norm": 0.6386382579803467, "learning_rate": 3.312666666666667e-05, "loss": 0.0731, "step": 4970 }, { "epoch": 3.173996175908222, "grad_norm": 0.937657356262207, "learning_rate": 3.319333333333334e-05, "loss": 0.0612, "step": 4980 }, { "epoch": 3.1803696622052264, "grad_norm": 1.0628135204315186, "learning_rate": 3.3260000000000003e-05, "loss": 0.0671, "step": 4990 }, { "epoch": 3.186743148502231, "grad_norm": 1.070125937461853, "learning_rate": 3.332666666666667e-05, "loss": 0.0723, "step": 5000 }, { "epoch": 3.1931166347992352, "grad_norm": 1.0119048357009888, "learning_rate": 3.339333333333334e-05, "loss": 0.0728, "step": 5010 }, { "epoch": 3.1994901210962396, "grad_norm": 0.9777017831802368, "learning_rate": 3.346e-05, "loss": 0.0652, "step": 5020 }, { "epoch": 3.205863607393244, "grad_norm": 1.3063814640045166, "learning_rate": 3.3526666666666665e-05, "loss": 0.0728, "step": 5030 }, { "epoch": 3.2122370936902485, "grad_norm": 0.6333829164505005, "learning_rate": 3.359333333333333e-05, "loss": 0.0548, "step": 5040 }, { "epoch": 3.218610579987253, "grad_norm": 0.7940689325332642, "learning_rate": 3.366e-05, "loss": 0.0869, "step": 5050 }, { "epoch": 3.2249840662842573, "grad_norm": 0.7025421261787415, "learning_rate": 3.372666666666667e-05, "loss": 0.053, "step": 5060 }, { "epoch": 3.231357552581262, "grad_norm": 1.060881495475769, "learning_rate": 3.3793333333333334e-05, "loss": 0.0776, "step": 5070 }, { "epoch": 3.2377310388782665, "grad_norm": 0.6675065159797668, "learning_rate": 3.3860000000000004e-05, "loss": 0.0808, "step": 5080 }, { "epoch": 3.244104525175271, "grad_norm": 0.788928747177124, "learning_rate": 3.392666666666667e-05, "loss": 0.0651, "step": 5090 }, { "epoch": 3.2504780114722753, "grad_norm": 0.9996426105499268, "learning_rate": 3.399333333333333e-05, "loss": 0.0649, "step": 5100 }, { "epoch": 3.2568514977692797, "grad_norm": 0.7572866082191467, "learning_rate": 3.406e-05, "loss": 0.0704, "step": 5110 }, { "epoch": 3.263224984066284, "grad_norm": 1.0348575115203857, "learning_rate": 3.4126666666666666e-05, "loss": 0.0815, "step": 5120 }, { "epoch": 3.2695984703632885, "grad_norm": 1.1521161794662476, "learning_rate": 3.419333333333333e-05, "loss": 0.075, "step": 5130 }, { "epoch": 3.2759719566602934, "grad_norm": 0.6772708892822266, "learning_rate": 3.426e-05, "loss": 0.0643, "step": 5140 }, { "epoch": 3.282345442957298, "grad_norm": 0.8704012632369995, "learning_rate": 3.432666666666667e-05, "loss": 0.0706, "step": 5150 }, { "epoch": 3.288718929254302, "grad_norm": 0.8287102580070496, "learning_rate": 3.4393333333333335e-05, "loss": 0.0844, "step": 5160 }, { "epoch": 3.2950924155513066, "grad_norm": 0.8892394304275513, "learning_rate": 3.4460000000000005e-05, "loss": 0.0694, "step": 5170 }, { "epoch": 3.301465901848311, "grad_norm": 1.0021601915359497, "learning_rate": 3.452666666666667e-05, "loss": 0.0617, "step": 5180 }, { "epoch": 3.3078393881453154, "grad_norm": 1.0118314027786255, "learning_rate": 3.459333333333333e-05, "loss": 0.0706, "step": 5190 }, { "epoch": 3.31421287444232, "grad_norm": 0.93133544921875, "learning_rate": 3.4660000000000004e-05, "loss": 0.0634, "step": 5200 }, { "epoch": 3.3205863607393242, "grad_norm": 0.9466410279273987, "learning_rate": 3.472666666666667e-05, "loss": 0.068, "step": 5210 }, { "epoch": 3.3269598470363286, "grad_norm": 0.8412697911262512, "learning_rate": 3.479333333333333e-05, "loss": 0.0636, "step": 5220 }, { "epoch": 3.3333333333333335, "grad_norm": 0.7940225005149841, "learning_rate": 3.486e-05, "loss": 0.066, "step": 5230 }, { "epoch": 3.339706819630338, "grad_norm": 0.9359369874000549, "learning_rate": 3.4926666666666665e-05, "loss": 0.0719, "step": 5240 }, { "epoch": 3.3460803059273423, "grad_norm": 1.3645020723342896, "learning_rate": 3.4993333333333336e-05, "loss": 0.0707, "step": 5250 }, { "epoch": 3.3524537922243467, "grad_norm": 1.2952816486358643, "learning_rate": 3.5060000000000007e-05, "loss": 0.0685, "step": 5260 }, { "epoch": 3.358827278521351, "grad_norm": 0.7421273589134216, "learning_rate": 3.512666666666667e-05, "loss": 0.0744, "step": 5270 }, { "epoch": 3.3652007648183555, "grad_norm": 0.7207891941070557, "learning_rate": 3.5193333333333334e-05, "loss": 0.0781, "step": 5280 }, { "epoch": 3.37157425111536, "grad_norm": 0.9531093239784241, "learning_rate": 3.5260000000000005e-05, "loss": 0.0676, "step": 5290 }, { "epoch": 3.3779477374123648, "grad_norm": 0.5877613425254822, "learning_rate": 3.532666666666667e-05, "loss": 0.0574, "step": 5300 }, { "epoch": 3.384321223709369, "grad_norm": 0.6459888815879822, "learning_rate": 3.539333333333333e-05, "loss": 0.0681, "step": 5310 }, { "epoch": 3.3906947100063736, "grad_norm": 1.011889100074768, "learning_rate": 3.546e-05, "loss": 0.0724, "step": 5320 }, { "epoch": 3.397068196303378, "grad_norm": 0.5205362439155579, "learning_rate": 3.5526666666666666e-05, "loss": 0.0797, "step": 5330 }, { "epoch": 3.4034416826003824, "grad_norm": 0.8124321103096008, "learning_rate": 3.559333333333334e-05, "loss": 0.0727, "step": 5340 }, { "epoch": 3.409815168897387, "grad_norm": 0.7917166352272034, "learning_rate": 3.566e-05, "loss": 0.0701, "step": 5350 }, { "epoch": 3.416188655194391, "grad_norm": 0.8654747009277344, "learning_rate": 3.572666666666667e-05, "loss": 0.0672, "step": 5360 }, { "epoch": 3.4225621414913956, "grad_norm": 1.0261168479919434, "learning_rate": 3.5793333333333335e-05, "loss": 0.0593, "step": 5370 }, { "epoch": 3.4289356277884004, "grad_norm": 0.7219008207321167, "learning_rate": 3.586e-05, "loss": 0.068, "step": 5380 }, { "epoch": 3.435309114085405, "grad_norm": 0.8611841201782227, "learning_rate": 3.592666666666667e-05, "loss": 0.0607, "step": 5390 }, { "epoch": 3.4416826003824093, "grad_norm": 0.7135807275772095, "learning_rate": 3.599333333333333e-05, "loss": 0.0698, "step": 5400 }, { "epoch": 3.4480560866794137, "grad_norm": 0.9602314829826355, "learning_rate": 3.606e-05, "loss": 0.0678, "step": 5410 }, { "epoch": 3.454429572976418, "grad_norm": 1.1248433589935303, "learning_rate": 3.612666666666667e-05, "loss": 0.0614, "step": 5420 }, { "epoch": 3.4608030592734225, "grad_norm": 0.6453667879104614, "learning_rate": 3.619333333333333e-05, "loss": 0.0599, "step": 5430 }, { "epoch": 3.467176545570427, "grad_norm": 1.391933798789978, "learning_rate": 3.626e-05, "loss": 0.0909, "step": 5440 }, { "epoch": 3.4735500318674317, "grad_norm": 0.8513909578323364, "learning_rate": 3.632666666666667e-05, "loss": 0.0529, "step": 5450 }, { "epoch": 3.479923518164436, "grad_norm": 1.0372978448867798, "learning_rate": 3.6393333333333336e-05, "loss": 0.0663, "step": 5460 }, { "epoch": 3.4862970044614405, "grad_norm": 0.6336553692817688, "learning_rate": 3.646e-05, "loss": 0.0559, "step": 5470 }, { "epoch": 3.492670490758445, "grad_norm": 0.7353127598762512, "learning_rate": 3.652666666666667e-05, "loss": 0.0592, "step": 5480 }, { "epoch": 3.4990439770554493, "grad_norm": 0.48831209540367126, "learning_rate": 3.6593333333333334e-05, "loss": 0.0602, "step": 5490 }, { "epoch": 3.5054174633524537, "grad_norm": 0.9713586568832397, "learning_rate": 3.666e-05, "loss": 0.0717, "step": 5500 }, { "epoch": 3.511790949649458, "grad_norm": 0.9127499461174011, "learning_rate": 3.672666666666667e-05, "loss": 0.0747, "step": 5510 }, { "epoch": 3.5181644359464626, "grad_norm": 0.7769306302070618, "learning_rate": 3.679333333333333e-05, "loss": 0.0631, "step": 5520 }, { "epoch": 3.524537922243467, "grad_norm": 0.7864208817481995, "learning_rate": 3.686e-05, "loss": 0.059, "step": 5530 }, { "epoch": 3.5309114085404714, "grad_norm": 0.9494785666465759, "learning_rate": 3.6926666666666673e-05, "loss": 0.0561, "step": 5540 }, { "epoch": 3.537284894837476, "grad_norm": 0.6086204648017883, "learning_rate": 3.699333333333334e-05, "loss": 0.0484, "step": 5550 }, { "epoch": 3.5436583811344806, "grad_norm": 0.9108319282531738, "learning_rate": 3.706e-05, "loss": 0.0739, "step": 5560 }, { "epoch": 3.550031867431485, "grad_norm": 0.8275277018547058, "learning_rate": 3.712666666666667e-05, "loss": 0.0649, "step": 5570 }, { "epoch": 3.5564053537284894, "grad_norm": 0.8741169571876526, "learning_rate": 3.7193333333333335e-05, "loss": 0.0711, "step": 5580 }, { "epoch": 3.562778840025494, "grad_norm": 0.5867934226989746, "learning_rate": 3.726e-05, "loss": 0.0666, "step": 5590 }, { "epoch": 3.5691523263224987, "grad_norm": 0.9893175363540649, "learning_rate": 3.732666666666667e-05, "loss": 0.0707, "step": 5600 }, { "epoch": 3.575525812619503, "grad_norm": 1.4422320127487183, "learning_rate": 3.739333333333333e-05, "loss": 0.0654, "step": 5610 }, { "epoch": 3.5818992989165075, "grad_norm": 1.280185580253601, "learning_rate": 3.7460000000000004e-05, "loss": 0.0674, "step": 5620 }, { "epoch": 3.588272785213512, "grad_norm": 1.0105539560317993, "learning_rate": 3.7526666666666674e-05, "loss": 0.0645, "step": 5630 }, { "epoch": 3.5946462715105163, "grad_norm": 0.8940385580062866, "learning_rate": 3.759333333333334e-05, "loss": 0.0635, "step": 5640 }, { "epoch": 3.6010197578075207, "grad_norm": 1.0427013635635376, "learning_rate": 3.766e-05, "loss": 0.0773, "step": 5650 }, { "epoch": 3.607393244104525, "grad_norm": 0.7193406224250793, "learning_rate": 3.7726666666666666e-05, "loss": 0.0583, "step": 5660 }, { "epoch": 3.6137667304015295, "grad_norm": 0.7543081641197205, "learning_rate": 3.7793333333333336e-05, "loss": 0.0707, "step": 5670 }, { "epoch": 3.620140216698534, "grad_norm": 0.8943052291870117, "learning_rate": 3.786e-05, "loss": 0.0679, "step": 5680 }, { "epoch": 3.6265137029955383, "grad_norm": 0.6760386228561401, "learning_rate": 3.7926666666666664e-05, "loss": 0.0647, "step": 5690 }, { "epoch": 3.632887189292543, "grad_norm": 0.9791440367698669, "learning_rate": 3.7993333333333334e-05, "loss": 0.0727, "step": 5700 }, { "epoch": 3.6392606755895476, "grad_norm": 0.9366337060928345, "learning_rate": 3.806e-05, "loss": 0.08, "step": 5710 }, { "epoch": 3.645634161886552, "grad_norm": 0.9262896180152893, "learning_rate": 3.812666666666667e-05, "loss": 0.072, "step": 5720 }, { "epoch": 3.6520076481835564, "grad_norm": 1.2537277936935425, "learning_rate": 3.819333333333334e-05, "loss": 0.0691, "step": 5730 }, { "epoch": 3.658381134480561, "grad_norm": 1.117384672164917, "learning_rate": 3.826e-05, "loss": 0.0754, "step": 5740 }, { "epoch": 3.664754620777565, "grad_norm": 0.6348002552986145, "learning_rate": 3.832666666666667e-05, "loss": 0.0603, "step": 5750 }, { "epoch": 3.67112810707457, "grad_norm": 0.7391278743743896, "learning_rate": 3.839333333333334e-05, "loss": 0.0656, "step": 5760 }, { "epoch": 3.6775015933715745, "grad_norm": 0.7247368097305298, "learning_rate": 3.846e-05, "loss": 0.064, "step": 5770 }, { "epoch": 3.683875079668579, "grad_norm": 0.6577849388122559, "learning_rate": 3.8526666666666665e-05, "loss": 0.0842, "step": 5780 }, { "epoch": 3.6902485659655833, "grad_norm": 0.8825833797454834, "learning_rate": 3.8593333333333335e-05, "loss": 0.06, "step": 5790 }, { "epoch": 3.6966220522625877, "grad_norm": 0.5579012632369995, "learning_rate": 3.866e-05, "loss": 0.0787, "step": 5800 }, { "epoch": 3.702995538559592, "grad_norm": 0.6694490909576416, "learning_rate": 3.872666666666667e-05, "loss": 0.0594, "step": 5810 }, { "epoch": 3.7093690248565965, "grad_norm": 0.8608837127685547, "learning_rate": 3.879333333333334e-05, "loss": 0.0632, "step": 5820 }, { "epoch": 3.715742511153601, "grad_norm": 0.8467159271240234, "learning_rate": 3.8860000000000004e-05, "loss": 0.0635, "step": 5830 }, { "epoch": 3.7221159974506053, "grad_norm": 0.8490839004516602, "learning_rate": 3.892666666666667e-05, "loss": 0.058, "step": 5840 }, { "epoch": 3.7284894837476097, "grad_norm": 1.2588324546813965, "learning_rate": 3.899333333333334e-05, "loss": 0.0562, "step": 5850 }, { "epoch": 3.7348629700446145, "grad_norm": 0.9957313537597656, "learning_rate": 3.906e-05, "loss": 0.0723, "step": 5860 }, { "epoch": 3.741236456341619, "grad_norm": 1.0650949478149414, "learning_rate": 3.9126666666666666e-05, "loss": 0.0862, "step": 5870 }, { "epoch": 3.7476099426386233, "grad_norm": 0.5892736911773682, "learning_rate": 3.9193333333333336e-05, "loss": 0.0807, "step": 5880 }, { "epoch": 3.7539834289356278, "grad_norm": 0.6370896697044373, "learning_rate": 3.926e-05, "loss": 0.0636, "step": 5890 }, { "epoch": 3.760356915232632, "grad_norm": 0.5198182463645935, "learning_rate": 3.9326666666666664e-05, "loss": 0.0796, "step": 5900 }, { "epoch": 3.7667304015296366, "grad_norm": 0.8571606278419495, "learning_rate": 3.9393333333333335e-05, "loss": 0.0782, "step": 5910 }, { "epoch": 3.7731038878266414, "grad_norm": 0.7116801738739014, "learning_rate": 3.9460000000000005e-05, "loss": 0.052, "step": 5920 }, { "epoch": 3.779477374123646, "grad_norm": 0.6442444920539856, "learning_rate": 3.952666666666667e-05, "loss": 0.0588, "step": 5930 }, { "epoch": 3.78585086042065, "grad_norm": 1.0575484037399292, "learning_rate": 3.959333333333334e-05, "loss": 0.0758, "step": 5940 }, { "epoch": 3.7922243467176546, "grad_norm": 0.8098411560058594, "learning_rate": 3.966e-05, "loss": 0.0496, "step": 5950 }, { "epoch": 3.798597833014659, "grad_norm": 0.7580195069313049, "learning_rate": 3.972666666666667e-05, "loss": 0.0574, "step": 5960 }, { "epoch": 3.8049713193116634, "grad_norm": 0.9306786060333252, "learning_rate": 3.979333333333333e-05, "loss": 0.0608, "step": 5970 }, { "epoch": 3.811344805608668, "grad_norm": 1.4234120845794678, "learning_rate": 3.986e-05, "loss": 0.0893, "step": 5980 }, { "epoch": 3.8177182919056722, "grad_norm": 0.7272818684577942, "learning_rate": 3.9926666666666665e-05, "loss": 0.0803, "step": 5990 }, { "epoch": 3.8240917782026767, "grad_norm": 0.7398123145103455, "learning_rate": 3.9993333333333336e-05, "loss": 0.0733, "step": 6000 }, { "epoch": 3.830465264499681, "grad_norm": 0.8781506419181824, "learning_rate": 4.0060000000000006e-05, "loss": 0.0715, "step": 6010 }, { "epoch": 3.836838750796686, "grad_norm": 0.8230180144309998, "learning_rate": 4.012666666666667e-05, "loss": 0.0831, "step": 6020 }, { "epoch": 3.8432122370936903, "grad_norm": 1.5217406749725342, "learning_rate": 4.0193333333333334e-05, "loss": 0.0818, "step": 6030 }, { "epoch": 3.8495857233906947, "grad_norm": 1.135979175567627, "learning_rate": 4.0260000000000004e-05, "loss": 0.0771, "step": 6040 }, { "epoch": 3.855959209687699, "grad_norm": 1.0574519634246826, "learning_rate": 4.032666666666667e-05, "loss": 0.0725, "step": 6050 }, { "epoch": 3.8623326959847035, "grad_norm": 1.0695313215255737, "learning_rate": 4.039333333333333e-05, "loss": 0.0555, "step": 6060 }, { "epoch": 3.868706182281708, "grad_norm": 1.1515299081802368, "learning_rate": 4.046e-05, "loss": 0.0753, "step": 6070 }, { "epoch": 3.8750796685787128, "grad_norm": 0.9582716822624207, "learning_rate": 4.0526666666666666e-05, "loss": 0.0961, "step": 6080 }, { "epoch": 3.881453154875717, "grad_norm": 0.7768780589103699, "learning_rate": 4.0593333333333337e-05, "loss": 0.0693, "step": 6090 }, { "epoch": 3.8878266411727216, "grad_norm": 0.7570106983184814, "learning_rate": 4.066e-05, "loss": 0.0826, "step": 6100 }, { "epoch": 3.894200127469726, "grad_norm": 0.531088650226593, "learning_rate": 4.072666666666667e-05, "loss": 0.0521, "step": 6110 }, { "epoch": 3.9005736137667304, "grad_norm": 0.9833720922470093, "learning_rate": 4.0793333333333335e-05, "loss": 0.0692, "step": 6120 }, { "epoch": 3.906947100063735, "grad_norm": 0.6525686383247375, "learning_rate": 4.0860000000000005e-05, "loss": 0.0628, "step": 6130 }, { "epoch": 3.913320586360739, "grad_norm": 1.07793128490448, "learning_rate": 4.092666666666667e-05, "loss": 0.0693, "step": 6140 }, { "epoch": 3.9196940726577436, "grad_norm": 0.8308563828468323, "learning_rate": 4.099333333333333e-05, "loss": 0.0807, "step": 6150 }, { "epoch": 3.926067558954748, "grad_norm": 0.7288112640380859, "learning_rate": 4.106e-05, "loss": 0.0723, "step": 6160 }, { "epoch": 3.932441045251753, "grad_norm": 0.8844555616378784, "learning_rate": 4.112666666666667e-05, "loss": 0.0737, "step": 6170 }, { "epoch": 3.9388145315487573, "grad_norm": 0.6023966073989868, "learning_rate": 4.119333333333333e-05, "loss": 0.0572, "step": 6180 }, { "epoch": 3.9451880178457617, "grad_norm": 0.6090594530105591, "learning_rate": 4.126e-05, "loss": 0.0747, "step": 6190 }, { "epoch": 3.951561504142766, "grad_norm": 0.8953942656517029, "learning_rate": 4.132666666666667e-05, "loss": 0.0598, "step": 6200 }, { "epoch": 3.9579349904397705, "grad_norm": 0.8100000023841858, "learning_rate": 4.1393333333333336e-05, "loss": 0.0667, "step": 6210 }, { "epoch": 3.964308476736775, "grad_norm": 0.7682720422744751, "learning_rate": 4.1460000000000006e-05, "loss": 0.067, "step": 6220 }, { "epoch": 3.9706819630337797, "grad_norm": 1.0075936317443848, "learning_rate": 4.152666666666667e-05, "loss": 0.0737, "step": 6230 }, { "epoch": 3.977055449330784, "grad_norm": 0.7655991911888123, "learning_rate": 4.1593333333333334e-05, "loss": 0.0686, "step": 6240 }, { "epoch": 3.9834289356277885, "grad_norm": 1.4795335531234741, "learning_rate": 4.1660000000000004e-05, "loss": 0.0937, "step": 6250 }, { "epoch": 3.989802421924793, "grad_norm": 0.5925967693328857, "learning_rate": 4.172666666666667e-05, "loss": 0.0531, "step": 6260 }, { "epoch": 3.9961759082217974, "grad_norm": 1.1187046766281128, "learning_rate": 4.179333333333333e-05, "loss": 0.0732, "step": 6270 }, { "epoch": 4.002549394518802, "grad_norm": 0.9150164127349854, "learning_rate": 4.186e-05, "loss": 0.0666, "step": 6280 }, { "epoch": 4.008922880815806, "grad_norm": 0.49149367213249207, "learning_rate": 4.192666666666667e-05, "loss": 0.0548, "step": 6290 }, { "epoch": 4.015296367112811, "grad_norm": 0.6512823104858398, "learning_rate": 4.199333333333334e-05, "loss": 0.0641, "step": 6300 }, { "epoch": 4.021669853409815, "grad_norm": 0.7342813014984131, "learning_rate": 4.206e-05, "loss": 0.0582, "step": 6310 }, { "epoch": 4.028043339706819, "grad_norm": 0.8499735593795776, "learning_rate": 4.212666666666667e-05, "loss": 0.0668, "step": 6320 }, { "epoch": 4.034416826003824, "grad_norm": 1.5677454471588135, "learning_rate": 4.2193333333333335e-05, "loss": 0.0775, "step": 6330 }, { "epoch": 4.040790312300828, "grad_norm": 0.5655584931373596, "learning_rate": 4.226e-05, "loss": 0.0737, "step": 6340 }, { "epoch": 4.047163798597833, "grad_norm": 0.5354357957839966, "learning_rate": 4.232666666666667e-05, "loss": 0.0576, "step": 6350 }, { "epoch": 4.053537284894838, "grad_norm": 0.6646163463592529, "learning_rate": 4.239333333333333e-05, "loss": 0.0713, "step": 6360 }, { "epoch": 4.059910771191842, "grad_norm": 0.8326684236526489, "learning_rate": 4.246e-05, "loss": 0.0512, "step": 6370 }, { "epoch": 4.066284257488847, "grad_norm": 0.6551247239112854, "learning_rate": 4.252666666666667e-05, "loss": 0.0547, "step": 6380 }, { "epoch": 4.072657743785851, "grad_norm": 0.8713095784187317, "learning_rate": 4.259333333333334e-05, "loss": 0.0754, "step": 6390 }, { "epoch": 4.0790312300828555, "grad_norm": 0.8737226724624634, "learning_rate": 4.266e-05, "loss": 0.0649, "step": 6400 }, { "epoch": 4.08540471637986, "grad_norm": 1.3375325202941895, "learning_rate": 4.272666666666667e-05, "loss": 0.0572, "step": 6410 }, { "epoch": 4.091778202676864, "grad_norm": 1.0645341873168945, "learning_rate": 4.2793333333333336e-05, "loss": 0.0579, "step": 6420 }, { "epoch": 4.098151688973869, "grad_norm": 0.6480242013931274, "learning_rate": 4.286e-05, "loss": 0.0657, "step": 6430 }, { "epoch": 4.104525175270873, "grad_norm": 0.8946130275726318, "learning_rate": 4.292666666666667e-05, "loss": 0.0584, "step": 6440 }, { "epoch": 4.1108986615678775, "grad_norm": 0.82254558801651, "learning_rate": 4.2993333333333334e-05, "loss": 0.0621, "step": 6450 }, { "epoch": 4.117272147864882, "grad_norm": 0.8301029205322266, "learning_rate": 4.306e-05, "loss": 0.0519, "step": 6460 }, { "epoch": 4.123645634161886, "grad_norm": 0.9191573262214661, "learning_rate": 4.312666666666667e-05, "loss": 0.0648, "step": 6470 }, { "epoch": 4.130019120458891, "grad_norm": 0.6837590336799622, "learning_rate": 4.319333333333334e-05, "loss": 0.0578, "step": 6480 }, { "epoch": 4.136392606755895, "grad_norm": 0.8105356097221375, "learning_rate": 4.326e-05, "loss": 0.0776, "step": 6490 }, { "epoch": 4.1427660930529, "grad_norm": 0.8152103424072266, "learning_rate": 4.332666666666667e-05, "loss": 0.0684, "step": 6500 }, { "epoch": 4.149139579349905, "grad_norm": 0.9396456480026245, "learning_rate": 4.339333333333334e-05, "loss": 0.0523, "step": 6510 }, { "epoch": 4.155513065646909, "grad_norm": 0.8193104863166809, "learning_rate": 4.346e-05, "loss": 0.0574, "step": 6520 }, { "epoch": 4.161886551943914, "grad_norm": 0.7371982336044312, "learning_rate": 4.352666666666667e-05, "loss": 0.0645, "step": 6530 }, { "epoch": 4.168260038240918, "grad_norm": 1.027175784111023, "learning_rate": 4.3593333333333335e-05, "loss": 0.0729, "step": 6540 }, { "epoch": 4.1746335245379225, "grad_norm": 0.6634669303894043, "learning_rate": 4.366e-05, "loss": 0.0523, "step": 6550 }, { "epoch": 4.181007010834927, "grad_norm": 0.7883057594299316, "learning_rate": 4.372666666666667e-05, "loss": 0.0818, "step": 6560 }, { "epoch": 4.187380497131931, "grad_norm": 1.2026309967041016, "learning_rate": 4.379333333333333e-05, "loss": 0.0756, "step": 6570 }, { "epoch": 4.193753983428936, "grad_norm": 1.1520119905471802, "learning_rate": 4.3860000000000004e-05, "loss": 0.0725, "step": 6580 }, { "epoch": 4.20012746972594, "grad_norm": 1.096225380897522, "learning_rate": 4.3926666666666674e-05, "loss": 0.0737, "step": 6590 }, { "epoch": 4.2065009560229445, "grad_norm": 1.2484513521194458, "learning_rate": 4.399333333333334e-05, "loss": 0.0759, "step": 6600 }, { "epoch": 4.212874442319949, "grad_norm": 0.33926501870155334, "learning_rate": 4.406e-05, "loss": 0.0621, "step": 6610 }, { "epoch": 4.219247928616953, "grad_norm": 0.8289417624473572, "learning_rate": 4.4126666666666665e-05, "loss": 0.0551, "step": 6620 }, { "epoch": 4.225621414913958, "grad_norm": 1.2392948865890503, "learning_rate": 4.4193333333333336e-05, "loss": 0.067, "step": 6630 }, { "epoch": 4.231994901210962, "grad_norm": 0.6331199407577515, "learning_rate": 4.426e-05, "loss": 0.0573, "step": 6640 }, { "epoch": 4.2383683875079665, "grad_norm": 0.7887037396430969, "learning_rate": 4.4326666666666664e-05, "loss": 0.0767, "step": 6650 }, { "epoch": 4.244741873804971, "grad_norm": 0.6712101101875305, "learning_rate": 4.4393333333333334e-05, "loss": 0.0501, "step": 6660 }, { "epoch": 4.251115360101975, "grad_norm": 0.5873392820358276, "learning_rate": 4.4460000000000005e-05, "loss": 0.0629, "step": 6670 }, { "epoch": 4.257488846398981, "grad_norm": 0.8280028104782104, "learning_rate": 4.452666666666667e-05, "loss": 0.0693, "step": 6680 }, { "epoch": 4.263862332695985, "grad_norm": 1.1430957317352295, "learning_rate": 4.459333333333334e-05, "loss": 0.055, "step": 6690 }, { "epoch": 4.270235818992989, "grad_norm": 0.5776487588882446, "learning_rate": 4.466e-05, "loss": 0.0741, "step": 6700 }, { "epoch": 4.276609305289994, "grad_norm": 0.657620906829834, "learning_rate": 4.4726666666666666e-05, "loss": 0.0602, "step": 6710 }, { "epoch": 4.282982791586998, "grad_norm": 0.7546238899230957, "learning_rate": 4.479333333333334e-05, "loss": 0.0772, "step": 6720 }, { "epoch": 4.289356277884003, "grad_norm": 0.7995972633361816, "learning_rate": 4.486e-05, "loss": 0.078, "step": 6730 }, { "epoch": 4.295729764181007, "grad_norm": 0.6277379989624023, "learning_rate": 4.4926666666666665e-05, "loss": 0.08, "step": 6740 }, { "epoch": 4.3021032504780115, "grad_norm": 0.7915437817573547, "learning_rate": 4.4993333333333335e-05, "loss": 0.0641, "step": 6750 }, { "epoch": 4.308476736775016, "grad_norm": 0.7979368567466736, "learning_rate": 4.506e-05, "loss": 0.0663, "step": 6760 }, { "epoch": 4.31485022307202, "grad_norm": 0.43743348121643066, "learning_rate": 4.512666666666667e-05, "loss": 0.0508, "step": 6770 }, { "epoch": 4.321223709369025, "grad_norm": 0.5177498459815979, "learning_rate": 4.519333333333334e-05, "loss": 0.0656, "step": 6780 }, { "epoch": 4.327597195666029, "grad_norm": 0.6387514472007751, "learning_rate": 4.5260000000000004e-05, "loss": 0.0755, "step": 6790 }, { "epoch": 4.3339706819630335, "grad_norm": 0.8713686466217041, "learning_rate": 4.532666666666667e-05, "loss": 0.0658, "step": 6800 }, { "epoch": 4.340344168260038, "grad_norm": 0.4039125442504883, "learning_rate": 4.539333333333334e-05, "loss": 0.0559, "step": 6810 }, { "epoch": 4.346717654557042, "grad_norm": 0.5320805907249451, "learning_rate": 4.546e-05, "loss": 0.0692, "step": 6820 }, { "epoch": 4.353091140854048, "grad_norm": 1.0053350925445557, "learning_rate": 4.5526666666666666e-05, "loss": 0.0738, "step": 6830 }, { "epoch": 4.359464627151052, "grad_norm": 0.8067516088485718, "learning_rate": 4.5593333333333336e-05, "loss": 0.0754, "step": 6840 }, { "epoch": 4.365838113448056, "grad_norm": 0.8567308783531189, "learning_rate": 4.566e-05, "loss": 0.0709, "step": 6850 }, { "epoch": 4.372211599745061, "grad_norm": 0.8834486603736877, "learning_rate": 4.572666666666667e-05, "loss": 0.054, "step": 6860 }, { "epoch": 4.378585086042065, "grad_norm": 0.8575840592384338, "learning_rate": 4.579333333333334e-05, "loss": 0.0828, "step": 6870 }, { "epoch": 4.38495857233907, "grad_norm": 0.5521349310874939, "learning_rate": 4.5860000000000005e-05, "loss": 0.0721, "step": 6880 }, { "epoch": 4.391332058636074, "grad_norm": 0.5701805949211121, "learning_rate": 4.592666666666667e-05, "loss": 0.0562, "step": 6890 }, { "epoch": 4.397705544933078, "grad_norm": 0.5939077138900757, "learning_rate": 4.599333333333334e-05, "loss": 0.0599, "step": 6900 }, { "epoch": 4.404079031230083, "grad_norm": 0.5022218823432922, "learning_rate": 4.606e-05, "loss": 0.0574, "step": 6910 }, { "epoch": 4.410452517527087, "grad_norm": 0.9328749775886536, "learning_rate": 4.612666666666667e-05, "loss": 0.0684, "step": 6920 }, { "epoch": 4.416826003824092, "grad_norm": 0.8469638824462891, "learning_rate": 4.619333333333333e-05, "loss": 0.0641, "step": 6930 }, { "epoch": 4.423199490121096, "grad_norm": 0.7458369731903076, "learning_rate": 4.626e-05, "loss": 0.057, "step": 6940 }, { "epoch": 4.4295729764181, "grad_norm": 0.7582130432128906, "learning_rate": 4.632666666666667e-05, "loss": 0.0622, "step": 6950 }, { "epoch": 4.435946462715105, "grad_norm": 0.7022218108177185, "learning_rate": 4.6393333333333335e-05, "loss": 0.0607, "step": 6960 }, { "epoch": 4.442319949012109, "grad_norm": 1.1240204572677612, "learning_rate": 4.6460000000000006e-05, "loss": 0.0668, "step": 6970 }, { "epoch": 4.448693435309114, "grad_norm": 0.8612990379333496, "learning_rate": 4.652666666666667e-05, "loss": 0.0726, "step": 6980 }, { "epoch": 4.455066921606119, "grad_norm": 0.8591046929359436, "learning_rate": 4.659333333333333e-05, "loss": 0.0598, "step": 6990 }, { "epoch": 4.461440407903123, "grad_norm": 0.9302867650985718, "learning_rate": 4.6660000000000004e-05, "loss": 0.0778, "step": 7000 }, { "epoch": 4.467813894200128, "grad_norm": 0.7860726714134216, "learning_rate": 4.672666666666667e-05, "loss": 0.0745, "step": 7010 }, { "epoch": 4.474187380497132, "grad_norm": 0.8013750910758972, "learning_rate": 4.679333333333333e-05, "loss": 0.0594, "step": 7020 }, { "epoch": 4.480560866794137, "grad_norm": 1.153072476387024, "learning_rate": 4.686e-05, "loss": 0.0818, "step": 7030 }, { "epoch": 4.486934353091141, "grad_norm": 0.8857479691505432, "learning_rate": 4.6926666666666666e-05, "loss": 0.0751, "step": 7040 }, { "epoch": 4.493307839388145, "grad_norm": 0.6867552399635315, "learning_rate": 4.6993333333333336e-05, "loss": 0.0498, "step": 7050 }, { "epoch": 4.49968132568515, "grad_norm": 0.696652889251709, "learning_rate": 4.706000000000001e-05, "loss": 0.0578, "step": 7060 }, { "epoch": 4.506054811982154, "grad_norm": 0.5428764820098877, "learning_rate": 4.712666666666667e-05, "loss": 0.0595, "step": 7070 }, { "epoch": 4.512428298279159, "grad_norm": 1.3017147779464722, "learning_rate": 4.7193333333333334e-05, "loss": 0.0929, "step": 7080 }, { "epoch": 4.518801784576163, "grad_norm": 0.595328688621521, "learning_rate": 4.7260000000000005e-05, "loss": 0.0652, "step": 7090 }, { "epoch": 4.525175270873167, "grad_norm": 0.9018744826316833, "learning_rate": 4.732666666666667e-05, "loss": 0.0642, "step": 7100 }, { "epoch": 4.531548757170172, "grad_norm": 0.7857619524002075, "learning_rate": 4.739333333333333e-05, "loss": 0.0836, "step": 7110 }, { "epoch": 4.537922243467176, "grad_norm": 0.6209820508956909, "learning_rate": 4.746e-05, "loss": 0.0481, "step": 7120 }, { "epoch": 4.5442957297641815, "grad_norm": 0.9665376543998718, "learning_rate": 4.752666666666667e-05, "loss": 0.0557, "step": 7130 }, { "epoch": 4.550669216061186, "grad_norm": 1.2759356498718262, "learning_rate": 4.759333333333334e-05, "loss": 0.0737, "step": 7140 }, { "epoch": 4.55704270235819, "grad_norm": 0.674159824848175, "learning_rate": 4.766000000000001e-05, "loss": 0.0615, "step": 7150 }, { "epoch": 4.563416188655195, "grad_norm": 1.0083224773406982, "learning_rate": 4.772666666666667e-05, "loss": 0.0712, "step": 7160 }, { "epoch": 4.569789674952199, "grad_norm": 0.5094786882400513, "learning_rate": 4.7793333333333335e-05, "loss": 0.0684, "step": 7170 }, { "epoch": 4.5761631612492035, "grad_norm": 0.8377692103385925, "learning_rate": 4.7860000000000006e-05, "loss": 0.0599, "step": 7180 }, { "epoch": 4.582536647546208, "grad_norm": 0.43251752853393555, "learning_rate": 4.792666666666667e-05, "loss": 0.0707, "step": 7190 }, { "epoch": 4.588910133843212, "grad_norm": 1.0879417657852173, "learning_rate": 4.7993333333333333e-05, "loss": 0.0525, "step": 7200 }, { "epoch": 4.595283620140217, "grad_norm": 0.9131572842597961, "learning_rate": 4.8060000000000004e-05, "loss": 0.0667, "step": 7210 }, { "epoch": 4.601657106437221, "grad_norm": 0.4205023944377899, "learning_rate": 4.812666666666667e-05, "loss": 0.0656, "step": 7220 }, { "epoch": 4.6080305927342256, "grad_norm": 0.4257108271121979, "learning_rate": 4.819333333333333e-05, "loss": 0.0538, "step": 7230 }, { "epoch": 4.61440407903123, "grad_norm": 0.9521909952163696, "learning_rate": 4.826e-05, "loss": 0.0673, "step": 7240 }, { "epoch": 4.620777565328234, "grad_norm": 0.8961038589477539, "learning_rate": 4.832666666666667e-05, "loss": 0.0717, "step": 7250 }, { "epoch": 4.627151051625239, "grad_norm": 0.6734869480133057, "learning_rate": 4.8393333333333336e-05, "loss": 0.0622, "step": 7260 }, { "epoch": 4.633524537922243, "grad_norm": 0.5155494213104248, "learning_rate": 4.846e-05, "loss": 0.0624, "step": 7270 }, { "epoch": 4.639898024219248, "grad_norm": 0.8553544282913208, "learning_rate": 4.852666666666667e-05, "loss": 0.059, "step": 7280 }, { "epoch": 4.646271510516252, "grad_norm": 1.181668758392334, "learning_rate": 4.8593333333333335e-05, "loss": 0.0696, "step": 7290 }, { "epoch": 4.652644996813256, "grad_norm": 0.7553808093070984, "learning_rate": 4.866e-05, "loss": 0.0633, "step": 7300 }, { "epoch": 4.659018483110262, "grad_norm": 0.834030270576477, "learning_rate": 4.872666666666667e-05, "loss": 0.0611, "step": 7310 }, { "epoch": 4.665391969407266, "grad_norm": 0.5208401679992676, "learning_rate": 4.879333333333333e-05, "loss": 0.0529, "step": 7320 }, { "epoch": 4.6717654557042705, "grad_norm": 0.8630608320236206, "learning_rate": 4.886e-05, "loss": 0.0572, "step": 7330 }, { "epoch": 4.678138942001275, "grad_norm": 1.0148797035217285, "learning_rate": 4.8926666666666674e-05, "loss": 0.0787, "step": 7340 }, { "epoch": 4.684512428298279, "grad_norm": 0.5508243441581726, "learning_rate": 4.899333333333334e-05, "loss": 0.0485, "step": 7350 }, { "epoch": 4.690885914595284, "grad_norm": 0.783965528011322, "learning_rate": 4.906e-05, "loss": 0.0709, "step": 7360 }, { "epoch": 4.697259400892288, "grad_norm": 0.7949073910713196, "learning_rate": 4.912666666666667e-05, "loss": 0.0656, "step": 7370 }, { "epoch": 4.7036328871892925, "grad_norm": 0.8483895659446716, "learning_rate": 4.9193333333333336e-05, "loss": 0.0712, "step": 7380 }, { "epoch": 4.710006373486297, "grad_norm": 0.504065752029419, "learning_rate": 4.926e-05, "loss": 0.064, "step": 7390 }, { "epoch": 4.716379859783301, "grad_norm": 0.6554313898086548, "learning_rate": 4.932666666666667e-05, "loss": 0.0585, "step": 7400 }, { "epoch": 4.722753346080306, "grad_norm": 0.6560725569725037, "learning_rate": 4.9393333333333334e-05, "loss": 0.068, "step": 7410 }, { "epoch": 4.72912683237731, "grad_norm": 0.7313477396965027, "learning_rate": 4.946e-05, "loss": 0.066, "step": 7420 }, { "epoch": 4.7355003186743145, "grad_norm": 0.8558791279792786, "learning_rate": 4.952666666666667e-05, "loss": 0.0774, "step": 7430 }, { "epoch": 4.741873804971319, "grad_norm": 0.6880666017532349, "learning_rate": 4.959333333333334e-05, "loss": 0.0684, "step": 7440 }, { "epoch": 4.748247291268324, "grad_norm": 0.9492466449737549, "learning_rate": 4.966e-05, "loss": 0.067, "step": 7450 }, { "epoch": 4.754620777565329, "grad_norm": 0.786095142364502, "learning_rate": 4.972666666666667e-05, "loss": 0.0584, "step": 7460 }, { "epoch": 4.760994263862333, "grad_norm": 0.6420959234237671, "learning_rate": 4.9793333333333337e-05, "loss": 0.072, "step": 7470 }, { "epoch": 4.7673677501593374, "grad_norm": 0.598442792892456, "learning_rate": 4.986e-05, "loss": 0.0655, "step": 7480 }, { "epoch": 4.773741236456342, "grad_norm": 1.136210560798645, "learning_rate": 4.992666666666667e-05, "loss": 0.0667, "step": 7490 }, { "epoch": 4.780114722753346, "grad_norm": 1.0655301809310913, "learning_rate": 4.9993333333333335e-05, "loss": 0.0606, "step": 7500 }, { "epoch": 4.786488209050351, "grad_norm": 0.7716424465179443, "learning_rate": 5.0060000000000005e-05, "loss": 0.0501, "step": 7510 }, { "epoch": 4.792861695347355, "grad_norm": 0.7212883234024048, "learning_rate": 5.012666666666667e-05, "loss": 0.0641, "step": 7520 }, { "epoch": 4.7992351816443595, "grad_norm": 0.8329992294311523, "learning_rate": 5.019333333333334e-05, "loss": 0.0429, "step": 7530 }, { "epoch": 4.805608667941364, "grad_norm": 0.5419008731842041, "learning_rate": 5.026000000000001e-05, "loss": 0.0558, "step": 7540 }, { "epoch": 4.811982154238368, "grad_norm": 0.4046176075935364, "learning_rate": 5.032666666666667e-05, "loss": 0.0522, "step": 7550 }, { "epoch": 4.818355640535373, "grad_norm": 0.7823494076728821, "learning_rate": 5.039333333333334e-05, "loss": 0.0629, "step": 7560 }, { "epoch": 4.824729126832377, "grad_norm": 1.1260772943496704, "learning_rate": 5.046000000000001e-05, "loss": 0.0743, "step": 7570 }, { "epoch": 4.8311026131293815, "grad_norm": 1.0845251083374023, "learning_rate": 5.0526666666666665e-05, "loss": 0.0669, "step": 7580 }, { "epoch": 4.837476099426386, "grad_norm": 0.9632096886634827, "learning_rate": 5.0593333333333336e-05, "loss": 0.0655, "step": 7590 }, { "epoch": 4.84384958572339, "grad_norm": 0.7473251223564148, "learning_rate": 5.0660000000000006e-05, "loss": 0.0737, "step": 7600 }, { "epoch": 4.850223072020395, "grad_norm": 0.8181489109992981, "learning_rate": 5.072666666666667e-05, "loss": 0.0695, "step": 7610 }, { "epoch": 4.8565965583174, "grad_norm": 0.8925595879554749, "learning_rate": 5.079333333333334e-05, "loss": 0.0621, "step": 7620 }, { "epoch": 4.862970044614404, "grad_norm": 0.9254628419876099, "learning_rate": 5.0860000000000004e-05, "loss": 0.0559, "step": 7630 }, { "epoch": 4.869343530911409, "grad_norm": 0.7745023965835571, "learning_rate": 5.092666666666667e-05, "loss": 0.0587, "step": 7640 }, { "epoch": 4.875717017208413, "grad_norm": 1.0639039278030396, "learning_rate": 5.099333333333334e-05, "loss": 0.0738, "step": 7650 }, { "epoch": 4.882090503505418, "grad_norm": 0.8726522922515869, "learning_rate": 5.106000000000001e-05, "loss": 0.0648, "step": 7660 }, { "epoch": 4.888463989802422, "grad_norm": 1.01661217212677, "learning_rate": 5.1126666666666666e-05, "loss": 0.0696, "step": 7670 }, { "epoch": 4.894837476099426, "grad_norm": 0.8254091739654541, "learning_rate": 5.119333333333334e-05, "loss": 0.0679, "step": 7680 }, { "epoch": 4.901210962396431, "grad_norm": 1.0373421907424927, "learning_rate": 5.1259999999999994e-05, "loss": 0.0675, "step": 7690 }, { "epoch": 4.907584448693435, "grad_norm": 1.0218459367752075, "learning_rate": 5.1326666666666664e-05, "loss": 0.0655, "step": 7700 }, { "epoch": 4.91395793499044, "grad_norm": 1.1121819019317627, "learning_rate": 5.1393333333333335e-05, "loss": 0.0672, "step": 7710 }, { "epoch": 4.920331421287444, "grad_norm": 0.6847512125968933, "learning_rate": 5.146e-05, "loss": 0.0812, "step": 7720 }, { "epoch": 4.9267049075844485, "grad_norm": 0.5663353204727173, "learning_rate": 5.152666666666667e-05, "loss": 0.0619, "step": 7730 }, { "epoch": 4.933078393881453, "grad_norm": 0.7508721947669983, "learning_rate": 5.159333333333334e-05, "loss": 0.0695, "step": 7740 }, { "epoch": 4.939451880178457, "grad_norm": 0.8792316317558289, "learning_rate": 5.166e-05, "loss": 0.0621, "step": 7750 }, { "epoch": 4.945825366475463, "grad_norm": 1.2483041286468506, "learning_rate": 5.172666666666667e-05, "loss": 0.0749, "step": 7760 }, { "epoch": 4.952198852772467, "grad_norm": 1.3614460229873657, "learning_rate": 5.179333333333334e-05, "loss": 0.0638, "step": 7770 }, { "epoch": 4.958572339069471, "grad_norm": 0.8524526953697205, "learning_rate": 5.1859999999999995e-05, "loss": 0.0772, "step": 7780 }, { "epoch": 4.964945825366476, "grad_norm": 0.7259225845336914, "learning_rate": 5.1926666666666665e-05, "loss": 0.0716, "step": 7790 }, { "epoch": 4.97131931166348, "grad_norm": 0.46326547861099243, "learning_rate": 5.1993333333333336e-05, "loss": 0.0617, "step": 7800 }, { "epoch": 4.977692797960485, "grad_norm": 0.9748802185058594, "learning_rate": 5.206e-05, "loss": 0.0578, "step": 7810 }, { "epoch": 4.984066284257489, "grad_norm": 0.8784742951393127, "learning_rate": 5.212666666666667e-05, "loss": 0.0532, "step": 7820 }, { "epoch": 4.990439770554493, "grad_norm": 0.6700788140296936, "learning_rate": 5.219333333333334e-05, "loss": 0.0597, "step": 7830 }, { "epoch": 4.996813256851498, "grad_norm": 0.5486948490142822, "learning_rate": 5.226e-05, "loss": 0.0622, "step": 7840 }, { "epoch": 5.003186743148502, "grad_norm": 0.5421954989433289, "learning_rate": 5.232666666666667e-05, "loss": 0.0669, "step": 7850 }, { "epoch": 5.009560229445507, "grad_norm": 0.5749859809875488, "learning_rate": 5.239333333333334e-05, "loss": 0.0637, "step": 7860 }, { "epoch": 5.015933715742511, "grad_norm": 0.7803623676300049, "learning_rate": 5.2459999999999996e-05, "loss": 0.0662, "step": 7870 }, { "epoch": 5.022307202039515, "grad_norm": 1.003782868385315, "learning_rate": 5.2526666666666666e-05, "loss": 0.0667, "step": 7880 }, { "epoch": 5.02868068833652, "grad_norm": 1.059926986694336, "learning_rate": 5.259333333333334e-05, "loss": 0.0552, "step": 7890 }, { "epoch": 5.035054174633524, "grad_norm": 0.7520424127578735, "learning_rate": 5.266e-05, "loss": 0.064, "step": 7900 }, { "epoch": 5.041427660930529, "grad_norm": 0.5341896414756775, "learning_rate": 5.272666666666667e-05, "loss": 0.0636, "step": 7910 }, { "epoch": 5.047801147227533, "grad_norm": 0.7045947313308716, "learning_rate": 5.279333333333334e-05, "loss": 0.0758, "step": 7920 }, { "epoch": 5.054174633524538, "grad_norm": 0.7141209840774536, "learning_rate": 5.286e-05, "loss": 0.053, "step": 7930 }, { "epoch": 5.060548119821543, "grad_norm": 0.7655815482139587, "learning_rate": 5.292666666666667e-05, "loss": 0.0647, "step": 7940 }, { "epoch": 5.066921606118547, "grad_norm": 0.6709862947463989, "learning_rate": 5.299333333333334e-05, "loss": 0.0603, "step": 7950 }, { "epoch": 5.0732950924155515, "grad_norm": 1.0763871669769287, "learning_rate": 5.306e-05, "loss": 0.0716, "step": 7960 }, { "epoch": 5.079668578712556, "grad_norm": 0.6360602974891663, "learning_rate": 5.312666666666667e-05, "loss": 0.0578, "step": 7970 }, { "epoch": 5.08604206500956, "grad_norm": 0.6845023036003113, "learning_rate": 5.319333333333334e-05, "loss": 0.0599, "step": 7980 }, { "epoch": 5.092415551306565, "grad_norm": 1.253475546836853, "learning_rate": 5.326e-05, "loss": 0.0686, "step": 7990 }, { "epoch": 5.098789037603569, "grad_norm": 0.881453812122345, "learning_rate": 5.332666666666667e-05, "loss": 0.0545, "step": 8000 }, { "epoch": 5.105162523900574, "grad_norm": 1.1186039447784424, "learning_rate": 5.339333333333334e-05, "loss": 0.063, "step": 8010 }, { "epoch": 5.111536010197578, "grad_norm": 0.542425811290741, "learning_rate": 5.346e-05, "loss": 0.0566, "step": 8020 }, { "epoch": 5.117909496494582, "grad_norm": 0.6032022833824158, "learning_rate": 5.352666666666667e-05, "loss": 0.0604, "step": 8030 }, { "epoch": 5.124282982791587, "grad_norm": 0.7036418318748474, "learning_rate": 5.359333333333334e-05, "loss": 0.0585, "step": 8040 }, { "epoch": 5.130656469088591, "grad_norm": 0.5876562595367432, "learning_rate": 5.366e-05, "loss": 0.0699, "step": 8050 }, { "epoch": 5.137029955385596, "grad_norm": 1.1779491901397705, "learning_rate": 5.372666666666667e-05, "loss": 0.0626, "step": 8060 }, { "epoch": 5.1434034416826, "grad_norm": 0.5704241991043091, "learning_rate": 5.379333333333334e-05, "loss": 0.0604, "step": 8070 }, { "epoch": 5.149776927979605, "grad_norm": 0.7771600484848022, "learning_rate": 5.386e-05, "loss": 0.0565, "step": 8080 }, { "epoch": 5.15615041427661, "grad_norm": 0.694642186164856, "learning_rate": 5.3926666666666666e-05, "loss": 0.0733, "step": 8090 }, { "epoch": 5.162523900573614, "grad_norm": 0.9646157026290894, "learning_rate": 5.399333333333334e-05, "loss": 0.0704, "step": 8100 }, { "epoch": 5.1688973868706185, "grad_norm": 0.5102450847625732, "learning_rate": 5.406e-05, "loss": 0.0633, "step": 8110 }, { "epoch": 5.175270873167623, "grad_norm": 0.8537219762802124, "learning_rate": 5.412666666666667e-05, "loss": 0.0601, "step": 8120 }, { "epoch": 5.181644359464627, "grad_norm": 0.8907761573791504, "learning_rate": 5.419333333333334e-05, "loss": 0.0616, "step": 8130 }, { "epoch": 5.188017845761632, "grad_norm": 0.9172653555870056, "learning_rate": 5.426e-05, "loss": 0.087, "step": 8140 }, { "epoch": 5.194391332058636, "grad_norm": 0.6922829747200012, "learning_rate": 5.432666666666667e-05, "loss": 0.0627, "step": 8150 }, { "epoch": 5.2007648183556405, "grad_norm": 0.6184099912643433, "learning_rate": 5.439333333333334e-05, "loss": 0.0682, "step": 8160 }, { "epoch": 5.207138304652645, "grad_norm": 0.3626927137374878, "learning_rate": 5.446e-05, "loss": 0.0694, "step": 8170 }, { "epoch": 5.213511790949649, "grad_norm": 0.7159704566001892, "learning_rate": 5.452666666666667e-05, "loss": 0.0532, "step": 8180 }, { "epoch": 5.219885277246654, "grad_norm": 0.42213982343673706, "learning_rate": 5.459333333333334e-05, "loss": 0.0588, "step": 8190 }, { "epoch": 5.226258763543658, "grad_norm": 1.0273957252502441, "learning_rate": 5.466e-05, "loss": 0.0654, "step": 8200 }, { "epoch": 5.2326322498406626, "grad_norm": 0.8868750929832458, "learning_rate": 5.472666666666667e-05, "loss": 0.0466, "step": 8210 }, { "epoch": 5.239005736137667, "grad_norm": 0.42625030875205994, "learning_rate": 5.479333333333334e-05, "loss": 0.0525, "step": 8220 }, { "epoch": 5.245379222434671, "grad_norm": 0.40689733624458313, "learning_rate": 5.486e-05, "loss": 0.05, "step": 8230 }, { "epoch": 5.251752708731676, "grad_norm": 0.838309645652771, "learning_rate": 5.492666666666667e-05, "loss": 0.0655, "step": 8240 }, { "epoch": 5.258126195028681, "grad_norm": 0.6672176122665405, "learning_rate": 5.499333333333334e-05, "loss": 0.0616, "step": 8250 }, { "epoch": 5.2644996813256855, "grad_norm": 0.6883916854858398, "learning_rate": 5.506e-05, "loss": 0.044, "step": 8260 }, { "epoch": 5.27087316762269, "grad_norm": 0.7790850400924683, "learning_rate": 5.512666666666667e-05, "loss": 0.0619, "step": 8270 }, { "epoch": 5.277246653919694, "grad_norm": 0.6095466613769531, "learning_rate": 5.519333333333334e-05, "loss": 0.0685, "step": 8280 }, { "epoch": 5.283620140216699, "grad_norm": 1.0407490730285645, "learning_rate": 5.526e-05, "loss": 0.0696, "step": 8290 }, { "epoch": 5.289993626513703, "grad_norm": 0.9175785183906555, "learning_rate": 5.5326666666666673e-05, "loss": 0.0649, "step": 8300 }, { "epoch": 5.2963671128107075, "grad_norm": 0.8231756687164307, "learning_rate": 5.5393333333333344e-05, "loss": 0.0678, "step": 8310 }, { "epoch": 5.302740599107712, "grad_norm": 0.7577474117279053, "learning_rate": 5.546e-05, "loss": 0.0609, "step": 8320 }, { "epoch": 5.309114085404716, "grad_norm": 0.8225215077400208, "learning_rate": 5.552666666666667e-05, "loss": 0.0566, "step": 8330 }, { "epoch": 5.315487571701721, "grad_norm": 0.7148070335388184, "learning_rate": 5.559333333333333e-05, "loss": 0.0538, "step": 8340 }, { "epoch": 5.321861057998725, "grad_norm": 0.5415747761726379, "learning_rate": 5.566e-05, "loss": 0.0656, "step": 8350 }, { "epoch": 5.3282345442957295, "grad_norm": 0.5351488590240479, "learning_rate": 5.572666666666667e-05, "loss": 0.0517, "step": 8360 }, { "epoch": 5.334608030592734, "grad_norm": 0.988207995891571, "learning_rate": 5.579333333333333e-05, "loss": 0.0588, "step": 8370 }, { "epoch": 5.340981516889738, "grad_norm": 0.5303449630737305, "learning_rate": 5.5860000000000004e-05, "loss": 0.0578, "step": 8380 }, { "epoch": 5.347355003186744, "grad_norm": 0.3682520091533661, "learning_rate": 5.5926666666666674e-05, "loss": 0.0539, "step": 8390 }, { "epoch": 5.353728489483748, "grad_norm": 0.9374811053276062, "learning_rate": 5.599333333333333e-05, "loss": 0.0565, "step": 8400 }, { "epoch": 5.360101975780752, "grad_norm": 0.8362928032875061, "learning_rate": 5.606e-05, "loss": 0.0605, "step": 8410 }, { "epoch": 5.366475462077757, "grad_norm": 0.6892371773719788, "learning_rate": 5.612666666666667e-05, "loss": 0.0514, "step": 8420 }, { "epoch": 5.372848948374761, "grad_norm": 0.6773945689201355, "learning_rate": 5.619333333333333e-05, "loss": 0.0512, "step": 8430 }, { "epoch": 5.379222434671766, "grad_norm": 0.5313262939453125, "learning_rate": 5.626e-05, "loss": 0.0533, "step": 8440 }, { "epoch": 5.38559592096877, "grad_norm": 0.554133951663971, "learning_rate": 5.632666666666667e-05, "loss": 0.0545, "step": 8450 }, { "epoch": 5.3919694072657744, "grad_norm": 0.7247320413589478, "learning_rate": 5.6393333333333334e-05, "loss": 0.0521, "step": 8460 }, { "epoch": 5.398342893562779, "grad_norm": 0.561715304851532, "learning_rate": 5.6460000000000005e-05, "loss": 0.0433, "step": 8470 }, { "epoch": 5.404716379859783, "grad_norm": 0.6583260297775269, "learning_rate": 5.6526666666666675e-05, "loss": 0.0461, "step": 8480 }, { "epoch": 5.411089866156788, "grad_norm": 1.1962138414382935, "learning_rate": 5.659333333333333e-05, "loss": 0.0592, "step": 8490 }, { "epoch": 5.417463352453792, "grad_norm": 0.8981515765190125, "learning_rate": 5.666e-05, "loss": 0.084, "step": 8500 }, { "epoch": 5.4238368387507965, "grad_norm": 0.5341464877128601, "learning_rate": 5.6726666666666674e-05, "loss": 0.0573, "step": 8510 }, { "epoch": 5.430210325047801, "grad_norm": 0.5179601311683655, "learning_rate": 5.679333333333333e-05, "loss": 0.0512, "step": 8520 }, { "epoch": 5.436583811344805, "grad_norm": 0.8599264025688171, "learning_rate": 5.686e-05, "loss": 0.0575, "step": 8530 }, { "epoch": 5.44295729764181, "grad_norm": 0.9956772923469543, "learning_rate": 5.692666666666667e-05, "loss": 0.0611, "step": 8540 }, { "epoch": 5.449330783938814, "grad_norm": 0.5431164503097534, "learning_rate": 5.699333333333333e-05, "loss": 0.0617, "step": 8550 }, { "epoch": 5.455704270235819, "grad_norm": 0.3895617723464966, "learning_rate": 5.706e-05, "loss": 0.0623, "step": 8560 }, { "epoch": 5.462077756532824, "grad_norm": 0.5155817270278931, "learning_rate": 5.712666666666667e-05, "loss": 0.077, "step": 8570 }, { "epoch": 5.468451242829828, "grad_norm": 0.7464935779571533, "learning_rate": 5.7193333333333333e-05, "loss": 0.0585, "step": 8580 }, { "epoch": 5.474824729126833, "grad_norm": 0.47496673464775085, "learning_rate": 5.7260000000000004e-05, "loss": 0.0725, "step": 8590 }, { "epoch": 5.481198215423837, "grad_norm": 0.48573318123817444, "learning_rate": 5.7326666666666675e-05, "loss": 0.0423, "step": 8600 }, { "epoch": 5.487571701720841, "grad_norm": 0.47488465905189514, "learning_rate": 5.739333333333333e-05, "loss": 0.0457, "step": 8610 }, { "epoch": 5.493945188017846, "grad_norm": 0.698132336139679, "learning_rate": 5.746e-05, "loss": 0.0593, "step": 8620 }, { "epoch": 5.50031867431485, "grad_norm": 0.6548746824264526, "learning_rate": 5.752666666666667e-05, "loss": 0.073, "step": 8630 }, { "epoch": 5.506692160611855, "grad_norm": 0.8008704781532288, "learning_rate": 5.759333333333333e-05, "loss": 0.0519, "step": 8640 }, { "epoch": 5.513065646908859, "grad_norm": 0.5915107131004333, "learning_rate": 5.766e-05, "loss": 0.069, "step": 8650 }, { "epoch": 5.519439133205863, "grad_norm": 0.5799915790557861, "learning_rate": 5.772666666666667e-05, "loss": 0.0645, "step": 8660 }, { "epoch": 5.525812619502868, "grad_norm": 1.300042748451233, "learning_rate": 5.7793333333333335e-05, "loss": 0.0705, "step": 8670 }, { "epoch": 5.532186105799872, "grad_norm": 0.764645516872406, "learning_rate": 5.7860000000000005e-05, "loss": 0.0476, "step": 8680 }, { "epoch": 5.538559592096877, "grad_norm": 0.7687389850616455, "learning_rate": 5.7926666666666676e-05, "loss": 0.0611, "step": 8690 }, { "epoch": 5.544933078393882, "grad_norm": 1.612897276878357, "learning_rate": 5.799333333333333e-05, "loss": 0.0652, "step": 8700 }, { "epoch": 5.551306564690886, "grad_norm": 0.5542522668838501, "learning_rate": 5.806e-05, "loss": 0.0563, "step": 8710 }, { "epoch": 5.557680050987891, "grad_norm": 0.8147121667861938, "learning_rate": 5.8126666666666674e-05, "loss": 0.0678, "step": 8720 }, { "epoch": 5.564053537284895, "grad_norm": 0.9129625558853149, "learning_rate": 5.819333333333333e-05, "loss": 0.0637, "step": 8730 }, { "epoch": 5.5704270235819, "grad_norm": 0.6277623176574707, "learning_rate": 5.826e-05, "loss": 0.0552, "step": 8740 }, { "epoch": 5.576800509878904, "grad_norm": 0.5057451128959656, "learning_rate": 5.832666666666667e-05, "loss": 0.0566, "step": 8750 }, { "epoch": 5.583173996175908, "grad_norm": 0.5900144577026367, "learning_rate": 5.8393333333333336e-05, "loss": 0.0577, "step": 8760 }, { "epoch": 5.589547482472913, "grad_norm": 0.6781315803527832, "learning_rate": 5.8460000000000006e-05, "loss": 0.0549, "step": 8770 }, { "epoch": 5.595920968769917, "grad_norm": 0.7219472527503967, "learning_rate": 5.852666666666668e-05, "loss": 0.063, "step": 8780 }, { "epoch": 5.602294455066922, "grad_norm": 0.7543862462043762, "learning_rate": 5.8593333333333334e-05, "loss": 0.0578, "step": 8790 }, { "epoch": 5.608667941363926, "grad_norm": 0.8206136226654053, "learning_rate": 5.8660000000000004e-05, "loss": 0.0491, "step": 8800 }, { "epoch": 5.61504142766093, "grad_norm": 0.9566856622695923, "learning_rate": 5.8726666666666675e-05, "loss": 0.0557, "step": 8810 }, { "epoch": 5.621414913957935, "grad_norm": 0.9145086407661438, "learning_rate": 5.879333333333333e-05, "loss": 0.0543, "step": 8820 }, { "epoch": 5.627788400254939, "grad_norm": 0.48449772596359253, "learning_rate": 5.886e-05, "loss": 0.0548, "step": 8830 }, { "epoch": 5.634161886551944, "grad_norm": 0.7416876554489136, "learning_rate": 5.892666666666667e-05, "loss": 0.0565, "step": 8840 }, { "epoch": 5.640535372848948, "grad_norm": 0.5322074890136719, "learning_rate": 5.8993333333333337e-05, "loss": 0.0528, "step": 8850 }, { "epoch": 5.646908859145952, "grad_norm": 0.7602063417434692, "learning_rate": 5.906000000000001e-05, "loss": 0.0569, "step": 8860 }, { "epoch": 5.653282345442957, "grad_norm": 0.6393805146217346, "learning_rate": 5.912666666666668e-05, "loss": 0.0476, "step": 8870 }, { "epoch": 5.659655831739962, "grad_norm": 0.8366895914077759, "learning_rate": 5.9193333333333335e-05, "loss": 0.0758, "step": 8880 }, { "epoch": 5.6660293180369665, "grad_norm": 0.6291630268096924, "learning_rate": 5.9260000000000005e-05, "loss": 0.0673, "step": 8890 }, { "epoch": 5.672402804333971, "grad_norm": 0.8264247179031372, "learning_rate": 5.9326666666666676e-05, "loss": 0.0578, "step": 8900 }, { "epoch": 5.678776290630975, "grad_norm": 0.8180743455886841, "learning_rate": 5.939333333333333e-05, "loss": 0.0611, "step": 8910 }, { "epoch": 5.68514977692798, "grad_norm": 1.2748998403549194, "learning_rate": 5.946e-05, "loss": 0.0734, "step": 8920 }, { "epoch": 5.691523263224984, "grad_norm": 0.7235389351844788, "learning_rate": 5.9526666666666674e-05, "loss": 0.0628, "step": 8930 }, { "epoch": 5.6978967495219885, "grad_norm": 0.7432854175567627, "learning_rate": 5.959333333333334e-05, "loss": 0.0727, "step": 8940 }, { "epoch": 5.704270235818993, "grad_norm": 0.45918038487434387, "learning_rate": 5.966000000000001e-05, "loss": 0.0519, "step": 8950 }, { "epoch": 5.710643722115997, "grad_norm": 0.797531008720398, "learning_rate": 5.9726666666666665e-05, "loss": 0.0549, "step": 8960 }, { "epoch": 5.717017208413002, "grad_norm": 0.8184925317764282, "learning_rate": 5.9793333333333336e-05, "loss": 0.0537, "step": 8970 }, { "epoch": 5.723390694710006, "grad_norm": 0.7258854508399963, "learning_rate": 5.9860000000000006e-05, "loss": 0.0672, "step": 8980 }, { "epoch": 5.729764181007011, "grad_norm": 0.8069808483123779, "learning_rate": 5.992666666666666e-05, "loss": 0.0845, "step": 8990 }, { "epoch": 5.736137667304015, "grad_norm": 0.6261904239654541, "learning_rate": 5.9993333333333334e-05, "loss": 0.066, "step": 9000 }, { "epoch": 5.742511153601019, "grad_norm": 0.8198938369750977, "learning_rate": 6.0060000000000004e-05, "loss": 0.0539, "step": 9010 }, { "epoch": 5.748884639898025, "grad_norm": 0.5557858347892761, "learning_rate": 6.012666666666666e-05, "loss": 0.0566, "step": 9020 }, { "epoch": 5.755258126195029, "grad_norm": 0.5786111354827881, "learning_rate": 6.019333333333333e-05, "loss": 0.0729, "step": 9030 }, { "epoch": 5.7616316124920335, "grad_norm": 0.4758230149745941, "learning_rate": 6.026e-05, "loss": 0.0581, "step": 9040 }, { "epoch": 5.768005098789038, "grad_norm": 0.7065956592559814, "learning_rate": 6.0326666666666666e-05, "loss": 0.0617, "step": 9050 }, { "epoch": 5.774378585086042, "grad_norm": 0.8448079228401184, "learning_rate": 6.039333333333334e-05, "loss": 0.0676, "step": 9060 }, { "epoch": 5.780752071383047, "grad_norm": 0.6785176992416382, "learning_rate": 6.046000000000001e-05, "loss": 0.0585, "step": 9070 }, { "epoch": 5.787125557680051, "grad_norm": 1.0336339473724365, "learning_rate": 6.0526666666666664e-05, "loss": 0.0655, "step": 9080 }, { "epoch": 5.7934990439770555, "grad_norm": 0.620617151260376, "learning_rate": 6.0593333333333335e-05, "loss": 0.0644, "step": 9090 }, { "epoch": 5.79987253027406, "grad_norm": 0.9712967872619629, "learning_rate": 6.0660000000000005e-05, "loss": 0.0697, "step": 9100 }, { "epoch": 5.806246016571064, "grad_norm": 0.8093079328536987, "learning_rate": 6.072666666666666e-05, "loss": 0.0604, "step": 9110 }, { "epoch": 5.812619502868069, "grad_norm": 0.714555025100708, "learning_rate": 6.079333333333333e-05, "loss": 0.0614, "step": 9120 }, { "epoch": 5.818992989165073, "grad_norm": 0.6867609620094299, "learning_rate": 6.0860000000000003e-05, "loss": 0.0707, "step": 9130 }, { "epoch": 5.8253664754620775, "grad_norm": 0.6431583762168884, "learning_rate": 6.092666666666667e-05, "loss": 0.0573, "step": 9140 }, { "epoch": 5.831739961759082, "grad_norm": 0.5615859031677246, "learning_rate": 6.099333333333334e-05, "loss": 0.07, "step": 9150 }, { "epoch": 5.838113448056086, "grad_norm": 0.4224008619785309, "learning_rate": 6.106e-05, "loss": 0.0514, "step": 9160 }, { "epoch": 5.844486934353091, "grad_norm": 0.769130289554596, "learning_rate": 6.112666666666667e-05, "loss": 0.0467, "step": 9170 }, { "epoch": 5.850860420650095, "grad_norm": 1.0572727918624878, "learning_rate": 6.119333333333333e-05, "loss": 0.0726, "step": 9180 }, { "epoch": 5.8572339069471, "grad_norm": 0.9408935308456421, "learning_rate": 6.126e-05, "loss": 0.0555, "step": 9190 }, { "epoch": 5.863607393244105, "grad_norm": 0.7971311211585999, "learning_rate": 6.132666666666667e-05, "loss": 0.0618, "step": 9200 }, { "epoch": 5.869980879541109, "grad_norm": 0.6989256739616394, "learning_rate": 6.139333333333333e-05, "loss": 0.0646, "step": 9210 }, { "epoch": 5.876354365838114, "grad_norm": 0.7526812553405762, "learning_rate": 6.146000000000001e-05, "loss": 0.0731, "step": 9220 }, { "epoch": 5.882727852135118, "grad_norm": 0.9330829381942749, "learning_rate": 6.152666666666666e-05, "loss": 0.0659, "step": 9230 }, { "epoch": 5.8891013384321225, "grad_norm": 0.5467667579650879, "learning_rate": 6.159333333333334e-05, "loss": 0.0655, "step": 9240 }, { "epoch": 5.895474824729127, "grad_norm": 0.5973659157752991, "learning_rate": 6.166e-05, "loss": 0.0488, "step": 9250 }, { "epoch": 5.901848311026131, "grad_norm": 0.4567822813987732, "learning_rate": 6.172666666666667e-05, "loss": 0.0544, "step": 9260 }, { "epoch": 5.908221797323136, "grad_norm": 0.6740135550498962, "learning_rate": 6.179333333333333e-05, "loss": 0.056, "step": 9270 }, { "epoch": 5.91459528362014, "grad_norm": 0.6389235854148865, "learning_rate": 6.186000000000001e-05, "loss": 0.0704, "step": 9280 }, { "epoch": 5.9209687699171445, "grad_norm": 0.5664303302764893, "learning_rate": 6.192666666666667e-05, "loss": 0.0544, "step": 9290 }, { "epoch": 5.927342256214149, "grad_norm": 0.39746248722076416, "learning_rate": 6.199333333333333e-05, "loss": 0.0693, "step": 9300 }, { "epoch": 5.933715742511153, "grad_norm": 0.6555776596069336, "learning_rate": 6.206000000000001e-05, "loss": 0.0724, "step": 9310 }, { "epoch": 5.940089228808158, "grad_norm": 1.0497114658355713, "learning_rate": 6.212666666666666e-05, "loss": 0.067, "step": 9320 }, { "epoch": 5.946462715105163, "grad_norm": 0.6835218667984009, "learning_rate": 6.219333333333334e-05, "loss": 0.0704, "step": 9330 }, { "epoch": 5.952836201402167, "grad_norm": 0.6713403463363647, "learning_rate": 6.226e-05, "loss": 0.0713, "step": 9340 }, { "epoch": 5.959209687699172, "grad_norm": 0.7689964175224304, "learning_rate": 6.232666666666667e-05, "loss": 0.0566, "step": 9350 }, { "epoch": 5.965583173996176, "grad_norm": 0.7639724016189575, "learning_rate": 6.239333333333333e-05, "loss": 0.0516, "step": 9360 }, { "epoch": 5.971956660293181, "grad_norm": 0.4493684470653534, "learning_rate": 6.246000000000001e-05, "loss": 0.0686, "step": 9370 }, { "epoch": 5.978330146590185, "grad_norm": 0.5575623512268066, "learning_rate": 6.252666666666667e-05, "loss": 0.0507, "step": 9380 }, { "epoch": 5.984703632887189, "grad_norm": 0.4197111427783966, "learning_rate": 6.259333333333334e-05, "loss": 0.0624, "step": 9390 }, { "epoch": 5.991077119184194, "grad_norm": 0.8092382550239563, "learning_rate": 6.266000000000001e-05, "loss": 0.0648, "step": 9400 }, { "epoch": 5.997450605481198, "grad_norm": 0.5399866104125977, "learning_rate": 6.272666666666666e-05, "loss": 0.0617, "step": 9410 }, { "epoch": 6.003824091778203, "grad_norm": 0.5049969553947449, "learning_rate": 6.279333333333334e-05, "loss": 0.0478, "step": 9420 }, { "epoch": 6.010197578075207, "grad_norm": 0.5460518002510071, "learning_rate": 6.286e-05, "loss": 0.0553, "step": 9430 }, { "epoch": 6.0165710643722115, "grad_norm": 0.7110340595245361, "learning_rate": 6.292666666666667e-05, "loss": 0.0535, "step": 9440 }, { "epoch": 6.022944550669216, "grad_norm": 0.6920067071914673, "learning_rate": 6.299333333333333e-05, "loss": 0.0651, "step": 9450 }, { "epoch": 6.02931803696622, "grad_norm": 0.466105192899704, "learning_rate": 6.306000000000001e-05, "loss": 0.0488, "step": 9460 }, { "epoch": 6.035691523263225, "grad_norm": 0.4926980435848236, "learning_rate": 6.312666666666667e-05, "loss": 0.0718, "step": 9470 }, { "epoch": 6.042065009560229, "grad_norm": 0.6662371158599854, "learning_rate": 6.319333333333334e-05, "loss": 0.0612, "step": 9480 }, { "epoch": 6.0484384958572335, "grad_norm": 0.9834148287773132, "learning_rate": 6.326000000000001e-05, "loss": 0.0528, "step": 9490 }, { "epoch": 6.054811982154239, "grad_norm": 0.7669191360473633, "learning_rate": 6.332666666666666e-05, "loss": 0.054, "step": 9500 }, { "epoch": 6.061185468451243, "grad_norm": 0.9060104489326477, "learning_rate": 6.339333333333334e-05, "loss": 0.0551, "step": 9510 }, { "epoch": 6.067558954748248, "grad_norm": 0.5581040382385254, "learning_rate": 6.346e-05, "loss": 0.0454, "step": 9520 }, { "epoch": 6.073932441045252, "grad_norm": 0.6774866580963135, "learning_rate": 6.352666666666667e-05, "loss": 0.0656, "step": 9530 }, { "epoch": 6.080305927342256, "grad_norm": 0.6943061351776123, "learning_rate": 6.359333333333333e-05, "loss": 0.0607, "step": 9540 }, { "epoch": 6.086679413639261, "grad_norm": 0.7664393782615662, "learning_rate": 6.366000000000001e-05, "loss": 0.0647, "step": 9550 }, { "epoch": 6.093052899936265, "grad_norm": 0.6050956845283508, "learning_rate": 6.372666666666667e-05, "loss": 0.0576, "step": 9560 }, { "epoch": 6.09942638623327, "grad_norm": 0.8026940226554871, "learning_rate": 6.379333333333334e-05, "loss": 0.0586, "step": 9570 }, { "epoch": 6.105799872530274, "grad_norm": 0.7366113662719727, "learning_rate": 6.386e-05, "loss": 0.0788, "step": 9580 }, { "epoch": 6.112173358827278, "grad_norm": 0.6704623103141785, "learning_rate": 6.392666666666667e-05, "loss": 0.045, "step": 9590 }, { "epoch": 6.118546845124283, "grad_norm": 0.7459873557090759, "learning_rate": 6.399333333333334e-05, "loss": 0.0701, "step": 9600 }, { "epoch": 6.124920331421287, "grad_norm": 0.5031256675720215, "learning_rate": 6.405999999999999e-05, "loss": 0.0514, "step": 9610 }, { "epoch": 6.131293817718292, "grad_norm": 0.8962956070899963, "learning_rate": 6.412666666666667e-05, "loss": 0.0709, "step": 9620 }, { "epoch": 6.137667304015296, "grad_norm": 0.5765659213066101, "learning_rate": 6.419333333333333e-05, "loss": 0.0525, "step": 9630 }, { "epoch": 6.1440407903123, "grad_norm": 0.7239110469818115, "learning_rate": 6.426e-05, "loss": 0.0506, "step": 9640 }, { "epoch": 6.150414276609306, "grad_norm": 0.7703732848167419, "learning_rate": 6.432666666666666e-05, "loss": 0.0564, "step": 9650 }, { "epoch": 6.15678776290631, "grad_norm": 0.6895337104797363, "learning_rate": 6.439333333333334e-05, "loss": 0.0554, "step": 9660 }, { "epoch": 6.1631612492033145, "grad_norm": 0.8975412249565125, "learning_rate": 6.446e-05, "loss": 0.071, "step": 9670 }, { "epoch": 6.169534735500319, "grad_norm": 0.7354782223701477, "learning_rate": 6.452666666666667e-05, "loss": 0.0735, "step": 9680 }, { "epoch": 6.175908221797323, "grad_norm": 0.6420743465423584, "learning_rate": 6.459333333333334e-05, "loss": 0.0668, "step": 9690 }, { "epoch": 6.182281708094328, "grad_norm": 0.9225689768791199, "learning_rate": 6.466e-05, "loss": 0.0569, "step": 9700 }, { "epoch": 6.188655194391332, "grad_norm": 1.1190143823623657, "learning_rate": 6.472666666666667e-05, "loss": 0.0593, "step": 9710 }, { "epoch": 6.195028680688337, "grad_norm": 0.5006576180458069, "learning_rate": 6.479333333333334e-05, "loss": 0.0521, "step": 9720 }, { "epoch": 6.201402166985341, "grad_norm": 0.9682351350784302, "learning_rate": 6.486e-05, "loss": 0.0696, "step": 9730 }, { "epoch": 6.207775653282345, "grad_norm": 0.7219080328941345, "learning_rate": 6.492666666666666e-05, "loss": 0.0609, "step": 9740 }, { "epoch": 6.21414913957935, "grad_norm": 0.5833954811096191, "learning_rate": 6.499333333333334e-05, "loss": 0.0587, "step": 9750 }, { "epoch": 6.220522625876354, "grad_norm": 0.43249595165252686, "learning_rate": 6.506e-05, "loss": 0.0561, "step": 9760 }, { "epoch": 6.226896112173359, "grad_norm": 0.47439971566200256, "learning_rate": 6.512666666666667e-05, "loss": 0.0531, "step": 9770 }, { "epoch": 6.233269598470363, "grad_norm": 0.43036776781082153, "learning_rate": 6.519333333333334e-05, "loss": 0.0667, "step": 9780 }, { "epoch": 6.239643084767367, "grad_norm": 0.6738905906677246, "learning_rate": 6.526e-05, "loss": 0.0516, "step": 9790 }, { "epoch": 6.246016571064372, "grad_norm": 0.6552261114120483, "learning_rate": 6.532666666666667e-05, "loss": 0.0528, "step": 9800 }, { "epoch": 6.252390057361376, "grad_norm": 0.7906575202941895, "learning_rate": 6.539333333333334e-05, "loss": 0.053, "step": 9810 }, { "epoch": 6.2587635436583815, "grad_norm": 0.4689423441886902, "learning_rate": 6.546e-05, "loss": 0.0632, "step": 9820 }, { "epoch": 6.265137029955386, "grad_norm": 0.41271916031837463, "learning_rate": 6.552666666666666e-05, "loss": 0.0568, "step": 9830 }, { "epoch": 6.27151051625239, "grad_norm": 0.7616137862205505, "learning_rate": 6.559333333333334e-05, "loss": 0.0648, "step": 9840 }, { "epoch": 6.277884002549395, "grad_norm": 0.6170486807823181, "learning_rate": 6.566e-05, "loss": 0.0626, "step": 9850 }, { "epoch": 6.284257488846399, "grad_norm": 0.894328236579895, "learning_rate": 6.572666666666667e-05, "loss": 0.0673, "step": 9860 }, { "epoch": 6.2906309751434035, "grad_norm": 0.743172824382782, "learning_rate": 6.579333333333335e-05, "loss": 0.0527, "step": 9870 }, { "epoch": 6.297004461440408, "grad_norm": 0.8911871314048767, "learning_rate": 6.586e-05, "loss": 0.0708, "step": 9880 }, { "epoch": 6.303377947737412, "grad_norm": 0.7575047016143799, "learning_rate": 6.592666666666667e-05, "loss": 0.0528, "step": 9890 }, { "epoch": 6.309751434034417, "grad_norm": 0.6481773853302002, "learning_rate": 6.599333333333334e-05, "loss": 0.0582, "step": 9900 }, { "epoch": 6.316124920331421, "grad_norm": 0.6251875162124634, "learning_rate": 6.606e-05, "loss": 0.051, "step": 9910 }, { "epoch": 6.3224984066284255, "grad_norm": 1.0031306743621826, "learning_rate": 6.612666666666666e-05, "loss": 0.0596, "step": 9920 }, { "epoch": 6.32887189292543, "grad_norm": 0.934403121471405, "learning_rate": 6.619333333333334e-05, "loss": 0.0658, "step": 9930 }, { "epoch": 6.335245379222434, "grad_norm": 0.749822199344635, "learning_rate": 6.626e-05, "loss": 0.0563, "step": 9940 }, { "epoch": 6.341618865519439, "grad_norm": 0.7393651008605957, "learning_rate": 6.632666666666667e-05, "loss": 0.0547, "step": 9950 }, { "epoch": 6.347992351816444, "grad_norm": 0.6630147695541382, "learning_rate": 6.639333333333335e-05, "loss": 0.0574, "step": 9960 }, { "epoch": 6.3543658381134485, "grad_norm": 1.2484126091003418, "learning_rate": 6.646e-05, "loss": 0.0624, "step": 9970 }, { "epoch": 6.360739324410453, "grad_norm": 0.4653473496437073, "learning_rate": 6.652666666666667e-05, "loss": 0.0511, "step": 9980 }, { "epoch": 6.367112810707457, "grad_norm": 0.6409396529197693, "learning_rate": 6.659333333333334e-05, "loss": 0.0613, "step": 9990 }, { "epoch": 6.373486297004462, "grad_norm": 0.7591714262962341, "learning_rate": 6.666e-05, "loss": 0.059, "step": 10000 }, { "epoch": 6.379859783301466, "grad_norm": 0.5877882838249207, "learning_rate": 6.672666666666667e-05, "loss": 0.0732, "step": 10010 }, { "epoch": 6.3862332695984705, "grad_norm": 0.9828495979309082, "learning_rate": 6.679333333333334e-05, "loss": 0.0619, "step": 10020 }, { "epoch": 6.392606755895475, "grad_norm": 0.6354754567146301, "learning_rate": 6.686000000000001e-05, "loss": 0.0457, "step": 10030 }, { "epoch": 6.398980242192479, "grad_norm": 0.8597481846809387, "learning_rate": 6.692666666666667e-05, "loss": 0.0615, "step": 10040 }, { "epoch": 6.405353728489484, "grad_norm": 0.5619955062866211, "learning_rate": 6.699333333333333e-05, "loss": 0.0764, "step": 10050 }, { "epoch": 6.411727214786488, "grad_norm": 0.5440130829811096, "learning_rate": 6.706e-05, "loss": 0.0566, "step": 10060 }, { "epoch": 6.4181007010834925, "grad_norm": 0.3928253948688507, "learning_rate": 6.712666666666668e-05, "loss": 0.0571, "step": 10070 }, { "epoch": 6.424474187380497, "grad_norm": 0.9080679416656494, "learning_rate": 6.719333333333334e-05, "loss": 0.062, "step": 10080 }, { "epoch": 6.430847673677501, "grad_norm": 0.6312495470046997, "learning_rate": 6.726e-05, "loss": 0.0624, "step": 10090 }, { "epoch": 6.437221159974506, "grad_norm": 0.6919388771057129, "learning_rate": 6.732666666666667e-05, "loss": 0.0683, "step": 10100 }, { "epoch": 6.44359464627151, "grad_norm": 0.48411840200424194, "learning_rate": 6.739333333333334e-05, "loss": 0.0503, "step": 10110 }, { "epoch": 6.4499681325685145, "grad_norm": 0.8454859256744385, "learning_rate": 6.746e-05, "loss": 0.053, "step": 10120 }, { "epoch": 6.45634161886552, "grad_norm": 0.71438068151474, "learning_rate": 6.752666666666667e-05, "loss": 0.0549, "step": 10130 }, { "epoch": 6.462715105162524, "grad_norm": 0.8116707801818848, "learning_rate": 6.759333333333334e-05, "loss": 0.0712, "step": 10140 }, { "epoch": 6.469088591459529, "grad_norm": 0.5926610231399536, "learning_rate": 6.766e-05, "loss": 0.0688, "step": 10150 }, { "epoch": 6.475462077756533, "grad_norm": 0.6376394629478455, "learning_rate": 6.772666666666668e-05, "loss": 0.0586, "step": 10160 }, { "epoch": 6.4818355640535374, "grad_norm": 0.7720407247543335, "learning_rate": 6.779333333333334e-05, "loss": 0.0567, "step": 10170 }, { "epoch": 6.488209050350542, "grad_norm": 0.5263776183128357, "learning_rate": 6.786e-05, "loss": 0.0665, "step": 10180 }, { "epoch": 6.494582536647546, "grad_norm": 0.902793288230896, "learning_rate": 6.792666666666667e-05, "loss": 0.0677, "step": 10190 }, { "epoch": 6.500956022944551, "grad_norm": 0.6537356376647949, "learning_rate": 6.799333333333333e-05, "loss": 0.0629, "step": 10200 }, { "epoch": 6.507329509241555, "grad_norm": 0.9066295623779297, "learning_rate": 6.806e-05, "loss": 0.0605, "step": 10210 }, { "epoch": 6.5137029955385595, "grad_norm": 0.9434707164764404, "learning_rate": 6.812666666666667e-05, "loss": 0.0546, "step": 10220 }, { "epoch": 6.520076481835564, "grad_norm": 1.070278525352478, "learning_rate": 6.819333333333334e-05, "loss": 0.0595, "step": 10230 }, { "epoch": 6.526449968132568, "grad_norm": 0.5275779366493225, "learning_rate": 6.826e-05, "loss": 0.0563, "step": 10240 }, { "epoch": 6.532823454429573, "grad_norm": 0.4790928363800049, "learning_rate": 6.832666666666668e-05, "loss": 0.0577, "step": 10250 }, { "epoch": 6.539196940726577, "grad_norm": 0.6474365592002869, "learning_rate": 6.839333333333333e-05, "loss": 0.0521, "step": 10260 }, { "epoch": 6.545570427023582, "grad_norm": 0.6602675318717957, "learning_rate": 6.846e-05, "loss": 0.0606, "step": 10270 }, { "epoch": 6.551943913320587, "grad_norm": 1.0174161195755005, "learning_rate": 6.852666666666667e-05, "loss": 0.0693, "step": 10280 }, { "epoch": 6.558317399617591, "grad_norm": 0.5428268909454346, "learning_rate": 6.859333333333333e-05, "loss": 0.0638, "step": 10290 }, { "epoch": 6.564690885914596, "grad_norm": 0.6684935688972473, "learning_rate": 6.866e-05, "loss": 0.0545, "step": 10300 }, { "epoch": 6.5710643722116, "grad_norm": 0.4289809465408325, "learning_rate": 6.872666666666667e-05, "loss": 0.0492, "step": 10310 }, { "epoch": 6.577437858508604, "grad_norm": 0.619526743888855, "learning_rate": 6.879333333333334e-05, "loss": 0.0594, "step": 10320 }, { "epoch": 6.583811344805609, "grad_norm": 0.6994621753692627, "learning_rate": 6.886e-05, "loss": 0.053, "step": 10330 }, { "epoch": 6.590184831102613, "grad_norm": 0.7008401155471802, "learning_rate": 6.892666666666668e-05, "loss": 0.0593, "step": 10340 }, { "epoch": 6.596558317399618, "grad_norm": 0.8790857195854187, "learning_rate": 6.899333333333333e-05, "loss": 0.067, "step": 10350 }, { "epoch": 6.602931803696622, "grad_norm": 0.628646194934845, "learning_rate": 6.906e-05, "loss": 0.0629, "step": 10360 }, { "epoch": 6.609305289993626, "grad_norm": 0.5086628794670105, "learning_rate": 6.912666666666667e-05, "loss": 0.0522, "step": 10370 }, { "epoch": 6.615678776290631, "grad_norm": 0.5419976115226746, "learning_rate": 6.919333333333333e-05, "loss": 0.0593, "step": 10380 }, { "epoch": 6.622052262587635, "grad_norm": 0.4934231638908386, "learning_rate": 6.926e-05, "loss": 0.0443, "step": 10390 }, { "epoch": 6.62842574888464, "grad_norm": 0.552006185054779, "learning_rate": 6.932666666666667e-05, "loss": 0.0623, "step": 10400 }, { "epoch": 6.634799235181644, "grad_norm": 0.6392134428024292, "learning_rate": 6.939333333333334e-05, "loss": 0.0586, "step": 10410 }, { "epoch": 6.6411727214786485, "grad_norm": 0.8226262331008911, "learning_rate": 6.946e-05, "loss": 0.064, "step": 10420 }, { "epoch": 6.647546207775653, "grad_norm": 0.552422285079956, "learning_rate": 6.952666666666668e-05, "loss": 0.0427, "step": 10430 }, { "epoch": 6.653919694072657, "grad_norm": 0.590422511100769, "learning_rate": 6.959333333333333e-05, "loss": 0.0497, "step": 10440 }, { "epoch": 6.660293180369663, "grad_norm": 0.6564462184906006, "learning_rate": 6.966000000000001e-05, "loss": 0.0635, "step": 10450 }, { "epoch": 6.666666666666667, "grad_norm": 0.5087324380874634, "learning_rate": 6.972666666666667e-05, "loss": 0.0546, "step": 10460 }, { "epoch": 6.673040152963671, "grad_norm": 0.5284962058067322, "learning_rate": 6.979333333333333e-05, "loss": 0.0554, "step": 10470 }, { "epoch": 6.679413639260676, "grad_norm": 0.7105791568756104, "learning_rate": 6.986e-05, "loss": 0.0459, "step": 10480 }, { "epoch": 6.68578712555768, "grad_norm": 0.5709689259529114, "learning_rate": 6.992666666666668e-05, "loss": 0.0508, "step": 10490 }, { "epoch": 6.692160611854685, "grad_norm": 0.7488101720809937, "learning_rate": 6.999333333333334e-05, "loss": 0.0471, "step": 10500 }, { "epoch": 6.698534098151689, "grad_norm": 0.7387686371803284, "learning_rate": 7.006e-05, "loss": 0.0576, "step": 10510 }, { "epoch": 6.704907584448693, "grad_norm": 0.6705060005187988, "learning_rate": 7.012666666666667e-05, "loss": 0.0488, "step": 10520 }, { "epoch": 6.711281070745698, "grad_norm": 0.6967387795448303, "learning_rate": 7.019333333333333e-05, "loss": 0.0614, "step": 10530 }, { "epoch": 6.717654557042702, "grad_norm": 0.4333428144454956, "learning_rate": 7.026000000000001e-05, "loss": 0.0483, "step": 10540 }, { "epoch": 6.724028043339707, "grad_norm": 0.4872308373451233, "learning_rate": 7.032666666666667e-05, "loss": 0.0518, "step": 10550 }, { "epoch": 6.730401529636711, "grad_norm": 0.5130177736282349, "learning_rate": 7.039333333333334e-05, "loss": 0.0553, "step": 10560 }, { "epoch": 6.736775015933715, "grad_norm": 0.421485960483551, "learning_rate": 7.046e-05, "loss": 0.0494, "step": 10570 }, { "epoch": 6.74314850223072, "grad_norm": 0.8604413866996765, "learning_rate": 7.052666666666668e-05, "loss": 0.0643, "step": 10580 }, { "epoch": 6.749521988527725, "grad_norm": 0.7307182550430298, "learning_rate": 7.059333333333333e-05, "loss": 0.0555, "step": 10590 }, { "epoch": 6.7558954748247295, "grad_norm": 0.7127339839935303, "learning_rate": 7.066e-05, "loss": 0.0744, "step": 10600 }, { "epoch": 6.762268961121734, "grad_norm": 1.1702367067337036, "learning_rate": 7.072666666666667e-05, "loss": 0.0697, "step": 10610 }, { "epoch": 6.768642447418738, "grad_norm": 1.2847914695739746, "learning_rate": 7.079333333333333e-05, "loss": 0.0651, "step": 10620 }, { "epoch": 6.775015933715743, "grad_norm": 0.593812108039856, "learning_rate": 7.086000000000001e-05, "loss": 0.0742, "step": 10630 }, { "epoch": 6.781389420012747, "grad_norm": 0.706718385219574, "learning_rate": 7.092666666666667e-05, "loss": 0.0703, "step": 10640 }, { "epoch": 6.7877629063097515, "grad_norm": 0.6615965962409973, "learning_rate": 7.099333333333334e-05, "loss": 0.0654, "step": 10650 }, { "epoch": 6.794136392606756, "grad_norm": 0.6300081610679626, "learning_rate": 7.106e-05, "loss": 0.0664, "step": 10660 }, { "epoch": 6.80050987890376, "grad_norm": 0.547766387462616, "learning_rate": 7.112666666666668e-05, "loss": 0.0666, "step": 10670 }, { "epoch": 6.806883365200765, "grad_norm": 0.6945487260818481, "learning_rate": 7.119333333333333e-05, "loss": 0.057, "step": 10680 }, { "epoch": 6.813256851497769, "grad_norm": 1.0761439800262451, "learning_rate": 7.126e-05, "loss": 0.0545, "step": 10690 }, { "epoch": 6.819630337794774, "grad_norm": 0.5777052044868469, "learning_rate": 7.132666666666667e-05, "loss": 0.0634, "step": 10700 }, { "epoch": 6.826003824091778, "grad_norm": 0.4692111909389496, "learning_rate": 7.139333333333333e-05, "loss": 0.068, "step": 10710 }, { "epoch": 6.832377310388782, "grad_norm": 0.5378316044807434, "learning_rate": 7.146000000000001e-05, "loss": 0.0661, "step": 10720 }, { "epoch": 6.838750796685787, "grad_norm": 0.7132922410964966, "learning_rate": 7.152666666666667e-05, "loss": 0.0506, "step": 10730 }, { "epoch": 6.845124282982791, "grad_norm": 0.5381823182106018, "learning_rate": 7.159333333333334e-05, "loss": 0.0509, "step": 10740 }, { "epoch": 6.851497769279796, "grad_norm": 0.5761995315551758, "learning_rate": 7.166e-05, "loss": 0.0566, "step": 10750 }, { "epoch": 6.857871255576801, "grad_norm": 0.781377911567688, "learning_rate": 7.172666666666668e-05, "loss": 0.0657, "step": 10760 }, { "epoch": 6.864244741873805, "grad_norm": 0.6733283996582031, "learning_rate": 7.179333333333333e-05, "loss": 0.0577, "step": 10770 }, { "epoch": 6.87061822817081, "grad_norm": 0.5615326762199402, "learning_rate": 7.186e-05, "loss": 0.0652, "step": 10780 }, { "epoch": 6.876991714467814, "grad_norm": 0.6769412159919739, "learning_rate": 7.192666666666667e-05, "loss": 0.0573, "step": 10790 }, { "epoch": 6.8833652007648185, "grad_norm": 0.7781015038490295, "learning_rate": 7.199333333333333e-05, "loss": 0.0753, "step": 10800 }, { "epoch": 6.889738687061823, "grad_norm": 0.7718249559402466, "learning_rate": 7.206000000000001e-05, "loss": 0.052, "step": 10810 }, { "epoch": 6.896112173358827, "grad_norm": 0.4445921778678894, "learning_rate": 7.212666666666666e-05, "loss": 0.0563, "step": 10820 }, { "epoch": 6.902485659655832, "grad_norm": 0.5905616283416748, "learning_rate": 7.219333333333334e-05, "loss": 0.0609, "step": 10830 }, { "epoch": 6.908859145952836, "grad_norm": 0.9250876903533936, "learning_rate": 7.226e-05, "loss": 0.0698, "step": 10840 }, { "epoch": 6.9152326322498405, "grad_norm": 0.6766232252120972, "learning_rate": 7.232666666666667e-05, "loss": 0.0522, "step": 10850 }, { "epoch": 6.921606118546845, "grad_norm": 0.8802348375320435, "learning_rate": 7.239333333333333e-05, "loss": 0.0596, "step": 10860 }, { "epoch": 6.927979604843849, "grad_norm": 0.48499956727027893, "learning_rate": 7.246000000000001e-05, "loss": 0.0647, "step": 10870 }, { "epoch": 6.934353091140854, "grad_norm": 0.6914782524108887, "learning_rate": 7.252666666666667e-05, "loss": 0.0642, "step": 10880 }, { "epoch": 6.940726577437858, "grad_norm": 1.1734741926193237, "learning_rate": 7.259333333333334e-05, "loss": 0.0668, "step": 10890 }, { "epoch": 6.947100063734863, "grad_norm": 0.9609723687171936, "learning_rate": 7.266000000000001e-05, "loss": 0.0726, "step": 10900 }, { "epoch": 6.953473550031868, "grad_norm": 0.6944030523300171, "learning_rate": 7.272666666666666e-05, "loss": 0.057, "step": 10910 }, { "epoch": 6.959847036328872, "grad_norm": 0.8145813941955566, "learning_rate": 7.279333333333334e-05, "loss": 0.0574, "step": 10920 }, { "epoch": 6.966220522625877, "grad_norm": 0.518361508846283, "learning_rate": 7.286e-05, "loss": 0.0637, "step": 10930 }, { "epoch": 6.972594008922881, "grad_norm": 0.7921425104141235, "learning_rate": 7.292666666666667e-05, "loss": 0.0622, "step": 10940 }, { "epoch": 6.9789674952198855, "grad_norm": 0.44566255807876587, "learning_rate": 7.299333333333333e-05, "loss": 0.0716, "step": 10950 }, { "epoch": 6.98534098151689, "grad_norm": 0.7134578824043274, "learning_rate": 7.306000000000001e-05, "loss": 0.0636, "step": 10960 }, { "epoch": 6.991714467813894, "grad_norm": 0.7634859681129456, "learning_rate": 7.312666666666666e-05, "loss": 0.0587, "step": 10970 }, { "epoch": 6.998087954110899, "grad_norm": 0.620833694934845, "learning_rate": 7.319333333333334e-05, "loss": 0.0572, "step": 10980 }, { "epoch": 7.004461440407903, "grad_norm": 0.8117725253105164, "learning_rate": 7.326e-05, "loss": 0.0549, "step": 10990 }, { "epoch": 7.0108349267049075, "grad_norm": 0.9135079383850098, "learning_rate": 7.332666666666666e-05, "loss": 0.0686, "step": 11000 }, { "epoch": 7.017208413001912, "grad_norm": 0.4884015917778015, "learning_rate": 7.339333333333334e-05, "loss": 0.057, "step": 11010 }, { "epoch": 7.023581899298916, "grad_norm": 1.0910640954971313, "learning_rate": 7.346e-05, "loss": 0.0727, "step": 11020 }, { "epoch": 7.029955385595921, "grad_norm": 0.8113101720809937, "learning_rate": 7.352666666666667e-05, "loss": 0.0687, "step": 11030 }, { "epoch": 7.036328871892925, "grad_norm": 0.9775698184967041, "learning_rate": 7.359333333333333e-05, "loss": 0.0618, "step": 11040 }, { "epoch": 7.0427023581899295, "grad_norm": 0.9888875484466553, "learning_rate": 7.366000000000001e-05, "loss": 0.0568, "step": 11050 }, { "epoch": 7.049075844486934, "grad_norm": 0.5883205533027649, "learning_rate": 7.372666666666666e-05, "loss": 0.0589, "step": 11060 }, { "epoch": 7.055449330783939, "grad_norm": 0.42911091446876526, "learning_rate": 7.379333333333334e-05, "loss": 0.06, "step": 11070 }, { "epoch": 7.061822817080944, "grad_norm": 0.4612443745136261, "learning_rate": 7.386e-05, "loss": 0.0504, "step": 11080 }, { "epoch": 7.068196303377948, "grad_norm": 0.8134729266166687, "learning_rate": 7.392666666666666e-05, "loss": 0.0673, "step": 11090 }, { "epoch": 7.074569789674952, "grad_norm": 0.68392014503479, "learning_rate": 7.399333333333334e-05, "loss": 0.0466, "step": 11100 }, { "epoch": 7.080943275971957, "grad_norm": 0.5725587606430054, "learning_rate": 7.406e-05, "loss": 0.0655, "step": 11110 }, { "epoch": 7.087316762268961, "grad_norm": 0.8250055313110352, "learning_rate": 7.412666666666667e-05, "loss": 0.0615, "step": 11120 }, { "epoch": 7.093690248565966, "grad_norm": 0.6771624684333801, "learning_rate": 7.419333333333333e-05, "loss": 0.0619, "step": 11130 }, { "epoch": 7.10006373486297, "grad_norm": 0.5988875031471252, "learning_rate": 7.426000000000001e-05, "loss": 0.0585, "step": 11140 }, { "epoch": 7.1064372211599744, "grad_norm": 0.4609868824481964, "learning_rate": 7.432666666666666e-05, "loss": 0.0344, "step": 11150 }, { "epoch": 7.112810707456979, "grad_norm": 0.5410454869270325, "learning_rate": 7.439333333333334e-05, "loss": 0.0512, "step": 11160 }, { "epoch": 7.119184193753983, "grad_norm": 0.7448944449424744, "learning_rate": 7.446e-05, "loss": 0.0532, "step": 11170 }, { "epoch": 7.125557680050988, "grad_norm": 0.7174357175827026, "learning_rate": 7.452666666666667e-05, "loss": 0.0531, "step": 11180 }, { "epoch": 7.131931166347992, "grad_norm": 0.43492981791496277, "learning_rate": 7.459333333333334e-05, "loss": 0.0562, "step": 11190 }, { "epoch": 7.1383046526449965, "grad_norm": 0.4520013928413391, "learning_rate": 7.466000000000001e-05, "loss": 0.0584, "step": 11200 }, { "epoch": 7.144678138942001, "grad_norm": 0.6327972412109375, "learning_rate": 7.472666666666667e-05, "loss": 0.0542, "step": 11210 }, { "epoch": 7.151051625239006, "grad_norm": 0.8993403911590576, "learning_rate": 7.479333333333333e-05, "loss": 0.0687, "step": 11220 }, { "epoch": 7.157425111536011, "grad_norm": 0.8367977738380432, "learning_rate": 7.486000000000001e-05, "loss": 0.0896, "step": 11230 }, { "epoch": 7.163798597833015, "grad_norm": 0.4823082685470581, "learning_rate": 7.492666666666666e-05, "loss": 0.0556, "step": 11240 }, { "epoch": 7.170172084130019, "grad_norm": 0.7253256440162659, "learning_rate": 7.499333333333334e-05, "loss": 0.0449, "step": 11250 }, { "epoch": 7.176545570427024, "grad_norm": 0.42491480708122253, "learning_rate": 7.506e-05, "loss": 0.0462, "step": 11260 }, { "epoch": 7.182919056724028, "grad_norm": 0.3749558627605438, "learning_rate": 7.512666666666667e-05, "loss": 0.0577, "step": 11270 }, { "epoch": 7.189292543021033, "grad_norm": 1.0233200788497925, "learning_rate": 7.519333333333334e-05, "loss": 0.0658, "step": 11280 }, { "epoch": 7.195666029318037, "grad_norm": 0.6975396275520325, "learning_rate": 7.526000000000001e-05, "loss": 0.0684, "step": 11290 }, { "epoch": 7.202039515615041, "grad_norm": 0.6800739169120789, "learning_rate": 7.532666666666667e-05, "loss": 0.0527, "step": 11300 }, { "epoch": 7.208413001912046, "grad_norm": 0.6258073449134827, "learning_rate": 7.539333333333334e-05, "loss": 0.0577, "step": 11310 }, { "epoch": 7.21478648820905, "grad_norm": 0.6886470317840576, "learning_rate": 7.546000000000001e-05, "loss": 0.0655, "step": 11320 }, { "epoch": 7.221159974506055, "grad_norm": 0.8964318037033081, "learning_rate": 7.552666666666666e-05, "loss": 0.0394, "step": 11330 }, { "epoch": 7.227533460803059, "grad_norm": 0.6888375282287598, "learning_rate": 7.559333333333334e-05, "loss": 0.0471, "step": 11340 }, { "epoch": 7.233906947100063, "grad_norm": 0.5329063534736633, "learning_rate": 7.566e-05, "loss": 0.0699, "step": 11350 }, { "epoch": 7.240280433397068, "grad_norm": 0.638798177242279, "learning_rate": 7.572666666666667e-05, "loss": 0.0492, "step": 11360 }, { "epoch": 7.246653919694072, "grad_norm": 0.6056517958641052, "learning_rate": 7.579333333333335e-05, "loss": 0.053, "step": 11370 }, { "epoch": 7.253027405991077, "grad_norm": 0.5125308632850647, "learning_rate": 7.586000000000001e-05, "loss": 0.0481, "step": 11380 }, { "epoch": 7.259400892288082, "grad_norm": 0.5904332399368286, "learning_rate": 7.592666666666667e-05, "loss": 0.0647, "step": 11390 }, { "epoch": 7.265774378585086, "grad_norm": 0.8789222240447998, "learning_rate": 7.599333333333334e-05, "loss": 0.0627, "step": 11400 }, { "epoch": 7.272147864882091, "grad_norm": 0.4663279950618744, "learning_rate": 7.606000000000001e-05, "loss": 0.0491, "step": 11410 }, { "epoch": 7.278521351179095, "grad_norm": 0.4175785779953003, "learning_rate": 7.612666666666666e-05, "loss": 0.0694, "step": 11420 }, { "epoch": 7.2848948374761, "grad_norm": 0.6583784222602844, "learning_rate": 7.619333333333334e-05, "loss": 0.0385, "step": 11430 }, { "epoch": 7.291268323773104, "grad_norm": 0.49252942204475403, "learning_rate": 7.625999999999999e-05, "loss": 0.056, "step": 11440 }, { "epoch": 7.297641810070108, "grad_norm": 0.6419488191604614, "learning_rate": 7.632666666666667e-05, "loss": 0.0538, "step": 11450 }, { "epoch": 7.304015296367113, "grad_norm": 0.4881172478199005, "learning_rate": 7.639333333333333e-05, "loss": 0.0673, "step": 11460 }, { "epoch": 7.310388782664117, "grad_norm": 0.6967611908912659, "learning_rate": 7.646e-05, "loss": 0.0476, "step": 11470 }, { "epoch": 7.316762268961122, "grad_norm": 0.9382917881011963, "learning_rate": 7.652666666666667e-05, "loss": 0.0533, "step": 11480 }, { "epoch": 7.323135755258126, "grad_norm": 0.5202028155326843, "learning_rate": 7.659333333333334e-05, "loss": 0.0597, "step": 11490 }, { "epoch": 7.32950924155513, "grad_norm": 0.6725322604179382, "learning_rate": 7.666e-05, "loss": 0.0612, "step": 11500 }, { "epoch": 7.335882727852135, "grad_norm": 0.9001889228820801, "learning_rate": 7.672666666666667e-05, "loss": 0.0528, "step": 11510 }, { "epoch": 7.342256214149139, "grad_norm": 1.0062668323516846, "learning_rate": 7.679333333333334e-05, "loss": 0.0594, "step": 11520 }, { "epoch": 7.3486297004461445, "grad_norm": 0.3967103660106659, "learning_rate": 7.685999999999999e-05, "loss": 0.0456, "step": 11530 }, { "epoch": 7.355003186743149, "grad_norm": 0.8786386847496033, "learning_rate": 7.692666666666667e-05, "loss": 0.0513, "step": 11540 }, { "epoch": 7.361376673040153, "grad_norm": 0.5655531883239746, "learning_rate": 7.699333333333333e-05, "loss": 0.0648, "step": 11550 }, { "epoch": 7.367750159337158, "grad_norm": 0.5367156863212585, "learning_rate": 7.706e-05, "loss": 0.0525, "step": 11560 }, { "epoch": 7.374123645634162, "grad_norm": 0.49696165323257446, "learning_rate": 7.712666666666667e-05, "loss": 0.0468, "step": 11570 }, { "epoch": 7.3804971319311665, "grad_norm": 0.6176717281341553, "learning_rate": 7.719333333333334e-05, "loss": 0.0599, "step": 11580 }, { "epoch": 7.386870618228171, "grad_norm": 0.5491601228713989, "learning_rate": 7.726e-05, "loss": 0.06, "step": 11590 }, { "epoch": 7.393244104525175, "grad_norm": 0.5185981392860413, "learning_rate": 7.732666666666667e-05, "loss": 0.0436, "step": 11600 }, { "epoch": 7.39961759082218, "grad_norm": 0.7929900884628296, "learning_rate": 7.739333333333334e-05, "loss": 0.0691, "step": 11610 }, { "epoch": 7.405991077119184, "grad_norm": 0.8049429655075073, "learning_rate": 7.746e-05, "loss": 0.0616, "step": 11620 }, { "epoch": 7.4123645634161885, "grad_norm": 1.1048583984375, "learning_rate": 7.752666666666667e-05, "loss": 0.0815, "step": 11630 }, { "epoch": 7.418738049713193, "grad_norm": 0.6788186430931091, "learning_rate": 7.759333333333333e-05, "loss": 0.0635, "step": 11640 }, { "epoch": 7.425111536010197, "grad_norm": 0.6070594787597656, "learning_rate": 7.766e-05, "loss": 0.0576, "step": 11650 }, { "epoch": 7.431485022307202, "grad_norm": 0.6177842617034912, "learning_rate": 7.772666666666668e-05, "loss": 0.063, "step": 11660 }, { "epoch": 7.437858508604206, "grad_norm": 0.502066969871521, "learning_rate": 7.779333333333334e-05, "loss": 0.0389, "step": 11670 }, { "epoch": 7.444231994901211, "grad_norm": 0.552257239818573, "learning_rate": 7.786e-05, "loss": 0.056, "step": 11680 }, { "epoch": 7.450605481198215, "grad_norm": 0.593451738357544, "learning_rate": 7.792666666666667e-05, "loss": 0.0525, "step": 11690 }, { "epoch": 7.45697896749522, "grad_norm": 0.437322199344635, "learning_rate": 7.799333333333334e-05, "loss": 0.0647, "step": 11700 }, { "epoch": 7.463352453792225, "grad_norm": 0.7175058722496033, "learning_rate": 7.806e-05, "loss": 0.0544, "step": 11710 }, { "epoch": 7.469725940089229, "grad_norm": 0.3645232617855072, "learning_rate": 7.812666666666667e-05, "loss": 0.0575, "step": 11720 }, { "epoch": 7.4760994263862335, "grad_norm": 0.9268646836280823, "learning_rate": 7.819333333333334e-05, "loss": 0.0551, "step": 11730 }, { "epoch": 7.482472912683238, "grad_norm": 0.8866416811943054, "learning_rate": 7.826e-05, "loss": 0.0542, "step": 11740 }, { "epoch": 7.488846398980242, "grad_norm": 0.7859415411949158, "learning_rate": 7.832666666666668e-05, "loss": 0.0612, "step": 11750 }, { "epoch": 7.495219885277247, "grad_norm": 0.6096539497375488, "learning_rate": 7.839333333333334e-05, "loss": 0.0526, "step": 11760 }, { "epoch": 7.501593371574251, "grad_norm": 0.6642547845840454, "learning_rate": 7.846e-05, "loss": 0.0473, "step": 11770 }, { "epoch": 7.5079668578712555, "grad_norm": 0.580796480178833, "learning_rate": 7.852666666666667e-05, "loss": 0.0554, "step": 11780 }, { "epoch": 7.51434034416826, "grad_norm": 0.5509164929389954, "learning_rate": 7.859333333333335e-05, "loss": 0.064, "step": 11790 }, { "epoch": 7.520713830465264, "grad_norm": 0.5541414618492126, "learning_rate": 7.866e-05, "loss": 0.0575, "step": 11800 }, { "epoch": 7.527087316762269, "grad_norm": 0.4667292535305023, "learning_rate": 7.872666666666667e-05, "loss": 0.0751, "step": 11810 }, { "epoch": 7.533460803059273, "grad_norm": 0.7596257328987122, "learning_rate": 7.879333333333334e-05, "loss": 0.0572, "step": 11820 }, { "epoch": 7.5398342893562775, "grad_norm": 0.8888062834739685, "learning_rate": 7.886e-05, "loss": 0.0741, "step": 11830 }, { "epoch": 7.546207775653283, "grad_norm": 1.5570415258407593, "learning_rate": 7.892666666666666e-05, "loss": 0.068, "step": 11840 }, { "epoch": 7.552581261950287, "grad_norm": 0.6390975117683411, "learning_rate": 7.899333333333334e-05, "loss": 0.0627, "step": 11850 }, { "epoch": 7.558954748247292, "grad_norm": 0.5417156219482422, "learning_rate": 7.906e-05, "loss": 0.0684, "step": 11860 }, { "epoch": 7.565328234544296, "grad_norm": 0.7917418479919434, "learning_rate": 7.912666666666667e-05, "loss": 0.0521, "step": 11870 }, { "epoch": 7.5717017208413, "grad_norm": 0.9175390601158142, "learning_rate": 7.919333333333335e-05, "loss": 0.094, "step": 11880 }, { "epoch": 7.578075207138305, "grad_norm": 0.6375300288200378, "learning_rate": 7.926e-05, "loss": 0.0583, "step": 11890 }, { "epoch": 7.584448693435309, "grad_norm": 0.7368698120117188, "learning_rate": 7.932666666666667e-05, "loss": 0.0567, "step": 11900 }, { "epoch": 7.590822179732314, "grad_norm": 0.42117226123809814, "learning_rate": 7.939333333333334e-05, "loss": 0.0657, "step": 11910 }, { "epoch": 7.597195666029318, "grad_norm": 0.7985228896141052, "learning_rate": 7.946e-05, "loss": 0.0551, "step": 11920 }, { "epoch": 7.6035691523263225, "grad_norm": 0.5908305048942566, "learning_rate": 7.952666666666667e-05, "loss": 0.0581, "step": 11930 }, { "epoch": 7.609942638623327, "grad_norm": 0.6016819477081299, "learning_rate": 7.959333333333334e-05, "loss": 0.0512, "step": 11940 }, { "epoch": 7.616316124920331, "grad_norm": 0.6850925087928772, "learning_rate": 7.966e-05, "loss": 0.0517, "step": 11950 }, { "epoch": 7.622689611217336, "grad_norm": 0.7231594324111938, "learning_rate": 7.972666666666667e-05, "loss": 0.0501, "step": 11960 }, { "epoch": 7.62906309751434, "grad_norm": 0.5239086747169495, "learning_rate": 7.979333333333335e-05, "loss": 0.0461, "step": 11970 }, { "epoch": 7.6354365838113445, "grad_norm": 0.6435546875, "learning_rate": 7.986e-05, "loss": 0.061, "step": 11980 }, { "epoch": 7.641810070108349, "grad_norm": 0.5770428776741028, "learning_rate": 7.992666666666668e-05, "loss": 0.0643, "step": 11990 }, { "epoch": 7.648183556405353, "grad_norm": 0.7032604813575745, "learning_rate": 7.999333333333334e-05, "loss": 0.0626, "step": 12000 }, { "epoch": 7.654557042702358, "grad_norm": 0.7884879112243652, "learning_rate": 8.006e-05, "loss": 0.0692, "step": 12010 }, { "epoch": 7.660930528999363, "grad_norm": 0.61533522605896, "learning_rate": 8.012666666666667e-05, "loss": 0.0492, "step": 12020 }, { "epoch": 7.667304015296367, "grad_norm": 0.41875240206718445, "learning_rate": 8.019333333333334e-05, "loss": 0.0464, "step": 12030 }, { "epoch": 7.673677501593372, "grad_norm": 0.27094653248786926, "learning_rate": 8.026000000000001e-05, "loss": 0.0438, "step": 12040 }, { "epoch": 7.680050987890376, "grad_norm": 0.724066436290741, "learning_rate": 8.032666666666667e-05, "loss": 0.0619, "step": 12050 }, { "epoch": 7.686424474187381, "grad_norm": 0.5979686975479126, "learning_rate": 8.039333333333335e-05, "loss": 0.0578, "step": 12060 }, { "epoch": 7.692797960484385, "grad_norm": 0.8723482489585876, "learning_rate": 8.046e-05, "loss": 0.0808, "step": 12070 }, { "epoch": 7.699171446781389, "grad_norm": 0.8083339929580688, "learning_rate": 8.052666666666668e-05, "loss": 0.0446, "step": 12080 }, { "epoch": 7.705544933078394, "grad_norm": 0.6310730576515198, "learning_rate": 8.059333333333333e-05, "loss": 0.0489, "step": 12090 }, { "epoch": 7.711918419375398, "grad_norm": 0.7891052961349487, "learning_rate": 8.066e-05, "loss": 0.0607, "step": 12100 }, { "epoch": 7.718291905672403, "grad_norm": 0.44173556566238403, "learning_rate": 8.072666666666667e-05, "loss": 0.0583, "step": 12110 }, { "epoch": 7.724665391969407, "grad_norm": 0.7001399397850037, "learning_rate": 8.079333333333333e-05, "loss": 0.0444, "step": 12120 }, { "epoch": 7.7310388782664115, "grad_norm": 0.7543017268180847, "learning_rate": 8.086000000000001e-05, "loss": 0.0674, "step": 12130 }, { "epoch": 7.737412364563416, "grad_norm": 0.5297566056251526, "learning_rate": 8.092666666666667e-05, "loss": 0.0475, "step": 12140 }, { "epoch": 7.743785850860421, "grad_norm": 0.4382438361644745, "learning_rate": 8.099333333333334e-05, "loss": 0.0507, "step": 12150 }, { "epoch": 7.7501593371574256, "grad_norm": 0.7441959977149963, "learning_rate": 8.106e-05, "loss": 0.0505, "step": 12160 }, { "epoch": 7.75653282345443, "grad_norm": 0.7295777201652527, "learning_rate": 8.112666666666668e-05, "loss": 0.0585, "step": 12170 }, { "epoch": 7.762906309751434, "grad_norm": 0.7317323088645935, "learning_rate": 8.119333333333333e-05, "loss": 0.0499, "step": 12180 }, { "epoch": 7.769279796048439, "grad_norm": 0.5636889934539795, "learning_rate": 8.126e-05, "loss": 0.0497, "step": 12190 }, { "epoch": 7.775653282345443, "grad_norm": 0.7215498685836792, "learning_rate": 8.132666666666667e-05, "loss": 0.073, "step": 12200 }, { "epoch": 7.782026768642448, "grad_norm": 0.8393809199333191, "learning_rate": 8.139333333333333e-05, "loss": 0.072, "step": 12210 }, { "epoch": 7.788400254939452, "grad_norm": 0.8304045796394348, "learning_rate": 8.146000000000001e-05, "loss": 0.0584, "step": 12220 }, { "epoch": 7.794773741236456, "grad_norm": 0.7566322684288025, "learning_rate": 8.152666666666667e-05, "loss": 0.0581, "step": 12230 }, { "epoch": 7.801147227533461, "grad_norm": 0.3760037124156952, "learning_rate": 8.159333333333334e-05, "loss": 0.0516, "step": 12240 }, { "epoch": 7.807520713830465, "grad_norm": 0.5697731971740723, "learning_rate": 8.166e-05, "loss": 0.0728, "step": 12250 }, { "epoch": 7.81389420012747, "grad_norm": 0.5997262001037598, "learning_rate": 8.172666666666668e-05, "loss": 0.0607, "step": 12260 }, { "epoch": 7.820267686424474, "grad_norm": 0.48271444439888, "learning_rate": 8.179333333333333e-05, "loss": 0.0587, "step": 12270 }, { "epoch": 7.826641172721478, "grad_norm": 0.4417996406555176, "learning_rate": 8.186e-05, "loss": 0.0498, "step": 12280 }, { "epoch": 7.833014659018483, "grad_norm": 0.34783264994621277, "learning_rate": 8.192666666666667e-05, "loss": 0.0531, "step": 12290 }, { "epoch": 7.839388145315487, "grad_norm": 0.7350356578826904, "learning_rate": 8.199333333333333e-05, "loss": 0.0543, "step": 12300 }, { "epoch": 7.845761631612492, "grad_norm": 0.878852367401123, "learning_rate": 8.206e-05, "loss": 0.0563, "step": 12310 }, { "epoch": 7.852135117909496, "grad_norm": 0.7118002772331238, "learning_rate": 8.212666666666667e-05, "loss": 0.06, "step": 12320 }, { "epoch": 7.858508604206501, "grad_norm": 0.4555775225162506, "learning_rate": 8.219333333333334e-05, "loss": 0.0556, "step": 12330 }, { "epoch": 7.864882090503506, "grad_norm": 0.711097002029419, "learning_rate": 8.226e-05, "loss": 0.0652, "step": 12340 }, { "epoch": 7.87125557680051, "grad_norm": 0.7447531223297119, "learning_rate": 8.232666666666668e-05, "loss": 0.0703, "step": 12350 }, { "epoch": 7.8776290630975145, "grad_norm": 0.48015764355659485, "learning_rate": 8.239333333333333e-05, "loss": 0.0651, "step": 12360 }, { "epoch": 7.884002549394519, "grad_norm": 0.8711662888526917, "learning_rate": 8.246e-05, "loss": 0.0475, "step": 12370 }, { "epoch": 7.890376035691523, "grad_norm": 1.0463588237762451, "learning_rate": 8.252666666666667e-05, "loss": 0.0508, "step": 12380 }, { "epoch": 7.896749521988528, "grad_norm": 0.67466801404953, "learning_rate": 8.259333333333333e-05, "loss": 0.0695, "step": 12390 }, { "epoch": 7.903123008285532, "grad_norm": 0.824320375919342, "learning_rate": 8.266e-05, "loss": 0.066, "step": 12400 }, { "epoch": 7.909496494582537, "grad_norm": 0.7048618197441101, "learning_rate": 8.272666666666668e-05, "loss": 0.0647, "step": 12410 }, { "epoch": 7.915869980879541, "grad_norm": 0.5582943558692932, "learning_rate": 8.279333333333334e-05, "loss": 0.0588, "step": 12420 }, { "epoch": 7.922243467176545, "grad_norm": 0.5646951794624329, "learning_rate": 8.286e-05, "loss": 0.0628, "step": 12430 }, { "epoch": 7.92861695347355, "grad_norm": 0.48201754689216614, "learning_rate": 8.292666666666668e-05, "loss": 0.0726, "step": 12440 }, { "epoch": 7.934990439770554, "grad_norm": 0.45752134919166565, "learning_rate": 8.299333333333333e-05, "loss": 0.0551, "step": 12450 }, { "epoch": 7.941363926067559, "grad_norm": 0.5571115612983704, "learning_rate": 8.306000000000001e-05, "loss": 0.0554, "step": 12460 }, { "epoch": 7.947737412364564, "grad_norm": 0.6592827439308167, "learning_rate": 8.312666666666667e-05, "loss": 0.0715, "step": 12470 }, { "epoch": 7.954110898661568, "grad_norm": 1.0221611261367798, "learning_rate": 8.319333333333334e-05, "loss": 0.0474, "step": 12480 }, { "epoch": 7.960484384958573, "grad_norm": 0.4267323911190033, "learning_rate": 8.326e-05, "loss": 0.0526, "step": 12490 }, { "epoch": 7.966857871255577, "grad_norm": 1.247265100479126, "learning_rate": 8.332666666666668e-05, "loss": 0.0757, "step": 12500 }, { "epoch": 7.9732313575525815, "grad_norm": 0.6217453479766846, "learning_rate": 8.339333333333334e-05, "loss": 0.0644, "step": 12510 }, { "epoch": 7.979604843849586, "grad_norm": 0.7512181401252747, "learning_rate": 8.346e-05, "loss": 0.0827, "step": 12520 }, { "epoch": 7.98597833014659, "grad_norm": 0.606593906879425, "learning_rate": 8.352666666666668e-05, "loss": 0.0625, "step": 12530 }, { "epoch": 7.992351816443595, "grad_norm": 0.5596364140510559, "learning_rate": 8.359333333333333e-05, "loss": 0.0633, "step": 12540 }, { "epoch": 7.998725302740599, "grad_norm": 0.6581247448921204, "learning_rate": 8.366000000000001e-05, "loss": 0.0799, "step": 12550 }, { "epoch": 8.005098789037604, "grad_norm": 0.7492228746414185, "learning_rate": 8.372666666666667e-05, "loss": 0.0564, "step": 12560 }, { "epoch": 8.011472275334608, "grad_norm": 0.8005379438400269, "learning_rate": 8.379333333333334e-05, "loss": 0.0585, "step": 12570 }, { "epoch": 8.017845761631612, "grad_norm": 0.7363327145576477, "learning_rate": 8.386e-05, "loss": 0.0621, "step": 12580 }, { "epoch": 8.024219247928617, "grad_norm": 0.5561438798904419, "learning_rate": 8.392666666666668e-05, "loss": 0.0647, "step": 12590 }, { "epoch": 8.030592734225621, "grad_norm": 1.0865379571914673, "learning_rate": 8.399333333333334e-05, "loss": 0.0646, "step": 12600 }, { "epoch": 8.036966220522626, "grad_norm": 0.8478971719741821, "learning_rate": 8.406e-05, "loss": 0.0562, "step": 12610 }, { "epoch": 8.04333970681963, "grad_norm": 0.874611496925354, "learning_rate": 8.412666666666668e-05, "loss": 0.0602, "step": 12620 }, { "epoch": 8.049713193116634, "grad_norm": 0.5931339263916016, "learning_rate": 8.419333333333333e-05, "loss": 0.0406, "step": 12630 }, { "epoch": 8.056086679413639, "grad_norm": 0.917545735836029, "learning_rate": 8.426000000000001e-05, "loss": 0.0618, "step": 12640 }, { "epoch": 8.062460165710643, "grad_norm": 0.6005682945251465, "learning_rate": 8.432666666666667e-05, "loss": 0.0652, "step": 12650 }, { "epoch": 8.068833652007648, "grad_norm": 0.6540262699127197, "learning_rate": 8.439333333333334e-05, "loss": 0.0546, "step": 12660 }, { "epoch": 8.075207138304652, "grad_norm": 0.8592332601547241, "learning_rate": 8.446e-05, "loss": 0.0631, "step": 12670 }, { "epoch": 8.081580624601656, "grad_norm": 0.7033776640892029, "learning_rate": 8.452666666666668e-05, "loss": 0.0495, "step": 12680 }, { "epoch": 8.08795411089866, "grad_norm": 0.6539597511291504, "learning_rate": 8.459333333333334e-05, "loss": 0.0661, "step": 12690 }, { "epoch": 8.094327597195665, "grad_norm": 0.7465230822563171, "learning_rate": 8.466e-05, "loss": 0.057, "step": 12700 }, { "epoch": 8.10070108349267, "grad_norm": 0.9498315453529358, "learning_rate": 8.472666666666667e-05, "loss": 0.0645, "step": 12710 }, { "epoch": 8.107074569789676, "grad_norm": 0.562738299369812, "learning_rate": 8.479333333333333e-05, "loss": 0.0557, "step": 12720 }, { "epoch": 8.11344805608668, "grad_norm": 0.6365389227867126, "learning_rate": 8.486000000000001e-05, "loss": 0.0452, "step": 12730 }, { "epoch": 8.119821542383685, "grad_norm": 0.7993060946464539, "learning_rate": 8.492666666666666e-05, "loss": 0.0633, "step": 12740 }, { "epoch": 8.126195028680689, "grad_norm": 0.7765477895736694, "learning_rate": 8.499333333333334e-05, "loss": 0.0632, "step": 12750 }, { "epoch": 8.132568514977693, "grad_norm": 0.8996695280075073, "learning_rate": 8.506e-05, "loss": 0.0668, "step": 12760 }, { "epoch": 8.138942001274698, "grad_norm": 0.6463532447814941, "learning_rate": 8.512666666666667e-05, "loss": 0.0758, "step": 12770 }, { "epoch": 8.145315487571702, "grad_norm": 0.7349516153335571, "learning_rate": 8.519333333333333e-05, "loss": 0.0669, "step": 12780 }, { "epoch": 8.151688973868707, "grad_norm": 0.6427165865898132, "learning_rate": 8.526000000000001e-05, "loss": 0.0617, "step": 12790 }, { "epoch": 8.158062460165711, "grad_norm": 0.742243230342865, "learning_rate": 8.532666666666667e-05, "loss": 0.0618, "step": 12800 }, { "epoch": 8.164435946462715, "grad_norm": 0.5856829285621643, "learning_rate": 8.539333333333333e-05, "loss": 0.0564, "step": 12810 }, { "epoch": 8.17080943275972, "grad_norm": 0.8810951709747314, "learning_rate": 8.546000000000001e-05, "loss": 0.0706, "step": 12820 }, { "epoch": 8.177182919056724, "grad_norm": 0.6175669431686401, "learning_rate": 8.552666666666666e-05, "loss": 0.0576, "step": 12830 }, { "epoch": 8.183556405353729, "grad_norm": 0.6240518093109131, "learning_rate": 8.559333333333334e-05, "loss": 0.0535, "step": 12840 }, { "epoch": 8.189929891650733, "grad_norm": 0.924400806427002, "learning_rate": 8.566e-05, "loss": 0.0549, "step": 12850 }, { "epoch": 8.196303377947737, "grad_norm": 0.7381389737129211, "learning_rate": 8.572666666666667e-05, "loss": 0.0648, "step": 12860 }, { "epoch": 8.202676864244742, "grad_norm": 0.6493406295776367, "learning_rate": 8.579333333333333e-05, "loss": 0.0478, "step": 12870 }, { "epoch": 8.209050350541746, "grad_norm": 0.49466875195503235, "learning_rate": 8.586000000000001e-05, "loss": 0.0479, "step": 12880 }, { "epoch": 8.21542383683875, "grad_norm": 0.5850510001182556, "learning_rate": 8.592666666666667e-05, "loss": 0.06, "step": 12890 }, { "epoch": 8.221797323135755, "grad_norm": 0.7878484725952148, "learning_rate": 8.599333333333334e-05, "loss": 0.074, "step": 12900 }, { "epoch": 8.22817080943276, "grad_norm": 0.8049789071083069, "learning_rate": 8.606000000000001e-05, "loss": 0.0603, "step": 12910 }, { "epoch": 8.234544295729764, "grad_norm": 0.7117138504981995, "learning_rate": 8.612666666666666e-05, "loss": 0.0422, "step": 12920 }, { "epoch": 8.240917782026768, "grad_norm": 1.0727053880691528, "learning_rate": 8.619333333333334e-05, "loss": 0.0529, "step": 12930 }, { "epoch": 8.247291268323773, "grad_norm": 0.576468825340271, "learning_rate": 8.626e-05, "loss": 0.0501, "step": 12940 }, { "epoch": 8.253664754620777, "grad_norm": 0.6107454299926758, "learning_rate": 8.632666666666667e-05, "loss": 0.0636, "step": 12950 }, { "epoch": 8.260038240917781, "grad_norm": 0.7168861627578735, "learning_rate": 8.639333333333333e-05, "loss": 0.0449, "step": 12960 }, { "epoch": 8.266411727214786, "grad_norm": 0.5779502391815186, "learning_rate": 8.646000000000001e-05, "loss": 0.0498, "step": 12970 }, { "epoch": 8.27278521351179, "grad_norm": 0.42147284746170044, "learning_rate": 8.652666666666667e-05, "loss": 0.0487, "step": 12980 }, { "epoch": 8.279158699808795, "grad_norm": 0.435723215341568, "learning_rate": 8.659333333333334e-05, "loss": 0.0493, "step": 12990 }, { "epoch": 8.2855321861058, "grad_norm": 0.8571350574493408, "learning_rate": 8.666000000000001e-05, "loss": 0.0562, "step": 13000 }, { "epoch": 8.291905672402804, "grad_norm": 0.9729090929031372, "learning_rate": 8.672666666666666e-05, "loss": 0.0666, "step": 13010 }, { "epoch": 8.29827915869981, "grad_norm": 0.5585812926292419, "learning_rate": 8.679333333333334e-05, "loss": 0.0569, "step": 13020 }, { "epoch": 8.304652644996814, "grad_norm": 0.9320511817932129, "learning_rate": 8.686e-05, "loss": 0.0748, "step": 13030 }, { "epoch": 8.311026131293819, "grad_norm": 0.5809386968612671, "learning_rate": 8.692666666666667e-05, "loss": 0.0504, "step": 13040 }, { "epoch": 8.317399617590823, "grad_norm": 0.6346476078033447, "learning_rate": 8.699333333333333e-05, "loss": 0.0596, "step": 13050 }, { "epoch": 8.323773103887827, "grad_norm": 0.6254171133041382, "learning_rate": 8.706000000000001e-05, "loss": 0.0483, "step": 13060 }, { "epoch": 8.330146590184832, "grad_norm": 0.6562458872795105, "learning_rate": 8.712666666666667e-05, "loss": 0.0645, "step": 13070 }, { "epoch": 8.336520076481836, "grad_norm": 0.6064593195915222, "learning_rate": 8.719333333333334e-05, "loss": 0.0661, "step": 13080 }, { "epoch": 8.34289356277884, "grad_norm": 1.0466936826705933, "learning_rate": 8.726000000000001e-05, "loss": 0.064, "step": 13090 }, { "epoch": 8.349267049075845, "grad_norm": 0.708739161491394, "learning_rate": 8.732666666666667e-05, "loss": 0.0827, "step": 13100 }, { "epoch": 8.35564053537285, "grad_norm": 0.4305594861507416, "learning_rate": 8.739333333333334e-05, "loss": 0.0573, "step": 13110 }, { "epoch": 8.362014021669854, "grad_norm": 0.581961989402771, "learning_rate": 8.746e-05, "loss": 0.0661, "step": 13120 }, { "epoch": 8.368387507966858, "grad_norm": 0.5669571161270142, "learning_rate": 8.752666666666667e-05, "loss": 0.0654, "step": 13130 }, { "epoch": 8.374760994263863, "grad_norm": 0.7762645483016968, "learning_rate": 8.759333333333333e-05, "loss": 0.0656, "step": 13140 }, { "epoch": 8.381134480560867, "grad_norm": 0.5275634527206421, "learning_rate": 8.766000000000001e-05, "loss": 0.0713, "step": 13150 }, { "epoch": 8.387507966857871, "grad_norm": 0.5054892897605896, "learning_rate": 8.772666666666666e-05, "loss": 0.0544, "step": 13160 }, { "epoch": 8.393881453154876, "grad_norm": 0.7584949731826782, "learning_rate": 8.779333333333334e-05, "loss": 0.0508, "step": 13170 }, { "epoch": 8.40025493945188, "grad_norm": 0.8117373585700989, "learning_rate": 8.786e-05, "loss": 0.065, "step": 13180 }, { "epoch": 8.406628425748885, "grad_norm": 0.5433506965637207, "learning_rate": 8.792666666666667e-05, "loss": 0.0513, "step": 13190 }, { "epoch": 8.413001912045889, "grad_norm": 0.6876084804534912, "learning_rate": 8.799333333333334e-05, "loss": 0.0525, "step": 13200 }, { "epoch": 8.419375398342893, "grad_norm": 0.40120765566825867, "learning_rate": 8.806000000000001e-05, "loss": 0.0678, "step": 13210 }, { "epoch": 8.425748884639898, "grad_norm": 0.7087835669517517, "learning_rate": 8.812666666666667e-05, "loss": 0.0603, "step": 13220 }, { "epoch": 8.432122370936902, "grad_norm": 0.6004530787467957, "learning_rate": 8.819333333333333e-05, "loss": 0.0559, "step": 13230 }, { "epoch": 8.438495857233907, "grad_norm": 0.4934057295322418, "learning_rate": 8.826000000000001e-05, "loss": 0.0649, "step": 13240 }, { "epoch": 8.444869343530911, "grad_norm": 0.9917402863502502, "learning_rate": 8.832666666666666e-05, "loss": 0.0628, "step": 13250 }, { "epoch": 8.451242829827915, "grad_norm": 0.7195178866386414, "learning_rate": 8.839333333333334e-05, "loss": 0.0511, "step": 13260 }, { "epoch": 8.45761631612492, "grad_norm": 0.8430331349372864, "learning_rate": 8.846e-05, "loss": 0.0555, "step": 13270 }, { "epoch": 8.463989802421924, "grad_norm": 0.42990657687187195, "learning_rate": 8.852666666666667e-05, "loss": 0.0472, "step": 13280 }, { "epoch": 8.470363288718929, "grad_norm": 0.6589018106460571, "learning_rate": 8.859333333333334e-05, "loss": 0.0627, "step": 13290 }, { "epoch": 8.476736775015933, "grad_norm": 0.5364127159118652, "learning_rate": 8.866000000000001e-05, "loss": 0.0499, "step": 13300 }, { "epoch": 8.483110261312937, "grad_norm": 0.8255322575569153, "learning_rate": 8.872666666666667e-05, "loss": 0.0555, "step": 13310 }, { "epoch": 8.489483747609942, "grad_norm": 0.7342777252197266, "learning_rate": 8.879333333333334e-05, "loss": 0.062, "step": 13320 }, { "epoch": 8.495857233906946, "grad_norm": 0.6865614652633667, "learning_rate": 8.886e-05, "loss": 0.0653, "step": 13330 }, { "epoch": 8.50223072020395, "grad_norm": 0.886336624622345, "learning_rate": 8.892666666666666e-05, "loss": 0.0651, "step": 13340 }, { "epoch": 8.508604206500957, "grad_norm": 0.8968199491500854, "learning_rate": 8.899333333333334e-05, "loss": 0.0542, "step": 13350 }, { "epoch": 8.514977692797961, "grad_norm": 0.6704058647155762, "learning_rate": 8.906e-05, "loss": 0.0604, "step": 13360 }, { "epoch": 8.521351179094966, "grad_norm": 0.47855129837989807, "learning_rate": 8.912666666666667e-05, "loss": 0.0649, "step": 13370 }, { "epoch": 8.52772466539197, "grad_norm": 0.5004100799560547, "learning_rate": 8.919333333333335e-05, "loss": 0.0473, "step": 13380 }, { "epoch": 8.534098151688974, "grad_norm": 0.5227582454681396, "learning_rate": 8.926e-05, "loss": 0.0782, "step": 13390 }, { "epoch": 8.540471637985979, "grad_norm": 0.6243178248405457, "learning_rate": 8.932666666666667e-05, "loss": 0.0713, "step": 13400 }, { "epoch": 8.546845124282983, "grad_norm": 0.6021842360496521, "learning_rate": 8.939333333333334e-05, "loss": 0.0601, "step": 13410 }, { "epoch": 8.553218610579988, "grad_norm": 0.8270822167396545, "learning_rate": 8.946e-05, "loss": 0.0643, "step": 13420 }, { "epoch": 8.559592096876992, "grad_norm": 0.7009769678115845, "learning_rate": 8.952666666666666e-05, "loss": 0.0477, "step": 13430 }, { "epoch": 8.565965583173996, "grad_norm": 0.7166087031364441, "learning_rate": 8.959333333333334e-05, "loss": 0.0633, "step": 13440 }, { "epoch": 8.572339069471, "grad_norm": 0.7389613389968872, "learning_rate": 8.966e-05, "loss": 0.0631, "step": 13450 }, { "epoch": 8.578712555768005, "grad_norm": 0.6592992544174194, "learning_rate": 8.972666666666667e-05, "loss": 0.077, "step": 13460 }, { "epoch": 8.58508604206501, "grad_norm": 0.5244159698486328, "learning_rate": 8.979333333333335e-05, "loss": 0.0703, "step": 13470 }, { "epoch": 8.591459528362014, "grad_norm": 0.8833013772964478, "learning_rate": 8.986e-05, "loss": 0.0618, "step": 13480 }, { "epoch": 8.597833014659019, "grad_norm": 0.38170257210731506, "learning_rate": 8.992666666666667e-05, "loss": 0.0694, "step": 13490 }, { "epoch": 8.604206500956023, "grad_norm": 0.8855859041213989, "learning_rate": 8.999333333333334e-05, "loss": 0.0506, "step": 13500 }, { "epoch": 8.610579987253027, "grad_norm": 0.6504349112510681, "learning_rate": 9.006e-05, "loss": 0.0692, "step": 13510 }, { "epoch": 8.616953473550032, "grad_norm": 0.6342761516571045, "learning_rate": 9.012666666666667e-05, "loss": 0.083, "step": 13520 }, { "epoch": 8.623326959847036, "grad_norm": 0.7263931632041931, "learning_rate": 9.019333333333334e-05, "loss": 0.0715, "step": 13530 }, { "epoch": 8.62970044614404, "grad_norm": 0.625174880027771, "learning_rate": 9.026e-05, "loss": 0.0668, "step": 13540 }, { "epoch": 8.636073932441045, "grad_norm": 0.6043142080307007, "learning_rate": 9.032666666666667e-05, "loss": 0.0574, "step": 13550 }, { "epoch": 8.64244741873805, "grad_norm": 1.0910696983337402, "learning_rate": 9.039333333333335e-05, "loss": 0.0701, "step": 13560 }, { "epoch": 8.648820905035054, "grad_norm": 0.7509660720825195, "learning_rate": 9.046e-05, "loss": 0.0426, "step": 13570 }, { "epoch": 8.655194391332058, "grad_norm": 0.7571632862091064, "learning_rate": 9.052666666666668e-05, "loss": 0.0551, "step": 13580 }, { "epoch": 8.661567877629063, "grad_norm": 0.9736087918281555, "learning_rate": 9.059333333333334e-05, "loss": 0.0666, "step": 13590 }, { "epoch": 8.667941363926067, "grad_norm": 0.7005413770675659, "learning_rate": 9.066e-05, "loss": 0.0648, "step": 13600 }, { "epoch": 8.674314850223071, "grad_norm": 0.64781254529953, "learning_rate": 9.072666666666667e-05, "loss": 0.0552, "step": 13610 }, { "epoch": 8.680688336520076, "grad_norm": 0.6549044847488403, "learning_rate": 9.079333333333334e-05, "loss": 0.0625, "step": 13620 }, { "epoch": 8.68706182281708, "grad_norm": 0.8877256512641907, "learning_rate": 9.086e-05, "loss": 0.0557, "step": 13630 }, { "epoch": 8.693435309114085, "grad_norm": 0.7637243270874023, "learning_rate": 9.092666666666667e-05, "loss": 0.0504, "step": 13640 }, { "epoch": 8.69980879541109, "grad_norm": 0.7398011088371277, "learning_rate": 9.099333333333334e-05, "loss": 0.0626, "step": 13650 }, { "epoch": 8.706182281708095, "grad_norm": 0.4594363272190094, "learning_rate": 9.106e-05, "loss": 0.0644, "step": 13660 }, { "epoch": 8.7125557680051, "grad_norm": 0.5682175755500793, "learning_rate": 9.112666666666668e-05, "loss": 0.077, "step": 13670 }, { "epoch": 8.718929254302104, "grad_norm": 0.4889727830886841, "learning_rate": 9.119333333333334e-05, "loss": 0.0686, "step": 13680 }, { "epoch": 8.725302740599108, "grad_norm": 0.7815690040588379, "learning_rate": 9.126e-05, "loss": 0.0561, "step": 13690 }, { "epoch": 8.731676226896113, "grad_norm": 0.756198525428772, "learning_rate": 9.132666666666667e-05, "loss": 0.0658, "step": 13700 }, { "epoch": 8.738049713193117, "grad_norm": 0.4668671190738678, "learning_rate": 9.139333333333334e-05, "loss": 0.0552, "step": 13710 }, { "epoch": 8.744423199490122, "grad_norm": 0.7657446265220642, "learning_rate": 9.146e-05, "loss": 0.056, "step": 13720 }, { "epoch": 8.750796685787126, "grad_norm": 0.5875611305236816, "learning_rate": 9.152666666666667e-05, "loss": 0.049, "step": 13730 }, { "epoch": 8.75717017208413, "grad_norm": 0.5725999474525452, "learning_rate": 9.159333333333334e-05, "loss": 0.0641, "step": 13740 }, { "epoch": 8.763543658381135, "grad_norm": 0.5676485300064087, "learning_rate": 9.166e-05, "loss": 0.0556, "step": 13750 }, { "epoch": 8.76991714467814, "grad_norm": 0.6588584780693054, "learning_rate": 9.172666666666668e-05, "loss": 0.0559, "step": 13760 }, { "epoch": 8.776290630975144, "grad_norm": 0.7619849443435669, "learning_rate": 9.179333333333334e-05, "loss": 0.0781, "step": 13770 }, { "epoch": 8.782664117272148, "grad_norm": 0.5273615717887878, "learning_rate": 9.186e-05, "loss": 0.0504, "step": 13780 }, { "epoch": 8.789037603569152, "grad_norm": 0.6352676153182983, "learning_rate": 9.192666666666667e-05, "loss": 0.0545, "step": 13790 }, { "epoch": 8.795411089866157, "grad_norm": 0.5762425661087036, "learning_rate": 9.199333333333335e-05, "loss": 0.0541, "step": 13800 }, { "epoch": 8.801784576163161, "grad_norm": 0.6435329914093018, "learning_rate": 9.206e-05, "loss": 0.0551, "step": 13810 }, { "epoch": 8.808158062460166, "grad_norm": 0.6709913015365601, "learning_rate": 9.212666666666667e-05, "loss": 0.0658, "step": 13820 }, { "epoch": 8.81453154875717, "grad_norm": 0.794775664806366, "learning_rate": 9.219333333333334e-05, "loss": 0.0687, "step": 13830 }, { "epoch": 8.820905035054174, "grad_norm": 0.5512056946754456, "learning_rate": 9.226e-05, "loss": 0.0581, "step": 13840 }, { "epoch": 8.827278521351179, "grad_norm": 0.4298240840435028, "learning_rate": 9.232666666666668e-05, "loss": 0.0494, "step": 13850 }, { "epoch": 8.833652007648183, "grad_norm": 0.425416499376297, "learning_rate": 9.239333333333334e-05, "loss": 0.065, "step": 13860 }, { "epoch": 8.840025493945188, "grad_norm": 0.5146231651306152, "learning_rate": 9.246e-05, "loss": 0.0577, "step": 13870 }, { "epoch": 8.846398980242192, "grad_norm": 0.5774735808372498, "learning_rate": 9.252666666666667e-05, "loss": 0.0407, "step": 13880 }, { "epoch": 8.852772466539196, "grad_norm": 0.580212414264679, "learning_rate": 9.259333333333335e-05, "loss": 0.0366, "step": 13890 }, { "epoch": 8.8591459528362, "grad_norm": 0.46737149357795715, "learning_rate": 9.266e-05, "loss": 0.0604, "step": 13900 }, { "epoch": 8.865519439133205, "grad_norm": 0.6332247853279114, "learning_rate": 9.272666666666667e-05, "loss": 0.0602, "step": 13910 }, { "epoch": 8.87189292543021, "grad_norm": 0.38513123989105225, "learning_rate": 9.279333333333334e-05, "loss": 0.0618, "step": 13920 }, { "epoch": 8.878266411727214, "grad_norm": 0.7670353055000305, "learning_rate": 9.286e-05, "loss": 0.0665, "step": 13930 }, { "epoch": 8.884639898024218, "grad_norm": 0.5132635831832886, "learning_rate": 9.292666666666668e-05, "loss": 0.058, "step": 13940 }, { "epoch": 8.891013384321223, "grad_norm": 0.5709401369094849, "learning_rate": 9.299333333333333e-05, "loss": 0.0545, "step": 13950 }, { "epoch": 8.897386870618227, "grad_norm": 0.469900518655777, "learning_rate": 9.306000000000001e-05, "loss": 0.0579, "step": 13960 }, { "epoch": 8.903760356915232, "grad_norm": 0.3554374873638153, "learning_rate": 9.312666666666667e-05, "loss": 0.076, "step": 13970 }, { "epoch": 8.910133843212238, "grad_norm": 0.5653266310691833, "learning_rate": 9.319333333333333e-05, "loss": 0.0668, "step": 13980 }, { "epoch": 8.916507329509242, "grad_norm": 0.9526334404945374, "learning_rate": 9.326e-05, "loss": 0.0744, "step": 13990 }, { "epoch": 8.922880815806247, "grad_norm": 0.5525512099266052, "learning_rate": 9.332666666666668e-05, "loss": 0.0585, "step": 14000 }, { "epoch": 8.929254302103251, "grad_norm": 0.5350542068481445, "learning_rate": 9.339333333333334e-05, "loss": 0.0504, "step": 14010 }, { "epoch": 8.935627788400256, "grad_norm": 0.6447840929031372, "learning_rate": 9.346e-05, "loss": 0.069, "step": 14020 }, { "epoch": 8.94200127469726, "grad_norm": 0.6802664995193481, "learning_rate": 9.352666666666667e-05, "loss": 0.0651, "step": 14030 }, { "epoch": 8.948374760994264, "grad_norm": 0.7058776617050171, "learning_rate": 9.359333333333333e-05, "loss": 0.0524, "step": 14040 }, { "epoch": 8.954748247291269, "grad_norm": 0.9878630638122559, "learning_rate": 9.366000000000001e-05, "loss": 0.052, "step": 14050 }, { "epoch": 8.961121733588273, "grad_norm": 0.7848569750785828, "learning_rate": 9.372666666666667e-05, "loss": 0.0616, "step": 14060 }, { "epoch": 8.967495219885278, "grad_norm": 0.5310623049736023, "learning_rate": 9.379333333333334e-05, "loss": 0.0505, "step": 14070 }, { "epoch": 8.973868706182282, "grad_norm": 0.4780806303024292, "learning_rate": 9.386e-05, "loss": 0.0493, "step": 14080 }, { "epoch": 8.980242192479286, "grad_norm": 0.5557371377944946, "learning_rate": 9.392666666666668e-05, "loss": 0.0494, "step": 14090 }, { "epoch": 8.98661567877629, "grad_norm": 0.9081065058708191, "learning_rate": 9.399333333333333e-05, "loss": 0.059, "step": 14100 }, { "epoch": 8.992989165073295, "grad_norm": 1.0077979564666748, "learning_rate": 9.406e-05, "loss": 0.0605, "step": 14110 }, { "epoch": 8.9993626513703, "grad_norm": 0.7335388660430908, "learning_rate": 9.412666666666667e-05, "loss": 0.0495, "step": 14120 }, { "epoch": 9.005736137667304, "grad_norm": 0.7435501217842102, "learning_rate": 9.419333333333333e-05, "loss": 0.055, "step": 14130 }, { "epoch": 9.012109623964308, "grad_norm": 0.6691561937332153, "learning_rate": 9.426000000000001e-05, "loss": 0.072, "step": 14140 }, { "epoch": 9.018483110261313, "grad_norm": 0.4774843156337738, "learning_rate": 9.432666666666667e-05, "loss": 0.041, "step": 14150 }, { "epoch": 9.024856596558317, "grad_norm": 0.8308414220809937, "learning_rate": 9.439333333333334e-05, "loss": 0.0728, "step": 14160 }, { "epoch": 9.031230082855322, "grad_norm": 0.8462230563163757, "learning_rate": 9.446e-05, "loss": 0.0631, "step": 14170 }, { "epoch": 9.037603569152326, "grad_norm": 0.435079962015152, "learning_rate": 9.452666666666668e-05, "loss": 0.0539, "step": 14180 }, { "epoch": 9.04397705544933, "grad_norm": 0.46461808681488037, "learning_rate": 9.459333333333333e-05, "loss": 0.0646, "step": 14190 }, { "epoch": 9.050350541746335, "grad_norm": 0.6792469620704651, "learning_rate": 9.466e-05, "loss": 0.0594, "step": 14200 }, { "epoch": 9.05672402804334, "grad_norm": 0.6812877058982849, "learning_rate": 9.472666666666667e-05, "loss": 0.0655, "step": 14210 }, { "epoch": 9.063097514340344, "grad_norm": 0.5176808834075928, "learning_rate": 9.479333333333333e-05, "loss": 0.0489, "step": 14220 }, { "epoch": 9.069471000637348, "grad_norm": 0.5170048475265503, "learning_rate": 9.486000000000001e-05, "loss": 0.0583, "step": 14230 }, { "epoch": 9.075844486934352, "grad_norm": 0.5931413769721985, "learning_rate": 9.492666666666667e-05, "loss": 0.0473, "step": 14240 }, { "epoch": 9.082217973231357, "grad_norm": 0.7961833477020264, "learning_rate": 9.499333333333334e-05, "loss": 0.0671, "step": 14250 }, { "epoch": 9.088591459528361, "grad_norm": 0.5670689940452576, "learning_rate": 9.506e-05, "loss": 0.0483, "step": 14260 }, { "epoch": 9.094964945825366, "grad_norm": 0.598980724811554, "learning_rate": 9.512666666666668e-05, "loss": 0.0544, "step": 14270 }, { "epoch": 9.101338432122372, "grad_norm": 0.6023126244544983, "learning_rate": 9.519333333333333e-05, "loss": 0.0638, "step": 14280 }, { "epoch": 9.107711918419376, "grad_norm": 0.6292657256126404, "learning_rate": 9.526e-05, "loss": 0.0595, "step": 14290 }, { "epoch": 9.11408540471638, "grad_norm": 0.715130627155304, "learning_rate": 9.532666666666667e-05, "loss": 0.054, "step": 14300 }, { "epoch": 9.120458891013385, "grad_norm": 0.6806856989860535, "learning_rate": 9.539333333333333e-05, "loss": 0.0517, "step": 14310 }, { "epoch": 9.12683237731039, "grad_norm": 0.505490243434906, "learning_rate": 9.546000000000001e-05, "loss": 0.071, "step": 14320 }, { "epoch": 9.133205863607394, "grad_norm": 0.6481647491455078, "learning_rate": 9.552666666666667e-05, "loss": 0.0511, "step": 14330 }, { "epoch": 9.139579349904398, "grad_norm": 0.41041502356529236, "learning_rate": 9.559333333333334e-05, "loss": 0.0504, "step": 14340 }, { "epoch": 9.145952836201403, "grad_norm": 0.4806886315345764, "learning_rate": 9.566e-05, "loss": 0.0588, "step": 14350 }, { "epoch": 9.152326322498407, "grad_norm": 1.0263683795928955, "learning_rate": 9.572666666666668e-05, "loss": 0.0579, "step": 14360 }, { "epoch": 9.158699808795411, "grad_norm": 0.3896993398666382, "learning_rate": 9.579333333333333e-05, "loss": 0.0475, "step": 14370 }, { "epoch": 9.165073295092416, "grad_norm": 0.6258089542388916, "learning_rate": 9.586000000000001e-05, "loss": 0.0552, "step": 14380 }, { "epoch": 9.17144678138942, "grad_norm": 0.675748884677887, "learning_rate": 9.592666666666667e-05, "loss": 0.0624, "step": 14390 }, { "epoch": 9.177820267686425, "grad_norm": 0.8424323201179504, "learning_rate": 9.599333333333333e-05, "loss": 0.0698, "step": 14400 }, { "epoch": 9.184193753983429, "grad_norm": 0.703313946723938, "learning_rate": 9.606000000000001e-05, "loss": 0.0445, "step": 14410 }, { "epoch": 9.190567240280433, "grad_norm": 0.6200817227363586, "learning_rate": 9.612666666666668e-05, "loss": 0.0578, "step": 14420 }, { "epoch": 9.196940726577438, "grad_norm": 0.3376152217388153, "learning_rate": 9.619333333333334e-05, "loss": 0.055, "step": 14430 }, { "epoch": 9.203314212874442, "grad_norm": 0.4722602665424347, "learning_rate": 9.626e-05, "loss": 0.0646, "step": 14440 }, { "epoch": 9.209687699171447, "grad_norm": 0.7795982360839844, "learning_rate": 9.632666666666668e-05, "loss": 0.0671, "step": 14450 }, { "epoch": 9.216061185468451, "grad_norm": 0.8860408663749695, "learning_rate": 9.639333333333333e-05, "loss": 0.057, "step": 14460 }, { "epoch": 9.222434671765456, "grad_norm": 0.9807704091072083, "learning_rate": 9.646000000000001e-05, "loss": 0.0587, "step": 14470 }, { "epoch": 9.22880815806246, "grad_norm": 0.4594630300998688, "learning_rate": 9.652666666666667e-05, "loss": 0.0416, "step": 14480 }, { "epoch": 9.235181644359464, "grad_norm": 0.6166505813598633, "learning_rate": 9.659333333333334e-05, "loss": 0.0478, "step": 14490 }, { "epoch": 9.241555130656469, "grad_norm": 0.45989012718200684, "learning_rate": 9.666e-05, "loss": 0.0597, "step": 14500 }, { "epoch": 9.247928616953473, "grad_norm": 0.7244271636009216, "learning_rate": 9.672666666666668e-05, "loss": 0.0591, "step": 14510 }, { "epoch": 9.254302103250478, "grad_norm": 0.635843813419342, "learning_rate": 9.679333333333334e-05, "loss": 0.0509, "step": 14520 }, { "epoch": 9.260675589547482, "grad_norm": 0.6153807640075684, "learning_rate": 9.686e-05, "loss": 0.0752, "step": 14530 }, { "epoch": 9.267049075844486, "grad_norm": 0.6309207081794739, "learning_rate": 9.692666666666668e-05, "loss": 0.0523, "step": 14540 }, { "epoch": 9.27342256214149, "grad_norm": 0.6113809943199158, "learning_rate": 9.699333333333333e-05, "loss": 0.0533, "step": 14550 }, { "epoch": 9.279796048438495, "grad_norm": 0.7630850672721863, "learning_rate": 9.706000000000001e-05, "loss": 0.0497, "step": 14560 }, { "epoch": 9.2861695347355, "grad_norm": 0.7670928835868835, "learning_rate": 9.712666666666666e-05, "loss": 0.0447, "step": 14570 }, { "epoch": 9.292543021032504, "grad_norm": 0.5984125137329102, "learning_rate": 9.719333333333334e-05, "loss": 0.0516, "step": 14580 }, { "epoch": 9.298916507329508, "grad_norm": 0.5009859800338745, "learning_rate": 9.726e-05, "loss": 0.0647, "step": 14590 }, { "epoch": 9.305289993626515, "grad_norm": 1.0185586214065552, "learning_rate": 9.732666666666666e-05, "loss": 0.0597, "step": 14600 }, { "epoch": 9.311663479923519, "grad_norm": 0.6494755744934082, "learning_rate": 9.739333333333334e-05, "loss": 0.0504, "step": 14610 }, { "epoch": 9.318036966220523, "grad_norm": 0.7207589149475098, "learning_rate": 9.746e-05, "loss": 0.0532, "step": 14620 }, { "epoch": 9.324410452517528, "grad_norm": 0.5885887145996094, "learning_rate": 9.752666666666667e-05, "loss": 0.0533, "step": 14630 }, { "epoch": 9.330783938814532, "grad_norm": 0.9548124670982361, "learning_rate": 9.759333333333333e-05, "loss": 0.0541, "step": 14640 }, { "epoch": 9.337157425111537, "grad_norm": 0.6898685097694397, "learning_rate": 9.766000000000001e-05, "loss": 0.048, "step": 14650 }, { "epoch": 9.343530911408541, "grad_norm": 0.8245808482170105, "learning_rate": 9.772666666666666e-05, "loss": 0.0583, "step": 14660 }, { "epoch": 9.349904397705545, "grad_norm": 0.6053099632263184, "learning_rate": 9.779333333333334e-05, "loss": 0.066, "step": 14670 }, { "epoch": 9.35627788400255, "grad_norm": 0.6548680067062378, "learning_rate": 9.786e-05, "loss": 0.0485, "step": 14680 }, { "epoch": 9.362651370299554, "grad_norm": 0.6803520917892456, "learning_rate": 9.792666666666667e-05, "loss": 0.0548, "step": 14690 }, { "epoch": 9.369024856596559, "grad_norm": 0.6984466314315796, "learning_rate": 9.799333333333334e-05, "loss": 0.0462, "step": 14700 }, { "epoch": 9.375398342893563, "grad_norm": 0.5420809984207153, "learning_rate": 9.806e-05, "loss": 0.0531, "step": 14710 }, { "epoch": 9.381771829190567, "grad_norm": 0.70583575963974, "learning_rate": 9.812666666666667e-05, "loss": 0.0716, "step": 14720 }, { "epoch": 9.388145315487572, "grad_norm": 0.9024677872657776, "learning_rate": 9.819333333333333e-05, "loss": 0.0508, "step": 14730 }, { "epoch": 9.394518801784576, "grad_norm": 0.9504936337471008, "learning_rate": 9.826000000000001e-05, "loss": 0.0914, "step": 14740 }, { "epoch": 9.40089228808158, "grad_norm": 0.5353940725326538, "learning_rate": 9.832666666666666e-05, "loss": 0.0551, "step": 14750 }, { "epoch": 9.407265774378585, "grad_norm": 0.9119231700897217, "learning_rate": 9.839333333333334e-05, "loss": 0.0744, "step": 14760 }, { "epoch": 9.41363926067559, "grad_norm": 0.8097683787345886, "learning_rate": 9.846e-05, "loss": 0.0662, "step": 14770 }, { "epoch": 9.420012746972594, "grad_norm": 0.7556730508804321, "learning_rate": 9.852666666666667e-05, "loss": 0.0588, "step": 14780 }, { "epoch": 9.426386233269598, "grad_norm": 0.5082266330718994, "learning_rate": 9.859333333333334e-05, "loss": 0.0523, "step": 14790 }, { "epoch": 9.432759719566603, "grad_norm": 0.4727637767791748, "learning_rate": 9.866000000000001e-05, "loss": 0.0565, "step": 14800 }, { "epoch": 9.439133205863607, "grad_norm": 0.7173406481742859, "learning_rate": 9.872666666666667e-05, "loss": 0.0618, "step": 14810 }, { "epoch": 9.445506692160611, "grad_norm": 0.6705032587051392, "learning_rate": 9.879333333333333e-05, "loss": 0.0496, "step": 14820 }, { "epoch": 9.451880178457616, "grad_norm": 0.5275325179100037, "learning_rate": 9.886000000000001e-05, "loss": 0.0737, "step": 14830 }, { "epoch": 9.45825366475462, "grad_norm": 0.8255791068077087, "learning_rate": 9.892666666666666e-05, "loss": 0.0759, "step": 14840 }, { "epoch": 9.464627151051625, "grad_norm": 0.8190925717353821, "learning_rate": 9.899333333333334e-05, "loss": 0.0459, "step": 14850 }, { "epoch": 9.471000637348629, "grad_norm": 0.967989444732666, "learning_rate": 9.906e-05, "loss": 0.0828, "step": 14860 }, { "epoch": 9.477374123645633, "grad_norm": 0.6458868384361267, "learning_rate": 9.912666666666667e-05, "loss": 0.0643, "step": 14870 }, { "epoch": 9.483747609942638, "grad_norm": 0.577491283416748, "learning_rate": 9.919333333333334e-05, "loss": 0.056, "step": 14880 }, { "epoch": 9.490121096239642, "grad_norm": 0.48064032196998596, "learning_rate": 9.926000000000001e-05, "loss": 0.0672, "step": 14890 }, { "epoch": 9.496494582536647, "grad_norm": 0.7018259763717651, "learning_rate": 9.932666666666667e-05, "loss": 0.0638, "step": 14900 }, { "epoch": 9.502868068833653, "grad_norm": 0.5664310455322266, "learning_rate": 9.939333333333334e-05, "loss": 0.0641, "step": 14910 }, { "epoch": 9.509241555130657, "grad_norm": 0.786356508731842, "learning_rate": 9.946000000000001e-05, "loss": 0.0559, "step": 14920 }, { "epoch": 9.515615041427662, "grad_norm": 0.5695542097091675, "learning_rate": 9.952666666666666e-05, "loss": 0.0567, "step": 14930 }, { "epoch": 9.521988527724666, "grad_norm": 0.7495202422142029, "learning_rate": 9.959333333333334e-05, "loss": 0.0528, "step": 14940 }, { "epoch": 9.52836201402167, "grad_norm": 0.7040093541145325, "learning_rate": 9.966e-05, "loss": 0.0584, "step": 14950 }, { "epoch": 9.534735500318675, "grad_norm": 0.6281925439834595, "learning_rate": 9.972666666666667e-05, "loss": 0.0637, "step": 14960 }, { "epoch": 9.54110898661568, "grad_norm": 0.5572054386138916, "learning_rate": 9.979333333333333e-05, "loss": 0.0604, "step": 14970 }, { "epoch": 9.547482472912684, "grad_norm": 0.7755315899848938, "learning_rate": 9.986000000000001e-05, "loss": 0.0568, "step": 14980 }, { "epoch": 9.553855959209688, "grad_norm": 0.4848751723766327, "learning_rate": 9.992666666666667e-05, "loss": 0.0624, "step": 14990 }, { "epoch": 9.560229445506693, "grad_norm": 0.6210675835609436, "learning_rate": 9.999333333333334e-05, "loss": 0.0678, "step": 15000 }, { "epoch": 9.566602931803697, "grad_norm": 0.42524453997612, "learning_rate": 9.999999975394338e-05, "loss": 0.0494, "step": 15010 }, { "epoch": 9.572976418100701, "grad_norm": 0.36582857370376587, "learning_rate": 9.999999890337731e-05, "loss": 0.0519, "step": 15020 }, { "epoch": 9.579349904397706, "grad_norm": 0.47146666049957275, "learning_rate": 9.999999744526401e-05, "loss": 0.0613, "step": 15030 }, { "epoch": 9.58572339069471, "grad_norm": 0.5648419857025146, "learning_rate": 9.999999537960356e-05, "loss": 0.0738, "step": 15040 }, { "epoch": 9.592096876991715, "grad_norm": 0.6596403121948242, "learning_rate": 9.999999270639596e-05, "loss": 0.0512, "step": 15050 }, { "epoch": 9.598470363288719, "grad_norm": 0.6999940276145935, "learning_rate": 9.999998942564122e-05, "loss": 0.0812, "step": 15060 }, { "epoch": 9.604843849585723, "grad_norm": 1.080168604850769, "learning_rate": 9.99999855373394e-05, "loss": 0.0596, "step": 15070 }, { "epoch": 9.611217335882728, "grad_norm": 0.8175768256187439, "learning_rate": 9.999998104149056e-05, "loss": 0.0555, "step": 15080 }, { "epoch": 9.617590822179732, "grad_norm": 0.5036365985870361, "learning_rate": 9.999997593809473e-05, "loss": 0.0606, "step": 15090 }, { "epoch": 9.623964308476737, "grad_norm": 0.7283940315246582, "learning_rate": 9.999997022715201e-05, "loss": 0.0679, "step": 15100 }, { "epoch": 9.630337794773741, "grad_norm": 0.4878939390182495, "learning_rate": 9.999996390866243e-05, "loss": 0.0527, "step": 15110 }, { "epoch": 9.636711281070745, "grad_norm": 0.5333446860313416, "learning_rate": 9.999995698262608e-05, "loss": 0.0503, "step": 15120 }, { "epoch": 9.64308476736775, "grad_norm": 0.5796135067939758, "learning_rate": 9.999994944904306e-05, "loss": 0.0565, "step": 15130 }, { "epoch": 9.649458253664754, "grad_norm": 0.6852944493293762, "learning_rate": 9.999994130791341e-05, "loss": 0.0422, "step": 15140 }, { "epoch": 9.655831739961759, "grad_norm": 0.8792035579681396, "learning_rate": 9.99999325592373e-05, "loss": 0.0655, "step": 15150 }, { "epoch": 9.662205226258763, "grad_norm": 0.5844212770462036, "learning_rate": 9.999992320301478e-05, "loss": 0.0566, "step": 15160 }, { "epoch": 9.668578712555767, "grad_norm": 0.5131065845489502, "learning_rate": 9.9999913239246e-05, "loss": 0.0602, "step": 15170 }, { "epoch": 9.674952198852772, "grad_norm": 0.8779581785202026, "learning_rate": 9.999990266793106e-05, "loss": 0.0692, "step": 15180 }, { "epoch": 9.681325685149776, "grad_norm": 0.4830127954483032, "learning_rate": 9.99998914890701e-05, "loss": 0.0568, "step": 15190 }, { "epoch": 9.68769917144678, "grad_norm": 0.6114006042480469, "learning_rate": 9.999987970266323e-05, "loss": 0.0726, "step": 15200 }, { "epoch": 9.694072657743785, "grad_norm": 0.8563700914382935, "learning_rate": 9.999986730871064e-05, "loss": 0.0685, "step": 15210 }, { "epoch": 9.70044614404079, "grad_norm": 0.7654949426651001, "learning_rate": 9.999985430721243e-05, "loss": 0.051, "step": 15220 }, { "epoch": 9.706819630337796, "grad_norm": 0.8609695434570312, "learning_rate": 9.999984069816881e-05, "loss": 0.0626, "step": 15230 }, { "epoch": 9.7131931166348, "grad_norm": 0.7848291397094727, "learning_rate": 9.99998264815799e-05, "loss": 0.0457, "step": 15240 }, { "epoch": 9.719566602931804, "grad_norm": 0.631909966468811, "learning_rate": 9.99998116574459e-05, "loss": 0.0645, "step": 15250 }, { "epoch": 9.725940089228809, "grad_norm": 0.6485551595687866, "learning_rate": 9.999979622576698e-05, "loss": 0.0753, "step": 15260 }, { "epoch": 9.732313575525813, "grad_norm": 1.1122710704803467, "learning_rate": 9.999978018654333e-05, "loss": 0.0663, "step": 15270 }, { "epoch": 9.738687061822818, "grad_norm": 0.516740620136261, "learning_rate": 9.999976353977511e-05, "loss": 0.053, "step": 15280 }, { "epoch": 9.745060548119822, "grad_norm": 0.7186670899391174, "learning_rate": 9.99997462854626e-05, "loss": 0.0623, "step": 15290 }, { "epoch": 9.751434034416826, "grad_norm": 0.48222893476486206, "learning_rate": 9.999972842360594e-05, "loss": 0.0624, "step": 15300 }, { "epoch": 9.75780752071383, "grad_norm": 0.6196928024291992, "learning_rate": 9.999970995420537e-05, "loss": 0.0857, "step": 15310 }, { "epoch": 9.764181007010835, "grad_norm": 0.7647284865379333, "learning_rate": 9.999969087726114e-05, "loss": 0.0486, "step": 15320 }, { "epoch": 9.77055449330784, "grad_norm": 0.629514217376709, "learning_rate": 9.999967119277344e-05, "loss": 0.0475, "step": 15330 }, { "epoch": 9.776927979604844, "grad_norm": 0.4425369203090668, "learning_rate": 9.999965090074254e-05, "loss": 0.0557, "step": 15340 }, { "epoch": 9.783301465901848, "grad_norm": 0.7280276417732239, "learning_rate": 9.999963000116865e-05, "loss": 0.0549, "step": 15350 }, { "epoch": 9.789674952198853, "grad_norm": 0.8114652037620544, "learning_rate": 9.999960849405208e-05, "loss": 0.0519, "step": 15360 }, { "epoch": 9.796048438495857, "grad_norm": 0.7324187159538269, "learning_rate": 9.999958637939304e-05, "loss": 0.0611, "step": 15370 }, { "epoch": 9.802421924792862, "grad_norm": 0.5522525310516357, "learning_rate": 9.999956365719182e-05, "loss": 0.0573, "step": 15380 }, { "epoch": 9.808795411089866, "grad_norm": 0.664623498916626, "learning_rate": 9.99995403274487e-05, "loss": 0.04, "step": 15390 }, { "epoch": 9.81516889738687, "grad_norm": 0.696927547454834, "learning_rate": 9.999951639016395e-05, "loss": 0.0543, "step": 15400 }, { "epoch": 9.821542383683875, "grad_norm": 1.258500099182129, "learning_rate": 9.999949184533788e-05, "loss": 0.0614, "step": 15410 }, { "epoch": 9.82791586998088, "grad_norm": 0.6368541121482849, "learning_rate": 9.999946669297076e-05, "loss": 0.057, "step": 15420 }, { "epoch": 9.834289356277884, "grad_norm": 0.6270894408226013, "learning_rate": 9.999944093306293e-05, "loss": 0.0686, "step": 15430 }, { "epoch": 9.840662842574888, "grad_norm": 0.5678898692131042, "learning_rate": 9.999941456561465e-05, "loss": 0.0553, "step": 15440 }, { "epoch": 9.847036328871893, "grad_norm": 0.5162479281425476, "learning_rate": 9.999938759062633e-05, "loss": 0.0637, "step": 15450 }, { "epoch": 9.853409815168897, "grad_norm": 1.0005308389663696, "learning_rate": 9.999936000809819e-05, "loss": 0.0607, "step": 15460 }, { "epoch": 9.859783301465901, "grad_norm": 0.5103743672370911, "learning_rate": 9.999933181803066e-05, "loss": 0.0464, "step": 15470 }, { "epoch": 9.866156787762906, "grad_norm": 0.70570969581604, "learning_rate": 9.999930302042402e-05, "loss": 0.0477, "step": 15480 }, { "epoch": 9.87253027405991, "grad_norm": 0.3010031282901764, "learning_rate": 9.999927361527866e-05, "loss": 0.0578, "step": 15490 }, { "epoch": 9.878903760356915, "grad_norm": 0.5186561346054077, "learning_rate": 9.99992436025949e-05, "loss": 0.0531, "step": 15500 }, { "epoch": 9.885277246653919, "grad_norm": 0.5804335474967957, "learning_rate": 9.999921298237311e-05, "loss": 0.0633, "step": 15510 }, { "epoch": 9.891650732950923, "grad_norm": 0.824968159198761, "learning_rate": 9.999918175461371e-05, "loss": 0.0533, "step": 15520 }, { "epoch": 9.89802421924793, "grad_norm": 0.6402810215950012, "learning_rate": 9.999914991931702e-05, "loss": 0.0539, "step": 15530 }, { "epoch": 9.904397705544934, "grad_norm": 1.041672706604004, "learning_rate": 9.999911747648346e-05, "loss": 0.0579, "step": 15540 }, { "epoch": 9.910771191841938, "grad_norm": 0.5121114253997803, "learning_rate": 9.999908442611341e-05, "loss": 0.0424, "step": 15550 }, { "epoch": 9.917144678138943, "grad_norm": 0.3592798411846161, "learning_rate": 9.999905076820728e-05, "loss": 0.0579, "step": 15560 }, { "epoch": 9.923518164435947, "grad_norm": 0.6497433185577393, "learning_rate": 9.999901650276549e-05, "loss": 0.0585, "step": 15570 }, { "epoch": 9.929891650732952, "grad_norm": 0.6613918542861938, "learning_rate": 9.999898162978843e-05, "loss": 0.0824, "step": 15580 }, { "epoch": 9.936265137029956, "grad_norm": 0.5419838428497314, "learning_rate": 9.999894614927653e-05, "loss": 0.0525, "step": 15590 }, { "epoch": 9.94263862332696, "grad_norm": 0.41045868396759033, "learning_rate": 9.999891006123023e-05, "loss": 0.0507, "step": 15600 }, { "epoch": 9.949012109623965, "grad_norm": 0.6143380403518677, "learning_rate": 9.999887336564998e-05, "loss": 0.0612, "step": 15610 }, { "epoch": 9.95538559592097, "grad_norm": 0.7102112174034119, "learning_rate": 9.99988360625362e-05, "loss": 0.0799, "step": 15620 }, { "epoch": 9.961759082217974, "grad_norm": 0.5740867853164673, "learning_rate": 9.999879815188936e-05, "loss": 0.0605, "step": 15630 }, { "epoch": 9.968132568514978, "grad_norm": 0.4061245024204254, "learning_rate": 9.999875963370991e-05, "loss": 0.0647, "step": 15640 }, { "epoch": 9.974506054811982, "grad_norm": 0.7223836183547974, "learning_rate": 9.999872050799833e-05, "loss": 0.0654, "step": 15650 }, { "epoch": 9.980879541108987, "grad_norm": 0.5730621218681335, "learning_rate": 9.999868077475509e-05, "loss": 0.0509, "step": 15660 }, { "epoch": 9.987253027405991, "grad_norm": 0.5930061936378479, "learning_rate": 9.999864043398066e-05, "loss": 0.0638, "step": 15670 }, { "epoch": 9.993626513702996, "grad_norm": 0.6156326532363892, "learning_rate": 9.999859948567557e-05, "loss": 0.0514, "step": 15680 }, { "epoch": 10.0, "grad_norm": 0.7921385765075684, "learning_rate": 9.999855792984025e-05, "loss": 0.072, "step": 15690 }, { "epoch": 10.006373486297004, "grad_norm": 0.5710336565971375, "learning_rate": 9.999851576647528e-05, "loss": 0.0582, "step": 15700 }, { "epoch": 10.012746972594009, "grad_norm": 0.6952888369560242, "learning_rate": 9.999847299558111e-05, "loss": 0.0716, "step": 15710 }, { "epoch": 10.019120458891013, "grad_norm": 0.6671890616416931, "learning_rate": 9.999842961715831e-05, "loss": 0.059, "step": 15720 }, { "epoch": 10.025493945188018, "grad_norm": 0.5703734755516052, "learning_rate": 9.999838563120737e-05, "loss": 0.0665, "step": 15730 }, { "epoch": 10.031867431485022, "grad_norm": 0.7158628702163696, "learning_rate": 9.999834103772883e-05, "loss": 0.0625, "step": 15740 }, { "epoch": 10.038240917782026, "grad_norm": 0.7780296206474304, "learning_rate": 9.999829583672327e-05, "loss": 0.0554, "step": 15750 }, { "epoch": 10.04461440407903, "grad_norm": 0.7620775103569031, "learning_rate": 9.999825002819119e-05, "loss": 0.0526, "step": 15760 }, { "epoch": 10.050987890376035, "grad_norm": 0.7956362962722778, "learning_rate": 9.999820361213316e-05, "loss": 0.0603, "step": 15770 }, { "epoch": 10.05736137667304, "grad_norm": 0.5731815099716187, "learning_rate": 9.999815658854976e-05, "loss": 0.0446, "step": 15780 }, { "epoch": 10.063734862970044, "grad_norm": 0.9944173693656921, "learning_rate": 9.999810895744155e-05, "loss": 0.0616, "step": 15790 }, { "epoch": 10.070108349267048, "grad_norm": 0.4661408066749573, "learning_rate": 9.99980607188091e-05, "loss": 0.054, "step": 15800 }, { "epoch": 10.076481835564053, "grad_norm": 0.6796442866325378, "learning_rate": 9.999801187265301e-05, "loss": 0.0579, "step": 15810 }, { "epoch": 10.082855321861057, "grad_norm": 0.448863685131073, "learning_rate": 9.999796241897387e-05, "loss": 0.0556, "step": 15820 }, { "epoch": 10.089228808158062, "grad_norm": 0.644807755947113, "learning_rate": 9.99979123577723e-05, "loss": 0.0562, "step": 15830 }, { "epoch": 10.095602294455066, "grad_norm": 0.6758536100387573, "learning_rate": 9.999786168904886e-05, "loss": 0.049, "step": 15840 }, { "epoch": 10.101975780752072, "grad_norm": 0.6082631945610046, "learning_rate": 9.999781041280422e-05, "loss": 0.0472, "step": 15850 }, { "epoch": 10.108349267049077, "grad_norm": 1.315745234489441, "learning_rate": 9.999775852903896e-05, "loss": 0.0771, "step": 15860 }, { "epoch": 10.114722753346081, "grad_norm": 0.5523437857627869, "learning_rate": 9.999770603775373e-05, "loss": 0.0686, "step": 15870 }, { "epoch": 10.121096239643085, "grad_norm": 0.6934453845024109, "learning_rate": 9.999765293894918e-05, "loss": 0.0705, "step": 15880 }, { "epoch": 10.12746972594009, "grad_norm": 0.7601489424705505, "learning_rate": 9.999759923262593e-05, "loss": 0.0586, "step": 15890 }, { "epoch": 10.133843212237094, "grad_norm": 0.5398861765861511, "learning_rate": 9.999754491878464e-05, "loss": 0.0511, "step": 15900 }, { "epoch": 10.140216698534099, "grad_norm": 0.5415521264076233, "learning_rate": 9.999748999742597e-05, "loss": 0.0471, "step": 15910 }, { "epoch": 10.146590184831103, "grad_norm": 0.7853138446807861, "learning_rate": 9.999743446855061e-05, "loss": 0.0622, "step": 15920 }, { "epoch": 10.152963671128107, "grad_norm": 0.6591348052024841, "learning_rate": 9.99973783321592e-05, "loss": 0.064, "step": 15930 }, { "epoch": 10.159337157425112, "grad_norm": 0.7076617479324341, "learning_rate": 9.999732158825246e-05, "loss": 0.0599, "step": 15940 }, { "epoch": 10.165710643722116, "grad_norm": 0.6937175989151001, "learning_rate": 9.999726423683104e-05, "loss": 0.0727, "step": 15950 }, { "epoch": 10.17208413001912, "grad_norm": 0.5288292169570923, "learning_rate": 9.999720627789566e-05, "loss": 0.0526, "step": 15960 }, { "epoch": 10.178457616316125, "grad_norm": 0.6662324666976929, "learning_rate": 9.999714771144701e-05, "loss": 0.0488, "step": 15970 }, { "epoch": 10.18483110261313, "grad_norm": 0.3469081521034241, "learning_rate": 9.99970885374858e-05, "loss": 0.0599, "step": 15980 }, { "epoch": 10.191204588910134, "grad_norm": 0.5644394159317017, "learning_rate": 9.999702875601277e-05, "loss": 0.0518, "step": 15990 }, { "epoch": 10.197578075207138, "grad_norm": 0.38171422481536865, "learning_rate": 9.999696836702865e-05, "loss": 0.062, "step": 16000 }, { "epoch": 10.203951561504143, "grad_norm": 0.6088692545890808, "learning_rate": 9.999690737053416e-05, "loss": 0.0637, "step": 16010 }, { "epoch": 10.210325047801147, "grad_norm": 0.8294176459312439, "learning_rate": 9.999684576653001e-05, "loss": 0.0608, "step": 16020 }, { "epoch": 10.216698534098152, "grad_norm": 0.5365344285964966, "learning_rate": 9.999678355501701e-05, "loss": 0.0564, "step": 16030 }, { "epoch": 10.223072020395156, "grad_norm": 0.4190749526023865, "learning_rate": 9.999672073599586e-05, "loss": 0.049, "step": 16040 }, { "epoch": 10.22944550669216, "grad_norm": 0.7462962865829468, "learning_rate": 9.999665730946736e-05, "loss": 0.0517, "step": 16050 }, { "epoch": 10.235818992989165, "grad_norm": 0.47202739119529724, "learning_rate": 9.999659327543228e-05, "loss": 0.0506, "step": 16060 }, { "epoch": 10.24219247928617, "grad_norm": 0.598147988319397, "learning_rate": 9.999652863389138e-05, "loss": 0.0497, "step": 16070 }, { "epoch": 10.248565965583174, "grad_norm": 0.941378116607666, "learning_rate": 9.999646338484545e-05, "loss": 0.0555, "step": 16080 }, { "epoch": 10.254939451880178, "grad_norm": 0.6677988767623901, "learning_rate": 9.999639752829528e-05, "loss": 0.0526, "step": 16090 }, { "epoch": 10.261312938177182, "grad_norm": 0.6185131669044495, "learning_rate": 9.999633106424169e-05, "loss": 0.0551, "step": 16100 }, { "epoch": 10.267686424474187, "grad_norm": 0.5114786028862, "learning_rate": 9.999626399268546e-05, "loss": 0.0612, "step": 16110 }, { "epoch": 10.274059910771191, "grad_norm": 0.600007176399231, "learning_rate": 9.999619631362742e-05, "loss": 0.0668, "step": 16120 }, { "epoch": 10.280433397068196, "grad_norm": 0.6373008489608765, "learning_rate": 9.99961280270684e-05, "loss": 0.068, "step": 16130 }, { "epoch": 10.2868068833652, "grad_norm": 0.49713751673698425, "learning_rate": 9.99960591330092e-05, "loss": 0.0608, "step": 16140 }, { "epoch": 10.293180369662204, "grad_norm": 0.8552665114402771, "learning_rate": 9.999598963145069e-05, "loss": 0.0654, "step": 16150 }, { "epoch": 10.29955385595921, "grad_norm": 0.7540515065193176, "learning_rate": 9.999591952239371e-05, "loss": 0.0679, "step": 16160 }, { "epoch": 10.305927342256215, "grad_norm": 1.0121618509292603, "learning_rate": 9.99958488058391e-05, "loss": 0.0522, "step": 16170 }, { "epoch": 10.31230082855322, "grad_norm": 0.602270245552063, "learning_rate": 9.999577748178773e-05, "loss": 0.0565, "step": 16180 }, { "epoch": 10.318674314850224, "grad_norm": 0.7276262044906616, "learning_rate": 9.999570555024043e-05, "loss": 0.0559, "step": 16190 }, { "epoch": 10.325047801147228, "grad_norm": 0.6497871279716492, "learning_rate": 9.999563301119811e-05, "loss": 0.0641, "step": 16200 }, { "epoch": 10.331421287444233, "grad_norm": 0.5377486348152161, "learning_rate": 9.999555986466166e-05, "loss": 0.0569, "step": 16210 }, { "epoch": 10.337794773741237, "grad_norm": 0.6051101684570312, "learning_rate": 9.999548611063197e-05, "loss": 0.0524, "step": 16220 }, { "epoch": 10.344168260038241, "grad_norm": 0.8002655506134033, "learning_rate": 9.99954117491099e-05, "loss": 0.0597, "step": 16230 }, { "epoch": 10.350541746335246, "grad_norm": 0.507246732711792, "learning_rate": 9.999533678009639e-05, "loss": 0.0581, "step": 16240 }, { "epoch": 10.35691523263225, "grad_norm": 0.48768991231918335, "learning_rate": 9.99952612035923e-05, "loss": 0.0688, "step": 16250 }, { "epoch": 10.363288718929255, "grad_norm": 0.795545220375061, "learning_rate": 9.999518501959861e-05, "loss": 0.0643, "step": 16260 }, { "epoch": 10.369662205226259, "grad_norm": 0.6196532845497131, "learning_rate": 9.999510822811621e-05, "loss": 0.0582, "step": 16270 }, { "epoch": 10.376035691523263, "grad_norm": 0.6797712445259094, "learning_rate": 9.999503082914605e-05, "loss": 0.0608, "step": 16280 }, { "epoch": 10.382409177820268, "grad_norm": 0.9830362200737, "learning_rate": 9.999495282268905e-05, "loss": 0.065, "step": 16290 }, { "epoch": 10.388782664117272, "grad_norm": 0.676116943359375, "learning_rate": 9.999487420874618e-05, "loss": 0.0498, "step": 16300 }, { "epoch": 10.395156150414277, "grad_norm": 0.795210599899292, "learning_rate": 9.999479498731839e-05, "loss": 0.0505, "step": 16310 }, { "epoch": 10.401529636711281, "grad_norm": 0.7064853310585022, "learning_rate": 9.999471515840662e-05, "loss": 0.0558, "step": 16320 }, { "epoch": 10.407903123008285, "grad_norm": 0.5844698548316956, "learning_rate": 9.999463472201187e-05, "loss": 0.0658, "step": 16330 }, { "epoch": 10.41427660930529, "grad_norm": 0.5698778033256531, "learning_rate": 9.999455367813511e-05, "loss": 0.0532, "step": 16340 }, { "epoch": 10.420650095602294, "grad_norm": 0.6656918525695801, "learning_rate": 9.999447202677732e-05, "loss": 0.0536, "step": 16350 }, { "epoch": 10.427023581899299, "grad_norm": 0.5382378101348877, "learning_rate": 9.999438976793947e-05, "loss": 0.0574, "step": 16360 }, { "epoch": 10.433397068196303, "grad_norm": 0.8292346000671387, "learning_rate": 9.999430690162261e-05, "loss": 0.0612, "step": 16370 }, { "epoch": 10.439770554493307, "grad_norm": 0.7137431502342224, "learning_rate": 9.99942234278277e-05, "loss": 0.0666, "step": 16380 }, { "epoch": 10.446144040790312, "grad_norm": 0.7976674437522888, "learning_rate": 9.999413934655578e-05, "loss": 0.0576, "step": 16390 }, { "epoch": 10.452517527087316, "grad_norm": 0.7755173444747925, "learning_rate": 9.999405465780787e-05, "loss": 0.0648, "step": 16400 }, { "epoch": 10.45889101338432, "grad_norm": 0.519244372844696, "learning_rate": 9.999396936158498e-05, "loss": 0.0504, "step": 16410 }, { "epoch": 10.465264499681325, "grad_norm": 0.46889597177505493, "learning_rate": 9.999388345788819e-05, "loss": 0.0606, "step": 16420 }, { "epoch": 10.47163798597833, "grad_norm": 0.5492410659790039, "learning_rate": 9.999379694671847e-05, "loss": 0.0579, "step": 16430 }, { "epoch": 10.478011472275334, "grad_norm": 0.5425655245780945, "learning_rate": 9.999370982807695e-05, "loss": 0.0572, "step": 16440 }, { "epoch": 10.484384958572338, "grad_norm": 0.5655606985092163, "learning_rate": 9.999362210196463e-05, "loss": 0.0519, "step": 16450 }, { "epoch": 10.490758444869343, "grad_norm": 0.475955992937088, "learning_rate": 9.99935337683826e-05, "loss": 0.0534, "step": 16460 }, { "epoch": 10.497131931166347, "grad_norm": 0.462833970785141, "learning_rate": 9.999344482733197e-05, "loss": 0.0565, "step": 16470 }, { "epoch": 10.503505417463352, "grad_norm": 0.5556862354278564, "learning_rate": 9.999335527881375e-05, "loss": 0.049, "step": 16480 }, { "epoch": 10.509878903760358, "grad_norm": 0.8238747715950012, "learning_rate": 9.999326512282908e-05, "loss": 0.0707, "step": 16490 }, { "epoch": 10.516252390057362, "grad_norm": 0.9749704599380493, "learning_rate": 9.9993174359379e-05, "loss": 0.0549, "step": 16500 }, { "epoch": 10.522625876354367, "grad_norm": 0.5116440653800964, "learning_rate": 9.999308298846468e-05, "loss": 0.0558, "step": 16510 }, { "epoch": 10.528999362651371, "grad_norm": 0.5135734677314758, "learning_rate": 9.999299101008721e-05, "loss": 0.0598, "step": 16520 }, { "epoch": 10.535372848948375, "grad_norm": 0.5870323777198792, "learning_rate": 9.999289842424768e-05, "loss": 0.0581, "step": 16530 }, { "epoch": 10.54174633524538, "grad_norm": 0.6875941157341003, "learning_rate": 9.999280523094724e-05, "loss": 0.0499, "step": 16540 }, { "epoch": 10.548119821542384, "grad_norm": 0.4197682738304138, "learning_rate": 9.999271143018702e-05, "loss": 0.0662, "step": 16550 }, { "epoch": 10.554493307839389, "grad_norm": 0.840517520904541, "learning_rate": 9.999261702196813e-05, "loss": 0.0674, "step": 16560 }, { "epoch": 10.560866794136393, "grad_norm": 0.6064279675483704, "learning_rate": 9.999252200629176e-05, "loss": 0.0608, "step": 16570 }, { "epoch": 10.567240280433397, "grad_norm": 0.6202387809753418, "learning_rate": 9.999242638315906e-05, "loss": 0.0605, "step": 16580 }, { "epoch": 10.573613766730402, "grad_norm": 0.674312174320221, "learning_rate": 9.999233015257116e-05, "loss": 0.0517, "step": 16590 }, { "epoch": 10.579987253027406, "grad_norm": 0.3645886778831482, "learning_rate": 9.999223331452926e-05, "loss": 0.0444, "step": 16600 }, { "epoch": 10.58636073932441, "grad_norm": 1.1831674575805664, "learning_rate": 9.999213586903452e-05, "loss": 0.0578, "step": 16610 }, { "epoch": 10.592734225621415, "grad_norm": 0.5315924882888794, "learning_rate": 9.999203781608813e-05, "loss": 0.0432, "step": 16620 }, { "epoch": 10.59910771191842, "grad_norm": 0.5224555134773254, "learning_rate": 9.999193915569127e-05, "loss": 0.0422, "step": 16630 }, { "epoch": 10.605481198215424, "grad_norm": 0.3949601650238037, "learning_rate": 9.999183988784516e-05, "loss": 0.0509, "step": 16640 }, { "epoch": 10.611854684512428, "grad_norm": 0.6192469596862793, "learning_rate": 9.999174001255099e-05, "loss": 0.0613, "step": 16650 }, { "epoch": 10.618228170809433, "grad_norm": 0.557607114315033, "learning_rate": 9.999163952981e-05, "loss": 0.0606, "step": 16660 }, { "epoch": 10.624601657106437, "grad_norm": 0.9131451845169067, "learning_rate": 9.999153843962336e-05, "loss": 0.0428, "step": 16670 }, { "epoch": 10.630975143403441, "grad_norm": 0.3558861315250397, "learning_rate": 9.999143674199234e-05, "loss": 0.0611, "step": 16680 }, { "epoch": 10.637348629700446, "grad_norm": 0.7525888681411743, "learning_rate": 9.999133443691817e-05, "loss": 0.0635, "step": 16690 }, { "epoch": 10.64372211599745, "grad_norm": 0.6017816066741943, "learning_rate": 9.999123152440208e-05, "loss": 0.0566, "step": 16700 }, { "epoch": 10.650095602294455, "grad_norm": 0.5555442571640015, "learning_rate": 9.999112800444533e-05, "loss": 0.057, "step": 16710 }, { "epoch": 10.656469088591459, "grad_norm": 0.8302058577537537, "learning_rate": 9.999102387704917e-05, "loss": 0.059, "step": 16720 }, { "epoch": 10.662842574888463, "grad_norm": 0.7159201502799988, "learning_rate": 9.999091914221486e-05, "loss": 0.0647, "step": 16730 }, { "epoch": 10.669216061185468, "grad_norm": 0.5368846654891968, "learning_rate": 9.99908137999437e-05, "loss": 0.0516, "step": 16740 }, { "epoch": 10.675589547482472, "grad_norm": 0.4882710576057434, "learning_rate": 9.999070785023696e-05, "loss": 0.0496, "step": 16750 }, { "epoch": 10.681963033779477, "grad_norm": 0.8025274276733398, "learning_rate": 9.99906012930959e-05, "loss": 0.0626, "step": 16760 }, { "epoch": 10.688336520076481, "grad_norm": 0.7932184338569641, "learning_rate": 9.999049412852186e-05, "loss": 0.0549, "step": 16770 }, { "epoch": 10.694710006373487, "grad_norm": 0.7863204479217529, "learning_rate": 9.99903863565161e-05, "loss": 0.0638, "step": 16780 }, { "epoch": 10.701083492670492, "grad_norm": 0.6114450693130493, "learning_rate": 9.999027797707994e-05, "loss": 0.059, "step": 16790 }, { "epoch": 10.707456978967496, "grad_norm": 0.35417380928993225, "learning_rate": 9.999016899021472e-05, "loss": 0.0677, "step": 16800 }, { "epoch": 10.7138304652645, "grad_norm": 0.6127309799194336, "learning_rate": 9.999005939592174e-05, "loss": 0.0523, "step": 16810 }, { "epoch": 10.720203951561505, "grad_norm": 0.7677340507507324, "learning_rate": 9.998994919420234e-05, "loss": 0.0572, "step": 16820 }, { "epoch": 10.72657743785851, "grad_norm": 0.41578641533851624, "learning_rate": 9.998983838505786e-05, "loss": 0.0533, "step": 16830 }, { "epoch": 10.732950924155514, "grad_norm": 0.7479081153869629, "learning_rate": 9.998972696848965e-05, "loss": 0.0488, "step": 16840 }, { "epoch": 10.739324410452518, "grad_norm": 0.5703463554382324, "learning_rate": 9.998961494449906e-05, "loss": 0.0424, "step": 16850 }, { "epoch": 10.745697896749522, "grad_norm": 0.7256620526313782, "learning_rate": 9.998950231308744e-05, "loss": 0.0473, "step": 16860 }, { "epoch": 10.752071383046527, "grad_norm": 0.38508841395378113, "learning_rate": 9.998938907425616e-05, "loss": 0.0587, "step": 16870 }, { "epoch": 10.758444869343531, "grad_norm": 0.39906227588653564, "learning_rate": 9.998927522800663e-05, "loss": 0.0418, "step": 16880 }, { "epoch": 10.764818355640536, "grad_norm": 0.42261654138565063, "learning_rate": 9.998916077434019e-05, "loss": 0.0505, "step": 16890 }, { "epoch": 10.77119184193754, "grad_norm": 0.8031429648399353, "learning_rate": 9.998904571325827e-05, "loss": 0.0528, "step": 16900 }, { "epoch": 10.777565328234544, "grad_norm": 0.29207149147987366, "learning_rate": 9.998893004476224e-05, "loss": 0.0494, "step": 16910 }, { "epoch": 10.783938814531549, "grad_norm": 0.6371474862098694, "learning_rate": 9.99888137688535e-05, "loss": 0.0546, "step": 16920 }, { "epoch": 10.790312300828553, "grad_norm": 0.7001533508300781, "learning_rate": 9.998869688553348e-05, "loss": 0.052, "step": 16930 }, { "epoch": 10.796685787125558, "grad_norm": 0.5313620567321777, "learning_rate": 9.998857939480361e-05, "loss": 0.0553, "step": 16940 }, { "epoch": 10.803059273422562, "grad_norm": 0.470103919506073, "learning_rate": 9.998846129666529e-05, "loss": 0.0477, "step": 16950 }, { "epoch": 10.809432759719567, "grad_norm": 0.8891826868057251, "learning_rate": 9.998834259111996e-05, "loss": 0.0509, "step": 16960 }, { "epoch": 10.815806246016571, "grad_norm": 0.6524693369865417, "learning_rate": 9.998822327816908e-05, "loss": 0.0565, "step": 16970 }, { "epoch": 10.822179732313575, "grad_norm": 0.7200140357017517, "learning_rate": 9.99881033578141e-05, "loss": 0.0557, "step": 16980 }, { "epoch": 10.82855321861058, "grad_norm": 0.6414451003074646, "learning_rate": 9.998798283005647e-05, "loss": 0.061, "step": 16990 }, { "epoch": 10.834926704907584, "grad_norm": 0.398269385099411, "learning_rate": 9.998786169489762e-05, "loss": 0.0512, "step": 17000 }, { "epoch": 10.841300191204589, "grad_norm": 0.6985471844673157, "learning_rate": 9.998773995233908e-05, "loss": 0.0615, "step": 17010 }, { "epoch": 10.847673677501593, "grad_norm": 0.5928990840911865, "learning_rate": 9.99876176023823e-05, "loss": 0.0631, "step": 17020 }, { "epoch": 10.854047163798597, "grad_norm": 0.5422001481056213, "learning_rate": 9.998749464502877e-05, "loss": 0.0546, "step": 17030 }, { "epoch": 10.860420650095602, "grad_norm": 0.5168808102607727, "learning_rate": 9.998737108027998e-05, "loss": 0.0441, "step": 17040 }, { "epoch": 10.866794136392606, "grad_norm": 0.4506094753742218, "learning_rate": 9.998724690813744e-05, "loss": 0.0393, "step": 17050 }, { "epoch": 10.87316762268961, "grad_norm": 0.7235383987426758, "learning_rate": 9.998712212860264e-05, "loss": 0.0453, "step": 17060 }, { "epoch": 10.879541108986615, "grad_norm": 0.7417569756507874, "learning_rate": 9.998699674167712e-05, "loss": 0.0526, "step": 17070 }, { "epoch": 10.88591459528362, "grad_norm": 0.7921003103256226, "learning_rate": 9.998687074736241e-05, "loss": 0.0502, "step": 17080 }, { "epoch": 10.892288081580624, "grad_norm": 0.58517986536026, "learning_rate": 9.998674414566001e-05, "loss": 0.0582, "step": 17090 }, { "epoch": 10.898661567877628, "grad_norm": 0.48357248306274414, "learning_rate": 9.998661693657148e-05, "loss": 0.0438, "step": 17100 }, { "epoch": 10.905035054174633, "grad_norm": 0.5111726522445679, "learning_rate": 9.998648912009835e-05, "loss": 0.0503, "step": 17110 }, { "epoch": 10.911408540471639, "grad_norm": 0.3606181740760803, "learning_rate": 9.998636069624219e-05, "loss": 0.0579, "step": 17120 }, { "epoch": 10.917782026768643, "grad_norm": 0.8223302960395813, "learning_rate": 9.998623166500454e-05, "loss": 0.0471, "step": 17130 }, { "epoch": 10.924155513065648, "grad_norm": 0.47486552596092224, "learning_rate": 9.998610202638699e-05, "loss": 0.0386, "step": 17140 }, { "epoch": 10.930528999362652, "grad_norm": 0.7364180684089661, "learning_rate": 9.998597178039111e-05, "loss": 0.0638, "step": 17150 }, { "epoch": 10.936902485659656, "grad_norm": 0.6910980939865112, "learning_rate": 9.998584092701848e-05, "loss": 0.0577, "step": 17160 }, { "epoch": 10.94327597195666, "grad_norm": 0.40659859776496887, "learning_rate": 9.998570946627067e-05, "loss": 0.0568, "step": 17170 }, { "epoch": 10.949649458253665, "grad_norm": 0.4734037220478058, "learning_rate": 9.998557739814931e-05, "loss": 0.0587, "step": 17180 }, { "epoch": 10.95602294455067, "grad_norm": 0.626246452331543, "learning_rate": 9.998544472265598e-05, "loss": 0.0556, "step": 17190 }, { "epoch": 10.962396430847674, "grad_norm": 1.0859874486923218, "learning_rate": 9.998531143979232e-05, "loss": 0.0494, "step": 17200 }, { "epoch": 10.968769917144678, "grad_norm": 0.5683014392852783, "learning_rate": 9.998517754955992e-05, "loss": 0.0611, "step": 17210 }, { "epoch": 10.975143403441683, "grad_norm": 0.6202378273010254, "learning_rate": 9.998504305196042e-05, "loss": 0.0566, "step": 17220 }, { "epoch": 10.981516889738687, "grad_norm": 0.5997171998023987, "learning_rate": 9.998490794699546e-05, "loss": 0.0432, "step": 17230 }, { "epoch": 10.987890376035692, "grad_norm": 0.5708463191986084, "learning_rate": 9.998477223466667e-05, "loss": 0.045, "step": 17240 }, { "epoch": 10.994263862332696, "grad_norm": 0.5977245569229126, "learning_rate": 9.998463591497569e-05, "loss": 0.0844, "step": 17250 }, { "epoch": 11.0006373486297, "grad_norm": 0.7333068251609802, "learning_rate": 9.998449898792422e-05, "loss": 0.0603, "step": 17260 }, { "epoch": 11.007010834926705, "grad_norm": 0.584259569644928, "learning_rate": 9.998436145351386e-05, "loss": 0.0456, "step": 17270 }, { "epoch": 11.01338432122371, "grad_norm": 0.43404531478881836, "learning_rate": 9.998422331174634e-05, "loss": 0.0522, "step": 17280 }, { "epoch": 11.019757807520714, "grad_norm": 1.031726360321045, "learning_rate": 9.99840845626233e-05, "loss": 0.0598, "step": 17290 }, { "epoch": 11.026131293817718, "grad_norm": 0.798851728439331, "learning_rate": 9.998394520614645e-05, "loss": 0.052, "step": 17300 }, { "epoch": 11.032504780114722, "grad_norm": 0.46099844574928284, "learning_rate": 9.998380524231747e-05, "loss": 0.0574, "step": 17310 }, { "epoch": 11.038878266411727, "grad_norm": 0.6638126373291016, "learning_rate": 9.998366467113807e-05, "loss": 0.0344, "step": 17320 }, { "epoch": 11.045251752708731, "grad_norm": 0.6141223907470703, "learning_rate": 9.998352349260993e-05, "loss": 0.0466, "step": 17330 }, { "epoch": 11.051625239005736, "grad_norm": 0.7332136034965515, "learning_rate": 9.99833817067348e-05, "loss": 0.0467, "step": 17340 }, { "epoch": 11.05799872530274, "grad_norm": 0.565891683101654, "learning_rate": 9.998323931351441e-05, "loss": 0.0442, "step": 17350 }, { "epoch": 11.064372211599744, "grad_norm": 0.6422147154808044, "learning_rate": 9.998309631295045e-05, "loss": 0.0495, "step": 17360 }, { "epoch": 11.070745697896749, "grad_norm": 0.6818075180053711, "learning_rate": 9.998295270504468e-05, "loss": 0.0532, "step": 17370 }, { "epoch": 11.077119184193753, "grad_norm": 0.7112347483634949, "learning_rate": 9.998280848979884e-05, "loss": 0.0665, "step": 17380 }, { "epoch": 11.083492670490758, "grad_norm": 0.8333026766777039, "learning_rate": 9.99826636672147e-05, "loss": 0.0528, "step": 17390 }, { "epoch": 11.089866156787762, "grad_norm": 0.6337835192680359, "learning_rate": 9.998251823729399e-05, "loss": 0.0444, "step": 17400 }, { "epoch": 11.096239643084767, "grad_norm": 0.5257356762886047, "learning_rate": 9.998237220003848e-05, "loss": 0.0559, "step": 17410 }, { "epoch": 11.102613129381773, "grad_norm": 0.6054905652999878, "learning_rate": 9.998222555544999e-05, "loss": 0.0477, "step": 17420 }, { "epoch": 11.108986615678777, "grad_norm": 0.47715023159980774, "learning_rate": 9.998207830353024e-05, "loss": 0.0562, "step": 17430 }, { "epoch": 11.115360101975782, "grad_norm": 1.4385261535644531, "learning_rate": 9.998193044428106e-05, "loss": 0.0393, "step": 17440 }, { "epoch": 11.121733588272786, "grad_norm": 0.53253173828125, "learning_rate": 9.998178197770422e-05, "loss": 0.0644, "step": 17450 }, { "epoch": 11.12810707456979, "grad_norm": 0.7377299070358276, "learning_rate": 9.998163290380156e-05, "loss": 0.0491, "step": 17460 }, { "epoch": 11.134480560866795, "grad_norm": 0.31071048974990845, "learning_rate": 9.998148322257486e-05, "loss": 0.0494, "step": 17470 }, { "epoch": 11.1408540471638, "grad_norm": 0.428337424993515, "learning_rate": 9.998133293402595e-05, "loss": 0.0396, "step": 17480 }, { "epoch": 11.147227533460804, "grad_norm": 0.6270567178726196, "learning_rate": 9.998118203815665e-05, "loss": 0.0583, "step": 17490 }, { "epoch": 11.153601019757808, "grad_norm": 0.36213913559913635, "learning_rate": 9.998103053496882e-05, "loss": 0.043, "step": 17500 }, { "epoch": 11.159974506054812, "grad_norm": 0.4144083857536316, "learning_rate": 9.998087842446426e-05, "loss": 0.042, "step": 17510 }, { "epoch": 11.166347992351817, "grad_norm": 0.45778176188468933, "learning_rate": 9.998072570664483e-05, "loss": 0.0467, "step": 17520 }, { "epoch": 11.172721478648821, "grad_norm": 0.4047306776046753, "learning_rate": 9.99805723815124e-05, "loss": 0.0574, "step": 17530 }, { "epoch": 11.179094964945826, "grad_norm": 0.958371102809906, "learning_rate": 9.998041844906883e-05, "loss": 0.0617, "step": 17540 }, { "epoch": 11.18546845124283, "grad_norm": 0.7824062705039978, "learning_rate": 9.998026390931598e-05, "loss": 0.0434, "step": 17550 }, { "epoch": 11.191841937539834, "grad_norm": 0.48062580823898315, "learning_rate": 9.998010876225574e-05, "loss": 0.0562, "step": 17560 }, { "epoch": 11.198215423836839, "grad_norm": 0.6278024315834045, "learning_rate": 9.997995300789e-05, "loss": 0.0632, "step": 17570 }, { "epoch": 11.204588910133843, "grad_norm": 0.7428111433982849, "learning_rate": 9.997979664622062e-05, "loss": 0.0563, "step": 17580 }, { "epoch": 11.210962396430848, "grad_norm": 0.9911885857582092, "learning_rate": 9.997963967724952e-05, "loss": 0.0588, "step": 17590 }, { "epoch": 11.217335882727852, "grad_norm": 0.631261944770813, "learning_rate": 9.997948210097863e-05, "loss": 0.0539, "step": 17600 }, { "epoch": 11.223709369024856, "grad_norm": 0.8290361166000366, "learning_rate": 9.997932391740983e-05, "loss": 0.0761, "step": 17610 }, { "epoch": 11.23008285532186, "grad_norm": 0.7124612331390381, "learning_rate": 9.997916512654507e-05, "loss": 0.0587, "step": 17620 }, { "epoch": 11.236456341618865, "grad_norm": 0.30079036951065063, "learning_rate": 9.997900572838626e-05, "loss": 0.0404, "step": 17630 }, { "epoch": 11.24282982791587, "grad_norm": 0.7114712595939636, "learning_rate": 9.997884572293534e-05, "loss": 0.06, "step": 17640 }, { "epoch": 11.249203314212874, "grad_norm": 0.6289411783218384, "learning_rate": 9.997868511019425e-05, "loss": 0.0498, "step": 17650 }, { "epoch": 11.255576800509878, "grad_norm": 1.10546875, "learning_rate": 9.997852389016496e-05, "loss": 0.0443, "step": 17660 }, { "epoch": 11.261950286806883, "grad_norm": 0.8365715742111206, "learning_rate": 9.997836206284941e-05, "loss": 0.0646, "step": 17670 }, { "epoch": 11.268323773103887, "grad_norm": 0.629873514175415, "learning_rate": 9.997819962824957e-05, "loss": 0.061, "step": 17680 }, { "epoch": 11.274697259400892, "grad_norm": 0.8870893120765686, "learning_rate": 9.997803658636742e-05, "loss": 0.0548, "step": 17690 }, { "epoch": 11.281070745697896, "grad_norm": 0.4535408318042755, "learning_rate": 9.997787293720495e-05, "loss": 0.0551, "step": 17700 }, { "epoch": 11.2874442319949, "grad_norm": 1.0586836338043213, "learning_rate": 9.997770868076413e-05, "loss": 0.0565, "step": 17710 }, { "epoch": 11.293817718291905, "grad_norm": 0.4891342222690582, "learning_rate": 9.997754381704696e-05, "loss": 0.0463, "step": 17720 }, { "epoch": 11.30019120458891, "grad_norm": 0.3861699402332306, "learning_rate": 9.997737834605545e-05, "loss": 0.0494, "step": 17730 }, { "epoch": 11.306564690885915, "grad_norm": 0.7491887211799622, "learning_rate": 9.99772122677916e-05, "loss": 0.0532, "step": 17740 }, { "epoch": 11.31293817718292, "grad_norm": 0.3956933319568634, "learning_rate": 9.997704558225743e-05, "loss": 0.0579, "step": 17750 }, { "epoch": 11.319311663479924, "grad_norm": 0.5724512934684753, "learning_rate": 9.997687828945499e-05, "loss": 0.0657, "step": 17760 }, { "epoch": 11.325685149776929, "grad_norm": 0.6473221182823181, "learning_rate": 9.997671038938628e-05, "loss": 0.0658, "step": 17770 }, { "epoch": 11.332058636073933, "grad_norm": 0.7001258134841919, "learning_rate": 9.997654188205334e-05, "loss": 0.0496, "step": 17780 }, { "epoch": 11.338432122370937, "grad_norm": 0.560114860534668, "learning_rate": 9.997637276745826e-05, "loss": 0.0615, "step": 17790 }, { "epoch": 11.344805608667942, "grad_norm": 0.6450788378715515, "learning_rate": 9.997620304560304e-05, "loss": 0.0521, "step": 17800 }, { "epoch": 11.351179094964946, "grad_norm": 0.4742850959300995, "learning_rate": 9.997603271648979e-05, "loss": 0.0454, "step": 17810 }, { "epoch": 11.35755258126195, "grad_norm": 1.1459397077560425, "learning_rate": 9.997586178012052e-05, "loss": 0.0524, "step": 17820 }, { "epoch": 11.363926067558955, "grad_norm": 0.4577571749687195, "learning_rate": 9.997569023649737e-05, "loss": 0.0503, "step": 17830 }, { "epoch": 11.37029955385596, "grad_norm": 0.6457903981208801, "learning_rate": 9.997551808562241e-05, "loss": 0.0497, "step": 17840 }, { "epoch": 11.376673040152964, "grad_norm": 0.9518571496009827, "learning_rate": 9.997534532749771e-05, "loss": 0.058, "step": 17850 }, { "epoch": 11.383046526449968, "grad_norm": 0.5197674632072449, "learning_rate": 9.997517196212538e-05, "loss": 0.0513, "step": 17860 }, { "epoch": 11.389420012746973, "grad_norm": 0.6322747468948364, "learning_rate": 9.997499798950751e-05, "loss": 0.0563, "step": 17870 }, { "epoch": 11.395793499043977, "grad_norm": 0.6855553984642029, "learning_rate": 9.997482340964625e-05, "loss": 0.0491, "step": 17880 }, { "epoch": 11.402166985340981, "grad_norm": 0.37808260321617126, "learning_rate": 9.99746482225437e-05, "loss": 0.0497, "step": 17890 }, { "epoch": 11.408540471637986, "grad_norm": 0.7514583468437195, "learning_rate": 9.997447242820199e-05, "loss": 0.0503, "step": 17900 }, { "epoch": 11.41491395793499, "grad_norm": 0.5832131505012512, "learning_rate": 9.997429602662325e-05, "loss": 0.0438, "step": 17910 }, { "epoch": 11.421287444231995, "grad_norm": 0.5786134004592896, "learning_rate": 9.997411901780963e-05, "loss": 0.0539, "step": 17920 }, { "epoch": 11.427660930529, "grad_norm": 0.474710613489151, "learning_rate": 9.997394140176328e-05, "loss": 0.0454, "step": 17930 }, { "epoch": 11.434034416826004, "grad_norm": 0.8023557662963867, "learning_rate": 9.997376317848636e-05, "loss": 0.0593, "step": 17940 }, { "epoch": 11.440407903123008, "grad_norm": 1.0486323833465576, "learning_rate": 9.997358434798103e-05, "loss": 0.0554, "step": 17950 }, { "epoch": 11.446781389420012, "grad_norm": 0.5281891822814941, "learning_rate": 9.997340491024948e-05, "loss": 0.0503, "step": 17960 }, { "epoch": 11.453154875717017, "grad_norm": 0.6257161498069763, "learning_rate": 9.997322486529387e-05, "loss": 0.0533, "step": 17970 }, { "epoch": 11.459528362014021, "grad_norm": 0.7398996949195862, "learning_rate": 9.997304421311639e-05, "loss": 0.0537, "step": 17980 }, { "epoch": 11.465901848311026, "grad_norm": 0.47160854935646057, "learning_rate": 9.997286295371926e-05, "loss": 0.05, "step": 17990 }, { "epoch": 11.47227533460803, "grad_norm": 0.5948900580406189, "learning_rate": 9.997268108710464e-05, "loss": 0.0575, "step": 18000 }, { "epoch": 11.478648820905034, "grad_norm": 0.5790464878082275, "learning_rate": 9.997249861327478e-05, "loss": 0.044, "step": 18010 }, { "epoch": 11.485022307202039, "grad_norm": 0.5215743184089661, "learning_rate": 9.997231553223186e-05, "loss": 0.0634, "step": 18020 }, { "epoch": 11.491395793499043, "grad_norm": 0.7115923762321472, "learning_rate": 9.997213184397814e-05, "loss": 0.0559, "step": 18030 }, { "epoch": 11.49776927979605, "grad_norm": 0.6699000000953674, "learning_rate": 9.997194754851584e-05, "loss": 0.0547, "step": 18040 }, { "epoch": 11.504142766093054, "grad_norm": 0.7512786388397217, "learning_rate": 9.997176264584717e-05, "loss": 0.0627, "step": 18050 }, { "epoch": 11.510516252390058, "grad_norm": 0.5977685451507568, "learning_rate": 9.997157713597444e-05, "loss": 0.0505, "step": 18060 }, { "epoch": 11.516889738687063, "grad_norm": 1.0020976066589355, "learning_rate": 9.997139101889986e-05, "loss": 0.0602, "step": 18070 }, { "epoch": 11.523263224984067, "grad_norm": 0.6246644258499146, "learning_rate": 9.997120429462567e-05, "loss": 0.0536, "step": 18080 }, { "epoch": 11.529636711281071, "grad_norm": 0.7035977840423584, "learning_rate": 9.997101696315418e-05, "loss": 0.0597, "step": 18090 }, { "epoch": 11.536010197578076, "grad_norm": 0.7818805575370789, "learning_rate": 9.997082902448767e-05, "loss": 0.0601, "step": 18100 }, { "epoch": 11.54238368387508, "grad_norm": 0.9022552967071533, "learning_rate": 9.997064047862839e-05, "loss": 0.0569, "step": 18110 }, { "epoch": 11.548757170172085, "grad_norm": 0.6126748919487, "learning_rate": 9.997045132557866e-05, "loss": 0.056, "step": 18120 }, { "epoch": 11.555130656469089, "grad_norm": 0.6667356491088867, "learning_rate": 9.997026156534075e-05, "loss": 0.0488, "step": 18130 }, { "epoch": 11.561504142766093, "grad_norm": 0.4176846444606781, "learning_rate": 9.9970071197917e-05, "loss": 0.045, "step": 18140 }, { "epoch": 11.567877629063098, "grad_norm": 0.6282650232315063, "learning_rate": 9.996988022330969e-05, "loss": 0.0547, "step": 18150 }, { "epoch": 11.574251115360102, "grad_norm": 0.46769824624061584, "learning_rate": 9.996968864152116e-05, "loss": 0.0562, "step": 18160 }, { "epoch": 11.580624601657107, "grad_norm": 0.6001009941101074, "learning_rate": 9.996949645255374e-05, "loss": 0.053, "step": 18170 }, { "epoch": 11.586998087954111, "grad_norm": 0.4135037362575531, "learning_rate": 9.996930365640975e-05, "loss": 0.0403, "step": 18180 }, { "epoch": 11.593371574251115, "grad_norm": 0.563538134098053, "learning_rate": 9.996911025309155e-05, "loss": 0.0516, "step": 18190 }, { "epoch": 11.59974506054812, "grad_norm": 0.6317570209503174, "learning_rate": 9.996891624260147e-05, "loss": 0.0551, "step": 18200 }, { "epoch": 11.606118546845124, "grad_norm": 0.591675341129303, "learning_rate": 9.996872162494189e-05, "loss": 0.0524, "step": 18210 }, { "epoch": 11.612492033142129, "grad_norm": 0.8648858070373535, "learning_rate": 9.996852640011516e-05, "loss": 0.0417, "step": 18220 }, { "epoch": 11.618865519439133, "grad_norm": 0.63719242811203, "learning_rate": 9.996833056812367e-05, "loss": 0.0472, "step": 18230 }, { "epoch": 11.625239005736137, "grad_norm": 0.6004825234413147, "learning_rate": 9.996813412896977e-05, "loss": 0.0563, "step": 18240 }, { "epoch": 11.631612492033142, "grad_norm": 1.0001980066299438, "learning_rate": 9.996793708265586e-05, "loss": 0.0636, "step": 18250 }, { "epoch": 11.637985978330146, "grad_norm": 1.067087173461914, "learning_rate": 9.996773942918435e-05, "loss": 0.0608, "step": 18260 }, { "epoch": 11.64435946462715, "grad_norm": 0.847445547580719, "learning_rate": 9.996754116855763e-05, "loss": 0.0565, "step": 18270 }, { "epoch": 11.650732950924155, "grad_norm": 0.3594801425933838, "learning_rate": 9.99673423007781e-05, "loss": 0.0517, "step": 18280 }, { "epoch": 11.65710643722116, "grad_norm": 0.6227310299873352, "learning_rate": 9.996714282584818e-05, "loss": 0.0522, "step": 18290 }, { "epoch": 11.663479923518164, "grad_norm": 0.4727412760257721, "learning_rate": 9.99669427437703e-05, "loss": 0.0482, "step": 18300 }, { "epoch": 11.669853409815168, "grad_norm": 0.9138240814208984, "learning_rate": 9.996674205454688e-05, "loss": 0.0668, "step": 18310 }, { "epoch": 11.676226896112173, "grad_norm": 0.6075490117073059, "learning_rate": 9.996654075818039e-05, "loss": 0.0509, "step": 18320 }, { "epoch": 11.682600382409177, "grad_norm": 0.7808625102043152, "learning_rate": 9.996633885467324e-05, "loss": 0.0592, "step": 18330 }, { "epoch": 11.688973868706181, "grad_norm": 0.7400175929069519, "learning_rate": 9.996613634402789e-05, "loss": 0.0629, "step": 18340 }, { "epoch": 11.695347355003186, "grad_norm": 0.755127489566803, "learning_rate": 9.996593322624682e-05, "loss": 0.058, "step": 18350 }, { "epoch": 11.70172084130019, "grad_norm": 0.44330671429634094, "learning_rate": 9.996572950133248e-05, "loss": 0.0517, "step": 18360 }, { "epoch": 11.708094327597196, "grad_norm": 0.5726228356361389, "learning_rate": 9.996552516928735e-05, "loss": 0.0571, "step": 18370 }, { "epoch": 11.7144678138942, "grad_norm": 0.591275691986084, "learning_rate": 9.996532023011392e-05, "loss": 0.054, "step": 18380 }, { "epoch": 11.720841300191205, "grad_norm": 1.0331820249557495, "learning_rate": 9.996511468381467e-05, "loss": 0.0497, "step": 18390 }, { "epoch": 11.72721478648821, "grad_norm": 0.9469217658042908, "learning_rate": 9.996490853039211e-05, "loss": 0.0519, "step": 18400 }, { "epoch": 11.733588272785214, "grad_norm": 0.5066688060760498, "learning_rate": 9.996470176984873e-05, "loss": 0.0558, "step": 18410 }, { "epoch": 11.739961759082219, "grad_norm": 0.586126446723938, "learning_rate": 9.996449440218704e-05, "loss": 0.0772, "step": 18420 }, { "epoch": 11.746335245379223, "grad_norm": 0.5571174621582031, "learning_rate": 9.996428642740958e-05, "loss": 0.055, "step": 18430 }, { "epoch": 11.752708731676227, "grad_norm": 0.6182481050491333, "learning_rate": 9.996407784551887e-05, "loss": 0.0509, "step": 18440 }, { "epoch": 11.759082217973232, "grad_norm": 0.4466564357280731, "learning_rate": 9.996386865651744e-05, "loss": 0.0388, "step": 18450 }, { "epoch": 11.765455704270236, "grad_norm": 0.4253198206424713, "learning_rate": 9.996365886040782e-05, "loss": 0.0555, "step": 18460 }, { "epoch": 11.77182919056724, "grad_norm": 0.7634193897247314, "learning_rate": 9.996344845719257e-05, "loss": 0.0592, "step": 18470 }, { "epoch": 11.778202676864245, "grad_norm": 0.7888400554656982, "learning_rate": 9.996323744687426e-05, "loss": 0.061, "step": 18480 }, { "epoch": 11.78457616316125, "grad_norm": 0.29007822275161743, "learning_rate": 9.996302582945543e-05, "loss": 0.0378, "step": 18490 }, { "epoch": 11.790949649458254, "grad_norm": 0.3867831230163574, "learning_rate": 9.996281360493866e-05, "loss": 0.0545, "step": 18500 }, { "epoch": 11.797323135755258, "grad_norm": 0.8648092150688171, "learning_rate": 9.996260077332656e-05, "loss": 0.0539, "step": 18510 }, { "epoch": 11.803696622052263, "grad_norm": 0.5546586513519287, "learning_rate": 9.996238733462167e-05, "loss": 0.045, "step": 18520 }, { "epoch": 11.810070108349267, "grad_norm": 0.6702616810798645, "learning_rate": 9.99621732888266e-05, "loss": 0.0542, "step": 18530 }, { "epoch": 11.816443594646271, "grad_norm": 0.5179563760757446, "learning_rate": 9.996195863594396e-05, "loss": 0.0494, "step": 18540 }, { "epoch": 11.822817080943276, "grad_norm": 0.5806443095207214, "learning_rate": 9.996174337597632e-05, "loss": 0.0441, "step": 18550 }, { "epoch": 11.82919056724028, "grad_norm": 0.6398141384124756, "learning_rate": 9.996152750892635e-05, "loss": 0.0592, "step": 18560 }, { "epoch": 11.835564053537285, "grad_norm": 0.4812298119068146, "learning_rate": 9.996131103479667e-05, "loss": 0.0609, "step": 18570 }, { "epoch": 11.841937539834289, "grad_norm": 0.4223327338695526, "learning_rate": 9.996109395358986e-05, "loss": 0.0545, "step": 18580 }, { "epoch": 11.848311026131293, "grad_norm": 0.585468590259552, "learning_rate": 9.99608762653086e-05, "loss": 0.0539, "step": 18590 }, { "epoch": 11.854684512428298, "grad_norm": 0.5610795617103577, "learning_rate": 9.996065796995552e-05, "loss": 0.0601, "step": 18600 }, { "epoch": 11.861057998725302, "grad_norm": 1.0543278455734253, "learning_rate": 9.996043906753328e-05, "loss": 0.0565, "step": 18610 }, { "epoch": 11.867431485022307, "grad_norm": 0.8825480341911316, "learning_rate": 9.996021955804453e-05, "loss": 0.046, "step": 18620 }, { "epoch": 11.873804971319311, "grad_norm": 0.6645362973213196, "learning_rate": 9.995999944149192e-05, "loss": 0.0795, "step": 18630 }, { "epoch": 11.880178457616315, "grad_norm": 0.6627642512321472, "learning_rate": 9.995977871787818e-05, "loss": 0.0591, "step": 18640 }, { "epoch": 11.88655194391332, "grad_norm": 0.5781305432319641, "learning_rate": 9.995955738720595e-05, "loss": 0.0549, "step": 18650 }, { "epoch": 11.892925430210324, "grad_norm": 0.28119561076164246, "learning_rate": 9.995933544947792e-05, "loss": 0.0444, "step": 18660 }, { "epoch": 11.89929891650733, "grad_norm": 0.49158886075019836, "learning_rate": 9.995911290469679e-05, "loss": 0.0555, "step": 18670 }, { "epoch": 11.905672402804335, "grad_norm": 1.0449167490005493, "learning_rate": 9.995888975286529e-05, "loss": 0.0584, "step": 18680 }, { "epoch": 11.91204588910134, "grad_norm": 0.4692186117172241, "learning_rate": 9.99586659939861e-05, "loss": 0.0498, "step": 18690 }, { "epoch": 11.918419375398344, "grad_norm": 0.6601914167404175, "learning_rate": 9.995844162806195e-05, "loss": 0.0345, "step": 18700 }, { "epoch": 11.924792861695348, "grad_norm": 0.7166926264762878, "learning_rate": 9.995821665509558e-05, "loss": 0.0573, "step": 18710 }, { "epoch": 11.931166347992352, "grad_norm": 0.507392942905426, "learning_rate": 9.995799107508968e-05, "loss": 0.0398, "step": 18720 }, { "epoch": 11.937539834289357, "grad_norm": 0.569351315498352, "learning_rate": 9.995776488804702e-05, "loss": 0.0726, "step": 18730 }, { "epoch": 11.943913320586361, "grad_norm": 0.5389679074287415, "learning_rate": 9.995753809397037e-05, "loss": 0.0484, "step": 18740 }, { "epoch": 11.950286806883366, "grad_norm": 0.7490147948265076, "learning_rate": 9.995731069286246e-05, "loss": 0.0383, "step": 18750 }, { "epoch": 11.95666029318037, "grad_norm": 0.38610783219337463, "learning_rate": 9.995708268472606e-05, "loss": 0.0375, "step": 18760 }, { "epoch": 11.963033779477374, "grad_norm": 1.1472092866897583, "learning_rate": 9.995685406956392e-05, "loss": 0.0632, "step": 18770 }, { "epoch": 11.969407265774379, "grad_norm": 1.152759313583374, "learning_rate": 9.995662484737885e-05, "loss": 0.0506, "step": 18780 }, { "epoch": 11.975780752071383, "grad_norm": 0.6192075610160828, "learning_rate": 9.995639501817362e-05, "loss": 0.059, "step": 18790 }, { "epoch": 11.982154238368388, "grad_norm": 0.3730803430080414, "learning_rate": 9.995616458195103e-05, "loss": 0.0435, "step": 18800 }, { "epoch": 11.988527724665392, "grad_norm": 0.4230964481830597, "learning_rate": 9.995593353871386e-05, "loss": 0.0552, "step": 18810 }, { "epoch": 11.994901210962396, "grad_norm": 0.28632041811943054, "learning_rate": 9.995570188846495e-05, "loss": 0.0394, "step": 18820 }, { "epoch": 12.0012746972594, "grad_norm": 0.6016038060188293, "learning_rate": 9.995546963120708e-05, "loss": 0.0531, "step": 18830 }, { "epoch": 12.007648183556405, "grad_norm": 1.32315993309021, "learning_rate": 9.995523676694309e-05, "loss": 0.0474, "step": 18840 }, { "epoch": 12.01402166985341, "grad_norm": 0.4533834755420685, "learning_rate": 9.995500329567583e-05, "loss": 0.0464, "step": 18850 }, { "epoch": 12.020395156150414, "grad_norm": 0.4621172249317169, "learning_rate": 9.995476921740807e-05, "loss": 0.0544, "step": 18860 }, { "epoch": 12.026768642447419, "grad_norm": 0.5771495699882507, "learning_rate": 9.995453453214274e-05, "loss": 0.047, "step": 18870 }, { "epoch": 12.033142128744423, "grad_norm": 0.677911102771759, "learning_rate": 9.995429923988263e-05, "loss": 0.0484, "step": 18880 }, { "epoch": 12.039515615041427, "grad_norm": 0.4545127749443054, "learning_rate": 9.995406334063062e-05, "loss": 0.0438, "step": 18890 }, { "epoch": 12.045889101338432, "grad_norm": 0.4634072482585907, "learning_rate": 9.995382683438958e-05, "loss": 0.064, "step": 18900 }, { "epoch": 12.052262587635436, "grad_norm": 0.5201743245124817, "learning_rate": 9.995358972116238e-05, "loss": 0.0466, "step": 18910 }, { "epoch": 12.05863607393244, "grad_norm": 0.7461873888969421, "learning_rate": 9.99533520009519e-05, "loss": 0.0449, "step": 18920 }, { "epoch": 12.065009560229445, "grad_norm": 0.5392987728118896, "learning_rate": 9.995311367376103e-05, "loss": 0.0491, "step": 18930 }, { "epoch": 12.07138304652645, "grad_norm": 0.7738301753997803, "learning_rate": 9.995287473959267e-05, "loss": 0.0515, "step": 18940 }, { "epoch": 12.077756532823454, "grad_norm": 0.7333182096481323, "learning_rate": 9.995263519844971e-05, "loss": 0.0426, "step": 18950 }, { "epoch": 12.084130019120458, "grad_norm": 0.7453256845474243, "learning_rate": 9.995239505033508e-05, "loss": 0.0409, "step": 18960 }, { "epoch": 12.090503505417463, "grad_norm": 0.6855975389480591, "learning_rate": 9.995215429525169e-05, "loss": 0.0542, "step": 18970 }, { "epoch": 12.096876991714467, "grad_norm": 0.35131192207336426, "learning_rate": 9.995191293320244e-05, "loss": 0.0532, "step": 18980 }, { "epoch": 12.103250478011473, "grad_norm": 0.5471280813217163, "learning_rate": 9.99516709641903e-05, "loss": 0.0464, "step": 18990 }, { "epoch": 12.109623964308478, "grad_norm": 0.5725041627883911, "learning_rate": 9.99514283882182e-05, "loss": 0.0531, "step": 19000 }, { "epoch": 12.115997450605482, "grad_norm": 0.4529055953025818, "learning_rate": 9.995118520528907e-05, "loss": 0.0507, "step": 19010 }, { "epoch": 12.122370936902486, "grad_norm": 0.8099885582923889, "learning_rate": 9.995094141540588e-05, "loss": 0.0496, "step": 19020 }, { "epoch": 12.12874442319949, "grad_norm": 0.37661561369895935, "learning_rate": 9.995069701857161e-05, "loss": 0.0471, "step": 19030 }, { "epoch": 12.135117909496495, "grad_norm": 0.5382967591285706, "learning_rate": 9.995045201478918e-05, "loss": 0.0595, "step": 19040 }, { "epoch": 12.1414913957935, "grad_norm": 0.5374127626419067, "learning_rate": 9.995020640406162e-05, "loss": 0.0439, "step": 19050 }, { "epoch": 12.147864882090504, "grad_norm": 0.27735188603401184, "learning_rate": 9.994996018639189e-05, "loss": 0.0458, "step": 19060 }, { "epoch": 12.154238368387508, "grad_norm": 0.6636207699775696, "learning_rate": 9.994971336178298e-05, "loss": 0.0477, "step": 19070 }, { "epoch": 12.160611854684513, "grad_norm": 0.6746484041213989, "learning_rate": 9.994946593023789e-05, "loss": 0.0471, "step": 19080 }, { "epoch": 12.166985340981517, "grad_norm": 0.32199937105178833, "learning_rate": 9.994921789175962e-05, "loss": 0.0509, "step": 19090 }, { "epoch": 12.173358827278522, "grad_norm": 0.6574446558952332, "learning_rate": 9.99489692463512e-05, "loss": 0.0636, "step": 19100 }, { "epoch": 12.179732313575526, "grad_norm": 0.7455927133560181, "learning_rate": 9.994871999401565e-05, "loss": 0.0415, "step": 19110 }, { "epoch": 12.18610579987253, "grad_norm": 0.6799868941307068, "learning_rate": 9.9948470134756e-05, "loss": 0.0504, "step": 19120 }, { "epoch": 12.192479286169535, "grad_norm": 0.3111349940299988, "learning_rate": 9.994821966857527e-05, "loss": 0.0478, "step": 19130 }, { "epoch": 12.19885277246654, "grad_norm": 0.4255180358886719, "learning_rate": 9.99479685954765e-05, "loss": 0.0604, "step": 19140 }, { "epoch": 12.205226258763544, "grad_norm": 0.6373127698898315, "learning_rate": 9.994771691546277e-05, "loss": 0.0404, "step": 19150 }, { "epoch": 12.211599745060548, "grad_norm": 0.7585551142692566, "learning_rate": 9.99474646285371e-05, "loss": 0.0459, "step": 19160 }, { "epoch": 12.217973231357552, "grad_norm": 0.5794774889945984, "learning_rate": 9.994721173470259e-05, "loss": 0.0457, "step": 19170 }, { "epoch": 12.224346717654557, "grad_norm": 0.5366019606590271, "learning_rate": 9.994695823396229e-05, "loss": 0.0562, "step": 19180 }, { "epoch": 12.230720203951561, "grad_norm": 0.7718575596809387, "learning_rate": 9.99467041263193e-05, "loss": 0.0578, "step": 19190 }, { "epoch": 12.237093690248566, "grad_norm": 0.4321316182613373, "learning_rate": 9.99464494117767e-05, "loss": 0.0435, "step": 19200 }, { "epoch": 12.24346717654557, "grad_norm": 0.5822618007659912, "learning_rate": 9.994619409033759e-05, "loss": 0.0535, "step": 19210 }, { "epoch": 12.249840662842574, "grad_norm": 0.6593590378761292, "learning_rate": 9.994593816200504e-05, "loss": 0.042, "step": 19220 }, { "epoch": 12.256214149139579, "grad_norm": 0.5769529342651367, "learning_rate": 9.994568162678219e-05, "loss": 0.0542, "step": 19230 }, { "epoch": 12.262587635436583, "grad_norm": 0.42017069458961487, "learning_rate": 9.994542448467216e-05, "loss": 0.0502, "step": 19240 }, { "epoch": 12.268961121733588, "grad_norm": 0.8386192321777344, "learning_rate": 9.994516673567807e-05, "loss": 0.0521, "step": 19250 }, { "epoch": 12.275334608030592, "grad_norm": 0.4258641004562378, "learning_rate": 9.994490837980305e-05, "loss": 0.0323, "step": 19260 }, { "epoch": 12.281708094327596, "grad_norm": 0.5408238172531128, "learning_rate": 9.994464941705022e-05, "loss": 0.036, "step": 19270 }, { "epoch": 12.2880815806246, "grad_norm": 0.423775851726532, "learning_rate": 9.994438984742275e-05, "loss": 0.0405, "step": 19280 }, { "epoch": 12.294455066921605, "grad_norm": 0.4911399185657501, "learning_rate": 9.99441296709238e-05, "loss": 0.0466, "step": 19290 }, { "epoch": 12.300828553218611, "grad_norm": 0.5672997832298279, "learning_rate": 9.994386888755651e-05, "loss": 0.0489, "step": 19300 }, { "epoch": 12.307202039515616, "grad_norm": 0.6688255071640015, "learning_rate": 9.994360749732407e-05, "loss": 0.0418, "step": 19310 }, { "epoch": 12.31357552581262, "grad_norm": 0.8784617185592651, "learning_rate": 9.994334550022964e-05, "loss": 0.0504, "step": 19320 }, { "epoch": 12.319949012109625, "grad_norm": 0.6308597922325134, "learning_rate": 9.994308289627639e-05, "loss": 0.0626, "step": 19330 }, { "epoch": 12.326322498406629, "grad_norm": 0.5725502371788025, "learning_rate": 9.994281968546755e-05, "loss": 0.0591, "step": 19340 }, { "epoch": 12.332695984703633, "grad_norm": 0.6846663951873779, "learning_rate": 9.994255586780629e-05, "loss": 0.0693, "step": 19350 }, { "epoch": 12.339069471000638, "grad_norm": 0.6701339483261108, "learning_rate": 9.99422914432958e-05, "loss": 0.0443, "step": 19360 }, { "epoch": 12.345442957297642, "grad_norm": 0.618817925453186, "learning_rate": 9.994202641193935e-05, "loss": 0.045, "step": 19370 }, { "epoch": 12.351816443594647, "grad_norm": 0.5576841831207275, "learning_rate": 9.994176077374012e-05, "loss": 0.0372, "step": 19380 }, { "epoch": 12.358189929891651, "grad_norm": 0.8428722620010376, "learning_rate": 9.994149452870133e-05, "loss": 0.0697, "step": 19390 }, { "epoch": 12.364563416188656, "grad_norm": 0.930770218372345, "learning_rate": 9.994122767682623e-05, "loss": 0.0519, "step": 19400 }, { "epoch": 12.37093690248566, "grad_norm": 1.0066298246383667, "learning_rate": 9.994096021811808e-05, "loss": 0.0484, "step": 19410 }, { "epoch": 12.377310388782664, "grad_norm": 0.42919373512268066, "learning_rate": 9.994069215258008e-05, "loss": 0.0488, "step": 19420 }, { "epoch": 12.383683875079669, "grad_norm": 0.5092325806617737, "learning_rate": 9.994042348021556e-05, "loss": 0.0602, "step": 19430 }, { "epoch": 12.390057361376673, "grad_norm": 0.5449823141098022, "learning_rate": 9.994015420102771e-05, "loss": 0.0655, "step": 19440 }, { "epoch": 12.396430847673678, "grad_norm": 0.5374361872673035, "learning_rate": 9.993988431501983e-05, "loss": 0.0463, "step": 19450 }, { "epoch": 12.402804333970682, "grad_norm": 0.6612887382507324, "learning_rate": 9.993961382219523e-05, "loss": 0.0533, "step": 19460 }, { "epoch": 12.409177820267686, "grad_norm": 0.5763975381851196, "learning_rate": 9.993934272255715e-05, "loss": 0.0492, "step": 19470 }, { "epoch": 12.41555130656469, "grad_norm": 0.46687546372413635, "learning_rate": 9.993907101610893e-05, "loss": 0.0537, "step": 19480 }, { "epoch": 12.421924792861695, "grad_norm": 0.6873669028282166, "learning_rate": 9.993879870285383e-05, "loss": 0.0477, "step": 19490 }, { "epoch": 12.4282982791587, "grad_norm": 0.793414294719696, "learning_rate": 9.993852578279518e-05, "loss": 0.0648, "step": 19500 }, { "epoch": 12.434671765455704, "grad_norm": 0.38007280230522156, "learning_rate": 9.993825225593629e-05, "loss": 0.0558, "step": 19510 }, { "epoch": 12.441045251752708, "grad_norm": 0.39757072925567627, "learning_rate": 9.993797812228048e-05, "loss": 0.0463, "step": 19520 }, { "epoch": 12.447418738049713, "grad_norm": 0.5992366075515747, "learning_rate": 9.99377033818311e-05, "loss": 0.0492, "step": 19530 }, { "epoch": 12.453792224346717, "grad_norm": 0.6468036770820618, "learning_rate": 9.993742803459146e-05, "loss": 0.0485, "step": 19540 }, { "epoch": 12.460165710643722, "grad_norm": 0.8821713328361511, "learning_rate": 9.993715208056493e-05, "loss": 0.053, "step": 19550 }, { "epoch": 12.466539196940726, "grad_norm": 3.29703950881958, "learning_rate": 9.993687551975485e-05, "loss": 0.0622, "step": 19560 }, { "epoch": 12.47291268323773, "grad_norm": 0.6634793281555176, "learning_rate": 9.99365983521646e-05, "loss": 0.0607, "step": 19570 }, { "epoch": 12.479286169534735, "grad_norm": 0.628406822681427, "learning_rate": 9.993632057779752e-05, "loss": 0.0526, "step": 19580 }, { "epoch": 12.48565965583174, "grad_norm": 0.41636964678764343, "learning_rate": 9.993604219665698e-05, "loss": 0.0444, "step": 19590 }, { "epoch": 12.492033142128744, "grad_norm": 0.7125439643859863, "learning_rate": 9.99357632087464e-05, "loss": 0.0574, "step": 19600 }, { "epoch": 12.498406628425748, "grad_norm": 0.37637677788734436, "learning_rate": 9.993548361406914e-05, "loss": 0.0467, "step": 19610 }, { "epoch": 12.504780114722752, "grad_norm": 0.4394405484199524, "learning_rate": 9.99352034126286e-05, "loss": 0.0483, "step": 19620 }, { "epoch": 12.511153601019759, "grad_norm": 0.5288838744163513, "learning_rate": 9.993492260442822e-05, "loss": 0.0607, "step": 19630 }, { "epoch": 12.517527087316763, "grad_norm": 0.5851248502731323, "learning_rate": 9.993464118947137e-05, "loss": 0.0349, "step": 19640 }, { "epoch": 12.523900573613767, "grad_norm": 0.663963258266449, "learning_rate": 9.993435916776149e-05, "loss": 0.0535, "step": 19650 }, { "epoch": 12.530274059910772, "grad_norm": 0.6779718399047852, "learning_rate": 9.993407653930198e-05, "loss": 0.0556, "step": 19660 }, { "epoch": 12.536647546207776, "grad_norm": 0.881489634513855, "learning_rate": 9.993379330409632e-05, "loss": 0.0498, "step": 19670 }, { "epoch": 12.54302103250478, "grad_norm": 0.5240219235420227, "learning_rate": 9.99335094621479e-05, "loss": 0.0413, "step": 19680 }, { "epoch": 12.549394518801785, "grad_norm": 0.5868619680404663, "learning_rate": 9.993322501346021e-05, "loss": 0.0538, "step": 19690 }, { "epoch": 12.55576800509879, "grad_norm": 0.7972108721733093, "learning_rate": 9.993293995803669e-05, "loss": 0.0508, "step": 19700 }, { "epoch": 12.562141491395794, "grad_norm": 0.739250659942627, "learning_rate": 9.993265429588079e-05, "loss": 0.0543, "step": 19710 }, { "epoch": 12.568514977692798, "grad_norm": 0.6294760704040527, "learning_rate": 9.993236802699601e-05, "loss": 0.055, "step": 19720 }, { "epoch": 12.574888463989803, "grad_norm": 0.6474262475967407, "learning_rate": 9.993208115138581e-05, "loss": 0.0474, "step": 19730 }, { "epoch": 12.581261950286807, "grad_norm": 0.49055972695350647, "learning_rate": 9.99317936690537e-05, "loss": 0.0554, "step": 19740 }, { "epoch": 12.587635436583811, "grad_norm": 0.5933214426040649, "learning_rate": 9.993150558000313e-05, "loss": 0.0365, "step": 19750 }, { "epoch": 12.594008922880816, "grad_norm": 0.6509645581245422, "learning_rate": 9.993121688423764e-05, "loss": 0.0509, "step": 19760 }, { "epoch": 12.60038240917782, "grad_norm": 0.588777482509613, "learning_rate": 9.99309275817607e-05, "loss": 0.055, "step": 19770 }, { "epoch": 12.606755895474825, "grad_norm": 0.5462532639503479, "learning_rate": 9.993063767257586e-05, "loss": 0.0506, "step": 19780 }, { "epoch": 12.613129381771829, "grad_norm": 0.6147275567054749, "learning_rate": 9.993034715668664e-05, "loss": 0.0534, "step": 19790 }, { "epoch": 12.619502868068833, "grad_norm": 0.687687873840332, "learning_rate": 9.993005603409655e-05, "loss": 0.0459, "step": 19800 }, { "epoch": 12.625876354365838, "grad_norm": 0.6746910810470581, "learning_rate": 9.992976430480914e-05, "loss": 0.054, "step": 19810 }, { "epoch": 12.632249840662842, "grad_norm": 0.7588403820991516, "learning_rate": 9.992947196882796e-05, "loss": 0.043, "step": 19820 }, { "epoch": 12.638623326959847, "grad_norm": 0.4694322943687439, "learning_rate": 9.992917902615654e-05, "loss": 0.0445, "step": 19830 }, { "epoch": 12.644996813256851, "grad_norm": 0.36447465419769287, "learning_rate": 9.992888547679846e-05, "loss": 0.0415, "step": 19840 }, { "epoch": 12.651370299553856, "grad_norm": 0.5364222526550293, "learning_rate": 9.992859132075728e-05, "loss": 0.0565, "step": 19850 }, { "epoch": 12.65774378585086, "grad_norm": 0.5305570960044861, "learning_rate": 9.992829655803657e-05, "loss": 0.0423, "step": 19860 }, { "epoch": 12.664117272147864, "grad_norm": 0.6378289461135864, "learning_rate": 9.992800118863992e-05, "loss": 0.0553, "step": 19870 }, { "epoch": 12.670490758444869, "grad_norm": 0.5443289875984192, "learning_rate": 9.992770521257092e-05, "loss": 0.0628, "step": 19880 }, { "epoch": 12.676864244741873, "grad_norm": 0.7340859174728394, "learning_rate": 9.992740862983316e-05, "loss": 0.0515, "step": 19890 }, { "epoch": 12.683237731038878, "grad_norm": 0.4966719448566437, "learning_rate": 9.992711144043024e-05, "loss": 0.0453, "step": 19900 }, { "epoch": 12.689611217335882, "grad_norm": 0.7671192288398743, "learning_rate": 9.992681364436578e-05, "loss": 0.0662, "step": 19910 }, { "epoch": 12.695984703632888, "grad_norm": 0.41733086109161377, "learning_rate": 9.992651524164341e-05, "loss": 0.0329, "step": 19920 }, { "epoch": 12.702358189929893, "grad_norm": 0.5677781701087952, "learning_rate": 9.992621623226672e-05, "loss": 0.0574, "step": 19930 }, { "epoch": 12.708731676226897, "grad_norm": 0.637280285358429, "learning_rate": 9.992591661623937e-05, "loss": 0.0483, "step": 19940 }, { "epoch": 12.715105162523901, "grad_norm": 0.5569368600845337, "learning_rate": 9.992561639356502e-05, "loss": 0.0458, "step": 19950 }, { "epoch": 12.721478648820906, "grad_norm": 0.556477963924408, "learning_rate": 9.992531556424726e-05, "loss": 0.0544, "step": 19960 }, { "epoch": 12.72785213511791, "grad_norm": 0.5600087642669678, "learning_rate": 9.992501412828978e-05, "loss": 0.0579, "step": 19970 }, { "epoch": 12.734225621414915, "grad_norm": 0.6145752668380737, "learning_rate": 9.992471208569626e-05, "loss": 0.0516, "step": 19980 }, { "epoch": 12.740599107711919, "grad_norm": 0.38638541102409363, "learning_rate": 9.992440943647035e-05, "loss": 0.0557, "step": 19990 }, { "epoch": 12.746972594008923, "grad_norm": 0.6183023452758789, "learning_rate": 9.992410618061572e-05, "loss": 0.0551, "step": 20000 }, { "epoch": 12.753346080305928, "grad_norm": 0.6596343517303467, "learning_rate": 9.992380231813607e-05, "loss": 0.0601, "step": 20010 }, { "epoch": 12.759719566602932, "grad_norm": 0.7001132965087891, "learning_rate": 9.992349784903508e-05, "loss": 0.0427, "step": 20020 }, { "epoch": 12.766093052899937, "grad_norm": 0.4830794632434845, "learning_rate": 9.992319277331646e-05, "loss": 0.0522, "step": 20030 }, { "epoch": 12.772466539196941, "grad_norm": 0.8132880330085754, "learning_rate": 9.992288709098389e-05, "loss": 0.0561, "step": 20040 }, { "epoch": 12.778840025493945, "grad_norm": 0.8195405006408691, "learning_rate": 9.992258080204113e-05, "loss": 0.0504, "step": 20050 }, { "epoch": 12.78521351179095, "grad_norm": 0.6251553297042847, "learning_rate": 9.99222739064919e-05, "loss": 0.0474, "step": 20060 }, { "epoch": 12.791586998087954, "grad_norm": 0.38529151678085327, "learning_rate": 9.992196640433986e-05, "loss": 0.0425, "step": 20070 }, { "epoch": 12.797960484384959, "grad_norm": 0.7417767643928528, "learning_rate": 9.992165829558881e-05, "loss": 0.063, "step": 20080 }, { "epoch": 12.804333970681963, "grad_norm": 0.8575668334960938, "learning_rate": 9.99213495802425e-05, "loss": 0.0565, "step": 20090 }, { "epoch": 12.810707456978967, "grad_norm": 0.5553948879241943, "learning_rate": 9.992104025830465e-05, "loss": 0.0422, "step": 20100 }, { "epoch": 12.817080943275972, "grad_norm": 0.41762441396713257, "learning_rate": 9.992073032977903e-05, "loss": 0.0461, "step": 20110 }, { "epoch": 12.823454429572976, "grad_norm": 0.5332140922546387, "learning_rate": 9.992041979466937e-05, "loss": 0.0499, "step": 20120 }, { "epoch": 12.82982791586998, "grad_norm": 0.6717177629470825, "learning_rate": 9.992010865297952e-05, "loss": 0.0499, "step": 20130 }, { "epoch": 12.836201402166985, "grad_norm": 0.5670310258865356, "learning_rate": 9.991979690471319e-05, "loss": 0.0427, "step": 20140 }, { "epoch": 12.84257488846399, "grad_norm": 0.5739524960517883, "learning_rate": 9.991948454987421e-05, "loss": 0.0538, "step": 20150 }, { "epoch": 12.848948374760994, "grad_norm": 0.5475471019744873, "learning_rate": 9.991917158846635e-05, "loss": 0.0419, "step": 20160 }, { "epoch": 12.855321861057998, "grad_norm": 0.7085890173912048, "learning_rate": 9.991885802049343e-05, "loss": 0.0581, "step": 20170 }, { "epoch": 12.861695347355003, "grad_norm": 0.9820379018783569, "learning_rate": 9.991854384595926e-05, "loss": 0.0481, "step": 20180 }, { "epoch": 12.868068833652007, "grad_norm": 0.2957818806171417, "learning_rate": 9.991822906486763e-05, "loss": 0.051, "step": 20190 }, { "epoch": 12.874442319949011, "grad_norm": 0.6987794041633606, "learning_rate": 9.991791367722241e-05, "loss": 0.0514, "step": 20200 }, { "epoch": 12.880815806246016, "grad_norm": 0.4746705889701843, "learning_rate": 9.991759768302739e-05, "loss": 0.0597, "step": 20210 }, { "epoch": 12.88718929254302, "grad_norm": 0.6064462661743164, "learning_rate": 9.991728108228645e-05, "loss": 0.0557, "step": 20220 }, { "epoch": 12.893562778840025, "grad_norm": 0.8060480356216431, "learning_rate": 9.99169638750034e-05, "loss": 0.0404, "step": 20230 }, { "epoch": 12.899936265137029, "grad_norm": 0.4107417166233063, "learning_rate": 9.991664606118212e-05, "loss": 0.0589, "step": 20240 }, { "epoch": 12.906309751434035, "grad_norm": 0.42253613471984863, "learning_rate": 9.991632764082644e-05, "loss": 0.0564, "step": 20250 }, { "epoch": 12.91268323773104, "grad_norm": 0.39724990725517273, "learning_rate": 9.991600861394026e-05, "loss": 0.0346, "step": 20260 }, { "epoch": 12.919056724028044, "grad_norm": 0.6950128674507141, "learning_rate": 9.991568898052745e-05, "loss": 0.0507, "step": 20270 }, { "epoch": 12.925430210325048, "grad_norm": 0.6304432153701782, "learning_rate": 9.99153687405919e-05, "loss": 0.0404, "step": 20280 }, { "epoch": 12.931803696622053, "grad_norm": 0.6180141568183899, "learning_rate": 9.991504789413748e-05, "loss": 0.0604, "step": 20290 }, { "epoch": 12.938177182919057, "grad_norm": 0.5379759669303894, "learning_rate": 9.99147264411681e-05, "loss": 0.0483, "step": 20300 }, { "epoch": 12.944550669216062, "grad_norm": 0.5783708691596985, "learning_rate": 9.991440438168767e-05, "loss": 0.0463, "step": 20310 }, { "epoch": 12.950924155513066, "grad_norm": 0.48275479674339294, "learning_rate": 9.99140817157001e-05, "loss": 0.0452, "step": 20320 }, { "epoch": 12.95729764181007, "grad_norm": 0.4784011244773865, "learning_rate": 9.991375844320931e-05, "loss": 0.0467, "step": 20330 }, { "epoch": 12.963671128107075, "grad_norm": 1.0397515296936035, "learning_rate": 9.991343456421923e-05, "loss": 0.0517, "step": 20340 }, { "epoch": 12.97004461440408, "grad_norm": 0.3007799983024597, "learning_rate": 9.99131100787338e-05, "loss": 0.0532, "step": 20350 }, { "epoch": 12.976418100701084, "grad_norm": 0.6348592042922974, "learning_rate": 9.991278498675694e-05, "loss": 0.0567, "step": 20360 }, { "epoch": 12.982791586998088, "grad_norm": 0.8217516541481018, "learning_rate": 9.991245928829262e-05, "loss": 0.0585, "step": 20370 }, { "epoch": 12.989165073295093, "grad_norm": 0.5914630889892578, "learning_rate": 9.991213298334481e-05, "loss": 0.054, "step": 20380 }, { "epoch": 12.995538559592097, "grad_norm": 0.6380867958068848, "learning_rate": 9.991180607191745e-05, "loss": 0.0722, "step": 20390 }, { "epoch": 13.001912045889101, "grad_norm": 0.6516058444976807, "learning_rate": 9.991147855401453e-05, "loss": 0.0449, "step": 20400 }, { "epoch": 13.008285532186106, "grad_norm": 0.6741889119148254, "learning_rate": 9.991115042964002e-05, "loss": 0.0488, "step": 20410 }, { "epoch": 13.01465901848311, "grad_norm": 0.6972917318344116, "learning_rate": 9.991082169879788e-05, "loss": 0.0532, "step": 20420 }, { "epoch": 13.021032504780115, "grad_norm": 0.5252955555915833, "learning_rate": 9.991049236149216e-05, "loss": 0.0419, "step": 20430 }, { "epoch": 13.027405991077119, "grad_norm": 0.5976403951644897, "learning_rate": 9.991016241772683e-05, "loss": 0.0523, "step": 20440 }, { "epoch": 13.033779477374123, "grad_norm": 0.7129591107368469, "learning_rate": 9.99098318675059e-05, "loss": 0.0552, "step": 20450 }, { "epoch": 13.040152963671128, "grad_norm": 0.541470468044281, "learning_rate": 9.99095007108334e-05, "loss": 0.0608, "step": 20460 }, { "epoch": 13.046526449968132, "grad_norm": 0.4547019898891449, "learning_rate": 9.990916894771334e-05, "loss": 0.0557, "step": 20470 }, { "epoch": 13.052899936265137, "grad_norm": 0.5617836713790894, "learning_rate": 9.990883657814975e-05, "loss": 0.0472, "step": 20480 }, { "epoch": 13.059273422562141, "grad_norm": 0.9097099304199219, "learning_rate": 9.990850360214667e-05, "loss": 0.044, "step": 20490 }, { "epoch": 13.065646908859145, "grad_norm": 0.421392560005188, "learning_rate": 9.990817001970816e-05, "loss": 0.0434, "step": 20500 }, { "epoch": 13.07202039515615, "grad_norm": 0.574134349822998, "learning_rate": 9.990783583083825e-05, "loss": 0.0433, "step": 20510 }, { "epoch": 13.078393881453154, "grad_norm": 0.6722016930580139, "learning_rate": 9.990750103554102e-05, "loss": 0.048, "step": 20520 }, { "epoch": 13.084767367750159, "grad_norm": 0.6953365206718445, "learning_rate": 9.990716563382055e-05, "loss": 0.0494, "step": 20530 }, { "epoch": 13.091140854047163, "grad_norm": 0.4717896580696106, "learning_rate": 9.990682962568087e-05, "loss": 0.0369, "step": 20540 }, { "epoch": 13.097514340344167, "grad_norm": 0.5848617553710938, "learning_rate": 9.990649301112611e-05, "loss": 0.0479, "step": 20550 }, { "epoch": 13.103887826641174, "grad_norm": 0.7929871082305908, "learning_rate": 9.990615579016033e-05, "loss": 0.0469, "step": 20560 }, { "epoch": 13.110261312938178, "grad_norm": 0.7508995532989502, "learning_rate": 9.990581796278764e-05, "loss": 0.0497, "step": 20570 }, { "epoch": 13.116634799235182, "grad_norm": 0.6095130443572998, "learning_rate": 9.990547952901215e-05, "loss": 0.0559, "step": 20580 }, { "epoch": 13.123008285532187, "grad_norm": 0.6503527164459229, "learning_rate": 9.990514048883794e-05, "loss": 0.0453, "step": 20590 }, { "epoch": 13.129381771829191, "grad_norm": 0.5425077676773071, "learning_rate": 9.990480084226919e-05, "loss": 0.0428, "step": 20600 }, { "epoch": 13.135755258126196, "grad_norm": 1.0947233438491821, "learning_rate": 9.990446058930996e-05, "loss": 0.0582, "step": 20610 }, { "epoch": 13.1421287444232, "grad_norm": 0.7393929362297058, "learning_rate": 9.990411972996443e-05, "loss": 0.051, "step": 20620 }, { "epoch": 13.148502230720204, "grad_norm": 0.5818952918052673, "learning_rate": 9.990377826423673e-05, "loss": 0.0481, "step": 20630 }, { "epoch": 13.154875717017209, "grad_norm": 0.42319175601005554, "learning_rate": 9.990343619213101e-05, "loss": 0.0478, "step": 20640 }, { "epoch": 13.161249203314213, "grad_norm": 0.5444415807723999, "learning_rate": 9.990309351365141e-05, "loss": 0.044, "step": 20650 }, { "epoch": 13.167622689611218, "grad_norm": 0.4666403532028198, "learning_rate": 9.990275022880213e-05, "loss": 0.035, "step": 20660 }, { "epoch": 13.173996175908222, "grad_norm": 0.31161317229270935, "learning_rate": 9.990240633758731e-05, "loss": 0.0374, "step": 20670 }, { "epoch": 13.180369662205226, "grad_norm": 0.6608639359474182, "learning_rate": 9.990206184001113e-05, "loss": 0.0455, "step": 20680 }, { "epoch": 13.18674314850223, "grad_norm": 0.3203983008861542, "learning_rate": 9.990171673607777e-05, "loss": 0.0356, "step": 20690 }, { "epoch": 13.193116634799235, "grad_norm": 0.4987826645374298, "learning_rate": 9.990137102579146e-05, "loss": 0.0436, "step": 20700 }, { "epoch": 13.19949012109624, "grad_norm": 0.8196752071380615, "learning_rate": 9.990102470915637e-05, "loss": 0.0511, "step": 20710 }, { "epoch": 13.205863607393244, "grad_norm": 0.6709547638893127, "learning_rate": 9.990067778617673e-05, "loss": 0.0503, "step": 20720 }, { "epoch": 13.212237093690248, "grad_norm": 0.513703465461731, "learning_rate": 9.990033025685672e-05, "loss": 0.0438, "step": 20730 }, { "epoch": 13.218610579987253, "grad_norm": 0.49383261799812317, "learning_rate": 9.98999821212006e-05, "loss": 0.0524, "step": 20740 }, { "epoch": 13.224984066284257, "grad_norm": 0.5461325645446777, "learning_rate": 9.989963337921256e-05, "loss": 0.0616, "step": 20750 }, { "epoch": 13.231357552581262, "grad_norm": 0.34099167585372925, "learning_rate": 9.989928403089688e-05, "loss": 0.0506, "step": 20760 }, { "epoch": 13.237731038878266, "grad_norm": 0.5841230154037476, "learning_rate": 9.989893407625779e-05, "loss": 0.0579, "step": 20770 }, { "epoch": 13.24410452517527, "grad_norm": 0.7240725159645081, "learning_rate": 9.989858351529953e-05, "loss": 0.0658, "step": 20780 }, { "epoch": 13.250478011472275, "grad_norm": 0.4097904860973358, "learning_rate": 9.989823234802636e-05, "loss": 0.0559, "step": 20790 }, { "epoch": 13.25685149776928, "grad_norm": 0.5467256307601929, "learning_rate": 9.989788057444256e-05, "loss": 0.0573, "step": 20800 }, { "epoch": 13.263224984066284, "grad_norm": 0.605908989906311, "learning_rate": 9.989752819455241e-05, "loss": 0.0524, "step": 20810 }, { "epoch": 13.269598470363288, "grad_norm": 0.3435722589492798, "learning_rate": 9.989717520836017e-05, "loss": 0.0529, "step": 20820 }, { "epoch": 13.275971956660293, "grad_norm": 0.6609158515930176, "learning_rate": 9.989682161587015e-05, "loss": 0.0516, "step": 20830 }, { "epoch": 13.282345442957297, "grad_norm": 1.0408082008361816, "learning_rate": 9.989646741708662e-05, "loss": 0.0555, "step": 20840 }, { "epoch": 13.288718929254301, "grad_norm": 0.620171308517456, "learning_rate": 9.989611261201393e-05, "loss": 0.0472, "step": 20850 }, { "epoch": 13.295092415551306, "grad_norm": 0.5598809719085693, "learning_rate": 9.989575720065633e-05, "loss": 0.0462, "step": 20860 }, { "epoch": 13.30146590184831, "grad_norm": 0.6512726545333862, "learning_rate": 9.98954011830182e-05, "loss": 0.0438, "step": 20870 }, { "epoch": 13.307839388145316, "grad_norm": 0.6691747307777405, "learning_rate": 9.989504455910381e-05, "loss": 0.0564, "step": 20880 }, { "epoch": 13.31421287444232, "grad_norm": 0.4432419240474701, "learning_rate": 9.989468732891754e-05, "loss": 0.0448, "step": 20890 }, { "epoch": 13.320586360739325, "grad_norm": 0.7473397254943848, "learning_rate": 9.98943294924637e-05, "loss": 0.0459, "step": 20900 }, { "epoch": 13.32695984703633, "grad_norm": 0.8737781643867493, "learning_rate": 9.989397104974665e-05, "loss": 0.0478, "step": 20910 }, { "epoch": 13.333333333333334, "grad_norm": 0.5254886150360107, "learning_rate": 9.989361200077074e-05, "loss": 0.0544, "step": 20920 }, { "epoch": 13.339706819630338, "grad_norm": 0.836776614189148, "learning_rate": 9.989325234554034e-05, "loss": 0.0445, "step": 20930 }, { "epoch": 13.346080305927343, "grad_norm": 0.8743926286697388, "learning_rate": 9.989289208405983e-05, "loss": 0.0532, "step": 20940 }, { "epoch": 13.352453792224347, "grad_norm": 0.4669855833053589, "learning_rate": 9.989253121633356e-05, "loss": 0.0457, "step": 20950 }, { "epoch": 13.358827278521352, "grad_norm": 0.4714788794517517, "learning_rate": 9.989216974236594e-05, "loss": 0.066, "step": 20960 }, { "epoch": 13.365200764818356, "grad_norm": 0.4179537296295166, "learning_rate": 9.989180766216134e-05, "loss": 0.045, "step": 20970 }, { "epoch": 13.37157425111536, "grad_norm": 0.508573055267334, "learning_rate": 9.989144497572418e-05, "loss": 0.0499, "step": 20980 }, { "epoch": 13.377947737412365, "grad_norm": 0.450468510389328, "learning_rate": 9.989108168305885e-05, "loss": 0.0481, "step": 20990 }, { "epoch": 13.38432122370937, "grad_norm": 0.3956815004348755, "learning_rate": 9.989071778416977e-05, "loss": 0.0382, "step": 21000 }, { "epoch": 13.390694710006374, "grad_norm": 0.6461054086685181, "learning_rate": 9.989035327906138e-05, "loss": 0.0493, "step": 21010 }, { "epoch": 13.397068196303378, "grad_norm": 0.6376568078994751, "learning_rate": 9.988998816773808e-05, "loss": 0.0483, "step": 21020 }, { "epoch": 13.403441682600382, "grad_norm": 0.4341847002506256, "learning_rate": 9.988962245020433e-05, "loss": 0.0407, "step": 21030 }, { "epoch": 13.409815168897387, "grad_norm": 0.5661268830299377, "learning_rate": 9.988925612646455e-05, "loss": 0.0591, "step": 21040 }, { "epoch": 13.416188655194391, "grad_norm": 1.0918251276016235, "learning_rate": 9.988888919652322e-05, "loss": 0.0512, "step": 21050 }, { "epoch": 13.422562141491396, "grad_norm": 0.6603458523750305, "learning_rate": 9.988852166038477e-05, "loss": 0.0584, "step": 21060 }, { "epoch": 13.4289356277884, "grad_norm": 0.4041292071342468, "learning_rate": 9.988815351805369e-05, "loss": 0.055, "step": 21070 }, { "epoch": 13.435309114085404, "grad_norm": 0.8125314116477966, "learning_rate": 9.988778476953443e-05, "loss": 0.0388, "step": 21080 }, { "epoch": 13.441682600382409, "grad_norm": 1.019276738166809, "learning_rate": 9.988741541483148e-05, "loss": 0.0569, "step": 21090 }, { "epoch": 13.448056086679413, "grad_norm": 0.7024986743927002, "learning_rate": 9.988704545394936e-05, "loss": 0.0638, "step": 21100 }, { "epoch": 13.454429572976418, "grad_norm": 0.9689950942993164, "learning_rate": 9.988667488689252e-05, "loss": 0.0516, "step": 21110 }, { "epoch": 13.460803059273422, "grad_norm": 0.707746684551239, "learning_rate": 9.988630371366548e-05, "loss": 0.0425, "step": 21120 }, { "epoch": 13.467176545570426, "grad_norm": 0.5330038666725159, "learning_rate": 9.988593193427274e-05, "loss": 0.06, "step": 21130 }, { "epoch": 13.47355003186743, "grad_norm": 0.5291359424591064, "learning_rate": 9.988555954871884e-05, "loss": 0.0518, "step": 21140 }, { "epoch": 13.479923518164435, "grad_norm": 0.5407653450965881, "learning_rate": 9.988518655700828e-05, "loss": 0.0593, "step": 21150 }, { "epoch": 13.48629700446144, "grad_norm": 0.5795421004295349, "learning_rate": 9.988481295914561e-05, "loss": 0.0509, "step": 21160 }, { "epoch": 13.492670490758444, "grad_norm": 1.0260026454925537, "learning_rate": 9.988443875513535e-05, "loss": 0.0569, "step": 21170 }, { "epoch": 13.49904397705545, "grad_norm": 0.4161250591278076, "learning_rate": 9.988406394498209e-05, "loss": 0.0465, "step": 21180 }, { "epoch": 13.505417463352455, "grad_norm": 0.734157919883728, "learning_rate": 9.988368852869034e-05, "loss": 0.055, "step": 21190 }, { "epoch": 13.511790949649459, "grad_norm": 0.3861832320690155, "learning_rate": 9.988331250626467e-05, "loss": 0.0569, "step": 21200 }, { "epoch": 13.518164435946463, "grad_norm": 0.3449721038341522, "learning_rate": 9.988293587770966e-05, "loss": 0.0455, "step": 21210 }, { "epoch": 13.524537922243468, "grad_norm": 0.7911856174468994, "learning_rate": 9.988255864302988e-05, "loss": 0.0453, "step": 21220 }, { "epoch": 13.530911408540472, "grad_norm": 0.7439214587211609, "learning_rate": 9.988218080222992e-05, "loss": 0.0483, "step": 21230 }, { "epoch": 13.537284894837477, "grad_norm": 0.2943658232688904, "learning_rate": 9.988180235531437e-05, "loss": 0.0461, "step": 21240 }, { "epoch": 13.543658381134481, "grad_norm": 0.5325548648834229, "learning_rate": 9.988142330228781e-05, "loss": 0.0476, "step": 21250 }, { "epoch": 13.550031867431485, "grad_norm": 0.9646657109260559, "learning_rate": 9.988104364315487e-05, "loss": 0.0494, "step": 21260 }, { "epoch": 13.55640535372849, "grad_norm": 0.6305964589118958, "learning_rate": 9.988066337792015e-05, "loss": 0.0658, "step": 21270 }, { "epoch": 13.562778840025494, "grad_norm": 0.39909276366233826, "learning_rate": 9.988028250658829e-05, "loss": 0.0388, "step": 21280 }, { "epoch": 13.569152326322499, "grad_norm": 0.35210394859313965, "learning_rate": 9.987990102916389e-05, "loss": 0.0558, "step": 21290 }, { "epoch": 13.575525812619503, "grad_norm": 0.4933224022388458, "learning_rate": 9.98795189456516e-05, "loss": 0.0503, "step": 21300 }, { "epoch": 13.581899298916507, "grad_norm": 0.5310596227645874, "learning_rate": 9.987913625605606e-05, "loss": 0.044, "step": 21310 }, { "epoch": 13.588272785213512, "grad_norm": 0.6643886566162109, "learning_rate": 9.987875296038192e-05, "loss": 0.0424, "step": 21320 }, { "epoch": 13.594646271510516, "grad_norm": 0.4461294412612915, "learning_rate": 9.987836905863384e-05, "loss": 0.044, "step": 21330 }, { "epoch": 13.60101975780752, "grad_norm": 0.731680691242218, "learning_rate": 9.987798455081648e-05, "loss": 0.051, "step": 21340 }, { "epoch": 13.607393244104525, "grad_norm": 0.4990619421005249, "learning_rate": 9.987759943693452e-05, "loss": 0.0543, "step": 21350 }, { "epoch": 13.61376673040153, "grad_norm": 0.6637498736381531, "learning_rate": 9.987721371699264e-05, "loss": 0.0402, "step": 21360 }, { "epoch": 13.620140216698534, "grad_norm": 0.5051818490028381, "learning_rate": 9.987682739099551e-05, "loss": 0.0548, "step": 21370 }, { "epoch": 13.626513702995538, "grad_norm": 0.47658732533454895, "learning_rate": 9.987644045894784e-05, "loss": 0.04, "step": 21380 }, { "epoch": 13.632887189292543, "grad_norm": 0.9316199421882629, "learning_rate": 9.987605292085432e-05, "loss": 0.0412, "step": 21390 }, { "epoch": 13.639260675589547, "grad_norm": 0.6131792664527893, "learning_rate": 9.987566477671968e-05, "loss": 0.0662, "step": 21400 }, { "epoch": 13.645634161886552, "grad_norm": 0.7972255349159241, "learning_rate": 9.987527602654861e-05, "loss": 0.047, "step": 21410 }, { "epoch": 13.652007648183556, "grad_norm": 0.34587010741233826, "learning_rate": 9.987488667034586e-05, "loss": 0.0682, "step": 21420 }, { "epoch": 13.65838113448056, "grad_norm": 0.5820200443267822, "learning_rate": 9.987449670811612e-05, "loss": 0.0534, "step": 21430 }, { "epoch": 13.664754620777565, "grad_norm": 0.6251576542854309, "learning_rate": 9.987410613986419e-05, "loss": 0.0469, "step": 21440 }, { "epoch": 13.67112810707457, "grad_norm": 0.6655545234680176, "learning_rate": 9.987371496559477e-05, "loss": 0.0496, "step": 21450 }, { "epoch": 13.677501593371574, "grad_norm": 0.8633976578712463, "learning_rate": 9.987332318531262e-05, "loss": 0.0533, "step": 21460 }, { "epoch": 13.683875079668578, "grad_norm": 0.6545677185058594, "learning_rate": 9.98729307990225e-05, "loss": 0.0684, "step": 21470 }, { "epoch": 13.690248565965582, "grad_norm": 0.3638930916786194, "learning_rate": 9.987253780672919e-05, "loss": 0.0498, "step": 21480 }, { "epoch": 13.696622052262587, "grad_norm": 0.7899186611175537, "learning_rate": 9.987214420843746e-05, "loss": 0.0564, "step": 21490 }, { "epoch": 13.702995538559591, "grad_norm": 0.6777027249336243, "learning_rate": 9.987175000415208e-05, "loss": 0.0661, "step": 21500 }, { "epoch": 13.709369024856597, "grad_norm": 0.5900900959968567, "learning_rate": 9.987135519387787e-05, "loss": 0.0597, "step": 21510 }, { "epoch": 13.715742511153602, "grad_norm": 0.47955411672592163, "learning_rate": 9.98709597776196e-05, "loss": 0.0503, "step": 21520 }, { "epoch": 13.722115997450606, "grad_norm": 0.8849825263023376, "learning_rate": 9.987056375538206e-05, "loss": 0.0448, "step": 21530 }, { "epoch": 13.72848948374761, "grad_norm": 0.5264932513237, "learning_rate": 9.987016712717011e-05, "loss": 0.0436, "step": 21540 }, { "epoch": 13.734862970044615, "grad_norm": 0.6136640310287476, "learning_rate": 9.986976989298854e-05, "loss": 0.0493, "step": 21550 }, { "epoch": 13.74123645634162, "grad_norm": 0.6094130873680115, "learning_rate": 9.986937205284216e-05, "loss": 0.0506, "step": 21560 }, { "epoch": 13.747609942638624, "grad_norm": 0.6434550285339355, "learning_rate": 9.986897360673585e-05, "loss": 0.0473, "step": 21570 }, { "epoch": 13.753983428935628, "grad_norm": 0.4452361762523651, "learning_rate": 9.986857455467443e-05, "loss": 0.0425, "step": 21580 }, { "epoch": 13.760356915232633, "grad_norm": 0.5930310487747192, "learning_rate": 9.986817489666273e-05, "loss": 0.0474, "step": 21590 }, { "epoch": 13.766730401529637, "grad_norm": 0.5597618222236633, "learning_rate": 9.986777463270563e-05, "loss": 0.0571, "step": 21600 }, { "epoch": 13.773103887826641, "grad_norm": 0.6503089070320129, "learning_rate": 9.986737376280799e-05, "loss": 0.045, "step": 21610 }, { "epoch": 13.779477374123646, "grad_norm": 0.5297825932502747, "learning_rate": 9.986697228697467e-05, "loss": 0.0597, "step": 21620 }, { "epoch": 13.78585086042065, "grad_norm": 0.3579823970794678, "learning_rate": 9.986657020521056e-05, "loss": 0.0629, "step": 21630 }, { "epoch": 13.792224346717655, "grad_norm": 0.5140018463134766, "learning_rate": 9.986616751752052e-05, "loss": 0.0446, "step": 21640 }, { "epoch": 13.798597833014659, "grad_norm": 1.2022029161453247, "learning_rate": 9.986576422390948e-05, "loss": 0.0474, "step": 21650 }, { "epoch": 13.804971319311663, "grad_norm": 0.6232237815856934, "learning_rate": 9.986536032438232e-05, "loss": 0.0457, "step": 21660 }, { "epoch": 13.811344805608668, "grad_norm": 0.7051699161529541, "learning_rate": 9.986495581894395e-05, "loss": 0.0479, "step": 21670 }, { "epoch": 13.817718291905672, "grad_norm": 0.6805098652839661, "learning_rate": 9.986455070759929e-05, "loss": 0.0488, "step": 21680 }, { "epoch": 13.824091778202677, "grad_norm": 0.5452792644500732, "learning_rate": 9.986414499035327e-05, "loss": 0.0403, "step": 21690 }, { "epoch": 13.830465264499681, "grad_norm": 0.5022963285446167, "learning_rate": 9.986373866721079e-05, "loss": 0.0508, "step": 21700 }, { "epoch": 13.836838750796685, "grad_norm": 0.3037646412849426, "learning_rate": 9.986333173817681e-05, "loss": 0.0526, "step": 21710 }, { "epoch": 13.84321223709369, "grad_norm": 0.753787100315094, "learning_rate": 9.986292420325627e-05, "loss": 0.055, "step": 21720 }, { "epoch": 13.849585723390694, "grad_norm": 0.38492351770401, "learning_rate": 9.986251606245412e-05, "loss": 0.0507, "step": 21730 }, { "epoch": 13.855959209687699, "grad_norm": 0.5820807814598083, "learning_rate": 9.986210731577534e-05, "loss": 0.0392, "step": 21740 }, { "epoch": 13.862332695984703, "grad_norm": 0.40073642134666443, "learning_rate": 9.986169796322487e-05, "loss": 0.0366, "step": 21750 }, { "epoch": 13.868706182281707, "grad_norm": 0.4889290928840637, "learning_rate": 9.986128800480768e-05, "loss": 0.0506, "step": 21760 }, { "epoch": 13.875079668578712, "grad_norm": 0.7286251783370972, "learning_rate": 9.986087744052876e-05, "loss": 0.0491, "step": 21770 }, { "epoch": 13.881453154875716, "grad_norm": 0.31427904963493347, "learning_rate": 9.986046627039311e-05, "loss": 0.0426, "step": 21780 }, { "epoch": 13.88782664117272, "grad_norm": 0.6496139168739319, "learning_rate": 9.986005449440574e-05, "loss": 0.0569, "step": 21790 }, { "epoch": 13.894200127469727, "grad_norm": 0.7005490660667419, "learning_rate": 9.985964211257162e-05, "loss": 0.048, "step": 21800 }, { "epoch": 13.900573613766731, "grad_norm": 0.5604730844497681, "learning_rate": 9.985922912489575e-05, "loss": 0.0659, "step": 21810 }, { "epoch": 13.906947100063736, "grad_norm": 0.5922862887382507, "learning_rate": 9.98588155313832e-05, "loss": 0.0377, "step": 21820 }, { "epoch": 13.91332058636074, "grad_norm": 0.5051929950714111, "learning_rate": 9.985840133203896e-05, "loss": 0.0511, "step": 21830 }, { "epoch": 13.919694072657744, "grad_norm": 0.5479230880737305, "learning_rate": 9.985798652686807e-05, "loss": 0.0533, "step": 21840 }, { "epoch": 13.926067558954749, "grad_norm": 0.424010306596756, "learning_rate": 9.985757111587556e-05, "loss": 0.0384, "step": 21850 }, { "epoch": 13.932441045251753, "grad_norm": 0.5242854952812195, "learning_rate": 9.985715509906649e-05, "loss": 0.0451, "step": 21860 }, { "epoch": 13.938814531548758, "grad_norm": 0.6178439855575562, "learning_rate": 9.985673847644592e-05, "loss": 0.0417, "step": 21870 }, { "epoch": 13.945188017845762, "grad_norm": 0.46104398369789124, "learning_rate": 9.98563212480189e-05, "loss": 0.0351, "step": 21880 }, { "epoch": 13.951561504142767, "grad_norm": 0.5755635499954224, "learning_rate": 9.985590341379051e-05, "loss": 0.0526, "step": 21890 }, { "epoch": 13.957934990439771, "grad_norm": 0.5781133770942688, "learning_rate": 9.985548497376582e-05, "loss": 0.037, "step": 21900 }, { "epoch": 13.964308476736775, "grad_norm": 0.5582672357559204, "learning_rate": 9.98550659279499e-05, "loss": 0.0444, "step": 21910 }, { "epoch": 13.97068196303378, "grad_norm": 0.5559388399124146, "learning_rate": 9.985464627634789e-05, "loss": 0.0575, "step": 21920 }, { "epoch": 13.977055449330784, "grad_norm": 0.5852295160293579, "learning_rate": 9.985422601896484e-05, "loss": 0.0433, "step": 21930 }, { "epoch": 13.983428935627789, "grad_norm": 0.48233315348625183, "learning_rate": 9.985380515580587e-05, "loss": 0.0389, "step": 21940 }, { "epoch": 13.989802421924793, "grad_norm": 0.760404109954834, "learning_rate": 9.985338368687609e-05, "loss": 0.0446, "step": 21950 }, { "epoch": 13.996175908221797, "grad_norm": 0.4950686991214752, "learning_rate": 9.985296161218064e-05, "loss": 0.0479, "step": 21960 }, { "epoch": 14.002549394518802, "grad_norm": 0.7322604656219482, "learning_rate": 9.985253893172464e-05, "loss": 0.0448, "step": 21970 }, { "epoch": 14.008922880815806, "grad_norm": 0.5258325934410095, "learning_rate": 9.985211564551322e-05, "loss": 0.0426, "step": 21980 }, { "epoch": 14.01529636711281, "grad_norm": 0.5011958479881287, "learning_rate": 9.985169175355153e-05, "loss": 0.0378, "step": 21990 }, { "epoch": 14.021669853409815, "grad_norm": 0.4282132685184479, "learning_rate": 9.985126725584471e-05, "loss": 0.0469, "step": 22000 }, { "epoch": 14.02804333970682, "grad_norm": 0.40375763177871704, "learning_rate": 9.985084215239792e-05, "loss": 0.0473, "step": 22010 }, { "epoch": 14.034416826003824, "grad_norm": 0.443429172039032, "learning_rate": 9.985041644321635e-05, "loss": 0.0485, "step": 22020 }, { "epoch": 14.040790312300828, "grad_norm": 0.5961761474609375, "learning_rate": 9.984999012830514e-05, "loss": 0.0607, "step": 22030 }, { "epoch": 14.047163798597833, "grad_norm": 0.5841917395591736, "learning_rate": 9.984956320766949e-05, "loss": 0.0372, "step": 22040 }, { "epoch": 14.053537284894837, "grad_norm": 0.5729540586471558, "learning_rate": 9.984913568131457e-05, "loss": 0.0376, "step": 22050 }, { "epoch": 14.059910771191841, "grad_norm": 0.6272742748260498, "learning_rate": 9.98487075492456e-05, "loss": 0.0403, "step": 22060 }, { "epoch": 14.066284257488846, "grad_norm": 0.7974969148635864, "learning_rate": 9.984827881146777e-05, "loss": 0.0416, "step": 22070 }, { "epoch": 14.07265774378585, "grad_norm": 0.4142572581768036, "learning_rate": 9.984784946798628e-05, "loss": 0.0292, "step": 22080 }, { "epoch": 14.079031230082855, "grad_norm": 0.4337879717350006, "learning_rate": 9.984741951880635e-05, "loss": 0.0537, "step": 22090 }, { "epoch": 14.085404716379859, "grad_norm": 0.6759001612663269, "learning_rate": 9.984698896393321e-05, "loss": 0.0491, "step": 22100 }, { "epoch": 14.091778202676863, "grad_norm": 0.5249837636947632, "learning_rate": 9.984655780337212e-05, "loss": 0.0414, "step": 22110 }, { "epoch": 14.098151688973868, "grad_norm": 0.570059597492218, "learning_rate": 9.984612603712825e-05, "loss": 0.0597, "step": 22120 }, { "epoch": 14.104525175270874, "grad_norm": 0.39427268505096436, "learning_rate": 9.98456936652069e-05, "loss": 0.0468, "step": 22130 }, { "epoch": 14.110898661567878, "grad_norm": 0.37906205654144287, "learning_rate": 9.98452606876133e-05, "loss": 0.0539, "step": 22140 }, { "epoch": 14.117272147864883, "grad_norm": 0.2791944444179535, "learning_rate": 9.984482710435274e-05, "loss": 0.0436, "step": 22150 }, { "epoch": 14.123645634161887, "grad_norm": 0.6249826550483704, "learning_rate": 9.984439291543046e-05, "loss": 0.0383, "step": 22160 }, { "epoch": 14.130019120458892, "grad_norm": 0.4677196741104126, "learning_rate": 9.984395812085175e-05, "loss": 0.0413, "step": 22170 }, { "epoch": 14.136392606755896, "grad_norm": 0.6214688420295715, "learning_rate": 9.98435227206219e-05, "loss": 0.0365, "step": 22180 }, { "epoch": 14.1427660930529, "grad_norm": 0.5297069549560547, "learning_rate": 9.984308671474616e-05, "loss": 0.0441, "step": 22190 }, { "epoch": 14.149139579349905, "grad_norm": 0.4164693057537079, "learning_rate": 9.984265010322987e-05, "loss": 0.0379, "step": 22200 }, { "epoch": 14.15551306564691, "grad_norm": 0.7650898098945618, "learning_rate": 9.984221288607834e-05, "loss": 0.0489, "step": 22210 }, { "epoch": 14.161886551943914, "grad_norm": 0.6391487121582031, "learning_rate": 9.984177506329684e-05, "loss": 0.0514, "step": 22220 }, { "epoch": 14.168260038240918, "grad_norm": 0.5873070359230042, "learning_rate": 9.984133663489072e-05, "loss": 0.039, "step": 22230 }, { "epoch": 14.174633524537922, "grad_norm": 0.48284444212913513, "learning_rate": 9.98408976008653e-05, "loss": 0.0455, "step": 22240 }, { "epoch": 14.181007010834927, "grad_norm": 0.4807731509208679, "learning_rate": 9.984045796122591e-05, "loss": 0.0455, "step": 22250 }, { "epoch": 14.187380497131931, "grad_norm": 0.4245636761188507, "learning_rate": 9.984001771597792e-05, "loss": 0.0318, "step": 22260 }, { "epoch": 14.193753983428936, "grad_norm": 0.5932088494300842, "learning_rate": 9.983957686512663e-05, "loss": 0.0542, "step": 22270 }, { "epoch": 14.20012746972594, "grad_norm": 1.124269962310791, "learning_rate": 9.983913540867747e-05, "loss": 0.0393, "step": 22280 }, { "epoch": 14.206500956022944, "grad_norm": 0.7120072841644287, "learning_rate": 9.983869334663571e-05, "loss": 0.0616, "step": 22290 }, { "epoch": 14.212874442319949, "grad_norm": 0.3253026306629181, "learning_rate": 9.983825067900681e-05, "loss": 0.0371, "step": 22300 }, { "epoch": 14.219247928616953, "grad_norm": 0.42797449231147766, "learning_rate": 9.983780740579608e-05, "loss": 0.0477, "step": 22310 }, { "epoch": 14.225621414913958, "grad_norm": 0.4663847088813782, "learning_rate": 9.983736352700894e-05, "loss": 0.0419, "step": 22320 }, { "epoch": 14.231994901210962, "grad_norm": 0.5582894086837769, "learning_rate": 9.983691904265079e-05, "loss": 0.0398, "step": 22330 }, { "epoch": 14.238368387507967, "grad_norm": 0.4541654884815216, "learning_rate": 9.983647395272701e-05, "loss": 0.0524, "step": 22340 }, { "epoch": 14.244741873804971, "grad_norm": 0.6301718354225159, "learning_rate": 9.983602825724303e-05, "loss": 0.0372, "step": 22350 }, { "epoch": 14.251115360101975, "grad_norm": 0.5594810843467712, "learning_rate": 9.983558195620424e-05, "loss": 0.0468, "step": 22360 }, { "epoch": 14.25748884639898, "grad_norm": 0.6863843202590942, "learning_rate": 9.983513504961608e-05, "loss": 0.0383, "step": 22370 }, { "epoch": 14.263862332695984, "grad_norm": 0.9575958847999573, "learning_rate": 9.983468753748397e-05, "loss": 0.0587, "step": 22380 }, { "epoch": 14.270235818992989, "grad_norm": 0.40769726037979126, "learning_rate": 9.983423941981338e-05, "loss": 0.0341, "step": 22390 }, { "epoch": 14.276609305289993, "grad_norm": 0.39516618847846985, "learning_rate": 9.983379069660971e-05, "loss": 0.0411, "step": 22400 }, { "epoch": 14.282982791586997, "grad_norm": 0.6157066822052002, "learning_rate": 9.983334136787843e-05, "loss": 0.0678, "step": 22410 }, { "epoch": 14.289356277884002, "grad_norm": 0.583670973777771, "learning_rate": 9.983289143362503e-05, "loss": 0.0547, "step": 22420 }, { "epoch": 14.295729764181006, "grad_norm": 0.8311648368835449, "learning_rate": 9.98324408938549e-05, "loss": 0.0499, "step": 22430 }, { "epoch": 14.302103250478012, "grad_norm": 0.6779030561447144, "learning_rate": 9.98319897485736e-05, "loss": 0.0453, "step": 22440 }, { "epoch": 14.308476736775017, "grad_norm": 0.6761898398399353, "learning_rate": 9.983153799778658e-05, "loss": 0.0516, "step": 22450 }, { "epoch": 14.314850223072021, "grad_norm": 0.42458826303482056, "learning_rate": 9.98310856414993e-05, "loss": 0.0404, "step": 22460 }, { "epoch": 14.321223709369026, "grad_norm": 0.464655339717865, "learning_rate": 9.983063267971729e-05, "loss": 0.0415, "step": 22470 }, { "epoch": 14.32759719566603, "grad_norm": 0.5490202307701111, "learning_rate": 9.983017911244604e-05, "loss": 0.0465, "step": 22480 }, { "epoch": 14.333970681963034, "grad_norm": 0.6828319430351257, "learning_rate": 9.982972493969108e-05, "loss": 0.0536, "step": 22490 }, { "epoch": 14.340344168260039, "grad_norm": 0.9010695219039917, "learning_rate": 9.982927016145791e-05, "loss": 0.0432, "step": 22500 }, { "epoch": 14.346717654557043, "grad_norm": 0.6490613222122192, "learning_rate": 9.982881477775206e-05, "loss": 0.0412, "step": 22510 }, { "epoch": 14.353091140854048, "grad_norm": 0.5261130332946777, "learning_rate": 9.982835878857905e-05, "loss": 0.0377, "step": 22520 }, { "epoch": 14.359464627151052, "grad_norm": 0.2353847175836563, "learning_rate": 9.982790219394445e-05, "loss": 0.0595, "step": 22530 }, { "epoch": 14.365838113448056, "grad_norm": 0.39870527386665344, "learning_rate": 9.98274449938538e-05, "loss": 0.0473, "step": 22540 }, { "epoch": 14.37221159974506, "grad_norm": 0.8602566719055176, "learning_rate": 9.982698718831262e-05, "loss": 0.0481, "step": 22550 }, { "epoch": 14.378585086042065, "grad_norm": 0.5441707968711853, "learning_rate": 9.982652877732653e-05, "loss": 0.0464, "step": 22560 }, { "epoch": 14.38495857233907, "grad_norm": 0.42394423484802246, "learning_rate": 9.982606976090107e-05, "loss": 0.0494, "step": 22570 }, { "epoch": 14.391332058636074, "grad_norm": 0.5326185822486877, "learning_rate": 9.98256101390418e-05, "loss": 0.0399, "step": 22580 }, { "epoch": 14.397705544933078, "grad_norm": 0.33314624428749084, "learning_rate": 9.982514991175433e-05, "loss": 0.0426, "step": 22590 }, { "epoch": 14.404079031230083, "grad_norm": 0.8561475872993469, "learning_rate": 9.982468907904426e-05, "loss": 0.0408, "step": 22600 }, { "epoch": 14.410452517527087, "grad_norm": 0.635107696056366, "learning_rate": 9.982422764091717e-05, "loss": 0.0335, "step": 22610 }, { "epoch": 14.416826003824092, "grad_norm": 0.4229608178138733, "learning_rate": 9.982376559737866e-05, "loss": 0.049, "step": 22620 }, { "epoch": 14.423199490121096, "grad_norm": 0.6325505971908569, "learning_rate": 9.982330294843436e-05, "loss": 0.0345, "step": 22630 }, { "epoch": 14.4295729764181, "grad_norm": 0.4519380033016205, "learning_rate": 9.98228396940899e-05, "loss": 0.0695, "step": 22640 }, { "epoch": 14.435946462715105, "grad_norm": 0.5471426248550415, "learning_rate": 9.982237583435089e-05, "loss": 0.0514, "step": 22650 }, { "epoch": 14.44231994901211, "grad_norm": 0.5457211136817932, "learning_rate": 9.982191136922296e-05, "loss": 0.0504, "step": 22660 }, { "epoch": 14.448693435309114, "grad_norm": 0.6787260174751282, "learning_rate": 9.982144629871179e-05, "loss": 0.0448, "step": 22670 }, { "epoch": 14.455066921606118, "grad_norm": 0.46625715494155884, "learning_rate": 9.982098062282299e-05, "loss": 0.0491, "step": 22680 }, { "epoch": 14.461440407903122, "grad_norm": 1.2135145664215088, "learning_rate": 9.982051434156225e-05, "loss": 0.0526, "step": 22690 }, { "epoch": 14.467813894200127, "grad_norm": 0.8382881283760071, "learning_rate": 9.982004745493523e-05, "loss": 0.0506, "step": 22700 }, { "epoch": 14.474187380497131, "grad_norm": 0.5003844499588013, "learning_rate": 9.981957996294757e-05, "loss": 0.0487, "step": 22710 }, { "epoch": 14.480560866794136, "grad_norm": 0.45827680826187134, "learning_rate": 9.9819111865605e-05, "loss": 0.0541, "step": 22720 }, { "epoch": 14.48693435309114, "grad_norm": 0.9332799911499023, "learning_rate": 9.981864316291316e-05, "loss": 0.0494, "step": 22730 }, { "epoch": 14.493307839388144, "grad_norm": 0.7069278359413147, "learning_rate": 9.98181738548778e-05, "loss": 0.0446, "step": 22740 }, { "epoch": 14.499681325685149, "grad_norm": 0.630792498588562, "learning_rate": 9.981770394150456e-05, "loss": 0.0468, "step": 22750 }, { "epoch": 14.506054811982153, "grad_norm": 0.5244588851928711, "learning_rate": 9.98172334227992e-05, "loss": 0.066, "step": 22760 }, { "epoch": 14.51242829827916, "grad_norm": 0.42685291171073914, "learning_rate": 9.981676229876741e-05, "loss": 0.0488, "step": 22770 }, { "epoch": 14.518801784576164, "grad_norm": 0.778305172920227, "learning_rate": 9.981629056941493e-05, "loss": 0.0507, "step": 22780 }, { "epoch": 14.525175270873168, "grad_norm": 0.622222900390625, "learning_rate": 9.981581823474749e-05, "loss": 0.042, "step": 22790 }, { "epoch": 14.531548757170173, "grad_norm": 0.3114643692970276, "learning_rate": 9.98153452947708e-05, "loss": 0.038, "step": 22800 }, { "epoch": 14.537922243467177, "grad_norm": 0.434457391500473, "learning_rate": 9.981487174949065e-05, "loss": 0.0376, "step": 22810 }, { "epoch": 14.544295729764182, "grad_norm": 0.4828667938709259, "learning_rate": 9.981439759891278e-05, "loss": 0.0571, "step": 22820 }, { "epoch": 14.550669216061186, "grad_norm": 0.5315593481063843, "learning_rate": 9.981392284304292e-05, "loss": 0.0456, "step": 22830 }, { "epoch": 14.55704270235819, "grad_norm": 0.5096555948257446, "learning_rate": 9.98134474818869e-05, "loss": 0.0476, "step": 22840 }, { "epoch": 14.563416188655195, "grad_norm": 0.6486849188804626, "learning_rate": 9.981297151545044e-05, "loss": 0.0625, "step": 22850 }, { "epoch": 14.5697896749522, "grad_norm": 0.7857367992401123, "learning_rate": 9.981249494373935e-05, "loss": 0.0538, "step": 22860 }, { "epoch": 14.576163161249204, "grad_norm": 0.6177077889442444, "learning_rate": 9.981201776675942e-05, "loss": 0.0403, "step": 22870 }, { "epoch": 14.582536647546208, "grad_norm": 0.601658284664154, "learning_rate": 9.981153998451645e-05, "loss": 0.0544, "step": 22880 }, { "epoch": 14.588910133843212, "grad_norm": 0.80660480260849, "learning_rate": 9.981106159701622e-05, "loss": 0.0541, "step": 22890 }, { "epoch": 14.595283620140217, "grad_norm": 0.5773524641990662, "learning_rate": 9.981058260426456e-05, "loss": 0.0586, "step": 22900 }, { "epoch": 14.601657106437221, "grad_norm": 0.5855287313461304, "learning_rate": 9.981010300626732e-05, "loss": 0.0657, "step": 22910 }, { "epoch": 14.608030592734226, "grad_norm": 0.690717875957489, "learning_rate": 9.980962280303028e-05, "loss": 0.0527, "step": 22920 }, { "epoch": 14.61440407903123, "grad_norm": 0.7315030097961426, "learning_rate": 9.980914199455929e-05, "loss": 0.0493, "step": 22930 }, { "epoch": 14.620777565328234, "grad_norm": 0.5379112958908081, "learning_rate": 9.980866058086021e-05, "loss": 0.0535, "step": 22940 }, { "epoch": 14.627151051625239, "grad_norm": 0.5226651430130005, "learning_rate": 9.980817856193887e-05, "loss": 0.0413, "step": 22950 }, { "epoch": 14.633524537922243, "grad_norm": 0.7389031052589417, "learning_rate": 9.980769593780115e-05, "loss": 0.0648, "step": 22960 }, { "epoch": 14.639898024219248, "grad_norm": 0.7381865978240967, "learning_rate": 9.980721270845288e-05, "loss": 0.0463, "step": 22970 }, { "epoch": 14.646271510516252, "grad_norm": 0.4846133887767792, "learning_rate": 9.980672887389994e-05, "loss": 0.0415, "step": 22980 }, { "epoch": 14.652644996813256, "grad_norm": 0.48646196722984314, "learning_rate": 9.980624443414824e-05, "loss": 0.0464, "step": 22990 }, { "epoch": 14.65901848311026, "grad_norm": 0.8300036191940308, "learning_rate": 9.980575938920364e-05, "loss": 0.0415, "step": 23000 }, { "epoch": 14.665391969407265, "grad_norm": 0.5487350225448608, "learning_rate": 9.980527373907204e-05, "loss": 0.0385, "step": 23010 }, { "epoch": 14.67176545570427, "grad_norm": 0.5324499011039734, "learning_rate": 9.980478748375935e-05, "loss": 0.0383, "step": 23020 }, { "epoch": 14.678138942001274, "grad_norm": 0.7573229074478149, "learning_rate": 9.980430062327144e-05, "loss": 0.0423, "step": 23030 }, { "epoch": 14.684512428298278, "grad_norm": 0.6033254861831665, "learning_rate": 9.980381315761426e-05, "loss": 0.0513, "step": 23040 }, { "epoch": 14.690885914595283, "grad_norm": 1.2765908241271973, "learning_rate": 9.980332508679375e-05, "loss": 0.0589, "step": 23050 }, { "epoch": 14.697259400892289, "grad_norm": 0.42224645614624023, "learning_rate": 9.98028364108158e-05, "loss": 0.0374, "step": 23060 }, { "epoch": 14.703632887189293, "grad_norm": 0.4443477392196655, "learning_rate": 9.980234712968638e-05, "loss": 0.0437, "step": 23070 }, { "epoch": 14.710006373486298, "grad_norm": 0.44286397099494934, "learning_rate": 9.980185724341142e-05, "loss": 0.0426, "step": 23080 }, { "epoch": 14.716379859783302, "grad_norm": 0.3904381990432739, "learning_rate": 9.980136675199686e-05, "loss": 0.0417, "step": 23090 }, { "epoch": 14.722753346080307, "grad_norm": 0.5800367593765259, "learning_rate": 9.980087565544869e-05, "loss": 0.0498, "step": 23100 }, { "epoch": 14.729126832377311, "grad_norm": 0.6309531927108765, "learning_rate": 9.980038395377284e-05, "loss": 0.0576, "step": 23110 }, { "epoch": 14.735500318674315, "grad_norm": 0.5551360249519348, "learning_rate": 9.979989164697531e-05, "loss": 0.034, "step": 23120 }, { "epoch": 14.74187380497132, "grad_norm": 0.4391493797302246, "learning_rate": 9.97993987350621e-05, "loss": 0.0439, "step": 23130 }, { "epoch": 14.748247291268324, "grad_norm": 0.4850173592567444, "learning_rate": 9.979890521803917e-05, "loss": 0.0336, "step": 23140 }, { "epoch": 14.754620777565329, "grad_norm": 0.7128845453262329, "learning_rate": 9.97984110959125e-05, "loss": 0.0627, "step": 23150 }, { "epoch": 14.760994263862333, "grad_norm": 0.6060341000556946, "learning_rate": 9.979791636868813e-05, "loss": 0.0537, "step": 23160 }, { "epoch": 14.767367750159337, "grad_norm": 1.037618637084961, "learning_rate": 9.979742103637206e-05, "loss": 0.0354, "step": 23170 }, { "epoch": 14.773741236456342, "grad_norm": 0.6059609651565552, "learning_rate": 9.979692509897032e-05, "loss": 0.0477, "step": 23180 }, { "epoch": 14.780114722753346, "grad_norm": 1.1166725158691406, "learning_rate": 9.979642855648893e-05, "loss": 0.0542, "step": 23190 }, { "epoch": 14.78648820905035, "grad_norm": 0.3213508427143097, "learning_rate": 9.979593140893388e-05, "loss": 0.0393, "step": 23200 }, { "epoch": 14.792861695347355, "grad_norm": 0.3468857407569885, "learning_rate": 9.979543365631128e-05, "loss": 0.0438, "step": 23210 }, { "epoch": 14.79923518164436, "grad_norm": 0.6471170783042908, "learning_rate": 9.979493529862714e-05, "loss": 0.043, "step": 23220 }, { "epoch": 14.805608667941364, "grad_norm": 0.5723495483398438, "learning_rate": 9.979443633588749e-05, "loss": 0.0412, "step": 23230 }, { "epoch": 14.811982154238368, "grad_norm": 0.6439129710197449, "learning_rate": 9.979393676809847e-05, "loss": 0.0453, "step": 23240 }, { "epoch": 14.818355640535373, "grad_norm": 0.4173094928264618, "learning_rate": 9.979343659526608e-05, "loss": 0.0606, "step": 23250 }, { "epoch": 14.824729126832377, "grad_norm": 0.5355782508850098, "learning_rate": 9.979293581739644e-05, "loss": 0.0405, "step": 23260 }, { "epoch": 14.831102613129381, "grad_norm": 0.5931435227394104, "learning_rate": 9.97924344344956e-05, "loss": 0.0375, "step": 23270 }, { "epoch": 14.837476099426386, "grad_norm": 0.5154598951339722, "learning_rate": 9.979193244656968e-05, "loss": 0.0489, "step": 23280 }, { "epoch": 14.84384958572339, "grad_norm": 0.9111235737800598, "learning_rate": 9.979142985362478e-05, "loss": 0.0434, "step": 23290 }, { "epoch": 14.850223072020395, "grad_norm": 0.8295645713806152, "learning_rate": 9.979092665566697e-05, "loss": 0.0525, "step": 23300 }, { "epoch": 14.8565965583174, "grad_norm": 0.38093316555023193, "learning_rate": 9.97904228527024e-05, "loss": 0.0491, "step": 23310 }, { "epoch": 14.862970044614404, "grad_norm": 0.7697515487670898, "learning_rate": 9.978991844473719e-05, "loss": 0.0538, "step": 23320 }, { "epoch": 14.869343530911408, "grad_norm": 0.5359776616096497, "learning_rate": 9.978941343177746e-05, "loss": 0.0512, "step": 23330 }, { "epoch": 14.875717017208412, "grad_norm": 0.632086992263794, "learning_rate": 9.978890781382935e-05, "loss": 0.0535, "step": 23340 }, { "epoch": 14.882090503505417, "grad_norm": 0.40464523434638977, "learning_rate": 9.978840159089899e-05, "loss": 0.0458, "step": 23350 }, { "epoch": 14.888463989802421, "grad_norm": 0.3309280276298523, "learning_rate": 9.978789476299255e-05, "loss": 0.0448, "step": 23360 }, { "epoch": 14.894837476099426, "grad_norm": 0.4444672167301178, "learning_rate": 9.97873873301162e-05, "loss": 0.0565, "step": 23370 }, { "epoch": 14.90121096239643, "grad_norm": 0.5160964727401733, "learning_rate": 9.978687929227606e-05, "loss": 0.0502, "step": 23380 }, { "epoch": 14.907584448693436, "grad_norm": 0.5577272176742554, "learning_rate": 9.978637064947834e-05, "loss": 0.0502, "step": 23390 }, { "epoch": 14.91395793499044, "grad_norm": 0.7256981134414673, "learning_rate": 9.97858614017292e-05, "loss": 0.0492, "step": 23400 }, { "epoch": 14.920331421287445, "grad_norm": 0.6995009183883667, "learning_rate": 9.978535154903484e-05, "loss": 0.0585, "step": 23410 }, { "epoch": 14.92670490758445, "grad_norm": 0.5200843214988708, "learning_rate": 9.978484109140147e-05, "loss": 0.0391, "step": 23420 }, { "epoch": 14.933078393881454, "grad_norm": 0.6342918872833252, "learning_rate": 9.978433002883527e-05, "loss": 0.042, "step": 23430 }, { "epoch": 14.939451880178458, "grad_norm": 0.45928239822387695, "learning_rate": 9.978381836134244e-05, "loss": 0.0388, "step": 23440 }, { "epoch": 14.945825366475463, "grad_norm": 0.45357558131217957, "learning_rate": 9.978330608892923e-05, "loss": 0.0376, "step": 23450 }, { "epoch": 14.952198852772467, "grad_norm": 0.5008507370948792, "learning_rate": 9.978279321160184e-05, "loss": 0.0428, "step": 23460 }, { "epoch": 14.958572339069471, "grad_norm": 0.5462914705276489, "learning_rate": 9.978227972936652e-05, "loss": 0.0446, "step": 23470 }, { "epoch": 14.964945825366476, "grad_norm": 0.9730573892593384, "learning_rate": 9.978176564222948e-05, "loss": 0.0544, "step": 23480 }, { "epoch": 14.97131931166348, "grad_norm": 0.37451326847076416, "learning_rate": 9.978125095019701e-05, "loss": 0.0389, "step": 23490 }, { "epoch": 14.977692797960485, "grad_norm": 0.37431076169013977, "learning_rate": 9.978073565327533e-05, "loss": 0.0432, "step": 23500 }, { "epoch": 14.984066284257489, "grad_norm": 0.49148106575012207, "learning_rate": 9.978021975147069e-05, "loss": 0.0635, "step": 23510 }, { "epoch": 14.990439770554493, "grad_norm": 0.4281140863895416, "learning_rate": 9.97797032447894e-05, "loss": 0.0526, "step": 23520 }, { "epoch": 14.996813256851498, "grad_norm": 0.6121013164520264, "learning_rate": 9.977918613323771e-05, "loss": 0.0539, "step": 23530 }, { "epoch": 15.003186743148502, "grad_norm": 0.36562085151672363, "learning_rate": 9.977866841682192e-05, "loss": 0.0473, "step": 23540 }, { "epoch": 15.009560229445507, "grad_norm": 0.4976012110710144, "learning_rate": 9.97781500955483e-05, "loss": 0.0468, "step": 23550 }, { "epoch": 15.015933715742511, "grad_norm": 0.6325187683105469, "learning_rate": 9.977763116942315e-05, "loss": 0.0391, "step": 23560 }, { "epoch": 15.022307202039515, "grad_norm": 0.3797138035297394, "learning_rate": 9.97771116384528e-05, "loss": 0.0455, "step": 23570 }, { "epoch": 15.02868068833652, "grad_norm": 0.4993521273136139, "learning_rate": 9.977659150264355e-05, "loss": 0.0347, "step": 23580 }, { "epoch": 15.035054174633524, "grad_norm": 1.0263770818710327, "learning_rate": 9.97760707620017e-05, "loss": 0.0446, "step": 23590 }, { "epoch": 15.041427660930529, "grad_norm": 0.5573161244392395, "learning_rate": 9.97755494165336e-05, "loss": 0.0519, "step": 23600 }, { "epoch": 15.047801147227533, "grad_norm": 0.923227846622467, "learning_rate": 9.977502746624559e-05, "loss": 0.0532, "step": 23610 }, { "epoch": 15.054174633524537, "grad_norm": 0.9130227565765381, "learning_rate": 9.977450491114398e-05, "loss": 0.041, "step": 23620 }, { "epoch": 15.060548119821542, "grad_norm": 0.36806720495224, "learning_rate": 9.977398175123515e-05, "loss": 0.0395, "step": 23630 }, { "epoch": 15.066921606118546, "grad_norm": 0.4509584903717041, "learning_rate": 9.977345798652545e-05, "loss": 0.04, "step": 23640 }, { "epoch": 15.07329509241555, "grad_norm": 0.7122308611869812, "learning_rate": 9.977293361702125e-05, "loss": 0.0573, "step": 23650 }, { "epoch": 15.079668578712555, "grad_norm": 0.410665363073349, "learning_rate": 9.97724086427289e-05, "loss": 0.0452, "step": 23660 }, { "epoch": 15.08604206500956, "grad_norm": 0.3860701024532318, "learning_rate": 9.97718830636548e-05, "loss": 0.0413, "step": 23670 }, { "epoch": 15.092415551306564, "grad_norm": 0.33533331751823425, "learning_rate": 9.977135687980531e-05, "loss": 0.0471, "step": 23680 }, { "epoch": 15.098789037603568, "grad_norm": 0.5211670994758606, "learning_rate": 9.977083009118686e-05, "loss": 0.051, "step": 23690 }, { "epoch": 15.105162523900574, "grad_norm": 0.3898063004016876, "learning_rate": 9.977030269780583e-05, "loss": 0.0326, "step": 23700 }, { "epoch": 15.111536010197579, "grad_norm": 0.42077144980430603, "learning_rate": 9.976977469966862e-05, "loss": 0.0427, "step": 23710 }, { "epoch": 15.117909496494583, "grad_norm": 0.4900611340999603, "learning_rate": 9.976924609678167e-05, "loss": 0.0383, "step": 23720 }, { "epoch": 15.124282982791588, "grad_norm": 0.49385595321655273, "learning_rate": 9.976871688915139e-05, "loss": 0.0364, "step": 23730 }, { "epoch": 15.130656469088592, "grad_norm": 0.587699830532074, "learning_rate": 9.97681870767842e-05, "loss": 0.053, "step": 23740 }, { "epoch": 15.137029955385596, "grad_norm": 0.6034412384033203, "learning_rate": 9.976765665968654e-05, "loss": 0.0388, "step": 23750 }, { "epoch": 15.1434034416826, "grad_norm": 0.5031957030296326, "learning_rate": 9.97671256378649e-05, "loss": 0.0479, "step": 23760 }, { "epoch": 15.149776927979605, "grad_norm": 0.44968873262405396, "learning_rate": 9.976659401132566e-05, "loss": 0.0414, "step": 23770 }, { "epoch": 15.15615041427661, "grad_norm": 0.4617210030555725, "learning_rate": 9.976606178007532e-05, "loss": 0.0368, "step": 23780 }, { "epoch": 15.162523900573614, "grad_norm": 1.0046274662017822, "learning_rate": 9.976552894412037e-05, "loss": 0.0488, "step": 23790 }, { "epoch": 15.168897386870619, "grad_norm": 0.665860116481781, "learning_rate": 9.976499550346723e-05, "loss": 0.0443, "step": 23800 }, { "epoch": 15.175270873167623, "grad_norm": 0.5657275915145874, "learning_rate": 9.976446145812242e-05, "loss": 0.0307, "step": 23810 }, { "epoch": 15.181644359464627, "grad_norm": 0.4063440263271332, "learning_rate": 9.976392680809242e-05, "loss": 0.0431, "step": 23820 }, { "epoch": 15.188017845761632, "grad_norm": 0.5577939748764038, "learning_rate": 9.976339155338371e-05, "loss": 0.0368, "step": 23830 }, { "epoch": 15.194391332058636, "grad_norm": 0.4961129426956177, "learning_rate": 9.976285569400281e-05, "loss": 0.0439, "step": 23840 }, { "epoch": 15.20076481835564, "grad_norm": 0.5882431864738464, "learning_rate": 9.976231922995624e-05, "loss": 0.0441, "step": 23850 }, { "epoch": 15.207138304652645, "grad_norm": 0.8021961450576782, "learning_rate": 9.97617821612505e-05, "loss": 0.0448, "step": 23860 }, { "epoch": 15.21351179094965, "grad_norm": 0.7422100901603699, "learning_rate": 9.976124448789214e-05, "loss": 0.0447, "step": 23870 }, { "epoch": 15.219885277246654, "grad_norm": 0.5307670831680298, "learning_rate": 9.976070620988767e-05, "loss": 0.0415, "step": 23880 }, { "epoch": 15.226258763543658, "grad_norm": 0.5787435173988342, "learning_rate": 9.976016732724362e-05, "loss": 0.0328, "step": 23890 }, { "epoch": 15.232632249840663, "grad_norm": 0.6310264468193054, "learning_rate": 9.975962783996657e-05, "loss": 0.0503, "step": 23900 }, { "epoch": 15.239005736137667, "grad_norm": 0.6409702301025391, "learning_rate": 9.975908774806307e-05, "loss": 0.0436, "step": 23910 }, { "epoch": 15.245379222434671, "grad_norm": 0.7289593815803528, "learning_rate": 9.975854705153965e-05, "loss": 0.0417, "step": 23920 }, { "epoch": 15.251752708731676, "grad_norm": 0.5920099020004272, "learning_rate": 9.975800575040291e-05, "loss": 0.0494, "step": 23930 }, { "epoch": 15.25812619502868, "grad_norm": 0.42223116755485535, "learning_rate": 9.975746384465943e-05, "loss": 0.0367, "step": 23940 }, { "epoch": 15.264499681325685, "grad_norm": 0.3632909655570984, "learning_rate": 9.975692133431579e-05, "loss": 0.0353, "step": 23950 }, { "epoch": 15.270873167622689, "grad_norm": 0.546451985836029, "learning_rate": 9.975637821937857e-05, "loss": 0.0431, "step": 23960 }, { "epoch": 15.277246653919693, "grad_norm": 0.5947668552398682, "learning_rate": 9.975583449985437e-05, "loss": 0.0534, "step": 23970 }, { "epoch": 15.283620140216698, "grad_norm": 0.8924487233161926, "learning_rate": 9.97552901757498e-05, "loss": 0.0435, "step": 23980 }, { "epoch": 15.289993626513702, "grad_norm": 0.5419203042984009, "learning_rate": 9.97547452470715e-05, "loss": 0.0456, "step": 23990 }, { "epoch": 15.296367112810707, "grad_norm": 0.46796733140945435, "learning_rate": 9.975419971382606e-05, "loss": 0.0476, "step": 24000 }, { "epoch": 15.302740599107711, "grad_norm": 0.5434220433235168, "learning_rate": 9.97536535760201e-05, "loss": 0.0372, "step": 24010 }, { "epoch": 15.309114085404717, "grad_norm": 0.5641089081764221, "learning_rate": 9.97531068336603e-05, "loss": 0.0387, "step": 24020 }, { "epoch": 15.315487571701722, "grad_norm": 0.48083871603012085, "learning_rate": 9.975255948675326e-05, "loss": 0.0379, "step": 24030 }, { "epoch": 15.321861057998726, "grad_norm": 0.5026900172233582, "learning_rate": 9.975201153530566e-05, "loss": 0.0353, "step": 24040 }, { "epoch": 15.32823454429573, "grad_norm": 0.5611559152603149, "learning_rate": 9.975146297932414e-05, "loss": 0.054, "step": 24050 }, { "epoch": 15.334608030592735, "grad_norm": 0.4676796495914459, "learning_rate": 9.975091381881536e-05, "loss": 0.0449, "step": 24060 }, { "epoch": 15.34098151688974, "grad_norm": 0.5819118022918701, "learning_rate": 9.975036405378602e-05, "loss": 0.0488, "step": 24070 }, { "epoch": 15.347355003186744, "grad_norm": 0.870334804058075, "learning_rate": 9.974981368424278e-05, "loss": 0.047, "step": 24080 }, { "epoch": 15.353728489483748, "grad_norm": 0.3387395441532135, "learning_rate": 9.974926271019234e-05, "loss": 0.0528, "step": 24090 }, { "epoch": 15.360101975780752, "grad_norm": 0.4962310194969177, "learning_rate": 9.974871113164137e-05, "loss": 0.0517, "step": 24100 }, { "epoch": 15.366475462077757, "grad_norm": 0.7542265057563782, "learning_rate": 9.97481589485966e-05, "loss": 0.0435, "step": 24110 }, { "epoch": 15.372848948374761, "grad_norm": 0.8023645281791687, "learning_rate": 9.974760616106473e-05, "loss": 0.0349, "step": 24120 }, { "epoch": 15.379222434671766, "grad_norm": 0.6609359383583069, "learning_rate": 9.974705276905248e-05, "loss": 0.055, "step": 24130 }, { "epoch": 15.38559592096877, "grad_norm": 0.4856813848018646, "learning_rate": 9.974649877256656e-05, "loss": 0.041, "step": 24140 }, { "epoch": 15.391969407265774, "grad_norm": 0.4110677242279053, "learning_rate": 9.974594417161371e-05, "loss": 0.0521, "step": 24150 }, { "epoch": 15.398342893562779, "grad_norm": 0.5060667395591736, "learning_rate": 9.974538896620067e-05, "loss": 0.0448, "step": 24160 }, { "epoch": 15.404716379859783, "grad_norm": 0.3165733516216278, "learning_rate": 9.97448331563342e-05, "loss": 0.0339, "step": 24170 }, { "epoch": 15.411089866156788, "grad_norm": 0.39315980672836304, "learning_rate": 9.974427674202102e-05, "loss": 0.0491, "step": 24180 }, { "epoch": 15.417463352453792, "grad_norm": 0.6105425953865051, "learning_rate": 9.974371972326792e-05, "loss": 0.0387, "step": 24190 }, { "epoch": 15.423836838750796, "grad_norm": 0.44018280506134033, "learning_rate": 9.974316210008167e-05, "loss": 0.0332, "step": 24200 }, { "epoch": 15.4302103250478, "grad_norm": 0.6559227705001831, "learning_rate": 9.974260387246903e-05, "loss": 0.0459, "step": 24210 }, { "epoch": 15.436583811344805, "grad_norm": 0.6357271671295166, "learning_rate": 9.974204504043679e-05, "loss": 0.0457, "step": 24220 }, { "epoch": 15.44295729764181, "grad_norm": 0.34703901410102844, "learning_rate": 9.974148560399173e-05, "loss": 0.0412, "step": 24230 }, { "epoch": 15.449330783938814, "grad_norm": 0.37697798013687134, "learning_rate": 9.974092556314065e-05, "loss": 0.0466, "step": 24240 }, { "epoch": 15.455704270235818, "grad_norm": 0.8878664970397949, "learning_rate": 9.974036491789038e-05, "loss": 0.045, "step": 24250 }, { "epoch": 15.462077756532823, "grad_norm": 0.44537651538848877, "learning_rate": 9.973980366824772e-05, "loss": 0.0466, "step": 24260 }, { "epoch": 15.468451242829827, "grad_norm": 0.42828378081321716, "learning_rate": 9.973924181421945e-05, "loss": 0.0437, "step": 24270 }, { "epoch": 15.474824729126832, "grad_norm": 0.4945051074028015, "learning_rate": 9.973867935581245e-05, "loss": 0.0473, "step": 24280 }, { "epoch": 15.481198215423836, "grad_norm": 0.4891318380832672, "learning_rate": 9.973811629303354e-05, "loss": 0.0377, "step": 24290 }, { "epoch": 15.48757170172084, "grad_norm": 0.4379228353500366, "learning_rate": 9.973755262588955e-05, "loss": 0.058, "step": 24300 }, { "epoch": 15.493945188017845, "grad_norm": 0.5696465373039246, "learning_rate": 9.973698835438733e-05, "loss": 0.0434, "step": 24310 }, { "epoch": 15.500318674314851, "grad_norm": 0.40631696581840515, "learning_rate": 9.973642347853374e-05, "loss": 0.0478, "step": 24320 }, { "epoch": 15.506692160611856, "grad_norm": 0.26752984523773193, "learning_rate": 9.973585799833568e-05, "loss": 0.0517, "step": 24330 }, { "epoch": 15.51306564690886, "grad_norm": 0.9819962382316589, "learning_rate": 9.973529191379993e-05, "loss": 0.0402, "step": 24340 }, { "epoch": 15.519439133205864, "grad_norm": 0.5621557235717773, "learning_rate": 9.973472522493347e-05, "loss": 0.0516, "step": 24350 }, { "epoch": 15.525812619502869, "grad_norm": 0.5048269629478455, "learning_rate": 9.973415793174313e-05, "loss": 0.0451, "step": 24360 }, { "epoch": 15.532186105799873, "grad_norm": 0.5996031761169434, "learning_rate": 9.97335900342358e-05, "loss": 0.0364, "step": 24370 }, { "epoch": 15.538559592096878, "grad_norm": 0.5816421508789062, "learning_rate": 9.973302153241841e-05, "loss": 0.0367, "step": 24380 }, { "epoch": 15.544933078393882, "grad_norm": 0.3382156193256378, "learning_rate": 9.973245242629786e-05, "loss": 0.0352, "step": 24390 }, { "epoch": 15.551306564690886, "grad_norm": 0.5436844229698181, "learning_rate": 9.973188271588103e-05, "loss": 0.0402, "step": 24400 }, { "epoch": 15.55768005098789, "grad_norm": 0.5301566123962402, "learning_rate": 9.973131240117489e-05, "loss": 0.0482, "step": 24410 }, { "epoch": 15.564053537284895, "grad_norm": 0.7537147998809814, "learning_rate": 9.973074148218636e-05, "loss": 0.0433, "step": 24420 }, { "epoch": 15.5704270235819, "grad_norm": 0.5716974139213562, "learning_rate": 9.973016995892234e-05, "loss": 0.0534, "step": 24430 }, { "epoch": 15.576800509878904, "grad_norm": 0.5674687623977661, "learning_rate": 9.972959783138981e-05, "loss": 0.043, "step": 24440 }, { "epoch": 15.583173996175908, "grad_norm": 0.8504915237426758, "learning_rate": 9.972902509959572e-05, "loss": 0.0438, "step": 24450 }, { "epoch": 15.589547482472913, "grad_norm": 0.493272066116333, "learning_rate": 9.972845176354703e-05, "loss": 0.0493, "step": 24460 }, { "epoch": 15.595920968769917, "grad_norm": 0.683525800704956, "learning_rate": 9.972787782325068e-05, "loss": 0.0425, "step": 24470 }, { "epoch": 15.602294455066922, "grad_norm": 0.32319724559783936, "learning_rate": 9.97273032787137e-05, "loss": 0.0404, "step": 24480 }, { "epoch": 15.608667941363926, "grad_norm": 0.6681970357894897, "learning_rate": 9.972672812994299e-05, "loss": 0.0476, "step": 24490 }, { "epoch": 15.61504142766093, "grad_norm": 0.7049742341041565, "learning_rate": 9.97261523769456e-05, "loss": 0.0438, "step": 24500 }, { "epoch": 15.621414913957935, "grad_norm": 0.34148627519607544, "learning_rate": 9.972557601972852e-05, "loss": 0.0418, "step": 24510 }, { "epoch": 15.62778840025494, "grad_norm": 0.4473116397857666, "learning_rate": 9.972499905829875e-05, "loss": 0.0486, "step": 24520 }, { "epoch": 15.634161886551944, "grad_norm": 0.4250977635383606, "learning_rate": 9.972442149266326e-05, "loss": 0.0367, "step": 24530 }, { "epoch": 15.640535372848948, "grad_norm": 0.3553280830383301, "learning_rate": 9.972384332282914e-05, "loss": 0.0475, "step": 24540 }, { "epoch": 15.646908859145952, "grad_norm": 0.4400824308395386, "learning_rate": 9.972326454880334e-05, "loss": 0.0366, "step": 24550 }, { "epoch": 15.653282345442957, "grad_norm": 0.49200350046157837, "learning_rate": 9.972268517059296e-05, "loss": 0.0407, "step": 24560 }, { "epoch": 15.659655831739961, "grad_norm": 0.4620760679244995, "learning_rate": 9.9722105188205e-05, "loss": 0.0362, "step": 24570 }, { "epoch": 15.666029318036966, "grad_norm": 0.36829373240470886, "learning_rate": 9.97215246016465e-05, "loss": 0.034, "step": 24580 }, { "epoch": 15.67240280433397, "grad_norm": 0.8021905422210693, "learning_rate": 9.972094341092455e-05, "loss": 0.0472, "step": 24590 }, { "epoch": 15.678776290630974, "grad_norm": 0.2785084545612335, "learning_rate": 9.972036161604618e-05, "loss": 0.0683, "step": 24600 }, { "epoch": 15.685149776927979, "grad_norm": 0.35500892996788025, "learning_rate": 9.971977921701848e-05, "loss": 0.0422, "step": 24610 }, { "epoch": 15.691523263224983, "grad_norm": 1.3361483812332153, "learning_rate": 9.971919621384851e-05, "loss": 0.049, "step": 24620 }, { "epoch": 15.697896749521988, "grad_norm": 0.5870903134346008, "learning_rate": 9.971861260654338e-05, "loss": 0.0439, "step": 24630 }, { "epoch": 15.704270235818992, "grad_norm": 0.7056443691253662, "learning_rate": 9.971802839511015e-05, "loss": 0.0565, "step": 24640 }, { "epoch": 15.710643722115998, "grad_norm": 0.4484405517578125, "learning_rate": 9.971744357955595e-05, "loss": 0.037, "step": 24650 }, { "epoch": 15.717017208413003, "grad_norm": 0.5729345083236694, "learning_rate": 9.971685815988786e-05, "loss": 0.0411, "step": 24660 }, { "epoch": 15.723390694710007, "grad_norm": 0.5862831473350525, "learning_rate": 9.9716272136113e-05, "loss": 0.0384, "step": 24670 }, { "epoch": 15.729764181007011, "grad_norm": 0.27969372272491455, "learning_rate": 9.971568550823849e-05, "loss": 0.0464, "step": 24680 }, { "epoch": 15.736137667304016, "grad_norm": 0.6362307667732239, "learning_rate": 9.971509827627147e-05, "loss": 0.0463, "step": 24690 }, { "epoch": 15.74251115360102, "grad_norm": 0.5723720788955688, "learning_rate": 9.971451044021906e-05, "loss": 0.0403, "step": 24700 }, { "epoch": 15.748884639898025, "grad_norm": 0.41518616676330566, "learning_rate": 9.971392200008842e-05, "loss": 0.0571, "step": 24710 }, { "epoch": 15.755258126195029, "grad_norm": 0.43931353092193604, "learning_rate": 9.971333295588667e-05, "loss": 0.0397, "step": 24720 }, { "epoch": 15.761631612492033, "grad_norm": 0.6646326184272766, "learning_rate": 9.971274330762102e-05, "loss": 0.0465, "step": 24730 }, { "epoch": 15.768005098789038, "grad_norm": 0.8012535572052002, "learning_rate": 9.971215305529858e-05, "loss": 0.0484, "step": 24740 }, { "epoch": 15.774378585086042, "grad_norm": 0.6311519742012024, "learning_rate": 9.971156219892654e-05, "loss": 0.0537, "step": 24750 }, { "epoch": 15.780752071383047, "grad_norm": 0.6525386571884155, "learning_rate": 9.97109707385121e-05, "loss": 0.0413, "step": 24760 }, { "epoch": 15.787125557680051, "grad_norm": 0.6937606334686279, "learning_rate": 9.971037867406241e-05, "loss": 0.0571, "step": 24770 }, { "epoch": 15.793499043977056, "grad_norm": 0.2991196811199188, "learning_rate": 9.97097860055847e-05, "loss": 0.0422, "step": 24780 }, { "epoch": 15.79987253027406, "grad_norm": 0.5672038793563843, "learning_rate": 9.970919273308615e-05, "loss": 0.0359, "step": 24790 }, { "epoch": 15.806246016571064, "grad_norm": 0.4974217116832733, "learning_rate": 9.970859885657399e-05, "loss": 0.0447, "step": 24800 }, { "epoch": 15.812619502868069, "grad_norm": 0.4844803512096405, "learning_rate": 9.970800437605541e-05, "loss": 0.0373, "step": 24810 }, { "epoch": 15.818992989165073, "grad_norm": 0.5815979242324829, "learning_rate": 9.970740929153763e-05, "loss": 0.0498, "step": 24820 }, { "epoch": 15.825366475462078, "grad_norm": 0.24050360918045044, "learning_rate": 9.970681360302791e-05, "loss": 0.0433, "step": 24830 }, { "epoch": 15.831739961759082, "grad_norm": 0.32661160826683044, "learning_rate": 9.970621731053348e-05, "loss": 0.0349, "step": 24840 }, { "epoch": 15.838113448056086, "grad_norm": 0.37772035598754883, "learning_rate": 9.970562041406158e-05, "loss": 0.052, "step": 24850 }, { "epoch": 15.84448693435309, "grad_norm": 0.6252054572105408, "learning_rate": 9.970502291361946e-05, "loss": 0.0541, "step": 24860 }, { "epoch": 15.850860420650095, "grad_norm": 0.6228477954864502, "learning_rate": 9.970442480921436e-05, "loss": 0.0363, "step": 24870 }, { "epoch": 15.8572339069471, "grad_norm": 0.6973903775215149, "learning_rate": 9.97038261008536e-05, "loss": 0.0499, "step": 24880 }, { "epoch": 15.863607393244104, "grad_norm": 0.413540244102478, "learning_rate": 9.97032267885444e-05, "loss": 0.0475, "step": 24890 }, { "epoch": 15.869980879541108, "grad_norm": 0.4371397793292999, "learning_rate": 9.970262687229408e-05, "loss": 0.0417, "step": 24900 }, { "epoch": 15.876354365838113, "grad_norm": 0.41675591468811035, "learning_rate": 9.970202635210992e-05, "loss": 0.0462, "step": 24910 }, { "epoch": 15.882727852135117, "grad_norm": 0.4219648540019989, "learning_rate": 9.970142522799921e-05, "loss": 0.0377, "step": 24920 }, { "epoch": 15.889101338432122, "grad_norm": 0.6078414916992188, "learning_rate": 9.970082349996925e-05, "loss": 0.0325, "step": 24930 }, { "epoch": 15.895474824729128, "grad_norm": 0.4558377265930176, "learning_rate": 9.970022116802737e-05, "loss": 0.0394, "step": 24940 }, { "epoch": 15.901848311026132, "grad_norm": 0.7134793996810913, "learning_rate": 9.969961823218087e-05, "loss": 0.0376, "step": 24950 }, { "epoch": 15.908221797323137, "grad_norm": 0.35169529914855957, "learning_rate": 9.969901469243708e-05, "loss": 0.043, "step": 24960 }, { "epoch": 15.914595283620141, "grad_norm": 0.5167200565338135, "learning_rate": 9.969841054880336e-05, "loss": 0.0385, "step": 24970 }, { "epoch": 15.920968769917145, "grad_norm": 0.6052188873291016, "learning_rate": 9.9697805801287e-05, "loss": 0.0434, "step": 24980 }, { "epoch": 15.92734225621415, "grad_norm": 0.6890076398849487, "learning_rate": 9.96972004498954e-05, "loss": 0.0431, "step": 24990 }, { "epoch": 15.933715742511154, "grad_norm": 0.8521940112113953, "learning_rate": 9.969659449463588e-05, "loss": 0.0443, "step": 25000 }, { "epoch": 15.940089228808159, "grad_norm": 0.6303184628486633, "learning_rate": 9.969598793551582e-05, "loss": 0.0484, "step": 25010 }, { "epoch": 15.946462715105163, "grad_norm": 0.6598095297813416, "learning_rate": 9.969538077254259e-05, "loss": 0.0392, "step": 25020 }, { "epoch": 15.952836201402167, "grad_norm": 0.4992329478263855, "learning_rate": 9.969477300572356e-05, "loss": 0.0527, "step": 25030 }, { "epoch": 15.959209687699172, "grad_norm": 0.45346710085868835, "learning_rate": 9.969416463506612e-05, "loss": 0.0496, "step": 25040 }, { "epoch": 15.965583173996176, "grad_norm": 0.5461167097091675, "learning_rate": 9.969355566057766e-05, "loss": 0.0479, "step": 25050 }, { "epoch": 15.97195666029318, "grad_norm": 0.60860675573349, "learning_rate": 9.969294608226556e-05, "loss": 0.0443, "step": 25060 }, { "epoch": 15.978330146590185, "grad_norm": 0.5992637276649475, "learning_rate": 9.969233590013727e-05, "loss": 0.0436, "step": 25070 }, { "epoch": 15.98470363288719, "grad_norm": 0.9202326536178589, "learning_rate": 9.969172511420017e-05, "loss": 0.0403, "step": 25080 }, { "epoch": 15.991077119184194, "grad_norm": 0.5138540267944336, "learning_rate": 9.969111372446171e-05, "loss": 0.0446, "step": 25090 }, { "epoch": 15.997450605481198, "grad_norm": 0.5955187678337097, "learning_rate": 9.969050173092928e-05, "loss": 0.0496, "step": 25100 }, { "epoch": 16.003824091778203, "grad_norm": 0.5476959347724915, "learning_rate": 9.968988913361035e-05, "loss": 0.0444, "step": 25110 }, { "epoch": 16.010197578075207, "grad_norm": 0.37764671444892883, "learning_rate": 9.968927593251235e-05, "loss": 0.0352, "step": 25120 }, { "epoch": 16.01657106437221, "grad_norm": 0.5696294903755188, "learning_rate": 9.968866212764274e-05, "loss": 0.0455, "step": 25130 }, { "epoch": 16.022944550669216, "grad_norm": 0.41611307859420776, "learning_rate": 9.968804771900897e-05, "loss": 0.0412, "step": 25140 }, { "epoch": 16.02931803696622, "grad_norm": 0.4574086368083954, "learning_rate": 9.96874327066185e-05, "loss": 0.0456, "step": 25150 }, { "epoch": 16.035691523263225, "grad_norm": 0.8568750619888306, "learning_rate": 9.968681709047881e-05, "loss": 0.0476, "step": 25160 }, { "epoch": 16.04206500956023, "grad_norm": 0.5008062124252319, "learning_rate": 9.968620087059738e-05, "loss": 0.0274, "step": 25170 }, { "epoch": 16.048438495857233, "grad_norm": 0.5712280869483948, "learning_rate": 9.968558404698171e-05, "loss": 0.0426, "step": 25180 }, { "epoch": 16.054811982154238, "grad_norm": 0.4087081849575043, "learning_rate": 9.968496661963926e-05, "loss": 0.0454, "step": 25190 }, { "epoch": 16.061185468451242, "grad_norm": 0.4734458327293396, "learning_rate": 9.968434858857758e-05, "loss": 0.0366, "step": 25200 }, { "epoch": 16.067558954748247, "grad_norm": 0.4051951467990875, "learning_rate": 9.968372995380415e-05, "loss": 0.0343, "step": 25210 }, { "epoch": 16.07393244104525, "grad_norm": 0.49625903367996216, "learning_rate": 9.968311071532647e-05, "loss": 0.0421, "step": 25220 }, { "epoch": 16.080305927342256, "grad_norm": 0.8067222833633423, "learning_rate": 9.968249087315212e-05, "loss": 0.0388, "step": 25230 }, { "epoch": 16.08667941363926, "grad_norm": 0.5366131067276001, "learning_rate": 9.968187042728857e-05, "loss": 0.0491, "step": 25240 }, { "epoch": 16.093052899936264, "grad_norm": 0.5047993659973145, "learning_rate": 9.968124937774339e-05, "loss": 0.0385, "step": 25250 }, { "epoch": 16.09942638623327, "grad_norm": 0.5189358592033386, "learning_rate": 9.968062772452413e-05, "loss": 0.0323, "step": 25260 }, { "epoch": 16.105799872530273, "grad_norm": 0.47940099239349365, "learning_rate": 9.968000546763834e-05, "loss": 0.0348, "step": 25270 }, { "epoch": 16.112173358827278, "grad_norm": 0.5000100135803223, "learning_rate": 9.967938260709357e-05, "loss": 0.0442, "step": 25280 }, { "epoch": 16.118546845124282, "grad_norm": 0.4666995704174042, "learning_rate": 9.96787591428974e-05, "loss": 0.0401, "step": 25290 }, { "epoch": 16.124920331421286, "grad_norm": 0.5550605654716492, "learning_rate": 9.967813507505741e-05, "loss": 0.0426, "step": 25300 }, { "epoch": 16.13129381771829, "grad_norm": 0.4802536964416504, "learning_rate": 9.967751040358116e-05, "loss": 0.031, "step": 25310 }, { "epoch": 16.137667304015295, "grad_norm": 0.7131120562553406, "learning_rate": 9.967688512847626e-05, "loss": 0.0358, "step": 25320 }, { "epoch": 16.1440407903123, "grad_norm": 0.7372891902923584, "learning_rate": 9.967625924975031e-05, "loss": 0.0462, "step": 25330 }, { "epoch": 16.150414276609304, "grad_norm": 0.5742558836936951, "learning_rate": 9.967563276741091e-05, "loss": 0.0379, "step": 25340 }, { "epoch": 16.15678776290631, "grad_norm": 0.6399430632591248, "learning_rate": 9.967500568146566e-05, "loss": 0.0401, "step": 25350 }, { "epoch": 16.163161249203313, "grad_norm": 0.7068963050842285, "learning_rate": 9.96743779919222e-05, "loss": 0.0402, "step": 25360 }, { "epoch": 16.169534735500317, "grad_norm": 0.7580180168151855, "learning_rate": 9.967374969878814e-05, "loss": 0.0409, "step": 25370 }, { "epoch": 16.17590822179732, "grad_norm": 1.066556453704834, "learning_rate": 9.967312080207112e-05, "loss": 0.0359, "step": 25380 }, { "epoch": 16.182281708094326, "grad_norm": 0.40891554951667786, "learning_rate": 9.967249130177878e-05, "loss": 0.0295, "step": 25390 }, { "epoch": 16.18865519439133, "grad_norm": 0.3703814744949341, "learning_rate": 9.967186119791878e-05, "loss": 0.0402, "step": 25400 }, { "epoch": 16.195028680688335, "grad_norm": 0.6365624666213989, "learning_rate": 9.967123049049877e-05, "loss": 0.0401, "step": 25410 }, { "epoch": 16.20140216698534, "grad_norm": 0.5917449593544006, "learning_rate": 9.96705991795264e-05, "loss": 0.0378, "step": 25420 }, { "epoch": 16.207775653282347, "grad_norm": 0.4612136483192444, "learning_rate": 9.966996726500936e-05, "loss": 0.0399, "step": 25430 }, { "epoch": 16.21414913957935, "grad_norm": 0.541172981262207, "learning_rate": 9.966933474695532e-05, "loss": 0.0453, "step": 25440 }, { "epoch": 16.220522625876356, "grad_norm": 0.2582317292690277, "learning_rate": 9.966870162537199e-05, "loss": 0.0416, "step": 25450 }, { "epoch": 16.22689611217336, "grad_norm": 0.45396140217781067, "learning_rate": 9.966806790026701e-05, "loss": 0.0382, "step": 25460 }, { "epoch": 16.233269598470365, "grad_norm": 0.5514276027679443, "learning_rate": 9.966743357164811e-05, "loss": 0.0382, "step": 25470 }, { "epoch": 16.23964308476737, "grad_norm": 0.9388208985328674, "learning_rate": 9.966679863952302e-05, "loss": 0.0411, "step": 25480 }, { "epoch": 16.246016571064374, "grad_norm": 0.45645856857299805, "learning_rate": 9.966616310389942e-05, "loss": 0.0361, "step": 25490 }, { "epoch": 16.252390057361378, "grad_norm": 0.2370976358652115, "learning_rate": 9.966552696478504e-05, "loss": 0.036, "step": 25500 }, { "epoch": 16.258763543658382, "grad_norm": 0.2736761271953583, "learning_rate": 9.966489022218763e-05, "loss": 0.0446, "step": 25510 }, { "epoch": 16.265137029955387, "grad_norm": 0.3377840518951416, "learning_rate": 9.96642528761149e-05, "loss": 0.033, "step": 25520 }, { "epoch": 16.27151051625239, "grad_norm": 0.4454249441623688, "learning_rate": 9.966361492657461e-05, "loss": 0.0484, "step": 25530 }, { "epoch": 16.277884002549396, "grad_norm": 0.4505033493041992, "learning_rate": 9.966297637357452e-05, "loss": 0.0475, "step": 25540 }, { "epoch": 16.2842574888464, "grad_norm": 0.6705693006515503, "learning_rate": 9.966233721712237e-05, "loss": 0.0421, "step": 25550 }, { "epoch": 16.290630975143404, "grad_norm": 0.5719900131225586, "learning_rate": 9.966169745722593e-05, "loss": 0.0469, "step": 25560 }, { "epoch": 16.29700446144041, "grad_norm": 0.3500453233718872, "learning_rate": 9.966105709389297e-05, "loss": 0.0461, "step": 25570 }, { "epoch": 16.303377947737413, "grad_norm": 0.5628396272659302, "learning_rate": 9.966041612713129e-05, "loss": 0.0433, "step": 25580 }, { "epoch": 16.309751434034418, "grad_norm": 0.423896849155426, "learning_rate": 9.965977455694867e-05, "loss": 0.0437, "step": 25590 }, { "epoch": 16.316124920331422, "grad_norm": 0.6233727931976318, "learning_rate": 9.96591323833529e-05, "loss": 0.0449, "step": 25600 }, { "epoch": 16.322498406628426, "grad_norm": 0.5242258310317993, "learning_rate": 9.965848960635179e-05, "loss": 0.0383, "step": 25610 }, { "epoch": 16.32887189292543, "grad_norm": 0.6171064972877502, "learning_rate": 9.965784622595313e-05, "loss": 0.0403, "step": 25620 }, { "epoch": 16.335245379222435, "grad_norm": 0.7143844962120056, "learning_rate": 9.965720224216476e-05, "loss": 0.0392, "step": 25630 }, { "epoch": 16.34161886551944, "grad_norm": 0.4803479313850403, "learning_rate": 9.96565576549945e-05, "loss": 0.047, "step": 25640 }, { "epoch": 16.347992351816444, "grad_norm": 0.9446491599082947, "learning_rate": 9.965591246445018e-05, "loss": 0.0414, "step": 25650 }, { "epoch": 16.35436583811345, "grad_norm": 0.9977034330368042, "learning_rate": 9.965526667053963e-05, "loss": 0.0539, "step": 25660 }, { "epoch": 16.360739324410453, "grad_norm": 0.5730120539665222, "learning_rate": 9.965462027327073e-05, "loss": 0.0426, "step": 25670 }, { "epoch": 16.367112810707457, "grad_norm": 0.8035280108451843, "learning_rate": 9.965397327265132e-05, "loss": 0.0536, "step": 25680 }, { "epoch": 16.37348629700446, "grad_norm": 1.0698623657226562, "learning_rate": 9.965332566868923e-05, "loss": 0.0438, "step": 25690 }, { "epoch": 16.379859783301466, "grad_norm": 0.5060622096061707, "learning_rate": 9.965267746139235e-05, "loss": 0.0447, "step": 25700 }, { "epoch": 16.38623326959847, "grad_norm": 0.3704836070537567, "learning_rate": 9.965202865076857e-05, "loss": 0.0489, "step": 25710 }, { "epoch": 16.392606755895475, "grad_norm": 0.836049497127533, "learning_rate": 9.965137923682577e-05, "loss": 0.0446, "step": 25720 }, { "epoch": 16.39898024219248, "grad_norm": 0.886762797832489, "learning_rate": 9.965072921957181e-05, "loss": 0.0415, "step": 25730 }, { "epoch": 16.405353728489484, "grad_norm": 0.6053810119628906, "learning_rate": 9.965007859901464e-05, "loss": 0.0411, "step": 25740 }, { "epoch": 16.411727214786488, "grad_norm": 0.8052540421485901, "learning_rate": 9.964942737516211e-05, "loss": 0.0333, "step": 25750 }, { "epoch": 16.418100701083493, "grad_norm": 0.5735458731651306, "learning_rate": 9.964877554802217e-05, "loss": 0.0492, "step": 25760 }, { "epoch": 16.424474187380497, "grad_norm": 0.31387919187545776, "learning_rate": 9.964812311760275e-05, "loss": 0.0443, "step": 25770 }, { "epoch": 16.4308476736775, "grad_norm": 0.4805189371109009, "learning_rate": 9.964747008391173e-05, "loss": 0.0368, "step": 25780 }, { "epoch": 16.437221159974506, "grad_norm": 0.3303167521953583, "learning_rate": 9.964681644695708e-05, "loss": 0.0343, "step": 25790 }, { "epoch": 16.44359464627151, "grad_norm": 0.3784751296043396, "learning_rate": 9.964616220674675e-05, "loss": 0.0397, "step": 25800 }, { "epoch": 16.449968132568515, "grad_norm": 0.35350027680397034, "learning_rate": 9.964550736328866e-05, "loss": 0.0396, "step": 25810 }, { "epoch": 16.45634161886552, "grad_norm": 0.38310375809669495, "learning_rate": 9.964485191659079e-05, "loss": 0.039, "step": 25820 }, { "epoch": 16.462715105162523, "grad_norm": 0.362540066242218, "learning_rate": 9.964419586666109e-05, "loss": 0.0363, "step": 25830 }, { "epoch": 16.469088591459528, "grad_norm": 0.5591185092926025, "learning_rate": 9.964353921350755e-05, "loss": 0.0435, "step": 25840 }, { "epoch": 16.475462077756532, "grad_norm": 0.9858211278915405, "learning_rate": 9.964288195713813e-05, "loss": 0.0386, "step": 25850 }, { "epoch": 16.481835564053537, "grad_norm": 0.766990602016449, "learning_rate": 9.964222409756083e-05, "loss": 0.0438, "step": 25860 }, { "epoch": 16.48820905035054, "grad_norm": 0.6600323915481567, "learning_rate": 9.964156563478364e-05, "loss": 0.0424, "step": 25870 }, { "epoch": 16.494582536647545, "grad_norm": 0.46155497431755066, "learning_rate": 9.964090656881455e-05, "loss": 0.0548, "step": 25880 }, { "epoch": 16.50095602294455, "grad_norm": 0.5408865809440613, "learning_rate": 9.964024689966158e-05, "loss": 0.0495, "step": 25890 }, { "epoch": 16.507329509241554, "grad_norm": 0.6564632058143616, "learning_rate": 9.963958662733274e-05, "loss": 0.0283, "step": 25900 }, { "epoch": 16.51370299553856, "grad_norm": 0.9540555477142334, "learning_rate": 9.963892575183604e-05, "loss": 0.0453, "step": 25910 }, { "epoch": 16.520076481835563, "grad_norm": 0.6405223608016968, "learning_rate": 9.963826427317955e-05, "loss": 0.0506, "step": 25920 }, { "epoch": 16.526449968132567, "grad_norm": 0.5533133149147034, "learning_rate": 9.963760219137127e-05, "loss": 0.0432, "step": 25930 }, { "epoch": 16.532823454429572, "grad_norm": 0.5807307958602905, "learning_rate": 9.963693950641926e-05, "loss": 0.0482, "step": 25940 }, { "epoch": 16.539196940726576, "grad_norm": 0.48776960372924805, "learning_rate": 9.963627621833157e-05, "loss": 0.0366, "step": 25950 }, { "epoch": 16.54557042702358, "grad_norm": 0.6296766996383667, "learning_rate": 9.963561232711626e-05, "loss": 0.0525, "step": 25960 }, { "epoch": 16.551943913320585, "grad_norm": 0.6705887913703918, "learning_rate": 9.963494783278141e-05, "loss": 0.0369, "step": 25970 }, { "epoch": 16.55831739961759, "grad_norm": 0.5784368515014648, "learning_rate": 9.963428273533506e-05, "loss": 0.0399, "step": 25980 }, { "epoch": 16.564690885914594, "grad_norm": 0.6684627532958984, "learning_rate": 9.963361703478531e-05, "loss": 0.0465, "step": 25990 }, { "epoch": 16.5710643722116, "grad_norm": 0.528326153755188, "learning_rate": 9.963295073114026e-05, "loss": 0.0321, "step": 26000 }, { "epoch": 16.577437858508603, "grad_norm": 0.5851904153823853, "learning_rate": 9.9632283824408e-05, "loss": 0.0404, "step": 26010 }, { "epoch": 16.583811344805607, "grad_norm": 0.6672981977462769, "learning_rate": 9.963161631459663e-05, "loss": 0.06, "step": 26020 }, { "epoch": 16.59018483110261, "grad_norm": 0.5629258751869202, "learning_rate": 9.963094820171426e-05, "loss": 0.0419, "step": 26030 }, { "epoch": 16.59655831739962, "grad_norm": 0.7017990350723267, "learning_rate": 9.963027948576902e-05, "loss": 0.0642, "step": 26040 }, { "epoch": 16.60293180369662, "grad_norm": 0.6739966869354248, "learning_rate": 9.9629610166769e-05, "loss": 0.0465, "step": 26050 }, { "epoch": 16.609305289993628, "grad_norm": 0.5004031658172607, "learning_rate": 9.962894024472237e-05, "loss": 0.0368, "step": 26060 }, { "epoch": 16.615678776290633, "grad_norm": 0.4587778151035309, "learning_rate": 9.962826971963728e-05, "loss": 0.0398, "step": 26070 }, { "epoch": 16.622052262587637, "grad_norm": 0.654233455657959, "learning_rate": 9.962759859152184e-05, "loss": 0.0525, "step": 26080 }, { "epoch": 16.62842574888464, "grad_norm": 0.4020244777202606, "learning_rate": 9.962692686038422e-05, "loss": 0.0573, "step": 26090 }, { "epoch": 16.634799235181646, "grad_norm": 0.49503394961357117, "learning_rate": 9.962625452623258e-05, "loss": 0.0562, "step": 26100 }, { "epoch": 16.64117272147865, "grad_norm": 0.6481123566627502, "learning_rate": 9.962558158907509e-05, "loss": 0.0498, "step": 26110 }, { "epoch": 16.647546207775655, "grad_norm": 0.582618236541748, "learning_rate": 9.962490804891993e-05, "loss": 0.0337, "step": 26120 }, { "epoch": 16.65391969407266, "grad_norm": 0.553983747959137, "learning_rate": 9.962423390577529e-05, "loss": 0.0483, "step": 26130 }, { "epoch": 16.660293180369663, "grad_norm": 0.7455496191978455, "learning_rate": 9.962355915964936e-05, "loss": 0.0472, "step": 26140 }, { "epoch": 16.666666666666668, "grad_norm": 0.4680227041244507, "learning_rate": 9.962288381055031e-05, "loss": 0.0507, "step": 26150 }, { "epoch": 16.673040152963672, "grad_norm": 0.5346328616142273, "learning_rate": 9.96222078584864e-05, "loss": 0.0485, "step": 26160 }, { "epoch": 16.679413639260677, "grad_norm": 0.3488355875015259, "learning_rate": 9.96215313034658e-05, "loss": 0.0306, "step": 26170 }, { "epoch": 16.68578712555768, "grad_norm": 0.38558948040008545, "learning_rate": 9.962085414549673e-05, "loss": 0.0443, "step": 26180 }, { "epoch": 16.692160611854685, "grad_norm": 0.575023889541626, "learning_rate": 9.962017638458746e-05, "loss": 0.0528, "step": 26190 }, { "epoch": 16.69853409815169, "grad_norm": 0.5307489037513733, "learning_rate": 9.961949802074618e-05, "loss": 0.0438, "step": 26200 }, { "epoch": 16.704907584448694, "grad_norm": 0.7933769226074219, "learning_rate": 9.961881905398117e-05, "loss": 0.046, "step": 26210 }, { "epoch": 16.7112810707457, "grad_norm": 0.5165237188339233, "learning_rate": 9.961813948430064e-05, "loss": 0.053, "step": 26220 }, { "epoch": 16.717654557042703, "grad_norm": 0.6027774810791016, "learning_rate": 9.961745931171287e-05, "loss": 0.0516, "step": 26230 }, { "epoch": 16.724028043339707, "grad_norm": 0.475593626499176, "learning_rate": 9.961677853622613e-05, "loss": 0.044, "step": 26240 }, { "epoch": 16.730401529636712, "grad_norm": 0.5969756245613098, "learning_rate": 9.961609715784868e-05, "loss": 0.0369, "step": 26250 }, { "epoch": 16.736775015933716, "grad_norm": 0.44374409317970276, "learning_rate": 9.96154151765888e-05, "loss": 0.0534, "step": 26260 }, { "epoch": 16.74314850223072, "grad_norm": 0.6982294917106628, "learning_rate": 9.961473259245478e-05, "loss": 0.0562, "step": 26270 }, { "epoch": 16.749521988527725, "grad_norm": 0.5568380951881409, "learning_rate": 9.961404940545492e-05, "loss": 0.0571, "step": 26280 }, { "epoch": 16.75589547482473, "grad_norm": 0.351503849029541, "learning_rate": 9.961336561559753e-05, "loss": 0.0532, "step": 26290 }, { "epoch": 16.762268961121734, "grad_norm": 0.5804415941238403, "learning_rate": 9.961268122289089e-05, "loss": 0.0633, "step": 26300 }, { "epoch": 16.76864244741874, "grad_norm": 0.6005134582519531, "learning_rate": 9.961199622734333e-05, "loss": 0.0562, "step": 26310 }, { "epoch": 16.775015933715743, "grad_norm": 0.3110552430152893, "learning_rate": 9.961131062896318e-05, "loss": 0.0582, "step": 26320 }, { "epoch": 16.781389420012747, "grad_norm": 0.5617906451225281, "learning_rate": 9.961062442775876e-05, "loss": 0.0458, "step": 26330 }, { "epoch": 16.78776290630975, "grad_norm": 0.7597768306732178, "learning_rate": 9.960993762373842e-05, "loss": 0.0432, "step": 26340 }, { "epoch": 16.794136392606756, "grad_norm": 0.6495307683944702, "learning_rate": 9.960925021691049e-05, "loss": 0.0441, "step": 26350 }, { "epoch": 16.80050987890376, "grad_norm": 0.4253827631473541, "learning_rate": 9.960856220728333e-05, "loss": 0.054, "step": 26360 }, { "epoch": 16.806883365200765, "grad_norm": 0.5670692920684814, "learning_rate": 9.960787359486531e-05, "loss": 0.0478, "step": 26370 }, { "epoch": 16.81325685149777, "grad_norm": 0.5977931618690491, "learning_rate": 9.960718437966479e-05, "loss": 0.0434, "step": 26380 }, { "epoch": 16.819630337794774, "grad_norm": 0.41143953800201416, "learning_rate": 9.960649456169014e-05, "loss": 0.0427, "step": 26390 }, { "epoch": 16.826003824091778, "grad_norm": 0.47518542408943176, "learning_rate": 9.960580414094973e-05, "loss": 0.0518, "step": 26400 }, { "epoch": 16.832377310388782, "grad_norm": 0.6711757183074951, "learning_rate": 9.960511311745198e-05, "loss": 0.0433, "step": 26410 }, { "epoch": 16.838750796685787, "grad_norm": 0.582792341709137, "learning_rate": 9.960442149120526e-05, "loss": 0.0357, "step": 26420 }, { "epoch": 16.84512428298279, "grad_norm": 0.7889184355735779, "learning_rate": 9.9603729262218e-05, "loss": 0.0499, "step": 26430 }, { "epoch": 16.851497769279796, "grad_norm": 0.4969421923160553, "learning_rate": 9.960303643049858e-05, "loss": 0.0587, "step": 26440 }, { "epoch": 16.8578712555768, "grad_norm": 0.9252558350563049, "learning_rate": 9.960234299605544e-05, "loss": 0.0454, "step": 26450 }, { "epoch": 16.864244741873804, "grad_norm": 0.5752676725387573, "learning_rate": 9.9601648958897e-05, "loss": 0.0336, "step": 26460 }, { "epoch": 16.87061822817081, "grad_norm": 0.42766857147216797, "learning_rate": 9.96009543190317e-05, "loss": 0.0486, "step": 26470 }, { "epoch": 16.876991714467813, "grad_norm": 0.8740835785865784, "learning_rate": 9.960025907646797e-05, "loss": 0.042, "step": 26480 }, { "epoch": 16.883365200764818, "grad_norm": 0.458851158618927, "learning_rate": 9.959956323121425e-05, "loss": 0.0374, "step": 26490 }, { "epoch": 16.889738687061822, "grad_norm": 0.35557204484939575, "learning_rate": 9.959886678327903e-05, "loss": 0.0366, "step": 26500 }, { "epoch": 16.896112173358826, "grad_norm": 0.6742868423461914, "learning_rate": 9.959816973267073e-05, "loss": 0.0675, "step": 26510 }, { "epoch": 16.90248565965583, "grad_norm": 0.6167845726013184, "learning_rate": 9.959747207939785e-05, "loss": 0.0437, "step": 26520 }, { "epoch": 16.908859145952835, "grad_norm": 0.636807918548584, "learning_rate": 9.959677382346884e-05, "loss": 0.0467, "step": 26530 }, { "epoch": 16.91523263224984, "grad_norm": 0.6801275610923767, "learning_rate": 9.959607496489223e-05, "loss": 0.0531, "step": 26540 }, { "epoch": 16.921606118546844, "grad_norm": 0.3105388581752777, "learning_rate": 9.959537550367646e-05, "loss": 0.04, "step": 26550 }, { "epoch": 16.92797960484385, "grad_norm": 0.6922913193702698, "learning_rate": 9.959467543983005e-05, "loss": 0.043, "step": 26560 }, { "epoch": 16.934353091140853, "grad_norm": 0.5376835465431213, "learning_rate": 9.959397477336152e-05, "loss": 0.0496, "step": 26570 }, { "epoch": 16.940726577437857, "grad_norm": 0.5774103403091431, "learning_rate": 9.959327350427937e-05, "loss": 0.0368, "step": 26580 }, { "epoch": 16.94710006373486, "grad_norm": 0.4151640832424164, "learning_rate": 9.959257163259212e-05, "loss": 0.0359, "step": 26590 }, { "epoch": 16.953473550031866, "grad_norm": 0.5616617202758789, "learning_rate": 9.95918691583083e-05, "loss": 0.0367, "step": 26600 }, { "epoch": 16.95984703632887, "grad_norm": 0.6855557560920715, "learning_rate": 9.959116608143646e-05, "loss": 0.0442, "step": 26610 }, { "epoch": 16.966220522625875, "grad_norm": 0.561180591583252, "learning_rate": 9.959046240198511e-05, "loss": 0.0406, "step": 26620 }, { "epoch": 16.97259400892288, "grad_norm": 0.4963124692440033, "learning_rate": 9.958975811996284e-05, "loss": 0.0578, "step": 26630 }, { "epoch": 16.978967495219884, "grad_norm": 0.6078842878341675, "learning_rate": 9.958905323537817e-05, "loss": 0.0433, "step": 26640 }, { "epoch": 16.985340981516888, "grad_norm": 0.5090770721435547, "learning_rate": 9.958834774823969e-05, "loss": 0.0333, "step": 26650 }, { "epoch": 16.991714467813892, "grad_norm": 0.676506519317627, "learning_rate": 9.958764165855598e-05, "loss": 0.0353, "step": 26660 }, { "epoch": 16.9980879541109, "grad_norm": 0.6919217109680176, "learning_rate": 9.95869349663356e-05, "loss": 0.0417, "step": 26670 }, { "epoch": 17.004461440407905, "grad_norm": 0.6133757829666138, "learning_rate": 9.958622767158713e-05, "loss": 0.0329, "step": 26680 }, { "epoch": 17.01083492670491, "grad_norm": 0.8568653464317322, "learning_rate": 9.958551977431918e-05, "loss": 0.0363, "step": 26690 }, { "epoch": 17.017208413001914, "grad_norm": 0.38847360014915466, "learning_rate": 9.958481127454036e-05, "loss": 0.0414, "step": 26700 }, { "epoch": 17.023581899298918, "grad_norm": 0.4781525731086731, "learning_rate": 9.958410217225925e-05, "loss": 0.0541, "step": 26710 }, { "epoch": 17.029955385595922, "grad_norm": 0.5136752724647522, "learning_rate": 9.95833924674845e-05, "loss": 0.0322, "step": 26720 }, { "epoch": 17.036328871892927, "grad_norm": 0.3825243413448334, "learning_rate": 9.958268216022471e-05, "loss": 0.0296, "step": 26730 }, { "epoch": 17.04270235818993, "grad_norm": 0.3564102053642273, "learning_rate": 9.958197125048852e-05, "loss": 0.0368, "step": 26740 }, { "epoch": 17.049075844486936, "grad_norm": 0.6764777302742004, "learning_rate": 9.958125973828454e-05, "loss": 0.0383, "step": 26750 }, { "epoch": 17.05544933078394, "grad_norm": 0.6747488975524902, "learning_rate": 9.958054762362148e-05, "loss": 0.0453, "step": 26760 }, { "epoch": 17.061822817080945, "grad_norm": 0.5109786987304688, "learning_rate": 9.957983490650793e-05, "loss": 0.0417, "step": 26770 }, { "epoch": 17.06819630337795, "grad_norm": 0.3799660801887512, "learning_rate": 9.95791215869526e-05, "loss": 0.0433, "step": 26780 }, { "epoch": 17.074569789674953, "grad_norm": 0.6049870252609253, "learning_rate": 9.957840766496411e-05, "loss": 0.0413, "step": 26790 }, { "epoch": 17.080943275971958, "grad_norm": 0.9162454009056091, "learning_rate": 9.957769314055117e-05, "loss": 0.0419, "step": 26800 }, { "epoch": 17.087316762268962, "grad_norm": 0.30667006969451904, "learning_rate": 9.957697801372243e-05, "loss": 0.0443, "step": 26810 }, { "epoch": 17.093690248565967, "grad_norm": 0.5703659057617188, "learning_rate": 9.957626228448662e-05, "loss": 0.04, "step": 26820 }, { "epoch": 17.10006373486297, "grad_norm": 0.7101849913597107, "learning_rate": 9.95755459528524e-05, "loss": 0.0392, "step": 26830 }, { "epoch": 17.106437221159975, "grad_norm": 0.36343079805374146, "learning_rate": 9.95748290188285e-05, "loss": 0.0369, "step": 26840 }, { "epoch": 17.11281070745698, "grad_norm": 0.6599883437156677, "learning_rate": 9.957411148242361e-05, "loss": 0.0543, "step": 26850 }, { "epoch": 17.119184193753984, "grad_norm": 0.35964056849479675, "learning_rate": 9.957339334364648e-05, "loss": 0.0376, "step": 26860 }, { "epoch": 17.12555768005099, "grad_norm": 0.35593265295028687, "learning_rate": 9.957267460250579e-05, "loss": 0.0325, "step": 26870 }, { "epoch": 17.131931166347993, "grad_norm": 0.5668849349021912, "learning_rate": 9.957195525901032e-05, "loss": 0.0382, "step": 26880 }, { "epoch": 17.138304652644997, "grad_norm": 0.6566433906555176, "learning_rate": 9.957123531316878e-05, "loss": 0.051, "step": 26890 }, { "epoch": 17.144678138942, "grad_norm": 0.5178101658821106, "learning_rate": 9.957051476498994e-05, "loss": 0.034, "step": 26900 }, { "epoch": 17.151051625239006, "grad_norm": 1.091646671295166, "learning_rate": 9.956979361448253e-05, "loss": 0.038, "step": 26910 }, { "epoch": 17.15742511153601, "grad_norm": 0.3322891592979431, "learning_rate": 9.956907186165532e-05, "loss": 0.038, "step": 26920 }, { "epoch": 17.163798597833015, "grad_norm": 0.6825461387634277, "learning_rate": 9.95683495065171e-05, "loss": 0.0374, "step": 26930 }, { "epoch": 17.17017208413002, "grad_norm": 0.4510245621204376, "learning_rate": 9.956762654907664e-05, "loss": 0.0425, "step": 26940 }, { "epoch": 17.176545570427024, "grad_norm": 0.739872395992279, "learning_rate": 9.95669029893427e-05, "loss": 0.0569, "step": 26950 }, { "epoch": 17.182919056724028, "grad_norm": 0.42860573530197144, "learning_rate": 9.956617882732409e-05, "loss": 0.0385, "step": 26960 }, { "epoch": 17.189292543021033, "grad_norm": 0.46240487694740295, "learning_rate": 9.956545406302962e-05, "loss": 0.0421, "step": 26970 }, { "epoch": 17.195666029318037, "grad_norm": 0.6773727536201477, "learning_rate": 9.956472869646807e-05, "loss": 0.0476, "step": 26980 }, { "epoch": 17.20203951561504, "grad_norm": 0.6244494915008545, "learning_rate": 9.956400272764829e-05, "loss": 0.0347, "step": 26990 }, { "epoch": 17.208413001912046, "grad_norm": 0.5109573602676392, "learning_rate": 9.956327615657907e-05, "loss": 0.0393, "step": 27000 }, { "epoch": 17.21478648820905, "grad_norm": 0.6456117033958435, "learning_rate": 9.956254898326924e-05, "loss": 0.0323, "step": 27010 }, { "epoch": 17.221159974506055, "grad_norm": 0.582136869430542, "learning_rate": 9.956182120772765e-05, "loss": 0.0305, "step": 27020 }, { "epoch": 17.22753346080306, "grad_norm": 0.6260345578193665, "learning_rate": 9.956109282996315e-05, "loss": 0.0356, "step": 27030 }, { "epoch": 17.233906947100063, "grad_norm": 0.564299464225769, "learning_rate": 9.956036384998457e-05, "loss": 0.0418, "step": 27040 }, { "epoch": 17.240280433397068, "grad_norm": 0.45445767045021057, "learning_rate": 9.95596342678008e-05, "loss": 0.0432, "step": 27050 }, { "epoch": 17.246653919694072, "grad_norm": 0.604080855846405, "learning_rate": 9.955890408342065e-05, "loss": 0.037, "step": 27060 }, { "epoch": 17.253027405991077, "grad_norm": 0.3245340883731842, "learning_rate": 9.955817329685303e-05, "loss": 0.0321, "step": 27070 }, { "epoch": 17.25940089228808, "grad_norm": 0.4026404917240143, "learning_rate": 9.955744190810682e-05, "loss": 0.0258, "step": 27080 }, { "epoch": 17.265774378585085, "grad_norm": 0.7042772173881531, "learning_rate": 9.95567099171909e-05, "loss": 0.0497, "step": 27090 }, { "epoch": 17.27214786488209, "grad_norm": 0.4526478350162506, "learning_rate": 9.955597732411417e-05, "loss": 0.0355, "step": 27100 }, { "epoch": 17.278521351179094, "grad_norm": 0.37130364775657654, "learning_rate": 9.95552441288855e-05, "loss": 0.0354, "step": 27110 }, { "epoch": 17.2848948374761, "grad_norm": 0.4773696959018707, "learning_rate": 9.955451033151387e-05, "loss": 0.0466, "step": 27120 }, { "epoch": 17.291268323773103, "grad_norm": 0.30388012528419495, "learning_rate": 9.955377593200813e-05, "loss": 0.0415, "step": 27130 }, { "epoch": 17.297641810070107, "grad_norm": 0.4556509256362915, "learning_rate": 9.955304093037721e-05, "loss": 0.0456, "step": 27140 }, { "epoch": 17.304015296367112, "grad_norm": 0.7264136075973511, "learning_rate": 9.955230532663007e-05, "loss": 0.0424, "step": 27150 }, { "epoch": 17.310388782664116, "grad_norm": 0.4282618761062622, "learning_rate": 9.955156912077565e-05, "loss": 0.0314, "step": 27160 }, { "epoch": 17.31676226896112, "grad_norm": 0.5929535627365112, "learning_rate": 9.955083231282286e-05, "loss": 0.0329, "step": 27170 }, { "epoch": 17.323135755258125, "grad_norm": 0.4965292811393738, "learning_rate": 9.955009490278068e-05, "loss": 0.0445, "step": 27180 }, { "epoch": 17.32950924155513, "grad_norm": 0.2812654972076416, "learning_rate": 9.954935689065808e-05, "loss": 0.0499, "step": 27190 }, { "epoch": 17.335882727852134, "grad_norm": 0.5554900169372559, "learning_rate": 9.954861827646398e-05, "loss": 0.0417, "step": 27200 }, { "epoch": 17.34225621414914, "grad_norm": 0.5372625589370728, "learning_rate": 9.95478790602074e-05, "loss": 0.0549, "step": 27210 }, { "epoch": 17.348629700446143, "grad_norm": 0.42561760544776917, "learning_rate": 9.954713924189731e-05, "loss": 0.033, "step": 27220 }, { "epoch": 17.355003186743147, "grad_norm": 0.5412494540214539, "learning_rate": 9.954639882154271e-05, "loss": 0.039, "step": 27230 }, { "epoch": 17.36137667304015, "grad_norm": 0.4809313416481018, "learning_rate": 9.954565779915257e-05, "loss": 0.0317, "step": 27240 }, { "epoch": 17.367750159337156, "grad_norm": 0.7674503922462463, "learning_rate": 9.954491617473591e-05, "loss": 0.048, "step": 27250 }, { "epoch": 17.37412364563416, "grad_norm": 0.5053706169128418, "learning_rate": 9.954417394830176e-05, "loss": 0.0362, "step": 27260 }, { "epoch": 17.380497131931165, "grad_norm": 0.40367111563682556, "learning_rate": 9.954343111985908e-05, "loss": 0.0322, "step": 27270 }, { "epoch": 17.38687061822817, "grad_norm": 0.5624961853027344, "learning_rate": 9.954268768941697e-05, "loss": 0.0264, "step": 27280 }, { "epoch": 17.393244104525174, "grad_norm": 0.44181734323501587, "learning_rate": 9.954194365698441e-05, "loss": 0.0348, "step": 27290 }, { "epoch": 17.39961759082218, "grad_norm": 0.743519127368927, "learning_rate": 9.954119902257047e-05, "loss": 0.0321, "step": 27300 }, { "epoch": 17.405991077119186, "grad_norm": 0.38629281520843506, "learning_rate": 9.954045378618418e-05, "loss": 0.0382, "step": 27310 }, { "epoch": 17.41236456341619, "grad_norm": 0.656932532787323, "learning_rate": 9.95397079478346e-05, "loss": 0.0481, "step": 27320 }, { "epoch": 17.418738049713195, "grad_norm": 0.628499448299408, "learning_rate": 9.953896150753082e-05, "loss": 0.035, "step": 27330 }, { "epoch": 17.4251115360102, "grad_norm": 0.5086801052093506, "learning_rate": 9.953821446528186e-05, "loss": 0.0286, "step": 27340 }, { "epoch": 17.431485022307204, "grad_norm": 0.7382622361183167, "learning_rate": 9.953746682109683e-05, "loss": 0.0444, "step": 27350 }, { "epoch": 17.437858508604208, "grad_norm": 0.30309590697288513, "learning_rate": 9.95367185749848e-05, "loss": 0.0465, "step": 27360 }, { "epoch": 17.444231994901212, "grad_norm": 0.7429732084274292, "learning_rate": 9.953596972695487e-05, "loss": 0.0378, "step": 27370 }, { "epoch": 17.450605481198217, "grad_norm": 0.9609372019767761, "learning_rate": 9.953522027701614e-05, "loss": 0.0381, "step": 27380 }, { "epoch": 17.45697896749522, "grad_norm": 0.735106885433197, "learning_rate": 9.953447022517772e-05, "loss": 0.044, "step": 27390 }, { "epoch": 17.463352453792226, "grad_norm": 0.5321285724639893, "learning_rate": 9.953371957144872e-05, "loss": 0.0351, "step": 27400 }, { "epoch": 17.46972594008923, "grad_norm": 0.7595342993736267, "learning_rate": 9.953296831583825e-05, "loss": 0.0373, "step": 27410 }, { "epoch": 17.476099426386234, "grad_norm": 0.6421709060668945, "learning_rate": 9.953221645835545e-05, "loss": 0.0425, "step": 27420 }, { "epoch": 17.48247291268324, "grad_norm": 0.5293576717376709, "learning_rate": 9.953146399900945e-05, "loss": 0.0324, "step": 27430 }, { "epoch": 17.488846398980243, "grad_norm": 0.5224354863166809, "learning_rate": 9.953071093780941e-05, "loss": 0.0533, "step": 27440 }, { "epoch": 17.495219885277248, "grad_norm": 0.5056937336921692, "learning_rate": 9.952995727476446e-05, "loss": 0.0495, "step": 27450 }, { "epoch": 17.501593371574252, "grad_norm": 0.6087573766708374, "learning_rate": 9.952920300988378e-05, "loss": 0.0395, "step": 27460 }, { "epoch": 17.507966857871256, "grad_norm": 0.7454321980476379, "learning_rate": 9.95284481431765e-05, "loss": 0.0296, "step": 27470 }, { "epoch": 17.51434034416826, "grad_norm": 0.4375068247318268, "learning_rate": 9.952769267465182e-05, "loss": 0.0609, "step": 27480 }, { "epoch": 17.520713830465265, "grad_norm": 0.49835067987442017, "learning_rate": 9.95269366043189e-05, "loss": 0.036, "step": 27490 }, { "epoch": 17.52708731676227, "grad_norm": 0.6412295699119568, "learning_rate": 9.952617993218696e-05, "loss": 0.044, "step": 27500 }, { "epoch": 17.533460803059274, "grad_norm": 0.6280710101127625, "learning_rate": 9.952542265826516e-05, "loss": 0.0655, "step": 27510 }, { "epoch": 17.53983428935628, "grad_norm": 0.4177946150302887, "learning_rate": 9.952466478256273e-05, "loss": 0.0352, "step": 27520 }, { "epoch": 17.546207775653283, "grad_norm": 0.370174765586853, "learning_rate": 9.952390630508884e-05, "loss": 0.0436, "step": 27530 }, { "epoch": 17.552581261950287, "grad_norm": 0.45558491349220276, "learning_rate": 9.952314722585274e-05, "loss": 0.0508, "step": 27540 }, { "epoch": 17.55895474824729, "grad_norm": 0.7363624572753906, "learning_rate": 9.952238754486365e-05, "loss": 0.0337, "step": 27550 }, { "epoch": 17.565328234544296, "grad_norm": 0.6047791242599487, "learning_rate": 9.952162726213078e-05, "loss": 0.0392, "step": 27560 }, { "epoch": 17.5717017208413, "grad_norm": 0.30739152431488037, "learning_rate": 9.95208663776634e-05, "loss": 0.0289, "step": 27570 }, { "epoch": 17.578075207138305, "grad_norm": 0.36529111862182617, "learning_rate": 9.952010489147074e-05, "loss": 0.038, "step": 27580 }, { "epoch": 17.58444869343531, "grad_norm": 0.27576979994773865, "learning_rate": 9.951934280356203e-05, "loss": 0.0382, "step": 27590 }, { "epoch": 17.590822179732314, "grad_norm": 0.4822746515274048, "learning_rate": 9.951858011394656e-05, "loss": 0.0383, "step": 27600 }, { "epoch": 17.597195666029318, "grad_norm": 0.5201099514961243, "learning_rate": 9.95178168226336e-05, "loss": 0.0552, "step": 27610 }, { "epoch": 17.603569152326322, "grad_norm": 0.33975285291671753, "learning_rate": 9.95170529296324e-05, "loss": 0.0527, "step": 27620 }, { "epoch": 17.609942638623327, "grad_norm": 0.39848917722702026, "learning_rate": 9.951628843495226e-05, "loss": 0.0374, "step": 27630 }, { "epoch": 17.61631612492033, "grad_norm": 0.4005988836288452, "learning_rate": 9.951552333860246e-05, "loss": 0.0494, "step": 27640 }, { "epoch": 17.622689611217336, "grad_norm": 0.4525968134403229, "learning_rate": 9.951475764059231e-05, "loss": 0.0449, "step": 27650 }, { "epoch": 17.62906309751434, "grad_norm": 0.731062650680542, "learning_rate": 9.95139913409311e-05, "loss": 0.0455, "step": 27660 }, { "epoch": 17.635436583811344, "grad_norm": 0.7559607625007629, "learning_rate": 9.951322443962815e-05, "loss": 0.0634, "step": 27670 }, { "epoch": 17.64181007010835, "grad_norm": 0.6006494164466858, "learning_rate": 9.951245693669277e-05, "loss": 0.0416, "step": 27680 }, { "epoch": 17.648183556405353, "grad_norm": 0.5289352536201477, "learning_rate": 9.951168883213428e-05, "loss": 0.0424, "step": 27690 }, { "epoch": 17.654557042702358, "grad_norm": 0.6971225142478943, "learning_rate": 9.951092012596204e-05, "loss": 0.0398, "step": 27700 }, { "epoch": 17.660930528999362, "grad_norm": 0.6305059790611267, "learning_rate": 9.951015081818537e-05, "loss": 0.0427, "step": 27710 }, { "epoch": 17.667304015296367, "grad_norm": 0.8132280111312866, "learning_rate": 9.950938090881362e-05, "loss": 0.0355, "step": 27720 }, { "epoch": 17.67367750159337, "grad_norm": 0.3476380705833435, "learning_rate": 9.950861039785614e-05, "loss": 0.0357, "step": 27730 }, { "epoch": 17.680050987890375, "grad_norm": 0.39315930008888245, "learning_rate": 9.950783928532232e-05, "loss": 0.0357, "step": 27740 }, { "epoch": 17.68642447418738, "grad_norm": 0.41412943601608276, "learning_rate": 9.950706757122151e-05, "loss": 0.0349, "step": 27750 }, { "epoch": 17.692797960484384, "grad_norm": 0.424165815114975, "learning_rate": 9.950629525556306e-05, "loss": 0.0455, "step": 27760 }, { "epoch": 17.69917144678139, "grad_norm": 0.3875099718570709, "learning_rate": 9.95055223383564e-05, "loss": 0.0466, "step": 27770 }, { "epoch": 17.705544933078393, "grad_norm": 0.5878488421440125, "learning_rate": 9.95047488196109e-05, "loss": 0.047, "step": 27780 }, { "epoch": 17.711918419375397, "grad_norm": 0.8434865474700928, "learning_rate": 9.950397469933597e-05, "loss": 0.0475, "step": 27790 }, { "epoch": 17.7182919056724, "grad_norm": 0.43690919876098633, "learning_rate": 9.950319997754098e-05, "loss": 0.0484, "step": 27800 }, { "epoch": 17.724665391969406, "grad_norm": 0.7100998163223267, "learning_rate": 9.950242465423541e-05, "loss": 0.0516, "step": 27810 }, { "epoch": 17.73103887826641, "grad_norm": 0.660385251045227, "learning_rate": 9.950164872942861e-05, "loss": 0.0491, "step": 27820 }, { "epoch": 17.737412364563415, "grad_norm": 0.8436694145202637, "learning_rate": 9.950087220313005e-05, "loss": 0.0445, "step": 27830 }, { "epoch": 17.74378585086042, "grad_norm": 0.5409918427467346, "learning_rate": 9.950009507534917e-05, "loss": 0.043, "step": 27840 }, { "epoch": 17.750159337157424, "grad_norm": 0.4436703324317932, "learning_rate": 9.949931734609538e-05, "loss": 0.0445, "step": 27850 }, { "epoch": 17.756532823454428, "grad_norm": 0.4926864206790924, "learning_rate": 9.949853901537815e-05, "loss": 0.0342, "step": 27860 }, { "epoch": 17.762906309751433, "grad_norm": 0.4110957384109497, "learning_rate": 9.949776008320695e-05, "loss": 0.0287, "step": 27870 }, { "epoch": 17.769279796048437, "grad_norm": 0.571541428565979, "learning_rate": 9.949698054959123e-05, "loss": 0.0411, "step": 27880 }, { "epoch": 17.77565328234544, "grad_norm": 1.0171386003494263, "learning_rate": 9.949620041454046e-05, "loss": 0.0625, "step": 27890 }, { "epoch": 17.782026768642446, "grad_norm": 0.44043421745300293, "learning_rate": 9.94954196780641e-05, "loss": 0.0314, "step": 27900 }, { "epoch": 17.78840025493945, "grad_norm": 0.5718318223953247, "learning_rate": 9.94946383401717e-05, "loss": 0.0412, "step": 27910 }, { "epoch": 17.794773741236455, "grad_norm": 0.5614866614341736, "learning_rate": 9.94938564008727e-05, "loss": 0.0387, "step": 27920 }, { "epoch": 17.801147227533463, "grad_norm": 0.6895573735237122, "learning_rate": 9.94930738601766e-05, "loss": 0.043, "step": 27930 }, { "epoch": 17.807520713830467, "grad_norm": 0.6153039932250977, "learning_rate": 9.949229071809293e-05, "loss": 0.0366, "step": 27940 }, { "epoch": 17.81389420012747, "grad_norm": 0.3587973415851593, "learning_rate": 9.949150697463121e-05, "loss": 0.0421, "step": 27950 }, { "epoch": 17.820267686424476, "grad_norm": 0.5704420208930969, "learning_rate": 9.949072262980094e-05, "loss": 0.0377, "step": 27960 }, { "epoch": 17.82664117272148, "grad_norm": 0.374412477016449, "learning_rate": 9.948993768361166e-05, "loss": 0.0455, "step": 27970 }, { "epoch": 17.833014659018485, "grad_norm": 0.6191907525062561, "learning_rate": 9.948915213607293e-05, "loss": 0.0397, "step": 27980 }, { "epoch": 17.83938814531549, "grad_norm": 0.4485732316970825, "learning_rate": 9.948836598719425e-05, "loss": 0.0374, "step": 27990 }, { "epoch": 17.845761631612493, "grad_norm": 0.3787826597690582, "learning_rate": 9.948757923698522e-05, "loss": 0.0433, "step": 28000 }, { "epoch": 17.852135117909498, "grad_norm": 0.44880014657974243, "learning_rate": 9.948679188545538e-05, "loss": 0.0549, "step": 28010 }, { "epoch": 17.858508604206502, "grad_norm": 0.9366286396980286, "learning_rate": 9.948600393261427e-05, "loss": 0.0409, "step": 28020 }, { "epoch": 17.864882090503507, "grad_norm": 0.5621974468231201, "learning_rate": 9.948521537847151e-05, "loss": 0.0503, "step": 28030 }, { "epoch": 17.87125557680051, "grad_norm": 0.6286352276802063, "learning_rate": 9.948442622303667e-05, "loss": 0.0391, "step": 28040 }, { "epoch": 17.877629063097515, "grad_norm": 0.7673023343086243, "learning_rate": 9.948363646631931e-05, "loss": 0.0523, "step": 28050 }, { "epoch": 17.88400254939452, "grad_norm": 0.8487848043441772, "learning_rate": 9.948284610832905e-05, "loss": 0.0437, "step": 28060 }, { "epoch": 17.890376035691524, "grad_norm": 0.5971744060516357, "learning_rate": 9.94820551490755e-05, "loss": 0.0527, "step": 28070 }, { "epoch": 17.89674952198853, "grad_norm": 0.6268824934959412, "learning_rate": 9.948126358856825e-05, "loss": 0.0427, "step": 28080 }, { "epoch": 17.903123008285533, "grad_norm": 0.6447497606277466, "learning_rate": 9.948047142681694e-05, "loss": 0.0339, "step": 28090 }, { "epoch": 17.909496494582537, "grad_norm": 0.5796855092048645, "learning_rate": 9.947967866383117e-05, "loss": 0.057, "step": 28100 }, { "epoch": 17.915869980879542, "grad_norm": 0.6518093943595886, "learning_rate": 9.947888529962061e-05, "loss": 0.036, "step": 28110 }, { "epoch": 17.922243467176546, "grad_norm": 0.6476771235466003, "learning_rate": 9.947809133419486e-05, "loss": 0.0389, "step": 28120 }, { "epoch": 17.92861695347355, "grad_norm": 0.6231204867362976, "learning_rate": 9.947729676756359e-05, "loss": 0.0375, "step": 28130 }, { "epoch": 17.934990439770555, "grad_norm": 0.5564508438110352, "learning_rate": 9.947650159973645e-05, "loss": 0.0341, "step": 28140 }, { "epoch": 17.94136392606756, "grad_norm": 0.6180081367492676, "learning_rate": 9.94757058307231e-05, "loss": 0.0304, "step": 28150 }, { "epoch": 17.947737412364564, "grad_norm": 0.7775779962539673, "learning_rate": 9.947490946053321e-05, "loss": 0.0567, "step": 28160 }, { "epoch": 17.95411089866157, "grad_norm": 0.4301908016204834, "learning_rate": 9.947411248917647e-05, "loss": 0.034, "step": 28170 }, { "epoch": 17.960484384958573, "grad_norm": 0.4990338683128357, "learning_rate": 9.947331491666255e-05, "loss": 0.0491, "step": 28180 }, { "epoch": 17.966857871255577, "grad_norm": 0.3363014757633209, "learning_rate": 9.947251674300113e-05, "loss": 0.0458, "step": 28190 }, { "epoch": 17.97323135755258, "grad_norm": 0.7903397083282471, "learning_rate": 9.947171796820192e-05, "loss": 0.0575, "step": 28200 }, { "epoch": 17.979604843849586, "grad_norm": 0.610996425151825, "learning_rate": 9.947091859227464e-05, "loss": 0.0517, "step": 28210 }, { "epoch": 17.98597833014659, "grad_norm": 0.4906894266605377, "learning_rate": 9.947011861522899e-05, "loss": 0.0275, "step": 28220 }, { "epoch": 17.992351816443595, "grad_norm": 0.5095095634460449, "learning_rate": 9.946931803707468e-05, "loss": 0.044, "step": 28230 }, { "epoch": 17.9987253027406, "grad_norm": 0.4036194980144501, "learning_rate": 9.946851685782147e-05, "loss": 0.045, "step": 28240 }, { "epoch": 18.005098789037604, "grad_norm": 0.42551663517951965, "learning_rate": 9.946771507747906e-05, "loss": 0.047, "step": 28250 }, { "epoch": 18.011472275334608, "grad_norm": 0.643452525138855, "learning_rate": 9.946691269605721e-05, "loss": 0.0497, "step": 28260 }, { "epoch": 18.017845761631612, "grad_norm": 0.5258587002754211, "learning_rate": 9.946610971356567e-05, "loss": 0.0392, "step": 28270 }, { "epoch": 18.024219247928617, "grad_norm": 0.32409507036209106, "learning_rate": 9.946530613001418e-05, "loss": 0.0347, "step": 28280 }, { "epoch": 18.03059273422562, "grad_norm": 0.48933738470077515, "learning_rate": 9.946450194541253e-05, "loss": 0.033, "step": 28290 }, { "epoch": 18.036966220522626, "grad_norm": 0.5729212760925293, "learning_rate": 9.946369715977047e-05, "loss": 0.0334, "step": 28300 }, { "epoch": 18.04333970681963, "grad_norm": 0.6380464434623718, "learning_rate": 9.94628917730978e-05, "loss": 0.0378, "step": 28310 }, { "epoch": 18.049713193116634, "grad_norm": 0.6099402904510498, "learning_rate": 9.946208578540428e-05, "loss": 0.0388, "step": 28320 }, { "epoch": 18.05608667941364, "grad_norm": 0.7217431664466858, "learning_rate": 9.946127919669972e-05, "loss": 0.0374, "step": 28330 }, { "epoch": 18.062460165710643, "grad_norm": 0.8351543545722961, "learning_rate": 9.946047200699393e-05, "loss": 0.048, "step": 28340 }, { "epoch": 18.068833652007648, "grad_norm": 0.507265031337738, "learning_rate": 9.94596642162967e-05, "loss": 0.038, "step": 28350 }, { "epoch": 18.075207138304652, "grad_norm": 0.38116517663002014, "learning_rate": 9.945885582461785e-05, "loss": 0.0356, "step": 28360 }, { "epoch": 18.081580624601656, "grad_norm": 0.6168778538703918, "learning_rate": 9.94580468319672e-05, "loss": 0.0355, "step": 28370 }, { "epoch": 18.08795411089866, "grad_norm": 0.5395082831382751, "learning_rate": 9.94572372383546e-05, "loss": 0.0375, "step": 28380 }, { "epoch": 18.094327597195665, "grad_norm": 0.5192626118659973, "learning_rate": 9.945642704378986e-05, "loss": 0.036, "step": 28390 }, { "epoch": 18.10070108349267, "grad_norm": 0.6316375136375427, "learning_rate": 9.945561624828285e-05, "loss": 0.0493, "step": 28400 }, { "epoch": 18.107074569789674, "grad_norm": 0.5782895684242249, "learning_rate": 9.94548048518434e-05, "loss": 0.0338, "step": 28410 }, { "epoch": 18.11344805608668, "grad_norm": 0.5730478167533875, "learning_rate": 9.945399285448137e-05, "loss": 0.0543, "step": 28420 }, { "epoch": 18.119821542383683, "grad_norm": 0.5736696124076843, "learning_rate": 9.945318025620665e-05, "loss": 0.0467, "step": 28430 }, { "epoch": 18.126195028680687, "grad_norm": 0.9262635707855225, "learning_rate": 9.945236705702908e-05, "loss": 0.0419, "step": 28440 }, { "epoch": 18.13256851497769, "grad_norm": 0.5693608522415161, "learning_rate": 9.945155325695858e-05, "loss": 0.0401, "step": 28450 }, { "epoch": 18.138942001274696, "grad_norm": 0.45460256934165955, "learning_rate": 9.9450738856005e-05, "loss": 0.0326, "step": 28460 }, { "epoch": 18.1453154875717, "grad_norm": 0.4986417293548584, "learning_rate": 9.944992385417826e-05, "loss": 0.0331, "step": 28470 }, { "epoch": 18.151688973868705, "grad_norm": 0.44941723346710205, "learning_rate": 9.944910825148826e-05, "loss": 0.0449, "step": 28480 }, { "epoch": 18.15806246016571, "grad_norm": 0.5503662824630737, "learning_rate": 9.94482920479449e-05, "loss": 0.0555, "step": 28490 }, { "epoch": 18.164435946462714, "grad_norm": 0.7120167016983032, "learning_rate": 9.944747524355811e-05, "loss": 0.0314, "step": 28500 }, { "epoch": 18.170809432759718, "grad_norm": 0.819176435470581, "learning_rate": 9.944665783833782e-05, "loss": 0.0474, "step": 28510 }, { "epoch": 18.177182919056722, "grad_norm": 0.5169119238853455, "learning_rate": 9.944583983229394e-05, "loss": 0.0481, "step": 28520 }, { "epoch": 18.183556405353727, "grad_norm": 0.5913349986076355, "learning_rate": 9.944502122543641e-05, "loss": 0.0402, "step": 28530 }, { "epoch": 18.18992989165073, "grad_norm": 0.8381208181381226, "learning_rate": 9.94442020177752e-05, "loss": 0.0516, "step": 28540 }, { "epoch": 18.196303377947736, "grad_norm": 0.9311365485191345, "learning_rate": 9.944338220932025e-05, "loss": 0.0477, "step": 28550 }, { "epoch": 18.202676864244744, "grad_norm": 0.8508256077766418, "learning_rate": 9.944256180008154e-05, "loss": 0.0374, "step": 28560 }, { "epoch": 18.209050350541748, "grad_norm": 1.5421631336212158, "learning_rate": 9.9441740790069e-05, "loss": 0.0826, "step": 28570 }, { "epoch": 18.215423836838752, "grad_norm": 0.5102139711380005, "learning_rate": 9.944091917929264e-05, "loss": 0.042, "step": 28580 }, { "epoch": 18.221797323135757, "grad_norm": 0.60739666223526, "learning_rate": 9.944009696776244e-05, "loss": 0.0454, "step": 28590 }, { "epoch": 18.22817080943276, "grad_norm": 0.6634073257446289, "learning_rate": 9.943927415548837e-05, "loss": 0.0418, "step": 28600 }, { "epoch": 18.234544295729766, "grad_norm": 0.6340003609657288, "learning_rate": 9.943845074248044e-05, "loss": 0.0534, "step": 28610 }, { "epoch": 18.24091778202677, "grad_norm": 0.3241119980812073, "learning_rate": 9.943762672874865e-05, "loss": 0.0435, "step": 28620 }, { "epoch": 18.247291268323774, "grad_norm": 0.4300916790962219, "learning_rate": 9.943680211430304e-05, "loss": 0.0414, "step": 28630 }, { "epoch": 18.25366475462078, "grad_norm": 0.310761034488678, "learning_rate": 9.943597689915361e-05, "loss": 0.0357, "step": 28640 }, { "epoch": 18.260038240917783, "grad_norm": 0.7176051139831543, "learning_rate": 9.943515108331036e-05, "loss": 0.0393, "step": 28650 }, { "epoch": 18.266411727214788, "grad_norm": 0.5310367941856384, "learning_rate": 9.943432466678336e-05, "loss": 0.0361, "step": 28660 }, { "epoch": 18.272785213511792, "grad_norm": 0.3363970220088959, "learning_rate": 9.943349764958266e-05, "loss": 0.0426, "step": 28670 }, { "epoch": 18.279158699808796, "grad_norm": 1.1641829013824463, "learning_rate": 9.943267003171825e-05, "loss": 0.0407, "step": 28680 }, { "epoch": 18.2855321861058, "grad_norm": 0.9257823824882507, "learning_rate": 9.943184181320026e-05, "loss": 0.0446, "step": 28690 }, { "epoch": 18.291905672402805, "grad_norm": 0.8590275645256042, "learning_rate": 9.943101299403871e-05, "loss": 0.0389, "step": 28700 }, { "epoch": 18.29827915869981, "grad_norm": 0.4272668659687042, "learning_rate": 9.943018357424369e-05, "loss": 0.0373, "step": 28710 }, { "epoch": 18.304652644996814, "grad_norm": 0.33628958463668823, "learning_rate": 9.942935355382525e-05, "loss": 0.0455, "step": 28720 }, { "epoch": 18.31102613129382, "grad_norm": 0.5181383490562439, "learning_rate": 9.942852293279351e-05, "loss": 0.0503, "step": 28730 }, { "epoch": 18.317399617590823, "grad_norm": 0.5718725919723511, "learning_rate": 9.942769171115856e-05, "loss": 0.0615, "step": 28740 }, { "epoch": 18.323773103887827, "grad_norm": 0.7419205904006958, "learning_rate": 9.942685988893045e-05, "loss": 0.0352, "step": 28750 }, { "epoch": 18.33014659018483, "grad_norm": 0.6127808094024658, "learning_rate": 9.942602746611936e-05, "loss": 0.0517, "step": 28760 }, { "epoch": 18.336520076481836, "grad_norm": 0.40552347898483276, "learning_rate": 9.942519444273537e-05, "loss": 0.0312, "step": 28770 }, { "epoch": 18.34289356277884, "grad_norm": 0.3745090067386627, "learning_rate": 9.942436081878859e-05, "loss": 0.03, "step": 28780 }, { "epoch": 18.349267049075845, "grad_norm": 0.4008125960826874, "learning_rate": 9.942352659428916e-05, "loss": 0.0398, "step": 28790 }, { "epoch": 18.35564053537285, "grad_norm": 0.6054003834724426, "learning_rate": 9.942269176924722e-05, "loss": 0.0464, "step": 28800 }, { "epoch": 18.362014021669854, "grad_norm": 0.47119346261024475, "learning_rate": 9.942185634367292e-05, "loss": 0.0341, "step": 28810 }, { "epoch": 18.368387507966858, "grad_norm": 0.6217157244682312, "learning_rate": 9.942102031757639e-05, "loss": 0.0334, "step": 28820 }, { "epoch": 18.374760994263863, "grad_norm": 0.6488839387893677, "learning_rate": 9.942018369096781e-05, "loss": 0.0328, "step": 28830 }, { "epoch": 18.381134480560867, "grad_norm": 0.5669146776199341, "learning_rate": 9.941934646385734e-05, "loss": 0.0361, "step": 28840 }, { "epoch": 18.38750796685787, "grad_norm": 0.8450927138328552, "learning_rate": 9.941850863625515e-05, "loss": 0.0348, "step": 28850 }, { "epoch": 18.393881453154876, "grad_norm": 0.5762073397636414, "learning_rate": 9.941767020817142e-05, "loss": 0.0549, "step": 28860 }, { "epoch": 18.40025493945188, "grad_norm": 0.490662544965744, "learning_rate": 9.941683117961634e-05, "loss": 0.0394, "step": 28870 }, { "epoch": 18.406628425748885, "grad_norm": 0.8585314154624939, "learning_rate": 9.941599155060011e-05, "loss": 0.0541, "step": 28880 }, { "epoch": 18.41300191204589, "grad_norm": 0.70564866065979, "learning_rate": 9.941515132113292e-05, "loss": 0.0391, "step": 28890 }, { "epoch": 18.419375398342893, "grad_norm": 0.5997070670127869, "learning_rate": 9.941431049122498e-05, "loss": 0.0425, "step": 28900 }, { "epoch": 18.425748884639898, "grad_norm": 0.4389268755912781, "learning_rate": 9.94134690608865e-05, "loss": 0.0424, "step": 28910 }, { "epoch": 18.432122370936902, "grad_norm": 0.694188117980957, "learning_rate": 9.941262703012775e-05, "loss": 0.0403, "step": 28920 }, { "epoch": 18.438495857233907, "grad_norm": 0.6166446208953857, "learning_rate": 9.941178439895891e-05, "loss": 0.0342, "step": 28930 }, { "epoch": 18.44486934353091, "grad_norm": 0.502869725227356, "learning_rate": 9.941094116739025e-05, "loss": 0.0365, "step": 28940 }, { "epoch": 18.451242829827915, "grad_norm": 0.46384647488594055, "learning_rate": 9.941009733543199e-05, "loss": 0.0431, "step": 28950 }, { "epoch": 18.45761631612492, "grad_norm": 0.6524788737297058, "learning_rate": 9.940925290309439e-05, "loss": 0.0441, "step": 28960 }, { "epoch": 18.463989802421924, "grad_norm": 0.6036084294319153, "learning_rate": 9.940840787038773e-05, "loss": 0.0306, "step": 28970 }, { "epoch": 18.47036328871893, "grad_norm": 0.387070894241333, "learning_rate": 9.940756223732226e-05, "loss": 0.0387, "step": 28980 }, { "epoch": 18.476736775015933, "grad_norm": 0.3576427102088928, "learning_rate": 9.940671600390828e-05, "loss": 0.0289, "step": 28990 }, { "epoch": 18.483110261312937, "grad_norm": 0.7115057706832886, "learning_rate": 9.940586917015602e-05, "loss": 0.0429, "step": 29000 }, { "epoch": 18.489483747609942, "grad_norm": 0.5900391936302185, "learning_rate": 9.940502173607582e-05, "loss": 0.0415, "step": 29010 }, { "epoch": 18.495857233906946, "grad_norm": 0.2714010775089264, "learning_rate": 9.940417370167796e-05, "loss": 0.0449, "step": 29020 }, { "epoch": 18.50223072020395, "grad_norm": 0.4475362300872803, "learning_rate": 9.940332506697274e-05, "loss": 0.0322, "step": 29030 }, { "epoch": 18.508604206500955, "grad_norm": 0.7675207257270813, "learning_rate": 9.940247583197047e-05, "loss": 0.0409, "step": 29040 }, { "epoch": 18.51497769279796, "grad_norm": 0.626508355140686, "learning_rate": 9.940162599668149e-05, "loss": 0.0387, "step": 29050 }, { "epoch": 18.521351179094964, "grad_norm": 0.2268603891134262, "learning_rate": 9.94007755611161e-05, "loss": 0.0392, "step": 29060 }, { "epoch": 18.52772466539197, "grad_norm": 0.4541628062725067, "learning_rate": 9.939992452528466e-05, "loss": 0.0448, "step": 29070 }, { "epoch": 18.534098151688973, "grad_norm": 0.6295639872550964, "learning_rate": 9.939907288919747e-05, "loss": 0.0316, "step": 29080 }, { "epoch": 18.540471637985977, "grad_norm": 0.6493943333625793, "learning_rate": 9.939822065286493e-05, "loss": 0.0334, "step": 29090 }, { "epoch": 18.54684512428298, "grad_norm": 0.5731953978538513, "learning_rate": 9.939736781629736e-05, "loss": 0.036, "step": 29100 }, { "epoch": 18.553218610579986, "grad_norm": 0.4753711223602295, "learning_rate": 9.939651437950512e-05, "loss": 0.0426, "step": 29110 }, { "epoch": 18.55959209687699, "grad_norm": 0.6338264346122742, "learning_rate": 9.939566034249861e-05, "loss": 0.0332, "step": 29120 }, { "epoch": 18.565965583173995, "grad_norm": 0.47118523716926575, "learning_rate": 9.939480570528819e-05, "loss": 0.0382, "step": 29130 }, { "epoch": 18.572339069471, "grad_norm": 0.7295191287994385, "learning_rate": 9.939395046788422e-05, "loss": 0.0261, "step": 29140 }, { "epoch": 18.578712555768004, "grad_norm": 0.4889342486858368, "learning_rate": 9.939309463029713e-05, "loss": 0.0411, "step": 29150 }, { "epoch": 18.585086042065008, "grad_norm": 0.6443138718605042, "learning_rate": 9.939223819253731e-05, "loss": 0.0415, "step": 29160 }, { "epoch": 18.591459528362012, "grad_norm": 0.5513317584991455, "learning_rate": 9.939138115461516e-05, "loss": 0.0398, "step": 29170 }, { "epoch": 18.597833014659017, "grad_norm": 0.6576837301254272, "learning_rate": 9.939052351654111e-05, "loss": 0.0587, "step": 29180 }, { "epoch": 18.604206500956025, "grad_norm": 0.48124411702156067, "learning_rate": 9.938966527832555e-05, "loss": 0.0348, "step": 29190 }, { "epoch": 18.61057998725303, "grad_norm": 0.9200423359870911, "learning_rate": 9.938880643997893e-05, "loss": 0.0438, "step": 29200 }, { "epoch": 18.616953473550033, "grad_norm": 0.7226598858833313, "learning_rate": 9.938794700151168e-05, "loss": 0.0392, "step": 29210 }, { "epoch": 18.623326959847038, "grad_norm": 0.5152060985565186, "learning_rate": 9.938708696293424e-05, "loss": 0.0454, "step": 29220 }, { "epoch": 18.629700446144042, "grad_norm": 0.41204386949539185, "learning_rate": 9.938622632425707e-05, "loss": 0.0463, "step": 29230 }, { "epoch": 18.636073932441047, "grad_norm": 0.6263889670372009, "learning_rate": 9.93853650854906e-05, "loss": 0.0498, "step": 29240 }, { "epoch": 18.64244741873805, "grad_norm": 0.4043098986148834, "learning_rate": 9.938450324664535e-05, "loss": 0.0297, "step": 29250 }, { "epoch": 18.648820905035056, "grad_norm": 0.47623881697654724, "learning_rate": 9.938364080773175e-05, "loss": 0.0359, "step": 29260 }, { "epoch": 18.65519439133206, "grad_norm": 0.49849221110343933, "learning_rate": 9.938277776876028e-05, "loss": 0.0391, "step": 29270 }, { "epoch": 18.661567877629064, "grad_norm": 0.3075621724128723, "learning_rate": 9.938191412974143e-05, "loss": 0.0381, "step": 29280 }, { "epoch": 18.66794136392607, "grad_norm": 0.7212318778038025, "learning_rate": 9.93810498906857e-05, "loss": 0.0391, "step": 29290 }, { "epoch": 18.674314850223073, "grad_norm": 0.44030001759529114, "learning_rate": 9.93801850516036e-05, "loss": 0.0359, "step": 29300 }, { "epoch": 18.680688336520078, "grad_norm": 0.324235737323761, "learning_rate": 9.937931961250564e-05, "loss": 0.0378, "step": 29310 }, { "epoch": 18.687061822817082, "grad_norm": 0.6790275573730469, "learning_rate": 9.93784535734023e-05, "loss": 0.0457, "step": 29320 }, { "epoch": 18.693435309114086, "grad_norm": 0.508127748966217, "learning_rate": 9.937758693430414e-05, "loss": 0.0323, "step": 29330 }, { "epoch": 18.69980879541109, "grad_norm": 0.569007933139801, "learning_rate": 9.937671969522169e-05, "loss": 0.0318, "step": 29340 }, { "epoch": 18.706182281708095, "grad_norm": 0.45339545607566833, "learning_rate": 9.937585185616546e-05, "loss": 0.0361, "step": 29350 }, { "epoch": 18.7125557680051, "grad_norm": 0.6050662994384766, "learning_rate": 9.937498341714602e-05, "loss": 0.0458, "step": 29360 }, { "epoch": 18.718929254302104, "grad_norm": 0.5710495710372925, "learning_rate": 9.93741143781739e-05, "loss": 0.041, "step": 29370 }, { "epoch": 18.72530274059911, "grad_norm": 0.4593247175216675, "learning_rate": 9.937324473925969e-05, "loss": 0.0367, "step": 29380 }, { "epoch": 18.731676226896113, "grad_norm": 0.6179113388061523, "learning_rate": 9.937237450041393e-05, "loss": 0.0425, "step": 29390 }, { "epoch": 18.738049713193117, "grad_norm": 0.7028769850730896, "learning_rate": 9.93715036616472e-05, "loss": 0.0343, "step": 29400 }, { "epoch": 18.74442319949012, "grad_norm": 0.4269255995750427, "learning_rate": 9.93706322229701e-05, "loss": 0.037, "step": 29410 }, { "epoch": 18.750796685787126, "grad_norm": 0.4630488157272339, "learning_rate": 9.936976018439321e-05, "loss": 0.041, "step": 29420 }, { "epoch": 18.75717017208413, "grad_norm": 0.43804648518562317, "learning_rate": 9.936888754592711e-05, "loss": 0.0302, "step": 29430 }, { "epoch": 18.763543658381135, "grad_norm": 0.7423017024993896, "learning_rate": 9.93680143075824e-05, "loss": 0.0425, "step": 29440 }, { "epoch": 18.76991714467814, "grad_norm": 0.6526033878326416, "learning_rate": 9.936714046936972e-05, "loss": 0.0429, "step": 29450 }, { "epoch": 18.776290630975144, "grad_norm": 0.8027708530426025, "learning_rate": 9.936626603129969e-05, "loss": 0.0431, "step": 29460 }, { "epoch": 18.782664117272148, "grad_norm": 0.4835878610610962, "learning_rate": 9.936539099338289e-05, "loss": 0.0277, "step": 29470 }, { "epoch": 18.789037603569152, "grad_norm": 0.5432160496711731, "learning_rate": 9.936451535562999e-05, "loss": 0.0497, "step": 29480 }, { "epoch": 18.795411089866157, "grad_norm": 0.5552000999450684, "learning_rate": 9.936363911805162e-05, "loss": 0.0337, "step": 29490 }, { "epoch": 18.80178457616316, "grad_norm": 0.2930489778518677, "learning_rate": 9.936276228065842e-05, "loss": 0.046, "step": 29500 }, { "epoch": 18.808158062460166, "grad_norm": 0.541323184967041, "learning_rate": 9.936188484346107e-05, "loss": 0.0472, "step": 29510 }, { "epoch": 18.81453154875717, "grad_norm": 0.4334292709827423, "learning_rate": 9.936100680647021e-05, "loss": 0.0328, "step": 29520 }, { "epoch": 18.820905035054174, "grad_norm": 0.6419758200645447, "learning_rate": 9.93601281696965e-05, "loss": 0.0344, "step": 29530 }, { "epoch": 18.82727852135118, "grad_norm": 0.44829273223876953, "learning_rate": 9.935924893315065e-05, "loss": 0.0416, "step": 29540 }, { "epoch": 18.833652007648183, "grad_norm": 0.3829452395439148, "learning_rate": 9.93583690968433e-05, "loss": 0.0447, "step": 29550 }, { "epoch": 18.840025493945188, "grad_norm": 0.4967927038669586, "learning_rate": 9.935748866078518e-05, "loss": 0.0357, "step": 29560 }, { "epoch": 18.846398980242192, "grad_norm": 0.7680585980415344, "learning_rate": 9.935660762498696e-05, "loss": 0.0333, "step": 29570 }, { "epoch": 18.852772466539196, "grad_norm": 0.4204786419868469, "learning_rate": 9.935572598945937e-05, "loss": 0.0454, "step": 29580 }, { "epoch": 18.8591459528362, "grad_norm": 0.6508587002754211, "learning_rate": 9.935484375421309e-05, "loss": 0.0428, "step": 29590 }, { "epoch": 18.865519439133205, "grad_norm": 0.3644666373729706, "learning_rate": 9.935396091925889e-05, "loss": 0.0302, "step": 29600 }, { "epoch": 18.87189292543021, "grad_norm": 0.5341609716415405, "learning_rate": 9.935307748460744e-05, "loss": 0.0337, "step": 29610 }, { "epoch": 18.878266411727214, "grad_norm": 0.38523614406585693, "learning_rate": 9.935219345026951e-05, "loss": 0.0219, "step": 29620 }, { "epoch": 18.88463989802422, "grad_norm": 0.39375802874565125, "learning_rate": 9.935130881625583e-05, "loss": 0.0418, "step": 29630 }, { "epoch": 18.891013384321223, "grad_norm": 0.35671064257621765, "learning_rate": 9.935042358257715e-05, "loss": 0.0314, "step": 29640 }, { "epoch": 18.897386870618227, "grad_norm": 0.525241494178772, "learning_rate": 9.934953774924424e-05, "loss": 0.0336, "step": 29650 }, { "epoch": 18.90376035691523, "grad_norm": 0.487211674451828, "learning_rate": 9.934865131626784e-05, "loss": 0.0406, "step": 29660 }, { "epoch": 18.910133843212236, "grad_norm": 0.32760319113731384, "learning_rate": 9.934776428365875e-05, "loss": 0.0475, "step": 29670 }, { "epoch": 18.91650732950924, "grad_norm": 1.010451078414917, "learning_rate": 9.934687665142773e-05, "loss": 0.0371, "step": 29680 }, { "epoch": 18.922880815806245, "grad_norm": 0.4842472970485687, "learning_rate": 9.934598841958555e-05, "loss": 0.0337, "step": 29690 }, { "epoch": 18.92925430210325, "grad_norm": 0.7316793203353882, "learning_rate": 9.934509958814304e-05, "loss": 0.0368, "step": 29700 }, { "epoch": 18.935627788400254, "grad_norm": 0.5729722380638123, "learning_rate": 9.934421015711097e-05, "loss": 0.0417, "step": 29710 }, { "epoch": 18.942001274697258, "grad_norm": 0.34655457735061646, "learning_rate": 9.934332012650017e-05, "loss": 0.0445, "step": 29720 }, { "epoch": 18.948374760994263, "grad_norm": 0.4603334367275238, "learning_rate": 9.934242949632143e-05, "loss": 0.0372, "step": 29730 }, { "epoch": 18.954748247291267, "grad_norm": 0.3202730119228363, "learning_rate": 9.934153826658557e-05, "loss": 0.0451, "step": 29740 }, { "epoch": 18.96112173358827, "grad_norm": 0.3960217833518982, "learning_rate": 9.934064643730345e-05, "loss": 0.0363, "step": 29750 }, { "epoch": 18.967495219885276, "grad_norm": 0.5772948861122131, "learning_rate": 9.93397540084859e-05, "loss": 0.0362, "step": 29760 }, { "epoch": 18.97386870618228, "grad_norm": 0.5629610419273376, "learning_rate": 9.933886098014374e-05, "loss": 0.0417, "step": 29770 }, { "epoch": 18.980242192479285, "grad_norm": 0.7069763541221619, "learning_rate": 9.933796735228785e-05, "loss": 0.0328, "step": 29780 }, { "epoch": 18.98661567877629, "grad_norm": 0.8418090343475342, "learning_rate": 9.933707312492906e-05, "loss": 0.0418, "step": 29790 }, { "epoch": 18.992989165073293, "grad_norm": 0.6206852793693542, "learning_rate": 9.933617829807825e-05, "loss": 0.032, "step": 29800 }, { "epoch": 18.999362651370298, "grad_norm": 0.9068872928619385, "learning_rate": 9.933528287174629e-05, "loss": 0.0359, "step": 29810 }, { "epoch": 19.005736137667306, "grad_norm": 0.5571058988571167, "learning_rate": 9.933438684594407e-05, "loss": 0.0353, "step": 29820 }, { "epoch": 19.01210962396431, "grad_norm": 0.7168715000152588, "learning_rate": 9.933349022068248e-05, "loss": 0.0412, "step": 29830 }, { "epoch": 19.018483110261315, "grad_norm": 0.45915642380714417, "learning_rate": 9.93325929959724e-05, "loss": 0.0587, "step": 29840 }, { "epoch": 19.02485659655832, "grad_norm": 0.8532593250274658, "learning_rate": 9.933169517182473e-05, "loss": 0.0415, "step": 29850 }, { "epoch": 19.031230082855323, "grad_norm": 0.6767350435256958, "learning_rate": 9.933079674825038e-05, "loss": 0.0539, "step": 29860 }, { "epoch": 19.037603569152328, "grad_norm": 0.4505847692489624, "learning_rate": 9.93298977252603e-05, "loss": 0.0402, "step": 29870 }, { "epoch": 19.043977055449332, "grad_norm": 0.40878984332084656, "learning_rate": 9.932899810286535e-05, "loss": 0.035, "step": 29880 }, { "epoch": 19.050350541746337, "grad_norm": 0.6644253134727478, "learning_rate": 9.932809788107653e-05, "loss": 0.0441, "step": 29890 }, { "epoch": 19.05672402804334, "grad_norm": 0.8221628069877625, "learning_rate": 9.932719705990474e-05, "loss": 0.0394, "step": 29900 }, { "epoch": 19.063097514340345, "grad_norm": 0.4734356105327606, "learning_rate": 9.932629563936094e-05, "loss": 0.0459, "step": 29910 }, { "epoch": 19.06947100063735, "grad_norm": 0.7842265367507935, "learning_rate": 9.932539361945606e-05, "loss": 0.0428, "step": 29920 }, { "epoch": 19.075844486934354, "grad_norm": 0.6345791816711426, "learning_rate": 9.932449100020107e-05, "loss": 0.0313, "step": 29930 }, { "epoch": 19.08221797323136, "grad_norm": 0.4702848196029663, "learning_rate": 9.932358778160697e-05, "loss": 0.0467, "step": 29940 }, { "epoch": 19.088591459528363, "grad_norm": 0.46002575755119324, "learning_rate": 9.932268396368469e-05, "loss": 0.0483, "step": 29950 }, { "epoch": 19.094964945825367, "grad_norm": 0.6662471890449524, "learning_rate": 9.932177954644524e-05, "loss": 0.0331, "step": 29960 }, { "epoch": 19.101338432122372, "grad_norm": 0.425302118062973, "learning_rate": 9.93208745298996e-05, "loss": 0.0374, "step": 29970 }, { "epoch": 19.107711918419376, "grad_norm": 0.5735731720924377, "learning_rate": 9.931996891405877e-05, "loss": 0.0371, "step": 29980 }, { "epoch": 19.11408540471638, "grad_norm": 0.4919684827327728, "learning_rate": 9.931906269893373e-05, "loss": 0.0411, "step": 29990 }, { "epoch": 19.120458891013385, "grad_norm": 0.4763234257698059, "learning_rate": 9.931815588453555e-05, "loss": 0.037, "step": 30000 }, { "epoch": 19.12683237731039, "grad_norm": 0.5986319780349731, "learning_rate": 9.93172484708752e-05, "loss": 0.0327, "step": 30010 }, { "epoch": 19.133205863607394, "grad_norm": 0.5059881806373596, "learning_rate": 9.93163404579637e-05, "loss": 0.0348, "step": 30020 }, { "epoch": 19.1395793499044, "grad_norm": 0.8306627869606018, "learning_rate": 9.931543184581211e-05, "loss": 0.0388, "step": 30030 }, { "epoch": 19.145952836201403, "grad_norm": 0.28949907422065735, "learning_rate": 9.931452263443146e-05, "loss": 0.026, "step": 30040 }, { "epoch": 19.152326322498407, "grad_norm": 0.6049891114234924, "learning_rate": 9.931361282383282e-05, "loss": 0.0407, "step": 30050 }, { "epoch": 19.15869980879541, "grad_norm": 0.7492976188659668, "learning_rate": 9.93127024140272e-05, "loss": 0.057, "step": 30060 }, { "epoch": 19.165073295092416, "grad_norm": 0.6889173984527588, "learning_rate": 9.931179140502569e-05, "loss": 0.036, "step": 30070 }, { "epoch": 19.17144678138942, "grad_norm": 0.6662917137145996, "learning_rate": 9.931087979683937e-05, "loss": 0.0403, "step": 30080 }, { "epoch": 19.177820267686425, "grad_norm": 0.6832749247550964, "learning_rate": 9.930996758947929e-05, "loss": 0.0317, "step": 30090 }, { "epoch": 19.18419375398343, "grad_norm": 0.3887522220611572, "learning_rate": 9.930905478295655e-05, "loss": 0.0325, "step": 30100 }, { "epoch": 19.190567240280433, "grad_norm": 0.49981802701950073, "learning_rate": 9.930814137728224e-05, "loss": 0.0403, "step": 30110 }, { "epoch": 19.196940726577438, "grad_norm": 0.5295567512512207, "learning_rate": 9.930722737246747e-05, "loss": 0.0534, "step": 30120 }, { "epoch": 19.203314212874442, "grad_norm": 0.7078784108161926, "learning_rate": 9.930631276852331e-05, "loss": 0.0375, "step": 30130 }, { "epoch": 19.209687699171447, "grad_norm": 0.45943304896354675, "learning_rate": 9.930539756546091e-05, "loss": 0.0362, "step": 30140 }, { "epoch": 19.21606118546845, "grad_norm": 0.622878909111023, "learning_rate": 9.930448176329139e-05, "loss": 0.0317, "step": 30150 }, { "epoch": 19.222434671765456, "grad_norm": 0.4926847517490387, "learning_rate": 9.930356536202585e-05, "loss": 0.0352, "step": 30160 }, { "epoch": 19.22880815806246, "grad_norm": 0.9150048494338989, "learning_rate": 9.930264836167544e-05, "loss": 0.0372, "step": 30170 }, { "epoch": 19.235181644359464, "grad_norm": 0.4262911081314087, "learning_rate": 9.93017307622513e-05, "loss": 0.0362, "step": 30180 }, { "epoch": 19.24155513065647, "grad_norm": 0.41691434383392334, "learning_rate": 9.930081256376461e-05, "loss": 0.0275, "step": 30190 }, { "epoch": 19.247928616953473, "grad_norm": 0.8108784556388855, "learning_rate": 9.929989376622646e-05, "loss": 0.0361, "step": 30200 }, { "epoch": 19.254302103250478, "grad_norm": 0.41191670298576355, "learning_rate": 9.929897436964807e-05, "loss": 0.0368, "step": 30210 }, { "epoch": 19.260675589547482, "grad_norm": 0.4684663414955139, "learning_rate": 9.92980543740406e-05, "loss": 0.0315, "step": 30220 }, { "epoch": 19.267049075844486, "grad_norm": 0.555474042892456, "learning_rate": 9.929713377941523e-05, "loss": 0.0308, "step": 30230 }, { "epoch": 19.27342256214149, "grad_norm": 0.43855366110801697, "learning_rate": 9.929621258578312e-05, "loss": 0.0361, "step": 30240 }, { "epoch": 19.279796048438495, "grad_norm": 0.6779509782791138, "learning_rate": 9.929529079315551e-05, "loss": 0.0462, "step": 30250 }, { "epoch": 19.2861695347355, "grad_norm": 0.6538847088813782, "learning_rate": 9.929436840154355e-05, "loss": 0.0373, "step": 30260 }, { "epoch": 19.292543021032504, "grad_norm": 0.48029494285583496, "learning_rate": 9.929344541095848e-05, "loss": 0.0259, "step": 30270 }, { "epoch": 19.29891650732951, "grad_norm": 0.41328009963035583, "learning_rate": 9.92925218214115e-05, "loss": 0.0445, "step": 30280 }, { "epoch": 19.305289993626513, "grad_norm": 0.44898107647895813, "learning_rate": 9.929159763291387e-05, "loss": 0.0513, "step": 30290 }, { "epoch": 19.311663479923517, "grad_norm": 0.48360171914100647, "learning_rate": 9.929067284547677e-05, "loss": 0.0424, "step": 30300 }, { "epoch": 19.31803696622052, "grad_norm": 0.5419996380805969, "learning_rate": 9.928974745911146e-05, "loss": 0.0305, "step": 30310 }, { "epoch": 19.324410452517526, "grad_norm": 0.4238566756248474, "learning_rate": 9.928882147382917e-05, "loss": 0.0511, "step": 30320 }, { "epoch": 19.33078393881453, "grad_norm": 0.44084325432777405, "learning_rate": 9.928789488964118e-05, "loss": 0.0418, "step": 30330 }, { "epoch": 19.337157425111535, "grad_norm": 0.48497602343559265, "learning_rate": 9.928696770655872e-05, "loss": 0.0459, "step": 30340 }, { "epoch": 19.34353091140854, "grad_norm": 0.734505832195282, "learning_rate": 9.928603992459306e-05, "loss": 0.0358, "step": 30350 }, { "epoch": 19.349904397705544, "grad_norm": 0.9555234313011169, "learning_rate": 9.928511154375551e-05, "loss": 0.0411, "step": 30360 }, { "epoch": 19.356277884002548, "grad_norm": 0.42453572154045105, "learning_rate": 9.928418256405729e-05, "loss": 0.0488, "step": 30370 }, { "epoch": 19.362651370299552, "grad_norm": 0.35866954922676086, "learning_rate": 9.928325298550973e-05, "loss": 0.0491, "step": 30380 }, { "epoch": 19.369024856596557, "grad_norm": 0.6244329810142517, "learning_rate": 9.928232280812413e-05, "loss": 0.0489, "step": 30390 }, { "epoch": 19.37539834289356, "grad_norm": 0.6762614250183105, "learning_rate": 9.928139203191176e-05, "loss": 0.0467, "step": 30400 }, { "epoch": 19.381771829190566, "grad_norm": 0.5204678177833557, "learning_rate": 9.928046065688396e-05, "loss": 0.0358, "step": 30410 }, { "epoch": 19.38814531548757, "grad_norm": 0.6249240040779114, "learning_rate": 9.927952868305203e-05, "loss": 0.0328, "step": 30420 }, { "epoch": 19.394518801784574, "grad_norm": 0.5215117931365967, "learning_rate": 9.927859611042729e-05, "loss": 0.0343, "step": 30430 }, { "epoch": 19.40089228808158, "grad_norm": 0.569284975528717, "learning_rate": 9.92776629390211e-05, "loss": 0.0354, "step": 30440 }, { "epoch": 19.407265774378587, "grad_norm": 0.7911409139633179, "learning_rate": 9.927672916884476e-05, "loss": 0.047, "step": 30450 }, { "epoch": 19.41363926067559, "grad_norm": 0.5741820931434631, "learning_rate": 9.927579479990964e-05, "loss": 0.0331, "step": 30460 }, { "epoch": 19.420012746972596, "grad_norm": 0.4142441749572754, "learning_rate": 9.927485983222709e-05, "loss": 0.0338, "step": 30470 }, { "epoch": 19.4263862332696, "grad_norm": 0.44358593225479126, "learning_rate": 9.927392426580847e-05, "loss": 0.0386, "step": 30480 }, { "epoch": 19.432759719566604, "grad_norm": 0.3871499300003052, "learning_rate": 9.927298810066514e-05, "loss": 0.0264, "step": 30490 }, { "epoch": 19.43913320586361, "grad_norm": 0.7500891089439392, "learning_rate": 9.92720513368085e-05, "loss": 0.0438, "step": 30500 }, { "epoch": 19.445506692160613, "grad_norm": 0.7373768091201782, "learning_rate": 9.927111397424991e-05, "loss": 0.0371, "step": 30510 }, { "epoch": 19.451880178457618, "grad_norm": 0.5835720896720886, "learning_rate": 9.927017601300077e-05, "loss": 0.0439, "step": 30520 }, { "epoch": 19.458253664754622, "grad_norm": 0.519656240940094, "learning_rate": 9.926923745307246e-05, "loss": 0.04, "step": 30530 }, { "epoch": 19.464627151051626, "grad_norm": 0.5708519220352173, "learning_rate": 9.926829829447641e-05, "loss": 0.0388, "step": 30540 }, { "epoch": 19.47100063734863, "grad_norm": 0.45617032051086426, "learning_rate": 9.926735853722401e-05, "loss": 0.0366, "step": 30550 }, { "epoch": 19.477374123645635, "grad_norm": 0.4663662016391754, "learning_rate": 9.926641818132669e-05, "loss": 0.0312, "step": 30560 }, { "epoch": 19.48374760994264, "grad_norm": 0.5140721797943115, "learning_rate": 9.926547722679586e-05, "loss": 0.0541, "step": 30570 }, { "epoch": 19.490121096239644, "grad_norm": 0.5515722036361694, "learning_rate": 9.9264535673643e-05, "loss": 0.0473, "step": 30580 }, { "epoch": 19.49649458253665, "grad_norm": 0.6542114615440369, "learning_rate": 9.92635935218795e-05, "loss": 0.0513, "step": 30590 }, { "epoch": 19.502868068833653, "grad_norm": 0.5618543028831482, "learning_rate": 9.926265077151682e-05, "loss": 0.0337, "step": 30600 }, { "epoch": 19.509241555130657, "grad_norm": 0.3177219033241272, "learning_rate": 9.926170742256643e-05, "loss": 0.0483, "step": 30610 }, { "epoch": 19.51561504142766, "grad_norm": 0.41662949323654175, "learning_rate": 9.926076347503979e-05, "loss": 0.035, "step": 30620 }, { "epoch": 19.521988527724666, "grad_norm": 0.5761436223983765, "learning_rate": 9.925981892894834e-05, "loss": 0.0446, "step": 30630 }, { "epoch": 19.52836201402167, "grad_norm": 0.6705889701843262, "learning_rate": 9.925887378430361e-05, "loss": 0.0337, "step": 30640 }, { "epoch": 19.534735500318675, "grad_norm": 0.4133850336074829, "learning_rate": 9.925792804111703e-05, "loss": 0.0305, "step": 30650 }, { "epoch": 19.54110898661568, "grad_norm": 0.69058758020401, "learning_rate": 9.925698169940013e-05, "loss": 0.0458, "step": 30660 }, { "epoch": 19.547482472912684, "grad_norm": 0.7333258390426636, "learning_rate": 9.925603475916439e-05, "loss": 0.0335, "step": 30670 }, { "epoch": 19.553855959209688, "grad_norm": 0.5526683330535889, "learning_rate": 9.925508722042134e-05, "loss": 0.0362, "step": 30680 }, { "epoch": 19.560229445506693, "grad_norm": 0.6082180738449097, "learning_rate": 9.925413908318247e-05, "loss": 0.0399, "step": 30690 }, { "epoch": 19.566602931803697, "grad_norm": 0.38242268562316895, "learning_rate": 9.925319034745928e-05, "loss": 0.0382, "step": 30700 }, { "epoch": 19.5729764181007, "grad_norm": 0.4181293547153473, "learning_rate": 9.925224101326335e-05, "loss": 0.0414, "step": 30710 }, { "epoch": 19.579349904397706, "grad_norm": 0.4419921040534973, "learning_rate": 9.925129108060618e-05, "loss": 0.0382, "step": 30720 }, { "epoch": 19.58572339069471, "grad_norm": 0.5485382080078125, "learning_rate": 9.925034054949933e-05, "loss": 0.0428, "step": 30730 }, { "epoch": 19.592096876991715, "grad_norm": 1.3604178428649902, "learning_rate": 9.924938941995434e-05, "loss": 0.0469, "step": 30740 }, { "epoch": 19.59847036328872, "grad_norm": 0.7106276750564575, "learning_rate": 9.924843769198276e-05, "loss": 0.0431, "step": 30750 }, { "epoch": 19.604843849585723, "grad_norm": 0.6182977557182312, "learning_rate": 9.924748536559618e-05, "loss": 0.0423, "step": 30760 }, { "epoch": 19.611217335882728, "grad_norm": 0.7213919758796692, "learning_rate": 9.924653244080614e-05, "loss": 0.036, "step": 30770 }, { "epoch": 19.617590822179732, "grad_norm": 0.4512563645839691, "learning_rate": 9.924557891762425e-05, "loss": 0.0374, "step": 30780 }, { "epoch": 19.623964308476737, "grad_norm": 0.8976032137870789, "learning_rate": 9.924462479606207e-05, "loss": 0.0396, "step": 30790 }, { "epoch": 19.63033779477374, "grad_norm": 0.44669491052627563, "learning_rate": 9.92436700761312e-05, "loss": 0.035, "step": 30800 }, { "epoch": 19.636711281070745, "grad_norm": 0.49275678396224976, "learning_rate": 9.924271475784325e-05, "loss": 0.0396, "step": 30810 }, { "epoch": 19.64308476736775, "grad_norm": 0.40826186537742615, "learning_rate": 9.924175884120983e-05, "loss": 0.0418, "step": 30820 }, { "epoch": 19.649458253664754, "grad_norm": 0.7369924783706665, "learning_rate": 9.924080232624254e-05, "loss": 0.033, "step": 30830 }, { "epoch": 19.65583173996176, "grad_norm": 0.5148134231567383, "learning_rate": 9.923984521295302e-05, "loss": 0.0342, "step": 30840 }, { "epoch": 19.662205226258763, "grad_norm": 0.734971284866333, "learning_rate": 9.923888750135289e-05, "loss": 0.0339, "step": 30850 }, { "epoch": 19.668578712555767, "grad_norm": 0.6121981739997864, "learning_rate": 9.923792919145377e-05, "loss": 0.031, "step": 30860 }, { "epoch": 19.674952198852772, "grad_norm": 0.49764755368232727, "learning_rate": 9.923697028326733e-05, "loss": 0.0441, "step": 30870 }, { "epoch": 19.681325685149776, "grad_norm": 0.5548027753829956, "learning_rate": 9.923601077680523e-05, "loss": 0.0362, "step": 30880 }, { "epoch": 19.68769917144678, "grad_norm": 0.8668046593666077, "learning_rate": 9.923505067207909e-05, "loss": 0.04, "step": 30890 }, { "epoch": 19.694072657743785, "grad_norm": 0.4896691143512726, "learning_rate": 9.92340899691006e-05, "loss": 0.0469, "step": 30900 }, { "epoch": 19.70044614404079, "grad_norm": 0.49435439705848694, "learning_rate": 9.923312866788145e-05, "loss": 0.0384, "step": 30910 }, { "epoch": 19.706819630337794, "grad_norm": 0.9273898601531982, "learning_rate": 9.923216676843329e-05, "loss": 0.0528, "step": 30920 }, { "epoch": 19.7131931166348, "grad_norm": 0.7938147187232971, "learning_rate": 9.923120427076782e-05, "loss": 0.041, "step": 30930 }, { "epoch": 19.719566602931803, "grad_norm": 0.4727533459663391, "learning_rate": 9.923024117489674e-05, "loss": 0.033, "step": 30940 }, { "epoch": 19.725940089228807, "grad_norm": 0.790953516960144, "learning_rate": 9.922927748083175e-05, "loss": 0.0432, "step": 30950 }, { "epoch": 19.73231357552581, "grad_norm": 0.3349590301513672, "learning_rate": 9.922831318858457e-05, "loss": 0.029, "step": 30960 }, { "epoch": 19.738687061822816, "grad_norm": 0.6204319000244141, "learning_rate": 9.922734829816688e-05, "loss": 0.044, "step": 30970 }, { "epoch": 19.74506054811982, "grad_norm": 0.5555143356323242, "learning_rate": 9.922638280959045e-05, "loss": 0.0315, "step": 30980 }, { "epoch": 19.751434034416825, "grad_norm": 0.36220285296440125, "learning_rate": 9.922541672286699e-05, "loss": 0.0336, "step": 30990 }, { "epoch": 19.75780752071383, "grad_norm": 0.49983927607536316, "learning_rate": 9.922445003800823e-05, "loss": 0.0423, "step": 31000 }, { "epoch": 19.764181007010833, "grad_norm": 0.6305146813392639, "learning_rate": 9.922348275502593e-05, "loss": 0.0393, "step": 31010 }, { "epoch": 19.770554493307838, "grad_norm": 0.4823150634765625, "learning_rate": 9.922251487393184e-05, "loss": 0.0446, "step": 31020 }, { "epoch": 19.776927979604842, "grad_norm": 0.2731166183948517, "learning_rate": 9.922154639473771e-05, "loss": 0.0274, "step": 31030 }, { "epoch": 19.783301465901847, "grad_norm": 0.3890707492828369, "learning_rate": 9.922057731745535e-05, "loss": 0.0336, "step": 31040 }, { "epoch": 19.78967495219885, "grad_norm": 0.33899155259132385, "learning_rate": 9.921960764209648e-05, "loss": 0.0323, "step": 31050 }, { "epoch": 19.79604843849586, "grad_norm": 0.5514169335365295, "learning_rate": 9.92186373686729e-05, "loss": 0.0322, "step": 31060 }, { "epoch": 19.80242192479286, "grad_norm": 0.5769526958465576, "learning_rate": 9.921766649719642e-05, "loss": 0.0316, "step": 31070 }, { "epoch": 19.808795411089868, "grad_norm": 0.41242578625679016, "learning_rate": 9.921669502767882e-05, "loss": 0.0327, "step": 31080 }, { "epoch": 19.815168897386872, "grad_norm": 0.8164010047912598, "learning_rate": 9.921572296013191e-05, "loss": 0.0351, "step": 31090 }, { "epoch": 19.821542383683877, "grad_norm": 0.4675033390522003, "learning_rate": 9.921475029456749e-05, "loss": 0.0426, "step": 31100 }, { "epoch": 19.82791586998088, "grad_norm": 0.6598914265632629, "learning_rate": 9.92137770309974e-05, "loss": 0.0341, "step": 31110 }, { "epoch": 19.834289356277885, "grad_norm": 0.6407289505004883, "learning_rate": 9.921280316943344e-05, "loss": 0.0456, "step": 31120 }, { "epoch": 19.84066284257489, "grad_norm": 0.2580413520336151, "learning_rate": 9.921182870988745e-05, "loss": 0.0338, "step": 31130 }, { "epoch": 19.847036328871894, "grad_norm": 0.7133857011795044, "learning_rate": 9.92108536523713e-05, "loss": 0.0433, "step": 31140 }, { "epoch": 19.8534098151689, "grad_norm": 0.5073107481002808, "learning_rate": 9.92098779968968e-05, "loss": 0.0324, "step": 31150 }, { "epoch": 19.859783301465903, "grad_norm": 0.6896771192550659, "learning_rate": 9.920890174347583e-05, "loss": 0.0329, "step": 31160 }, { "epoch": 19.866156787762907, "grad_norm": 0.4341394007205963, "learning_rate": 9.920792489212023e-05, "loss": 0.0395, "step": 31170 }, { "epoch": 19.872530274059912, "grad_norm": 0.5641313195228577, "learning_rate": 9.920694744284189e-05, "loss": 0.0466, "step": 31180 }, { "epoch": 19.878903760356916, "grad_norm": 0.5229228138923645, "learning_rate": 9.920596939565268e-05, "loss": 0.0379, "step": 31190 }, { "epoch": 19.88527724665392, "grad_norm": 0.4775829315185547, "learning_rate": 9.920499075056448e-05, "loss": 0.0355, "step": 31200 }, { "epoch": 19.891650732950925, "grad_norm": 0.46678051352500916, "learning_rate": 9.92040115075892e-05, "loss": 0.0316, "step": 31210 }, { "epoch": 19.89802421924793, "grad_norm": 0.4558700621128082, "learning_rate": 9.920303166673872e-05, "loss": 0.0375, "step": 31220 }, { "epoch": 19.904397705544934, "grad_norm": 0.4194766581058502, "learning_rate": 9.920205122802494e-05, "loss": 0.0394, "step": 31230 }, { "epoch": 19.91077119184194, "grad_norm": 0.4703217148780823, "learning_rate": 9.920107019145978e-05, "loss": 0.058, "step": 31240 }, { "epoch": 19.917144678138943, "grad_norm": 0.6007516384124756, "learning_rate": 9.920008855705519e-05, "loss": 0.032, "step": 31250 }, { "epoch": 19.923518164435947, "grad_norm": 0.5609459280967712, "learning_rate": 9.919910632482303e-05, "loss": 0.0452, "step": 31260 }, { "epoch": 19.92989165073295, "grad_norm": 0.5445705652236938, "learning_rate": 9.919812349477531e-05, "loss": 0.049, "step": 31270 }, { "epoch": 19.936265137029956, "grad_norm": 0.6331278085708618, "learning_rate": 9.919714006692393e-05, "loss": 0.0316, "step": 31280 }, { "epoch": 19.94263862332696, "grad_norm": 0.8607646822929382, "learning_rate": 9.919615604128084e-05, "loss": 0.0602, "step": 31290 }, { "epoch": 19.949012109623965, "grad_norm": 0.7605261206626892, "learning_rate": 9.919517141785802e-05, "loss": 0.0471, "step": 31300 }, { "epoch": 19.95538559592097, "grad_norm": 0.5033801198005676, "learning_rate": 9.91941861966674e-05, "loss": 0.0615, "step": 31310 }, { "epoch": 19.961759082217974, "grad_norm": 1.1008477210998535, "learning_rate": 9.919320037772098e-05, "loss": 0.0626, "step": 31320 }, { "epoch": 19.968132568514978, "grad_norm": 0.6704133749008179, "learning_rate": 9.919221396103073e-05, "loss": 0.0372, "step": 31330 }, { "epoch": 19.974506054811982, "grad_norm": 0.43298137187957764, "learning_rate": 9.919122694660863e-05, "loss": 0.0319, "step": 31340 }, { "epoch": 19.980879541108987, "grad_norm": 0.7401928901672363, "learning_rate": 9.919023933446669e-05, "loss": 0.0363, "step": 31350 }, { "epoch": 19.98725302740599, "grad_norm": 0.8605602979660034, "learning_rate": 9.918925112461688e-05, "loss": 0.0396, "step": 31360 }, { "epoch": 19.993626513702996, "grad_norm": 0.8451436758041382, "learning_rate": 9.918826231707126e-05, "loss": 0.0551, "step": 31370 }, { "epoch": 20.0, "grad_norm": 0.7706111669540405, "learning_rate": 9.918727291184178e-05, "loss": 0.0472, "step": 31380 }, { "epoch": 20.006373486297004, "grad_norm": 0.4842163920402527, "learning_rate": 9.918628290894049e-05, "loss": 0.0331, "step": 31390 }, { "epoch": 20.01274697259401, "grad_norm": 0.588696300983429, "learning_rate": 9.918529230837945e-05, "loss": 0.039, "step": 31400 }, { "epoch": 20.019120458891013, "grad_norm": 0.6585944294929504, "learning_rate": 9.918430111017066e-05, "loss": 0.0373, "step": 31410 }, { "epoch": 20.025493945188018, "grad_norm": 0.5827052593231201, "learning_rate": 9.918330931432616e-05, "loss": 0.035, "step": 31420 }, { "epoch": 20.031867431485022, "grad_norm": 0.6124544739723206, "learning_rate": 9.918231692085803e-05, "loss": 0.0437, "step": 31430 }, { "epoch": 20.038240917782026, "grad_norm": 0.41697749495506287, "learning_rate": 9.918132392977832e-05, "loss": 0.036, "step": 31440 }, { "epoch": 20.04461440407903, "grad_norm": 0.7197396755218506, "learning_rate": 9.918033034109908e-05, "loss": 0.0472, "step": 31450 }, { "epoch": 20.050987890376035, "grad_norm": 0.47255319356918335, "learning_rate": 9.917933615483238e-05, "loss": 0.0327, "step": 31460 }, { "epoch": 20.05736137667304, "grad_norm": 0.7539225816726685, "learning_rate": 9.917834137099032e-05, "loss": 0.0362, "step": 31470 }, { "epoch": 20.063734862970044, "grad_norm": 0.395638108253479, "learning_rate": 9.9177345989585e-05, "loss": 0.0233, "step": 31480 }, { "epoch": 20.07010834926705, "grad_norm": 0.5284397006034851, "learning_rate": 9.917635001062847e-05, "loss": 0.0387, "step": 31490 }, { "epoch": 20.076481835564053, "grad_norm": 0.42776015400886536, "learning_rate": 9.917535343413287e-05, "loss": 0.0469, "step": 31500 }, { "epoch": 20.082855321861057, "grad_norm": 0.4647214412689209, "learning_rate": 9.917435626011029e-05, "loss": 0.0305, "step": 31510 }, { "epoch": 20.08922880815806, "grad_norm": 0.5136334896087646, "learning_rate": 9.917335848857286e-05, "loss": 0.0334, "step": 31520 }, { "epoch": 20.095602294455066, "grad_norm": 0.6021164655685425, "learning_rate": 9.917236011953268e-05, "loss": 0.0501, "step": 31530 }, { "epoch": 20.10197578075207, "grad_norm": 0.47220155596733093, "learning_rate": 9.917136115300192e-05, "loss": 0.0482, "step": 31540 }, { "epoch": 20.108349267049075, "grad_norm": 0.5563257336616516, "learning_rate": 9.917036158899269e-05, "loss": 0.0298, "step": 31550 }, { "epoch": 20.11472275334608, "grad_norm": 0.44852110743522644, "learning_rate": 9.916936142751713e-05, "loss": 0.0278, "step": 31560 }, { "epoch": 20.121096239643084, "grad_norm": 0.4002387523651123, "learning_rate": 9.916836066858743e-05, "loss": 0.0316, "step": 31570 }, { "epoch": 20.127469725940088, "grad_norm": 0.6808817386627197, "learning_rate": 9.91673593122157e-05, "loss": 0.0382, "step": 31580 }, { "epoch": 20.133843212237093, "grad_norm": 0.47998136281967163, "learning_rate": 9.916635735841415e-05, "loss": 0.028, "step": 31590 }, { "epoch": 20.140216698534097, "grad_norm": 0.7633264064788818, "learning_rate": 9.916535480719493e-05, "loss": 0.0331, "step": 31600 }, { "epoch": 20.1465901848311, "grad_norm": 0.6634160280227661, "learning_rate": 9.916435165857023e-05, "loss": 0.0339, "step": 31610 }, { "epoch": 20.152963671128106, "grad_norm": 0.48326966166496277, "learning_rate": 9.916334791255224e-05, "loss": 0.0488, "step": 31620 }, { "epoch": 20.15933715742511, "grad_norm": 0.39859503507614136, "learning_rate": 9.916234356915315e-05, "loss": 0.0455, "step": 31630 }, { "epoch": 20.165710643722115, "grad_norm": 0.6801707148551941, "learning_rate": 9.916133862838517e-05, "loss": 0.0389, "step": 31640 }, { "epoch": 20.17208413001912, "grad_norm": 0.617017388343811, "learning_rate": 9.916033309026051e-05, "loss": 0.0456, "step": 31650 }, { "epoch": 20.178457616316123, "grad_norm": 0.79550701379776, "learning_rate": 9.91593269547914e-05, "loss": 0.048, "step": 31660 }, { "epoch": 20.184831102613128, "grad_norm": 0.7285943031311035, "learning_rate": 9.915832022199003e-05, "loss": 0.029, "step": 31670 }, { "epoch": 20.191204588910132, "grad_norm": 0.49208498001098633, "learning_rate": 9.915731289186866e-05, "loss": 0.0355, "step": 31680 }, { "epoch": 20.197578075207137, "grad_norm": 0.5854390859603882, "learning_rate": 9.915630496443955e-05, "loss": 0.0369, "step": 31690 }, { "epoch": 20.203951561504145, "grad_norm": 0.7893449068069458, "learning_rate": 9.91552964397149e-05, "loss": 0.0406, "step": 31700 }, { "epoch": 20.21032504780115, "grad_norm": 0.5641257762908936, "learning_rate": 9.9154287317707e-05, "loss": 0.037, "step": 31710 }, { "epoch": 20.216698534098153, "grad_norm": 0.37174034118652344, "learning_rate": 9.91532775984281e-05, "loss": 0.0388, "step": 31720 }, { "epoch": 20.223072020395158, "grad_norm": 0.4521788954734802, "learning_rate": 9.915226728189047e-05, "loss": 0.047, "step": 31730 }, { "epoch": 20.229445506692162, "grad_norm": 0.43040764331817627, "learning_rate": 9.915125636810638e-05, "loss": 0.0293, "step": 31740 }, { "epoch": 20.235818992989167, "grad_norm": 0.5899751782417297, "learning_rate": 9.915024485708812e-05, "loss": 0.038, "step": 31750 }, { "epoch": 20.24219247928617, "grad_norm": 0.5041904449462891, "learning_rate": 9.914923274884798e-05, "loss": 0.042, "step": 31760 }, { "epoch": 20.248565965583175, "grad_norm": 0.3741162419319153, "learning_rate": 9.914822004339825e-05, "loss": 0.0455, "step": 31770 }, { "epoch": 20.25493945188018, "grad_norm": 0.5811709761619568, "learning_rate": 9.914720674075125e-05, "loss": 0.0395, "step": 31780 }, { "epoch": 20.261312938177184, "grad_norm": 0.5764346718788147, "learning_rate": 9.914619284091929e-05, "loss": 0.0522, "step": 31790 }, { "epoch": 20.26768642447419, "grad_norm": 0.9789597392082214, "learning_rate": 9.914517834391468e-05, "loss": 0.0552, "step": 31800 }, { "epoch": 20.274059910771193, "grad_norm": 0.58121657371521, "learning_rate": 9.914416324974975e-05, "loss": 0.0504, "step": 31810 }, { "epoch": 20.280433397068197, "grad_norm": 0.6952957510948181, "learning_rate": 9.914314755843683e-05, "loss": 0.0434, "step": 31820 }, { "epoch": 20.2868068833652, "grad_norm": 0.6010002493858337, "learning_rate": 9.914213126998828e-05, "loss": 0.0421, "step": 31830 }, { "epoch": 20.293180369662206, "grad_norm": 0.728855311870575, "learning_rate": 9.914111438441644e-05, "loss": 0.0627, "step": 31840 }, { "epoch": 20.29955385595921, "grad_norm": 0.623874843120575, "learning_rate": 9.914009690173364e-05, "loss": 0.0442, "step": 31850 }, { "epoch": 20.305927342256215, "grad_norm": 0.411680668592453, "learning_rate": 9.91390788219523e-05, "loss": 0.0457, "step": 31860 }, { "epoch": 20.31230082855322, "grad_norm": 0.48180603981018066, "learning_rate": 9.913806014508474e-05, "loss": 0.0356, "step": 31870 }, { "epoch": 20.318674314850224, "grad_norm": 0.4155711233615875, "learning_rate": 9.913704087114335e-05, "loss": 0.0393, "step": 31880 }, { "epoch": 20.325047801147228, "grad_norm": 0.8110759258270264, "learning_rate": 9.913602100014053e-05, "loss": 0.0428, "step": 31890 }, { "epoch": 20.331421287444233, "grad_norm": 0.6050277352333069, "learning_rate": 9.913500053208866e-05, "loss": 0.0563, "step": 31900 }, { "epoch": 20.337794773741237, "grad_norm": 0.3016187846660614, "learning_rate": 9.913397946700014e-05, "loss": 0.0332, "step": 31910 }, { "epoch": 20.34416826003824, "grad_norm": 0.3538776934146881, "learning_rate": 9.913295780488738e-05, "loss": 0.0499, "step": 31920 }, { "epoch": 20.350541746335246, "grad_norm": 0.5012523531913757, "learning_rate": 9.91319355457628e-05, "loss": 0.0589, "step": 31930 }, { "epoch": 20.35691523263225, "grad_norm": 0.39226800203323364, "learning_rate": 9.913091268963881e-05, "loss": 0.0291, "step": 31940 }, { "epoch": 20.363288718929255, "grad_norm": 0.4774281978607178, "learning_rate": 9.912988923652784e-05, "loss": 0.0358, "step": 31950 }, { "epoch": 20.36966220522626, "grad_norm": 0.6373028755187988, "learning_rate": 9.912886518644234e-05, "loss": 0.0442, "step": 31960 }, { "epoch": 20.376035691523263, "grad_norm": 0.44271302223205566, "learning_rate": 9.912784053939473e-05, "loss": 0.0403, "step": 31970 }, { "epoch": 20.382409177820268, "grad_norm": 0.47209858894348145, "learning_rate": 9.912681529539748e-05, "loss": 0.0339, "step": 31980 }, { "epoch": 20.388782664117272, "grad_norm": 0.4574304223060608, "learning_rate": 9.912578945446305e-05, "loss": 0.0347, "step": 31990 }, { "epoch": 20.395156150414277, "grad_norm": 0.6760044693946838, "learning_rate": 9.912476301660387e-05, "loss": 0.0495, "step": 32000 }, { "epoch": 20.40152963671128, "grad_norm": 0.4937826097011566, "learning_rate": 9.912373598183246e-05, "loss": 0.0375, "step": 32010 }, { "epoch": 20.407903123008285, "grad_norm": 0.4969538748264313, "learning_rate": 9.912270835016126e-05, "loss": 0.0375, "step": 32020 }, { "epoch": 20.41427660930529, "grad_norm": 0.7108178734779358, "learning_rate": 9.912168012160278e-05, "loss": 0.0442, "step": 32030 }, { "epoch": 20.420650095602294, "grad_norm": 0.3456435203552246, "learning_rate": 9.912065129616952e-05, "loss": 0.0299, "step": 32040 }, { "epoch": 20.4270235818993, "grad_norm": 0.5581284165382385, "learning_rate": 9.911962187387397e-05, "loss": 0.0461, "step": 32050 }, { "epoch": 20.433397068196303, "grad_norm": 0.40117985010147095, "learning_rate": 9.911859185472862e-05, "loss": 0.036, "step": 32060 }, { "epoch": 20.439770554493307, "grad_norm": 0.8666346669197083, "learning_rate": 9.911756123874602e-05, "loss": 0.0381, "step": 32070 }, { "epoch": 20.446144040790312, "grad_norm": 0.397077739238739, "learning_rate": 9.911653002593867e-05, "loss": 0.0427, "step": 32080 }, { "epoch": 20.452517527087316, "grad_norm": 0.9644503593444824, "learning_rate": 9.91154982163191e-05, "loss": 0.0338, "step": 32090 }, { "epoch": 20.45889101338432, "grad_norm": 0.339598149061203, "learning_rate": 9.911446580989988e-05, "loss": 0.0366, "step": 32100 }, { "epoch": 20.465264499681325, "grad_norm": 0.7253049612045288, "learning_rate": 9.911343280669349e-05, "loss": 0.035, "step": 32110 }, { "epoch": 20.47163798597833, "grad_norm": 0.7502745389938354, "learning_rate": 9.911239920671254e-05, "loss": 0.0399, "step": 32120 }, { "epoch": 20.478011472275334, "grad_norm": 0.4695591330528259, "learning_rate": 9.911136500996956e-05, "loss": 0.0286, "step": 32130 }, { "epoch": 20.48438495857234, "grad_norm": 0.5460295081138611, "learning_rate": 9.911033021647715e-05, "loss": 0.0384, "step": 32140 }, { "epoch": 20.490758444869343, "grad_norm": 0.8266238570213318, "learning_rate": 9.910929482624784e-05, "loss": 0.0304, "step": 32150 }, { "epoch": 20.497131931166347, "grad_norm": 0.4289080798625946, "learning_rate": 9.910825883929424e-05, "loss": 0.0269, "step": 32160 }, { "epoch": 20.50350541746335, "grad_norm": 0.42191484570503235, "learning_rate": 9.910722225562891e-05, "loss": 0.0337, "step": 32170 }, { "epoch": 20.509878903760356, "grad_norm": 0.3428705334663391, "learning_rate": 9.910618507526448e-05, "loss": 0.0394, "step": 32180 }, { "epoch": 20.51625239005736, "grad_norm": 0.44297927618026733, "learning_rate": 9.910514729821354e-05, "loss": 0.0272, "step": 32190 }, { "epoch": 20.522625876354365, "grad_norm": 0.7043351531028748, "learning_rate": 9.91041089244887e-05, "loss": 0.0286, "step": 32200 }, { "epoch": 20.52899936265137, "grad_norm": 0.5189981460571289, "learning_rate": 9.910306995410257e-05, "loss": 0.041, "step": 32210 }, { "epoch": 20.535372848948374, "grad_norm": 0.7327574491500854, "learning_rate": 9.910203038706778e-05, "loss": 0.0361, "step": 32220 }, { "epoch": 20.541746335245378, "grad_norm": 0.5154539346694946, "learning_rate": 9.910099022339696e-05, "loss": 0.0375, "step": 32230 }, { "epoch": 20.548119821542382, "grad_norm": 0.42485809326171875, "learning_rate": 9.909994946310275e-05, "loss": 0.0438, "step": 32240 }, { "epoch": 20.554493307839387, "grad_norm": 0.46833348274230957, "learning_rate": 9.909890810619779e-05, "loss": 0.0362, "step": 32250 }, { "epoch": 20.56086679413639, "grad_norm": 0.44683268666267395, "learning_rate": 9.909786615269474e-05, "loss": 0.0463, "step": 32260 }, { "epoch": 20.567240280433396, "grad_norm": 0.39845260977745056, "learning_rate": 9.909682360260625e-05, "loss": 0.0333, "step": 32270 }, { "epoch": 20.5736137667304, "grad_norm": 0.3318699896335602, "learning_rate": 9.909578045594503e-05, "loss": 0.0304, "step": 32280 }, { "epoch": 20.579987253027404, "grad_norm": 0.46723198890686035, "learning_rate": 9.909473671272371e-05, "loss": 0.0258, "step": 32290 }, { "epoch": 20.58636073932441, "grad_norm": 0.2995181977748871, "learning_rate": 9.909369237295498e-05, "loss": 0.0318, "step": 32300 }, { "epoch": 20.592734225621413, "grad_norm": 1.0364646911621094, "learning_rate": 9.909264743665152e-05, "loss": 0.0389, "step": 32310 }, { "epoch": 20.59910771191842, "grad_norm": 0.6058845520019531, "learning_rate": 9.909160190382607e-05, "loss": 0.0493, "step": 32320 }, { "epoch": 20.605481198215422, "grad_norm": 0.6111282706260681, "learning_rate": 9.90905557744913e-05, "loss": 0.0314, "step": 32330 }, { "epoch": 20.61185468451243, "grad_norm": 0.5166767239570618, "learning_rate": 9.908950904865993e-05, "loss": 0.0378, "step": 32340 }, { "epoch": 20.618228170809434, "grad_norm": 0.3108362555503845, "learning_rate": 9.908846172634467e-05, "loss": 0.0363, "step": 32350 }, { "epoch": 20.62460165710644, "grad_norm": 0.3442677855491638, "learning_rate": 9.908741380755826e-05, "loss": 0.0319, "step": 32360 }, { "epoch": 20.630975143403443, "grad_norm": 0.3573359549045563, "learning_rate": 9.908636529231341e-05, "loss": 0.0324, "step": 32370 }, { "epoch": 20.637348629700448, "grad_norm": 0.3774619698524475, "learning_rate": 9.90853161806229e-05, "loss": 0.0397, "step": 32380 }, { "epoch": 20.643722115997452, "grad_norm": 0.351256400346756, "learning_rate": 9.908426647249942e-05, "loss": 0.0327, "step": 32390 }, { "epoch": 20.650095602294456, "grad_norm": 0.5157501101493835, "learning_rate": 9.90832161679558e-05, "loss": 0.0303, "step": 32400 }, { "epoch": 20.65646908859146, "grad_norm": 0.543287456035614, "learning_rate": 9.908216526700474e-05, "loss": 0.037, "step": 32410 }, { "epoch": 20.662842574888465, "grad_norm": 0.47728314995765686, "learning_rate": 9.908111376965903e-05, "loss": 0.0366, "step": 32420 }, { "epoch": 20.66921606118547, "grad_norm": 0.6593474745750427, "learning_rate": 9.908006167593145e-05, "loss": 0.0418, "step": 32430 }, { "epoch": 20.675589547482474, "grad_norm": 0.2726775109767914, "learning_rate": 9.907900898583477e-05, "loss": 0.0408, "step": 32440 }, { "epoch": 20.68196303377948, "grad_norm": 0.5829458832740784, "learning_rate": 9.90779556993818e-05, "loss": 0.0328, "step": 32450 }, { "epoch": 20.688336520076483, "grad_norm": 0.6999732255935669, "learning_rate": 9.907690181658534e-05, "loss": 0.0383, "step": 32460 }, { "epoch": 20.694710006373487, "grad_norm": 0.47810623049736023, "learning_rate": 9.907584733745818e-05, "loss": 0.0327, "step": 32470 }, { "epoch": 20.70108349267049, "grad_norm": 0.6252797245979309, "learning_rate": 9.907479226201315e-05, "loss": 0.0522, "step": 32480 }, { "epoch": 20.707456978967496, "grad_norm": 0.47271057963371277, "learning_rate": 9.907373659026305e-05, "loss": 0.0301, "step": 32490 }, { "epoch": 20.7138304652645, "grad_norm": 0.3314191997051239, "learning_rate": 9.907268032222071e-05, "loss": 0.0419, "step": 32500 }, { "epoch": 20.720203951561505, "grad_norm": 0.8567216992378235, "learning_rate": 9.907162345789898e-05, "loss": 0.042, "step": 32510 }, { "epoch": 20.72657743785851, "grad_norm": 0.6678280234336853, "learning_rate": 9.907056599731069e-05, "loss": 0.0335, "step": 32520 }, { "epoch": 20.732950924155514, "grad_norm": 0.4286373257637024, "learning_rate": 9.90695079404687e-05, "loss": 0.034, "step": 32530 }, { "epoch": 20.739324410452518, "grad_norm": 0.340762734413147, "learning_rate": 9.906844928738585e-05, "loss": 0.03, "step": 32540 }, { "epoch": 20.745697896749522, "grad_norm": 0.36485153436660767, "learning_rate": 9.906739003807503e-05, "loss": 0.051, "step": 32550 }, { "epoch": 20.752071383046527, "grad_norm": 0.3059370517730713, "learning_rate": 9.906633019254907e-05, "loss": 0.0523, "step": 32560 }, { "epoch": 20.75844486934353, "grad_norm": 0.5903476476669312, "learning_rate": 9.906526975082088e-05, "loss": 0.0455, "step": 32570 }, { "epoch": 20.764818355640536, "grad_norm": 0.932595431804657, "learning_rate": 9.906420871290335e-05, "loss": 0.0448, "step": 32580 }, { "epoch": 20.77119184193754, "grad_norm": 0.6965762376785278, "learning_rate": 9.906314707880935e-05, "loss": 0.0416, "step": 32590 }, { "epoch": 20.777565328234544, "grad_norm": 0.44755029678344727, "learning_rate": 9.906208484855179e-05, "loss": 0.034, "step": 32600 }, { "epoch": 20.78393881453155, "grad_norm": 0.42917704582214355, "learning_rate": 9.906102202214358e-05, "loss": 0.0418, "step": 32610 }, { "epoch": 20.790312300828553, "grad_norm": 0.37115615606307983, "learning_rate": 9.905995859959762e-05, "loss": 0.0399, "step": 32620 }, { "epoch": 20.796685787125558, "grad_norm": 0.44704997539520264, "learning_rate": 9.905889458092686e-05, "loss": 0.0424, "step": 32630 }, { "epoch": 20.803059273422562, "grad_norm": 0.3460333049297333, "learning_rate": 9.90578299661442e-05, "loss": 0.0311, "step": 32640 }, { "epoch": 20.809432759719567, "grad_norm": 0.7044121026992798, "learning_rate": 9.90567647552626e-05, "loss": 0.0318, "step": 32650 }, { "epoch": 20.81580624601657, "grad_norm": 0.5258469581604004, "learning_rate": 9.905569894829497e-05, "loss": 0.0443, "step": 32660 }, { "epoch": 20.822179732313575, "grad_norm": 0.33766207098960876, "learning_rate": 9.905463254525428e-05, "loss": 0.0448, "step": 32670 }, { "epoch": 20.82855321861058, "grad_norm": 0.4253661334514618, "learning_rate": 9.905356554615353e-05, "loss": 0.0273, "step": 32680 }, { "epoch": 20.834926704907584, "grad_norm": 0.39972105622291565, "learning_rate": 9.905249795100561e-05, "loss": 0.0372, "step": 32690 }, { "epoch": 20.84130019120459, "grad_norm": 0.4974951148033142, "learning_rate": 9.905142975982354e-05, "loss": 0.0394, "step": 32700 }, { "epoch": 20.847673677501593, "grad_norm": 0.7446993589401245, "learning_rate": 9.905036097262027e-05, "loss": 0.0421, "step": 32710 }, { "epoch": 20.854047163798597, "grad_norm": 0.6133078932762146, "learning_rate": 9.904929158940881e-05, "loss": 0.0371, "step": 32720 }, { "epoch": 20.8604206500956, "grad_norm": 0.6737104058265686, "learning_rate": 9.904822161020216e-05, "loss": 0.0376, "step": 32730 }, { "epoch": 20.866794136392606, "grad_norm": 0.4237561523914337, "learning_rate": 9.904715103501329e-05, "loss": 0.0302, "step": 32740 }, { "epoch": 20.87316762268961, "grad_norm": 0.560462236404419, "learning_rate": 9.904607986385523e-05, "loss": 0.0329, "step": 32750 }, { "epoch": 20.879541108986615, "grad_norm": 0.49120181798934937, "learning_rate": 9.9045008096741e-05, "loss": 0.0415, "step": 32760 }, { "epoch": 20.88591459528362, "grad_norm": 0.5374075174331665, "learning_rate": 9.90439357336836e-05, "loss": 0.0356, "step": 32770 }, { "epoch": 20.892288081580624, "grad_norm": 0.29945650696754456, "learning_rate": 9.904286277469609e-05, "loss": 0.035, "step": 32780 }, { "epoch": 20.898661567877628, "grad_norm": 0.6709789633750916, "learning_rate": 9.90417892197915e-05, "loss": 0.0442, "step": 32790 }, { "epoch": 20.905035054174633, "grad_norm": 0.4916990399360657, "learning_rate": 9.904071506898285e-05, "loss": 0.0294, "step": 32800 }, { "epoch": 20.911408540471637, "grad_norm": 0.39094769954681396, "learning_rate": 9.903964032228323e-05, "loss": 0.0458, "step": 32810 }, { "epoch": 20.91778202676864, "grad_norm": 0.32712340354919434, "learning_rate": 9.903856497970568e-05, "loss": 0.0294, "step": 32820 }, { "epoch": 20.924155513065646, "grad_norm": 0.7401004433631897, "learning_rate": 9.903748904126325e-05, "loss": 0.0501, "step": 32830 }, { "epoch": 20.93052899936265, "grad_norm": 0.6848212480545044, "learning_rate": 9.903641250696902e-05, "loss": 0.049, "step": 32840 }, { "epoch": 20.936902485659655, "grad_norm": 0.9070122838020325, "learning_rate": 9.903533537683609e-05, "loss": 0.0443, "step": 32850 }, { "epoch": 20.94327597195666, "grad_norm": 0.6189029812812805, "learning_rate": 9.903425765087755e-05, "loss": 0.0377, "step": 32860 }, { "epoch": 20.949649458253663, "grad_norm": 0.4983058571815491, "learning_rate": 9.903317932910647e-05, "loss": 0.0499, "step": 32870 }, { "epoch": 20.956022944550668, "grad_norm": 0.5379558801651001, "learning_rate": 9.903210041153596e-05, "loss": 0.0448, "step": 32880 }, { "epoch": 20.962396430847672, "grad_norm": 0.29845455288887024, "learning_rate": 9.903102089817915e-05, "loss": 0.0388, "step": 32890 }, { "epoch": 20.968769917144677, "grad_norm": 0.38562801480293274, "learning_rate": 9.902994078904915e-05, "loss": 0.0292, "step": 32900 }, { "epoch": 20.97514340344168, "grad_norm": 0.2758936882019043, "learning_rate": 9.902886008415906e-05, "loss": 0.0482, "step": 32910 }, { "epoch": 20.981516889738685, "grad_norm": 0.573087215423584, "learning_rate": 9.902777878352203e-05, "loss": 0.0469, "step": 32920 }, { "epoch": 20.98789037603569, "grad_norm": 0.48148173093795776, "learning_rate": 9.90266968871512e-05, "loss": 0.0495, "step": 32930 }, { "epoch": 20.994263862332694, "grad_norm": 0.5030044913291931, "learning_rate": 9.902561439505973e-05, "loss": 0.0335, "step": 32940 }, { "epoch": 21.000637348629702, "grad_norm": 0.35676664113998413, "learning_rate": 9.902453130726075e-05, "loss": 0.029, "step": 32950 }, { "epoch": 21.007010834926707, "grad_norm": 0.7540311813354492, "learning_rate": 9.902344762376742e-05, "loss": 0.0499, "step": 32960 }, { "epoch": 21.01338432122371, "grad_norm": 0.5658154487609863, "learning_rate": 9.902236334459291e-05, "loss": 0.0375, "step": 32970 }, { "epoch": 21.019757807520715, "grad_norm": 0.43763744831085205, "learning_rate": 9.90212784697504e-05, "loss": 0.0295, "step": 32980 }, { "epoch": 21.02613129381772, "grad_norm": 0.6033055782318115, "learning_rate": 9.902019299925309e-05, "loss": 0.0415, "step": 32990 }, { "epoch": 21.032504780114724, "grad_norm": 0.6296896934509277, "learning_rate": 9.901910693311415e-05, "loss": 0.0297, "step": 33000 }, { "epoch": 21.03887826641173, "grad_norm": 0.6170443892478943, "learning_rate": 9.901802027134677e-05, "loss": 0.0404, "step": 33010 }, { "epoch": 21.045251752708733, "grad_norm": 0.5801374912261963, "learning_rate": 9.901693301396417e-05, "loss": 0.0359, "step": 33020 }, { "epoch": 21.051625239005737, "grad_norm": 0.6196602582931519, "learning_rate": 9.901584516097954e-05, "loss": 0.0357, "step": 33030 }, { "epoch": 21.057998725302742, "grad_norm": 0.4480137526988983, "learning_rate": 9.901475671240613e-05, "loss": 0.0291, "step": 33040 }, { "epoch": 21.064372211599746, "grad_norm": 0.5306084156036377, "learning_rate": 9.901366766825714e-05, "loss": 0.0334, "step": 33050 }, { "epoch": 21.07074569789675, "grad_norm": 0.525765597820282, "learning_rate": 9.901257802854582e-05, "loss": 0.0533, "step": 33060 }, { "epoch": 21.077119184193755, "grad_norm": 1.007337212562561, "learning_rate": 9.90114877932854e-05, "loss": 0.0443, "step": 33070 }, { "epoch": 21.08349267049076, "grad_norm": 0.7523496150970459, "learning_rate": 9.90103969624891e-05, "loss": 0.0296, "step": 33080 }, { "epoch": 21.089866156787764, "grad_norm": 0.37494054436683655, "learning_rate": 9.900930553617024e-05, "loss": 0.0588, "step": 33090 }, { "epoch": 21.09623964308477, "grad_norm": 0.6175310611724854, "learning_rate": 9.900821351434203e-05, "loss": 0.0337, "step": 33100 }, { "epoch": 21.102613129381773, "grad_norm": 2.121645450592041, "learning_rate": 9.900712089701776e-05, "loss": 0.0456, "step": 33110 }, { "epoch": 21.108986615678777, "grad_norm": 0.5083878040313721, "learning_rate": 9.90060276842107e-05, "loss": 0.0264, "step": 33120 }, { "epoch": 21.11536010197578, "grad_norm": 0.4423936903476715, "learning_rate": 9.900493387593412e-05, "loss": 0.0505, "step": 33130 }, { "epoch": 21.121733588272786, "grad_norm": 0.4551222622394562, "learning_rate": 9.900383947220133e-05, "loss": 0.0242, "step": 33140 }, { "epoch": 21.12810707456979, "grad_norm": 0.7340670228004456, "learning_rate": 9.900274447302564e-05, "loss": 0.0468, "step": 33150 }, { "epoch": 21.134480560866795, "grad_norm": 0.2980981469154358, "learning_rate": 9.900164887842032e-05, "loss": 0.0247, "step": 33160 }, { "epoch": 21.1408540471638, "grad_norm": 0.3659701645374298, "learning_rate": 9.900055268839872e-05, "loss": 0.0393, "step": 33170 }, { "epoch": 21.147227533460804, "grad_norm": 0.5428755283355713, "learning_rate": 9.899945590297412e-05, "loss": 0.0429, "step": 33180 }, { "epoch": 21.153601019757808, "grad_norm": 0.6634163856506348, "learning_rate": 9.899835852215988e-05, "loss": 0.0414, "step": 33190 }, { "epoch": 21.159974506054812, "grad_norm": 0.5601595044136047, "learning_rate": 9.899726054596932e-05, "loss": 0.0277, "step": 33200 }, { "epoch": 21.166347992351817, "grad_norm": 0.3913784623146057, "learning_rate": 9.899616197441579e-05, "loss": 0.0315, "step": 33210 }, { "epoch": 21.17272147864882, "grad_norm": 0.42655080556869507, "learning_rate": 9.89950628075126e-05, "loss": 0.035, "step": 33220 }, { "epoch": 21.179094964945826, "grad_norm": 0.5679159164428711, "learning_rate": 9.899396304527317e-05, "loss": 0.04, "step": 33230 }, { "epoch": 21.18546845124283, "grad_norm": 0.2692587375640869, "learning_rate": 9.899286268771084e-05, "loss": 0.0461, "step": 33240 }, { "epoch": 21.191841937539834, "grad_norm": 0.21358618140220642, "learning_rate": 9.899176173483894e-05, "loss": 0.0444, "step": 33250 }, { "epoch": 21.19821542383684, "grad_norm": 0.3576653301715851, "learning_rate": 9.899066018667089e-05, "loss": 0.042, "step": 33260 }, { "epoch": 21.204588910133843, "grad_norm": 0.3487298786640167, "learning_rate": 9.898955804322007e-05, "loss": 0.0387, "step": 33270 }, { "epoch": 21.210962396430848, "grad_norm": 0.5706133842468262, "learning_rate": 9.898845530449987e-05, "loss": 0.0414, "step": 33280 }, { "epoch": 21.217335882727852, "grad_norm": 0.460977703332901, "learning_rate": 9.898735197052367e-05, "loss": 0.0322, "step": 33290 }, { "epoch": 21.223709369024856, "grad_norm": 0.5471218228340149, "learning_rate": 9.898624804130489e-05, "loss": 0.0323, "step": 33300 }, { "epoch": 21.23008285532186, "grad_norm": 0.37643781304359436, "learning_rate": 9.898514351685697e-05, "loss": 0.0381, "step": 33310 }, { "epoch": 21.236456341618865, "grad_norm": 0.35387441515922546, "learning_rate": 9.898403839719329e-05, "loss": 0.047, "step": 33320 }, { "epoch": 21.24282982791587, "grad_norm": 0.5888198614120483, "learning_rate": 9.898293268232731e-05, "loss": 0.0359, "step": 33330 }, { "epoch": 21.249203314212874, "grad_norm": 0.48150134086608887, "learning_rate": 9.898182637227243e-05, "loss": 0.0403, "step": 33340 }, { "epoch": 21.25557680050988, "grad_norm": 0.22532965242862701, "learning_rate": 9.898071946704211e-05, "loss": 0.0346, "step": 33350 }, { "epoch": 21.261950286806883, "grad_norm": 0.34886297583580017, "learning_rate": 9.897961196664982e-05, "loss": 0.0377, "step": 33360 }, { "epoch": 21.268323773103887, "grad_norm": 0.325795978307724, "learning_rate": 9.8978503871109e-05, "loss": 0.0339, "step": 33370 }, { "epoch": 21.27469725940089, "grad_norm": 0.5327252745628357, "learning_rate": 9.89773951804331e-05, "loss": 0.0427, "step": 33380 }, { "epoch": 21.281070745697896, "grad_norm": 0.685027003288269, "learning_rate": 9.897628589463561e-05, "loss": 0.0307, "step": 33390 }, { "epoch": 21.2874442319949, "grad_norm": 0.37561291456222534, "learning_rate": 9.897517601373002e-05, "loss": 0.0379, "step": 33400 }, { "epoch": 21.293817718291905, "grad_norm": 0.34865817427635193, "learning_rate": 9.897406553772979e-05, "loss": 0.0214, "step": 33410 }, { "epoch": 21.30019120458891, "grad_norm": 0.4685518145561218, "learning_rate": 9.897295446664843e-05, "loss": 0.0344, "step": 33420 }, { "epoch": 21.306564690885914, "grad_norm": 0.7396595478057861, "learning_rate": 9.897184280049943e-05, "loss": 0.039, "step": 33430 }, { "epoch": 21.312938177182918, "grad_norm": 0.6175336837768555, "learning_rate": 9.89707305392963e-05, "loss": 0.0401, "step": 33440 }, { "epoch": 21.319311663479922, "grad_norm": 0.4183218479156494, "learning_rate": 9.896961768305255e-05, "loss": 0.032, "step": 33450 }, { "epoch": 21.325685149776927, "grad_norm": 0.5580785274505615, "learning_rate": 9.896850423178173e-05, "loss": 0.0272, "step": 33460 }, { "epoch": 21.33205863607393, "grad_norm": 0.2957943379878998, "learning_rate": 9.896739018549735e-05, "loss": 0.0275, "step": 33470 }, { "epoch": 21.338432122370936, "grad_norm": 0.48510852456092834, "learning_rate": 9.896627554421292e-05, "loss": 0.036, "step": 33480 }, { "epoch": 21.34480560866794, "grad_norm": 0.42194828391075134, "learning_rate": 9.896516030794204e-05, "loss": 0.0284, "step": 33490 }, { "epoch": 21.351179094964944, "grad_norm": 0.7790687680244446, "learning_rate": 9.896404447669823e-05, "loss": 0.0304, "step": 33500 }, { "epoch": 21.35755258126195, "grad_norm": 0.7340450286865234, "learning_rate": 9.896292805049503e-05, "loss": 0.0435, "step": 33510 }, { "epoch": 21.363926067558953, "grad_norm": 0.4435712695121765, "learning_rate": 9.896181102934603e-05, "loss": 0.0492, "step": 33520 }, { "epoch": 21.370299553855958, "grad_norm": 0.7536898255348206, "learning_rate": 9.896069341326481e-05, "loss": 0.031, "step": 33530 }, { "epoch": 21.376673040152962, "grad_norm": 0.5052396059036255, "learning_rate": 9.895957520226493e-05, "loss": 0.043, "step": 33540 }, { "epoch": 21.383046526449967, "grad_norm": 0.3777838349342346, "learning_rate": 9.895845639635999e-05, "loss": 0.0321, "step": 33550 }, { "epoch": 21.38942001274697, "grad_norm": 0.3997042775154114, "learning_rate": 9.895733699556357e-05, "loss": 0.0373, "step": 33560 }, { "epoch": 21.395793499043975, "grad_norm": 0.2579830288887024, "learning_rate": 9.895621699988928e-05, "loss": 0.0299, "step": 33570 }, { "epoch": 21.402166985340983, "grad_norm": 0.42862001061439514, "learning_rate": 9.895509640935075e-05, "loss": 0.0511, "step": 33580 }, { "epoch": 21.408540471637988, "grad_norm": 0.5525429844856262, "learning_rate": 9.895397522396158e-05, "loss": 0.0246, "step": 33590 }, { "epoch": 21.414913957934992, "grad_norm": 0.5243450403213501, "learning_rate": 9.895285344373537e-05, "loss": 0.031, "step": 33600 }, { "epoch": 21.421287444231996, "grad_norm": 0.7823039293289185, "learning_rate": 9.895173106868578e-05, "loss": 0.0271, "step": 33610 }, { "epoch": 21.427660930529, "grad_norm": 0.6277959942817688, "learning_rate": 9.895060809882644e-05, "loss": 0.0376, "step": 33620 }, { "epoch": 21.434034416826005, "grad_norm": 0.37120676040649414, "learning_rate": 9.894948453417098e-05, "loss": 0.0433, "step": 33630 }, { "epoch": 21.44040790312301, "grad_norm": 0.2807099521160126, "learning_rate": 9.89483603747331e-05, "loss": 0.0321, "step": 33640 }, { "epoch": 21.446781389420014, "grad_norm": 0.5695929527282715, "learning_rate": 9.894723562052641e-05, "loss": 0.0386, "step": 33650 }, { "epoch": 21.45315487571702, "grad_norm": 0.4366040527820587, "learning_rate": 9.894611027156458e-05, "loss": 0.0308, "step": 33660 }, { "epoch": 21.459528362014023, "grad_norm": 0.46178221702575684, "learning_rate": 9.894498432786132e-05, "loss": 0.0291, "step": 33670 }, { "epoch": 21.465901848311027, "grad_norm": 0.625882625579834, "learning_rate": 9.894385778943027e-05, "loss": 0.0432, "step": 33680 }, { "epoch": 21.47227533460803, "grad_norm": 0.529136061668396, "learning_rate": 9.894273065628514e-05, "loss": 0.0306, "step": 33690 }, { "epoch": 21.478648820905036, "grad_norm": 0.49732890725135803, "learning_rate": 9.894160292843964e-05, "loss": 0.034, "step": 33700 }, { "epoch": 21.48502230720204, "grad_norm": 0.4490704834461212, "learning_rate": 9.894047460590744e-05, "loss": 0.0317, "step": 33710 }, { "epoch": 21.491395793499045, "grad_norm": 0.5355561375617981, "learning_rate": 9.893934568870227e-05, "loss": 0.0284, "step": 33720 }, { "epoch": 21.49776927979605, "grad_norm": 0.5296640396118164, "learning_rate": 9.893821617683784e-05, "loss": 0.041, "step": 33730 }, { "epoch": 21.504142766093054, "grad_norm": 0.6401536464691162, "learning_rate": 9.893708607032787e-05, "loss": 0.0312, "step": 33740 }, { "epoch": 21.510516252390058, "grad_norm": 0.7757630348205566, "learning_rate": 9.893595536918612e-05, "loss": 0.0325, "step": 33750 }, { "epoch": 21.516889738687063, "grad_norm": 0.36943888664245605, "learning_rate": 9.893482407342632e-05, "loss": 0.0258, "step": 33760 }, { "epoch": 21.523263224984067, "grad_norm": 0.6172603368759155, "learning_rate": 9.893369218306217e-05, "loss": 0.0438, "step": 33770 }, { "epoch": 21.52963671128107, "grad_norm": 0.4445973336696625, "learning_rate": 9.893255969810748e-05, "loss": 0.0307, "step": 33780 }, { "epoch": 21.536010197578076, "grad_norm": 0.45456793904304504, "learning_rate": 9.893142661857597e-05, "loss": 0.0358, "step": 33790 }, { "epoch": 21.54238368387508, "grad_norm": 0.4152179956436157, "learning_rate": 9.893029294448145e-05, "loss": 0.0283, "step": 33800 }, { "epoch": 21.548757170172085, "grad_norm": 0.3421584665775299, "learning_rate": 9.892915867583767e-05, "loss": 0.0367, "step": 33810 }, { "epoch": 21.55513065646909, "grad_norm": 0.346899151802063, "learning_rate": 9.89280238126584e-05, "loss": 0.0234, "step": 33820 }, { "epoch": 21.561504142766093, "grad_norm": 0.5675371885299683, "learning_rate": 9.892688835495747e-05, "loss": 0.0314, "step": 33830 }, { "epoch": 21.567877629063098, "grad_norm": 0.24062864482402802, "learning_rate": 9.892575230274863e-05, "loss": 0.0251, "step": 33840 }, { "epoch": 21.574251115360102, "grad_norm": 0.44688624143600464, "learning_rate": 9.892461565604572e-05, "loss": 0.0371, "step": 33850 }, { "epoch": 21.580624601657107, "grad_norm": 0.6995373368263245, "learning_rate": 9.892347841486252e-05, "loss": 0.0356, "step": 33860 }, { "epoch": 21.58699808795411, "grad_norm": 0.44560351967811584, "learning_rate": 9.892234057921288e-05, "loss": 0.0388, "step": 33870 }, { "epoch": 21.593371574251115, "grad_norm": 0.46044862270355225, "learning_rate": 9.89212021491106e-05, "loss": 0.0315, "step": 33880 }, { "epoch": 21.59974506054812, "grad_norm": 0.4010196626186371, "learning_rate": 9.892006312456954e-05, "loss": 0.0324, "step": 33890 }, { "epoch": 21.606118546845124, "grad_norm": 0.5219210982322693, "learning_rate": 9.891892350560352e-05, "loss": 0.0386, "step": 33900 }, { "epoch": 21.61249203314213, "grad_norm": 0.6276299357414246, "learning_rate": 9.891778329222638e-05, "loss": 0.0347, "step": 33910 }, { "epoch": 21.618865519439133, "grad_norm": 0.4902973473072052, "learning_rate": 9.8916642484452e-05, "loss": 0.0485, "step": 33920 }, { "epoch": 21.625239005736137, "grad_norm": 0.3681551516056061, "learning_rate": 9.891550108229423e-05, "loss": 0.0476, "step": 33930 }, { "epoch": 21.631612492033142, "grad_norm": 0.39696118235588074, "learning_rate": 9.891435908576692e-05, "loss": 0.0317, "step": 33940 }, { "epoch": 21.637985978330146, "grad_norm": 0.8397636413574219, "learning_rate": 9.891321649488398e-05, "loss": 0.0381, "step": 33950 }, { "epoch": 21.64435946462715, "grad_norm": 0.6202315092086792, "learning_rate": 9.891207330965928e-05, "loss": 0.0527, "step": 33960 }, { "epoch": 21.650732950924155, "grad_norm": 0.8681958317756653, "learning_rate": 9.89109295301067e-05, "loss": 0.0276, "step": 33970 }, { "epoch": 21.65710643722116, "grad_norm": 0.5664786696434021, "learning_rate": 9.890978515624017e-05, "loss": 0.0279, "step": 33980 }, { "epoch": 21.663479923518164, "grad_norm": 0.5184953808784485, "learning_rate": 9.890864018807354e-05, "loss": 0.0389, "step": 33990 }, { "epoch": 21.66985340981517, "grad_norm": 0.5251326560974121, "learning_rate": 9.890749462562077e-05, "loss": 0.0313, "step": 34000 }, { "epoch": 21.676226896112173, "grad_norm": 0.5055752396583557, "learning_rate": 9.890634846889577e-05, "loss": 0.0436, "step": 34010 }, { "epoch": 21.682600382409177, "grad_norm": 0.3840938210487366, "learning_rate": 9.890520171791245e-05, "loss": 0.0325, "step": 34020 }, { "epoch": 21.68897386870618, "grad_norm": 0.5885211229324341, "learning_rate": 9.890405437268475e-05, "loss": 0.0356, "step": 34030 }, { "epoch": 21.695347355003186, "grad_norm": 0.5757576823234558, "learning_rate": 9.890290643322661e-05, "loss": 0.0336, "step": 34040 }, { "epoch": 21.70172084130019, "grad_norm": 0.7059485912322998, "learning_rate": 9.8901757899552e-05, "loss": 0.0392, "step": 34050 }, { "epoch": 21.708094327597195, "grad_norm": 0.3947299122810364, "learning_rate": 9.890060877167487e-05, "loss": 0.0351, "step": 34060 }, { "epoch": 21.7144678138942, "grad_norm": 0.6324647068977356, "learning_rate": 9.889945904960916e-05, "loss": 0.0321, "step": 34070 }, { "epoch": 21.720841300191204, "grad_norm": 0.375550776720047, "learning_rate": 9.889830873336885e-05, "loss": 0.0358, "step": 34080 }, { "epoch": 21.727214786488208, "grad_norm": 0.7949948906898499, "learning_rate": 9.889715782296794e-05, "loss": 0.036, "step": 34090 }, { "epoch": 21.733588272785212, "grad_norm": 0.3773329555988312, "learning_rate": 9.889600631842038e-05, "loss": 0.0325, "step": 34100 }, { "epoch": 21.739961759082217, "grad_norm": 0.4015154242515564, "learning_rate": 9.889485421974017e-05, "loss": 0.034, "step": 34110 }, { "epoch": 21.74633524537922, "grad_norm": 0.7220259308815002, "learning_rate": 9.889370152694133e-05, "loss": 0.0344, "step": 34120 }, { "epoch": 21.752708731676226, "grad_norm": 0.5984421968460083, "learning_rate": 9.889254824003785e-05, "loss": 0.0274, "step": 34130 }, { "epoch": 21.75908221797323, "grad_norm": 0.4008987545967102, "learning_rate": 9.889139435904375e-05, "loss": 0.0317, "step": 34140 }, { "epoch": 21.765455704270234, "grad_norm": 0.4927680194377899, "learning_rate": 9.889023988397303e-05, "loss": 0.0249, "step": 34150 }, { "epoch": 21.77182919056724, "grad_norm": 0.41582080721855164, "learning_rate": 9.888908481483976e-05, "loss": 0.0331, "step": 34160 }, { "epoch": 21.778202676864243, "grad_norm": 0.7995622754096985, "learning_rate": 9.888792915165795e-05, "loss": 0.0376, "step": 34170 }, { "epoch": 21.784576163161248, "grad_norm": 0.45375895500183105, "learning_rate": 9.888677289444163e-05, "loss": 0.0367, "step": 34180 }, { "epoch": 21.790949649458252, "grad_norm": 0.5788255929946899, "learning_rate": 9.888561604320486e-05, "loss": 0.0325, "step": 34190 }, { "epoch": 21.797323135755256, "grad_norm": 0.8217025995254517, "learning_rate": 9.888445859796171e-05, "loss": 0.0348, "step": 34200 }, { "epoch": 21.803696622052264, "grad_norm": 0.553433895111084, "learning_rate": 9.888330055872623e-05, "loss": 0.0381, "step": 34210 }, { "epoch": 21.81007010834927, "grad_norm": 0.35488733649253845, "learning_rate": 9.888214192551249e-05, "loss": 0.0413, "step": 34220 }, { "epoch": 21.816443594646273, "grad_norm": 0.37866947054862976, "learning_rate": 9.888098269833457e-05, "loss": 0.0305, "step": 34230 }, { "epoch": 21.822817080943278, "grad_norm": 0.600410521030426, "learning_rate": 9.887982287720657e-05, "loss": 0.0468, "step": 34240 }, { "epoch": 21.829190567240282, "grad_norm": 0.5504175424575806, "learning_rate": 9.887866246214257e-05, "loss": 0.03, "step": 34250 }, { "epoch": 21.835564053537286, "grad_norm": 0.8275286555290222, "learning_rate": 9.887750145315668e-05, "loss": 0.039, "step": 34260 }, { "epoch": 21.84193753983429, "grad_norm": 0.698927640914917, "learning_rate": 9.887633985026298e-05, "loss": 0.0356, "step": 34270 }, { "epoch": 21.848311026131295, "grad_norm": 0.3693823218345642, "learning_rate": 9.887517765347562e-05, "loss": 0.0462, "step": 34280 }, { "epoch": 21.8546845124283, "grad_norm": 0.345838725566864, "learning_rate": 9.88740148628087e-05, "loss": 0.0375, "step": 34290 }, { "epoch": 21.861057998725304, "grad_norm": 0.9541895389556885, "learning_rate": 9.887285147827634e-05, "loss": 0.036, "step": 34300 }, { "epoch": 21.86743148502231, "grad_norm": 0.5108804106712341, "learning_rate": 9.88716874998927e-05, "loss": 0.0353, "step": 34310 }, { "epoch": 21.873804971319313, "grad_norm": 0.7709117531776428, "learning_rate": 9.887052292767193e-05, "loss": 0.0418, "step": 34320 }, { "epoch": 21.880178457616317, "grad_norm": 0.5951215028762817, "learning_rate": 9.886935776162816e-05, "loss": 0.0341, "step": 34330 }, { "epoch": 21.88655194391332, "grad_norm": 0.6434561610221863, "learning_rate": 9.886819200177554e-05, "loss": 0.0388, "step": 34340 }, { "epoch": 21.892925430210326, "grad_norm": 0.7685309052467346, "learning_rate": 9.886702564812825e-05, "loss": 0.033, "step": 34350 }, { "epoch": 21.89929891650733, "grad_norm": 0.3714256286621094, "learning_rate": 9.886585870070047e-05, "loss": 0.0416, "step": 34360 }, { "epoch": 21.905672402804335, "grad_norm": 0.5191555619239807, "learning_rate": 9.886469115950636e-05, "loss": 0.0351, "step": 34370 }, { "epoch": 21.91204588910134, "grad_norm": 0.5067389011383057, "learning_rate": 9.886352302456012e-05, "loss": 0.0379, "step": 34380 }, { "epoch": 21.918419375398344, "grad_norm": 0.5698462128639221, "learning_rate": 9.886235429587594e-05, "loss": 0.0541, "step": 34390 }, { "epoch": 21.924792861695348, "grad_norm": 0.42807474732398987, "learning_rate": 9.886118497346803e-05, "loss": 0.0434, "step": 34400 }, { "epoch": 21.931166347992352, "grad_norm": 0.733651340007782, "learning_rate": 9.886001505735059e-05, "loss": 0.0405, "step": 34410 }, { "epoch": 21.937539834289357, "grad_norm": 0.6504823565483093, "learning_rate": 9.885884454753781e-05, "loss": 0.037, "step": 34420 }, { "epoch": 21.94391332058636, "grad_norm": 0.6256840229034424, "learning_rate": 9.885767344404396e-05, "loss": 0.0433, "step": 34430 }, { "epoch": 21.950286806883366, "grad_norm": 0.3689855635166168, "learning_rate": 9.885650174688326e-05, "loss": 0.038, "step": 34440 }, { "epoch": 21.95666029318037, "grad_norm": 0.7795689702033997, "learning_rate": 9.885532945606992e-05, "loss": 0.0443, "step": 34450 }, { "epoch": 21.963033779477374, "grad_norm": 0.40495598316192627, "learning_rate": 9.885415657161818e-05, "loss": 0.0447, "step": 34460 }, { "epoch": 21.96940726577438, "grad_norm": 0.35698962211608887, "learning_rate": 9.885298309354234e-05, "loss": 0.0416, "step": 34470 }, { "epoch": 21.975780752071383, "grad_norm": 0.435452938079834, "learning_rate": 9.885180902185662e-05, "loss": 0.0406, "step": 34480 }, { "epoch": 21.982154238368388, "grad_norm": 0.4199587106704712, "learning_rate": 9.885063435657532e-05, "loss": 0.0259, "step": 34490 }, { "epoch": 21.988527724665392, "grad_norm": 0.608440637588501, "learning_rate": 9.884945909771266e-05, "loss": 0.0356, "step": 34500 }, { "epoch": 21.994901210962396, "grad_norm": 0.48252299427986145, "learning_rate": 9.884828324528295e-05, "loss": 0.0238, "step": 34510 }, { "epoch": 22.0012746972594, "grad_norm": 0.4837629795074463, "learning_rate": 9.88471067993005e-05, "loss": 0.032, "step": 34520 }, { "epoch": 22.007648183556405, "grad_norm": 0.6715797185897827, "learning_rate": 9.884592975977956e-05, "loss": 0.0514, "step": 34530 }, { "epoch": 22.01402166985341, "grad_norm": 1.0237489938735962, "learning_rate": 9.884475212673447e-05, "loss": 0.046, "step": 34540 }, { "epoch": 22.020395156150414, "grad_norm": 0.43757814168930054, "learning_rate": 9.884357390017954e-05, "loss": 0.0245, "step": 34550 }, { "epoch": 22.02676864244742, "grad_norm": 0.4704092741012573, "learning_rate": 9.884239508012904e-05, "loss": 0.0383, "step": 34560 }, { "epoch": 22.033142128744423, "grad_norm": 0.5576807856559753, "learning_rate": 9.884121566659733e-05, "loss": 0.0264, "step": 34570 }, { "epoch": 22.039515615041427, "grad_norm": 0.6145924925804138, "learning_rate": 9.884003565959874e-05, "loss": 0.0451, "step": 34580 }, { "epoch": 22.04588910133843, "grad_norm": 0.4765033721923828, "learning_rate": 9.883885505914761e-05, "loss": 0.0377, "step": 34590 }, { "epoch": 22.052262587635436, "grad_norm": 0.5439814925193787, "learning_rate": 9.883767386525829e-05, "loss": 0.0435, "step": 34600 }, { "epoch": 22.05863607393244, "grad_norm": 0.64704430103302, "learning_rate": 9.88364920779451e-05, "loss": 0.0341, "step": 34610 }, { "epoch": 22.065009560229445, "grad_norm": 0.442614883184433, "learning_rate": 9.883530969722243e-05, "loss": 0.0411, "step": 34620 }, { "epoch": 22.07138304652645, "grad_norm": 0.40172383189201355, "learning_rate": 9.883412672310463e-05, "loss": 0.0296, "step": 34630 }, { "epoch": 22.077756532823454, "grad_norm": 0.3768358528614044, "learning_rate": 9.88329431556061e-05, "loss": 0.033, "step": 34640 }, { "epoch": 22.084130019120458, "grad_norm": 0.3362719714641571, "learning_rate": 9.883175899474119e-05, "loss": 0.038, "step": 34650 }, { "epoch": 22.090503505417463, "grad_norm": 0.24823905527591705, "learning_rate": 9.883057424052433e-05, "loss": 0.0344, "step": 34660 }, { "epoch": 22.096876991714467, "grad_norm": 0.5325015187263489, "learning_rate": 9.882938889296987e-05, "loss": 0.0244, "step": 34670 }, { "epoch": 22.10325047801147, "grad_norm": 0.6017010807991028, "learning_rate": 9.882820295209223e-05, "loss": 0.0315, "step": 34680 }, { "epoch": 22.109623964308476, "grad_norm": 0.7105106115341187, "learning_rate": 9.882701641790582e-05, "loss": 0.0367, "step": 34690 }, { "epoch": 22.11599745060548, "grad_norm": 0.5900437831878662, "learning_rate": 9.882582929042507e-05, "loss": 0.0353, "step": 34700 }, { "epoch": 22.122370936902485, "grad_norm": 0.697341799736023, "learning_rate": 9.88246415696644e-05, "loss": 0.0382, "step": 34710 }, { "epoch": 22.12874442319949, "grad_norm": 0.40661507844924927, "learning_rate": 9.882345325563822e-05, "loss": 0.0333, "step": 34720 }, { "epoch": 22.135117909496493, "grad_norm": 0.517068088054657, "learning_rate": 9.882226434836099e-05, "loss": 0.0466, "step": 34730 }, { "epoch": 22.141491395793498, "grad_norm": 0.4185454547405243, "learning_rate": 9.882107484784715e-05, "loss": 0.024, "step": 34740 }, { "epoch": 22.147864882090502, "grad_norm": 0.6745854020118713, "learning_rate": 9.881988475411116e-05, "loss": 0.0385, "step": 34750 }, { "epoch": 22.154238368387507, "grad_norm": 0.567823052406311, "learning_rate": 9.881869406716748e-05, "loss": 0.032, "step": 34760 }, { "epoch": 22.16061185468451, "grad_norm": 0.5652419328689575, "learning_rate": 9.881750278703057e-05, "loss": 0.0301, "step": 34770 }, { "epoch": 22.166985340981515, "grad_norm": 0.461007684469223, "learning_rate": 9.881631091371491e-05, "loss": 0.0234, "step": 34780 }, { "epoch": 22.17335882727852, "grad_norm": 0.3966669738292694, "learning_rate": 9.881511844723499e-05, "loss": 0.041, "step": 34790 }, { "epoch": 22.179732313575524, "grad_norm": 0.29091978073120117, "learning_rate": 9.881392538760529e-05, "loss": 0.0264, "step": 34800 }, { "epoch": 22.18610579987253, "grad_norm": 0.6113274097442627, "learning_rate": 9.88127317348403e-05, "loss": 0.0309, "step": 34810 }, { "epoch": 22.192479286169533, "grad_norm": 0.5449618697166443, "learning_rate": 9.881153748895455e-05, "loss": 0.0281, "step": 34820 }, { "epoch": 22.198852772466537, "grad_norm": 0.3119959235191345, "learning_rate": 9.881034264996252e-05, "loss": 0.0303, "step": 34830 }, { "epoch": 22.205226258763545, "grad_norm": 0.4033069312572479, "learning_rate": 9.880914721787872e-05, "loss": 0.0264, "step": 34840 }, { "epoch": 22.21159974506055, "grad_norm": 0.42602333426475525, "learning_rate": 9.880795119271773e-05, "loss": 0.029, "step": 34850 }, { "epoch": 22.217973231357554, "grad_norm": 0.625603973865509, "learning_rate": 9.880675457449404e-05, "loss": 0.0366, "step": 34860 }, { "epoch": 22.22434671765456, "grad_norm": 0.5646366477012634, "learning_rate": 9.880555736322221e-05, "loss": 0.0379, "step": 34870 }, { "epoch": 22.230720203951563, "grad_norm": 0.34572529792785645, "learning_rate": 9.880435955891676e-05, "loss": 0.0334, "step": 34880 }, { "epoch": 22.237093690248567, "grad_norm": 0.516228199005127, "learning_rate": 9.880316116159229e-05, "loss": 0.0368, "step": 34890 }, { "epoch": 22.243467176545572, "grad_norm": 0.4655141532421112, "learning_rate": 9.880196217126332e-05, "loss": 0.0504, "step": 34900 }, { "epoch": 22.249840662842576, "grad_norm": 0.5334170460700989, "learning_rate": 9.880076258794441e-05, "loss": 0.0269, "step": 34910 }, { "epoch": 22.25621414913958, "grad_norm": 0.493969202041626, "learning_rate": 9.87995624116502e-05, "loss": 0.0454, "step": 34920 }, { "epoch": 22.262587635436585, "grad_norm": 0.4887004494667053, "learning_rate": 9.87983616423952e-05, "loss": 0.0346, "step": 34930 }, { "epoch": 22.26896112173359, "grad_norm": 0.4698292315006256, "learning_rate": 9.879716028019405e-05, "loss": 0.0317, "step": 34940 }, { "epoch": 22.275334608030594, "grad_norm": 0.7241379022598267, "learning_rate": 9.879595832506132e-05, "loss": 0.0386, "step": 34950 }, { "epoch": 22.2817080943276, "grad_norm": 0.6733170747756958, "learning_rate": 9.879475577701163e-05, "loss": 0.0488, "step": 34960 }, { "epoch": 22.288081580624603, "grad_norm": 0.7493279576301575, "learning_rate": 9.879355263605958e-05, "loss": 0.0445, "step": 34970 }, { "epoch": 22.294455066921607, "grad_norm": 0.5190390944480896, "learning_rate": 9.879234890221982e-05, "loss": 0.0466, "step": 34980 }, { "epoch": 22.30082855321861, "grad_norm": 0.4218645989894867, "learning_rate": 9.879114457550693e-05, "loss": 0.0343, "step": 34990 }, { "epoch": 22.307202039515616, "grad_norm": 0.5315443873405457, "learning_rate": 9.878993965593557e-05, "loss": 0.0309, "step": 35000 }, { "epoch": 22.31357552581262, "grad_norm": 0.5222482085227966, "learning_rate": 9.878873414352039e-05, "loss": 0.0404, "step": 35010 }, { "epoch": 22.319949012109625, "grad_norm": 0.3544315695762634, "learning_rate": 9.878752803827602e-05, "loss": 0.0448, "step": 35020 }, { "epoch": 22.32632249840663, "grad_norm": 0.7255458235740662, "learning_rate": 9.878632134021712e-05, "loss": 0.0383, "step": 35030 }, { "epoch": 22.332695984703633, "grad_norm": 0.5964975953102112, "learning_rate": 9.878511404935835e-05, "loss": 0.0452, "step": 35040 }, { "epoch": 22.339069471000638, "grad_norm": 0.7118129730224609, "learning_rate": 9.878390616571438e-05, "loss": 0.0471, "step": 35050 }, { "epoch": 22.345442957297642, "grad_norm": 0.3512742519378662, "learning_rate": 9.87826976892999e-05, "loss": 0.0403, "step": 35060 }, { "epoch": 22.351816443594647, "grad_norm": 0.6572921276092529, "learning_rate": 9.878148862012957e-05, "loss": 0.0338, "step": 35070 }, { "epoch": 22.35818992989165, "grad_norm": 0.37757009267807007, "learning_rate": 9.878027895821812e-05, "loss": 0.0373, "step": 35080 }, { "epoch": 22.364563416188656, "grad_norm": 1.0088146924972534, "learning_rate": 9.877906870358022e-05, "loss": 0.0401, "step": 35090 }, { "epoch": 22.37093690248566, "grad_norm": 0.6984426975250244, "learning_rate": 9.877785785623056e-05, "loss": 0.0427, "step": 35100 }, { "epoch": 22.377310388782664, "grad_norm": 0.30489569902420044, "learning_rate": 9.877664641618389e-05, "loss": 0.0475, "step": 35110 }, { "epoch": 22.38368387507967, "grad_norm": 0.40591955184936523, "learning_rate": 9.877543438345491e-05, "loss": 0.0434, "step": 35120 }, { "epoch": 22.390057361376673, "grad_norm": 0.822650134563446, "learning_rate": 9.877422175805835e-05, "loss": 0.0425, "step": 35130 }, { "epoch": 22.396430847673678, "grad_norm": 0.6844353079795837, "learning_rate": 9.877300854000893e-05, "loss": 0.048, "step": 35140 }, { "epoch": 22.402804333970682, "grad_norm": 0.547126054763794, "learning_rate": 9.877179472932143e-05, "loss": 0.0606, "step": 35150 }, { "epoch": 22.409177820267686, "grad_norm": 0.5891717076301575, "learning_rate": 9.877058032601056e-05, "loss": 0.0316, "step": 35160 }, { "epoch": 22.41555130656469, "grad_norm": 0.5831828117370605, "learning_rate": 9.876936533009112e-05, "loss": 0.0369, "step": 35170 }, { "epoch": 22.421924792861695, "grad_norm": 0.38758018612861633, "learning_rate": 9.876814974157781e-05, "loss": 0.0349, "step": 35180 }, { "epoch": 22.4282982791587, "grad_norm": 0.5421608686447144, "learning_rate": 9.876693356048545e-05, "loss": 0.0436, "step": 35190 }, { "epoch": 22.434671765455704, "grad_norm": 0.34861406683921814, "learning_rate": 9.87657167868288e-05, "loss": 0.0283, "step": 35200 }, { "epoch": 22.44104525175271, "grad_norm": 0.7452977895736694, "learning_rate": 9.876449942062266e-05, "loss": 0.0599, "step": 35210 }, { "epoch": 22.447418738049713, "grad_norm": 0.5051636099815369, "learning_rate": 9.876328146188181e-05, "loss": 0.042, "step": 35220 }, { "epoch": 22.453792224346717, "grad_norm": 0.3058321177959442, "learning_rate": 9.876206291062104e-05, "loss": 0.0298, "step": 35230 }, { "epoch": 22.46016571064372, "grad_norm": 0.4911765456199646, "learning_rate": 9.876084376685517e-05, "loss": 0.0381, "step": 35240 }, { "epoch": 22.466539196940726, "grad_norm": 0.6163874268531799, "learning_rate": 9.8759624030599e-05, "loss": 0.0322, "step": 35250 }, { "epoch": 22.47291268323773, "grad_norm": 0.41075339913368225, "learning_rate": 9.875840370186738e-05, "loss": 0.0307, "step": 35260 }, { "epoch": 22.479286169534735, "grad_norm": 0.6550604104995728, "learning_rate": 9.87571827806751e-05, "loss": 0.032, "step": 35270 }, { "epoch": 22.48565965583174, "grad_norm": 0.6580571532249451, "learning_rate": 9.875596126703704e-05, "loss": 0.0359, "step": 35280 }, { "epoch": 22.492033142128744, "grad_norm": 0.4544246196746826, "learning_rate": 9.8754739160968e-05, "loss": 0.0504, "step": 35290 }, { "epoch": 22.498406628425748, "grad_norm": 1.0186104774475098, "learning_rate": 9.875351646248284e-05, "loss": 0.0398, "step": 35300 }, { "epoch": 22.504780114722752, "grad_norm": 0.6584447622299194, "learning_rate": 9.875229317159645e-05, "loss": 0.0325, "step": 35310 }, { "epoch": 22.511153601019757, "grad_norm": 0.3832549750804901, "learning_rate": 9.875106928832364e-05, "loss": 0.0441, "step": 35320 }, { "epoch": 22.51752708731676, "grad_norm": 0.6077988743782043, "learning_rate": 9.874984481267933e-05, "loss": 0.0388, "step": 35330 }, { "epoch": 22.523900573613766, "grad_norm": 0.4665283262729645, "learning_rate": 9.874861974467836e-05, "loss": 0.0342, "step": 35340 }, { "epoch": 22.53027405991077, "grad_norm": 0.4998858869075775, "learning_rate": 9.874739408433565e-05, "loss": 0.0321, "step": 35350 }, { "epoch": 22.536647546207774, "grad_norm": 0.5833396315574646, "learning_rate": 9.874616783166607e-05, "loss": 0.0366, "step": 35360 }, { "epoch": 22.54302103250478, "grad_norm": 0.3186429738998413, "learning_rate": 9.874494098668452e-05, "loss": 0.0226, "step": 35370 }, { "epoch": 22.549394518801783, "grad_norm": 0.3945305645465851, "learning_rate": 9.874371354940592e-05, "loss": 0.0259, "step": 35380 }, { "epoch": 22.555768005098788, "grad_norm": 0.8879762887954712, "learning_rate": 9.874248551984519e-05, "loss": 0.0441, "step": 35390 }, { "epoch": 22.562141491395792, "grad_norm": 0.3650534749031067, "learning_rate": 9.874125689801723e-05, "loss": 0.0395, "step": 35400 }, { "epoch": 22.568514977692796, "grad_norm": 0.3361234664916992, "learning_rate": 9.874002768393698e-05, "loss": 0.04, "step": 35410 }, { "epoch": 22.5748884639898, "grad_norm": 0.5994387865066528, "learning_rate": 9.873879787761938e-05, "loss": 0.0352, "step": 35420 }, { "epoch": 22.581261950286805, "grad_norm": 0.7112413048744202, "learning_rate": 9.873756747907935e-05, "loss": 0.0527, "step": 35430 }, { "epoch": 22.58763543658381, "grad_norm": 0.5254963040351868, "learning_rate": 9.873633648833187e-05, "loss": 0.032, "step": 35440 }, { "epoch": 22.594008922880818, "grad_norm": 0.4152343273162842, "learning_rate": 9.873510490539189e-05, "loss": 0.0381, "step": 35450 }, { "epoch": 22.60038240917782, "grad_norm": 0.5694442391395569, "learning_rate": 9.873387273027438e-05, "loss": 0.0317, "step": 35460 }, { "epoch": 22.606755895474826, "grad_norm": 0.1981181502342224, "learning_rate": 9.873263996299429e-05, "loss": 0.0288, "step": 35470 }, { "epoch": 22.61312938177183, "grad_norm": 0.6542591452598572, "learning_rate": 9.873140660356663e-05, "loss": 0.0337, "step": 35480 }, { "epoch": 22.619502868068835, "grad_norm": 0.8989945650100708, "learning_rate": 9.873017265200635e-05, "loss": 0.0337, "step": 35490 }, { "epoch": 22.62587635436584, "grad_norm": 0.7037742137908936, "learning_rate": 9.872893810832847e-05, "loss": 0.0297, "step": 35500 }, { "epoch": 22.632249840662844, "grad_norm": 0.5993256568908691, "learning_rate": 9.872770297254799e-05, "loss": 0.0362, "step": 35510 }, { "epoch": 22.63862332695985, "grad_norm": 0.4735136330127716, "learning_rate": 9.872646724467991e-05, "loss": 0.0368, "step": 35520 }, { "epoch": 22.644996813256853, "grad_norm": 0.2969745397567749, "learning_rate": 9.872523092473924e-05, "loss": 0.0417, "step": 35530 }, { "epoch": 22.651370299553857, "grad_norm": 0.4498269855976105, "learning_rate": 9.872399401274101e-05, "loss": 0.0333, "step": 35540 }, { "epoch": 22.65774378585086, "grad_norm": 0.6345143914222717, "learning_rate": 9.872275650870027e-05, "loss": 0.0402, "step": 35550 }, { "epoch": 22.664117272147866, "grad_norm": 0.6988514065742493, "learning_rate": 9.872151841263201e-05, "loss": 0.0405, "step": 35560 }, { "epoch": 22.67049075844487, "grad_norm": 0.7707237601280212, "learning_rate": 9.872027972455132e-05, "loss": 0.0364, "step": 35570 }, { "epoch": 22.676864244741875, "grad_norm": 0.6449145078659058, "learning_rate": 9.871904044447322e-05, "loss": 0.0431, "step": 35580 }, { "epoch": 22.68323773103888, "grad_norm": 0.5047450065612793, "learning_rate": 9.87178005724128e-05, "loss": 0.0315, "step": 35590 }, { "epoch": 22.689611217335884, "grad_norm": 0.594368040561676, "learning_rate": 9.871656010838508e-05, "loss": 0.0429, "step": 35600 }, { "epoch": 22.695984703632888, "grad_norm": 0.2974255084991455, "learning_rate": 9.871531905240517e-05, "loss": 0.0437, "step": 35610 }, { "epoch": 22.702358189929893, "grad_norm": 0.36203864216804504, "learning_rate": 9.871407740448815e-05, "loss": 0.0386, "step": 35620 }, { "epoch": 22.708731676226897, "grad_norm": 0.3179348409175873, "learning_rate": 9.871283516464908e-05, "loss": 0.0224, "step": 35630 }, { "epoch": 22.7151051625239, "grad_norm": 0.3596300482749939, "learning_rate": 9.871159233290307e-05, "loss": 0.029, "step": 35640 }, { "epoch": 22.721478648820906, "grad_norm": 0.6995904445648193, "learning_rate": 9.871034890926523e-05, "loss": 0.0421, "step": 35650 }, { "epoch": 22.72785213511791, "grad_norm": 0.5614590048789978, "learning_rate": 9.870910489375065e-05, "loss": 0.0377, "step": 35660 }, { "epoch": 22.734225621414915, "grad_norm": 0.616037905216217, "learning_rate": 9.870786028637446e-05, "loss": 0.0423, "step": 35670 }, { "epoch": 22.74059910771192, "grad_norm": 0.6043813824653625, "learning_rate": 9.870661508715178e-05, "loss": 0.0361, "step": 35680 }, { "epoch": 22.746972594008923, "grad_norm": 0.613956093788147, "learning_rate": 9.870536929609776e-05, "loss": 0.0312, "step": 35690 }, { "epoch": 22.753346080305928, "grad_norm": 0.48165464401245117, "learning_rate": 9.87041229132275e-05, "loss": 0.0341, "step": 35700 }, { "epoch": 22.759719566602932, "grad_norm": 0.3860820233821869, "learning_rate": 9.870287593855616e-05, "loss": 0.0358, "step": 35710 }, { "epoch": 22.766093052899937, "grad_norm": 0.5796486139297485, "learning_rate": 9.870162837209889e-05, "loss": 0.0284, "step": 35720 }, { "epoch": 22.77246653919694, "grad_norm": 0.7215161919593811, "learning_rate": 9.870038021387086e-05, "loss": 0.0273, "step": 35730 }, { "epoch": 22.778840025493945, "grad_norm": 0.6009196639060974, "learning_rate": 9.869913146388723e-05, "loss": 0.0333, "step": 35740 }, { "epoch": 22.78521351179095, "grad_norm": 0.7514901757240295, "learning_rate": 9.869788212216318e-05, "loss": 0.0375, "step": 35750 }, { "epoch": 22.791586998087954, "grad_norm": 0.7864081859588623, "learning_rate": 9.869663218871387e-05, "loss": 0.0359, "step": 35760 }, { "epoch": 22.79796048438496, "grad_norm": 0.2704887092113495, "learning_rate": 9.869538166355451e-05, "loss": 0.0384, "step": 35770 }, { "epoch": 22.804333970681963, "grad_norm": 0.5436603426933289, "learning_rate": 9.869413054670028e-05, "loss": 0.0461, "step": 35780 }, { "epoch": 22.810707456978967, "grad_norm": 0.519588053226471, "learning_rate": 9.869287883816641e-05, "loss": 0.031, "step": 35790 }, { "epoch": 22.817080943275972, "grad_norm": 0.4137604832649231, "learning_rate": 9.869162653796807e-05, "loss": 0.0376, "step": 35800 }, { "epoch": 22.823454429572976, "grad_norm": 0.29036200046539307, "learning_rate": 9.869037364612048e-05, "loss": 0.0403, "step": 35810 }, { "epoch": 22.82982791586998, "grad_norm": 0.5087034106254578, "learning_rate": 9.86891201626389e-05, "loss": 0.0248, "step": 35820 }, { "epoch": 22.836201402166985, "grad_norm": 0.35461094975471497, "learning_rate": 9.868786608753853e-05, "loss": 0.0325, "step": 35830 }, { "epoch": 22.84257488846399, "grad_norm": 0.2711033225059509, "learning_rate": 9.868661142083464e-05, "loss": 0.0261, "step": 35840 }, { "epoch": 22.848948374760994, "grad_norm": 0.536163330078125, "learning_rate": 9.868535616254243e-05, "loss": 0.0459, "step": 35850 }, { "epoch": 22.855321861058, "grad_norm": 0.5534836649894714, "learning_rate": 9.868410031267718e-05, "loss": 0.0429, "step": 35860 }, { "epoch": 22.861695347355003, "grad_norm": 0.4618101119995117, "learning_rate": 9.868284387125415e-05, "loss": 0.0293, "step": 35870 }, { "epoch": 22.868068833652007, "grad_norm": 0.6351956725120544, "learning_rate": 9.868158683828862e-05, "loss": 0.034, "step": 35880 }, { "epoch": 22.87444231994901, "grad_norm": 0.409598171710968, "learning_rate": 9.868032921379582e-05, "loss": 0.0361, "step": 35890 }, { "epoch": 22.880815806246016, "grad_norm": 0.3513404428958893, "learning_rate": 9.867907099779106e-05, "loss": 0.0328, "step": 35900 }, { "epoch": 22.88718929254302, "grad_norm": 0.46665483713150024, "learning_rate": 9.867781219028964e-05, "loss": 0.0254, "step": 35910 }, { "epoch": 22.893562778840025, "grad_norm": 0.6560744047164917, "learning_rate": 9.867655279130683e-05, "loss": 0.0531, "step": 35920 }, { "epoch": 22.89993626513703, "grad_norm": 0.4701434075832367, "learning_rate": 9.867529280085795e-05, "loss": 0.0261, "step": 35930 }, { "epoch": 22.906309751434033, "grad_norm": 0.5362236499786377, "learning_rate": 9.86740322189583e-05, "loss": 0.034, "step": 35940 }, { "epoch": 22.912683237731038, "grad_norm": 0.43048354983329773, "learning_rate": 9.867277104562323e-05, "loss": 0.0428, "step": 35950 }, { "epoch": 22.919056724028042, "grad_norm": 0.4683257043361664, "learning_rate": 9.867150928086801e-05, "loss": 0.0388, "step": 35960 }, { "epoch": 22.925430210325047, "grad_norm": 0.2844640612602234, "learning_rate": 9.8670246924708e-05, "loss": 0.0342, "step": 35970 }, { "epoch": 22.93180369662205, "grad_norm": 0.68386310338974, "learning_rate": 9.866898397715855e-05, "loss": 0.0412, "step": 35980 }, { "epoch": 22.938177182919055, "grad_norm": 0.5814060568809509, "learning_rate": 9.866772043823498e-05, "loss": 0.0475, "step": 35990 }, { "epoch": 22.94455066921606, "grad_norm": 0.49509575963020325, "learning_rate": 9.866645630795267e-05, "loss": 0.0342, "step": 36000 }, { "epoch": 22.950924155513064, "grad_norm": 0.24639937281608582, "learning_rate": 9.866519158632697e-05, "loss": 0.0394, "step": 36010 }, { "epoch": 22.95729764181007, "grad_norm": 0.26937276124954224, "learning_rate": 9.866392627337322e-05, "loss": 0.0292, "step": 36020 }, { "epoch": 22.963671128107073, "grad_norm": 0.587815523147583, "learning_rate": 9.866266036910683e-05, "loss": 0.0341, "step": 36030 }, { "epoch": 22.970044614404078, "grad_norm": 0.41701415181159973, "learning_rate": 9.866139387354318e-05, "loss": 0.0263, "step": 36040 }, { "epoch": 22.976418100701082, "grad_norm": 0.4268590807914734, "learning_rate": 9.866012678669764e-05, "loss": 0.0383, "step": 36050 }, { "epoch": 22.982791586998086, "grad_norm": 0.5448164343833923, "learning_rate": 9.865885910858564e-05, "loss": 0.0259, "step": 36060 }, { "epoch": 22.98916507329509, "grad_norm": 0.5478262901306152, "learning_rate": 9.865759083922253e-05, "loss": 0.0249, "step": 36070 }, { "epoch": 22.9955385595921, "grad_norm": 0.3121899366378784, "learning_rate": 9.865632197862377e-05, "loss": 0.0282, "step": 36080 }, { "epoch": 23.001912045889103, "grad_norm": 0.45514288544654846, "learning_rate": 9.865505252680476e-05, "loss": 0.0309, "step": 36090 }, { "epoch": 23.008285532186108, "grad_norm": 0.31081923842430115, "learning_rate": 9.865378248378091e-05, "loss": 0.0309, "step": 36100 }, { "epoch": 23.014659018483112, "grad_norm": 0.656318724155426, "learning_rate": 9.865251184956767e-05, "loss": 0.0498, "step": 36110 }, { "epoch": 23.021032504780116, "grad_norm": 0.42957475781440735, "learning_rate": 9.865124062418048e-05, "loss": 0.0208, "step": 36120 }, { "epoch": 23.02740599107712, "grad_norm": 0.5568720698356628, "learning_rate": 9.864996880763478e-05, "loss": 0.0248, "step": 36130 }, { "epoch": 23.033779477374125, "grad_norm": 0.2583651840686798, "learning_rate": 9.864869639994603e-05, "loss": 0.0328, "step": 36140 }, { "epoch": 23.04015296367113, "grad_norm": 0.5028584599494934, "learning_rate": 9.864742340112969e-05, "loss": 0.0318, "step": 36150 }, { "epoch": 23.046526449968134, "grad_norm": 0.8908265233039856, "learning_rate": 9.864614981120122e-05, "loss": 0.0349, "step": 36160 }, { "epoch": 23.05289993626514, "grad_norm": 0.6324557662010193, "learning_rate": 9.864487563017609e-05, "loss": 0.0337, "step": 36170 }, { "epoch": 23.059273422562143, "grad_norm": 0.5461339950561523, "learning_rate": 9.864360085806981e-05, "loss": 0.036, "step": 36180 }, { "epoch": 23.065646908859147, "grad_norm": 0.6034703254699707, "learning_rate": 9.864232549489782e-05, "loss": 0.0273, "step": 36190 }, { "epoch": 23.07202039515615, "grad_norm": 0.9316211938858032, "learning_rate": 9.864104954067567e-05, "loss": 0.0387, "step": 36200 }, { "epoch": 23.078393881453156, "grad_norm": 0.4064961373806, "learning_rate": 9.863977299541887e-05, "loss": 0.0243, "step": 36210 }, { "epoch": 23.08476736775016, "grad_norm": 0.4518246054649353, "learning_rate": 9.863849585914286e-05, "loss": 0.0499, "step": 36220 }, { "epoch": 23.091140854047165, "grad_norm": 0.7394051551818848, "learning_rate": 9.863721813186324e-05, "loss": 0.0354, "step": 36230 }, { "epoch": 23.09751434034417, "grad_norm": 0.3792189359664917, "learning_rate": 9.863593981359548e-05, "loss": 0.0345, "step": 36240 }, { "epoch": 23.103887826641174, "grad_norm": 0.6214410066604614, "learning_rate": 9.863466090435513e-05, "loss": 0.0222, "step": 36250 }, { "epoch": 23.110261312938178, "grad_norm": 0.34806138277053833, "learning_rate": 9.863338140415774e-05, "loss": 0.0322, "step": 36260 }, { "epoch": 23.116634799235182, "grad_norm": 0.434895396232605, "learning_rate": 9.863210131301883e-05, "loss": 0.0347, "step": 36270 }, { "epoch": 23.123008285532187, "grad_norm": 0.5578153133392334, "learning_rate": 9.863082063095399e-05, "loss": 0.0328, "step": 36280 }, { "epoch": 23.12938177182919, "grad_norm": 0.6369719505310059, "learning_rate": 9.862953935797876e-05, "loss": 0.0438, "step": 36290 }, { "epoch": 23.135755258126196, "grad_norm": 0.3536309003829956, "learning_rate": 9.862825749410871e-05, "loss": 0.042, "step": 36300 }, { "epoch": 23.1421287444232, "grad_norm": 0.5308420658111572, "learning_rate": 9.862697503935942e-05, "loss": 0.0247, "step": 36310 }, { "epoch": 23.148502230720204, "grad_norm": 0.7654992341995239, "learning_rate": 9.862569199374648e-05, "loss": 0.0301, "step": 36320 }, { "epoch": 23.15487571701721, "grad_norm": 0.6247915625572205, "learning_rate": 9.862440835728547e-05, "loss": 0.0402, "step": 36330 }, { "epoch": 23.161249203314213, "grad_norm": 0.36276599764823914, "learning_rate": 9.862312412999198e-05, "loss": 0.0314, "step": 36340 }, { "epoch": 23.167622689611218, "grad_norm": 0.4394279420375824, "learning_rate": 9.862183931188162e-05, "loss": 0.0284, "step": 36350 }, { "epoch": 23.173996175908222, "grad_norm": 0.5732686519622803, "learning_rate": 9.862055390297003e-05, "loss": 0.0392, "step": 36360 }, { "epoch": 23.180369662205226, "grad_norm": 0.4227307438850403, "learning_rate": 9.861926790327278e-05, "loss": 0.0313, "step": 36370 }, { "epoch": 23.18674314850223, "grad_norm": 0.6741951704025269, "learning_rate": 9.861798131280553e-05, "loss": 0.038, "step": 36380 }, { "epoch": 23.193116634799235, "grad_norm": 0.2974992096424103, "learning_rate": 9.861669413158391e-05, "loss": 0.0342, "step": 36390 }, { "epoch": 23.19949012109624, "grad_norm": 0.3867451548576355, "learning_rate": 9.861540635962354e-05, "loss": 0.0405, "step": 36400 }, { "epoch": 23.205863607393244, "grad_norm": 0.4575144052505493, "learning_rate": 9.861411799694009e-05, "loss": 0.035, "step": 36410 }, { "epoch": 23.21223709369025, "grad_norm": 0.5691337585449219, "learning_rate": 9.861282904354922e-05, "loss": 0.0283, "step": 36420 }, { "epoch": 23.218610579987253, "grad_norm": 0.5980804562568665, "learning_rate": 9.861153949946655e-05, "loss": 0.0356, "step": 36430 }, { "epoch": 23.224984066284257, "grad_norm": 0.5748704671859741, "learning_rate": 9.86102493647078e-05, "loss": 0.0463, "step": 36440 }, { "epoch": 23.23135755258126, "grad_norm": 0.752718448638916, "learning_rate": 9.860895863928863e-05, "loss": 0.04, "step": 36450 }, { "epoch": 23.237731038878266, "grad_norm": 0.4540983736515045, "learning_rate": 9.86076673232247e-05, "loss": 0.0421, "step": 36460 }, { "epoch": 23.24410452517527, "grad_norm": 0.8507975339889526, "learning_rate": 9.860637541653172e-05, "loss": 0.0428, "step": 36470 }, { "epoch": 23.250478011472275, "grad_norm": 0.5903734564781189, "learning_rate": 9.860508291922541e-05, "loss": 0.0328, "step": 36480 }, { "epoch": 23.25685149776928, "grad_norm": 0.4271882176399231, "learning_rate": 9.860378983132143e-05, "loss": 0.0281, "step": 36490 }, { "epoch": 23.263224984066284, "grad_norm": 0.3872644007205963, "learning_rate": 9.860249615283552e-05, "loss": 0.0307, "step": 36500 }, { "epoch": 23.269598470363288, "grad_norm": 0.4694727659225464, "learning_rate": 9.860120188378339e-05, "loss": 0.0271, "step": 36510 }, { "epoch": 23.275971956660293, "grad_norm": 0.7661231756210327, "learning_rate": 9.859990702418077e-05, "loss": 0.0299, "step": 36520 }, { "epoch": 23.282345442957297, "grad_norm": 0.5404526591300964, "learning_rate": 9.85986115740434e-05, "loss": 0.0304, "step": 36530 }, { "epoch": 23.2887189292543, "grad_norm": 0.5391700863838196, "learning_rate": 9.859731553338701e-05, "loss": 0.0323, "step": 36540 }, { "epoch": 23.295092415551306, "grad_norm": 0.5226530432701111, "learning_rate": 9.859601890222735e-05, "loss": 0.0359, "step": 36550 }, { "epoch": 23.30146590184831, "grad_norm": 0.4017866551876068, "learning_rate": 9.859472168058018e-05, "loss": 0.0281, "step": 36560 }, { "epoch": 23.307839388145315, "grad_norm": 0.5931110382080078, "learning_rate": 9.859342386846127e-05, "loss": 0.0359, "step": 36570 }, { "epoch": 23.31421287444232, "grad_norm": 0.39134860038757324, "learning_rate": 9.859212546588636e-05, "loss": 0.0236, "step": 36580 }, { "epoch": 23.320586360739323, "grad_norm": 0.5923264026641846, "learning_rate": 9.859082647287126e-05, "loss": 0.047, "step": 36590 }, { "epoch": 23.326959847036328, "grad_norm": 0.49127089977264404, "learning_rate": 9.858952688943174e-05, "loss": 0.0443, "step": 36600 }, { "epoch": 23.333333333333332, "grad_norm": 0.48142293095588684, "learning_rate": 9.85882267155836e-05, "loss": 0.0376, "step": 36610 }, { "epoch": 23.339706819630337, "grad_norm": 0.6156603097915649, "learning_rate": 9.858692595134263e-05, "loss": 0.0286, "step": 36620 }, { "epoch": 23.34608030592734, "grad_norm": 0.6895135641098022, "learning_rate": 9.858562459672462e-05, "loss": 0.032, "step": 36630 }, { "epoch": 23.352453792224345, "grad_norm": 0.45453163981437683, "learning_rate": 9.858432265174541e-05, "loss": 0.0419, "step": 36640 }, { "epoch": 23.35882727852135, "grad_norm": 0.603053867816925, "learning_rate": 9.85830201164208e-05, "loss": 0.0291, "step": 36650 }, { "epoch": 23.365200764818354, "grad_norm": 0.41972616314888, "learning_rate": 9.858171699076664e-05, "loss": 0.0347, "step": 36660 }, { "epoch": 23.37157425111536, "grad_norm": 0.5342583656311035, "learning_rate": 9.858041327479875e-05, "loss": 0.0324, "step": 36670 }, { "epoch": 23.377947737412363, "grad_norm": 0.35054877400398254, "learning_rate": 9.857910896853295e-05, "loss": 0.0427, "step": 36680 }, { "epoch": 23.384321223709367, "grad_norm": 0.49216216802597046, "learning_rate": 9.857780407198512e-05, "loss": 0.035, "step": 36690 }, { "epoch": 23.390694710006372, "grad_norm": 0.39326122403144836, "learning_rate": 9.85764985851711e-05, "loss": 0.0351, "step": 36700 }, { "epoch": 23.39706819630338, "grad_norm": 0.6145009994506836, "learning_rate": 9.857519250810678e-05, "loss": 0.0212, "step": 36710 }, { "epoch": 23.403441682600384, "grad_norm": 0.4676506519317627, "learning_rate": 9.8573885840808e-05, "loss": 0.0363, "step": 36720 }, { "epoch": 23.40981516889739, "grad_norm": 0.3322041928768158, "learning_rate": 9.857257858329063e-05, "loss": 0.0359, "step": 36730 }, { "epoch": 23.416188655194393, "grad_norm": 0.6079815030097961, "learning_rate": 9.857127073557057e-05, "loss": 0.039, "step": 36740 }, { "epoch": 23.422562141491397, "grad_norm": 0.5640106201171875, "learning_rate": 9.856996229766374e-05, "loss": 0.02, "step": 36750 }, { "epoch": 23.4289356277884, "grad_norm": 0.7194762229919434, "learning_rate": 9.856865326958599e-05, "loss": 0.0318, "step": 36760 }, { "epoch": 23.435309114085406, "grad_norm": 0.4033360183238983, "learning_rate": 9.856734365135324e-05, "loss": 0.0333, "step": 36770 }, { "epoch": 23.44168260038241, "grad_norm": 0.5370861887931824, "learning_rate": 9.856603344298143e-05, "loss": 0.0384, "step": 36780 }, { "epoch": 23.448056086679415, "grad_norm": 0.45059478282928467, "learning_rate": 9.856472264448645e-05, "loss": 0.0331, "step": 36790 }, { "epoch": 23.45442957297642, "grad_norm": 0.6513437628746033, "learning_rate": 9.856341125588423e-05, "loss": 0.0406, "step": 36800 }, { "epoch": 23.460803059273424, "grad_norm": 0.3299298882484436, "learning_rate": 9.856209927719071e-05, "loss": 0.0291, "step": 36810 }, { "epoch": 23.467176545570428, "grad_norm": 0.5888707041740417, "learning_rate": 9.856078670842186e-05, "loss": 0.0365, "step": 36820 }, { "epoch": 23.473550031867433, "grad_norm": 0.7050340175628662, "learning_rate": 9.855947354959357e-05, "loss": 0.0448, "step": 36830 }, { "epoch": 23.479923518164437, "grad_norm": 0.4835750460624695, "learning_rate": 9.855815980072185e-05, "loss": 0.0289, "step": 36840 }, { "epoch": 23.48629700446144, "grad_norm": 0.6583870053291321, "learning_rate": 9.855684546182262e-05, "loss": 0.0324, "step": 36850 }, { "epoch": 23.492670490758446, "grad_norm": 0.8079191446304321, "learning_rate": 9.85555305329119e-05, "loss": 0.0515, "step": 36860 }, { "epoch": 23.49904397705545, "grad_norm": 0.38979196548461914, "learning_rate": 9.855421501400562e-05, "loss": 0.035, "step": 36870 }, { "epoch": 23.505417463352455, "grad_norm": 0.421710342168808, "learning_rate": 9.855289890511979e-05, "loss": 0.0393, "step": 36880 }, { "epoch": 23.51179094964946, "grad_norm": 0.35926273465156555, "learning_rate": 9.85515822062704e-05, "loss": 0.0441, "step": 36890 }, { "epoch": 23.518164435946463, "grad_norm": 0.571291983127594, "learning_rate": 9.855026491747344e-05, "loss": 0.0398, "step": 36900 }, { "epoch": 23.524537922243468, "grad_norm": 0.7023671269416809, "learning_rate": 9.854894703874493e-05, "loss": 0.0367, "step": 36910 }, { "epoch": 23.530911408540472, "grad_norm": 0.2783041298389435, "learning_rate": 9.854762857010087e-05, "loss": 0.0406, "step": 36920 }, { "epoch": 23.537284894837477, "grad_norm": 1.0094852447509766, "learning_rate": 9.85463095115573e-05, "loss": 0.0425, "step": 36930 }, { "epoch": 23.54365838113448, "grad_norm": 0.47917866706848145, "learning_rate": 9.854498986313021e-05, "loss": 0.0348, "step": 36940 }, { "epoch": 23.550031867431485, "grad_norm": 0.4936361312866211, "learning_rate": 9.854366962483568e-05, "loss": 0.0393, "step": 36950 }, { "epoch": 23.55640535372849, "grad_norm": 0.9955158233642578, "learning_rate": 9.854234879668971e-05, "loss": 0.0262, "step": 36960 }, { "epoch": 23.562778840025494, "grad_norm": 0.8160923719406128, "learning_rate": 9.854102737870838e-05, "loss": 0.0577, "step": 36970 }, { "epoch": 23.5691523263225, "grad_norm": 0.6121181845664978, "learning_rate": 9.853970537090775e-05, "loss": 0.0481, "step": 36980 }, { "epoch": 23.575525812619503, "grad_norm": 0.5603047013282776, "learning_rate": 9.853838277330385e-05, "loss": 0.0405, "step": 36990 }, { "epoch": 23.581899298916507, "grad_norm": 0.25304895639419556, "learning_rate": 9.853705958591279e-05, "loss": 0.0271, "step": 37000 }, { "epoch": 23.588272785213512, "grad_norm": 0.415301650762558, "learning_rate": 9.853573580875062e-05, "loss": 0.037, "step": 37010 }, { "epoch": 23.594646271510516, "grad_norm": 0.43318790197372437, "learning_rate": 9.853441144183345e-05, "loss": 0.0335, "step": 37020 }, { "epoch": 23.60101975780752, "grad_norm": 0.4673224091529846, "learning_rate": 9.853308648517732e-05, "loss": 0.0359, "step": 37030 }, { "epoch": 23.607393244104525, "grad_norm": 0.47700899839401245, "learning_rate": 9.85317609387984e-05, "loss": 0.0279, "step": 37040 }, { "epoch": 23.61376673040153, "grad_norm": 0.9069885015487671, "learning_rate": 9.853043480271276e-05, "loss": 0.036, "step": 37050 }, { "epoch": 23.620140216698534, "grad_norm": 1.0800111293792725, "learning_rate": 9.85291080769365e-05, "loss": 0.0415, "step": 37060 }, { "epoch": 23.62651370299554, "grad_norm": 0.37277042865753174, "learning_rate": 9.852778076148576e-05, "loss": 0.0271, "step": 37070 }, { "epoch": 23.632887189292543, "grad_norm": 0.6269493699073792, "learning_rate": 9.852645285637668e-05, "loss": 0.0416, "step": 37080 }, { "epoch": 23.639260675589547, "grad_norm": 0.2793123424053192, "learning_rate": 9.852512436162538e-05, "loss": 0.0317, "step": 37090 }, { "epoch": 23.64563416188655, "grad_norm": 0.34734880924224854, "learning_rate": 9.8523795277248e-05, "loss": 0.0293, "step": 37100 }, { "epoch": 23.652007648183556, "grad_norm": 0.661919891834259, "learning_rate": 9.852246560326069e-05, "loss": 0.0383, "step": 37110 }, { "epoch": 23.65838113448056, "grad_norm": 0.33034586906433105, "learning_rate": 9.852113533967961e-05, "loss": 0.0295, "step": 37120 }, { "epoch": 23.664754620777565, "grad_norm": 0.5846928954124451, "learning_rate": 9.851980448652092e-05, "loss": 0.0381, "step": 37130 }, { "epoch": 23.67112810707457, "grad_norm": 0.41564038395881653, "learning_rate": 9.85184730438008e-05, "loss": 0.0362, "step": 37140 }, { "epoch": 23.677501593371574, "grad_norm": 0.39438435435295105, "learning_rate": 9.851714101153542e-05, "loss": 0.0311, "step": 37150 }, { "epoch": 23.683875079668578, "grad_norm": 0.6188374161720276, "learning_rate": 9.851580838974097e-05, "loss": 0.0284, "step": 37160 }, { "epoch": 23.690248565965582, "grad_norm": 0.4564496874809265, "learning_rate": 9.851447517843366e-05, "loss": 0.0383, "step": 37170 }, { "epoch": 23.696622052262587, "grad_norm": 0.6168647408485413, "learning_rate": 9.851314137762965e-05, "loss": 0.0311, "step": 37180 }, { "epoch": 23.70299553855959, "grad_norm": 0.416768878698349, "learning_rate": 9.851180698734517e-05, "loss": 0.05, "step": 37190 }, { "epoch": 23.709369024856596, "grad_norm": 0.6282676458358765, "learning_rate": 9.851047200759645e-05, "loss": 0.0391, "step": 37200 }, { "epoch": 23.7157425111536, "grad_norm": 0.5219295620918274, "learning_rate": 9.850913643839967e-05, "loss": 0.028, "step": 37210 }, { "epoch": 23.722115997450604, "grad_norm": 0.4956703782081604, "learning_rate": 9.850780027977109e-05, "loss": 0.0295, "step": 37220 }, { "epoch": 23.72848948374761, "grad_norm": 0.6324825882911682, "learning_rate": 9.850646353172694e-05, "loss": 0.0371, "step": 37230 }, { "epoch": 23.734862970044613, "grad_norm": 0.5696253180503845, "learning_rate": 9.850512619428346e-05, "loss": 0.0455, "step": 37240 }, { "epoch": 23.741236456341618, "grad_norm": 0.5835239887237549, "learning_rate": 9.85037882674569e-05, "loss": 0.0412, "step": 37250 }, { "epoch": 23.747609942638622, "grad_norm": 0.6295267939567566, "learning_rate": 9.850244975126352e-05, "loss": 0.0326, "step": 37260 }, { "epoch": 23.753983428935626, "grad_norm": 0.6781933307647705, "learning_rate": 9.850111064571958e-05, "loss": 0.0277, "step": 37270 }, { "epoch": 23.76035691523263, "grad_norm": 0.8851192593574524, "learning_rate": 9.849977095084135e-05, "loss": 0.035, "step": 37280 }, { "epoch": 23.766730401529635, "grad_norm": 0.7456137537956238, "learning_rate": 9.849843066664512e-05, "loss": 0.0455, "step": 37290 }, { "epoch": 23.77310388782664, "grad_norm": 1.1779190301895142, "learning_rate": 9.849708979314716e-05, "loss": 0.0334, "step": 37300 }, { "epoch": 23.779477374123644, "grad_norm": 0.801834762096405, "learning_rate": 9.849574833036377e-05, "loss": 0.0375, "step": 37310 }, { "epoch": 23.78585086042065, "grad_norm": 0.7612029910087585, "learning_rate": 9.849440627831126e-05, "loss": 0.0391, "step": 37320 }, { "epoch": 23.792224346717653, "grad_norm": 0.49355363845825195, "learning_rate": 9.849306363700591e-05, "loss": 0.0411, "step": 37330 }, { "epoch": 23.79859783301466, "grad_norm": 0.4595854580402374, "learning_rate": 9.849172040646406e-05, "loss": 0.0264, "step": 37340 }, { "epoch": 23.80497131931166, "grad_norm": 0.2429874986410141, "learning_rate": 9.849037658670202e-05, "loss": 0.039, "step": 37350 }, { "epoch": 23.81134480560867, "grad_norm": 0.6151285767555237, "learning_rate": 9.848903217773613e-05, "loss": 0.0438, "step": 37360 }, { "epoch": 23.817718291905674, "grad_norm": 0.49741554260253906, "learning_rate": 9.848768717958272e-05, "loss": 0.0286, "step": 37370 }, { "epoch": 23.82409177820268, "grad_norm": 0.5290849208831787, "learning_rate": 9.848634159225812e-05, "loss": 0.0279, "step": 37380 }, { "epoch": 23.830465264499683, "grad_norm": 0.6803759336471558, "learning_rate": 9.848499541577871e-05, "loss": 0.0368, "step": 37390 }, { "epoch": 23.836838750796687, "grad_norm": 0.5212733745574951, "learning_rate": 9.84836486501608e-05, "loss": 0.0338, "step": 37400 }, { "epoch": 23.84321223709369, "grad_norm": 0.3238081932067871, "learning_rate": 9.848230129542079e-05, "loss": 0.033, "step": 37410 }, { "epoch": 23.849585723390696, "grad_norm": 0.4323550760746002, "learning_rate": 9.848095335157506e-05, "loss": 0.0375, "step": 37420 }, { "epoch": 23.8559592096877, "grad_norm": 0.3778531849384308, "learning_rate": 9.847960481863997e-05, "loss": 0.0349, "step": 37430 }, { "epoch": 23.862332695984705, "grad_norm": 0.4219679832458496, "learning_rate": 9.847825569663189e-05, "loss": 0.0303, "step": 37440 }, { "epoch": 23.86870618228171, "grad_norm": 0.680701494216919, "learning_rate": 9.847690598556726e-05, "loss": 0.0582, "step": 37450 }, { "epoch": 23.875079668578714, "grad_norm": 0.5344898104667664, "learning_rate": 9.847555568546243e-05, "loss": 0.0448, "step": 37460 }, { "epoch": 23.881453154875718, "grad_norm": 0.798287570476532, "learning_rate": 9.847420479633384e-05, "loss": 0.0487, "step": 37470 }, { "epoch": 23.887826641172722, "grad_norm": 0.5545568466186523, "learning_rate": 9.847285331819788e-05, "loss": 0.0365, "step": 37480 }, { "epoch": 23.894200127469727, "grad_norm": 0.6355661153793335, "learning_rate": 9.8471501251071e-05, "loss": 0.0323, "step": 37490 }, { "epoch": 23.90057361376673, "grad_norm": 0.6961902976036072, "learning_rate": 9.847014859496963e-05, "loss": 0.045, "step": 37500 }, { "epoch": 23.906947100063736, "grad_norm": 0.7367477416992188, "learning_rate": 9.846879534991018e-05, "loss": 0.0417, "step": 37510 }, { "epoch": 23.91332058636074, "grad_norm": 0.4929988384246826, "learning_rate": 9.84674415159091e-05, "loss": 0.0315, "step": 37520 }, { "epoch": 23.919694072657744, "grad_norm": 1.0386641025543213, "learning_rate": 9.846608709298284e-05, "loss": 0.0454, "step": 37530 }, { "epoch": 23.92606755895475, "grad_norm": 0.4639098048210144, "learning_rate": 9.846473208114786e-05, "loss": 0.0316, "step": 37540 }, { "epoch": 23.932441045251753, "grad_norm": 0.9023931622505188, "learning_rate": 9.846337648042063e-05, "loss": 0.0409, "step": 37550 }, { "epoch": 23.938814531548758, "grad_norm": 0.3718760013580322, "learning_rate": 9.846202029081763e-05, "loss": 0.0325, "step": 37560 }, { "epoch": 23.945188017845762, "grad_norm": 0.43446189165115356, "learning_rate": 9.846066351235532e-05, "loss": 0.0382, "step": 37570 }, { "epoch": 23.951561504142767, "grad_norm": 0.30794772505760193, "learning_rate": 9.845930614505019e-05, "loss": 0.0305, "step": 37580 }, { "epoch": 23.95793499043977, "grad_norm": 0.4514079988002777, "learning_rate": 9.845794818891875e-05, "loss": 0.026, "step": 37590 }, { "epoch": 23.964308476736775, "grad_norm": 0.5350214838981628, "learning_rate": 9.845658964397748e-05, "loss": 0.0339, "step": 37600 }, { "epoch": 23.97068196303378, "grad_norm": 0.5738411545753479, "learning_rate": 9.84552305102429e-05, "loss": 0.0392, "step": 37610 }, { "epoch": 23.977055449330784, "grad_norm": 0.7394959330558777, "learning_rate": 9.84538707877315e-05, "loss": 0.0411, "step": 37620 }, { "epoch": 23.98342893562779, "grad_norm": 0.632942259311676, "learning_rate": 9.845251047645983e-05, "loss": 0.0328, "step": 37630 }, { "epoch": 23.989802421924793, "grad_norm": 0.37757253646850586, "learning_rate": 9.845114957644442e-05, "loss": 0.0332, "step": 37640 }, { "epoch": 23.996175908221797, "grad_norm": 0.46853557229042053, "learning_rate": 9.84497880877018e-05, "loss": 0.0318, "step": 37650 }, { "epoch": 24.0025493945188, "grad_norm": 0.4802883565425873, "learning_rate": 9.844842601024851e-05, "loss": 0.032, "step": 37660 }, { "epoch": 24.008922880815806, "grad_norm": 0.5748679637908936, "learning_rate": 9.844706334410108e-05, "loss": 0.053, "step": 37670 }, { "epoch": 24.01529636711281, "grad_norm": 0.6048776507377625, "learning_rate": 9.844570008927611e-05, "loss": 0.0357, "step": 37680 }, { "epoch": 24.021669853409815, "grad_norm": 0.7143999934196472, "learning_rate": 9.844433624579014e-05, "loss": 0.0361, "step": 37690 }, { "epoch": 24.02804333970682, "grad_norm": 0.4235288202762604, "learning_rate": 9.844297181365974e-05, "loss": 0.0297, "step": 37700 }, { "epoch": 24.034416826003824, "grad_norm": 0.4564266502857208, "learning_rate": 9.84416067929015e-05, "loss": 0.0368, "step": 37710 }, { "epoch": 24.040790312300828, "grad_norm": 0.6228170394897461, "learning_rate": 9.8440241183532e-05, "loss": 0.031, "step": 37720 }, { "epoch": 24.047163798597833, "grad_norm": 0.40617209672927856, "learning_rate": 9.843887498556783e-05, "loss": 0.0312, "step": 37730 }, { "epoch": 24.053537284894837, "grad_norm": 0.39136525988578796, "learning_rate": 9.84375081990256e-05, "loss": 0.0309, "step": 37740 }, { "epoch": 24.05991077119184, "grad_norm": 0.8294902443885803, "learning_rate": 9.843614082392192e-05, "loss": 0.0283, "step": 37750 }, { "epoch": 24.066284257488846, "grad_norm": 0.640822172164917, "learning_rate": 9.843477286027338e-05, "loss": 0.0438, "step": 37760 }, { "epoch": 24.07265774378585, "grad_norm": 0.9119243621826172, "learning_rate": 9.843340430809663e-05, "loss": 0.0555, "step": 37770 }, { "epoch": 24.079031230082855, "grad_norm": 0.5937010049819946, "learning_rate": 9.843203516740829e-05, "loss": 0.0377, "step": 37780 }, { "epoch": 24.08540471637986, "grad_norm": 0.515120804309845, "learning_rate": 9.8430665438225e-05, "loss": 0.0337, "step": 37790 }, { "epoch": 24.091778202676863, "grad_norm": 0.4904628396034241, "learning_rate": 9.842929512056338e-05, "loss": 0.0342, "step": 37800 }, { "epoch": 24.098151688973868, "grad_norm": 0.18511086702346802, "learning_rate": 9.842792421444011e-05, "loss": 0.0292, "step": 37810 }, { "epoch": 24.104525175270872, "grad_norm": 0.27261674404144287, "learning_rate": 9.842655271987185e-05, "loss": 0.042, "step": 37820 }, { "epoch": 24.110898661567877, "grad_norm": 0.5782015323638916, "learning_rate": 9.842518063687524e-05, "loss": 0.0451, "step": 37830 }, { "epoch": 24.11727214786488, "grad_norm": 0.44219568371772766, "learning_rate": 9.842380796546697e-05, "loss": 0.0304, "step": 37840 }, { "epoch": 24.123645634161885, "grad_norm": 0.5537877082824707, "learning_rate": 9.84224347056637e-05, "loss": 0.0391, "step": 37850 }, { "epoch": 24.13001912045889, "grad_norm": 0.4428641200065613, "learning_rate": 9.842106085748215e-05, "loss": 0.0279, "step": 37860 }, { "epoch": 24.136392606755894, "grad_norm": 0.4276747703552246, "learning_rate": 9.841968642093898e-05, "loss": 0.0341, "step": 37870 }, { "epoch": 24.1427660930529, "grad_norm": 0.3970367908477783, "learning_rate": 9.84183113960509e-05, "loss": 0.0313, "step": 37880 }, { "epoch": 24.149139579349903, "grad_norm": 0.46533265709877014, "learning_rate": 9.841693578283463e-05, "loss": 0.0461, "step": 37890 }, { "epoch": 24.155513065646907, "grad_norm": 0.5321467518806458, "learning_rate": 9.841555958130688e-05, "loss": 0.0318, "step": 37900 }, { "epoch": 24.161886551943912, "grad_norm": 0.43100494146347046, "learning_rate": 9.841418279148435e-05, "loss": 0.0305, "step": 37910 }, { "epoch": 24.168260038240916, "grad_norm": 0.4807056486606598, "learning_rate": 9.841280541338382e-05, "loss": 0.0396, "step": 37920 }, { "epoch": 24.17463352453792, "grad_norm": 0.565575361251831, "learning_rate": 9.841142744702197e-05, "loss": 0.0335, "step": 37930 }, { "epoch": 24.181007010834925, "grad_norm": 0.5753377676010132, "learning_rate": 9.841004889241559e-05, "loss": 0.0412, "step": 37940 }, { "epoch": 24.18738049713193, "grad_norm": 0.4535566568374634, "learning_rate": 9.840866974958138e-05, "loss": 0.0343, "step": 37950 }, { "epoch": 24.193753983428934, "grad_norm": 0.5226954221725464, "learning_rate": 9.840729001853615e-05, "loss": 0.0569, "step": 37960 }, { "epoch": 24.200127469725942, "grad_norm": 0.6045295596122742, "learning_rate": 9.840590969929664e-05, "loss": 0.0363, "step": 37970 }, { "epoch": 24.206500956022946, "grad_norm": 0.8840661644935608, "learning_rate": 9.840452879187962e-05, "loss": 0.0391, "step": 37980 }, { "epoch": 24.21287444231995, "grad_norm": 0.4054742455482483, "learning_rate": 9.840314729630186e-05, "loss": 0.0276, "step": 37990 }, { "epoch": 24.219247928616955, "grad_norm": 0.6850738525390625, "learning_rate": 9.840176521258017e-05, "loss": 0.0281, "step": 38000 }, { "epoch": 24.22562141491396, "grad_norm": 0.33180710673332214, "learning_rate": 9.840038254073134e-05, "loss": 0.0598, "step": 38010 }, { "epoch": 24.231994901210964, "grad_norm": 0.6189227104187012, "learning_rate": 9.839899928077216e-05, "loss": 0.0487, "step": 38020 }, { "epoch": 24.23836838750797, "grad_norm": 0.5323652625083923, "learning_rate": 9.839761543271946e-05, "loss": 0.0317, "step": 38030 }, { "epoch": 24.244741873804973, "grad_norm": 0.3720780313014984, "learning_rate": 9.839623099659002e-05, "loss": 0.0301, "step": 38040 }, { "epoch": 24.251115360101977, "grad_norm": 0.5797668695449829, "learning_rate": 9.839484597240067e-05, "loss": 0.0263, "step": 38050 }, { "epoch": 24.25748884639898, "grad_norm": 0.4359554648399353, "learning_rate": 9.839346036016827e-05, "loss": 0.0432, "step": 38060 }, { "epoch": 24.263862332695986, "grad_norm": 0.4080764949321747, "learning_rate": 9.839207415990963e-05, "loss": 0.0398, "step": 38070 }, { "epoch": 24.27023581899299, "grad_norm": 0.5639428496360779, "learning_rate": 9.83906873716416e-05, "loss": 0.0247, "step": 38080 }, { "epoch": 24.276609305289995, "grad_norm": 0.42230623960494995, "learning_rate": 9.838929999538102e-05, "loss": 0.035, "step": 38090 }, { "epoch": 24.282982791587, "grad_norm": 0.4116472005844116, "learning_rate": 9.838791203114477e-05, "loss": 0.0295, "step": 38100 }, { "epoch": 24.289356277884004, "grad_norm": 0.7337201833724976, "learning_rate": 9.83865234789497e-05, "loss": 0.0425, "step": 38110 }, { "epoch": 24.295729764181008, "grad_norm": 0.638992428779602, "learning_rate": 9.838513433881268e-05, "loss": 0.0399, "step": 38120 }, { "epoch": 24.302103250478012, "grad_norm": 0.7952844500541687, "learning_rate": 9.838374461075059e-05, "loss": 0.0317, "step": 38130 }, { "epoch": 24.308476736775017, "grad_norm": 0.6632501482963562, "learning_rate": 9.838235429478033e-05, "loss": 0.0347, "step": 38140 }, { "epoch": 24.31485022307202, "grad_norm": 0.4192325472831726, "learning_rate": 9.838096339091877e-05, "loss": 0.0491, "step": 38150 }, { "epoch": 24.321223709369026, "grad_norm": 0.4519707262516022, "learning_rate": 9.837957189918285e-05, "loss": 0.0279, "step": 38160 }, { "epoch": 24.32759719566603, "grad_norm": 0.36190348863601685, "learning_rate": 9.837817981958945e-05, "loss": 0.0358, "step": 38170 }, { "epoch": 24.333970681963034, "grad_norm": 0.7802234292030334, "learning_rate": 9.837678715215548e-05, "loss": 0.0391, "step": 38180 }, { "epoch": 24.34034416826004, "grad_norm": 0.32212376594543457, "learning_rate": 9.837539389689787e-05, "loss": 0.0392, "step": 38190 }, { "epoch": 24.346717654557043, "grad_norm": 0.5439888834953308, "learning_rate": 9.837400005383354e-05, "loss": 0.0362, "step": 38200 }, { "epoch": 24.353091140854048, "grad_norm": 0.549466609954834, "learning_rate": 9.837260562297946e-05, "loss": 0.0332, "step": 38210 }, { "epoch": 24.359464627151052, "grad_norm": 1.0089901685714722, "learning_rate": 9.837121060435254e-05, "loss": 0.0347, "step": 38220 }, { "epoch": 24.365838113448056, "grad_norm": 0.3550252914428711, "learning_rate": 9.836981499796974e-05, "loss": 0.0362, "step": 38230 }, { "epoch": 24.37221159974506, "grad_norm": 0.4326470196247101, "learning_rate": 9.836841880384802e-05, "loss": 0.0489, "step": 38240 }, { "epoch": 24.378585086042065, "grad_norm": 0.492059588432312, "learning_rate": 9.836702202200434e-05, "loss": 0.0314, "step": 38250 }, { "epoch": 24.38495857233907, "grad_norm": 0.4955512285232544, "learning_rate": 9.83656246524557e-05, "loss": 0.0298, "step": 38260 }, { "epoch": 24.391332058636074, "grad_norm": 0.7034401893615723, "learning_rate": 9.836422669521902e-05, "loss": 0.0355, "step": 38270 }, { "epoch": 24.39770554493308, "grad_norm": 0.594855785369873, "learning_rate": 9.836282815031134e-05, "loss": 0.0372, "step": 38280 }, { "epoch": 24.404079031230083, "grad_norm": 0.5855913758277893, "learning_rate": 9.836142901774963e-05, "loss": 0.0351, "step": 38290 }, { "epoch": 24.410452517527087, "grad_norm": 0.656150758266449, "learning_rate": 9.83600292975509e-05, "loss": 0.0267, "step": 38300 }, { "epoch": 24.41682600382409, "grad_norm": 0.49173814058303833, "learning_rate": 9.835862898973216e-05, "loss": 0.038, "step": 38310 }, { "epoch": 24.423199490121096, "grad_norm": 0.42426395416259766, "learning_rate": 9.83572280943104e-05, "loss": 0.0385, "step": 38320 }, { "epoch": 24.4295729764181, "grad_norm": 0.3419315814971924, "learning_rate": 9.83558266113027e-05, "loss": 0.0321, "step": 38330 }, { "epoch": 24.435946462715105, "grad_norm": 0.5565381050109863, "learning_rate": 9.8354424540726e-05, "loss": 0.0339, "step": 38340 }, { "epoch": 24.44231994901211, "grad_norm": 0.6314712762832642, "learning_rate": 9.835302188259743e-05, "loss": 0.0409, "step": 38350 }, { "epoch": 24.448693435309114, "grad_norm": 1.8308894634246826, "learning_rate": 9.835161863693398e-05, "loss": 0.0257, "step": 38360 }, { "epoch": 24.455066921606118, "grad_norm": 0.37085020542144775, "learning_rate": 9.83502148037527e-05, "loss": 0.0212, "step": 38370 }, { "epoch": 24.461440407903122, "grad_norm": 0.5847166180610657, "learning_rate": 9.834881038307068e-05, "loss": 0.0398, "step": 38380 }, { "epoch": 24.467813894200127, "grad_norm": 0.5794952511787415, "learning_rate": 9.834740537490495e-05, "loss": 0.0263, "step": 38390 }, { "epoch": 24.47418738049713, "grad_norm": 0.8091599345207214, "learning_rate": 9.83459997792726e-05, "loss": 0.0293, "step": 38400 }, { "epoch": 24.480560866794136, "grad_norm": 0.3204199969768524, "learning_rate": 9.834459359619071e-05, "loss": 0.0371, "step": 38410 }, { "epoch": 24.48693435309114, "grad_norm": 0.6389208436012268, "learning_rate": 9.834318682567635e-05, "loss": 0.0348, "step": 38420 }, { "epoch": 24.493307839388144, "grad_norm": 0.6860884428024292, "learning_rate": 9.834177946774665e-05, "loss": 0.0375, "step": 38430 }, { "epoch": 24.49968132568515, "grad_norm": 0.7466098666191101, "learning_rate": 9.834037152241868e-05, "loss": 0.0319, "step": 38440 }, { "epoch": 24.506054811982153, "grad_norm": 0.6468111276626587, "learning_rate": 9.833896298970955e-05, "loss": 0.0385, "step": 38450 }, { "epoch": 24.512428298279158, "grad_norm": 0.8163811564445496, "learning_rate": 9.833755386963637e-05, "loss": 0.0466, "step": 38460 }, { "epoch": 24.518801784576162, "grad_norm": 0.37384459376335144, "learning_rate": 9.833614416221628e-05, "loss": 0.0332, "step": 38470 }, { "epoch": 24.525175270873167, "grad_norm": 0.5225015878677368, "learning_rate": 9.833473386746641e-05, "loss": 0.037, "step": 38480 }, { "epoch": 24.53154875717017, "grad_norm": 0.872281551361084, "learning_rate": 9.833332298540388e-05, "loss": 0.041, "step": 38490 }, { "epoch": 24.537922243467175, "grad_norm": 0.6988385319709778, "learning_rate": 9.833191151604583e-05, "loss": 0.0462, "step": 38500 }, { "epoch": 24.54429572976418, "grad_norm": 0.42815154790878296, "learning_rate": 9.833049945940945e-05, "loss": 0.0348, "step": 38510 }, { "epoch": 24.550669216061184, "grad_norm": 0.7160035371780396, "learning_rate": 9.832908681551185e-05, "loss": 0.0348, "step": 38520 }, { "epoch": 24.55704270235819, "grad_norm": 0.44611823558807373, "learning_rate": 9.832767358437023e-05, "loss": 0.0396, "step": 38530 }, { "epoch": 24.563416188655193, "grad_norm": 0.5921107530593872, "learning_rate": 9.832625976600173e-05, "loss": 0.0339, "step": 38540 }, { "epoch": 24.569789674952197, "grad_norm": 0.354353129863739, "learning_rate": 9.832484536042355e-05, "loss": 0.0434, "step": 38550 }, { "epoch": 24.5761631612492, "grad_norm": 0.3932753801345825, "learning_rate": 9.832343036765288e-05, "loss": 0.0279, "step": 38560 }, { "epoch": 24.582536647546206, "grad_norm": 0.5793641209602356, "learning_rate": 9.83220147877069e-05, "loss": 0.0477, "step": 38570 }, { "epoch": 24.58891013384321, "grad_norm": 0.5384972095489502, "learning_rate": 9.832059862060282e-05, "loss": 0.0362, "step": 38580 }, { "epoch": 24.595283620140215, "grad_norm": 0.5239253044128418, "learning_rate": 9.831918186635785e-05, "loss": 0.0356, "step": 38590 }, { "epoch": 24.601657106437223, "grad_norm": 0.46745747327804565, "learning_rate": 9.831776452498919e-05, "loss": 0.0338, "step": 38600 }, { "epoch": 24.608030592734227, "grad_norm": 0.3033801317214966, "learning_rate": 9.831634659651407e-05, "loss": 0.0375, "step": 38610 }, { "epoch": 24.61440407903123, "grad_norm": 0.6810902953147888, "learning_rate": 9.831492808094973e-05, "loss": 0.0413, "step": 38620 }, { "epoch": 24.620777565328236, "grad_norm": 0.6167108416557312, "learning_rate": 9.83135089783134e-05, "loss": 0.0382, "step": 38630 }, { "epoch": 24.62715105162524, "grad_norm": 0.7084755301475525, "learning_rate": 9.83120892886223e-05, "loss": 0.0391, "step": 38640 }, { "epoch": 24.633524537922245, "grad_norm": 0.5848824977874756, "learning_rate": 9.831066901189372e-05, "loss": 0.0352, "step": 38650 }, { "epoch": 24.63989802421925, "grad_norm": 0.4761250615119934, "learning_rate": 9.830924814814488e-05, "loss": 0.0429, "step": 38660 }, { "epoch": 24.646271510516254, "grad_norm": 0.5063850283622742, "learning_rate": 9.830782669739308e-05, "loss": 0.0299, "step": 38670 }, { "epoch": 24.652644996813258, "grad_norm": 0.708981454372406, "learning_rate": 9.830640465965557e-05, "loss": 0.0325, "step": 38680 }, { "epoch": 24.659018483110263, "grad_norm": 0.4199800193309784, "learning_rate": 9.830498203494964e-05, "loss": 0.0359, "step": 38690 }, { "epoch": 24.665391969407267, "grad_norm": 0.6549193859100342, "learning_rate": 9.830355882329257e-05, "loss": 0.0389, "step": 38700 }, { "epoch": 24.67176545570427, "grad_norm": 0.9048917889595032, "learning_rate": 9.830213502470165e-05, "loss": 0.0456, "step": 38710 }, { "epoch": 24.678138942001276, "grad_norm": 0.47741028666496277, "learning_rate": 9.830071063919418e-05, "loss": 0.0266, "step": 38720 }, { "epoch": 24.68451242829828, "grad_norm": 0.4264077842235565, "learning_rate": 9.829928566678748e-05, "loss": 0.0396, "step": 38730 }, { "epoch": 24.690885914595285, "grad_norm": 0.32953059673309326, "learning_rate": 9.829786010749885e-05, "loss": 0.0326, "step": 38740 }, { "epoch": 24.69725940089229, "grad_norm": 0.3859044015407562, "learning_rate": 9.829643396134563e-05, "loss": 0.0247, "step": 38750 }, { "epoch": 24.703632887189293, "grad_norm": 0.816207766532898, "learning_rate": 9.829500722834513e-05, "loss": 0.0321, "step": 38760 }, { "epoch": 24.710006373486298, "grad_norm": 0.4959228038787842, "learning_rate": 9.829357990851468e-05, "loss": 0.0326, "step": 38770 }, { "epoch": 24.716379859783302, "grad_norm": 0.7457894086837769, "learning_rate": 9.829215200187167e-05, "loss": 0.0385, "step": 38780 }, { "epoch": 24.722753346080307, "grad_norm": 0.27765101194381714, "learning_rate": 9.82907235084334e-05, "loss": 0.0373, "step": 38790 }, { "epoch": 24.72912683237731, "grad_norm": 0.41058456897735596, "learning_rate": 9.828929442821725e-05, "loss": 0.0304, "step": 38800 }, { "epoch": 24.735500318674315, "grad_norm": 0.688689649105072, "learning_rate": 9.828786476124056e-05, "loss": 0.0329, "step": 38810 }, { "epoch": 24.74187380497132, "grad_norm": 0.5703933238983154, "learning_rate": 9.828643450752076e-05, "loss": 0.0408, "step": 38820 }, { "epoch": 24.748247291268324, "grad_norm": 0.48635345697402954, "learning_rate": 9.828500366707516e-05, "loss": 0.0401, "step": 38830 }, { "epoch": 24.75462077756533, "grad_norm": 0.6730559468269348, "learning_rate": 9.828357223992119e-05, "loss": 0.0396, "step": 38840 }, { "epoch": 24.760994263862333, "grad_norm": 0.4870309829711914, "learning_rate": 9.828214022607623e-05, "loss": 0.0399, "step": 38850 }, { "epoch": 24.767367750159337, "grad_norm": 1.1711093187332153, "learning_rate": 9.828070762555767e-05, "loss": 0.0524, "step": 38860 }, { "epoch": 24.773741236456342, "grad_norm": 0.3556297719478607, "learning_rate": 9.827927443838293e-05, "loss": 0.0215, "step": 38870 }, { "epoch": 24.780114722753346, "grad_norm": 0.40977638959884644, "learning_rate": 9.827784066456943e-05, "loss": 0.0302, "step": 38880 }, { "epoch": 24.78648820905035, "grad_norm": 0.42178672552108765, "learning_rate": 9.827640630413459e-05, "loss": 0.0387, "step": 38890 }, { "epoch": 24.792861695347355, "grad_norm": 0.4901495575904846, "learning_rate": 9.82749713570958e-05, "loss": 0.032, "step": 38900 }, { "epoch": 24.79923518164436, "grad_norm": 0.45226696133613586, "learning_rate": 9.827353582347057e-05, "loss": 0.0326, "step": 38910 }, { "epoch": 24.805608667941364, "grad_norm": 0.43765273690223694, "learning_rate": 9.827209970327629e-05, "loss": 0.0379, "step": 38920 }, { "epoch": 24.81198215423837, "grad_norm": 0.43268218636512756, "learning_rate": 9.82706629965304e-05, "loss": 0.0361, "step": 38930 }, { "epoch": 24.818355640535373, "grad_norm": 0.5586946606636047, "learning_rate": 9.82692257032504e-05, "loss": 0.0289, "step": 38940 }, { "epoch": 24.824729126832377, "grad_norm": 0.8792363405227661, "learning_rate": 9.826778782345374e-05, "loss": 0.0433, "step": 38950 }, { "epoch": 24.83110261312938, "grad_norm": 0.4022250175476074, "learning_rate": 9.826634935715788e-05, "loss": 0.0266, "step": 38960 }, { "epoch": 24.837476099426386, "grad_norm": 0.4029271900653839, "learning_rate": 9.82649103043803e-05, "loss": 0.0241, "step": 38970 }, { "epoch": 24.84384958572339, "grad_norm": 0.547328531742096, "learning_rate": 9.826347066513847e-05, "loss": 0.0329, "step": 38980 }, { "epoch": 24.850223072020395, "grad_norm": 0.480762243270874, "learning_rate": 9.826203043944994e-05, "loss": 0.0415, "step": 38990 }, { "epoch": 24.8565965583174, "grad_norm": 0.5345247387886047, "learning_rate": 9.826058962733216e-05, "loss": 0.036, "step": 39000 }, { "epoch": 24.862970044614404, "grad_norm": 0.6169059872627258, "learning_rate": 9.825914822880264e-05, "loss": 0.025, "step": 39010 }, { "epoch": 24.869343530911408, "grad_norm": 0.2947370111942291, "learning_rate": 9.825770624387891e-05, "loss": 0.0311, "step": 39020 }, { "epoch": 24.875717017208412, "grad_norm": 0.5213439464569092, "learning_rate": 9.82562636725785e-05, "loss": 0.0341, "step": 39030 }, { "epoch": 24.882090503505417, "grad_norm": 0.3553881049156189, "learning_rate": 9.82548205149189e-05, "loss": 0.0281, "step": 39040 }, { "epoch": 24.88846398980242, "grad_norm": 0.480663537979126, "learning_rate": 9.825337677091768e-05, "loss": 0.0327, "step": 39050 }, { "epoch": 24.894837476099426, "grad_norm": 0.7088951468467712, "learning_rate": 9.825193244059238e-05, "loss": 0.0461, "step": 39060 }, { "epoch": 24.90121096239643, "grad_norm": 0.5639773011207581, "learning_rate": 9.825048752396055e-05, "loss": 0.0312, "step": 39070 }, { "epoch": 24.907584448693434, "grad_norm": 0.6111724376678467, "learning_rate": 9.824904202103972e-05, "loss": 0.0296, "step": 39080 }, { "epoch": 24.91395793499044, "grad_norm": 0.876187801361084, "learning_rate": 9.824759593184749e-05, "loss": 0.029, "step": 39090 }, { "epoch": 24.920331421287443, "grad_norm": 0.4184933304786682, "learning_rate": 9.824614925640141e-05, "loss": 0.0339, "step": 39100 }, { "epoch": 24.926704907584448, "grad_norm": 0.3652113974094391, "learning_rate": 9.824470199471907e-05, "loss": 0.0328, "step": 39110 }, { "epoch": 24.933078393881452, "grad_norm": 0.3299172818660736, "learning_rate": 9.824325414681805e-05, "loss": 0.0352, "step": 39120 }, { "epoch": 24.939451880178456, "grad_norm": 0.608484148979187, "learning_rate": 9.824180571271594e-05, "loss": 0.0234, "step": 39130 }, { "epoch": 24.94582536647546, "grad_norm": 0.40828120708465576, "learning_rate": 9.824035669243035e-05, "loss": 0.03, "step": 39140 }, { "epoch": 24.952198852772465, "grad_norm": 0.6209337711334229, "learning_rate": 9.823890708597887e-05, "loss": 0.0285, "step": 39150 }, { "epoch": 24.95857233906947, "grad_norm": 0.6148239374160767, "learning_rate": 9.823745689337912e-05, "loss": 0.0516, "step": 39160 }, { "epoch": 24.964945825366474, "grad_norm": 0.5113382935523987, "learning_rate": 9.823600611464874e-05, "loss": 0.0316, "step": 39170 }, { "epoch": 24.97131931166348, "grad_norm": 0.8183391690254211, "learning_rate": 9.823455474980532e-05, "loss": 0.0265, "step": 39180 }, { "epoch": 24.977692797960483, "grad_norm": 0.43828994035720825, "learning_rate": 9.823310279886654e-05, "loss": 0.0259, "step": 39190 }, { "epoch": 24.984066284257487, "grad_norm": 0.49602243304252625, "learning_rate": 9.823165026185001e-05, "loss": 0.0326, "step": 39200 }, { "epoch": 24.99043977055449, "grad_norm": 0.5951496362686157, "learning_rate": 9.823019713877339e-05, "loss": 0.0243, "step": 39210 }, { "epoch": 24.996813256851496, "grad_norm": 0.6928075551986694, "learning_rate": 9.822874342965435e-05, "loss": 0.0402, "step": 39220 }, { "epoch": 25.003186743148504, "grad_norm": 0.623993456363678, "learning_rate": 9.822728913451053e-05, "loss": 0.032, "step": 39230 }, { "epoch": 25.00956022944551, "grad_norm": 0.5616344809532166, "learning_rate": 9.822583425335963e-05, "loss": 0.0401, "step": 39240 }, { "epoch": 25.015933715742513, "grad_norm": 0.45921823382377625, "learning_rate": 9.822437878621931e-05, "loss": 0.0328, "step": 39250 }, { "epoch": 25.022307202039517, "grad_norm": 0.7087714076042175, "learning_rate": 9.822292273310723e-05, "loss": 0.0261, "step": 39260 }, { "epoch": 25.02868068833652, "grad_norm": 0.5531269907951355, "learning_rate": 9.822146609404112e-05, "loss": 0.0353, "step": 39270 }, { "epoch": 25.035054174633526, "grad_norm": 0.7592794895172119, "learning_rate": 9.822000886903869e-05, "loss": 0.0349, "step": 39280 }, { "epoch": 25.04142766093053, "grad_norm": 0.38011661171913147, "learning_rate": 9.821855105811761e-05, "loss": 0.0387, "step": 39290 }, { "epoch": 25.047801147227535, "grad_norm": 0.6682592034339905, "learning_rate": 9.82170926612956e-05, "loss": 0.0573, "step": 39300 }, { "epoch": 25.05417463352454, "grad_norm": 0.6371383666992188, "learning_rate": 9.82156336785904e-05, "loss": 0.037, "step": 39310 }, { "epoch": 25.060548119821544, "grad_norm": 0.3597332835197449, "learning_rate": 9.821417411001971e-05, "loss": 0.0342, "step": 39320 }, { "epoch": 25.066921606118548, "grad_norm": 0.43058523535728455, "learning_rate": 9.82127139556013e-05, "loss": 0.0307, "step": 39330 }, { "epoch": 25.073295092415552, "grad_norm": 0.5836226344108582, "learning_rate": 9.82112532153529e-05, "loss": 0.0313, "step": 39340 }, { "epoch": 25.079668578712557, "grad_norm": 0.6565682888031006, "learning_rate": 9.820979188929225e-05, "loss": 0.0359, "step": 39350 }, { "epoch": 25.08604206500956, "grad_norm": 0.6128043532371521, "learning_rate": 9.82083299774371e-05, "loss": 0.0452, "step": 39360 }, { "epoch": 25.092415551306566, "grad_norm": 0.30971190333366394, "learning_rate": 9.820686747980525e-05, "loss": 0.0295, "step": 39370 }, { "epoch": 25.09878903760357, "grad_norm": 0.4215708374977112, "learning_rate": 9.820540439641441e-05, "loss": 0.0372, "step": 39380 }, { "epoch": 25.105162523900574, "grad_norm": 0.3842742145061493, "learning_rate": 9.820394072728243e-05, "loss": 0.0351, "step": 39390 }, { "epoch": 25.11153601019758, "grad_norm": 0.49108433723449707, "learning_rate": 9.820247647242703e-05, "loss": 0.0406, "step": 39400 }, { "epoch": 25.117909496494583, "grad_norm": 0.3654775023460388, "learning_rate": 9.820101163186603e-05, "loss": 0.0495, "step": 39410 }, { "epoch": 25.124282982791588, "grad_norm": 0.36959317326545715, "learning_rate": 9.819954620561725e-05, "loss": 0.0426, "step": 39420 }, { "epoch": 25.130656469088592, "grad_norm": 0.6964523792266846, "learning_rate": 9.819808019369847e-05, "loss": 0.0324, "step": 39430 }, { "epoch": 25.137029955385596, "grad_norm": 1.0358121395111084, "learning_rate": 9.81966135961275e-05, "loss": 0.0367, "step": 39440 }, { "epoch": 25.1434034416826, "grad_norm": 0.42308324575424194, "learning_rate": 9.819514641292217e-05, "loss": 0.0322, "step": 39450 }, { "epoch": 25.149776927979605, "grad_norm": 0.45675790309906006, "learning_rate": 9.819367864410032e-05, "loss": 0.0303, "step": 39460 }, { "epoch": 25.15615041427661, "grad_norm": 0.3328995406627655, "learning_rate": 9.819221028967975e-05, "loss": 0.0298, "step": 39470 }, { "epoch": 25.162523900573614, "grad_norm": 0.327166348695755, "learning_rate": 9.819074134967834e-05, "loss": 0.0279, "step": 39480 }, { "epoch": 25.16889738687062, "grad_norm": 0.8217535614967346, "learning_rate": 9.818927182411392e-05, "loss": 0.0379, "step": 39490 }, { "epoch": 25.175270873167623, "grad_norm": 0.4297102093696594, "learning_rate": 9.818780171300435e-05, "loss": 0.0411, "step": 39500 }, { "epoch": 25.181644359464627, "grad_norm": 0.34240713715553284, "learning_rate": 9.81863310163675e-05, "loss": 0.026, "step": 39510 }, { "epoch": 25.18801784576163, "grad_norm": 0.8828849196434021, "learning_rate": 9.818485973422121e-05, "loss": 0.0266, "step": 39520 }, { "epoch": 25.194391332058636, "grad_norm": 1.1191885471343994, "learning_rate": 9.818338786658339e-05, "loss": 0.0377, "step": 39530 }, { "epoch": 25.20076481835564, "grad_norm": 0.38518843054771423, "learning_rate": 9.818191541347193e-05, "loss": 0.0287, "step": 39540 }, { "epoch": 25.207138304652645, "grad_norm": 0.5294539928436279, "learning_rate": 9.818044237490469e-05, "loss": 0.0395, "step": 39550 }, { "epoch": 25.21351179094965, "grad_norm": 0.6108579039573669, "learning_rate": 9.817896875089959e-05, "loss": 0.0331, "step": 39560 }, { "epoch": 25.219885277246654, "grad_norm": 0.43169453740119934, "learning_rate": 9.817749454147454e-05, "loss": 0.0444, "step": 39570 }, { "epoch": 25.226258763543658, "grad_norm": 0.40731561183929443, "learning_rate": 9.817601974664744e-05, "loss": 0.0221, "step": 39580 }, { "epoch": 25.232632249840663, "grad_norm": 0.6495663523674011, "learning_rate": 9.81745443664362e-05, "loss": 0.042, "step": 39590 }, { "epoch": 25.239005736137667, "grad_norm": 0.7028456926345825, "learning_rate": 9.817306840085877e-05, "loss": 0.0406, "step": 39600 }, { "epoch": 25.24537922243467, "grad_norm": 0.2752927839756012, "learning_rate": 9.817159184993308e-05, "loss": 0.0267, "step": 39610 }, { "epoch": 25.251752708731676, "grad_norm": 0.4462147653102875, "learning_rate": 9.817011471367706e-05, "loss": 0.0324, "step": 39620 }, { "epoch": 25.25812619502868, "grad_norm": 0.46049419045448303, "learning_rate": 9.816863699210868e-05, "loss": 0.0281, "step": 39630 }, { "epoch": 25.264499681325685, "grad_norm": 0.3987809717655182, "learning_rate": 9.816715868524586e-05, "loss": 0.0369, "step": 39640 }, { "epoch": 25.27087316762269, "grad_norm": 2.0677452087402344, "learning_rate": 9.81656797931066e-05, "loss": 0.0446, "step": 39650 }, { "epoch": 25.277246653919693, "grad_norm": 0.4305398762226105, "learning_rate": 9.816420031570885e-05, "loss": 0.0489, "step": 39660 }, { "epoch": 25.283620140216698, "grad_norm": 0.5744912028312683, "learning_rate": 9.81627202530706e-05, "loss": 0.0243, "step": 39670 }, { "epoch": 25.289993626513702, "grad_norm": 0.6020693778991699, "learning_rate": 9.816123960520982e-05, "loss": 0.031, "step": 39680 }, { "epoch": 25.296367112810707, "grad_norm": 0.6658399701118469, "learning_rate": 9.815975837214451e-05, "loss": 0.0369, "step": 39690 }, { "epoch": 25.30274059910771, "grad_norm": 0.2149060070514679, "learning_rate": 9.815827655389267e-05, "loss": 0.0299, "step": 39700 }, { "epoch": 25.309114085404715, "grad_norm": 0.45360514521598816, "learning_rate": 9.815679415047228e-05, "loss": 0.0292, "step": 39710 }, { "epoch": 25.31548757170172, "grad_norm": 0.49588531255722046, "learning_rate": 9.815531116190139e-05, "loss": 0.0241, "step": 39720 }, { "epoch": 25.321861057998724, "grad_norm": 0.42829108238220215, "learning_rate": 9.8153827588198e-05, "loss": 0.0271, "step": 39730 }, { "epoch": 25.32823454429573, "grad_norm": 0.4518713653087616, "learning_rate": 9.815234342938014e-05, "loss": 0.055, "step": 39740 }, { "epoch": 25.334608030592733, "grad_norm": 0.8778572678565979, "learning_rate": 9.815085868546585e-05, "loss": 0.0406, "step": 39750 }, { "epoch": 25.340981516889737, "grad_norm": 0.6271068453788757, "learning_rate": 9.814937335647316e-05, "loss": 0.0308, "step": 39760 }, { "epoch": 25.347355003186742, "grad_norm": 0.393430233001709, "learning_rate": 9.814788744242012e-05, "loss": 0.0387, "step": 39770 }, { "epoch": 25.353728489483746, "grad_norm": 0.3414284884929657, "learning_rate": 9.814640094332478e-05, "loss": 0.038, "step": 39780 }, { "epoch": 25.36010197578075, "grad_norm": 0.29574844241142273, "learning_rate": 9.814491385920523e-05, "loss": 0.0365, "step": 39790 }, { "epoch": 25.366475462077755, "grad_norm": 0.3667440414428711, "learning_rate": 9.814342619007951e-05, "loss": 0.0315, "step": 39800 }, { "epoch": 25.37284894837476, "grad_norm": 0.742486834526062, "learning_rate": 9.81419379359657e-05, "loss": 0.0559, "step": 39810 }, { "epoch": 25.379222434671764, "grad_norm": 0.4952855706214905, "learning_rate": 9.814044909688191e-05, "loss": 0.0247, "step": 39820 }, { "epoch": 25.38559592096877, "grad_norm": 0.9497376680374146, "learning_rate": 9.813895967284621e-05, "loss": 0.0397, "step": 39830 }, { "epoch": 25.391969407265773, "grad_norm": 0.46765750646591187, "learning_rate": 9.813746966387669e-05, "loss": 0.0268, "step": 39840 }, { "epoch": 25.398342893562777, "grad_norm": 0.9075254797935486, "learning_rate": 9.813597906999146e-05, "loss": 0.0375, "step": 39850 }, { "epoch": 25.404716379859785, "grad_norm": 0.5852899551391602, "learning_rate": 9.813448789120864e-05, "loss": 0.0295, "step": 39860 }, { "epoch": 25.41108986615679, "grad_norm": 0.5586073994636536, "learning_rate": 9.813299612754634e-05, "loss": 0.0352, "step": 39870 }, { "epoch": 25.417463352453794, "grad_norm": 0.4361903965473175, "learning_rate": 9.81315037790227e-05, "loss": 0.0384, "step": 39880 }, { "epoch": 25.4238368387508, "grad_norm": 0.3150855600833893, "learning_rate": 9.813001084565585e-05, "loss": 0.0374, "step": 39890 }, { "epoch": 25.430210325047803, "grad_norm": 0.41138219833374023, "learning_rate": 9.812851732746391e-05, "loss": 0.0354, "step": 39900 }, { "epoch": 25.436583811344807, "grad_norm": 0.4973923861980438, "learning_rate": 9.812702322446505e-05, "loss": 0.0276, "step": 39910 }, { "epoch": 25.44295729764181, "grad_norm": 0.6138496398925781, "learning_rate": 9.812552853667742e-05, "loss": 0.042, "step": 39920 }, { "epoch": 25.449330783938816, "grad_norm": 0.5973079204559326, "learning_rate": 9.812403326411919e-05, "loss": 0.0339, "step": 39930 }, { "epoch": 25.45570427023582, "grad_norm": 0.43267983198165894, "learning_rate": 9.812253740680851e-05, "loss": 0.0351, "step": 39940 }, { "epoch": 25.462077756532825, "grad_norm": 0.4871577322483063, "learning_rate": 9.812104096476358e-05, "loss": 0.0325, "step": 39950 }, { "epoch": 25.46845124282983, "grad_norm": 0.5238043665885925, "learning_rate": 9.811954393800255e-05, "loss": 0.0271, "step": 39960 }, { "epoch": 25.474824729126833, "grad_norm": 0.9439226388931274, "learning_rate": 9.811804632654362e-05, "loss": 0.0315, "step": 39970 }, { "epoch": 25.481198215423838, "grad_norm": 0.22500838339328766, "learning_rate": 9.811654813040502e-05, "loss": 0.0346, "step": 39980 }, { "epoch": 25.487571701720842, "grad_norm": 0.5708286166191101, "learning_rate": 9.811504934960492e-05, "loss": 0.0473, "step": 39990 }, { "epoch": 25.493945188017847, "grad_norm": 0.7597689032554626, "learning_rate": 9.811354998416154e-05, "loss": 0.0318, "step": 40000 }, { "epoch": 25.50031867431485, "grad_norm": 0.5784670114517212, "learning_rate": 9.81120500340931e-05, "loss": 0.0394, "step": 40010 }, { "epoch": 25.506692160611856, "grad_norm": 0.5145370364189148, "learning_rate": 9.811054949941783e-05, "loss": 0.0412, "step": 40020 }, { "epoch": 25.51306564690886, "grad_norm": 0.281352698802948, "learning_rate": 9.810904838015397e-05, "loss": 0.0318, "step": 40030 }, { "epoch": 25.519439133205864, "grad_norm": 0.6966325044631958, "learning_rate": 9.810754667631973e-05, "loss": 0.0465, "step": 40040 }, { "epoch": 25.52581261950287, "grad_norm": 0.5496698617935181, "learning_rate": 9.810604438793339e-05, "loss": 0.0335, "step": 40050 }, { "epoch": 25.532186105799873, "grad_norm": 0.6323529481887817, "learning_rate": 9.810454151501319e-05, "loss": 0.0314, "step": 40060 }, { "epoch": 25.538559592096878, "grad_norm": 0.5735466480255127, "learning_rate": 9.810303805757739e-05, "loss": 0.0302, "step": 40070 }, { "epoch": 25.544933078393882, "grad_norm": 0.7447389364242554, "learning_rate": 9.810153401564426e-05, "loss": 0.0485, "step": 40080 }, { "epoch": 25.551306564690886, "grad_norm": 1.0027998685836792, "learning_rate": 9.810002938923208e-05, "loss": 0.0255, "step": 40090 }, { "epoch": 25.55768005098789, "grad_norm": 0.48476648330688477, "learning_rate": 9.809852417835912e-05, "loss": 0.0409, "step": 40100 }, { "epoch": 25.564053537284895, "grad_norm": 0.8056094646453857, "learning_rate": 9.809701838304368e-05, "loss": 0.0342, "step": 40110 }, { "epoch": 25.5704270235819, "grad_norm": 0.28897541761398315, "learning_rate": 9.809551200330407e-05, "loss": 0.0254, "step": 40120 }, { "epoch": 25.576800509878904, "grad_norm": 0.5787471532821655, "learning_rate": 9.809400503915857e-05, "loss": 0.028, "step": 40130 }, { "epoch": 25.58317399617591, "grad_norm": 0.304202139377594, "learning_rate": 9.80924974906255e-05, "loss": 0.0339, "step": 40140 }, { "epoch": 25.589547482472913, "grad_norm": 0.907636284828186, "learning_rate": 9.809098935772318e-05, "loss": 0.0487, "step": 40150 }, { "epoch": 25.595920968769917, "grad_norm": 0.7451756000518799, "learning_rate": 9.808948064046993e-05, "loss": 0.0377, "step": 40160 }, { "epoch": 25.60229445506692, "grad_norm": 1.0392460823059082, "learning_rate": 9.80879713388841e-05, "loss": 0.045, "step": 40170 }, { "epoch": 25.608667941363926, "grad_norm": 0.9809366464614868, "learning_rate": 9.808646145298399e-05, "loss": 0.0348, "step": 40180 }, { "epoch": 25.61504142766093, "grad_norm": 0.7227786183357239, "learning_rate": 9.808495098278799e-05, "loss": 0.0391, "step": 40190 }, { "epoch": 25.621414913957935, "grad_norm": 0.33519279956817627, "learning_rate": 9.808343992831443e-05, "loss": 0.034, "step": 40200 }, { "epoch": 25.62778840025494, "grad_norm": 0.4900830388069153, "learning_rate": 9.808192828958169e-05, "loss": 0.0302, "step": 40210 }, { "epoch": 25.634161886551944, "grad_norm": 0.7125679850578308, "learning_rate": 9.808041606660811e-05, "loss": 0.0401, "step": 40220 }, { "epoch": 25.640535372848948, "grad_norm": 0.7391586899757385, "learning_rate": 9.80789032594121e-05, "loss": 0.0428, "step": 40230 }, { "epoch": 25.646908859145952, "grad_norm": 0.5739883184432983, "learning_rate": 9.807738986801198e-05, "loss": 0.0356, "step": 40240 }, { "epoch": 25.653282345442957, "grad_norm": 0.6630788445472717, "learning_rate": 9.807587589242621e-05, "loss": 0.0325, "step": 40250 }, { "epoch": 25.65965583173996, "grad_norm": 0.3680783808231354, "learning_rate": 9.807436133267316e-05, "loss": 0.0279, "step": 40260 }, { "epoch": 25.666029318036966, "grad_norm": 0.44966721534729004, "learning_rate": 9.807284618877121e-05, "loss": 0.0395, "step": 40270 }, { "epoch": 25.67240280433397, "grad_norm": 0.6302251815795898, "learning_rate": 9.807133046073881e-05, "loss": 0.0403, "step": 40280 }, { "epoch": 25.678776290630974, "grad_norm": 0.29586124420166016, "learning_rate": 9.806981414859435e-05, "loss": 0.0262, "step": 40290 }, { "epoch": 25.68514977692798, "grad_norm": 0.3809783160686493, "learning_rate": 9.806829725235627e-05, "loss": 0.0357, "step": 40300 }, { "epoch": 25.691523263224983, "grad_norm": 0.746821939945221, "learning_rate": 9.806677977204298e-05, "loss": 0.028, "step": 40310 }, { "epoch": 25.697896749521988, "grad_norm": 0.7282025814056396, "learning_rate": 9.806526170767294e-05, "loss": 0.0438, "step": 40320 }, { "epoch": 25.704270235818992, "grad_norm": 0.31228116154670715, "learning_rate": 9.80637430592646e-05, "loss": 0.0218, "step": 40330 }, { "epoch": 25.710643722115996, "grad_norm": 0.46417829394340515, "learning_rate": 9.806222382683639e-05, "loss": 0.0295, "step": 40340 }, { "epoch": 25.717017208413, "grad_norm": 0.6913856863975525, "learning_rate": 9.806070401040678e-05, "loss": 0.0374, "step": 40350 }, { "epoch": 25.723390694710005, "grad_norm": 0.4596313238143921, "learning_rate": 9.805918360999424e-05, "loss": 0.0257, "step": 40360 }, { "epoch": 25.72976418100701, "grad_norm": 0.45668748021125793, "learning_rate": 9.805766262561726e-05, "loss": 0.0306, "step": 40370 }, { "epoch": 25.736137667304014, "grad_norm": 0.48825353384017944, "learning_rate": 9.805614105729429e-05, "loss": 0.0334, "step": 40380 }, { "epoch": 25.74251115360102, "grad_norm": 0.8489535450935364, "learning_rate": 9.805461890504384e-05, "loss": 0.0371, "step": 40390 }, { "epoch": 25.748884639898023, "grad_norm": 0.6208127737045288, "learning_rate": 9.805309616888438e-05, "loss": 0.0256, "step": 40400 }, { "epoch": 25.755258126195027, "grad_norm": 0.41087955236434937, "learning_rate": 9.805157284883446e-05, "loss": 0.0368, "step": 40410 }, { "epoch": 25.76163161249203, "grad_norm": 0.42125561833381653, "learning_rate": 9.805004894491255e-05, "loss": 0.0269, "step": 40420 }, { "epoch": 25.768005098789036, "grad_norm": 0.3624279797077179, "learning_rate": 9.804852445713716e-05, "loss": 0.0455, "step": 40430 }, { "epoch": 25.77437858508604, "grad_norm": 0.3836914002895355, "learning_rate": 9.804699938552687e-05, "loss": 0.0386, "step": 40440 }, { "epoch": 25.780752071383045, "grad_norm": 0.3899528682231903, "learning_rate": 9.804547373010015e-05, "loss": 0.0281, "step": 40450 }, { "epoch": 25.78712555768005, "grad_norm": 0.3634029030799866, "learning_rate": 9.804394749087555e-05, "loss": 0.032, "step": 40460 }, { "epoch": 25.793499043977054, "grad_norm": 0.6043806076049805, "learning_rate": 9.804242066787167e-05, "loss": 0.0365, "step": 40470 }, { "epoch": 25.799872530274058, "grad_norm": 0.3717891275882721, "learning_rate": 9.804089326110697e-05, "loss": 0.0289, "step": 40480 }, { "epoch": 25.806246016571066, "grad_norm": 0.5614397525787354, "learning_rate": 9.803936527060009e-05, "loss": 0.0442, "step": 40490 }, { "epoch": 25.81261950286807, "grad_norm": 0.46124744415283203, "learning_rate": 9.803783669636955e-05, "loss": 0.0256, "step": 40500 }, { "epoch": 25.818992989165075, "grad_norm": 0.5526819229125977, "learning_rate": 9.803630753843395e-05, "loss": 0.0291, "step": 40510 }, { "epoch": 25.82536647546208, "grad_norm": 0.50849848985672, "learning_rate": 9.803477779681185e-05, "loss": 0.0357, "step": 40520 }, { "epoch": 25.831739961759084, "grad_norm": 0.7440577745437622, "learning_rate": 9.803324747152185e-05, "loss": 0.0316, "step": 40530 }, { "epoch": 25.838113448056088, "grad_norm": 0.5509979724884033, "learning_rate": 9.803171656258254e-05, "loss": 0.0315, "step": 40540 }, { "epoch": 25.844486934353093, "grad_norm": 0.5559027194976807, "learning_rate": 9.803018507001253e-05, "loss": 0.0323, "step": 40550 }, { "epoch": 25.850860420650097, "grad_norm": 0.5763866901397705, "learning_rate": 9.802865299383043e-05, "loss": 0.0373, "step": 40560 }, { "epoch": 25.8572339069471, "grad_norm": 0.4962155222892761, "learning_rate": 9.802712033405483e-05, "loss": 0.0365, "step": 40570 }, { "epoch": 25.863607393244106, "grad_norm": 0.44195908308029175, "learning_rate": 9.802558709070439e-05, "loss": 0.0275, "step": 40580 }, { "epoch": 25.86998087954111, "grad_norm": 0.5724771022796631, "learning_rate": 9.802405326379772e-05, "loss": 0.0333, "step": 40590 }, { "epoch": 25.876354365838115, "grad_norm": 0.42344096302986145, "learning_rate": 9.802251885335345e-05, "loss": 0.0264, "step": 40600 }, { "epoch": 25.88272785213512, "grad_norm": 0.5523172616958618, "learning_rate": 9.802098385939025e-05, "loss": 0.0309, "step": 40610 }, { "epoch": 25.889101338432123, "grad_norm": 0.43028560280799866, "learning_rate": 9.801944828192675e-05, "loss": 0.0317, "step": 40620 }, { "epoch": 25.895474824729128, "grad_norm": 0.4124147891998291, "learning_rate": 9.801791212098162e-05, "loss": 0.0296, "step": 40630 }, { "epoch": 25.901848311026132, "grad_norm": 0.41405194997787476, "learning_rate": 9.801637537657351e-05, "loss": 0.0284, "step": 40640 }, { "epoch": 25.908221797323137, "grad_norm": 0.5136837959289551, "learning_rate": 9.801483804872112e-05, "loss": 0.0324, "step": 40650 }, { "epoch": 25.91459528362014, "grad_norm": 0.30281969904899597, "learning_rate": 9.80133001374431e-05, "loss": 0.0262, "step": 40660 }, { "epoch": 25.920968769917145, "grad_norm": 0.4209826588630676, "learning_rate": 9.801176164275816e-05, "loss": 0.0487, "step": 40670 }, { "epoch": 25.92734225621415, "grad_norm": 0.3522534668445587, "learning_rate": 9.801022256468498e-05, "loss": 0.0382, "step": 40680 }, { "epoch": 25.933715742511154, "grad_norm": 0.6347562074661255, "learning_rate": 9.800868290324227e-05, "loss": 0.0238, "step": 40690 }, { "epoch": 25.94008922880816, "grad_norm": 0.3325772285461426, "learning_rate": 9.800714265844872e-05, "loss": 0.0377, "step": 40700 }, { "epoch": 25.946462715105163, "grad_norm": 0.4902306795120239, "learning_rate": 9.800560183032308e-05, "loss": 0.0263, "step": 40710 }, { "epoch": 25.952836201402167, "grad_norm": 0.2938218116760254, "learning_rate": 9.800406041888405e-05, "loss": 0.0252, "step": 40720 }, { "epoch": 25.959209687699172, "grad_norm": 0.5713808536529541, "learning_rate": 9.800251842415036e-05, "loss": 0.0402, "step": 40730 }, { "epoch": 25.965583173996176, "grad_norm": 0.5097112655639648, "learning_rate": 9.800097584614075e-05, "loss": 0.037, "step": 40740 }, { "epoch": 25.97195666029318, "grad_norm": 0.617712676525116, "learning_rate": 9.799943268487397e-05, "loss": 0.0323, "step": 40750 }, { "epoch": 25.978330146590185, "grad_norm": 0.5795984864234924, "learning_rate": 9.799788894036875e-05, "loss": 0.0393, "step": 40760 }, { "epoch": 25.98470363288719, "grad_norm": 0.5161428451538086, "learning_rate": 9.799634461264387e-05, "loss": 0.0252, "step": 40770 }, { "epoch": 25.991077119184194, "grad_norm": 0.35265636444091797, "learning_rate": 9.799479970171808e-05, "loss": 0.0313, "step": 40780 }, { "epoch": 25.9974506054812, "grad_norm": 0.3866861164569855, "learning_rate": 9.799325420761016e-05, "loss": 0.0318, "step": 40790 }, { "epoch": 26.003824091778203, "grad_norm": 0.8604021668434143, "learning_rate": 9.799170813033889e-05, "loss": 0.0272, "step": 40800 }, { "epoch": 26.010197578075207, "grad_norm": 0.47888457775115967, "learning_rate": 9.799016146992306e-05, "loss": 0.0258, "step": 40810 }, { "epoch": 26.01657106437221, "grad_norm": 0.4438333809375763, "learning_rate": 9.798861422638147e-05, "loss": 0.0368, "step": 40820 }, { "epoch": 26.022944550669216, "grad_norm": 0.80049067735672, "learning_rate": 9.798706639973288e-05, "loss": 0.0319, "step": 40830 }, { "epoch": 26.02931803696622, "grad_norm": 0.685234546661377, "learning_rate": 9.798551798999613e-05, "loss": 0.0245, "step": 40840 }, { "epoch": 26.035691523263225, "grad_norm": 0.6471700072288513, "learning_rate": 9.798396899719005e-05, "loss": 0.0376, "step": 40850 }, { "epoch": 26.04206500956023, "grad_norm": 0.5088632702827454, "learning_rate": 9.798241942133342e-05, "loss": 0.026, "step": 40860 }, { "epoch": 26.048438495857233, "grad_norm": 0.511092483997345, "learning_rate": 9.798086926244511e-05, "loss": 0.0274, "step": 40870 }, { "epoch": 26.054811982154238, "grad_norm": 0.56766676902771, "learning_rate": 9.797931852054393e-05, "loss": 0.0351, "step": 40880 }, { "epoch": 26.061185468451242, "grad_norm": 0.5463147759437561, "learning_rate": 9.797776719564873e-05, "loss": 0.035, "step": 40890 }, { "epoch": 26.067558954748247, "grad_norm": 0.7147695422172546, "learning_rate": 9.797621528777837e-05, "loss": 0.0385, "step": 40900 }, { "epoch": 26.07393244104525, "grad_norm": 0.5646417737007141, "learning_rate": 9.797466279695168e-05, "loss": 0.0325, "step": 40910 }, { "epoch": 26.080305927342256, "grad_norm": 0.3981863260269165, "learning_rate": 9.797310972318755e-05, "loss": 0.0364, "step": 40920 }, { "epoch": 26.08667941363926, "grad_norm": 0.7347411513328552, "learning_rate": 9.797155606650483e-05, "loss": 0.028, "step": 40930 }, { "epoch": 26.093052899936264, "grad_norm": 0.3876502215862274, "learning_rate": 9.797000182692243e-05, "loss": 0.0308, "step": 40940 }, { "epoch": 26.09942638623327, "grad_norm": 1.1200830936431885, "learning_rate": 9.796844700445921e-05, "loss": 0.0278, "step": 40950 }, { "epoch": 26.105799872530273, "grad_norm": 0.3290374279022217, "learning_rate": 9.796689159913406e-05, "loss": 0.0327, "step": 40960 }, { "epoch": 26.112173358827278, "grad_norm": 0.298495888710022, "learning_rate": 9.79653356109659e-05, "loss": 0.0255, "step": 40970 }, { "epoch": 26.118546845124282, "grad_norm": 0.7288366556167603, "learning_rate": 9.796377903997363e-05, "loss": 0.0415, "step": 40980 }, { "epoch": 26.124920331421286, "grad_norm": 0.4295289218425751, "learning_rate": 9.796222188617615e-05, "loss": 0.024, "step": 40990 }, { "epoch": 26.13129381771829, "grad_norm": 0.7116581797599792, "learning_rate": 9.796066414959238e-05, "loss": 0.0243, "step": 41000 }, { "epoch": 26.137667304015295, "grad_norm": 0.5832488536834717, "learning_rate": 9.795910583024129e-05, "loss": 0.0306, "step": 41010 }, { "epoch": 26.1440407903123, "grad_norm": 0.42752605676651, "learning_rate": 9.795754692814176e-05, "loss": 0.0246, "step": 41020 }, { "epoch": 26.150414276609304, "grad_norm": 0.7192235589027405, "learning_rate": 9.795598744331275e-05, "loss": 0.0288, "step": 41030 }, { "epoch": 26.15678776290631, "grad_norm": 0.7591318488121033, "learning_rate": 9.795442737577323e-05, "loss": 0.0432, "step": 41040 }, { "epoch": 26.163161249203313, "grad_norm": 0.5937816500663757, "learning_rate": 9.795286672554213e-05, "loss": 0.0392, "step": 41050 }, { "epoch": 26.169534735500317, "grad_norm": 0.4404557943344116, "learning_rate": 9.795130549263843e-05, "loss": 0.04, "step": 41060 }, { "epoch": 26.17590822179732, "grad_norm": 0.4103528559207916, "learning_rate": 9.794974367708109e-05, "loss": 0.0251, "step": 41070 }, { "epoch": 26.182281708094326, "grad_norm": 0.25227975845336914, "learning_rate": 9.79481812788891e-05, "loss": 0.0335, "step": 41080 }, { "epoch": 26.18865519439133, "grad_norm": 0.456488698720932, "learning_rate": 9.794661829808142e-05, "loss": 0.0239, "step": 41090 }, { "epoch": 26.195028680688335, "grad_norm": 0.48274660110473633, "learning_rate": 9.794505473467708e-05, "loss": 0.0544, "step": 41100 }, { "epoch": 26.20140216698534, "grad_norm": 0.6885577440261841, "learning_rate": 9.794349058869506e-05, "loss": 0.0334, "step": 41110 }, { "epoch": 26.207775653282347, "grad_norm": 0.45659247040748596, "learning_rate": 9.794192586015433e-05, "loss": 0.0383, "step": 41120 }, { "epoch": 26.21414913957935, "grad_norm": 0.672415554523468, "learning_rate": 9.794036054907396e-05, "loss": 0.0481, "step": 41130 }, { "epoch": 26.220522625876356, "grad_norm": 0.5611081123352051, "learning_rate": 9.793879465547296e-05, "loss": 0.0349, "step": 41140 }, { "epoch": 26.22689611217336, "grad_norm": 0.718242347240448, "learning_rate": 9.793722817937032e-05, "loss": 0.0342, "step": 41150 }, { "epoch": 26.233269598470365, "grad_norm": 0.41502007842063904, "learning_rate": 9.793566112078509e-05, "loss": 0.0227, "step": 41160 }, { "epoch": 26.23964308476737, "grad_norm": 0.8946523666381836, "learning_rate": 9.793409347973635e-05, "loss": 0.0387, "step": 41170 }, { "epoch": 26.246016571064374, "grad_norm": 0.9620791077613831, "learning_rate": 9.79325252562431e-05, "loss": 0.042, "step": 41180 }, { "epoch": 26.252390057361378, "grad_norm": 0.7109879851341248, "learning_rate": 9.793095645032441e-05, "loss": 0.0331, "step": 41190 }, { "epoch": 26.258763543658382, "grad_norm": 0.566011905670166, "learning_rate": 9.792938706199934e-05, "loss": 0.034, "step": 41200 }, { "epoch": 26.265137029955387, "grad_norm": 0.4160345792770386, "learning_rate": 9.792781709128698e-05, "loss": 0.023, "step": 41210 }, { "epoch": 26.27151051625239, "grad_norm": 0.2870670258998871, "learning_rate": 9.792624653820638e-05, "loss": 0.0265, "step": 41220 }, { "epoch": 26.277884002549396, "grad_norm": 0.6686944365501404, "learning_rate": 9.792467540277663e-05, "loss": 0.029, "step": 41230 }, { "epoch": 26.2842574888464, "grad_norm": 0.4624597132205963, "learning_rate": 9.792310368501683e-05, "loss": 0.0589, "step": 41240 }, { "epoch": 26.290630975143404, "grad_norm": 0.4988868534564972, "learning_rate": 9.792153138494608e-05, "loss": 0.026, "step": 41250 }, { "epoch": 26.29700446144041, "grad_norm": 0.6210262179374695, "learning_rate": 9.791995850258348e-05, "loss": 0.0369, "step": 41260 }, { "epoch": 26.303377947737413, "grad_norm": 0.3958175480365753, "learning_rate": 9.791838503794812e-05, "loss": 0.0404, "step": 41270 }, { "epoch": 26.309751434034418, "grad_norm": 0.4863801598548889, "learning_rate": 9.791681099105915e-05, "loss": 0.0412, "step": 41280 }, { "epoch": 26.316124920331422, "grad_norm": 0.3420000374317169, "learning_rate": 9.791523636193571e-05, "loss": 0.0239, "step": 41290 }, { "epoch": 26.322498406628426, "grad_norm": 0.4110923111438751, "learning_rate": 9.791366115059689e-05, "loss": 0.0226, "step": 41300 }, { "epoch": 26.32887189292543, "grad_norm": 0.576371967792511, "learning_rate": 9.791208535706184e-05, "loss": 0.0235, "step": 41310 }, { "epoch": 26.335245379222435, "grad_norm": 0.7259501218795776, "learning_rate": 9.791050898134972e-05, "loss": 0.0422, "step": 41320 }, { "epoch": 26.34161886551944, "grad_norm": 0.34121406078338623, "learning_rate": 9.790893202347969e-05, "loss": 0.0297, "step": 41330 }, { "epoch": 26.347992351816444, "grad_norm": 0.6072424054145813, "learning_rate": 9.79073544834709e-05, "loss": 0.0249, "step": 41340 }, { "epoch": 26.35436583811345, "grad_norm": 0.6023910641670227, "learning_rate": 9.79057763613425e-05, "loss": 0.0302, "step": 41350 }, { "epoch": 26.360739324410453, "grad_norm": 0.36167219281196594, "learning_rate": 9.790419765711372e-05, "loss": 0.0386, "step": 41360 }, { "epoch": 26.367112810707457, "grad_norm": 0.6131445169448853, "learning_rate": 9.790261837080368e-05, "loss": 0.0257, "step": 41370 }, { "epoch": 26.37348629700446, "grad_norm": 0.4489663243293762, "learning_rate": 9.790103850243161e-05, "loss": 0.0335, "step": 41380 }, { "epoch": 26.379859783301466, "grad_norm": 0.3466247618198395, "learning_rate": 9.789945805201672e-05, "loss": 0.0286, "step": 41390 }, { "epoch": 26.38623326959847, "grad_norm": 0.39985668659210205, "learning_rate": 9.789787701957815e-05, "loss": 0.0224, "step": 41400 }, { "epoch": 26.392606755895475, "grad_norm": 0.5911824107170105, "learning_rate": 9.789629540513518e-05, "loss": 0.0328, "step": 41410 }, { "epoch": 26.39898024219248, "grad_norm": 0.7254574298858643, "learning_rate": 9.789471320870699e-05, "loss": 0.0239, "step": 41420 }, { "epoch": 26.405353728489484, "grad_norm": 0.5336246490478516, "learning_rate": 9.789313043031281e-05, "loss": 0.029, "step": 41430 }, { "epoch": 26.411727214786488, "grad_norm": 0.4649886190891266, "learning_rate": 9.789154706997187e-05, "loss": 0.031, "step": 41440 }, { "epoch": 26.418100701083493, "grad_norm": 0.4161438047885895, "learning_rate": 9.788996312770344e-05, "loss": 0.0222, "step": 41450 }, { "epoch": 26.424474187380497, "grad_norm": 0.7702319622039795, "learning_rate": 9.788837860352673e-05, "loss": 0.0365, "step": 41460 }, { "epoch": 26.4308476736775, "grad_norm": 0.7023902535438538, "learning_rate": 9.788679349746101e-05, "loss": 0.0351, "step": 41470 }, { "epoch": 26.437221159974506, "grad_norm": 0.6471976637840271, "learning_rate": 9.788520780952553e-05, "loss": 0.0298, "step": 41480 }, { "epoch": 26.44359464627151, "grad_norm": 0.8786109089851379, "learning_rate": 9.788362153973955e-05, "loss": 0.0433, "step": 41490 }, { "epoch": 26.449968132568515, "grad_norm": 0.4884776175022125, "learning_rate": 9.788203468812238e-05, "loss": 0.0334, "step": 41500 }, { "epoch": 26.45634161886552, "grad_norm": 0.39006826281547546, "learning_rate": 9.788044725469327e-05, "loss": 0.0285, "step": 41510 }, { "epoch": 26.462715105162523, "grad_norm": 0.41547033190727234, "learning_rate": 9.787885923947153e-05, "loss": 0.0288, "step": 41520 }, { "epoch": 26.469088591459528, "grad_norm": 0.7170929908752441, "learning_rate": 9.787727064247643e-05, "loss": 0.0426, "step": 41530 }, { "epoch": 26.475462077756532, "grad_norm": 0.4527498185634613, "learning_rate": 9.787568146372729e-05, "loss": 0.0279, "step": 41540 }, { "epoch": 26.481835564053537, "grad_norm": 0.48785221576690674, "learning_rate": 9.787409170324345e-05, "loss": 0.0294, "step": 41550 }, { "epoch": 26.48820905035054, "grad_norm": 0.3759062886238098, "learning_rate": 9.787250136104415e-05, "loss": 0.0301, "step": 41560 }, { "epoch": 26.494582536647545, "grad_norm": 0.37527981400489807, "learning_rate": 9.787091043714879e-05, "loss": 0.0265, "step": 41570 }, { "epoch": 26.50095602294455, "grad_norm": 0.8074070811271667, "learning_rate": 9.786931893157666e-05, "loss": 0.0354, "step": 41580 }, { "epoch": 26.507329509241554, "grad_norm": 0.5236986875534058, "learning_rate": 9.786772684434711e-05, "loss": 0.0447, "step": 41590 }, { "epoch": 26.51370299553856, "grad_norm": 0.4914737641811371, "learning_rate": 9.786613417547947e-05, "loss": 0.034, "step": 41600 }, { "epoch": 26.520076481835563, "grad_norm": 0.3342273533344269, "learning_rate": 9.786454092499314e-05, "loss": 0.0297, "step": 41610 }, { "epoch": 26.526449968132567, "grad_norm": 0.5309076905250549, "learning_rate": 9.786294709290741e-05, "loss": 0.0402, "step": 41620 }, { "epoch": 26.532823454429572, "grad_norm": 0.5091167688369751, "learning_rate": 9.78613526792417e-05, "loss": 0.0336, "step": 41630 }, { "epoch": 26.539196940726576, "grad_norm": 0.4147992432117462, "learning_rate": 9.785975768401537e-05, "loss": 0.0366, "step": 41640 }, { "epoch": 26.54557042702358, "grad_norm": 0.4643586277961731, "learning_rate": 9.785816210724779e-05, "loss": 0.0232, "step": 41650 }, { "epoch": 26.551943913320585, "grad_norm": 0.35512903332710266, "learning_rate": 9.785656594895836e-05, "loss": 0.0326, "step": 41660 }, { "epoch": 26.55831739961759, "grad_norm": 0.502058744430542, "learning_rate": 9.785496920916646e-05, "loss": 0.0333, "step": 41670 }, { "epoch": 26.564690885914594, "grad_norm": 0.44213804602622986, "learning_rate": 9.785337188789151e-05, "loss": 0.0254, "step": 41680 }, { "epoch": 26.5710643722116, "grad_norm": 0.5241914987564087, "learning_rate": 9.78517739851529e-05, "loss": 0.0284, "step": 41690 }, { "epoch": 26.577437858508603, "grad_norm": 0.3741266429424286, "learning_rate": 9.785017550097006e-05, "loss": 0.0314, "step": 41700 }, { "epoch": 26.583811344805607, "grad_norm": 0.36622586846351624, "learning_rate": 9.784857643536243e-05, "loss": 0.0233, "step": 41710 }, { "epoch": 26.59018483110261, "grad_norm": 0.37584641575813293, "learning_rate": 9.78469767883494e-05, "loss": 0.0223, "step": 41720 }, { "epoch": 26.59655831739962, "grad_norm": 0.3701398968696594, "learning_rate": 9.784537655995043e-05, "loss": 0.0353, "step": 41730 }, { "epoch": 26.60293180369662, "grad_norm": 0.49583032727241516, "learning_rate": 9.784377575018496e-05, "loss": 0.0489, "step": 41740 }, { "epoch": 26.609305289993628, "grad_norm": 0.39954572916030884, "learning_rate": 9.784217435907244e-05, "loss": 0.0281, "step": 41750 }, { "epoch": 26.615678776290633, "grad_norm": 0.4347255229949951, "learning_rate": 9.784057238663234e-05, "loss": 0.0387, "step": 41760 }, { "epoch": 26.622052262587637, "grad_norm": 0.46338048577308655, "learning_rate": 9.783896983288412e-05, "loss": 0.0333, "step": 41770 }, { "epoch": 26.62842574888464, "grad_norm": 0.5483663082122803, "learning_rate": 9.783736669784725e-05, "loss": 0.0313, "step": 41780 }, { "epoch": 26.634799235181646, "grad_norm": 0.6989869475364685, "learning_rate": 9.783576298154119e-05, "loss": 0.0338, "step": 41790 }, { "epoch": 26.64117272147865, "grad_norm": 0.45994454622268677, "learning_rate": 9.783415868398545e-05, "loss": 0.0425, "step": 41800 }, { "epoch": 26.647546207775655, "grad_norm": 0.550617516040802, "learning_rate": 9.783255380519953e-05, "loss": 0.035, "step": 41810 }, { "epoch": 26.65391969407266, "grad_norm": 0.2228291928768158, "learning_rate": 9.783094834520291e-05, "loss": 0.0287, "step": 41820 }, { "epoch": 26.660293180369663, "grad_norm": 0.39090222120285034, "learning_rate": 9.782934230401512e-05, "loss": 0.0207, "step": 41830 }, { "epoch": 26.666666666666668, "grad_norm": 0.6888470649719238, "learning_rate": 9.782773568165565e-05, "loss": 0.0395, "step": 41840 }, { "epoch": 26.673040152963672, "grad_norm": 0.47411271929740906, "learning_rate": 9.782612847814404e-05, "loss": 0.032, "step": 41850 }, { "epoch": 26.679413639260677, "grad_norm": 0.5727260708808899, "learning_rate": 9.782452069349981e-05, "loss": 0.0322, "step": 41860 }, { "epoch": 26.68578712555768, "grad_norm": 0.7682517766952515, "learning_rate": 9.78229123277425e-05, "loss": 0.045, "step": 41870 }, { "epoch": 26.692160611854685, "grad_norm": 0.3485850989818573, "learning_rate": 9.782130338089165e-05, "loss": 0.0234, "step": 41880 }, { "epoch": 26.69853409815169, "grad_norm": 0.3398924469947815, "learning_rate": 9.781969385296683e-05, "loss": 0.023, "step": 41890 }, { "epoch": 26.704907584448694, "grad_norm": 0.4345161020755768, "learning_rate": 9.781808374398757e-05, "loss": 0.0243, "step": 41900 }, { "epoch": 26.7112810707457, "grad_norm": 0.36500152945518494, "learning_rate": 9.781647305397343e-05, "loss": 0.0356, "step": 41910 }, { "epoch": 26.717654557042703, "grad_norm": 0.4141809940338135, "learning_rate": 9.781486178294402e-05, "loss": 0.0366, "step": 41920 }, { "epoch": 26.724028043339707, "grad_norm": 0.42421048879623413, "learning_rate": 9.781324993091889e-05, "loss": 0.0219, "step": 41930 }, { "epoch": 26.730401529636712, "grad_norm": 0.28780397772789, "learning_rate": 9.781163749791762e-05, "loss": 0.0309, "step": 41940 }, { "epoch": 26.736775015933716, "grad_norm": 0.7716911435127258, "learning_rate": 9.781002448395982e-05, "loss": 0.0236, "step": 41950 }, { "epoch": 26.74314850223072, "grad_norm": 0.39955684542655945, "learning_rate": 9.780841088906508e-05, "loss": 0.029, "step": 41960 }, { "epoch": 26.749521988527725, "grad_norm": 0.32886803150177, "learning_rate": 9.7806796713253e-05, "loss": 0.0328, "step": 41970 }, { "epoch": 26.75589547482473, "grad_norm": 0.28118640184402466, "learning_rate": 9.780518195654322e-05, "loss": 0.0283, "step": 41980 }, { "epoch": 26.762268961121734, "grad_norm": 0.5915328860282898, "learning_rate": 9.780356661895534e-05, "loss": 0.0289, "step": 41990 }, { "epoch": 26.76864244741874, "grad_norm": 0.6661473512649536, "learning_rate": 9.780195070050898e-05, "loss": 0.0392, "step": 42000 }, { "epoch": 26.775015933715743, "grad_norm": 0.3593858480453491, "learning_rate": 9.78003342012238e-05, "loss": 0.0278, "step": 42010 }, { "epoch": 26.781389420012747, "grad_norm": 0.23564191162586212, "learning_rate": 9.779871712111942e-05, "loss": 0.0262, "step": 42020 }, { "epoch": 26.78776290630975, "grad_norm": 0.4596879482269287, "learning_rate": 9.77970994602155e-05, "loss": 0.033, "step": 42030 }, { "epoch": 26.794136392606756, "grad_norm": 0.4411656856536865, "learning_rate": 9.779548121853169e-05, "loss": 0.0438, "step": 42040 }, { "epoch": 26.80050987890376, "grad_norm": 0.3380880355834961, "learning_rate": 9.779386239608765e-05, "loss": 0.0357, "step": 42050 }, { "epoch": 26.806883365200765, "grad_norm": 0.5125289559364319, "learning_rate": 9.779224299290308e-05, "loss": 0.0506, "step": 42060 }, { "epoch": 26.81325685149777, "grad_norm": 0.46288588643074036, "learning_rate": 9.779062300899762e-05, "loss": 0.0321, "step": 42070 }, { "epoch": 26.819630337794774, "grad_norm": 0.6414213180541992, "learning_rate": 9.778900244439097e-05, "loss": 0.0287, "step": 42080 }, { "epoch": 26.826003824091778, "grad_norm": 0.679557740688324, "learning_rate": 9.778738129910281e-05, "loss": 0.0365, "step": 42090 }, { "epoch": 26.832377310388782, "grad_norm": 0.41695597767829895, "learning_rate": 9.778575957315287e-05, "loss": 0.0358, "step": 42100 }, { "epoch": 26.838750796685787, "grad_norm": 0.484231561422348, "learning_rate": 9.778413726656081e-05, "loss": 0.0435, "step": 42110 }, { "epoch": 26.84512428298279, "grad_norm": 0.31692075729370117, "learning_rate": 9.778251437934637e-05, "loss": 0.0278, "step": 42120 }, { "epoch": 26.851497769279796, "grad_norm": 0.6634456515312195, "learning_rate": 9.778089091152929e-05, "loss": 0.0365, "step": 42130 }, { "epoch": 26.8578712555768, "grad_norm": 0.48582810163497925, "learning_rate": 9.777926686312926e-05, "loss": 0.0286, "step": 42140 }, { "epoch": 26.864244741873804, "grad_norm": 0.6180738210678101, "learning_rate": 9.777764223416602e-05, "loss": 0.0249, "step": 42150 }, { "epoch": 26.87061822817081, "grad_norm": 0.6959978938102722, "learning_rate": 9.777601702465933e-05, "loss": 0.0301, "step": 42160 }, { "epoch": 26.876991714467813, "grad_norm": 0.5285813212394714, "learning_rate": 9.77743912346289e-05, "loss": 0.0289, "step": 42170 }, { "epoch": 26.883365200764818, "grad_norm": 0.5362122058868408, "learning_rate": 9.777276486409454e-05, "loss": 0.0257, "step": 42180 }, { "epoch": 26.889738687061822, "grad_norm": 0.39523330330848694, "learning_rate": 9.777113791307598e-05, "loss": 0.0301, "step": 42190 }, { "epoch": 26.896112173358826, "grad_norm": 0.6104212999343872, "learning_rate": 9.776951038159296e-05, "loss": 0.0375, "step": 42200 }, { "epoch": 26.90248565965583, "grad_norm": 0.5991176962852478, "learning_rate": 9.776788226966531e-05, "loss": 0.0317, "step": 42210 }, { "epoch": 26.908859145952835, "grad_norm": 0.5169550776481628, "learning_rate": 9.776625357731279e-05, "loss": 0.0343, "step": 42220 }, { "epoch": 26.91523263224984, "grad_norm": 0.6992713809013367, "learning_rate": 9.776462430455519e-05, "loss": 0.0322, "step": 42230 }, { "epoch": 26.921606118546844, "grad_norm": 0.44037771224975586, "learning_rate": 9.776299445141231e-05, "loss": 0.0311, "step": 42240 }, { "epoch": 26.92797960484385, "grad_norm": 0.6163039803504944, "learning_rate": 9.776136401790393e-05, "loss": 0.0473, "step": 42250 }, { "epoch": 26.934353091140853, "grad_norm": 0.5443461537361145, "learning_rate": 9.77597330040499e-05, "loss": 0.0253, "step": 42260 }, { "epoch": 26.940726577437857, "grad_norm": 0.5473769307136536, "learning_rate": 9.775810140987003e-05, "loss": 0.0355, "step": 42270 }, { "epoch": 26.94710006373486, "grad_norm": 0.4330480694770813, "learning_rate": 9.775646923538411e-05, "loss": 0.0319, "step": 42280 }, { "epoch": 26.953473550031866, "grad_norm": 0.36358991265296936, "learning_rate": 9.775483648061203e-05, "loss": 0.0371, "step": 42290 }, { "epoch": 26.95984703632887, "grad_norm": 0.6562947034835815, "learning_rate": 9.775320314557358e-05, "loss": 0.0349, "step": 42300 }, { "epoch": 26.966220522625875, "grad_norm": 0.4184749126434326, "learning_rate": 9.775156923028863e-05, "loss": 0.0232, "step": 42310 }, { "epoch": 26.97259400892288, "grad_norm": 0.3798682689666748, "learning_rate": 9.774993473477702e-05, "loss": 0.0298, "step": 42320 }, { "epoch": 26.978967495219884, "grad_norm": 0.7274406552314758, "learning_rate": 9.774829965905863e-05, "loss": 0.0337, "step": 42330 }, { "epoch": 26.985340981516888, "grad_norm": 0.5030524134635925, "learning_rate": 9.774666400315333e-05, "loss": 0.0316, "step": 42340 }, { "epoch": 26.991714467813892, "grad_norm": 0.4737909734249115, "learning_rate": 9.774502776708096e-05, "loss": 0.035, "step": 42350 }, { "epoch": 26.9980879541109, "grad_norm": 0.4415510892868042, "learning_rate": 9.774339095086144e-05, "loss": 0.0345, "step": 42360 }, { "epoch": 27.004461440407905, "grad_norm": 0.45854583382606506, "learning_rate": 9.774175355451463e-05, "loss": 0.0359, "step": 42370 }, { "epoch": 27.01083492670491, "grad_norm": 0.41550180315971375, "learning_rate": 9.774011557806044e-05, "loss": 0.0421, "step": 42380 }, { "epoch": 27.017208413001914, "grad_norm": 0.5666798949241638, "learning_rate": 9.773847702151877e-05, "loss": 0.0312, "step": 42390 }, { "epoch": 27.023581899298918, "grad_norm": 0.8952369689941406, "learning_rate": 9.773683788490955e-05, "loss": 0.0481, "step": 42400 }, { "epoch": 27.029955385595922, "grad_norm": 0.6514202952384949, "learning_rate": 9.773519816825265e-05, "loss": 0.035, "step": 42410 }, { "epoch": 27.036328871892927, "grad_norm": 0.519882082939148, "learning_rate": 9.773355787156804e-05, "loss": 0.0298, "step": 42420 }, { "epoch": 27.04270235818993, "grad_norm": 0.6521331071853638, "learning_rate": 9.773191699487564e-05, "loss": 0.0332, "step": 42430 }, { "epoch": 27.049075844486936, "grad_norm": 0.5467392802238464, "learning_rate": 9.773027553819537e-05, "loss": 0.0415, "step": 42440 }, { "epoch": 27.05544933078394, "grad_norm": 0.6759063005447388, "learning_rate": 9.77286335015472e-05, "loss": 0.0392, "step": 42450 }, { "epoch": 27.061822817080945, "grad_norm": 0.6634697914123535, "learning_rate": 9.772699088495106e-05, "loss": 0.0392, "step": 42460 }, { "epoch": 27.06819630337795, "grad_norm": 0.42603620886802673, "learning_rate": 9.772534768842692e-05, "loss": 0.0357, "step": 42470 }, { "epoch": 27.074569789674953, "grad_norm": 0.5501759648323059, "learning_rate": 9.772370391199473e-05, "loss": 0.0385, "step": 42480 }, { "epoch": 27.080943275971958, "grad_norm": 0.5600732564926147, "learning_rate": 9.77220595556745e-05, "loss": 0.0242, "step": 42490 }, { "epoch": 27.087316762268962, "grad_norm": 0.6579861640930176, "learning_rate": 9.772041461948618e-05, "loss": 0.0388, "step": 42500 }, { "epoch": 27.093690248565967, "grad_norm": 0.7068892121315002, "learning_rate": 9.771876910344977e-05, "loss": 0.0403, "step": 42510 }, { "epoch": 27.10006373486297, "grad_norm": 0.45586320757865906, "learning_rate": 9.771712300758526e-05, "loss": 0.0371, "step": 42520 }, { "epoch": 27.106437221159975, "grad_norm": 0.4014410376548767, "learning_rate": 9.771547633191266e-05, "loss": 0.0326, "step": 42530 }, { "epoch": 27.11281070745698, "grad_norm": 0.6864639520645142, "learning_rate": 9.771382907645197e-05, "loss": 0.0313, "step": 42540 }, { "epoch": 27.119184193753984, "grad_norm": 0.4354401230812073, "learning_rate": 9.77121812412232e-05, "loss": 0.032, "step": 42550 }, { "epoch": 27.12555768005099, "grad_norm": 0.4531353712081909, "learning_rate": 9.771053282624638e-05, "loss": 0.0267, "step": 42560 }, { "epoch": 27.131931166347993, "grad_norm": 0.2705150842666626, "learning_rate": 9.770888383154156e-05, "loss": 0.028, "step": 42570 }, { "epoch": 27.138304652644997, "grad_norm": 0.5971609950065613, "learning_rate": 9.770723425712873e-05, "loss": 0.0263, "step": 42580 }, { "epoch": 27.144678138942, "grad_norm": 0.6712452173233032, "learning_rate": 9.770558410302796e-05, "loss": 0.0265, "step": 42590 }, { "epoch": 27.151051625239006, "grad_norm": 0.40640750527381897, "learning_rate": 9.770393336925932e-05, "loss": 0.03, "step": 42600 }, { "epoch": 27.15742511153601, "grad_norm": 0.3879620432853699, "learning_rate": 9.770228205584284e-05, "loss": 0.0247, "step": 42610 }, { "epoch": 27.163798597833015, "grad_norm": 0.4280916154384613, "learning_rate": 9.77006301627986e-05, "loss": 0.0296, "step": 42620 }, { "epoch": 27.17017208413002, "grad_norm": 0.42630475759506226, "learning_rate": 9.769897769014667e-05, "loss": 0.039, "step": 42630 }, { "epoch": 27.176545570427024, "grad_norm": 0.3668208420276642, "learning_rate": 9.76973246379071e-05, "loss": 0.0248, "step": 42640 }, { "epoch": 27.182919056724028, "grad_norm": 0.4459036886692047, "learning_rate": 9.769567100610003e-05, "loss": 0.0256, "step": 42650 }, { "epoch": 27.189292543021033, "grad_norm": 0.4663746953010559, "learning_rate": 9.769401679474553e-05, "loss": 0.0404, "step": 42660 }, { "epoch": 27.195666029318037, "grad_norm": 0.7490636110305786, "learning_rate": 9.769236200386367e-05, "loss": 0.0247, "step": 42670 }, { "epoch": 27.20203951561504, "grad_norm": 0.3479316830635071, "learning_rate": 9.769070663347459e-05, "loss": 0.0201, "step": 42680 }, { "epoch": 27.208413001912046, "grad_norm": 0.2896709740161896, "learning_rate": 9.768905068359841e-05, "loss": 0.0267, "step": 42690 }, { "epoch": 27.21478648820905, "grad_norm": 0.7208948731422424, "learning_rate": 9.768739415425523e-05, "loss": 0.0451, "step": 42700 }, { "epoch": 27.221159974506055, "grad_norm": 0.38189494609832764, "learning_rate": 9.768573704546517e-05, "loss": 0.0242, "step": 42710 }, { "epoch": 27.22753346080306, "grad_norm": 0.48520195484161377, "learning_rate": 9.768407935724842e-05, "loss": 0.0238, "step": 42720 }, { "epoch": 27.233906947100063, "grad_norm": 0.5111380219459534, "learning_rate": 9.768242108962506e-05, "loss": 0.0421, "step": 42730 }, { "epoch": 27.240280433397068, "grad_norm": 0.8313961625099182, "learning_rate": 9.768076224261526e-05, "loss": 0.021, "step": 42740 }, { "epoch": 27.246653919694072, "grad_norm": 0.2813982367515564, "learning_rate": 9.767910281623919e-05, "loss": 0.0286, "step": 42750 }, { "epoch": 27.253027405991077, "grad_norm": 0.6580824255943298, "learning_rate": 9.767744281051701e-05, "loss": 0.0331, "step": 42760 }, { "epoch": 27.25940089228808, "grad_norm": 0.4643847644329071, "learning_rate": 9.767578222546888e-05, "loss": 0.0296, "step": 42770 }, { "epoch": 27.265774378585085, "grad_norm": 0.6691859364509583, "learning_rate": 9.767412106111496e-05, "loss": 0.0227, "step": 42780 }, { "epoch": 27.27214786488209, "grad_norm": 0.5022182464599609, "learning_rate": 9.767245931747549e-05, "loss": 0.0388, "step": 42790 }, { "epoch": 27.278521351179094, "grad_norm": 0.4062817394733429, "learning_rate": 9.76707969945706e-05, "loss": 0.0288, "step": 42800 }, { "epoch": 27.2848948374761, "grad_norm": 0.7121409177780151, "learning_rate": 9.766913409242053e-05, "loss": 0.0284, "step": 42810 }, { "epoch": 27.291268323773103, "grad_norm": 0.4174536168575287, "learning_rate": 9.766747061104548e-05, "loss": 0.0274, "step": 42820 }, { "epoch": 27.297641810070107, "grad_norm": 0.6136816740036011, "learning_rate": 9.766580655046566e-05, "loss": 0.0253, "step": 42830 }, { "epoch": 27.304015296367112, "grad_norm": 0.6404725313186646, "learning_rate": 9.766414191070128e-05, "loss": 0.0315, "step": 42840 }, { "epoch": 27.310388782664116, "grad_norm": 0.374038428068161, "learning_rate": 9.766247669177258e-05, "loss": 0.0536, "step": 42850 }, { "epoch": 27.31676226896112, "grad_norm": 0.6502777338027954, "learning_rate": 9.766081089369979e-05, "loss": 0.0274, "step": 42860 }, { "epoch": 27.323135755258125, "grad_norm": 0.43991732597351074, "learning_rate": 9.765914451650313e-05, "loss": 0.0299, "step": 42870 }, { "epoch": 27.32950924155513, "grad_norm": 0.38164645433425903, "learning_rate": 9.765747756020287e-05, "loss": 0.025, "step": 42880 }, { "epoch": 27.335882727852134, "grad_norm": 0.325210303068161, "learning_rate": 9.765581002481927e-05, "loss": 0.0293, "step": 42890 }, { "epoch": 27.34225621414914, "grad_norm": 0.5042225122451782, "learning_rate": 9.765414191037257e-05, "loss": 0.0389, "step": 42900 }, { "epoch": 27.348629700446143, "grad_norm": 0.4349748492240906, "learning_rate": 9.765247321688307e-05, "loss": 0.0282, "step": 42910 }, { "epoch": 27.355003186743147, "grad_norm": 0.5858556032180786, "learning_rate": 9.765080394437102e-05, "loss": 0.024, "step": 42920 }, { "epoch": 27.36137667304015, "grad_norm": 0.3326181173324585, "learning_rate": 9.764913409285672e-05, "loss": 0.028, "step": 42930 }, { "epoch": 27.367750159337156, "grad_norm": 0.34276527166366577, "learning_rate": 9.764746366236044e-05, "loss": 0.0356, "step": 42940 }, { "epoch": 27.37412364563416, "grad_norm": 0.5838547348976135, "learning_rate": 9.764579265290251e-05, "loss": 0.032, "step": 42950 }, { "epoch": 27.380497131931165, "grad_norm": 0.3988277316093445, "learning_rate": 9.76441210645032e-05, "loss": 0.0234, "step": 42960 }, { "epoch": 27.38687061822817, "grad_norm": 0.41003066301345825, "learning_rate": 9.764244889718284e-05, "loss": 0.0402, "step": 42970 }, { "epoch": 27.393244104525174, "grad_norm": 0.5355158448219299, "learning_rate": 9.764077615096173e-05, "loss": 0.0333, "step": 42980 }, { "epoch": 27.39961759082218, "grad_norm": 0.352802574634552, "learning_rate": 9.763910282586022e-05, "loss": 0.0259, "step": 42990 }, { "epoch": 27.405991077119186, "grad_norm": 0.40623608231544495, "learning_rate": 9.763742892189865e-05, "loss": 0.0331, "step": 43000 }, { "epoch": 27.41236456341619, "grad_norm": 0.37196511030197144, "learning_rate": 9.763575443909732e-05, "loss": 0.029, "step": 43010 }, { "epoch": 27.418738049713195, "grad_norm": 0.5398632287979126, "learning_rate": 9.763407937747659e-05, "loss": 0.0258, "step": 43020 }, { "epoch": 27.4251115360102, "grad_norm": 0.7087506651878357, "learning_rate": 9.763240373705684e-05, "loss": 0.0293, "step": 43030 }, { "epoch": 27.431485022307204, "grad_norm": 0.4795816242694855, "learning_rate": 9.76307275178584e-05, "loss": 0.0391, "step": 43040 }, { "epoch": 27.437858508604208, "grad_norm": 0.4953254461288452, "learning_rate": 9.762905071990165e-05, "loss": 0.0365, "step": 43050 }, { "epoch": 27.444231994901212, "grad_norm": 0.6698494553565979, "learning_rate": 9.762737334320697e-05, "loss": 0.0265, "step": 43060 }, { "epoch": 27.450605481198217, "grad_norm": 0.5150022506713867, "learning_rate": 9.762569538779474e-05, "loss": 0.0361, "step": 43070 }, { "epoch": 27.45697896749522, "grad_norm": 0.9207808971405029, "learning_rate": 9.762401685368534e-05, "loss": 0.0355, "step": 43080 }, { "epoch": 27.463352453792226, "grad_norm": 0.7661358714103699, "learning_rate": 9.762233774089917e-05, "loss": 0.0334, "step": 43090 }, { "epoch": 27.46972594008923, "grad_norm": 0.41314250230789185, "learning_rate": 9.762065804945663e-05, "loss": 0.0302, "step": 43100 }, { "epoch": 27.476099426386234, "grad_norm": 0.48619672656059265, "learning_rate": 9.761897777937813e-05, "loss": 0.0344, "step": 43110 }, { "epoch": 27.48247291268324, "grad_norm": 0.3556061387062073, "learning_rate": 9.761729693068411e-05, "loss": 0.0354, "step": 43120 }, { "epoch": 27.488846398980243, "grad_norm": 0.473741352558136, "learning_rate": 9.761561550339496e-05, "loss": 0.025, "step": 43130 }, { "epoch": 27.495219885277248, "grad_norm": 0.6408193111419678, "learning_rate": 9.761393349753114e-05, "loss": 0.034, "step": 43140 }, { "epoch": 27.501593371574252, "grad_norm": 0.5366620421409607, "learning_rate": 9.761225091311305e-05, "loss": 0.0306, "step": 43150 }, { "epoch": 27.507966857871256, "grad_norm": 0.554911732673645, "learning_rate": 9.761056775016117e-05, "loss": 0.0236, "step": 43160 }, { "epoch": 27.51434034416826, "grad_norm": 0.45023760199546814, "learning_rate": 9.760888400869595e-05, "loss": 0.0496, "step": 43170 }, { "epoch": 27.520713830465265, "grad_norm": 0.4123894274234772, "learning_rate": 9.760719968873781e-05, "loss": 0.0267, "step": 43180 }, { "epoch": 27.52708731676227, "grad_norm": 0.5496741533279419, "learning_rate": 9.760551479030728e-05, "loss": 0.043, "step": 43190 }, { "epoch": 27.533460803059274, "grad_norm": 0.5389478206634521, "learning_rate": 9.760382931342477e-05, "loss": 0.0255, "step": 43200 }, { "epoch": 27.53983428935628, "grad_norm": 0.5101574659347534, "learning_rate": 9.76021432581108e-05, "loss": 0.0399, "step": 43210 }, { "epoch": 27.546207775653283, "grad_norm": 0.42027056217193604, "learning_rate": 9.760045662438585e-05, "loss": 0.0299, "step": 43220 }, { "epoch": 27.552581261950287, "grad_norm": 0.4567740857601166, "learning_rate": 9.75987694122704e-05, "loss": 0.0237, "step": 43230 }, { "epoch": 27.55895474824729, "grad_norm": 0.5957986116409302, "learning_rate": 9.759708162178497e-05, "loss": 0.0333, "step": 43240 }, { "epoch": 27.565328234544296, "grad_norm": 0.582748532295227, "learning_rate": 9.759539325295005e-05, "loss": 0.0278, "step": 43250 }, { "epoch": 27.5717017208413, "grad_norm": 0.31856614351272583, "learning_rate": 9.759370430578618e-05, "loss": 0.0337, "step": 43260 }, { "epoch": 27.578075207138305, "grad_norm": 0.29704105854034424, "learning_rate": 9.759201478031385e-05, "loss": 0.0371, "step": 43270 }, { "epoch": 27.58444869343531, "grad_norm": 0.23343893885612488, "learning_rate": 9.75903246765536e-05, "loss": 0.0242, "step": 43280 }, { "epoch": 27.590822179732314, "grad_norm": 0.5061424374580383, "learning_rate": 9.7588633994526e-05, "loss": 0.0306, "step": 43290 }, { "epoch": 27.597195666029318, "grad_norm": 0.37801432609558105, "learning_rate": 9.758694273425155e-05, "loss": 0.0361, "step": 43300 }, { "epoch": 27.603569152326322, "grad_norm": 0.5899465680122375, "learning_rate": 9.758525089575081e-05, "loss": 0.0355, "step": 43310 }, { "epoch": 27.609942638623327, "grad_norm": 0.6520706415176392, "learning_rate": 9.758355847904435e-05, "loss": 0.0388, "step": 43320 }, { "epoch": 27.61631612492033, "grad_norm": 0.5751564502716064, "learning_rate": 9.758186548415273e-05, "loss": 0.0336, "step": 43330 }, { "epoch": 27.622689611217336, "grad_norm": 0.6411465406417847, "learning_rate": 9.758017191109652e-05, "loss": 0.0339, "step": 43340 }, { "epoch": 27.62906309751434, "grad_norm": 0.5868284702301025, "learning_rate": 9.757847775989631e-05, "loss": 0.0414, "step": 43350 }, { "epoch": 27.635436583811344, "grad_norm": 0.8106486201286316, "learning_rate": 9.757678303057265e-05, "loss": 0.0347, "step": 43360 }, { "epoch": 27.64181007010835, "grad_norm": 0.5329729914665222, "learning_rate": 9.757508772314616e-05, "loss": 0.0365, "step": 43370 }, { "epoch": 27.648183556405353, "grad_norm": 0.6391049027442932, "learning_rate": 9.757339183763745e-05, "loss": 0.0381, "step": 43380 }, { "epoch": 27.654557042702358, "grad_norm": 0.35752376914024353, "learning_rate": 9.757169537406709e-05, "loss": 0.0385, "step": 43390 }, { "epoch": 27.660930528999362, "grad_norm": 0.36805009841918945, "learning_rate": 9.756999833245575e-05, "loss": 0.0276, "step": 43400 }, { "epoch": 27.667304015296367, "grad_norm": 0.4045107066631317, "learning_rate": 9.756830071282398e-05, "loss": 0.024, "step": 43410 }, { "epoch": 27.67367750159337, "grad_norm": 0.3612765669822693, "learning_rate": 9.756660251519248e-05, "loss": 0.028, "step": 43420 }, { "epoch": 27.680050987890375, "grad_norm": 0.5063760280609131, "learning_rate": 9.756490373958183e-05, "loss": 0.0253, "step": 43430 }, { "epoch": 27.68642447418738, "grad_norm": 0.3555130958557129, "learning_rate": 9.756320438601269e-05, "loss": 0.0362, "step": 43440 }, { "epoch": 27.692797960484384, "grad_norm": 0.541277289390564, "learning_rate": 9.75615044545057e-05, "loss": 0.0323, "step": 43450 }, { "epoch": 27.69917144678139, "grad_norm": 0.5074331164360046, "learning_rate": 9.755980394508155e-05, "loss": 0.0296, "step": 43460 }, { "epoch": 27.705544933078393, "grad_norm": 0.6467015743255615, "learning_rate": 9.755810285776086e-05, "loss": 0.0322, "step": 43470 }, { "epoch": 27.711918419375397, "grad_norm": 0.5056699514389038, "learning_rate": 9.755640119256432e-05, "loss": 0.0319, "step": 43480 }, { "epoch": 27.7182919056724, "grad_norm": 0.35040026903152466, "learning_rate": 9.755469894951261e-05, "loss": 0.0204, "step": 43490 }, { "epoch": 27.724665391969406, "grad_norm": 0.33973732590675354, "learning_rate": 9.75529961286264e-05, "loss": 0.0355, "step": 43500 }, { "epoch": 27.73103887826641, "grad_norm": 0.46022918820381165, "learning_rate": 9.75512927299264e-05, "loss": 0.0298, "step": 43510 }, { "epoch": 27.737412364563415, "grad_norm": 0.45593592524528503, "learning_rate": 9.754958875343329e-05, "loss": 0.0269, "step": 43520 }, { "epoch": 27.74378585086042, "grad_norm": 0.6009355783462524, "learning_rate": 9.754788419916779e-05, "loss": 0.0271, "step": 43530 }, { "epoch": 27.750159337157424, "grad_norm": 0.5592274069786072, "learning_rate": 9.75461790671506e-05, "loss": 0.0333, "step": 43540 }, { "epoch": 27.756532823454428, "grad_norm": 0.7361756563186646, "learning_rate": 9.754447335740244e-05, "loss": 0.0306, "step": 43550 }, { "epoch": 27.762906309751433, "grad_norm": 0.5715871453285217, "learning_rate": 9.754276706994406e-05, "loss": 0.0195, "step": 43560 }, { "epoch": 27.769279796048437, "grad_norm": 0.7657013535499573, "learning_rate": 9.754106020479614e-05, "loss": 0.0361, "step": 43570 }, { "epoch": 27.77565328234544, "grad_norm": 0.5092758536338806, "learning_rate": 9.753935276197947e-05, "loss": 0.0384, "step": 43580 }, { "epoch": 27.782026768642446, "grad_norm": 0.35640034079551697, "learning_rate": 9.753764474151477e-05, "loss": 0.0388, "step": 43590 }, { "epoch": 27.78840025493945, "grad_norm": 0.23954685032367706, "learning_rate": 9.753593614342282e-05, "loss": 0.019, "step": 43600 }, { "epoch": 27.794773741236455, "grad_norm": 0.43784767389297485, "learning_rate": 9.753422696772435e-05, "loss": 0.0316, "step": 43610 }, { "epoch": 27.801147227533463, "grad_norm": 0.6333953142166138, "learning_rate": 9.753251721444014e-05, "loss": 0.0309, "step": 43620 }, { "epoch": 27.807520713830467, "grad_norm": 0.31105631589889526, "learning_rate": 9.753080688359099e-05, "loss": 0.0314, "step": 43630 }, { "epoch": 27.81389420012747, "grad_norm": 0.41747790575027466, "learning_rate": 9.752909597519763e-05, "loss": 0.0334, "step": 43640 }, { "epoch": 27.820267686424476, "grad_norm": 0.7120935916900635, "learning_rate": 9.75273844892809e-05, "loss": 0.0288, "step": 43650 }, { "epoch": 27.82664117272148, "grad_norm": 0.40811508893966675, "learning_rate": 9.752567242586157e-05, "loss": 0.0252, "step": 43660 }, { "epoch": 27.833014659018485, "grad_norm": 0.35748323798179626, "learning_rate": 9.752395978496045e-05, "loss": 0.0206, "step": 43670 }, { "epoch": 27.83938814531549, "grad_norm": 0.5098660588264465, "learning_rate": 9.752224656659833e-05, "loss": 0.0281, "step": 43680 }, { "epoch": 27.845761631612493, "grad_norm": 0.5578731894493103, "learning_rate": 9.752053277079607e-05, "loss": 0.0193, "step": 43690 }, { "epoch": 27.852135117909498, "grad_norm": 0.7307285666465759, "learning_rate": 9.751881839757446e-05, "loss": 0.0432, "step": 43700 }, { "epoch": 27.858508604206502, "grad_norm": 0.811240017414093, "learning_rate": 9.751710344695435e-05, "loss": 0.0405, "step": 43710 }, { "epoch": 27.864882090503507, "grad_norm": 0.5275760889053345, "learning_rate": 9.751538791895656e-05, "loss": 0.0339, "step": 43720 }, { "epoch": 27.87125557680051, "grad_norm": 0.7013908624649048, "learning_rate": 9.751367181360195e-05, "loss": 0.036, "step": 43730 }, { "epoch": 27.877629063097515, "grad_norm": 0.4571201801300049, "learning_rate": 9.751195513091136e-05, "loss": 0.0216, "step": 43740 }, { "epoch": 27.88400254939452, "grad_norm": 0.5197249054908752, "learning_rate": 9.751023787090568e-05, "loss": 0.035, "step": 43750 }, { "epoch": 27.890376035691524, "grad_norm": 0.4787781536579132, "learning_rate": 9.750852003360573e-05, "loss": 0.0374, "step": 43760 }, { "epoch": 27.89674952198853, "grad_norm": 0.7194153666496277, "learning_rate": 9.750680161903241e-05, "loss": 0.026, "step": 43770 }, { "epoch": 27.903123008285533, "grad_norm": 0.6144666075706482, "learning_rate": 9.750508262720659e-05, "loss": 0.0405, "step": 43780 }, { "epoch": 27.909496494582537, "grad_norm": 0.6022862195968628, "learning_rate": 9.750336305814917e-05, "loss": 0.0338, "step": 43790 }, { "epoch": 27.915869980879542, "grad_norm": 0.7967201471328735, "learning_rate": 9.750164291188104e-05, "loss": 0.0291, "step": 43800 }, { "epoch": 27.922243467176546, "grad_norm": 0.27110058069229126, "learning_rate": 9.74999221884231e-05, "loss": 0.0265, "step": 43810 }, { "epoch": 27.92861695347355, "grad_norm": 0.5907347798347473, "learning_rate": 9.749820088779625e-05, "loss": 0.0283, "step": 43820 }, { "epoch": 27.934990439770555, "grad_norm": 0.4597783088684082, "learning_rate": 9.749647901002142e-05, "loss": 0.0281, "step": 43830 }, { "epoch": 27.94136392606756, "grad_norm": 0.40914514660835266, "learning_rate": 9.749475655511952e-05, "loss": 0.0352, "step": 43840 }, { "epoch": 27.947737412364564, "grad_norm": 0.6103016138076782, "learning_rate": 9.74930335231115e-05, "loss": 0.0335, "step": 43850 }, { "epoch": 27.95411089866157, "grad_norm": 0.44958028197288513, "learning_rate": 9.749130991401827e-05, "loss": 0.0355, "step": 43860 }, { "epoch": 27.960484384958573, "grad_norm": 0.5218605995178223, "learning_rate": 9.748958572786079e-05, "loss": 0.0301, "step": 43870 }, { "epoch": 27.966857871255577, "grad_norm": 0.3141953945159912, "learning_rate": 9.748786096466001e-05, "loss": 0.0285, "step": 43880 }, { "epoch": 27.97323135755258, "grad_norm": 0.5665042400360107, "learning_rate": 9.748613562443687e-05, "loss": 0.0263, "step": 43890 }, { "epoch": 27.979604843849586, "grad_norm": 0.343961626291275, "learning_rate": 9.748440970721236e-05, "loss": 0.0278, "step": 43900 }, { "epoch": 27.98597833014659, "grad_norm": 0.5238397121429443, "learning_rate": 9.748268321300742e-05, "loss": 0.0389, "step": 43910 }, { "epoch": 27.992351816443595, "grad_norm": 0.6671867370605469, "learning_rate": 9.748095614184308e-05, "loss": 0.036, "step": 43920 }, { "epoch": 27.9987253027406, "grad_norm": 0.4362097382545471, "learning_rate": 9.747922849374027e-05, "loss": 0.0269, "step": 43930 }, { "epoch": 28.005098789037604, "grad_norm": 0.39225107431411743, "learning_rate": 9.747750026872e-05, "loss": 0.0291, "step": 43940 }, { "epoch": 28.011472275334608, "grad_norm": 0.5118345022201538, "learning_rate": 9.747577146680329e-05, "loss": 0.0277, "step": 43950 }, { "epoch": 28.017845761631612, "grad_norm": 0.7160177826881409, "learning_rate": 9.747404208801115e-05, "loss": 0.0399, "step": 43960 }, { "epoch": 28.024219247928617, "grad_norm": 0.4097060561180115, "learning_rate": 9.747231213236454e-05, "loss": 0.0313, "step": 43970 }, { "epoch": 28.03059273422562, "grad_norm": 0.5713366270065308, "learning_rate": 9.747058159988453e-05, "loss": 0.0385, "step": 43980 }, { "epoch": 28.036966220522626, "grad_norm": 0.5803492069244385, "learning_rate": 9.746885049059213e-05, "loss": 0.0329, "step": 43990 }, { "epoch": 28.04333970681963, "grad_norm": 0.45776650309562683, "learning_rate": 9.746711880450838e-05, "loss": 0.0465, "step": 44000 }, { "epoch": 28.049713193116634, "grad_norm": 0.4027117192745209, "learning_rate": 9.746538654165433e-05, "loss": 0.0364, "step": 44010 }, { "epoch": 28.05608667941364, "grad_norm": 0.7454760670661926, "learning_rate": 9.746365370205101e-05, "loss": 0.0285, "step": 44020 }, { "epoch": 28.062460165710643, "grad_norm": 0.5461715459823608, "learning_rate": 9.74619202857195e-05, "loss": 0.0333, "step": 44030 }, { "epoch": 28.068833652007648, "grad_norm": 0.6563729643821716, "learning_rate": 9.746018629268083e-05, "loss": 0.0276, "step": 44040 }, { "epoch": 28.075207138304652, "grad_norm": 0.28457748889923096, "learning_rate": 9.745845172295609e-05, "loss": 0.0251, "step": 44050 }, { "epoch": 28.081580624601656, "grad_norm": 0.6931522488594055, "learning_rate": 9.745671657656634e-05, "loss": 0.0317, "step": 44060 }, { "epoch": 28.08795411089866, "grad_norm": 0.7118549346923828, "learning_rate": 9.74549808535327e-05, "loss": 0.0314, "step": 44070 }, { "epoch": 28.094327597195665, "grad_norm": 0.5831851959228516, "learning_rate": 9.745324455387623e-05, "loss": 0.0265, "step": 44080 }, { "epoch": 28.10070108349267, "grad_norm": 0.5146048665046692, "learning_rate": 9.745150767761804e-05, "loss": 0.03, "step": 44090 }, { "epoch": 28.107074569789674, "grad_norm": 0.4322667717933655, "learning_rate": 9.744977022477921e-05, "loss": 0.0413, "step": 44100 }, { "epoch": 28.11344805608668, "grad_norm": 0.4913763999938965, "learning_rate": 9.74480321953809e-05, "loss": 0.0418, "step": 44110 }, { "epoch": 28.119821542383683, "grad_norm": 0.5145254731178284, "learning_rate": 9.744629358944418e-05, "loss": 0.0311, "step": 44120 }, { "epoch": 28.126195028680687, "grad_norm": 0.3814544379711151, "learning_rate": 9.74445544069902e-05, "loss": 0.0309, "step": 44130 }, { "epoch": 28.13256851497769, "grad_norm": 0.4900464415550232, "learning_rate": 9.74428146480401e-05, "loss": 0.0298, "step": 44140 }, { "epoch": 28.138942001274696, "grad_norm": 0.43112340569496155, "learning_rate": 9.7441074312615e-05, "loss": 0.021, "step": 44150 }, { "epoch": 28.1453154875717, "grad_norm": 0.5536928772926331, "learning_rate": 9.743933340073606e-05, "loss": 0.0307, "step": 44160 }, { "epoch": 28.151688973868705, "grad_norm": 0.5054657459259033, "learning_rate": 9.743759191242443e-05, "loss": 0.0217, "step": 44170 }, { "epoch": 28.15806246016571, "grad_norm": 0.3926021456718445, "learning_rate": 9.743584984770125e-05, "loss": 0.0233, "step": 44180 }, { "epoch": 28.164435946462714, "grad_norm": 0.38819441199302673, "learning_rate": 9.743410720658773e-05, "loss": 0.0301, "step": 44190 }, { "epoch": 28.170809432759718, "grad_norm": 0.6325846314430237, "learning_rate": 9.743236398910501e-05, "loss": 0.0417, "step": 44200 }, { "epoch": 28.177182919056722, "grad_norm": 0.9727165102958679, "learning_rate": 9.74306201952743e-05, "loss": 0.0342, "step": 44210 }, { "epoch": 28.183556405353727, "grad_norm": 0.6439951062202454, "learning_rate": 9.742887582511675e-05, "loss": 0.0289, "step": 44220 }, { "epoch": 28.18992989165073, "grad_norm": 0.5638203024864197, "learning_rate": 9.742713087865358e-05, "loss": 0.0314, "step": 44230 }, { "epoch": 28.196303377947736, "grad_norm": 0.45663660764694214, "learning_rate": 9.742538535590601e-05, "loss": 0.0267, "step": 44240 }, { "epoch": 28.202676864244744, "grad_norm": 0.3631056547164917, "learning_rate": 9.742363925689521e-05, "loss": 0.0238, "step": 44250 }, { "epoch": 28.209050350541748, "grad_norm": 0.4587206542491913, "learning_rate": 9.742189258164243e-05, "loss": 0.0329, "step": 44260 }, { "epoch": 28.215423836838752, "grad_norm": 0.34747639298439026, "learning_rate": 9.742014533016888e-05, "loss": 0.044, "step": 44270 }, { "epoch": 28.221797323135757, "grad_norm": 0.2990148365497589, "learning_rate": 9.74183975024958e-05, "loss": 0.0288, "step": 44280 }, { "epoch": 28.22817080943276, "grad_norm": 0.3446245789527893, "learning_rate": 9.741664909864441e-05, "loss": 0.0301, "step": 44290 }, { "epoch": 28.234544295729766, "grad_norm": 0.701809287071228, "learning_rate": 9.741490011863597e-05, "loss": 0.0361, "step": 44300 }, { "epoch": 28.24091778202677, "grad_norm": 0.32744964957237244, "learning_rate": 9.741315056249171e-05, "loss": 0.0269, "step": 44310 }, { "epoch": 28.247291268323774, "grad_norm": 0.5077762007713318, "learning_rate": 9.741140043023294e-05, "loss": 0.0322, "step": 44320 }, { "epoch": 28.25366475462078, "grad_norm": 0.5891860723495483, "learning_rate": 9.740964972188086e-05, "loss": 0.0434, "step": 44330 }, { "epoch": 28.260038240917783, "grad_norm": 0.8284059166908264, "learning_rate": 9.74078984374568e-05, "loss": 0.0284, "step": 44340 }, { "epoch": 28.266411727214788, "grad_norm": 0.2561104893684387, "learning_rate": 9.740614657698198e-05, "loss": 0.0315, "step": 44350 }, { "epoch": 28.272785213511792, "grad_norm": 0.4391784071922302, "learning_rate": 9.740439414047773e-05, "loss": 0.0295, "step": 44360 }, { "epoch": 28.279158699808796, "grad_norm": 0.5876372456550598, "learning_rate": 9.740264112796536e-05, "loss": 0.0346, "step": 44370 }, { "epoch": 28.2855321861058, "grad_norm": 0.6304070949554443, "learning_rate": 9.740088753946612e-05, "loss": 0.0252, "step": 44380 }, { "epoch": 28.291905672402805, "grad_norm": 0.5483896136283875, "learning_rate": 9.739913337500134e-05, "loss": 0.0399, "step": 44390 }, { "epoch": 28.29827915869981, "grad_norm": 0.42039191722869873, "learning_rate": 9.739737863459235e-05, "loss": 0.0291, "step": 44400 }, { "epoch": 28.304652644996814, "grad_norm": 0.42290419340133667, "learning_rate": 9.739562331826045e-05, "loss": 0.0387, "step": 44410 }, { "epoch": 28.31102613129382, "grad_norm": 0.47661104798316956, "learning_rate": 9.739386742602698e-05, "loss": 0.0531, "step": 44420 }, { "epoch": 28.317399617590823, "grad_norm": 0.41108641028404236, "learning_rate": 9.739211095791329e-05, "loss": 0.0418, "step": 44430 }, { "epoch": 28.323773103887827, "grad_norm": 0.6814137101173401, "learning_rate": 9.739035391394069e-05, "loss": 0.0332, "step": 44440 }, { "epoch": 28.33014659018483, "grad_norm": 0.376516729593277, "learning_rate": 9.738859629413055e-05, "loss": 0.0312, "step": 44450 }, { "epoch": 28.336520076481836, "grad_norm": 0.6428983211517334, "learning_rate": 9.738683809850422e-05, "loss": 0.0344, "step": 44460 }, { "epoch": 28.34289356277884, "grad_norm": 0.5290582776069641, "learning_rate": 9.738507932708307e-05, "loss": 0.0307, "step": 44470 }, { "epoch": 28.349267049075845, "grad_norm": 0.7155981063842773, "learning_rate": 9.738331997988845e-05, "loss": 0.0418, "step": 44480 }, { "epoch": 28.35564053537285, "grad_norm": 0.6242151856422424, "learning_rate": 9.738156005694179e-05, "loss": 0.029, "step": 44490 }, { "epoch": 28.362014021669854, "grad_norm": 0.36684373021125793, "learning_rate": 9.737979955826442e-05, "loss": 0.0253, "step": 44500 }, { "epoch": 28.368387507966858, "grad_norm": 0.6103434562683105, "learning_rate": 9.737803848387775e-05, "loss": 0.0322, "step": 44510 }, { "epoch": 28.374760994263863, "grad_norm": 0.5094860196113586, "learning_rate": 9.737627683380317e-05, "loss": 0.0385, "step": 44520 }, { "epoch": 28.381134480560867, "grad_norm": 0.5378496050834656, "learning_rate": 9.73745146080621e-05, "loss": 0.0328, "step": 44530 }, { "epoch": 28.38750796685787, "grad_norm": 0.45685452222824097, "learning_rate": 9.737275180667597e-05, "loss": 0.0305, "step": 44540 }, { "epoch": 28.393881453154876, "grad_norm": 0.44867458939552307, "learning_rate": 9.737098842966617e-05, "loss": 0.0392, "step": 44550 }, { "epoch": 28.40025493945188, "grad_norm": 0.355293869972229, "learning_rate": 9.736922447705413e-05, "loss": 0.0337, "step": 44560 }, { "epoch": 28.406628425748885, "grad_norm": 0.4744146168231964, "learning_rate": 9.736745994886127e-05, "loss": 0.037, "step": 44570 }, { "epoch": 28.41300191204589, "grad_norm": 0.6998198628425598, "learning_rate": 9.736569484510908e-05, "loss": 0.0442, "step": 44580 }, { "epoch": 28.419375398342893, "grad_norm": 0.3674408793449402, "learning_rate": 9.736392916581896e-05, "loss": 0.0495, "step": 44590 }, { "epoch": 28.425748884639898, "grad_norm": 0.49607911705970764, "learning_rate": 9.736216291101239e-05, "loss": 0.0242, "step": 44600 }, { "epoch": 28.432122370936902, "grad_norm": 0.3356236517429352, "learning_rate": 9.736039608071081e-05, "loss": 0.0367, "step": 44610 }, { "epoch": 28.438495857233907, "grad_norm": 0.5102330446243286, "learning_rate": 9.735862867493572e-05, "loss": 0.0293, "step": 44620 }, { "epoch": 28.44486934353091, "grad_norm": 0.5678808689117432, "learning_rate": 9.735686069370857e-05, "loss": 0.0379, "step": 44630 }, { "epoch": 28.451242829827915, "grad_norm": 0.4358641803264618, "learning_rate": 9.735509213705086e-05, "loss": 0.0289, "step": 44640 }, { "epoch": 28.45761631612492, "grad_norm": 0.5033252835273743, "learning_rate": 9.735332300498405e-05, "loss": 0.0369, "step": 44650 }, { "epoch": 28.463989802421924, "grad_norm": 0.23804186284542084, "learning_rate": 9.735155329752966e-05, "loss": 0.0327, "step": 44660 }, { "epoch": 28.47036328871893, "grad_norm": 0.5551007390022278, "learning_rate": 9.734978301470919e-05, "loss": 0.0305, "step": 44670 }, { "epoch": 28.476736775015933, "grad_norm": 0.5479502081871033, "learning_rate": 9.734801215654417e-05, "loss": 0.0384, "step": 44680 }, { "epoch": 28.483110261312937, "grad_norm": 0.5125648975372314, "learning_rate": 9.734624072305607e-05, "loss": 0.0367, "step": 44690 }, { "epoch": 28.489483747609942, "grad_norm": 0.5217462778091431, "learning_rate": 9.734446871426645e-05, "loss": 0.033, "step": 44700 }, { "epoch": 28.495857233906946, "grad_norm": 0.5010473728179932, "learning_rate": 9.734269613019687e-05, "loss": 0.0377, "step": 44710 }, { "epoch": 28.50223072020395, "grad_norm": 0.4423927962779999, "learning_rate": 9.734092297086879e-05, "loss": 0.028, "step": 44720 }, { "epoch": 28.508604206500955, "grad_norm": 0.4629896581172943, "learning_rate": 9.733914923630381e-05, "loss": 0.016, "step": 44730 }, { "epoch": 28.51497769279796, "grad_norm": 0.5018481016159058, "learning_rate": 9.733737492652348e-05, "loss": 0.0288, "step": 44740 }, { "epoch": 28.521351179094964, "grad_norm": 0.3976377844810486, "learning_rate": 9.733560004154934e-05, "loss": 0.0331, "step": 44750 }, { "epoch": 28.52772466539197, "grad_norm": 0.7726106643676758, "learning_rate": 9.7333824581403e-05, "loss": 0.0387, "step": 44760 }, { "epoch": 28.534098151688973, "grad_norm": 0.4114859402179718, "learning_rate": 9.733204854610596e-05, "loss": 0.0356, "step": 44770 }, { "epoch": 28.540471637985977, "grad_norm": 0.40384554862976074, "learning_rate": 9.733027193567986e-05, "loss": 0.0349, "step": 44780 }, { "epoch": 28.54684512428298, "grad_norm": 0.22549287974834442, "learning_rate": 9.732849475014625e-05, "loss": 0.038, "step": 44790 }, { "epoch": 28.553218610579986, "grad_norm": 0.6069677472114563, "learning_rate": 9.732671698952676e-05, "loss": 0.0379, "step": 44800 }, { "epoch": 28.55959209687699, "grad_norm": 0.42942240834236145, "learning_rate": 9.732493865384299e-05, "loss": 0.0244, "step": 44810 }, { "epoch": 28.565965583173995, "grad_norm": 0.6117925643920898, "learning_rate": 9.73231597431165e-05, "loss": 0.0326, "step": 44820 }, { "epoch": 28.572339069471, "grad_norm": 0.42230334877967834, "learning_rate": 9.732138025736895e-05, "loss": 0.0613, "step": 44830 }, { "epoch": 28.578712555768004, "grad_norm": 0.27693191170692444, "learning_rate": 9.731960019662196e-05, "loss": 0.0351, "step": 44840 }, { "epoch": 28.585086042065008, "grad_norm": 0.6259333491325378, "learning_rate": 9.731781956089714e-05, "loss": 0.0387, "step": 44850 }, { "epoch": 28.591459528362012, "grad_norm": 0.31598782539367676, "learning_rate": 9.731603835021614e-05, "loss": 0.0293, "step": 44860 }, { "epoch": 28.597833014659017, "grad_norm": 0.5598121881484985, "learning_rate": 9.73142565646006e-05, "loss": 0.0303, "step": 44870 }, { "epoch": 28.604206500956025, "grad_norm": 0.7967960238456726, "learning_rate": 9.731247420407219e-05, "loss": 0.0435, "step": 44880 }, { "epoch": 28.61057998725303, "grad_norm": 0.47927552461624146, "learning_rate": 9.731069126865254e-05, "loss": 0.029, "step": 44890 }, { "epoch": 28.616953473550033, "grad_norm": 0.6913169026374817, "learning_rate": 9.73089077583633e-05, "loss": 0.0354, "step": 44900 }, { "epoch": 28.623326959847038, "grad_norm": 0.3500741720199585, "learning_rate": 9.730712367322618e-05, "loss": 0.0276, "step": 44910 }, { "epoch": 28.629700446144042, "grad_norm": 0.805134117603302, "learning_rate": 9.730533901326285e-05, "loss": 0.0348, "step": 44920 }, { "epoch": 28.636073932441047, "grad_norm": 0.4991244077682495, "learning_rate": 9.730355377849496e-05, "loss": 0.0408, "step": 44930 }, { "epoch": 28.64244741873805, "grad_norm": 0.7973124980926514, "learning_rate": 9.730176796894425e-05, "loss": 0.0467, "step": 44940 }, { "epoch": 28.648820905035056, "grad_norm": 0.5614928007125854, "learning_rate": 9.72999815846324e-05, "loss": 0.0302, "step": 44950 }, { "epoch": 28.65519439133206, "grad_norm": 0.36722517013549805, "learning_rate": 9.72981946255811e-05, "loss": 0.0309, "step": 44960 }, { "epoch": 28.661567877629064, "grad_norm": 0.6403418779373169, "learning_rate": 9.72964070918121e-05, "loss": 0.0253, "step": 44970 }, { "epoch": 28.66794136392607, "grad_norm": 0.5140339136123657, "learning_rate": 9.729461898334709e-05, "loss": 0.0455, "step": 44980 }, { "epoch": 28.674314850223073, "grad_norm": 0.530879557132721, "learning_rate": 9.72928303002078e-05, "loss": 0.0456, "step": 44990 }, { "epoch": 28.680688336520078, "grad_norm": 0.8292019963264465, "learning_rate": 9.729104104241598e-05, "loss": 0.032, "step": 45000 }, { "epoch": 28.687061822817082, "grad_norm": 0.6900933384895325, "learning_rate": 9.728925120999334e-05, "loss": 0.0313, "step": 45010 }, { "epoch": 28.693435309114086, "grad_norm": 0.42918214201927185, "learning_rate": 9.728746080296167e-05, "loss": 0.0279, "step": 45020 }, { "epoch": 28.69980879541109, "grad_norm": 0.3363021910190582, "learning_rate": 9.728566982134271e-05, "loss": 0.0394, "step": 45030 }, { "epoch": 28.706182281708095, "grad_norm": 0.3593839406967163, "learning_rate": 9.728387826515819e-05, "loss": 0.0299, "step": 45040 }, { "epoch": 28.7125557680051, "grad_norm": 0.5878353714942932, "learning_rate": 9.728208613442993e-05, "loss": 0.0324, "step": 45050 }, { "epoch": 28.718929254302104, "grad_norm": 0.4548419415950775, "learning_rate": 9.728029342917967e-05, "loss": 0.0276, "step": 45060 }, { "epoch": 28.72530274059911, "grad_norm": 0.6807811260223389, "learning_rate": 9.727850014942921e-05, "loss": 0.0422, "step": 45070 }, { "epoch": 28.731676226896113, "grad_norm": 0.9300608038902283, "learning_rate": 9.727670629520034e-05, "loss": 0.0286, "step": 45080 }, { "epoch": 28.738049713193117, "grad_norm": 0.6083965301513672, "learning_rate": 9.727491186651484e-05, "loss": 0.0261, "step": 45090 }, { "epoch": 28.74442319949012, "grad_norm": 0.9100776314735413, "learning_rate": 9.727311686339453e-05, "loss": 0.0308, "step": 45100 }, { "epoch": 28.750796685787126, "grad_norm": 0.4473556578159332, "learning_rate": 9.727132128586122e-05, "loss": 0.0324, "step": 45110 }, { "epoch": 28.75717017208413, "grad_norm": 0.4234658181667328, "learning_rate": 9.726952513393672e-05, "loss": 0.0371, "step": 45120 }, { "epoch": 28.763543658381135, "grad_norm": 0.47622305154800415, "learning_rate": 9.726772840764287e-05, "loss": 0.0383, "step": 45130 }, { "epoch": 28.76991714467814, "grad_norm": 0.5168135166168213, "learning_rate": 9.726593110700148e-05, "loss": 0.0289, "step": 45140 }, { "epoch": 28.776290630975144, "grad_norm": 0.6061549782752991, "learning_rate": 9.72641332320344e-05, "loss": 0.026, "step": 45150 }, { "epoch": 28.782664117272148, "grad_norm": 0.6520146727561951, "learning_rate": 9.726233478276347e-05, "loss": 0.0307, "step": 45160 }, { "epoch": 28.789037603569152, "grad_norm": 0.5421724915504456, "learning_rate": 9.726053575921057e-05, "loss": 0.0409, "step": 45170 }, { "epoch": 28.795411089866157, "grad_norm": 0.5494058132171631, "learning_rate": 9.725873616139753e-05, "loss": 0.0332, "step": 45180 }, { "epoch": 28.80178457616316, "grad_norm": 0.4381203055381775, "learning_rate": 9.725693598934622e-05, "loss": 0.0367, "step": 45190 }, { "epoch": 28.808158062460166, "grad_norm": 0.6294469833374023, "learning_rate": 9.725513524307853e-05, "loss": 0.0264, "step": 45200 }, { "epoch": 28.81453154875717, "grad_norm": 0.41329479217529297, "learning_rate": 9.725333392261634e-05, "loss": 0.0276, "step": 45210 }, { "epoch": 28.820905035054174, "grad_norm": 0.6800749897956848, "learning_rate": 9.725153202798149e-05, "loss": 0.0262, "step": 45220 }, { "epoch": 28.82727852135118, "grad_norm": 0.4792836010456085, "learning_rate": 9.724972955919595e-05, "loss": 0.0226, "step": 45230 }, { "epoch": 28.833652007648183, "grad_norm": 0.46982795000076294, "learning_rate": 9.724792651628156e-05, "loss": 0.0344, "step": 45240 }, { "epoch": 28.840025493945188, "grad_norm": 0.345822274684906, "learning_rate": 9.724612289926028e-05, "loss": 0.0387, "step": 45250 }, { "epoch": 28.846398980242192, "grad_norm": 0.8648724555969238, "learning_rate": 9.724431870815398e-05, "loss": 0.0338, "step": 45260 }, { "epoch": 28.852772466539196, "grad_norm": 0.5537001490592957, "learning_rate": 9.724251394298461e-05, "loss": 0.0252, "step": 45270 }, { "epoch": 28.8591459528362, "grad_norm": 0.41298604011535645, "learning_rate": 9.724070860377408e-05, "loss": 0.0361, "step": 45280 }, { "epoch": 28.865519439133205, "grad_norm": 0.977382481098175, "learning_rate": 9.723890269054434e-05, "loss": 0.0328, "step": 45290 }, { "epoch": 28.87189292543021, "grad_norm": 0.5002893805503845, "learning_rate": 9.723709620331733e-05, "loss": 0.0324, "step": 45300 }, { "epoch": 28.878266411727214, "grad_norm": 0.6122251749038696, "learning_rate": 9.723528914211502e-05, "loss": 0.0458, "step": 45310 }, { "epoch": 28.88463989802422, "grad_norm": 0.35261765122413635, "learning_rate": 9.723348150695934e-05, "loss": 0.0446, "step": 45320 }, { "epoch": 28.891013384321223, "grad_norm": 0.31689614057540894, "learning_rate": 9.723167329787227e-05, "loss": 0.0223, "step": 45330 }, { "epoch": 28.897386870618227, "grad_norm": 0.4325798451900482, "learning_rate": 9.722986451487574e-05, "loss": 0.0291, "step": 45340 }, { "epoch": 28.90376035691523, "grad_norm": 0.5852455496788025, "learning_rate": 9.72280551579918e-05, "loss": 0.0374, "step": 45350 }, { "epoch": 28.910133843212236, "grad_norm": 0.3591966927051544, "learning_rate": 9.72262452272424e-05, "loss": 0.0306, "step": 45360 }, { "epoch": 28.91650732950924, "grad_norm": 0.6264025568962097, "learning_rate": 9.722443472264951e-05, "loss": 0.0419, "step": 45370 }, { "epoch": 28.922880815806245, "grad_norm": 1.1702104806900024, "learning_rate": 9.722262364423517e-05, "loss": 0.0391, "step": 45380 }, { "epoch": 28.92925430210325, "grad_norm": 0.5118148326873779, "learning_rate": 9.722081199202136e-05, "loss": 0.0302, "step": 45390 }, { "epoch": 28.935627788400254, "grad_norm": 0.6835926175117493, "learning_rate": 9.72189997660301e-05, "loss": 0.0392, "step": 45400 }, { "epoch": 28.942001274697258, "grad_norm": 0.47925686836242676, "learning_rate": 9.721718696628341e-05, "loss": 0.0212, "step": 45410 }, { "epoch": 28.948374760994263, "grad_norm": 0.3401205837726593, "learning_rate": 9.721537359280331e-05, "loss": 0.0351, "step": 45420 }, { "epoch": 28.954748247291267, "grad_norm": 0.36152106523513794, "learning_rate": 9.721355964561185e-05, "loss": 0.0207, "step": 45430 }, { "epoch": 28.96112173358827, "grad_norm": 0.46962159872055054, "learning_rate": 9.721174512473107e-05, "loss": 0.0253, "step": 45440 }, { "epoch": 28.967495219885276, "grad_norm": 0.3419113755226135, "learning_rate": 9.7209930030183e-05, "loss": 0.0296, "step": 45450 }, { "epoch": 28.97386870618228, "grad_norm": 0.39809948205947876, "learning_rate": 9.720811436198971e-05, "loss": 0.0382, "step": 45460 }, { "epoch": 28.980242192479285, "grad_norm": 0.44267067313194275, "learning_rate": 9.720629812017327e-05, "loss": 0.0355, "step": 45470 }, { "epoch": 28.98661567877629, "grad_norm": 0.79180508852005, "learning_rate": 9.720448130475572e-05, "loss": 0.03, "step": 45480 }, { "epoch": 28.992989165073293, "grad_norm": 0.4299890995025635, "learning_rate": 9.720266391575916e-05, "loss": 0.0354, "step": 45490 }, { "epoch": 28.999362651370298, "grad_norm": 0.40014269948005676, "learning_rate": 9.720084595320567e-05, "loss": 0.0239, "step": 45500 }, { "epoch": 29.005736137667306, "grad_norm": 0.5993176102638245, "learning_rate": 9.719902741711735e-05, "loss": 0.0298, "step": 45510 }, { "epoch": 29.01210962396431, "grad_norm": 0.6219274997711182, "learning_rate": 9.719720830751627e-05, "loss": 0.0348, "step": 45520 }, { "epoch": 29.018483110261315, "grad_norm": 0.28383973240852356, "learning_rate": 9.719538862442455e-05, "loss": 0.0297, "step": 45530 }, { "epoch": 29.02485659655832, "grad_norm": 0.6099619269371033, "learning_rate": 9.719356836786429e-05, "loss": 0.023, "step": 45540 }, { "epoch": 29.031230082855323, "grad_norm": 0.368915319442749, "learning_rate": 9.719174753785761e-05, "loss": 0.0222, "step": 45550 }, { "epoch": 29.037603569152328, "grad_norm": 0.7281890511512756, "learning_rate": 9.718992613442666e-05, "loss": 0.0282, "step": 45560 }, { "epoch": 29.043977055449332, "grad_norm": 0.5120526552200317, "learning_rate": 9.718810415759354e-05, "loss": 0.0306, "step": 45570 }, { "epoch": 29.050350541746337, "grad_norm": 0.5037716031074524, "learning_rate": 9.718628160738041e-05, "loss": 0.0244, "step": 45580 }, { "epoch": 29.05672402804334, "grad_norm": 0.4021316170692444, "learning_rate": 9.71844584838094e-05, "loss": 0.0355, "step": 45590 }, { "epoch": 29.063097514340345, "grad_norm": 0.448818176984787, "learning_rate": 9.718263478690268e-05, "loss": 0.02, "step": 45600 }, { "epoch": 29.06947100063735, "grad_norm": 0.5142409801483154, "learning_rate": 9.718081051668239e-05, "loss": 0.032, "step": 45610 }, { "epoch": 29.075844486934354, "grad_norm": 1.1114367246627808, "learning_rate": 9.717898567317071e-05, "loss": 0.0336, "step": 45620 }, { "epoch": 29.08221797323136, "grad_norm": 0.5713164806365967, "learning_rate": 9.71771602563898e-05, "loss": 0.0449, "step": 45630 }, { "epoch": 29.088591459528363, "grad_norm": 0.5510401129722595, "learning_rate": 9.717533426636187e-05, "loss": 0.0292, "step": 45640 }, { "epoch": 29.094964945825367, "grad_norm": 0.5720933079719543, "learning_rate": 9.717350770310907e-05, "loss": 0.0311, "step": 45650 }, { "epoch": 29.101338432122372, "grad_norm": 0.874988853931427, "learning_rate": 9.717168056665362e-05, "loss": 0.0362, "step": 45660 }, { "epoch": 29.107711918419376, "grad_norm": 0.9443122148513794, "learning_rate": 9.716985285701772e-05, "loss": 0.0333, "step": 45670 }, { "epoch": 29.11408540471638, "grad_norm": 0.5816103219985962, "learning_rate": 9.716802457422356e-05, "loss": 0.0327, "step": 45680 }, { "epoch": 29.120458891013385, "grad_norm": 0.40928781032562256, "learning_rate": 9.716619571829338e-05, "loss": 0.0412, "step": 45690 }, { "epoch": 29.12683237731039, "grad_norm": 0.5306153297424316, "learning_rate": 9.716436628924937e-05, "loss": 0.032, "step": 45700 }, { "epoch": 29.133205863607394, "grad_norm": 0.39029309153556824, "learning_rate": 9.71625362871138e-05, "loss": 0.0265, "step": 45710 }, { "epoch": 29.1395793499044, "grad_norm": 0.6047243475914001, "learning_rate": 9.716070571190887e-05, "loss": 0.0319, "step": 45720 }, { "epoch": 29.145952836201403, "grad_norm": 0.3802611231803894, "learning_rate": 9.715887456365684e-05, "loss": 0.0301, "step": 45730 }, { "epoch": 29.152326322498407, "grad_norm": 0.472699373960495, "learning_rate": 9.715704284237997e-05, "loss": 0.0226, "step": 45740 }, { "epoch": 29.15869980879541, "grad_norm": 0.41487348079681396, "learning_rate": 9.715521054810051e-05, "loss": 0.0259, "step": 45750 }, { "epoch": 29.165073295092416, "grad_norm": 0.43770483136177063, "learning_rate": 9.71533776808407e-05, "loss": 0.0316, "step": 45760 }, { "epoch": 29.17144678138942, "grad_norm": 0.5364236831665039, "learning_rate": 9.715154424062284e-05, "loss": 0.03, "step": 45770 }, { "epoch": 29.177820267686425, "grad_norm": 0.40557777881622314, "learning_rate": 9.71497102274692e-05, "loss": 0.0229, "step": 45780 }, { "epoch": 29.18419375398343, "grad_norm": 0.6140701770782471, "learning_rate": 9.714787564140206e-05, "loss": 0.0329, "step": 45790 }, { "epoch": 29.190567240280433, "grad_norm": 0.5551201701164246, "learning_rate": 9.714604048244373e-05, "loss": 0.0385, "step": 45800 }, { "epoch": 29.196940726577438, "grad_norm": 0.435820609331131, "learning_rate": 9.714420475061648e-05, "loss": 0.0305, "step": 45810 }, { "epoch": 29.203314212874442, "grad_norm": 0.5099499225616455, "learning_rate": 9.714236844594263e-05, "loss": 0.0314, "step": 45820 }, { "epoch": 29.209687699171447, "grad_norm": 0.326946884393692, "learning_rate": 9.71405315684445e-05, "loss": 0.0322, "step": 45830 }, { "epoch": 29.21606118546845, "grad_norm": 0.5033048987388611, "learning_rate": 9.713869411814441e-05, "loss": 0.0488, "step": 45840 }, { "epoch": 29.222434671765456, "grad_norm": 0.5972860455513, "learning_rate": 9.713685609506467e-05, "loss": 0.0464, "step": 45850 }, { "epoch": 29.22880815806246, "grad_norm": 0.7451785206794739, "learning_rate": 9.713501749922762e-05, "loss": 0.0367, "step": 45860 }, { "epoch": 29.235181644359464, "grad_norm": 0.36853545904159546, "learning_rate": 9.713317833065561e-05, "loss": 0.0298, "step": 45870 }, { "epoch": 29.24155513065647, "grad_norm": 0.4983586072921753, "learning_rate": 9.7131338589371e-05, "loss": 0.0311, "step": 45880 }, { "epoch": 29.247928616953473, "grad_norm": 0.4712297320365906, "learning_rate": 9.712949827539611e-05, "loss": 0.0319, "step": 45890 }, { "epoch": 29.254302103250478, "grad_norm": 0.5224852561950684, "learning_rate": 9.712765738875332e-05, "loss": 0.0331, "step": 45900 }, { "epoch": 29.260675589547482, "grad_norm": 0.7474142909049988, "learning_rate": 9.712581592946499e-05, "loss": 0.0292, "step": 45910 }, { "epoch": 29.267049075844486, "grad_norm": 0.5282825827598572, "learning_rate": 9.71239738975535e-05, "loss": 0.0414, "step": 45920 }, { "epoch": 29.27342256214149, "grad_norm": 0.3555707335472107, "learning_rate": 9.712213129304124e-05, "loss": 0.0287, "step": 45930 }, { "epoch": 29.279796048438495, "grad_norm": 0.36459726095199585, "learning_rate": 9.712028811595059e-05, "loss": 0.024, "step": 45940 }, { "epoch": 29.2861695347355, "grad_norm": 0.4337528944015503, "learning_rate": 9.711844436630397e-05, "loss": 0.0298, "step": 45950 }, { "epoch": 29.292543021032504, "grad_norm": 0.5455673336982727, "learning_rate": 9.711660004412375e-05, "loss": 0.0356, "step": 45960 }, { "epoch": 29.29891650732951, "grad_norm": 0.4543514549732208, "learning_rate": 9.711475514943236e-05, "loss": 0.0243, "step": 45970 }, { "epoch": 29.305289993626513, "grad_norm": 0.3699268698692322, "learning_rate": 9.711290968225219e-05, "loss": 0.0278, "step": 45980 }, { "epoch": 29.311663479923517, "grad_norm": 0.5745609402656555, "learning_rate": 9.711106364260571e-05, "loss": 0.0296, "step": 45990 }, { "epoch": 29.31803696622052, "grad_norm": 0.5258733034133911, "learning_rate": 9.710921703051533e-05, "loss": 0.0418, "step": 46000 }, { "epoch": 29.324410452517526, "grad_norm": 0.4699323773384094, "learning_rate": 9.710736984600346e-05, "loss": 0.03, "step": 46010 }, { "epoch": 29.33078393881453, "grad_norm": 1.4567075967788696, "learning_rate": 9.710552208909258e-05, "loss": 0.0338, "step": 46020 }, { "epoch": 29.337157425111535, "grad_norm": 0.4261760115623474, "learning_rate": 9.710367375980513e-05, "loss": 0.025, "step": 46030 }, { "epoch": 29.34353091140854, "grad_norm": 0.7398717999458313, "learning_rate": 9.710182485816358e-05, "loss": 0.0319, "step": 46040 }, { "epoch": 29.349904397705544, "grad_norm": 0.431633859872818, "learning_rate": 9.709997538419038e-05, "loss": 0.0359, "step": 46050 }, { "epoch": 29.356277884002548, "grad_norm": 0.4918813705444336, "learning_rate": 9.7098125337908e-05, "loss": 0.0283, "step": 46060 }, { "epoch": 29.362651370299552, "grad_norm": 0.6754379272460938, "learning_rate": 9.709627471933896e-05, "loss": 0.0326, "step": 46070 }, { "epoch": 29.369024856596557, "grad_norm": 0.8517892956733704, "learning_rate": 9.709442352850569e-05, "loss": 0.0467, "step": 46080 }, { "epoch": 29.37539834289356, "grad_norm": 0.5752487778663635, "learning_rate": 9.709257176543071e-05, "loss": 0.0313, "step": 46090 }, { "epoch": 29.381771829190566, "grad_norm": 0.3474980592727661, "learning_rate": 9.709071943013652e-05, "loss": 0.0279, "step": 46100 }, { "epoch": 29.38814531548757, "grad_norm": 0.48432061076164246, "learning_rate": 9.708886652264563e-05, "loss": 0.0375, "step": 46110 }, { "epoch": 29.394518801784574, "grad_norm": 0.3884345293045044, "learning_rate": 9.708701304298056e-05, "loss": 0.0472, "step": 46120 }, { "epoch": 29.40089228808158, "grad_norm": 0.492495059967041, "learning_rate": 9.708515899116382e-05, "loss": 0.0305, "step": 46130 }, { "epoch": 29.407265774378587, "grad_norm": 0.6608692407608032, "learning_rate": 9.708330436721794e-05, "loss": 0.0294, "step": 46140 }, { "epoch": 29.41363926067559, "grad_norm": 0.6116952300071716, "learning_rate": 9.708144917116546e-05, "loss": 0.0289, "step": 46150 }, { "epoch": 29.420012746972596, "grad_norm": 0.5953435897827148, "learning_rate": 9.707959340302893e-05, "loss": 0.0313, "step": 46160 }, { "epoch": 29.4263862332696, "grad_norm": 0.6745482683181763, "learning_rate": 9.707773706283087e-05, "loss": 0.0374, "step": 46170 }, { "epoch": 29.432759719566604, "grad_norm": 0.6373438835144043, "learning_rate": 9.707588015059386e-05, "loss": 0.0342, "step": 46180 }, { "epoch": 29.43913320586361, "grad_norm": 0.6488941311836243, "learning_rate": 9.707402266634047e-05, "loss": 0.0357, "step": 46190 }, { "epoch": 29.445506692160613, "grad_norm": 0.4686749279499054, "learning_rate": 9.707216461009325e-05, "loss": 0.0288, "step": 46200 }, { "epoch": 29.451880178457618, "grad_norm": 0.3598555624485016, "learning_rate": 9.70703059818748e-05, "loss": 0.0362, "step": 46210 }, { "epoch": 29.458253664754622, "grad_norm": 0.5486948490142822, "learning_rate": 9.706844678170767e-05, "loss": 0.0353, "step": 46220 }, { "epoch": 29.464627151051626, "grad_norm": 0.45492681860923767, "learning_rate": 9.706658700961448e-05, "loss": 0.0302, "step": 46230 }, { "epoch": 29.47100063734863, "grad_norm": 0.446791410446167, "learning_rate": 9.706472666561783e-05, "loss": 0.0254, "step": 46240 }, { "epoch": 29.477374123645635, "grad_norm": 1.3344231843948364, "learning_rate": 9.70628657497403e-05, "loss": 0.0357, "step": 46250 }, { "epoch": 29.48374760994264, "grad_norm": 0.4148443937301636, "learning_rate": 9.706100426200452e-05, "loss": 0.0242, "step": 46260 }, { "epoch": 29.490121096239644, "grad_norm": 0.6448552012443542, "learning_rate": 9.705914220243313e-05, "loss": 0.0361, "step": 46270 }, { "epoch": 29.49649458253665, "grad_norm": 0.47051575779914856, "learning_rate": 9.705727957104871e-05, "loss": 0.0228, "step": 46280 }, { "epoch": 29.502868068833653, "grad_norm": 0.35242611169815063, "learning_rate": 9.70554163678739e-05, "loss": 0.0294, "step": 46290 }, { "epoch": 29.509241555130657, "grad_norm": 0.6139636635780334, "learning_rate": 9.705355259293137e-05, "loss": 0.0264, "step": 46300 }, { "epoch": 29.51561504142766, "grad_norm": 0.8718332052230835, "learning_rate": 9.705168824624376e-05, "loss": 0.0261, "step": 46310 }, { "epoch": 29.521988527724666, "grad_norm": 0.558094322681427, "learning_rate": 9.70498233278337e-05, "loss": 0.0266, "step": 46320 }, { "epoch": 29.52836201402167, "grad_norm": 0.5590869188308716, "learning_rate": 9.704795783772386e-05, "loss": 0.0277, "step": 46330 }, { "epoch": 29.534735500318675, "grad_norm": 0.37498053908348083, "learning_rate": 9.704609177593691e-05, "loss": 0.0325, "step": 46340 }, { "epoch": 29.54110898661568, "grad_norm": 0.6030009984970093, "learning_rate": 9.704422514249554e-05, "loss": 0.0406, "step": 46350 }, { "epoch": 29.547482472912684, "grad_norm": 0.46655571460723877, "learning_rate": 9.70423579374224e-05, "loss": 0.0394, "step": 46360 }, { "epoch": 29.553855959209688, "grad_norm": 0.5389651656150818, "learning_rate": 9.704049016074022e-05, "loss": 0.036, "step": 46370 }, { "epoch": 29.560229445506693, "grad_norm": 0.5369834899902344, "learning_rate": 9.703862181247166e-05, "loss": 0.0275, "step": 46380 }, { "epoch": 29.566602931803697, "grad_norm": 0.8305455446243286, "learning_rate": 9.703675289263942e-05, "loss": 0.0311, "step": 46390 }, { "epoch": 29.5729764181007, "grad_norm": 0.545233428478241, "learning_rate": 9.703488340126623e-05, "loss": 0.0353, "step": 46400 }, { "epoch": 29.579349904397706, "grad_norm": 0.4404902160167694, "learning_rate": 9.70330133383748e-05, "loss": 0.0298, "step": 46410 }, { "epoch": 29.58572339069471, "grad_norm": 0.6201958060264587, "learning_rate": 9.703114270398785e-05, "loss": 0.0345, "step": 46420 }, { "epoch": 29.592096876991715, "grad_norm": 0.5541747212409973, "learning_rate": 9.702927149812812e-05, "loss": 0.0326, "step": 46430 }, { "epoch": 29.59847036328872, "grad_norm": 0.3940418064594269, "learning_rate": 9.702739972081834e-05, "loss": 0.0277, "step": 46440 }, { "epoch": 29.604843849585723, "grad_norm": 0.4481523633003235, "learning_rate": 9.702552737208124e-05, "loss": 0.0253, "step": 46450 }, { "epoch": 29.611217335882728, "grad_norm": 0.5187958478927612, "learning_rate": 9.702365445193958e-05, "loss": 0.0226, "step": 46460 }, { "epoch": 29.617590822179732, "grad_norm": 0.5064092874526978, "learning_rate": 9.702178096041612e-05, "loss": 0.0262, "step": 46470 }, { "epoch": 29.623964308476737, "grad_norm": 0.2739316523075104, "learning_rate": 9.701990689753365e-05, "loss": 0.0335, "step": 46480 }, { "epoch": 29.63033779477374, "grad_norm": 0.3782416880130768, "learning_rate": 9.701803226331489e-05, "loss": 0.0374, "step": 46490 }, { "epoch": 29.636711281070745, "grad_norm": 0.47051769495010376, "learning_rate": 9.701615705778267e-05, "loss": 0.0262, "step": 46500 }, { "epoch": 29.64308476736775, "grad_norm": 0.7086570858955383, "learning_rate": 9.701428128095972e-05, "loss": 0.0308, "step": 46510 }, { "epoch": 29.649458253664754, "grad_norm": 0.4280010163784027, "learning_rate": 9.701240493286888e-05, "loss": 0.0319, "step": 46520 }, { "epoch": 29.65583173996176, "grad_norm": 0.3365243673324585, "learning_rate": 9.701052801353294e-05, "loss": 0.0236, "step": 46530 }, { "epoch": 29.662205226258763, "grad_norm": 0.4565931558609009, "learning_rate": 9.70086505229747e-05, "loss": 0.0566, "step": 46540 }, { "epoch": 29.668578712555767, "grad_norm": 0.79093998670578, "learning_rate": 9.700677246121695e-05, "loss": 0.0376, "step": 46550 }, { "epoch": 29.674952198852772, "grad_norm": 0.7757982015609741, "learning_rate": 9.700489382828256e-05, "loss": 0.0257, "step": 46560 }, { "epoch": 29.681325685149776, "grad_norm": 0.3275720775127411, "learning_rate": 9.70030146241943e-05, "loss": 0.0223, "step": 46570 }, { "epoch": 29.68769917144678, "grad_norm": 0.5628404021263123, "learning_rate": 9.700113484897506e-05, "loss": 0.0296, "step": 46580 }, { "epoch": 29.694072657743785, "grad_norm": 0.40180811285972595, "learning_rate": 9.699925450264765e-05, "loss": 0.0364, "step": 46590 }, { "epoch": 29.70044614404079, "grad_norm": 0.35429057478904724, "learning_rate": 9.699737358523492e-05, "loss": 0.0252, "step": 46600 }, { "epoch": 29.706819630337794, "grad_norm": 0.356086403131485, "learning_rate": 9.699549209675973e-05, "loss": 0.0274, "step": 46610 }, { "epoch": 29.7131931166348, "grad_norm": 0.4660834074020386, "learning_rate": 9.699361003724495e-05, "loss": 0.0321, "step": 46620 }, { "epoch": 29.719566602931803, "grad_norm": 0.40020546317100525, "learning_rate": 9.699172740671342e-05, "loss": 0.0234, "step": 46630 }, { "epoch": 29.725940089228807, "grad_norm": 0.30107176303863525, "learning_rate": 9.698984420518804e-05, "loss": 0.0268, "step": 46640 }, { "epoch": 29.73231357552581, "grad_norm": 0.36909517645835876, "learning_rate": 9.69879604326917e-05, "loss": 0.0239, "step": 46650 }, { "epoch": 29.738687061822816, "grad_norm": 0.548409104347229, "learning_rate": 9.698607608924726e-05, "loss": 0.0326, "step": 46660 }, { "epoch": 29.74506054811982, "grad_norm": 0.601775050163269, "learning_rate": 9.698419117487762e-05, "loss": 0.025, "step": 46670 }, { "epoch": 29.751434034416825, "grad_norm": 0.7909955978393555, "learning_rate": 9.698230568960574e-05, "loss": 0.0377, "step": 46680 }, { "epoch": 29.75780752071383, "grad_norm": 0.6439356207847595, "learning_rate": 9.698041963345446e-05, "loss": 0.031, "step": 46690 }, { "epoch": 29.764181007010833, "grad_norm": 0.36052584648132324, "learning_rate": 9.697853300644672e-05, "loss": 0.0347, "step": 46700 }, { "epoch": 29.770554493307838, "grad_norm": 0.4040693938732147, "learning_rate": 9.697664580860546e-05, "loss": 0.0343, "step": 46710 }, { "epoch": 29.776927979604842, "grad_norm": 0.6904430389404297, "learning_rate": 9.69747580399536e-05, "loss": 0.0369, "step": 46720 }, { "epoch": 29.783301465901847, "grad_norm": 0.56150221824646, "learning_rate": 9.697286970051407e-05, "loss": 0.0253, "step": 46730 }, { "epoch": 29.78967495219885, "grad_norm": 0.5458069443702698, "learning_rate": 9.697098079030982e-05, "loss": 0.0352, "step": 46740 }, { "epoch": 29.79604843849586, "grad_norm": 0.4143088459968567, "learning_rate": 9.696909130936382e-05, "loss": 0.0261, "step": 46750 }, { "epoch": 29.80242192479286, "grad_norm": 0.4481980800628662, "learning_rate": 9.6967201257699e-05, "loss": 0.0283, "step": 46760 }, { "epoch": 29.808795411089868, "grad_norm": 0.5536223649978638, "learning_rate": 9.696531063533835e-05, "loss": 0.0382, "step": 46770 }, { "epoch": 29.815168897386872, "grad_norm": 0.6695423126220703, "learning_rate": 9.696341944230484e-05, "loss": 0.04, "step": 46780 }, { "epoch": 29.821542383683877, "grad_norm": 0.7825644612312317, "learning_rate": 9.696152767862143e-05, "loss": 0.0335, "step": 46790 }, { "epoch": 29.82791586998088, "grad_norm": 0.5314565896987915, "learning_rate": 9.695963534431113e-05, "loss": 0.0293, "step": 46800 }, { "epoch": 29.834289356277885, "grad_norm": 0.49897485971450806, "learning_rate": 9.695774243939691e-05, "loss": 0.0379, "step": 46810 }, { "epoch": 29.84066284257489, "grad_norm": 0.5014762282371521, "learning_rate": 9.695584896390179e-05, "loss": 0.0538, "step": 46820 }, { "epoch": 29.847036328871894, "grad_norm": 0.6763722896575928, "learning_rate": 9.695395491784878e-05, "loss": 0.0322, "step": 46830 }, { "epoch": 29.8534098151689, "grad_norm": 0.6262781023979187, "learning_rate": 9.695206030126087e-05, "loss": 0.033, "step": 46840 }, { "epoch": 29.859783301465903, "grad_norm": 0.43103867769241333, "learning_rate": 9.695016511416111e-05, "loss": 0.0204, "step": 46850 }, { "epoch": 29.866156787762907, "grad_norm": 0.5196895599365234, "learning_rate": 9.694826935657252e-05, "loss": 0.0274, "step": 46860 }, { "epoch": 29.872530274059912, "grad_norm": 0.6107072234153748, "learning_rate": 9.694637302851813e-05, "loss": 0.0245, "step": 46870 }, { "epoch": 29.878903760356916, "grad_norm": 0.4301851689815521, "learning_rate": 9.694447613002096e-05, "loss": 0.0283, "step": 46880 }, { "epoch": 29.88527724665392, "grad_norm": 0.5175921320915222, "learning_rate": 9.69425786611041e-05, "loss": 0.025, "step": 46890 }, { "epoch": 29.891650732950925, "grad_norm": 0.6600993871688843, "learning_rate": 9.694068062179058e-05, "loss": 0.0286, "step": 46900 }, { "epoch": 29.89802421924793, "grad_norm": 0.5901431441307068, "learning_rate": 9.69387820121035e-05, "loss": 0.0238, "step": 46910 }, { "epoch": 29.904397705544934, "grad_norm": 0.5341496467590332, "learning_rate": 9.693688283206586e-05, "loss": 0.0324, "step": 46920 }, { "epoch": 29.91077119184194, "grad_norm": 0.3416377604007721, "learning_rate": 9.693498308170079e-05, "loss": 0.0302, "step": 46930 }, { "epoch": 29.917144678138943, "grad_norm": 0.43900632858276367, "learning_rate": 9.693308276103137e-05, "loss": 0.0275, "step": 46940 }, { "epoch": 29.923518164435947, "grad_norm": 0.4261423349380493, "learning_rate": 9.693118187008068e-05, "loss": 0.0351, "step": 46950 }, { "epoch": 29.92989165073295, "grad_norm": 0.49773746728897095, "learning_rate": 9.692928040887183e-05, "loss": 0.032, "step": 46960 }, { "epoch": 29.936265137029956, "grad_norm": 0.38618946075439453, "learning_rate": 9.692737837742789e-05, "loss": 0.0211, "step": 46970 }, { "epoch": 29.94263862332696, "grad_norm": 0.35614651441574097, "learning_rate": 9.692547577577202e-05, "loss": 0.0325, "step": 46980 }, { "epoch": 29.949012109623965, "grad_norm": 0.3810688853263855, "learning_rate": 9.69235726039273e-05, "loss": 0.0243, "step": 46990 }, { "epoch": 29.95538559592097, "grad_norm": 0.4994458258152008, "learning_rate": 9.692166886191686e-05, "loss": 0.0244, "step": 47000 }, { "epoch": 29.961759082217974, "grad_norm": 0.6641340255737305, "learning_rate": 9.691976454976387e-05, "loss": 0.0257, "step": 47010 }, { "epoch": 29.968132568514978, "grad_norm": 0.45964425802230835, "learning_rate": 9.691785966749143e-05, "loss": 0.0324, "step": 47020 }, { "epoch": 29.974506054811982, "grad_norm": 0.46489396691322327, "learning_rate": 9.691595421512269e-05, "loss": 0.0298, "step": 47030 }, { "epoch": 29.980879541108987, "grad_norm": 0.37213489413261414, "learning_rate": 9.691404819268082e-05, "loss": 0.033, "step": 47040 }, { "epoch": 29.98725302740599, "grad_norm": 0.33924904465675354, "learning_rate": 9.691214160018895e-05, "loss": 0.0332, "step": 47050 }, { "epoch": 29.993626513702996, "grad_norm": 0.7121188044548035, "learning_rate": 9.691023443767029e-05, "loss": 0.029, "step": 47060 }, { "epoch": 30.0, "grad_norm": 0.5272395014762878, "learning_rate": 9.690832670514798e-05, "loss": 0.0206, "step": 47070 }, { "epoch": 30.006373486297004, "grad_norm": 0.5339630842208862, "learning_rate": 9.690641840264522e-05, "loss": 0.0316, "step": 47080 }, { "epoch": 30.01274697259401, "grad_norm": 0.36145028471946716, "learning_rate": 9.690450953018518e-05, "loss": 0.0208, "step": 47090 }, { "epoch": 30.019120458891013, "grad_norm": 0.231989324092865, "learning_rate": 9.690260008779106e-05, "loss": 0.0254, "step": 47100 }, { "epoch": 30.025493945188018, "grad_norm": 0.44678598642349243, "learning_rate": 9.690069007548608e-05, "loss": 0.0358, "step": 47110 }, { "epoch": 30.031867431485022, "grad_norm": 0.24146738648414612, "learning_rate": 9.689877949329342e-05, "loss": 0.0219, "step": 47120 }, { "epoch": 30.038240917782026, "grad_norm": 0.48522791266441345, "learning_rate": 9.689686834123632e-05, "loss": 0.0241, "step": 47130 }, { "epoch": 30.04461440407903, "grad_norm": 0.393105685710907, "learning_rate": 9.6894956619338e-05, "loss": 0.0265, "step": 47140 }, { "epoch": 30.050987890376035, "grad_norm": 0.5010115504264832, "learning_rate": 9.689304432762167e-05, "loss": 0.0377, "step": 47150 }, { "epoch": 30.05736137667304, "grad_norm": 0.7131636142730713, "learning_rate": 9.689113146611057e-05, "loss": 0.0225, "step": 47160 }, { "epoch": 30.063734862970044, "grad_norm": 0.5025869011878967, "learning_rate": 9.688921803482796e-05, "loss": 0.0331, "step": 47170 }, { "epoch": 30.07010834926705, "grad_norm": 0.319277822971344, "learning_rate": 9.688730403379708e-05, "loss": 0.0359, "step": 47180 }, { "epoch": 30.076481835564053, "grad_norm": 0.45344534516334534, "learning_rate": 9.688538946304118e-05, "loss": 0.0297, "step": 47190 }, { "epoch": 30.082855321861057, "grad_norm": 0.16791537404060364, "learning_rate": 9.688347432258353e-05, "loss": 0.0217, "step": 47200 }, { "epoch": 30.08922880815806, "grad_norm": 0.4167848825454712, "learning_rate": 9.68815586124474e-05, "loss": 0.0239, "step": 47210 }, { "epoch": 30.095602294455066, "grad_norm": 0.4025765359401703, "learning_rate": 9.687964233265608e-05, "loss": 0.02, "step": 47220 }, { "epoch": 30.10197578075207, "grad_norm": 0.5122728943824768, "learning_rate": 9.687772548323284e-05, "loss": 0.0226, "step": 47230 }, { "epoch": 30.108349267049075, "grad_norm": 0.5051019787788391, "learning_rate": 9.687580806420098e-05, "loss": 0.0342, "step": 47240 }, { "epoch": 30.11472275334608, "grad_norm": 0.6490054130554199, "learning_rate": 9.687389007558379e-05, "loss": 0.0265, "step": 47250 }, { "epoch": 30.121096239643084, "grad_norm": 0.4030517637729645, "learning_rate": 9.687197151740457e-05, "loss": 0.0271, "step": 47260 }, { "epoch": 30.127469725940088, "grad_norm": 0.32583263516426086, "learning_rate": 9.687005238968665e-05, "loss": 0.0373, "step": 47270 }, { "epoch": 30.133843212237093, "grad_norm": 0.48657089471817017, "learning_rate": 9.686813269245333e-05, "loss": 0.0434, "step": 47280 }, { "epoch": 30.140216698534097, "grad_norm": 0.4019191265106201, "learning_rate": 9.686621242572797e-05, "loss": 0.0284, "step": 47290 }, { "epoch": 30.1465901848311, "grad_norm": 0.46040210127830505, "learning_rate": 9.686429158953388e-05, "loss": 0.032, "step": 47300 }, { "epoch": 30.152963671128106, "grad_norm": 0.2923794090747833, "learning_rate": 9.686237018389438e-05, "loss": 0.0257, "step": 47310 }, { "epoch": 30.15933715742511, "grad_norm": 0.49327313899993896, "learning_rate": 9.686044820883285e-05, "loss": 0.0253, "step": 47320 }, { "epoch": 30.165710643722115, "grad_norm": 0.5814868807792664, "learning_rate": 9.685852566437261e-05, "loss": 0.0241, "step": 47330 }, { "epoch": 30.17208413001912, "grad_norm": 0.5563207864761353, "learning_rate": 9.685660255053706e-05, "loss": 0.0229, "step": 47340 }, { "epoch": 30.178457616316123, "grad_norm": 0.6287422180175781, "learning_rate": 9.685467886734956e-05, "loss": 0.0275, "step": 47350 }, { "epoch": 30.184831102613128, "grad_norm": 0.6853327751159668, "learning_rate": 9.685275461483346e-05, "loss": 0.0286, "step": 47360 }, { "epoch": 30.191204588910132, "grad_norm": 0.4149956703186035, "learning_rate": 9.685082979301214e-05, "loss": 0.0256, "step": 47370 }, { "epoch": 30.197578075207137, "grad_norm": 0.6666845679283142, "learning_rate": 9.684890440190903e-05, "loss": 0.0431, "step": 47380 }, { "epoch": 30.203951561504145, "grad_norm": 0.7017353177070618, "learning_rate": 9.684697844154747e-05, "loss": 0.0351, "step": 47390 }, { "epoch": 30.21032504780115, "grad_norm": 0.2848077416419983, "learning_rate": 9.684505191195092e-05, "loss": 0.0244, "step": 47400 }, { "epoch": 30.216698534098153, "grad_norm": 0.3562880754470825, "learning_rate": 9.684312481314275e-05, "loss": 0.0313, "step": 47410 }, { "epoch": 30.223072020395158, "grad_norm": 0.3509465157985687, "learning_rate": 9.684119714514638e-05, "loss": 0.0242, "step": 47420 }, { "epoch": 30.229445506692162, "grad_norm": 0.5636443495750427, "learning_rate": 9.683926890798525e-05, "loss": 0.0336, "step": 47430 }, { "epoch": 30.235818992989167, "grad_norm": 0.3845672011375427, "learning_rate": 9.683734010168276e-05, "loss": 0.0279, "step": 47440 }, { "epoch": 30.24219247928617, "grad_norm": 0.6842441558837891, "learning_rate": 9.683541072626239e-05, "loss": 0.0429, "step": 47450 }, { "epoch": 30.248565965583175, "grad_norm": 0.5303451418876648, "learning_rate": 9.683348078174754e-05, "loss": 0.0276, "step": 47460 }, { "epoch": 30.25493945188018, "grad_norm": 0.601003885269165, "learning_rate": 9.683155026816168e-05, "loss": 0.0459, "step": 47470 }, { "epoch": 30.261312938177184, "grad_norm": 0.6152337789535522, "learning_rate": 9.682961918552828e-05, "loss": 0.0302, "step": 47480 }, { "epoch": 30.26768642447419, "grad_norm": 0.8760467767715454, "learning_rate": 9.682768753387079e-05, "loss": 0.041, "step": 47490 }, { "epoch": 30.274059910771193, "grad_norm": 0.3496536612510681, "learning_rate": 9.682575531321269e-05, "loss": 0.0195, "step": 47500 }, { "epoch": 30.280433397068197, "grad_norm": 0.6106433868408203, "learning_rate": 9.682382252357745e-05, "loss": 0.0332, "step": 47510 }, { "epoch": 30.2868068833652, "grad_norm": 0.680897057056427, "learning_rate": 9.682188916498855e-05, "loss": 0.0254, "step": 47520 }, { "epoch": 30.293180369662206, "grad_norm": 0.4990352988243103, "learning_rate": 9.681995523746948e-05, "loss": 0.0291, "step": 47530 }, { "epoch": 30.29955385595921, "grad_norm": 0.3918863832950592, "learning_rate": 9.681802074104376e-05, "loss": 0.0249, "step": 47540 }, { "epoch": 30.305927342256215, "grad_norm": 0.5724256038665771, "learning_rate": 9.681608567573487e-05, "loss": 0.0291, "step": 47550 }, { "epoch": 30.31230082855322, "grad_norm": 0.45661425590515137, "learning_rate": 9.681415004156638e-05, "loss": 0.0343, "step": 47560 }, { "epoch": 30.318674314850224, "grad_norm": 0.42636817693710327, "learning_rate": 9.681221383856173e-05, "loss": 0.0354, "step": 47570 }, { "epoch": 30.325047801147228, "grad_norm": 0.5731152296066284, "learning_rate": 9.68102770667445e-05, "loss": 0.0294, "step": 47580 }, { "epoch": 30.331421287444233, "grad_norm": 0.629697859287262, "learning_rate": 9.680833972613818e-05, "loss": 0.0381, "step": 47590 }, { "epoch": 30.337794773741237, "grad_norm": 0.6697096228599548, "learning_rate": 9.680640181676636e-05, "loss": 0.0222, "step": 47600 }, { "epoch": 30.34416826003824, "grad_norm": 0.5257476568222046, "learning_rate": 9.680446333865258e-05, "loss": 0.0547, "step": 47610 }, { "epoch": 30.350541746335246, "grad_norm": 0.6578401923179626, "learning_rate": 9.680252429182035e-05, "loss": 0.025, "step": 47620 }, { "epoch": 30.35691523263225, "grad_norm": 0.6949588656425476, "learning_rate": 9.680058467629328e-05, "loss": 0.0324, "step": 47630 }, { "epoch": 30.363288718929255, "grad_norm": 0.4209093749523163, "learning_rate": 9.679864449209491e-05, "loss": 0.0481, "step": 47640 }, { "epoch": 30.36966220522626, "grad_norm": 0.4853968024253845, "learning_rate": 9.679670373924883e-05, "loss": 0.0282, "step": 47650 }, { "epoch": 30.376035691523263, "grad_norm": 0.30764615535736084, "learning_rate": 9.67947624177786e-05, "loss": 0.031, "step": 47660 }, { "epoch": 30.382409177820268, "grad_norm": 0.4162934720516205, "learning_rate": 9.679282052770785e-05, "loss": 0.0238, "step": 47670 }, { "epoch": 30.388782664117272, "grad_norm": 0.2692345678806305, "learning_rate": 9.679087806906014e-05, "loss": 0.0221, "step": 47680 }, { "epoch": 30.395156150414277, "grad_norm": 0.523062527179718, "learning_rate": 9.678893504185908e-05, "loss": 0.0315, "step": 47690 }, { "epoch": 30.40152963671128, "grad_norm": 0.4525114893913269, "learning_rate": 9.678699144612828e-05, "loss": 0.0275, "step": 47700 }, { "epoch": 30.407903123008285, "grad_norm": 0.42778345942497253, "learning_rate": 9.678504728189138e-05, "loss": 0.0178, "step": 47710 }, { "epoch": 30.41427660930529, "grad_norm": 0.6708573698997498, "learning_rate": 9.678310254917197e-05, "loss": 0.022, "step": 47720 }, { "epoch": 30.420650095602294, "grad_norm": 0.45474258065223694, "learning_rate": 9.678115724799369e-05, "loss": 0.0292, "step": 47730 }, { "epoch": 30.4270235818993, "grad_norm": 0.6631261706352234, "learning_rate": 9.677921137838018e-05, "loss": 0.0276, "step": 47740 }, { "epoch": 30.433397068196303, "grad_norm": 0.5106426477432251, "learning_rate": 9.677726494035508e-05, "loss": 0.0388, "step": 47750 }, { "epoch": 30.439770554493307, "grad_norm": 0.4893629550933838, "learning_rate": 9.677531793394205e-05, "loss": 0.0254, "step": 47760 }, { "epoch": 30.446144040790312, "grad_norm": 0.5849584937095642, "learning_rate": 9.677337035916477e-05, "loss": 0.0294, "step": 47770 }, { "epoch": 30.452517527087316, "grad_norm": 0.6654514670372009, "learning_rate": 9.677142221604685e-05, "loss": 0.0266, "step": 47780 }, { "epoch": 30.45889101338432, "grad_norm": 0.5773518681526184, "learning_rate": 9.676947350461199e-05, "loss": 0.0419, "step": 47790 }, { "epoch": 30.465264499681325, "grad_norm": 0.41494619846343994, "learning_rate": 9.676752422488388e-05, "loss": 0.0193, "step": 47800 }, { "epoch": 30.47163798597833, "grad_norm": 0.5592244267463684, "learning_rate": 9.67655743768862e-05, "loss": 0.0266, "step": 47810 }, { "epoch": 30.478011472275334, "grad_norm": 0.29879140853881836, "learning_rate": 9.676362396064263e-05, "loss": 0.0211, "step": 47820 }, { "epoch": 30.48438495857234, "grad_norm": 0.47534358501434326, "learning_rate": 9.676167297617686e-05, "loss": 0.0185, "step": 47830 }, { "epoch": 30.490758444869343, "grad_norm": 0.3677673637866974, "learning_rate": 9.675972142351264e-05, "loss": 0.0446, "step": 47840 }, { "epoch": 30.497131931166347, "grad_norm": 0.5857642889022827, "learning_rate": 9.675776930267364e-05, "loss": 0.0283, "step": 47850 }, { "epoch": 30.50350541746335, "grad_norm": 0.3543374538421631, "learning_rate": 9.67558166136836e-05, "loss": 0.0477, "step": 47860 }, { "epoch": 30.509878903760356, "grad_norm": 0.3732214570045471, "learning_rate": 9.675386335656624e-05, "loss": 0.0291, "step": 47870 }, { "epoch": 30.51625239005736, "grad_norm": 0.43065503239631653, "learning_rate": 9.67519095313453e-05, "loss": 0.0183, "step": 47880 }, { "epoch": 30.522625876354365, "grad_norm": 0.6613799929618835, "learning_rate": 9.674995513804452e-05, "loss": 0.0299, "step": 47890 }, { "epoch": 30.52899936265137, "grad_norm": 0.3799893856048584, "learning_rate": 9.674800017668765e-05, "loss": 0.0304, "step": 47900 }, { "epoch": 30.535372848948374, "grad_norm": 0.6712312698364258, "learning_rate": 9.674604464729844e-05, "loss": 0.0361, "step": 47910 }, { "epoch": 30.541746335245378, "grad_norm": 0.5331726670265198, "learning_rate": 9.674408854990064e-05, "loss": 0.0285, "step": 47920 }, { "epoch": 30.548119821542382, "grad_norm": 0.7846921682357788, "learning_rate": 9.674213188451804e-05, "loss": 0.0327, "step": 47930 }, { "epoch": 30.554493307839387, "grad_norm": 0.414529025554657, "learning_rate": 9.67401746511744e-05, "loss": 0.0283, "step": 47940 }, { "epoch": 30.56086679413639, "grad_norm": 0.5281225442886353, "learning_rate": 9.673821684989352e-05, "loss": 0.0289, "step": 47950 }, { "epoch": 30.567240280433396, "grad_norm": 0.6101946234703064, "learning_rate": 9.673625848069916e-05, "loss": 0.0224, "step": 47960 }, { "epoch": 30.5736137667304, "grad_norm": 0.4984627068042755, "learning_rate": 9.673429954361514e-05, "loss": 0.0316, "step": 47970 }, { "epoch": 30.579987253027404, "grad_norm": 0.525843620300293, "learning_rate": 9.673234003866526e-05, "loss": 0.0291, "step": 47980 }, { "epoch": 30.58636073932441, "grad_norm": 0.6882246732711792, "learning_rate": 9.673037996587333e-05, "loss": 0.0268, "step": 47990 }, { "epoch": 30.592734225621413, "grad_norm": 0.4233173727989197, "learning_rate": 9.672841932526315e-05, "loss": 0.0426, "step": 48000 }, { "epoch": 30.59910771191842, "grad_norm": 0.8273032903671265, "learning_rate": 9.672645811685857e-05, "loss": 0.0248, "step": 48010 }, { "epoch": 30.605481198215422, "grad_norm": 0.617384672164917, "learning_rate": 9.672449634068339e-05, "loss": 0.0301, "step": 48020 }, { "epoch": 30.61185468451243, "grad_norm": 0.6358820199966431, "learning_rate": 9.672253399676148e-05, "loss": 0.0378, "step": 48030 }, { "epoch": 30.618228170809434, "grad_norm": 0.41155993938446045, "learning_rate": 9.672057108511665e-05, "loss": 0.0319, "step": 48040 }, { "epoch": 30.62460165710644, "grad_norm": 0.6127799153327942, "learning_rate": 9.671860760577279e-05, "loss": 0.0313, "step": 48050 }, { "epoch": 30.630975143403443, "grad_norm": 0.5372704267501831, "learning_rate": 9.671664355875373e-05, "loss": 0.0345, "step": 48060 }, { "epoch": 30.637348629700448, "grad_norm": 0.6143667697906494, "learning_rate": 9.671467894408335e-05, "loss": 0.0288, "step": 48070 }, { "epoch": 30.643722115997452, "grad_norm": 0.6638042330741882, "learning_rate": 9.671271376178551e-05, "loss": 0.0431, "step": 48080 }, { "epoch": 30.650095602294456, "grad_norm": 0.566862940788269, "learning_rate": 9.671074801188408e-05, "loss": 0.0333, "step": 48090 }, { "epoch": 30.65646908859146, "grad_norm": 0.7406484484672546, "learning_rate": 9.670878169440297e-05, "loss": 0.0404, "step": 48100 }, { "epoch": 30.662842574888465, "grad_norm": 0.38298800587654114, "learning_rate": 9.670681480936606e-05, "loss": 0.0245, "step": 48110 }, { "epoch": 30.66921606118547, "grad_norm": 0.6056684255599976, "learning_rate": 9.670484735679725e-05, "loss": 0.0394, "step": 48120 }, { "epoch": 30.675589547482474, "grad_norm": 0.6433191895484924, "learning_rate": 9.670287933672046e-05, "loss": 0.0307, "step": 48130 }, { "epoch": 30.68196303377948, "grad_norm": 0.41848137974739075, "learning_rate": 9.670091074915959e-05, "loss": 0.0196, "step": 48140 }, { "epoch": 30.688336520076483, "grad_norm": 0.430083692073822, "learning_rate": 9.669894159413854e-05, "loss": 0.022, "step": 48150 }, { "epoch": 30.694710006373487, "grad_norm": 0.4374866187572479, "learning_rate": 9.669697187168127e-05, "loss": 0.0303, "step": 48160 }, { "epoch": 30.70108349267049, "grad_norm": 0.47613564133644104, "learning_rate": 9.669500158181172e-05, "loss": 0.0347, "step": 48170 }, { "epoch": 30.707456978967496, "grad_norm": 0.7548603415489197, "learning_rate": 9.66930307245538e-05, "loss": 0.0363, "step": 48180 }, { "epoch": 30.7138304652645, "grad_norm": 0.7053880095481873, "learning_rate": 9.669105929993147e-05, "loss": 0.0312, "step": 48190 }, { "epoch": 30.720203951561505, "grad_norm": 0.4214024841785431, "learning_rate": 9.668908730796868e-05, "loss": 0.0274, "step": 48200 }, { "epoch": 30.72657743785851, "grad_norm": 0.5566748976707458, "learning_rate": 9.668711474868941e-05, "loss": 0.0334, "step": 48210 }, { "epoch": 30.732950924155514, "grad_norm": 0.8187327980995178, "learning_rate": 9.668514162211761e-05, "loss": 0.0335, "step": 48220 }, { "epoch": 30.739324410452518, "grad_norm": 0.3835656940937042, "learning_rate": 9.668316792827725e-05, "loss": 0.0172, "step": 48230 }, { "epoch": 30.745697896749522, "grad_norm": 0.4533531367778778, "learning_rate": 9.668119366719235e-05, "loss": 0.0358, "step": 48240 }, { "epoch": 30.752071383046527, "grad_norm": 1.1595898866653442, "learning_rate": 9.667921883888685e-05, "loss": 0.0411, "step": 48250 }, { "epoch": 30.75844486934353, "grad_norm": 0.4471672773361206, "learning_rate": 9.667724344338478e-05, "loss": 0.039, "step": 48260 }, { "epoch": 30.764818355640536, "grad_norm": 0.4533998966217041, "learning_rate": 9.667526748071011e-05, "loss": 0.0374, "step": 48270 }, { "epoch": 30.77119184193754, "grad_norm": 0.4383496046066284, "learning_rate": 9.66732909508869e-05, "loss": 0.0223, "step": 48280 }, { "epoch": 30.777565328234544, "grad_norm": 0.4665463864803314, "learning_rate": 9.667131385393913e-05, "loss": 0.0256, "step": 48290 }, { "epoch": 30.78393881453155, "grad_norm": 0.48245003819465637, "learning_rate": 9.666933618989082e-05, "loss": 0.0339, "step": 48300 }, { "epoch": 30.790312300828553, "grad_norm": 0.5799371600151062, "learning_rate": 9.666735795876602e-05, "loss": 0.0197, "step": 48310 }, { "epoch": 30.796685787125558, "grad_norm": 0.4700395464897156, "learning_rate": 9.666537916058876e-05, "loss": 0.0271, "step": 48320 }, { "epoch": 30.803059273422562, "grad_norm": 0.769071102142334, "learning_rate": 9.666339979538308e-05, "loss": 0.0263, "step": 48330 }, { "epoch": 30.809432759719567, "grad_norm": 0.5273532271385193, "learning_rate": 9.666141986317305e-05, "loss": 0.0193, "step": 48340 }, { "epoch": 30.81580624601657, "grad_norm": 0.43907466530799866, "learning_rate": 9.66594393639827e-05, "loss": 0.0274, "step": 48350 }, { "epoch": 30.822179732313575, "grad_norm": 0.43947866559028625, "learning_rate": 9.66574582978361e-05, "loss": 0.0273, "step": 48360 }, { "epoch": 30.82855321861058, "grad_norm": 0.7551102638244629, "learning_rate": 9.665547666475734e-05, "loss": 0.0238, "step": 48370 }, { "epoch": 30.834926704907584, "grad_norm": 0.6669016480445862, "learning_rate": 9.66534944647705e-05, "loss": 0.0353, "step": 48380 }, { "epoch": 30.84130019120459, "grad_norm": 0.659733235836029, "learning_rate": 9.665151169789964e-05, "loss": 0.0287, "step": 48390 }, { "epoch": 30.847673677501593, "grad_norm": 0.34366127848625183, "learning_rate": 9.664952836416887e-05, "loss": 0.0295, "step": 48400 }, { "epoch": 30.854047163798597, "grad_norm": 0.5763676166534424, "learning_rate": 9.664754446360227e-05, "loss": 0.0306, "step": 48410 }, { "epoch": 30.8604206500956, "grad_norm": 0.45096004009246826, "learning_rate": 9.664555999622397e-05, "loss": 0.0205, "step": 48420 }, { "epoch": 30.866794136392606, "grad_norm": 0.5856727361679077, "learning_rate": 9.66435749620581e-05, "loss": 0.0285, "step": 48430 }, { "epoch": 30.87316762268961, "grad_norm": 0.6432901620864868, "learning_rate": 9.664158936112873e-05, "loss": 0.0362, "step": 48440 }, { "epoch": 30.879541108986615, "grad_norm": 0.34328076243400574, "learning_rate": 9.663960319346002e-05, "loss": 0.0254, "step": 48450 }, { "epoch": 30.88591459528362, "grad_norm": 0.4030042290687561, "learning_rate": 9.663761645907609e-05, "loss": 0.0302, "step": 48460 }, { "epoch": 30.892288081580624, "grad_norm": 0.41659975051879883, "learning_rate": 9.66356291580011e-05, "loss": 0.0387, "step": 48470 }, { "epoch": 30.898661567877628, "grad_norm": 0.5034846663475037, "learning_rate": 9.663364129025919e-05, "loss": 0.0307, "step": 48480 }, { "epoch": 30.905035054174633, "grad_norm": 0.5030421614646912, "learning_rate": 9.663165285587448e-05, "loss": 0.035, "step": 48490 }, { "epoch": 30.911408540471637, "grad_norm": 0.4905811548233032, "learning_rate": 9.662966385487118e-05, "loss": 0.0312, "step": 48500 }, { "epoch": 30.91778202676864, "grad_norm": 0.37994247674942017, "learning_rate": 9.662767428727343e-05, "loss": 0.0302, "step": 48510 }, { "epoch": 30.924155513065646, "grad_norm": 0.5185877084732056, "learning_rate": 9.662568415310543e-05, "loss": 0.0338, "step": 48520 }, { "epoch": 30.93052899936265, "grad_norm": 0.5717117190361023, "learning_rate": 9.662369345239132e-05, "loss": 0.0225, "step": 48530 }, { "epoch": 30.936902485659655, "grad_norm": 0.7184933423995972, "learning_rate": 9.662170218515533e-05, "loss": 0.0377, "step": 48540 }, { "epoch": 30.94327597195666, "grad_norm": 0.5517489314079285, "learning_rate": 9.661971035142164e-05, "loss": 0.0238, "step": 48550 }, { "epoch": 30.949649458253663, "grad_norm": 0.3884011209011078, "learning_rate": 9.661771795121444e-05, "loss": 0.0413, "step": 48560 }, { "epoch": 30.956022944550668, "grad_norm": 0.2566922605037689, "learning_rate": 9.661572498455798e-05, "loss": 0.0261, "step": 48570 }, { "epoch": 30.962396430847672, "grad_norm": 0.35354575514793396, "learning_rate": 9.661373145147643e-05, "loss": 0.0553, "step": 48580 }, { "epoch": 30.968769917144677, "grad_norm": 0.8680574297904968, "learning_rate": 9.661173735199402e-05, "loss": 0.0394, "step": 48590 }, { "epoch": 30.97514340344168, "grad_norm": 0.34516721963882446, "learning_rate": 9.660974268613501e-05, "loss": 0.025, "step": 48600 }, { "epoch": 30.981516889738685, "grad_norm": 0.3806225657463074, "learning_rate": 9.660774745392363e-05, "loss": 0.0274, "step": 48610 }, { "epoch": 30.98789037603569, "grad_norm": 0.884701132774353, "learning_rate": 9.66057516553841e-05, "loss": 0.0649, "step": 48620 }, { "epoch": 30.994263862332694, "grad_norm": 0.5785786509513855, "learning_rate": 9.660375529054068e-05, "loss": 0.0295, "step": 48630 }, { "epoch": 31.000637348629702, "grad_norm": 0.4000411033630371, "learning_rate": 9.660175835941763e-05, "loss": 0.0262, "step": 48640 }, { "epoch": 31.007010834926707, "grad_norm": 0.5221824049949646, "learning_rate": 9.659976086203923e-05, "loss": 0.0242, "step": 48650 }, { "epoch": 31.01338432122371, "grad_norm": 0.6969224810600281, "learning_rate": 9.659776279842972e-05, "loss": 0.0306, "step": 48660 }, { "epoch": 31.019757807520715, "grad_norm": 0.32996734976768494, "learning_rate": 9.659576416861342e-05, "loss": 0.0218, "step": 48670 }, { "epoch": 31.02613129381772, "grad_norm": 0.5652149319648743, "learning_rate": 9.659376497261457e-05, "loss": 0.0307, "step": 48680 }, { "epoch": 31.032504780114724, "grad_norm": 0.6288396716117859, "learning_rate": 9.65917652104575e-05, "loss": 0.0291, "step": 48690 }, { "epoch": 31.03887826641173, "grad_norm": 0.9513260722160339, "learning_rate": 9.658976488216648e-05, "loss": 0.036, "step": 48700 }, { "epoch": 31.045251752708733, "grad_norm": 0.40537938475608826, "learning_rate": 9.658776398776583e-05, "loss": 0.0283, "step": 48710 }, { "epoch": 31.051625239005737, "grad_norm": 0.5784873366355896, "learning_rate": 9.658576252727987e-05, "loss": 0.0272, "step": 48720 }, { "epoch": 31.057998725302742, "grad_norm": 0.425845742225647, "learning_rate": 9.65837605007329e-05, "loss": 0.0255, "step": 48730 }, { "epoch": 31.064372211599746, "grad_norm": 0.6609985828399658, "learning_rate": 9.658175790814927e-05, "loss": 0.0416, "step": 48740 }, { "epoch": 31.07074569789675, "grad_norm": 0.5502713322639465, "learning_rate": 9.65797547495533e-05, "loss": 0.0394, "step": 48750 }, { "epoch": 31.077119184193755, "grad_norm": 0.43580299615859985, "learning_rate": 9.657775102496933e-05, "loss": 0.0243, "step": 48760 }, { "epoch": 31.08349267049076, "grad_norm": 0.4713056683540344, "learning_rate": 9.65757467344217e-05, "loss": 0.028, "step": 48770 }, { "epoch": 31.089866156787764, "grad_norm": 0.5350761413574219, "learning_rate": 9.657374187793477e-05, "loss": 0.0302, "step": 48780 }, { "epoch": 31.09623964308477, "grad_norm": 0.6443381309509277, "learning_rate": 9.65717364555329e-05, "loss": 0.0203, "step": 48790 }, { "epoch": 31.102613129381773, "grad_norm": 0.6345484256744385, "learning_rate": 9.656973046724048e-05, "loss": 0.0217, "step": 48800 }, { "epoch": 31.108986615678777, "grad_norm": 0.6308213472366333, "learning_rate": 9.656772391308186e-05, "loss": 0.028, "step": 48810 }, { "epoch": 31.11536010197578, "grad_norm": 0.39703309535980225, "learning_rate": 9.656571679308141e-05, "loss": 0.0262, "step": 48820 }, { "epoch": 31.121733588272786, "grad_norm": 0.3152744770050049, "learning_rate": 9.656370910726355e-05, "loss": 0.0406, "step": 48830 }, { "epoch": 31.12810707456979, "grad_norm": 0.5273758172988892, "learning_rate": 9.656170085565269e-05, "loss": 0.037, "step": 48840 }, { "epoch": 31.134480560866795, "grad_norm": 0.47815024852752686, "learning_rate": 9.655969203827316e-05, "loss": 0.0305, "step": 48850 }, { "epoch": 31.1408540471638, "grad_norm": 0.7427622675895691, "learning_rate": 9.655768265514941e-05, "loss": 0.0353, "step": 48860 }, { "epoch": 31.147227533460804, "grad_norm": 0.32691600918769836, "learning_rate": 9.655567270630589e-05, "loss": 0.0314, "step": 48870 }, { "epoch": 31.153601019757808, "grad_norm": 0.7550122737884521, "learning_rate": 9.655366219176696e-05, "loss": 0.0277, "step": 48880 }, { "epoch": 31.159974506054812, "grad_norm": 0.4847235083580017, "learning_rate": 9.65516511115571e-05, "loss": 0.0366, "step": 48890 }, { "epoch": 31.166347992351817, "grad_norm": 0.5646427869796753, "learning_rate": 9.654963946570073e-05, "loss": 0.0253, "step": 48900 }, { "epoch": 31.17272147864882, "grad_norm": 0.41712501645088196, "learning_rate": 9.654762725422229e-05, "loss": 0.0399, "step": 48910 }, { "epoch": 31.179094964945826, "grad_norm": 0.462252676486969, "learning_rate": 9.654561447714621e-05, "loss": 0.0357, "step": 48920 }, { "epoch": 31.18546845124283, "grad_norm": 0.36987707018852234, "learning_rate": 9.654360113449698e-05, "loss": 0.0295, "step": 48930 }, { "epoch": 31.191841937539834, "grad_norm": 0.4034726619720459, "learning_rate": 9.654158722629906e-05, "loss": 0.0362, "step": 48940 }, { "epoch": 31.19821542383684, "grad_norm": 0.506635844707489, "learning_rate": 9.65395727525769e-05, "loss": 0.0268, "step": 48950 }, { "epoch": 31.204588910133843, "grad_norm": 0.7329914569854736, "learning_rate": 9.653755771335499e-05, "loss": 0.032, "step": 48960 }, { "epoch": 31.210962396430848, "grad_norm": 0.6020264029502869, "learning_rate": 9.653554210865783e-05, "loss": 0.0394, "step": 48970 }, { "epoch": 31.217335882727852, "grad_norm": 0.6978441476821899, "learning_rate": 9.653352593850988e-05, "loss": 0.0305, "step": 48980 }, { "epoch": 31.223709369024856, "grad_norm": 0.4542914032936096, "learning_rate": 9.653150920293568e-05, "loss": 0.0296, "step": 48990 }, { "epoch": 31.23008285532186, "grad_norm": 0.47818824648857117, "learning_rate": 9.652949190195967e-05, "loss": 0.0289, "step": 49000 }, { "epoch": 31.236456341618865, "grad_norm": 0.6142327785491943, "learning_rate": 9.652747403560644e-05, "loss": 0.0319, "step": 49010 }, { "epoch": 31.24282982791587, "grad_norm": 0.42436355352401733, "learning_rate": 9.652545560390045e-05, "loss": 0.0377, "step": 49020 }, { "epoch": 31.249203314212874, "grad_norm": 0.33607783913612366, "learning_rate": 9.652343660686626e-05, "loss": 0.0272, "step": 49030 }, { "epoch": 31.25557680050988, "grad_norm": 0.7621344923973083, "learning_rate": 9.652141704452837e-05, "loss": 0.0289, "step": 49040 }, { "epoch": 31.261950286806883, "grad_norm": 0.6974853873252869, "learning_rate": 9.651939691691137e-05, "loss": 0.0288, "step": 49050 }, { "epoch": 31.268323773103887, "grad_norm": 0.3906075954437256, "learning_rate": 9.651737622403974e-05, "loss": 0.0339, "step": 49060 }, { "epoch": 31.27469725940089, "grad_norm": 0.47291985154151917, "learning_rate": 9.651535496593809e-05, "loss": 0.0233, "step": 49070 }, { "epoch": 31.281070745697896, "grad_norm": 0.2513379454612732, "learning_rate": 9.651333314263096e-05, "loss": 0.0318, "step": 49080 }, { "epoch": 31.2874442319949, "grad_norm": 0.8028595447540283, "learning_rate": 9.651131075414292e-05, "loss": 0.0311, "step": 49090 }, { "epoch": 31.293817718291905, "grad_norm": 0.4388154447078705, "learning_rate": 9.650928780049853e-05, "loss": 0.0339, "step": 49100 }, { "epoch": 31.30019120458891, "grad_norm": 0.41119831800460815, "learning_rate": 9.650726428172239e-05, "loss": 0.0297, "step": 49110 }, { "epoch": 31.306564690885914, "grad_norm": 0.3754115700721741, "learning_rate": 9.650524019783906e-05, "loss": 0.0333, "step": 49120 }, { "epoch": 31.312938177182918, "grad_norm": 0.3825179636478424, "learning_rate": 9.650321554887318e-05, "loss": 0.0289, "step": 49130 }, { "epoch": 31.319311663479922, "grad_norm": 0.6352419257164001, "learning_rate": 9.65011903348493e-05, "loss": 0.0363, "step": 49140 }, { "epoch": 31.325685149776927, "grad_norm": 0.8752667903900146, "learning_rate": 9.649916455579206e-05, "loss": 0.0317, "step": 49150 }, { "epoch": 31.33205863607393, "grad_norm": 0.575877845287323, "learning_rate": 9.649713821172608e-05, "loss": 0.0234, "step": 49160 }, { "epoch": 31.338432122370936, "grad_norm": 0.4051464796066284, "learning_rate": 9.649511130267597e-05, "loss": 0.0307, "step": 49170 }, { "epoch": 31.34480560866794, "grad_norm": 0.32039958238601685, "learning_rate": 9.649308382866635e-05, "loss": 0.031, "step": 49180 }, { "epoch": 31.351179094964944, "grad_norm": 0.4514513313770294, "learning_rate": 9.649105578972186e-05, "loss": 0.0222, "step": 49190 }, { "epoch": 31.35755258126195, "grad_norm": 0.5966686010360718, "learning_rate": 9.648902718586717e-05, "loss": 0.0307, "step": 49200 }, { "epoch": 31.363926067558953, "grad_norm": 0.40468645095825195, "learning_rate": 9.648699801712688e-05, "loss": 0.0198, "step": 49210 }, { "epoch": 31.370299553855958, "grad_norm": 0.7375828623771667, "learning_rate": 9.648496828352569e-05, "loss": 0.0335, "step": 49220 }, { "epoch": 31.376673040152962, "grad_norm": 0.6003609299659729, "learning_rate": 9.648293798508823e-05, "loss": 0.0257, "step": 49230 }, { "epoch": 31.383046526449967, "grad_norm": 0.508283257484436, "learning_rate": 9.648090712183919e-05, "loss": 0.0287, "step": 49240 }, { "epoch": 31.38942001274697, "grad_norm": 0.3484508693218231, "learning_rate": 9.647887569380326e-05, "loss": 0.0257, "step": 49250 }, { "epoch": 31.395793499043975, "grad_norm": 0.5096694827079773, "learning_rate": 9.64768437010051e-05, "loss": 0.0362, "step": 49260 }, { "epoch": 31.402166985340983, "grad_norm": 0.3822021186351776, "learning_rate": 9.647481114346941e-05, "loss": 0.0237, "step": 49270 }, { "epoch": 31.408540471637988, "grad_norm": 0.6328350305557251, "learning_rate": 9.647277802122087e-05, "loss": 0.0289, "step": 49280 }, { "epoch": 31.414913957934992, "grad_norm": 0.3835970163345337, "learning_rate": 9.647074433428421e-05, "loss": 0.0291, "step": 49290 }, { "epoch": 31.421287444231996, "grad_norm": 0.6144000291824341, "learning_rate": 9.646871008268412e-05, "loss": 0.0254, "step": 49300 }, { "epoch": 31.427660930529, "grad_norm": 0.45338183641433716, "learning_rate": 9.646667526644535e-05, "loss": 0.0207, "step": 49310 }, { "epoch": 31.434034416826005, "grad_norm": 0.5927661657333374, "learning_rate": 9.64646398855926e-05, "loss": 0.0236, "step": 49320 }, { "epoch": 31.44040790312301, "grad_norm": 0.8147389888763428, "learning_rate": 9.646260394015059e-05, "loss": 0.0296, "step": 49330 }, { "epoch": 31.446781389420014, "grad_norm": 0.7373713254928589, "learning_rate": 9.646056743014407e-05, "loss": 0.0262, "step": 49340 }, { "epoch": 31.45315487571702, "grad_norm": 0.3558279275894165, "learning_rate": 9.645853035559781e-05, "loss": 0.0354, "step": 49350 }, { "epoch": 31.459528362014023, "grad_norm": 0.5918961763381958, "learning_rate": 9.645649271653654e-05, "loss": 0.0371, "step": 49360 }, { "epoch": 31.465901848311027, "grad_norm": 0.32327085733413696, "learning_rate": 9.645445451298501e-05, "loss": 0.0251, "step": 49370 }, { "epoch": 31.47227533460803, "grad_norm": 0.26392844319343567, "learning_rate": 9.6452415744968e-05, "loss": 0.0222, "step": 49380 }, { "epoch": 31.478648820905036, "grad_norm": 0.5202863216400146, "learning_rate": 9.645037641251029e-05, "loss": 0.0385, "step": 49390 }, { "epoch": 31.48502230720204, "grad_norm": 1.6697717905044556, "learning_rate": 9.644833651563664e-05, "loss": 0.0412, "step": 49400 }, { "epoch": 31.491395793499045, "grad_norm": 0.4944114089012146, "learning_rate": 9.644629605437186e-05, "loss": 0.0289, "step": 49410 }, { "epoch": 31.49776927979605, "grad_norm": 0.5295724868774414, "learning_rate": 9.64442550287407e-05, "loss": 0.0276, "step": 49420 }, { "epoch": 31.504142766093054, "grad_norm": 0.4605574607849121, "learning_rate": 9.644221343876801e-05, "loss": 0.0268, "step": 49430 }, { "epoch": 31.510516252390058, "grad_norm": 0.35252270102500916, "learning_rate": 9.644017128447859e-05, "loss": 0.0241, "step": 49440 }, { "epoch": 31.516889738687063, "grad_norm": 0.6916472315788269, "learning_rate": 9.643812856589723e-05, "loss": 0.0319, "step": 49450 }, { "epoch": 31.523263224984067, "grad_norm": 0.5201977491378784, "learning_rate": 9.643608528304877e-05, "loss": 0.0324, "step": 49460 }, { "epoch": 31.52963671128107, "grad_norm": 0.34854909777641296, "learning_rate": 9.643404143595801e-05, "loss": 0.0199, "step": 49470 }, { "epoch": 31.536010197578076, "grad_norm": 0.40513482689857483, "learning_rate": 9.643199702464982e-05, "loss": 0.0243, "step": 49480 }, { "epoch": 31.54238368387508, "grad_norm": 0.4340430796146393, "learning_rate": 9.642995204914902e-05, "loss": 0.0225, "step": 49490 }, { "epoch": 31.548757170172085, "grad_norm": 1.0601682662963867, "learning_rate": 9.642790650948048e-05, "loss": 0.0359, "step": 49500 }, { "epoch": 31.55513065646909, "grad_norm": 0.6526304483413696, "learning_rate": 9.642586040566903e-05, "loss": 0.0414, "step": 49510 }, { "epoch": 31.561504142766093, "grad_norm": 0.7227376103401184, "learning_rate": 9.642381373773955e-05, "loss": 0.0393, "step": 49520 }, { "epoch": 31.567877629063098, "grad_norm": 0.40774422883987427, "learning_rate": 9.642176650571691e-05, "loss": 0.0333, "step": 49530 }, { "epoch": 31.574251115360102, "grad_norm": 0.48242467641830444, "learning_rate": 9.641971870962597e-05, "loss": 0.0397, "step": 49540 }, { "epoch": 31.580624601657107, "grad_norm": 0.6240224242210388, "learning_rate": 9.641767034949161e-05, "loss": 0.0298, "step": 49550 }, { "epoch": 31.58699808795411, "grad_norm": 0.4099906086921692, "learning_rate": 9.641562142533874e-05, "loss": 0.0357, "step": 49560 }, { "epoch": 31.593371574251115, "grad_norm": 0.23491650819778442, "learning_rate": 9.641357193719223e-05, "loss": 0.0283, "step": 49570 }, { "epoch": 31.59974506054812, "grad_norm": 0.35146304965019226, "learning_rate": 9.6411521885077e-05, "loss": 0.0317, "step": 49580 }, { "epoch": 31.606118546845124, "grad_norm": 0.5646169185638428, "learning_rate": 9.640947126901798e-05, "loss": 0.0372, "step": 49590 }, { "epoch": 31.61249203314213, "grad_norm": 0.3802034556865692, "learning_rate": 9.640742008904005e-05, "loss": 0.041, "step": 49600 }, { "epoch": 31.618865519439133, "grad_norm": 0.3893008530139923, "learning_rate": 9.640536834516815e-05, "loss": 0.0363, "step": 49610 }, { "epoch": 31.625239005736137, "grad_norm": 0.6261109709739685, "learning_rate": 9.640331603742723e-05, "loss": 0.0341, "step": 49620 }, { "epoch": 31.631612492033142, "grad_norm": 0.5088911056518555, "learning_rate": 9.640126316584219e-05, "loss": 0.0392, "step": 49630 }, { "epoch": 31.637985978330146, "grad_norm": 0.9107030034065247, "learning_rate": 9.639920973043801e-05, "loss": 0.0324, "step": 49640 }, { "epoch": 31.64435946462715, "grad_norm": 0.9197406768798828, "learning_rate": 9.639715573123961e-05, "loss": 0.0286, "step": 49650 }, { "epoch": 31.650732950924155, "grad_norm": 0.39591729640960693, "learning_rate": 9.639510116827196e-05, "loss": 0.0288, "step": 49660 }, { "epoch": 31.65710643722116, "grad_norm": 0.3422752320766449, "learning_rate": 9.639304604156003e-05, "loss": 0.0285, "step": 49670 }, { "epoch": 31.663479923518164, "grad_norm": 0.4514029026031494, "learning_rate": 9.63909903511288e-05, "loss": 0.024, "step": 49680 }, { "epoch": 31.66985340981517, "grad_norm": 0.5566706657409668, "learning_rate": 9.638893409700321e-05, "loss": 0.0432, "step": 49690 }, { "epoch": 31.676226896112173, "grad_norm": 0.6060683131217957, "learning_rate": 9.638687727920828e-05, "loss": 0.0281, "step": 49700 }, { "epoch": 31.682600382409177, "grad_norm": 0.6065380573272705, "learning_rate": 9.638481989776901e-05, "loss": 0.0321, "step": 49710 }, { "epoch": 31.68897386870618, "grad_norm": 0.5228275656700134, "learning_rate": 9.638276195271037e-05, "loss": 0.0246, "step": 49720 }, { "epoch": 31.695347355003186, "grad_norm": 0.442179411649704, "learning_rate": 9.638070344405738e-05, "loss": 0.025, "step": 49730 }, { "epoch": 31.70172084130019, "grad_norm": 0.4860314130783081, "learning_rate": 9.637864437183507e-05, "loss": 0.0332, "step": 49740 }, { "epoch": 31.708094327597195, "grad_norm": 0.5204603672027588, "learning_rate": 9.637658473606843e-05, "loss": 0.0288, "step": 49750 }, { "epoch": 31.7144678138942, "grad_norm": 0.4228981137275696, "learning_rate": 9.637452453678251e-05, "loss": 0.0262, "step": 49760 }, { "epoch": 31.720841300191204, "grad_norm": 0.43562695384025574, "learning_rate": 9.63724637740023e-05, "loss": 0.0276, "step": 49770 }, { "epoch": 31.727214786488208, "grad_norm": 0.6446609497070312, "learning_rate": 9.637040244775291e-05, "loss": 0.0247, "step": 49780 }, { "epoch": 31.733588272785212, "grad_norm": 0.5088351964950562, "learning_rate": 9.636834055805933e-05, "loss": 0.0243, "step": 49790 }, { "epoch": 31.739961759082217, "grad_norm": 0.6405149102210999, "learning_rate": 9.636627810494663e-05, "loss": 0.0227, "step": 49800 }, { "epoch": 31.74633524537922, "grad_norm": 0.6196313500404358, "learning_rate": 9.636421508843988e-05, "loss": 0.0231, "step": 49810 }, { "epoch": 31.752708731676226, "grad_norm": 0.8192743062973022, "learning_rate": 9.636215150856414e-05, "loss": 0.0289, "step": 49820 }, { "epoch": 31.75908221797323, "grad_norm": 0.43991485238075256, "learning_rate": 9.636008736534448e-05, "loss": 0.0329, "step": 49830 }, { "epoch": 31.765455704270234, "grad_norm": 0.7172949910163879, "learning_rate": 9.6358022658806e-05, "loss": 0.0247, "step": 49840 }, { "epoch": 31.77182919056724, "grad_norm": 0.4867013394832611, "learning_rate": 9.635595738897377e-05, "loss": 0.0273, "step": 49850 }, { "epoch": 31.778202676864243, "grad_norm": 0.5021018981933594, "learning_rate": 9.635389155587289e-05, "loss": 0.0366, "step": 49860 }, { "epoch": 31.784576163161248, "grad_norm": 0.42545363306999207, "learning_rate": 9.635182515952847e-05, "loss": 0.0263, "step": 49870 }, { "epoch": 31.790949649458252, "grad_norm": 0.2706247568130493, "learning_rate": 9.63497581999656e-05, "loss": 0.0219, "step": 49880 }, { "epoch": 31.797323135755256, "grad_norm": 0.5228594541549683, "learning_rate": 9.634769067720941e-05, "loss": 0.032, "step": 49890 }, { "epoch": 31.803696622052264, "grad_norm": 0.42217695713043213, "learning_rate": 9.634562259128502e-05, "loss": 0.04, "step": 49900 }, { "epoch": 31.81007010834927, "grad_norm": 0.4393623173236847, "learning_rate": 9.634355394221756e-05, "loss": 0.0306, "step": 49910 }, { "epoch": 31.816443594646273, "grad_norm": 0.3546094596385956, "learning_rate": 9.634148473003216e-05, "loss": 0.0201, "step": 49920 }, { "epoch": 31.822817080943278, "grad_norm": 0.3944132328033447, "learning_rate": 9.633941495475396e-05, "loss": 0.0263, "step": 49930 }, { "epoch": 31.829190567240282, "grad_norm": 0.441276490688324, "learning_rate": 9.633734461640813e-05, "loss": 0.0178, "step": 49940 }, { "epoch": 31.835564053537286, "grad_norm": 0.3799179494380951, "learning_rate": 9.63352737150198e-05, "loss": 0.0268, "step": 49950 }, { "epoch": 31.84193753983429, "grad_norm": 0.46664437651634216, "learning_rate": 9.633320225061417e-05, "loss": 0.0354, "step": 49960 }, { "epoch": 31.848311026131295, "grad_norm": 0.2994579076766968, "learning_rate": 9.633113022321636e-05, "loss": 0.031, "step": 49970 }, { "epoch": 31.8546845124283, "grad_norm": 0.5496512651443481, "learning_rate": 9.632905763285158e-05, "loss": 0.0224, "step": 49980 }, { "epoch": 31.861057998725304, "grad_norm": 0.2695121467113495, "learning_rate": 9.632698447954503e-05, "loss": 0.022, "step": 49990 }, { "epoch": 31.86743148502231, "grad_norm": 0.4434918761253357, "learning_rate": 9.632491076332185e-05, "loss": 0.0284, "step": 50000 }, { "epoch": 31.873804971319313, "grad_norm": 0.5567672848701477, "learning_rate": 9.632283648420728e-05, "loss": 0.0339, "step": 50010 }, { "epoch": 31.880178457616317, "grad_norm": 0.5594049096107483, "learning_rate": 9.632076164222651e-05, "loss": 0.0381, "step": 50020 }, { "epoch": 31.88655194391332, "grad_norm": 0.32747602462768555, "learning_rate": 9.631868623740475e-05, "loss": 0.0192, "step": 50030 }, { "epoch": 31.892925430210326, "grad_norm": 0.41065672039985657, "learning_rate": 9.631661026976722e-05, "loss": 0.0282, "step": 50040 }, { "epoch": 31.89929891650733, "grad_norm": 0.44742387533187866, "learning_rate": 9.631453373933914e-05, "loss": 0.0371, "step": 50050 }, { "epoch": 31.905672402804335, "grad_norm": 0.5311201810836792, "learning_rate": 9.631245664614576e-05, "loss": 0.0387, "step": 50060 }, { "epoch": 31.91204588910134, "grad_norm": 0.6947133541107178, "learning_rate": 9.631037899021229e-05, "loss": 0.0325, "step": 50070 }, { "epoch": 31.918419375398344, "grad_norm": 0.526229977607727, "learning_rate": 9.630830077156401e-05, "loss": 0.0218, "step": 50080 }, { "epoch": 31.924792861695348, "grad_norm": 0.6932626962661743, "learning_rate": 9.630622199022614e-05, "loss": 0.0275, "step": 50090 }, { "epoch": 31.931166347992352, "grad_norm": 0.41859298944473267, "learning_rate": 9.630414264622397e-05, "loss": 0.0208, "step": 50100 }, { "epoch": 31.937539834289357, "grad_norm": 0.3434845507144928, "learning_rate": 9.630206273958272e-05, "loss": 0.0305, "step": 50110 }, { "epoch": 31.94391332058636, "grad_norm": 0.47382214665412903, "learning_rate": 9.62999822703277e-05, "loss": 0.0268, "step": 50120 }, { "epoch": 31.950286806883366, "grad_norm": 0.637179970741272, "learning_rate": 9.62979012384842e-05, "loss": 0.0285, "step": 50130 }, { "epoch": 31.95666029318037, "grad_norm": 0.5213580131530762, "learning_rate": 9.629581964407746e-05, "loss": 0.0279, "step": 50140 }, { "epoch": 31.963033779477374, "grad_norm": 1.0581862926483154, "learning_rate": 9.629373748713281e-05, "loss": 0.024, "step": 50150 }, { "epoch": 31.96940726577438, "grad_norm": 0.5908228754997253, "learning_rate": 9.629165476767555e-05, "loss": 0.0258, "step": 50160 }, { "epoch": 31.975780752071383, "grad_norm": 0.6178209185600281, "learning_rate": 9.628957148573098e-05, "loss": 0.0234, "step": 50170 }, { "epoch": 31.982154238368388, "grad_norm": 0.44477248191833496, "learning_rate": 9.628748764132441e-05, "loss": 0.0359, "step": 50180 }, { "epoch": 31.988527724665392, "grad_norm": 0.509045422077179, "learning_rate": 9.628540323448114e-05, "loss": 0.0273, "step": 50190 }, { "epoch": 31.994901210962396, "grad_norm": 0.404618501663208, "learning_rate": 9.628331826522654e-05, "loss": 0.0324, "step": 50200 }, { "epoch": 32.001274697259404, "grad_norm": 0.3674522638320923, "learning_rate": 9.628123273358594e-05, "loss": 0.0415, "step": 50210 }, { "epoch": 32.007648183556405, "grad_norm": 0.4915797710418701, "learning_rate": 9.627914663958463e-05, "loss": 0.0279, "step": 50220 }, { "epoch": 32.01402166985341, "grad_norm": 0.5052920579910278, "learning_rate": 9.627705998324802e-05, "loss": 0.0204, "step": 50230 }, { "epoch": 32.020395156150414, "grad_norm": 0.3605380058288574, "learning_rate": 9.627497276460142e-05, "loss": 0.0294, "step": 50240 }, { "epoch": 32.02676864244742, "grad_norm": 0.31230372190475464, "learning_rate": 9.627288498367022e-05, "loss": 0.0274, "step": 50250 }, { "epoch": 32.03314212874442, "grad_norm": 0.3686187267303467, "learning_rate": 9.627079664047979e-05, "loss": 0.02, "step": 50260 }, { "epoch": 32.03951561504143, "grad_norm": 0.511961042881012, "learning_rate": 9.626870773505548e-05, "loss": 0.0274, "step": 50270 }, { "epoch": 32.04588910133843, "grad_norm": 0.6962223052978516, "learning_rate": 9.62666182674227e-05, "loss": 0.0363, "step": 50280 }, { "epoch": 32.05226258763544, "grad_norm": 0.4706366956233978, "learning_rate": 9.626452823760681e-05, "loss": 0.0371, "step": 50290 }, { "epoch": 32.05863607393244, "grad_norm": 0.6609902381896973, "learning_rate": 9.626243764563321e-05, "loss": 0.0296, "step": 50300 }, { "epoch": 32.06500956022945, "grad_norm": 0.4340435564517975, "learning_rate": 9.626034649152733e-05, "loss": 0.0263, "step": 50310 }, { "epoch": 32.07138304652645, "grad_norm": 0.632668137550354, "learning_rate": 9.625825477531456e-05, "loss": 0.0258, "step": 50320 }, { "epoch": 32.07775653282346, "grad_norm": 0.5861436128616333, "learning_rate": 9.625616249702032e-05, "loss": 0.0415, "step": 50330 }, { "epoch": 32.08413001912046, "grad_norm": 0.40851935744285583, "learning_rate": 9.625406965667003e-05, "loss": 0.0309, "step": 50340 }, { "epoch": 32.090503505417466, "grad_norm": 0.5697430372238159, "learning_rate": 9.625197625428913e-05, "loss": 0.0297, "step": 50350 }, { "epoch": 32.09687699171447, "grad_norm": 0.4416266679763794, "learning_rate": 9.624988228990305e-05, "loss": 0.0264, "step": 50360 }, { "epoch": 32.103250478011475, "grad_norm": 0.8499746322631836, "learning_rate": 9.624778776353722e-05, "loss": 0.0295, "step": 50370 }, { "epoch": 32.109623964308476, "grad_norm": 0.3335327208042145, "learning_rate": 9.624569267521712e-05, "loss": 0.0236, "step": 50380 }, { "epoch": 32.115997450605484, "grad_norm": 0.3305918872356415, "learning_rate": 9.624359702496819e-05, "loss": 0.0269, "step": 50390 }, { "epoch": 32.122370936902485, "grad_norm": 0.46877411007881165, "learning_rate": 9.624150081281589e-05, "loss": 0.0357, "step": 50400 }, { "epoch": 32.12874442319949, "grad_norm": 0.4583114981651306, "learning_rate": 9.623940403878569e-05, "loss": 0.0304, "step": 50410 }, { "epoch": 32.13511790949649, "grad_norm": 0.4105294644832611, "learning_rate": 9.623730670290307e-05, "loss": 0.0304, "step": 50420 }, { "epoch": 32.1414913957935, "grad_norm": 0.596894383430481, "learning_rate": 9.623520880519352e-05, "loss": 0.0278, "step": 50430 }, { "epoch": 32.1478648820905, "grad_norm": 0.6973862648010254, "learning_rate": 9.623311034568254e-05, "loss": 0.0235, "step": 50440 }, { "epoch": 32.15423836838751, "grad_norm": 0.38368988037109375, "learning_rate": 9.623101132439561e-05, "loss": 0.0222, "step": 50450 }, { "epoch": 32.16061185468451, "grad_norm": 0.3807137906551361, "learning_rate": 9.622891174135825e-05, "loss": 0.0261, "step": 50460 }, { "epoch": 32.16698534098152, "grad_norm": 0.6037473678588867, "learning_rate": 9.622681159659597e-05, "loss": 0.0314, "step": 50470 }, { "epoch": 32.17335882727852, "grad_norm": 0.45340391993522644, "learning_rate": 9.622471089013427e-05, "loss": 0.0205, "step": 50480 }, { "epoch": 32.17973231357553, "grad_norm": 0.3425014019012451, "learning_rate": 9.622260962199869e-05, "loss": 0.0234, "step": 50490 }, { "epoch": 32.18610579987253, "grad_norm": 0.4424443244934082, "learning_rate": 9.622050779221477e-05, "loss": 0.024, "step": 50500 }, { "epoch": 32.19247928616954, "grad_norm": 0.5424277782440186, "learning_rate": 9.621840540080805e-05, "loss": 0.0309, "step": 50510 }, { "epoch": 32.19885277246654, "grad_norm": 0.7074220180511475, "learning_rate": 9.621630244780404e-05, "loss": 0.0236, "step": 50520 }, { "epoch": 32.205226258763545, "grad_norm": 0.9161022901535034, "learning_rate": 9.621419893322834e-05, "loss": 0.0443, "step": 50530 }, { "epoch": 32.211599745060546, "grad_norm": 0.48749053478240967, "learning_rate": 9.621209485710647e-05, "loss": 0.03, "step": 50540 }, { "epoch": 32.217973231357554, "grad_norm": 0.8491482734680176, "learning_rate": 9.620999021946404e-05, "loss": 0.0275, "step": 50550 }, { "epoch": 32.224346717654555, "grad_norm": 0.2147703766822815, "learning_rate": 9.620788502032658e-05, "loss": 0.0201, "step": 50560 }, { "epoch": 32.23072020395156, "grad_norm": 0.4647914469242096, "learning_rate": 9.62057792597197e-05, "loss": 0.0328, "step": 50570 }, { "epoch": 32.237093690248564, "grad_norm": 0.9474709630012512, "learning_rate": 9.620367293766898e-05, "loss": 0.0281, "step": 50580 }, { "epoch": 32.24346717654557, "grad_norm": 0.49123984575271606, "learning_rate": 9.620156605420001e-05, "loss": 0.028, "step": 50590 }, { "epoch": 32.24984066284257, "grad_norm": 0.3277948200702667, "learning_rate": 9.619945860933839e-05, "loss": 0.035, "step": 50600 }, { "epoch": 32.25621414913958, "grad_norm": 0.6156045198440552, "learning_rate": 9.619735060310972e-05, "loss": 0.0205, "step": 50610 }, { "epoch": 32.26258763543658, "grad_norm": 0.44093379378318787, "learning_rate": 9.619524203553962e-05, "loss": 0.0226, "step": 50620 }, { "epoch": 32.26896112173359, "grad_norm": 0.3858446478843689, "learning_rate": 9.619313290665373e-05, "loss": 0.0361, "step": 50630 }, { "epoch": 32.27533460803059, "grad_norm": 0.6209191083908081, "learning_rate": 9.619102321647765e-05, "loss": 0.0273, "step": 50640 }, { "epoch": 32.2817080943276, "grad_norm": 0.7387955784797668, "learning_rate": 9.618891296503702e-05, "loss": 0.0465, "step": 50650 }, { "epoch": 32.2880815806246, "grad_norm": 0.6510863304138184, "learning_rate": 9.618680215235751e-05, "loss": 0.0319, "step": 50660 }, { "epoch": 32.29445506692161, "grad_norm": 0.5081324577331543, "learning_rate": 9.618469077846473e-05, "loss": 0.0232, "step": 50670 }, { "epoch": 32.30082855321861, "grad_norm": 0.5236969590187073, "learning_rate": 9.618257884338435e-05, "loss": 0.0409, "step": 50680 }, { "epoch": 32.307202039515616, "grad_norm": 0.7517256736755371, "learning_rate": 9.618046634714205e-05, "loss": 0.0314, "step": 50690 }, { "epoch": 32.31357552581262, "grad_norm": 0.3475639522075653, "learning_rate": 9.617835328976345e-05, "loss": 0.0273, "step": 50700 }, { "epoch": 32.319949012109625, "grad_norm": 0.38038545846939087, "learning_rate": 9.617623967127429e-05, "loss": 0.0228, "step": 50710 }, { "epoch": 32.326322498406626, "grad_norm": 0.3361956477165222, "learning_rate": 9.61741254917002e-05, "loss": 0.0277, "step": 50720 }, { "epoch": 32.33269598470363, "grad_norm": 0.5007058382034302, "learning_rate": 9.61720107510669e-05, "loss": 0.0289, "step": 50730 }, { "epoch": 32.339069471000634, "grad_norm": 0.45574307441711426, "learning_rate": 9.616989544940009e-05, "loss": 0.0255, "step": 50740 }, { "epoch": 32.34544295729764, "grad_norm": 0.6038351058959961, "learning_rate": 9.616777958672543e-05, "loss": 0.0381, "step": 50750 }, { "epoch": 32.35181644359464, "grad_norm": 0.5222229957580566, "learning_rate": 9.616566316306867e-05, "loss": 0.0316, "step": 50760 }, { "epoch": 32.35818992989165, "grad_norm": 0.3345721662044525, "learning_rate": 9.616354617845552e-05, "loss": 0.0377, "step": 50770 }, { "epoch": 32.36456341618865, "grad_norm": 0.40095987915992737, "learning_rate": 9.616142863291169e-05, "loss": 0.0253, "step": 50780 }, { "epoch": 32.37093690248566, "grad_norm": 0.9229083061218262, "learning_rate": 9.615931052646293e-05, "loss": 0.0407, "step": 50790 }, { "epoch": 32.37731038878266, "grad_norm": 0.5151830315589905, "learning_rate": 9.615719185913496e-05, "loss": 0.0293, "step": 50800 }, { "epoch": 32.38368387507967, "grad_norm": 0.2776039242744446, "learning_rate": 9.615507263095353e-05, "loss": 0.0349, "step": 50810 }, { "epoch": 32.39005736137667, "grad_norm": 0.5033051371574402, "learning_rate": 9.61529528419444e-05, "loss": 0.0267, "step": 50820 }, { "epoch": 32.39643084767368, "grad_norm": 0.544008731842041, "learning_rate": 9.615083249213329e-05, "loss": 0.029, "step": 50830 }, { "epoch": 32.40280433397068, "grad_norm": 0.5722498297691345, "learning_rate": 9.614871158154601e-05, "loss": 0.026, "step": 50840 }, { "epoch": 32.409177820267686, "grad_norm": 0.46081364154815674, "learning_rate": 9.61465901102083e-05, "loss": 0.0224, "step": 50850 }, { "epoch": 32.415551306564694, "grad_norm": 0.3459358513355255, "learning_rate": 9.614446807814597e-05, "loss": 0.0283, "step": 50860 }, { "epoch": 32.421924792861695, "grad_norm": 0.33563676476478577, "learning_rate": 9.614234548538477e-05, "loss": 0.0316, "step": 50870 }, { "epoch": 32.4282982791587, "grad_norm": 0.3282836973667145, "learning_rate": 9.614022233195051e-05, "loss": 0.0232, "step": 50880 }, { "epoch": 32.434671765455704, "grad_norm": 0.44209009408950806, "learning_rate": 9.613809861786899e-05, "loss": 0.0357, "step": 50890 }, { "epoch": 32.44104525175271, "grad_norm": 0.6703401207923889, "learning_rate": 9.613597434316602e-05, "loss": 0.0256, "step": 50900 }, { "epoch": 32.44741873804971, "grad_norm": 0.4466400742530823, "learning_rate": 9.613384950786738e-05, "loss": 0.0292, "step": 50910 }, { "epoch": 32.45379222434672, "grad_norm": 0.49755531549453735, "learning_rate": 9.613172411199891e-05, "loss": 0.0291, "step": 50920 }, { "epoch": 32.46016571064372, "grad_norm": 0.354056179523468, "learning_rate": 9.612959815558645e-05, "loss": 0.0261, "step": 50930 }, { "epoch": 32.46653919694073, "grad_norm": 0.43406248092651367, "learning_rate": 9.612747163865582e-05, "loss": 0.025, "step": 50940 }, { "epoch": 32.47291268323773, "grad_norm": 0.32500842213630676, "learning_rate": 9.612534456123285e-05, "loss": 0.0256, "step": 50950 }, { "epoch": 32.47928616953474, "grad_norm": 0.2974274456501007, "learning_rate": 9.612321692334341e-05, "loss": 0.0306, "step": 50960 }, { "epoch": 32.48565965583174, "grad_norm": 0.5435872673988342, "learning_rate": 9.612108872501331e-05, "loss": 0.0419, "step": 50970 }, { "epoch": 32.49203314212875, "grad_norm": 0.2850504517555237, "learning_rate": 9.611895996626845e-05, "loss": 0.0225, "step": 50980 }, { "epoch": 32.49840662842575, "grad_norm": 0.564734160900116, "learning_rate": 9.611683064713469e-05, "loss": 0.0297, "step": 50990 }, { "epoch": 32.504780114722756, "grad_norm": 0.4351910650730133, "learning_rate": 9.611470076763788e-05, "loss": 0.0272, "step": 51000 }, { "epoch": 32.51115360101976, "grad_norm": 0.4046146869659424, "learning_rate": 9.611257032780394e-05, "loss": 0.035, "step": 51010 }, { "epoch": 32.517527087316765, "grad_norm": 0.4468485116958618, "learning_rate": 9.611043932765872e-05, "loss": 0.0279, "step": 51020 }, { "epoch": 32.523900573613766, "grad_norm": 0.40404176712036133, "learning_rate": 9.610830776722812e-05, "loss": 0.0288, "step": 51030 }, { "epoch": 32.530274059910774, "grad_norm": 0.21042655408382416, "learning_rate": 9.610617564653805e-05, "loss": 0.0247, "step": 51040 }, { "epoch": 32.536647546207774, "grad_norm": 0.29448527097702026, "learning_rate": 9.610404296561441e-05, "loss": 0.0273, "step": 51050 }, { "epoch": 32.54302103250478, "grad_norm": 0.7024926543235779, "learning_rate": 9.610190972448314e-05, "loss": 0.0216, "step": 51060 }, { "epoch": 32.54939451880178, "grad_norm": 0.39891892671585083, "learning_rate": 9.609977592317012e-05, "loss": 0.0323, "step": 51070 }, { "epoch": 32.55576800509879, "grad_norm": 0.34664294123649597, "learning_rate": 9.609764156170132e-05, "loss": 0.0197, "step": 51080 }, { "epoch": 32.56214149139579, "grad_norm": 0.6028919219970703, "learning_rate": 9.609550664010263e-05, "loss": 0.0254, "step": 51090 }, { "epoch": 32.5685149776928, "grad_norm": 0.5032874941825867, "learning_rate": 9.60933711584e-05, "loss": 0.0275, "step": 51100 }, { "epoch": 32.5748884639898, "grad_norm": 0.800987720489502, "learning_rate": 9.609123511661941e-05, "loss": 0.0366, "step": 51110 }, { "epoch": 32.58126195028681, "grad_norm": 0.4672933518886566, "learning_rate": 9.60890985147868e-05, "loss": 0.0417, "step": 51120 }, { "epoch": 32.58763543658381, "grad_norm": 0.5694705843925476, "learning_rate": 9.608696135292812e-05, "loss": 0.0278, "step": 51130 }, { "epoch": 32.59400892288082, "grad_norm": 0.47093504667282104, "learning_rate": 9.608482363106935e-05, "loss": 0.0255, "step": 51140 }, { "epoch": 32.60038240917782, "grad_norm": 0.5619297623634338, "learning_rate": 9.608268534923646e-05, "loss": 0.0222, "step": 51150 }, { "epoch": 32.606755895474826, "grad_norm": 0.38592392206192017, "learning_rate": 9.608054650745543e-05, "loss": 0.029, "step": 51160 }, { "epoch": 32.61312938177183, "grad_norm": 0.39945876598358154, "learning_rate": 9.607840710575226e-05, "loss": 0.0571, "step": 51170 }, { "epoch": 32.619502868068835, "grad_norm": 0.37257975339889526, "learning_rate": 9.607626714415294e-05, "loss": 0.0308, "step": 51180 }, { "epoch": 32.625876354365836, "grad_norm": 0.30795133113861084, "learning_rate": 9.607412662268346e-05, "loss": 0.0288, "step": 51190 }, { "epoch": 32.632249840662844, "grad_norm": 0.404982328414917, "learning_rate": 9.607198554136985e-05, "loss": 0.0286, "step": 51200 }, { "epoch": 32.638623326959845, "grad_norm": 0.5144482851028442, "learning_rate": 9.606984390023812e-05, "loss": 0.0297, "step": 51210 }, { "epoch": 32.64499681325685, "grad_norm": 0.49482566118240356, "learning_rate": 9.606770169931427e-05, "loss": 0.0469, "step": 51220 }, { "epoch": 32.651370299553854, "grad_norm": 1.3468883037567139, "learning_rate": 9.606555893862437e-05, "loss": 0.0313, "step": 51230 }, { "epoch": 32.65774378585086, "grad_norm": 0.5524077415466309, "learning_rate": 9.606341561819443e-05, "loss": 0.0287, "step": 51240 }, { "epoch": 32.66411727214786, "grad_norm": 0.5672935843467712, "learning_rate": 9.606127173805048e-05, "loss": 0.0275, "step": 51250 }, { "epoch": 32.67049075844487, "grad_norm": 0.3613284230232239, "learning_rate": 9.605912729821862e-05, "loss": 0.0292, "step": 51260 }, { "epoch": 32.67686424474187, "grad_norm": 0.7732415199279785, "learning_rate": 9.605698229872487e-05, "loss": 0.026, "step": 51270 }, { "epoch": 32.68323773103888, "grad_norm": 0.4910607635974884, "learning_rate": 9.605483673959528e-05, "loss": 0.0364, "step": 51280 }, { "epoch": 32.68961121733588, "grad_norm": 0.7407321929931641, "learning_rate": 9.605269062085596e-05, "loss": 0.0261, "step": 51290 }, { "epoch": 32.69598470363289, "grad_norm": 0.4703577756881714, "learning_rate": 9.605054394253295e-05, "loss": 0.0273, "step": 51300 }, { "epoch": 32.70235818992989, "grad_norm": 0.8633629679679871, "learning_rate": 9.604839670465236e-05, "loss": 0.0311, "step": 51310 }, { "epoch": 32.7087316762269, "grad_norm": 0.5776593685150146, "learning_rate": 9.604624890724027e-05, "loss": 0.0277, "step": 51320 }, { "epoch": 32.7151051625239, "grad_norm": 0.5495419502258301, "learning_rate": 9.60441005503228e-05, "loss": 0.0302, "step": 51330 }, { "epoch": 32.721478648820906, "grad_norm": 0.43936818838119507, "learning_rate": 9.604195163392602e-05, "loss": 0.026, "step": 51340 }, { "epoch": 32.72785213511791, "grad_norm": 0.41285133361816406, "learning_rate": 9.603980215807607e-05, "loss": 0.0353, "step": 51350 }, { "epoch": 32.734225621414915, "grad_norm": 0.45568233728408813, "learning_rate": 9.603765212279904e-05, "loss": 0.04, "step": 51360 }, { "epoch": 32.740599107711915, "grad_norm": 0.4877910912036896, "learning_rate": 9.603550152812106e-05, "loss": 0.0269, "step": 51370 }, { "epoch": 32.74697259400892, "grad_norm": 0.6525394916534424, "learning_rate": 9.60333503740683e-05, "loss": 0.0307, "step": 51380 }, { "epoch": 32.753346080305924, "grad_norm": 0.40296053886413574, "learning_rate": 9.603119866066686e-05, "loss": 0.0172, "step": 51390 }, { "epoch": 32.75971956660293, "grad_norm": 0.47951367497444153, "learning_rate": 9.602904638794288e-05, "loss": 0.0274, "step": 51400 }, { "epoch": 32.76609305289993, "grad_norm": 0.6736000180244446, "learning_rate": 9.602689355592254e-05, "loss": 0.0244, "step": 51410 }, { "epoch": 32.77246653919694, "grad_norm": 0.4851997196674347, "learning_rate": 9.602474016463199e-05, "loss": 0.0264, "step": 51420 }, { "epoch": 32.77884002549394, "grad_norm": 0.3178149163722992, "learning_rate": 9.602258621409739e-05, "loss": 0.0248, "step": 51430 }, { "epoch": 32.78521351179095, "grad_norm": 0.4149443507194519, "learning_rate": 9.602043170434491e-05, "loss": 0.0338, "step": 51440 }, { "epoch": 32.79158699808795, "grad_norm": 0.3450547754764557, "learning_rate": 9.601827663540075e-05, "loss": 0.0273, "step": 51450 }, { "epoch": 32.79796048438496, "grad_norm": 0.4411133825778961, "learning_rate": 9.601612100729106e-05, "loss": 0.0273, "step": 51460 }, { "epoch": 32.80433397068197, "grad_norm": 0.4255078136920929, "learning_rate": 9.601396482004206e-05, "loss": 0.0304, "step": 51470 }, { "epoch": 32.81070745697897, "grad_norm": 0.9307108521461487, "learning_rate": 9.601180807367995e-05, "loss": 0.0317, "step": 51480 }, { "epoch": 32.817080943275975, "grad_norm": 0.6013964414596558, "learning_rate": 9.600965076823092e-05, "loss": 0.0237, "step": 51490 }, { "epoch": 32.823454429572976, "grad_norm": 0.43137577176094055, "learning_rate": 9.600749290372119e-05, "loss": 0.029, "step": 51500 }, { "epoch": 32.829827915869984, "grad_norm": 0.47421103715896606, "learning_rate": 9.6005334480177e-05, "loss": 0.0301, "step": 51510 }, { "epoch": 32.836201402166985, "grad_norm": 0.4701686501502991, "learning_rate": 9.600317549762454e-05, "loss": 0.0259, "step": 51520 }, { "epoch": 32.84257488846399, "grad_norm": 0.32566794753074646, "learning_rate": 9.600101595609008e-05, "loss": 0.0296, "step": 51530 }, { "epoch": 32.848948374760994, "grad_norm": 0.6722625494003296, "learning_rate": 9.599885585559985e-05, "loss": 0.0319, "step": 51540 }, { "epoch": 32.855321861058, "grad_norm": 0.6065362095832825, "learning_rate": 9.599669519618006e-05, "loss": 0.0322, "step": 51550 }, { "epoch": 32.861695347355, "grad_norm": 0.7035350203514099, "learning_rate": 9.599453397785702e-05, "loss": 0.0371, "step": 51560 }, { "epoch": 32.86806883365201, "grad_norm": 0.6945169568061829, "learning_rate": 9.599237220065696e-05, "loss": 0.0346, "step": 51570 }, { "epoch": 32.87444231994901, "grad_norm": 0.895824134349823, "learning_rate": 9.599020986460616e-05, "loss": 0.0311, "step": 51580 }, { "epoch": 32.88081580624602, "grad_norm": 0.6051656603813171, "learning_rate": 9.598804696973087e-05, "loss": 0.0267, "step": 51590 }, { "epoch": 32.88718929254302, "grad_norm": 0.5784096717834473, "learning_rate": 9.598588351605742e-05, "loss": 0.0208, "step": 51600 }, { "epoch": 32.89356277884003, "grad_norm": 0.39280831813812256, "learning_rate": 9.598371950361205e-05, "loss": 0.0237, "step": 51610 }, { "epoch": 32.89993626513703, "grad_norm": 0.42418429255485535, "learning_rate": 9.598155493242106e-05, "loss": 0.0223, "step": 51620 }, { "epoch": 32.90630975143404, "grad_norm": 0.6621795296669006, "learning_rate": 9.597938980251078e-05, "loss": 0.028, "step": 51630 }, { "epoch": 32.91268323773104, "grad_norm": 0.3239007592201233, "learning_rate": 9.597722411390749e-05, "loss": 0.0254, "step": 51640 }, { "epoch": 32.919056724028046, "grad_norm": 0.45672088861465454, "learning_rate": 9.597505786663754e-05, "loss": 0.0223, "step": 51650 }, { "epoch": 32.92543021032505, "grad_norm": 0.7999601364135742, "learning_rate": 9.597289106072719e-05, "loss": 0.0321, "step": 51660 }, { "epoch": 32.931803696622055, "grad_norm": 0.9373602271080017, "learning_rate": 9.597072369620284e-05, "loss": 0.0302, "step": 51670 }, { "epoch": 32.938177182919055, "grad_norm": 0.5340915322303772, "learning_rate": 9.596855577309078e-05, "loss": 0.0271, "step": 51680 }, { "epoch": 32.94455066921606, "grad_norm": 0.5211122632026672, "learning_rate": 9.596638729141738e-05, "loss": 0.0466, "step": 51690 }, { "epoch": 32.950924155513064, "grad_norm": 0.38493096828460693, "learning_rate": 9.596421825120895e-05, "loss": 0.0293, "step": 51700 }, { "epoch": 32.95729764181007, "grad_norm": 0.6660537123680115, "learning_rate": 9.59620486524919e-05, "loss": 0.0285, "step": 51710 }, { "epoch": 32.96367112810707, "grad_norm": 0.5970039963722229, "learning_rate": 9.595987849529255e-05, "loss": 0.0295, "step": 51720 }, { "epoch": 32.97004461440408, "grad_norm": 0.401458740234375, "learning_rate": 9.595770777963728e-05, "loss": 0.0253, "step": 51730 }, { "epoch": 32.97641810070108, "grad_norm": 0.4435962438583374, "learning_rate": 9.595553650555248e-05, "loss": 0.0304, "step": 51740 }, { "epoch": 32.98279158699809, "grad_norm": 0.4422203302383423, "learning_rate": 9.595336467306451e-05, "loss": 0.0217, "step": 51750 }, { "epoch": 32.98916507329509, "grad_norm": 0.3956891596317291, "learning_rate": 9.595119228219978e-05, "loss": 0.0242, "step": 51760 }, { "epoch": 32.9955385595921, "grad_norm": 0.5664805769920349, "learning_rate": 9.594901933298468e-05, "loss": 0.0253, "step": 51770 }, { "epoch": 33.0019120458891, "grad_norm": 0.39192429184913635, "learning_rate": 9.594684582544559e-05, "loss": 0.031, "step": 51780 }, { "epoch": 33.00828553218611, "grad_norm": 0.6645326018333435, "learning_rate": 9.594467175960897e-05, "loss": 0.0294, "step": 51790 }, { "epoch": 33.01465901848311, "grad_norm": 0.8251082301139832, "learning_rate": 9.59424971355012e-05, "loss": 0.0323, "step": 51800 }, { "epoch": 33.021032504780116, "grad_norm": 0.5013106465339661, "learning_rate": 9.594032195314872e-05, "loss": 0.0222, "step": 51810 }, { "epoch": 33.02740599107712, "grad_norm": 0.4512283504009247, "learning_rate": 9.593814621257794e-05, "loss": 0.0299, "step": 51820 }, { "epoch": 33.033779477374125, "grad_norm": 0.4122163951396942, "learning_rate": 9.593596991381531e-05, "loss": 0.0271, "step": 51830 }, { "epoch": 33.040152963671126, "grad_norm": 0.5083367228507996, "learning_rate": 9.593379305688728e-05, "loss": 0.0203, "step": 51840 }, { "epoch": 33.046526449968134, "grad_norm": 0.6539205312728882, "learning_rate": 9.593161564182029e-05, "loss": 0.0352, "step": 51850 }, { "epoch": 33.052899936265135, "grad_norm": 0.5143155455589294, "learning_rate": 9.592943766864083e-05, "loss": 0.0223, "step": 51860 }, { "epoch": 33.05927342256214, "grad_norm": 0.6443670988082886, "learning_rate": 9.59272591373753e-05, "loss": 0.0339, "step": 51870 }, { "epoch": 33.065646908859144, "grad_norm": 0.5199339985847473, "learning_rate": 9.592508004805023e-05, "loss": 0.0371, "step": 51880 }, { "epoch": 33.07202039515615, "grad_norm": 0.6474002599716187, "learning_rate": 9.592290040069207e-05, "loss": 0.034, "step": 51890 }, { "epoch": 33.07839388145315, "grad_norm": 0.6562907695770264, "learning_rate": 9.592072019532731e-05, "loss": 0.0352, "step": 51900 }, { "epoch": 33.08476736775016, "grad_norm": 0.39504918456077576, "learning_rate": 9.591853943198245e-05, "loss": 0.0229, "step": 51910 }, { "epoch": 33.09114085404716, "grad_norm": 0.7300223708152771, "learning_rate": 9.5916358110684e-05, "loss": 0.0282, "step": 51920 }, { "epoch": 33.09751434034417, "grad_norm": 0.7987372875213623, "learning_rate": 9.591417623145843e-05, "loss": 0.0337, "step": 51930 }, { "epoch": 33.10388782664117, "grad_norm": 0.524081826210022, "learning_rate": 9.591199379433225e-05, "loss": 0.0297, "step": 51940 }, { "epoch": 33.11026131293818, "grad_norm": 0.28298643231391907, "learning_rate": 9.590981079933202e-05, "loss": 0.0201, "step": 51950 }, { "epoch": 33.11663479923518, "grad_norm": 0.4383884072303772, "learning_rate": 9.590762724648423e-05, "loss": 0.0306, "step": 51960 }, { "epoch": 33.12300828553219, "grad_norm": 0.42452484369277954, "learning_rate": 9.590544313581545e-05, "loss": 0.0325, "step": 51970 }, { "epoch": 33.12938177182919, "grad_norm": 0.7442895770072937, "learning_rate": 9.590325846735216e-05, "loss": 0.0312, "step": 51980 }, { "epoch": 33.135755258126196, "grad_norm": 0.42221128940582275, "learning_rate": 9.590107324112095e-05, "loss": 0.0249, "step": 51990 }, { "epoch": 33.1421287444232, "grad_norm": 0.3364163041114807, "learning_rate": 9.589888745714837e-05, "loss": 0.0246, "step": 52000 }, { "epoch": 33.148502230720204, "grad_norm": 0.759209930896759, "learning_rate": 9.589670111546097e-05, "loss": 0.0368, "step": 52010 }, { "epoch": 33.154875717017205, "grad_norm": 0.3822955787181854, "learning_rate": 9.589451421608531e-05, "loss": 0.0434, "step": 52020 }, { "epoch": 33.16124920331421, "grad_norm": 0.6044666767120361, "learning_rate": 9.589232675904798e-05, "loss": 0.0298, "step": 52030 }, { "epoch": 33.167622689611214, "grad_norm": 1.056251883506775, "learning_rate": 9.589013874437555e-05, "loss": 0.0308, "step": 52040 }, { "epoch": 33.17399617590822, "grad_norm": 0.48414215445518494, "learning_rate": 9.58879501720946e-05, "loss": 0.0446, "step": 52050 }, { "epoch": 33.18036966220522, "grad_norm": 0.29902100563049316, "learning_rate": 9.588576104223174e-05, "loss": 0.0269, "step": 52060 }, { "epoch": 33.18674314850223, "grad_norm": 0.5396019220352173, "learning_rate": 9.588357135481355e-05, "loss": 0.0387, "step": 52070 }, { "epoch": 33.19311663479923, "grad_norm": 0.43449875712394714, "learning_rate": 9.588138110986665e-05, "loss": 0.0289, "step": 52080 }, { "epoch": 33.19949012109624, "grad_norm": 0.4832167625427246, "learning_rate": 9.587919030741766e-05, "loss": 0.026, "step": 52090 }, { "epoch": 33.20586360739325, "grad_norm": 0.5578222870826721, "learning_rate": 9.587699894749317e-05, "loss": 0.0305, "step": 52100 }, { "epoch": 33.21223709369025, "grad_norm": 0.6065574884414673, "learning_rate": 9.587480703011984e-05, "loss": 0.0389, "step": 52110 }, { "epoch": 33.218610579987256, "grad_norm": 0.5374851822853088, "learning_rate": 9.587261455532429e-05, "loss": 0.0267, "step": 52120 }, { "epoch": 33.22498406628426, "grad_norm": 0.5225486755371094, "learning_rate": 9.587042152313317e-05, "loss": 0.0269, "step": 52130 }, { "epoch": 33.231357552581265, "grad_norm": 0.47357842326164246, "learning_rate": 9.586822793357311e-05, "loss": 0.0308, "step": 52140 }, { "epoch": 33.237731038878266, "grad_norm": 0.25090691447257996, "learning_rate": 9.586603378667077e-05, "loss": 0.0246, "step": 52150 }, { "epoch": 33.244104525175274, "grad_norm": 0.4330727159976959, "learning_rate": 9.586383908245283e-05, "loss": 0.0239, "step": 52160 }, { "epoch": 33.250478011472275, "grad_norm": 0.6424544453620911, "learning_rate": 9.586164382094593e-05, "loss": 0.0254, "step": 52170 }, { "epoch": 33.25685149776928, "grad_norm": 0.44416797161102295, "learning_rate": 9.585944800217677e-05, "loss": 0.0176, "step": 52180 }, { "epoch": 33.263224984066284, "grad_norm": 0.42545875906944275, "learning_rate": 9.5857251626172e-05, "loss": 0.037, "step": 52190 }, { "epoch": 33.26959847036329, "grad_norm": 0.4206511676311493, "learning_rate": 9.585505469295833e-05, "loss": 0.029, "step": 52200 }, { "epoch": 33.27597195666029, "grad_norm": 0.6629734635353088, "learning_rate": 9.585285720256246e-05, "loss": 0.0288, "step": 52210 }, { "epoch": 33.2823454429573, "grad_norm": 0.37238195538520813, "learning_rate": 9.585065915501108e-05, "loss": 0.0237, "step": 52220 }, { "epoch": 33.2887189292543, "grad_norm": 0.4782288372516632, "learning_rate": 9.58484605503309e-05, "loss": 0.0252, "step": 52230 }, { "epoch": 33.29509241555131, "grad_norm": 0.5243447422981262, "learning_rate": 9.584626138854864e-05, "loss": 0.0316, "step": 52240 }, { "epoch": 33.30146590184831, "grad_norm": 0.4381398558616638, "learning_rate": 9.584406166969101e-05, "loss": 0.0307, "step": 52250 }, { "epoch": 33.30783938814532, "grad_norm": 0.4061771333217621, "learning_rate": 9.584186139378474e-05, "loss": 0.0173, "step": 52260 }, { "epoch": 33.31421287444232, "grad_norm": 0.32514333724975586, "learning_rate": 9.58396605608566e-05, "loss": 0.0248, "step": 52270 }, { "epoch": 33.32058636073933, "grad_norm": 0.4628959596157074, "learning_rate": 9.583745917093327e-05, "loss": 0.0204, "step": 52280 }, { "epoch": 33.32695984703633, "grad_norm": 0.4943941533565521, "learning_rate": 9.583525722404157e-05, "loss": 0.019, "step": 52290 }, { "epoch": 33.333333333333336, "grad_norm": 0.5929449796676636, "learning_rate": 9.58330547202082e-05, "loss": 0.0221, "step": 52300 }, { "epoch": 33.33970681963034, "grad_norm": 0.6916621327400208, "learning_rate": 9.583085165945992e-05, "loss": 0.0221, "step": 52310 }, { "epoch": 33.346080305927345, "grad_norm": 0.4577091336250305, "learning_rate": 9.582864804182356e-05, "loss": 0.0222, "step": 52320 }, { "epoch": 33.352453792224345, "grad_norm": 0.6032233834266663, "learning_rate": 9.582644386732584e-05, "loss": 0.0299, "step": 52330 }, { "epoch": 33.35882727852135, "grad_norm": 0.48720675706863403, "learning_rate": 9.582423913599357e-05, "loss": 0.0351, "step": 52340 }, { "epoch": 33.365200764818354, "grad_norm": 0.5394951105117798, "learning_rate": 9.582203384785351e-05, "loss": 0.0269, "step": 52350 }, { "epoch": 33.37157425111536, "grad_norm": 0.528304934501648, "learning_rate": 9.581982800293248e-05, "loss": 0.031, "step": 52360 }, { "epoch": 33.37794773741236, "grad_norm": 0.5044497847557068, "learning_rate": 9.581762160125729e-05, "loss": 0.0252, "step": 52370 }, { "epoch": 33.38432122370937, "grad_norm": 0.3245835602283478, "learning_rate": 9.581541464285472e-05, "loss": 0.0241, "step": 52380 }, { "epoch": 33.39069471000637, "grad_norm": 0.5321206450462341, "learning_rate": 9.581320712775161e-05, "loss": 0.033, "step": 52390 }, { "epoch": 33.39706819630338, "grad_norm": 0.5315662622451782, "learning_rate": 9.58109990559748e-05, "loss": 0.0302, "step": 52400 }, { "epoch": 33.40344168260038, "grad_norm": 0.7033866047859192, "learning_rate": 9.580879042755109e-05, "loss": 0.0304, "step": 52410 }, { "epoch": 33.40981516889739, "grad_norm": 0.5001955032348633, "learning_rate": 9.580658124250731e-05, "loss": 0.0289, "step": 52420 }, { "epoch": 33.41618865519439, "grad_norm": 0.5968204736709595, "learning_rate": 9.580437150087033e-05, "loss": 0.0286, "step": 52430 }, { "epoch": 33.4225621414914, "grad_norm": 0.32420679926872253, "learning_rate": 9.580216120266698e-05, "loss": 0.019, "step": 52440 }, { "epoch": 33.4289356277884, "grad_norm": 0.6301494836807251, "learning_rate": 9.579995034792414e-05, "loss": 0.0248, "step": 52450 }, { "epoch": 33.435309114085406, "grad_norm": 0.44887086749076843, "learning_rate": 9.579773893666865e-05, "loss": 0.0211, "step": 52460 }, { "epoch": 33.44168260038241, "grad_norm": 0.32745301723480225, "learning_rate": 9.579552696892741e-05, "loss": 0.0333, "step": 52470 }, { "epoch": 33.448056086679415, "grad_norm": 0.44621390104293823, "learning_rate": 9.579331444472725e-05, "loss": 0.0257, "step": 52480 }, { "epoch": 33.454429572976416, "grad_norm": 0.4460764229297638, "learning_rate": 9.579110136409511e-05, "loss": 0.0329, "step": 52490 }, { "epoch": 33.460803059273424, "grad_norm": 0.6321269273757935, "learning_rate": 9.578888772705785e-05, "loss": 0.045, "step": 52500 }, { "epoch": 33.467176545570425, "grad_norm": 0.41663962602615356, "learning_rate": 9.578667353364237e-05, "loss": 0.0252, "step": 52510 }, { "epoch": 33.47355003186743, "grad_norm": 0.3875863254070282, "learning_rate": 9.578445878387556e-05, "loss": 0.0253, "step": 52520 }, { "epoch": 33.47992351816443, "grad_norm": 0.4657875895500183, "learning_rate": 9.578224347778437e-05, "loss": 0.026, "step": 52530 }, { "epoch": 33.48629700446144, "grad_norm": 0.6377483606338501, "learning_rate": 9.578002761539568e-05, "loss": 0.0245, "step": 52540 }, { "epoch": 33.49267049075844, "grad_norm": 0.6499089598655701, "learning_rate": 9.577781119673645e-05, "loss": 0.0355, "step": 52550 }, { "epoch": 33.49904397705545, "grad_norm": 0.45889222621917725, "learning_rate": 9.577559422183357e-05, "loss": 0.0239, "step": 52560 }, { "epoch": 33.50541746335245, "grad_norm": 0.5011894702911377, "learning_rate": 9.577337669071403e-05, "loss": 0.0304, "step": 52570 }, { "epoch": 33.51179094964946, "grad_norm": 0.37935492396354675, "learning_rate": 9.577115860340473e-05, "loss": 0.0266, "step": 52580 }, { "epoch": 33.51816443594646, "grad_norm": 0.3869948983192444, "learning_rate": 9.576893995993264e-05, "loss": 0.0344, "step": 52590 }, { "epoch": 33.52453792224347, "grad_norm": 0.5484520792961121, "learning_rate": 9.57667207603247e-05, "loss": 0.0314, "step": 52600 }, { "epoch": 33.53091140854047, "grad_norm": 0.3637152910232544, "learning_rate": 9.57645010046079e-05, "loss": 0.0284, "step": 52610 }, { "epoch": 33.53728489483748, "grad_norm": 0.3845587372779846, "learning_rate": 9.576228069280921e-05, "loss": 0.0223, "step": 52620 }, { "epoch": 33.54365838113448, "grad_norm": 0.46730637550354004, "learning_rate": 9.576005982495562e-05, "loss": 0.0221, "step": 52630 }, { "epoch": 33.550031867431485, "grad_norm": 0.44513535499572754, "learning_rate": 9.575783840107407e-05, "loss": 0.0245, "step": 52640 }, { "epoch": 33.556405353728486, "grad_norm": 0.40078699588775635, "learning_rate": 9.575561642119159e-05, "loss": 0.025, "step": 52650 }, { "epoch": 33.562778840025494, "grad_norm": 0.3006082773208618, "learning_rate": 9.575339388533517e-05, "loss": 0.025, "step": 52660 }, { "epoch": 33.569152326322495, "grad_norm": 0.3805831968784332, "learning_rate": 9.575117079353182e-05, "loss": 0.0305, "step": 52670 }, { "epoch": 33.5755258126195, "grad_norm": 0.4649515151977539, "learning_rate": 9.574894714580854e-05, "loss": 0.0316, "step": 52680 }, { "epoch": 33.581899298916504, "grad_norm": 0.446331650018692, "learning_rate": 9.574672294219237e-05, "loss": 0.0342, "step": 52690 }, { "epoch": 33.58827278521351, "grad_norm": 0.5373594760894775, "learning_rate": 9.574449818271032e-05, "loss": 0.0227, "step": 52700 }, { "epoch": 33.59464627151051, "grad_norm": 0.5900424718856812, "learning_rate": 9.574227286738942e-05, "loss": 0.0319, "step": 52710 }, { "epoch": 33.60101975780752, "grad_norm": 0.9526788592338562, "learning_rate": 9.574004699625671e-05, "loss": 0.0295, "step": 52720 }, { "epoch": 33.60739324410453, "grad_norm": 0.6568556427955627, "learning_rate": 9.573782056933927e-05, "loss": 0.0283, "step": 52730 }, { "epoch": 33.61376673040153, "grad_norm": 0.27466288208961487, "learning_rate": 9.57355935866641e-05, "loss": 0.0214, "step": 52740 }, { "epoch": 33.62014021669854, "grad_norm": 0.32533931732177734, "learning_rate": 9.573336604825831e-05, "loss": 0.0281, "step": 52750 }, { "epoch": 33.62651370299554, "grad_norm": 0.2134103775024414, "learning_rate": 9.573113795414891e-05, "loss": 0.0266, "step": 52760 }, { "epoch": 33.632887189292546, "grad_norm": 0.39872175455093384, "learning_rate": 9.572890930436305e-05, "loss": 0.026, "step": 52770 }, { "epoch": 33.63926067558955, "grad_norm": 0.6040008068084717, "learning_rate": 9.572668009892773e-05, "loss": 0.0437, "step": 52780 }, { "epoch": 33.645634161886555, "grad_norm": 0.49851861596107483, "learning_rate": 9.57244503378701e-05, "loss": 0.0274, "step": 52790 }, { "epoch": 33.652007648183556, "grad_norm": 0.34538549184799194, "learning_rate": 9.572222002121722e-05, "loss": 0.0386, "step": 52800 }, { "epoch": 33.658381134480564, "grad_norm": 0.28822439908981323, "learning_rate": 9.571998914899619e-05, "loss": 0.0329, "step": 52810 }, { "epoch": 33.664754620777565, "grad_norm": 0.5181159377098083, "learning_rate": 9.571775772123412e-05, "loss": 0.0314, "step": 52820 }, { "epoch": 33.67112810707457, "grad_norm": 0.4672962427139282, "learning_rate": 9.571552573795815e-05, "loss": 0.0314, "step": 52830 }, { "epoch": 33.677501593371574, "grad_norm": 0.509365439414978, "learning_rate": 9.571329319919535e-05, "loss": 0.0278, "step": 52840 }, { "epoch": 33.68387507966858, "grad_norm": 0.3330395519733429, "learning_rate": 9.571106010497291e-05, "loss": 0.0302, "step": 52850 }, { "epoch": 33.69024856596558, "grad_norm": 0.4655858874320984, "learning_rate": 9.57088264553179e-05, "loss": 0.0402, "step": 52860 }, { "epoch": 33.69662205226259, "grad_norm": 0.3230804204940796, "learning_rate": 9.570659225025751e-05, "loss": 0.0281, "step": 52870 }, { "epoch": 33.70299553855959, "grad_norm": 0.23053815960884094, "learning_rate": 9.570435748981888e-05, "loss": 0.0264, "step": 52880 }, { "epoch": 33.7093690248566, "grad_norm": 0.4922768175601959, "learning_rate": 9.570212217402913e-05, "loss": 0.0332, "step": 52890 }, { "epoch": 33.7157425111536, "grad_norm": 0.4251912534236908, "learning_rate": 9.569988630291545e-05, "loss": 0.0373, "step": 52900 }, { "epoch": 33.72211599745061, "grad_norm": 0.6231818795204163, "learning_rate": 9.569764987650502e-05, "loss": 0.0281, "step": 52910 }, { "epoch": 33.72848948374761, "grad_norm": 0.5457112193107605, "learning_rate": 9.569541289482498e-05, "loss": 0.0365, "step": 52920 }, { "epoch": 33.73486297004462, "grad_norm": 0.5576350092887878, "learning_rate": 9.569317535790255e-05, "loss": 0.0275, "step": 52930 }, { "epoch": 33.74123645634162, "grad_norm": 0.43591880798339844, "learning_rate": 9.569093726576488e-05, "loss": 0.0318, "step": 52940 }, { "epoch": 33.747609942638626, "grad_norm": 0.40061303973197937, "learning_rate": 9.568869861843919e-05, "loss": 0.0393, "step": 52950 }, { "epoch": 33.753983428935626, "grad_norm": 0.7621817588806152, "learning_rate": 9.568645941595267e-05, "loss": 0.0419, "step": 52960 }, { "epoch": 33.760356915232634, "grad_norm": 0.5062728524208069, "learning_rate": 9.568421965833254e-05, "loss": 0.0245, "step": 52970 }, { "epoch": 33.766730401529635, "grad_norm": 0.3432307243347168, "learning_rate": 9.568197934560602e-05, "loss": 0.0331, "step": 52980 }, { "epoch": 33.77310388782664, "grad_norm": 0.4563245475292206, "learning_rate": 9.567973847780029e-05, "loss": 0.0379, "step": 52990 }, { "epoch": 33.779477374123644, "grad_norm": 0.3369235098361969, "learning_rate": 9.567749705494263e-05, "loss": 0.0222, "step": 53000 }, { "epoch": 33.78585086042065, "grad_norm": 0.65120530128479, "learning_rate": 9.567525507706024e-05, "loss": 0.0305, "step": 53010 }, { "epoch": 33.79222434671765, "grad_norm": 0.572632372379303, "learning_rate": 9.567301254418038e-05, "loss": 0.0249, "step": 53020 }, { "epoch": 33.79859783301466, "grad_norm": 0.6785462498664856, "learning_rate": 9.567076945633031e-05, "loss": 0.0331, "step": 53030 }, { "epoch": 33.80497131931166, "grad_norm": 0.4580390155315399, "learning_rate": 9.566852581353726e-05, "loss": 0.0363, "step": 53040 }, { "epoch": 33.81134480560867, "grad_norm": 0.27569130063056946, "learning_rate": 9.56662816158285e-05, "loss": 0.0338, "step": 53050 }, { "epoch": 33.81771829190567, "grad_norm": 0.46244460344314575, "learning_rate": 9.566403686323131e-05, "loss": 0.0356, "step": 53060 }, { "epoch": 33.82409177820268, "grad_norm": 0.3413628935813904, "learning_rate": 9.566179155577294e-05, "loss": 0.0303, "step": 53070 }, { "epoch": 33.83046526449968, "grad_norm": 0.34665679931640625, "learning_rate": 9.565954569348072e-05, "loss": 0.0355, "step": 53080 }, { "epoch": 33.83683875079669, "grad_norm": 0.4611565172672272, "learning_rate": 9.565729927638188e-05, "loss": 0.051, "step": 53090 }, { "epoch": 33.84321223709369, "grad_norm": 0.6459419131278992, "learning_rate": 9.565505230450376e-05, "loss": 0.0478, "step": 53100 }, { "epoch": 33.849585723390696, "grad_norm": 0.39568012952804565, "learning_rate": 9.565280477787367e-05, "loss": 0.0346, "step": 53110 }, { "epoch": 33.8559592096877, "grad_norm": 0.47443461418151855, "learning_rate": 9.565055669651886e-05, "loss": 0.0283, "step": 53120 }, { "epoch": 33.862332695984705, "grad_norm": 0.6358253359794617, "learning_rate": 9.56483080604667e-05, "loss": 0.0392, "step": 53130 }, { "epoch": 33.868706182281706, "grad_norm": 0.4592743217945099, "learning_rate": 9.564605886974451e-05, "loss": 0.0295, "step": 53140 }, { "epoch": 33.875079668578714, "grad_norm": 0.42860621213912964, "learning_rate": 9.56438091243796e-05, "loss": 0.0276, "step": 53150 }, { "epoch": 33.881453154875715, "grad_norm": 0.5611061453819275, "learning_rate": 9.564155882439931e-05, "loss": 0.0396, "step": 53160 }, { "epoch": 33.88782664117272, "grad_norm": 0.48615095019340515, "learning_rate": 9.563930796983098e-05, "loss": 0.0377, "step": 53170 }, { "epoch": 33.89420012746972, "grad_norm": 0.42084577679634094, "learning_rate": 9.563705656070198e-05, "loss": 0.0301, "step": 53180 }, { "epoch": 33.90057361376673, "grad_norm": 0.567521333694458, "learning_rate": 9.563480459703966e-05, "loss": 0.0427, "step": 53190 }, { "epoch": 33.90694710006373, "grad_norm": 0.5445899963378906, "learning_rate": 9.563255207887136e-05, "loss": 0.0334, "step": 53200 }, { "epoch": 33.91332058636074, "grad_norm": 0.4061892330646515, "learning_rate": 9.563029900622449e-05, "loss": 0.0293, "step": 53210 }, { "epoch": 33.91969407265774, "grad_norm": 0.2565276026725769, "learning_rate": 9.562804537912639e-05, "loss": 0.0281, "step": 53220 }, { "epoch": 33.92606755895475, "grad_norm": 0.5534552335739136, "learning_rate": 9.562579119760447e-05, "loss": 0.0274, "step": 53230 }, { "epoch": 33.93244104525175, "grad_norm": 0.3412909209728241, "learning_rate": 9.56235364616861e-05, "loss": 0.0186, "step": 53240 }, { "epoch": 33.93881453154876, "grad_norm": 0.31539711356163025, "learning_rate": 9.562128117139868e-05, "loss": 0.036, "step": 53250 }, { "epoch": 33.94518801784576, "grad_norm": 0.3508473038673401, "learning_rate": 9.561902532676963e-05, "loss": 0.0246, "step": 53260 }, { "epoch": 33.95156150414277, "grad_norm": 0.48333466053009033, "learning_rate": 9.561676892782633e-05, "loss": 0.0296, "step": 53270 }, { "epoch": 33.95793499043977, "grad_norm": 0.5551915764808655, "learning_rate": 9.561451197459625e-05, "loss": 0.0302, "step": 53280 }, { "epoch": 33.964308476736775, "grad_norm": 0.33949464559555054, "learning_rate": 9.561225446710676e-05, "loss": 0.0323, "step": 53290 }, { "epoch": 33.970681963033776, "grad_norm": 0.32581931352615356, "learning_rate": 9.560999640538532e-05, "loss": 0.0208, "step": 53300 }, { "epoch": 33.977055449330784, "grad_norm": 0.39329954981803894, "learning_rate": 9.560773778945938e-05, "loss": 0.038, "step": 53310 }, { "epoch": 33.983428935627785, "grad_norm": 0.6731166243553162, "learning_rate": 9.560547861935634e-05, "loss": 0.0339, "step": 53320 }, { "epoch": 33.98980242192479, "grad_norm": 0.3781405985355377, "learning_rate": 9.560321889510368e-05, "loss": 0.0355, "step": 53330 }, { "epoch": 33.9961759082218, "grad_norm": 0.47978052496910095, "learning_rate": 9.560095861672885e-05, "loss": 0.0261, "step": 53340 }, { "epoch": 34.0025493945188, "grad_norm": 0.4785864055156708, "learning_rate": 9.559869778425933e-05, "loss": 0.0305, "step": 53350 }, { "epoch": 34.00892288081581, "grad_norm": 0.36711546778678894, "learning_rate": 9.559643639772258e-05, "loss": 0.0305, "step": 53360 }, { "epoch": 34.01529636711281, "grad_norm": 0.4253818094730377, "learning_rate": 9.559417445714607e-05, "loss": 0.0293, "step": 53370 }, { "epoch": 34.02166985340982, "grad_norm": 0.6237525343894958, "learning_rate": 9.55919119625573e-05, "loss": 0.0236, "step": 53380 }, { "epoch": 34.02804333970682, "grad_norm": 0.34273648262023926, "learning_rate": 9.558964891398374e-05, "loss": 0.0277, "step": 53390 }, { "epoch": 34.03441682600383, "grad_norm": 0.4689711332321167, "learning_rate": 9.55873853114529e-05, "loss": 0.0246, "step": 53400 }, { "epoch": 34.04079031230083, "grad_norm": 0.24570055305957794, "learning_rate": 9.55851211549923e-05, "loss": 0.0204, "step": 53410 }, { "epoch": 34.047163798597836, "grad_norm": 0.49715811014175415, "learning_rate": 9.558285644462943e-05, "loss": 0.0334, "step": 53420 }, { "epoch": 34.05353728489484, "grad_norm": 0.5334717631340027, "learning_rate": 9.558059118039183e-05, "loss": 0.0255, "step": 53430 }, { "epoch": 34.059910771191845, "grad_norm": 0.546329915523529, "learning_rate": 9.557832536230701e-05, "loss": 0.0222, "step": 53440 }, { "epoch": 34.066284257488846, "grad_norm": 1.4655948877334595, "learning_rate": 9.55760589904025e-05, "loss": 0.0274, "step": 53450 }, { "epoch": 34.072657743785854, "grad_norm": 0.3237227499485016, "learning_rate": 9.557379206470583e-05, "loss": 0.0282, "step": 53460 }, { "epoch": 34.079031230082855, "grad_norm": 0.6544840335845947, "learning_rate": 9.557152458524457e-05, "loss": 0.0323, "step": 53470 }, { "epoch": 34.08540471637986, "grad_norm": 0.2570575475692749, "learning_rate": 9.556925655204626e-05, "loss": 0.0191, "step": 53480 }, { "epoch": 34.09177820267686, "grad_norm": 0.3816926181316376, "learning_rate": 9.556698796513845e-05, "loss": 0.0226, "step": 53490 }, { "epoch": 34.09815168897387, "grad_norm": 0.49775460362434387, "learning_rate": 9.556471882454872e-05, "loss": 0.0448, "step": 53500 }, { "epoch": 34.10452517527087, "grad_norm": 0.4969298243522644, "learning_rate": 9.556244913030465e-05, "loss": 0.0225, "step": 53510 }, { "epoch": 34.11089866156788, "grad_norm": 0.4886188209056854, "learning_rate": 9.556017888243378e-05, "loss": 0.0261, "step": 53520 }, { "epoch": 34.11727214786488, "grad_norm": 0.5399004817008972, "learning_rate": 9.555790808096374e-05, "loss": 0.0265, "step": 53530 }, { "epoch": 34.12364563416189, "grad_norm": 0.28880321979522705, "learning_rate": 9.55556367259221e-05, "loss": 0.019, "step": 53540 }, { "epoch": 34.13001912045889, "grad_norm": 0.4671781361103058, "learning_rate": 9.555336481733644e-05, "loss": 0.0374, "step": 53550 }, { "epoch": 34.1363926067559, "grad_norm": 0.2675691246986389, "learning_rate": 9.55510923552344e-05, "loss": 0.0214, "step": 53560 }, { "epoch": 34.1427660930529, "grad_norm": 0.2760460674762726, "learning_rate": 9.554881933964359e-05, "loss": 0.029, "step": 53570 }, { "epoch": 34.14913957934991, "grad_norm": 0.5502656102180481, "learning_rate": 9.554654577059163e-05, "loss": 0.0203, "step": 53580 }, { "epoch": 34.15551306564691, "grad_norm": 0.5731120109558105, "learning_rate": 9.554427164810611e-05, "loss": 0.0248, "step": 53590 }, { "epoch": 34.161886551943915, "grad_norm": 0.5067721605300903, "learning_rate": 9.55419969722147e-05, "loss": 0.0273, "step": 53600 }, { "epoch": 34.168260038240916, "grad_norm": 0.3770888149738312, "learning_rate": 9.553972174294503e-05, "loss": 0.0276, "step": 53610 }, { "epoch": 34.174633524537924, "grad_norm": 0.45629173517227173, "learning_rate": 9.553744596032474e-05, "loss": 0.0292, "step": 53620 }, { "epoch": 34.181007010834925, "grad_norm": 0.6713749170303345, "learning_rate": 9.553516962438149e-05, "loss": 0.0247, "step": 53630 }, { "epoch": 34.18738049713193, "grad_norm": 0.6958839297294617, "learning_rate": 9.553289273514294e-05, "loss": 0.037, "step": 53640 }, { "epoch": 34.193753983428934, "grad_norm": 0.5203536152839661, "learning_rate": 9.553061529263673e-05, "loss": 0.0298, "step": 53650 }, { "epoch": 34.20012746972594, "grad_norm": 0.5386590957641602, "learning_rate": 9.55283372968906e-05, "loss": 0.0307, "step": 53660 }, { "epoch": 34.20650095602294, "grad_norm": 0.5252214074134827, "learning_rate": 9.552605874793214e-05, "loss": 0.0223, "step": 53670 }, { "epoch": 34.21287444231995, "grad_norm": 0.31366869807243347, "learning_rate": 9.552377964578912e-05, "loss": 0.0292, "step": 53680 }, { "epoch": 34.21924792861695, "grad_norm": 0.39047500491142273, "learning_rate": 9.552149999048917e-05, "loss": 0.0267, "step": 53690 }, { "epoch": 34.22562141491396, "grad_norm": 0.6531780362129211, "learning_rate": 9.551921978206001e-05, "loss": 0.0413, "step": 53700 }, { "epoch": 34.23199490121096, "grad_norm": 0.5505098700523376, "learning_rate": 9.551693902052938e-05, "loss": 0.0258, "step": 53710 }, { "epoch": 34.23836838750797, "grad_norm": 0.575623631477356, "learning_rate": 9.551465770592495e-05, "loss": 0.0319, "step": 53720 }, { "epoch": 34.24474187380497, "grad_norm": 0.4888959228992462, "learning_rate": 9.551237583827446e-05, "loss": 0.0325, "step": 53730 }, { "epoch": 34.25111536010198, "grad_norm": 0.2585799992084503, "learning_rate": 9.551009341760564e-05, "loss": 0.0349, "step": 53740 }, { "epoch": 34.25748884639898, "grad_norm": 0.5387828946113586, "learning_rate": 9.550781044394621e-05, "loss": 0.0234, "step": 53750 }, { "epoch": 34.263862332695986, "grad_norm": 0.5920343995094299, "learning_rate": 9.55055269173239e-05, "loss": 0.031, "step": 53760 }, { "epoch": 34.27023581899299, "grad_norm": 0.4863366186618805, "learning_rate": 9.55032428377665e-05, "loss": 0.0467, "step": 53770 }, { "epoch": 34.276609305289995, "grad_norm": 0.3027154505252838, "learning_rate": 9.550095820530172e-05, "loss": 0.0272, "step": 53780 }, { "epoch": 34.282982791586996, "grad_norm": 0.7009127140045166, "learning_rate": 9.549867301995734e-05, "loss": 0.0244, "step": 53790 }, { "epoch": 34.289356277884, "grad_norm": 0.816160261631012, "learning_rate": 9.549638728176114e-05, "loss": 0.0365, "step": 53800 }, { "epoch": 34.295729764181004, "grad_norm": 0.48131388425827026, "learning_rate": 9.549410099074085e-05, "loss": 0.0234, "step": 53810 }, { "epoch": 34.30210325047801, "grad_norm": 0.49172544479370117, "learning_rate": 9.549181414692431e-05, "loss": 0.033, "step": 53820 }, { "epoch": 34.30847673677501, "grad_norm": 0.35445553064346313, "learning_rate": 9.548952675033926e-05, "loss": 0.0307, "step": 53830 }, { "epoch": 34.31485022307202, "grad_norm": 0.24412131309509277, "learning_rate": 9.548723880101351e-05, "loss": 0.0259, "step": 53840 }, { "epoch": 34.32122370936902, "grad_norm": 0.627521276473999, "learning_rate": 9.548495029897486e-05, "loss": 0.0314, "step": 53850 }, { "epoch": 34.32759719566603, "grad_norm": 0.31412380933761597, "learning_rate": 9.548266124425113e-05, "loss": 0.0297, "step": 53860 }, { "epoch": 34.33397068196303, "grad_norm": 0.2314710021018982, "learning_rate": 9.548037163687011e-05, "loss": 0.02, "step": 53870 }, { "epoch": 34.34034416826004, "grad_norm": 0.7032288312911987, "learning_rate": 9.547808147685965e-05, "loss": 0.0356, "step": 53880 }, { "epoch": 34.34671765455704, "grad_norm": 0.43201225996017456, "learning_rate": 9.547579076424756e-05, "loss": 0.0228, "step": 53890 }, { "epoch": 34.35309114085405, "grad_norm": 0.18556854128837585, "learning_rate": 9.547349949906166e-05, "loss": 0.0156, "step": 53900 }, { "epoch": 34.35946462715105, "grad_norm": 0.32275819778442383, "learning_rate": 9.547120768132982e-05, "loss": 0.0193, "step": 53910 }, { "epoch": 34.365838113448056, "grad_norm": 0.7590206265449524, "learning_rate": 9.546891531107986e-05, "loss": 0.0311, "step": 53920 }, { "epoch": 34.37221159974506, "grad_norm": 0.5695422291755676, "learning_rate": 9.546662238833967e-05, "loss": 0.0305, "step": 53930 }, { "epoch": 34.378585086042065, "grad_norm": 0.5140156745910645, "learning_rate": 9.546432891313708e-05, "loss": 0.0426, "step": 53940 }, { "epoch": 34.384958572339066, "grad_norm": 0.3449597656726837, "learning_rate": 9.546203488549997e-05, "loss": 0.0425, "step": 53950 }, { "epoch": 34.391332058636074, "grad_norm": 0.44223153591156006, "learning_rate": 9.545974030545621e-05, "loss": 0.0258, "step": 53960 }, { "epoch": 34.397705544933075, "grad_norm": 0.780102014541626, "learning_rate": 9.545744517303368e-05, "loss": 0.0477, "step": 53970 }, { "epoch": 34.40407903123008, "grad_norm": 0.7015387415885925, "learning_rate": 9.545514948826028e-05, "loss": 0.0457, "step": 53980 }, { "epoch": 34.41045251752709, "grad_norm": 0.6713115572929382, "learning_rate": 9.54528532511639e-05, "loss": 0.0209, "step": 53990 }, { "epoch": 34.41682600382409, "grad_norm": 0.9521574974060059, "learning_rate": 9.545055646177243e-05, "loss": 0.0448, "step": 54000 }, { "epoch": 34.4231994901211, "grad_norm": 0.4362484812736511, "learning_rate": 9.544825912011379e-05, "loss": 0.022, "step": 54010 }, { "epoch": 34.4295729764181, "grad_norm": 0.46086394786834717, "learning_rate": 9.54459612262159e-05, "loss": 0.0295, "step": 54020 }, { "epoch": 34.43594646271511, "grad_norm": 0.4730875790119171, "learning_rate": 9.544366278010666e-05, "loss": 0.0276, "step": 54030 }, { "epoch": 34.44231994901211, "grad_norm": 0.5375385284423828, "learning_rate": 9.544136378181401e-05, "loss": 0.0375, "step": 54040 }, { "epoch": 34.44869343530912, "grad_norm": 0.4185850918292999, "learning_rate": 9.543906423136589e-05, "loss": 0.0377, "step": 54050 }, { "epoch": 34.45506692160612, "grad_norm": 0.4055817425251007, "learning_rate": 9.543676412879024e-05, "loss": 0.0207, "step": 54060 }, { "epoch": 34.461440407903126, "grad_norm": 0.27319684624671936, "learning_rate": 9.543446347411501e-05, "loss": 0.0205, "step": 54070 }, { "epoch": 34.46781389420013, "grad_norm": 0.43446555733680725, "learning_rate": 9.543216226736815e-05, "loss": 0.0282, "step": 54080 }, { "epoch": 34.474187380497135, "grad_norm": 0.3387250304222107, "learning_rate": 9.542986050857761e-05, "loss": 0.02, "step": 54090 }, { "epoch": 34.480560866794136, "grad_norm": 0.3333646357059479, "learning_rate": 9.542755819777139e-05, "loss": 0.0295, "step": 54100 }, { "epoch": 34.486934353091144, "grad_norm": 0.4125802516937256, "learning_rate": 9.542525533497743e-05, "loss": 0.0177, "step": 54110 }, { "epoch": 34.493307839388144, "grad_norm": 0.4044288694858551, "learning_rate": 9.542295192022374e-05, "loss": 0.0353, "step": 54120 }, { "epoch": 34.49968132568515, "grad_norm": 0.3926030695438385, "learning_rate": 9.542064795353831e-05, "loss": 0.0182, "step": 54130 }, { "epoch": 34.50605481198215, "grad_norm": 0.3920355439186096, "learning_rate": 9.541834343494911e-05, "loss": 0.0328, "step": 54140 }, { "epoch": 34.51242829827916, "grad_norm": 0.44078636169433594, "learning_rate": 9.541603836448417e-05, "loss": 0.02, "step": 54150 }, { "epoch": 34.51880178457616, "grad_norm": 0.31956973671913147, "learning_rate": 9.541373274217145e-05, "loss": 0.0377, "step": 54160 }, { "epoch": 34.52517527087317, "grad_norm": 0.3992490768432617, "learning_rate": 9.541142656803903e-05, "loss": 0.0325, "step": 54170 }, { "epoch": 34.53154875717017, "grad_norm": 0.8118324279785156, "learning_rate": 9.540911984211489e-05, "loss": 0.0326, "step": 54180 }, { "epoch": 34.53792224346718, "grad_norm": 0.4829874336719513, "learning_rate": 9.540681256442707e-05, "loss": 0.026, "step": 54190 }, { "epoch": 34.54429572976418, "grad_norm": 0.31661680340766907, "learning_rate": 9.54045047350036e-05, "loss": 0.0285, "step": 54200 }, { "epoch": 34.55066921606119, "grad_norm": 0.28740260004997253, "learning_rate": 9.540219635387253e-05, "loss": 0.0189, "step": 54210 }, { "epoch": 34.55704270235819, "grad_norm": 0.45540308952331543, "learning_rate": 9.539988742106191e-05, "loss": 0.0361, "step": 54220 }, { "epoch": 34.5634161886552, "grad_norm": 0.7181097865104675, "learning_rate": 9.53975779365998e-05, "loss": 0.0308, "step": 54230 }, { "epoch": 34.5697896749522, "grad_norm": 0.449716717004776, "learning_rate": 9.539526790051426e-05, "loss": 0.0214, "step": 54240 }, { "epoch": 34.576163161249205, "grad_norm": 0.5261098742485046, "learning_rate": 9.539295731283336e-05, "loss": 0.0302, "step": 54250 }, { "epoch": 34.582536647546206, "grad_norm": 0.5469397306442261, "learning_rate": 9.539064617358515e-05, "loss": 0.0325, "step": 54260 }, { "epoch": 34.588910133843214, "grad_norm": 0.5277042984962463, "learning_rate": 9.538833448279773e-05, "loss": 0.0227, "step": 54270 }, { "epoch": 34.595283620140215, "grad_norm": 0.4310486912727356, "learning_rate": 9.538602224049922e-05, "loss": 0.0258, "step": 54280 }, { "epoch": 34.60165710643722, "grad_norm": 0.5732216238975525, "learning_rate": 9.538370944671766e-05, "loss": 0.029, "step": 54290 }, { "epoch": 34.608030592734224, "grad_norm": 0.6305785775184631, "learning_rate": 9.53813961014812e-05, "loss": 0.0409, "step": 54300 }, { "epoch": 34.61440407903123, "grad_norm": 0.5343277454376221, "learning_rate": 9.537908220481791e-05, "loss": 0.0303, "step": 54310 }, { "epoch": 34.62077756532823, "grad_norm": 4.150912761688232, "learning_rate": 9.537676775675595e-05, "loss": 0.028, "step": 54320 }, { "epoch": 34.62715105162524, "grad_norm": 0.5228842496871948, "learning_rate": 9.537445275732341e-05, "loss": 0.0355, "step": 54330 }, { "epoch": 34.63352453792224, "grad_norm": 0.48362424969673157, "learning_rate": 9.537213720654843e-05, "loss": 0.035, "step": 54340 }, { "epoch": 34.63989802421925, "grad_norm": 0.5665836334228516, "learning_rate": 9.536982110445914e-05, "loss": 0.0269, "step": 54350 }, { "epoch": 34.64627151051625, "grad_norm": 0.64130699634552, "learning_rate": 9.536750445108367e-05, "loss": 0.0272, "step": 54360 }, { "epoch": 34.65264499681326, "grad_norm": 0.38070589303970337, "learning_rate": 9.536518724645022e-05, "loss": 0.0248, "step": 54370 }, { "epoch": 34.65901848311026, "grad_norm": 0.5269302129745483, "learning_rate": 9.53628694905869e-05, "loss": 0.0277, "step": 54380 }, { "epoch": 34.66539196940727, "grad_norm": 0.65503990650177, "learning_rate": 9.536055118352187e-05, "loss": 0.0479, "step": 54390 }, { "epoch": 34.67176545570427, "grad_norm": 0.45364704728126526, "learning_rate": 9.535823232528334e-05, "loss": 0.0215, "step": 54400 }, { "epoch": 34.678138942001276, "grad_norm": 0.4646311402320862, "learning_rate": 9.535591291589945e-05, "loss": 0.0329, "step": 54410 }, { "epoch": 34.68451242829828, "grad_norm": 0.9024292230606079, "learning_rate": 9.53535929553984e-05, "loss": 0.0274, "step": 54420 }, { "epoch": 34.690885914595285, "grad_norm": 0.7289599776268005, "learning_rate": 9.535127244380837e-05, "loss": 0.0274, "step": 54430 }, { "epoch": 34.697259400892285, "grad_norm": 0.42252904176712036, "learning_rate": 9.534895138115755e-05, "loss": 0.0196, "step": 54440 }, { "epoch": 34.70363288718929, "grad_norm": 0.6122227907180786, "learning_rate": 9.534662976747418e-05, "loss": 0.0498, "step": 54450 }, { "epoch": 34.710006373486294, "grad_norm": 0.48873403668403625, "learning_rate": 9.534430760278644e-05, "loss": 0.027, "step": 54460 }, { "epoch": 34.7163798597833, "grad_norm": 0.4227951169013977, "learning_rate": 9.534198488712253e-05, "loss": 0.0221, "step": 54470 }, { "epoch": 34.7227533460803, "grad_norm": 0.512873113155365, "learning_rate": 9.533966162051071e-05, "loss": 0.032, "step": 54480 }, { "epoch": 34.72912683237731, "grad_norm": 0.6129639744758606, "learning_rate": 9.533733780297919e-05, "loss": 0.0273, "step": 54490 }, { "epoch": 34.73550031867431, "grad_norm": 0.4442170262336731, "learning_rate": 9.53350134345562e-05, "loss": 0.0362, "step": 54500 }, { "epoch": 34.74187380497132, "grad_norm": 0.2548813819885254, "learning_rate": 9.533268851527001e-05, "loss": 0.0289, "step": 54510 }, { "epoch": 34.74824729126832, "grad_norm": 0.27000072598457336, "learning_rate": 9.533036304514884e-05, "loss": 0.0314, "step": 54520 }, { "epoch": 34.75462077756533, "grad_norm": 0.4531651437282562, "learning_rate": 9.532803702422098e-05, "loss": 0.0336, "step": 54530 }, { "epoch": 34.76099426386233, "grad_norm": 0.4736124277114868, "learning_rate": 9.532571045251465e-05, "loss": 0.0235, "step": 54540 }, { "epoch": 34.76736775015934, "grad_norm": 0.3686412274837494, "learning_rate": 9.532338333005815e-05, "loss": 0.0238, "step": 54550 }, { "epoch": 34.77374123645634, "grad_norm": 0.5054984092712402, "learning_rate": 9.532105565687976e-05, "loss": 0.0394, "step": 54560 }, { "epoch": 34.780114722753346, "grad_norm": 0.49800899624824524, "learning_rate": 9.531872743300774e-05, "loss": 0.021, "step": 54570 }, { "epoch": 34.78648820905035, "grad_norm": 0.3700025975704193, "learning_rate": 9.531639865847039e-05, "loss": 0.0244, "step": 54580 }, { "epoch": 34.792861695347355, "grad_norm": 0.39722928404808044, "learning_rate": 9.531406933329602e-05, "loss": 0.0259, "step": 54590 }, { "epoch": 34.79923518164436, "grad_norm": 0.5059988498687744, "learning_rate": 9.531173945751293e-05, "loss": 0.0236, "step": 54600 }, { "epoch": 34.805608667941364, "grad_norm": 0.48618635535240173, "learning_rate": 9.530940903114942e-05, "loss": 0.0219, "step": 54610 }, { "epoch": 34.81198215423837, "grad_norm": 0.49334630370140076, "learning_rate": 9.530707805423381e-05, "loss": 0.0348, "step": 54620 }, { "epoch": 34.81835564053537, "grad_norm": 0.4374242126941681, "learning_rate": 9.530474652679444e-05, "loss": 0.0409, "step": 54630 }, { "epoch": 34.82472912683238, "grad_norm": 0.661401093006134, "learning_rate": 9.530241444885961e-05, "loss": 0.0306, "step": 54640 }, { "epoch": 34.83110261312938, "grad_norm": 0.5589414834976196, "learning_rate": 9.530008182045766e-05, "loss": 0.0268, "step": 54650 }, { "epoch": 34.83747609942639, "grad_norm": 0.626943051815033, "learning_rate": 9.529774864161695e-05, "loss": 0.0254, "step": 54660 }, { "epoch": 34.84384958572339, "grad_norm": 0.5908910632133484, "learning_rate": 9.529541491236585e-05, "loss": 0.0273, "step": 54670 }, { "epoch": 34.8502230720204, "grad_norm": 0.3634604513645172, "learning_rate": 9.529308063273265e-05, "loss": 0.0218, "step": 54680 }, { "epoch": 34.8565965583174, "grad_norm": 0.5779263377189636, "learning_rate": 9.529074580274579e-05, "loss": 0.0247, "step": 54690 }, { "epoch": 34.86297004461441, "grad_norm": 0.4129598140716553, "learning_rate": 9.52884104224336e-05, "loss": 0.0397, "step": 54700 }, { "epoch": 34.86934353091141, "grad_norm": 0.5561365485191345, "learning_rate": 9.528607449182447e-05, "loss": 0.0272, "step": 54710 }, { "epoch": 34.875717017208416, "grad_norm": 0.7617877125740051, "learning_rate": 9.528373801094676e-05, "loss": 0.02, "step": 54720 }, { "epoch": 34.88209050350542, "grad_norm": 0.5030522346496582, "learning_rate": 9.528140097982889e-05, "loss": 0.035, "step": 54730 }, { "epoch": 34.888463989802425, "grad_norm": 0.4373551309108734, "learning_rate": 9.527906339849925e-05, "loss": 0.037, "step": 54740 }, { "epoch": 34.894837476099426, "grad_norm": 0.5599916577339172, "learning_rate": 9.527672526698624e-05, "loss": 0.0296, "step": 54750 }, { "epoch": 34.90121096239643, "grad_norm": 0.9060662984848022, "learning_rate": 9.527438658531826e-05, "loss": 0.0271, "step": 54760 }, { "epoch": 34.907584448693434, "grad_norm": 0.28670892119407654, "learning_rate": 9.527204735352374e-05, "loss": 0.026, "step": 54770 }, { "epoch": 34.91395793499044, "grad_norm": 0.49428582191467285, "learning_rate": 9.52697075716311e-05, "loss": 0.0417, "step": 54780 }, { "epoch": 34.92033142128744, "grad_norm": 0.6374714970588684, "learning_rate": 9.526736723966878e-05, "loss": 0.0271, "step": 54790 }, { "epoch": 34.92670490758445, "grad_norm": 0.5660468935966492, "learning_rate": 9.52650263576652e-05, "loss": 0.0277, "step": 54800 }, { "epoch": 34.93307839388145, "grad_norm": 0.3216094970703125, "learning_rate": 9.52626849256488e-05, "loss": 0.032, "step": 54810 }, { "epoch": 34.93945188017846, "grad_norm": 0.5788775682449341, "learning_rate": 9.526034294364807e-05, "loss": 0.0273, "step": 54820 }, { "epoch": 34.94582536647546, "grad_norm": 0.5020880699157715, "learning_rate": 9.525800041169142e-05, "loss": 0.0226, "step": 54830 }, { "epoch": 34.95219885277247, "grad_norm": 0.542186975479126, "learning_rate": 9.525565732980735e-05, "loss": 0.0286, "step": 54840 }, { "epoch": 34.95857233906947, "grad_norm": 0.2930257320404053, "learning_rate": 9.525331369802431e-05, "loss": 0.0336, "step": 54850 }, { "epoch": 34.96494582536648, "grad_norm": 0.3854008913040161, "learning_rate": 9.525096951637078e-05, "loss": 0.0238, "step": 54860 }, { "epoch": 34.97131931166348, "grad_norm": 0.4955374598503113, "learning_rate": 9.524862478487524e-05, "loss": 0.0243, "step": 54870 }, { "epoch": 34.977692797960486, "grad_norm": 0.4580172300338745, "learning_rate": 9.52462795035662e-05, "loss": 0.0214, "step": 54880 }, { "epoch": 34.98406628425749, "grad_norm": 0.5698228478431702, "learning_rate": 9.524393367247212e-05, "loss": 0.0247, "step": 54890 }, { "epoch": 34.990439770554495, "grad_norm": 0.955376386642456, "learning_rate": 9.524158729162155e-05, "loss": 0.0398, "step": 54900 }, { "epoch": 34.996813256851496, "grad_norm": 0.3933910131454468, "learning_rate": 9.523924036104297e-05, "loss": 0.0243, "step": 54910 }, { "epoch": 35.003186743148504, "grad_norm": 0.5582911372184753, "learning_rate": 9.523689288076491e-05, "loss": 0.0293, "step": 54920 }, { "epoch": 35.009560229445505, "grad_norm": 0.6082283854484558, "learning_rate": 9.52345448508159e-05, "loss": 0.0252, "step": 54930 }, { "epoch": 35.01593371574251, "grad_norm": 0.5585810542106628, "learning_rate": 9.523219627122445e-05, "loss": 0.0357, "step": 54940 }, { "epoch": 35.022307202039514, "grad_norm": 0.4258825480937958, "learning_rate": 9.522984714201911e-05, "loss": 0.022, "step": 54950 }, { "epoch": 35.02868068833652, "grad_norm": 0.522302508354187, "learning_rate": 9.522749746322844e-05, "loss": 0.0243, "step": 54960 }, { "epoch": 35.03505417463352, "grad_norm": 0.3039967715740204, "learning_rate": 9.522514723488095e-05, "loss": 0.0359, "step": 54970 }, { "epoch": 35.04142766093053, "grad_norm": 0.3911643624305725, "learning_rate": 9.522279645700522e-05, "loss": 0.0358, "step": 54980 }, { "epoch": 35.04780114722753, "grad_norm": 0.35110360383987427, "learning_rate": 9.522044512962984e-05, "loss": 0.0293, "step": 54990 }, { "epoch": 35.05417463352454, "grad_norm": 0.5617950558662415, "learning_rate": 9.521809325278335e-05, "loss": 0.0383, "step": 55000 }, { "epoch": 35.06054811982154, "grad_norm": 0.4283028841018677, "learning_rate": 9.521574082649431e-05, "loss": 0.0341, "step": 55010 }, { "epoch": 35.06692160611855, "grad_norm": 0.5666895508766174, "learning_rate": 9.521338785079134e-05, "loss": 0.0288, "step": 55020 }, { "epoch": 35.07329509241555, "grad_norm": 0.381862074136734, "learning_rate": 9.521103432570304e-05, "loss": 0.0352, "step": 55030 }, { "epoch": 35.07966857871256, "grad_norm": 1.3599945306777954, "learning_rate": 9.520868025125798e-05, "loss": 0.0284, "step": 55040 }, { "epoch": 35.08604206500956, "grad_norm": 0.6772026419639587, "learning_rate": 9.520632562748474e-05, "loss": 0.0346, "step": 55050 }, { "epoch": 35.092415551306566, "grad_norm": 0.4685651361942291, "learning_rate": 9.5203970454412e-05, "loss": 0.0257, "step": 55060 }, { "epoch": 35.09878903760357, "grad_norm": 0.5068112015724182, "learning_rate": 9.520161473206832e-05, "loss": 0.0318, "step": 55070 }, { "epoch": 35.105162523900574, "grad_norm": 0.38628146052360535, "learning_rate": 9.519925846048234e-05, "loss": 0.0299, "step": 55080 }, { "epoch": 35.111536010197575, "grad_norm": 0.4938599169254303, "learning_rate": 9.519690163968272e-05, "loss": 0.0234, "step": 55090 }, { "epoch": 35.11790949649458, "grad_norm": 0.4402371048927307, "learning_rate": 9.519454426969803e-05, "loss": 0.0356, "step": 55100 }, { "epoch": 35.124282982791584, "grad_norm": 1.0453944206237793, "learning_rate": 9.5192186350557e-05, "loss": 0.0309, "step": 55110 }, { "epoch": 35.13065646908859, "grad_norm": 0.3025345206260681, "learning_rate": 9.518982788228821e-05, "loss": 0.0282, "step": 55120 }, { "epoch": 35.13702995538559, "grad_norm": 0.4899325966835022, "learning_rate": 9.518746886492035e-05, "loss": 0.0252, "step": 55130 }, { "epoch": 35.1434034416826, "grad_norm": 0.4341687262058258, "learning_rate": 9.518510929848208e-05, "loss": 0.0301, "step": 55140 }, { "epoch": 35.1497769279796, "grad_norm": 0.5626493096351624, "learning_rate": 9.518274918300208e-05, "loss": 0.0349, "step": 55150 }, { "epoch": 35.15615041427661, "grad_norm": 0.3852962851524353, "learning_rate": 9.5180388518509e-05, "loss": 0.0283, "step": 55160 }, { "epoch": 35.16252390057361, "grad_norm": 0.5742796659469604, "learning_rate": 9.517802730503156e-05, "loss": 0.0266, "step": 55170 }, { "epoch": 35.16889738687062, "grad_norm": 0.35183802247047424, "learning_rate": 9.517566554259842e-05, "loss": 0.0202, "step": 55180 }, { "epoch": 35.17527087316762, "grad_norm": 0.7214723825454712, "learning_rate": 9.51733032312383e-05, "loss": 0.0347, "step": 55190 }, { "epoch": 35.18164435946463, "grad_norm": 0.5835719108581543, "learning_rate": 9.51709403709799e-05, "loss": 0.0312, "step": 55200 }, { "epoch": 35.18801784576163, "grad_norm": 0.3250637650489807, "learning_rate": 9.51685769618519e-05, "loss": 0.0226, "step": 55210 }, { "epoch": 35.194391332058636, "grad_norm": 0.37022122740745544, "learning_rate": 9.516621300388306e-05, "loss": 0.0436, "step": 55220 }, { "epoch": 35.20076481835564, "grad_norm": 0.469418466091156, "learning_rate": 9.51638484971021e-05, "loss": 0.0297, "step": 55230 }, { "epoch": 35.207138304652645, "grad_norm": 0.5068747401237488, "learning_rate": 9.516148344153773e-05, "loss": 0.0325, "step": 55240 }, { "epoch": 35.21351179094965, "grad_norm": 0.596798300743103, "learning_rate": 9.515911783721868e-05, "loss": 0.0405, "step": 55250 }, { "epoch": 35.219885277246654, "grad_norm": 0.4990018904209137, "learning_rate": 9.515675168417372e-05, "loss": 0.0284, "step": 55260 }, { "epoch": 35.22625876354366, "grad_norm": 0.44395509362220764, "learning_rate": 9.515438498243161e-05, "loss": 0.0233, "step": 55270 }, { "epoch": 35.23263224984066, "grad_norm": 0.22774653136730194, "learning_rate": 9.515201773202106e-05, "loss": 0.0316, "step": 55280 }, { "epoch": 35.23900573613767, "grad_norm": 0.5739774107933044, "learning_rate": 9.514964993297089e-05, "loss": 0.04, "step": 55290 }, { "epoch": 35.24537922243467, "grad_norm": 0.3615114390850067, "learning_rate": 9.514728158530983e-05, "loss": 0.0371, "step": 55300 }, { "epoch": 35.25175270873168, "grad_norm": 0.41419804096221924, "learning_rate": 9.51449126890667e-05, "loss": 0.0224, "step": 55310 }, { "epoch": 35.25812619502868, "grad_norm": 0.5439110994338989, "learning_rate": 9.514254324427022e-05, "loss": 0.0207, "step": 55320 }, { "epoch": 35.26449968132569, "grad_norm": 0.5592920184135437, "learning_rate": 9.514017325094923e-05, "loss": 0.0262, "step": 55330 }, { "epoch": 35.27087316762269, "grad_norm": 0.5115230083465576, "learning_rate": 9.513780270913253e-05, "loss": 0.0319, "step": 55340 }, { "epoch": 35.2772466539197, "grad_norm": 0.635106086730957, "learning_rate": 9.51354316188489e-05, "loss": 0.0295, "step": 55350 }, { "epoch": 35.2836201402167, "grad_norm": 0.3669269382953644, "learning_rate": 9.513305998012716e-05, "loss": 0.0231, "step": 55360 }, { "epoch": 35.289993626513706, "grad_norm": 0.6878235936164856, "learning_rate": 9.513068779299612e-05, "loss": 0.029, "step": 55370 }, { "epoch": 35.29636711281071, "grad_norm": 0.4709893763065338, "learning_rate": 9.512831505748463e-05, "loss": 0.0277, "step": 55380 }, { "epoch": 35.302740599107715, "grad_norm": 0.8442989587783813, "learning_rate": 9.512594177362148e-05, "loss": 0.0354, "step": 55390 }, { "epoch": 35.309114085404715, "grad_norm": 0.24564409255981445, "learning_rate": 9.512356794143556e-05, "loss": 0.0259, "step": 55400 }, { "epoch": 35.31548757170172, "grad_norm": 0.21884040534496307, "learning_rate": 9.512119356095567e-05, "loss": 0.0216, "step": 55410 }, { "epoch": 35.321861057998724, "grad_norm": 0.42159944772720337, "learning_rate": 9.511881863221068e-05, "loss": 0.0327, "step": 55420 }, { "epoch": 35.32823454429573, "grad_norm": 0.4503278434276581, "learning_rate": 9.511644315522945e-05, "loss": 0.0338, "step": 55430 }, { "epoch": 35.33460803059273, "grad_norm": 0.5629932284355164, "learning_rate": 9.511406713004084e-05, "loss": 0.0256, "step": 55440 }, { "epoch": 35.34098151688974, "grad_norm": 0.3847126364707947, "learning_rate": 9.511169055667371e-05, "loss": 0.0284, "step": 55450 }, { "epoch": 35.34735500318674, "grad_norm": 0.3572671413421631, "learning_rate": 9.510931343515694e-05, "loss": 0.0189, "step": 55460 }, { "epoch": 35.35372848948375, "grad_norm": 0.5695170760154724, "learning_rate": 9.510693576551943e-05, "loss": 0.0322, "step": 55470 }, { "epoch": 35.36010197578075, "grad_norm": 0.5786896347999573, "learning_rate": 9.510455754779006e-05, "loss": 0.0253, "step": 55480 }, { "epoch": 35.36647546207776, "grad_norm": 0.644467294216156, "learning_rate": 9.510217878199774e-05, "loss": 0.0231, "step": 55490 }, { "epoch": 35.37284894837476, "grad_norm": 0.47169211506843567, "learning_rate": 9.509979946817136e-05, "loss": 0.029, "step": 55500 }, { "epoch": 35.37922243467177, "grad_norm": 0.49006518721580505, "learning_rate": 9.509741960633981e-05, "loss": 0.0265, "step": 55510 }, { "epoch": 35.38559592096877, "grad_norm": 0.5282463431358337, "learning_rate": 9.509503919653207e-05, "loss": 0.0198, "step": 55520 }, { "epoch": 35.391969407265776, "grad_norm": 1.186556339263916, "learning_rate": 9.5092658238777e-05, "loss": 0.0295, "step": 55530 }, { "epoch": 35.39834289356278, "grad_norm": 0.3059556186199188, "learning_rate": 9.509027673310358e-05, "loss": 0.0318, "step": 55540 }, { "epoch": 35.404716379859785, "grad_norm": 0.41900452971458435, "learning_rate": 9.508789467954072e-05, "loss": 0.0278, "step": 55550 }, { "epoch": 35.411089866156786, "grad_norm": 0.7526875734329224, "learning_rate": 9.508551207811735e-05, "loss": 0.0279, "step": 55560 }, { "epoch": 35.417463352453794, "grad_norm": 0.5941905975341797, "learning_rate": 9.508312892886247e-05, "loss": 0.0369, "step": 55570 }, { "epoch": 35.423836838750795, "grad_norm": 0.37143465876579285, "learning_rate": 9.5080745231805e-05, "loss": 0.0272, "step": 55580 }, { "epoch": 35.4302103250478, "grad_norm": 0.45726409554481506, "learning_rate": 9.507836098697389e-05, "loss": 0.0271, "step": 55590 }, { "epoch": 35.4365838113448, "grad_norm": 0.38343968987464905, "learning_rate": 9.507597619439816e-05, "loss": 0.0245, "step": 55600 }, { "epoch": 35.44295729764181, "grad_norm": 0.298199862241745, "learning_rate": 9.507359085410675e-05, "loss": 0.027, "step": 55610 }, { "epoch": 35.44933078393881, "grad_norm": 0.6016449332237244, "learning_rate": 9.507120496612865e-05, "loss": 0.0313, "step": 55620 }, { "epoch": 35.45570427023582, "grad_norm": 0.34862378239631653, "learning_rate": 9.506881853049285e-05, "loss": 0.0174, "step": 55630 }, { "epoch": 35.46207775653282, "grad_norm": 0.29296115040779114, "learning_rate": 9.506643154722837e-05, "loss": 0.0238, "step": 55640 }, { "epoch": 35.46845124282983, "grad_norm": 0.4980362057685852, "learning_rate": 9.50640440163642e-05, "loss": 0.0333, "step": 55650 }, { "epoch": 35.47482472912683, "grad_norm": 0.40781158208847046, "learning_rate": 9.506165593792933e-05, "loss": 0.0246, "step": 55660 }, { "epoch": 35.48119821542384, "grad_norm": 0.6440739035606384, "learning_rate": 9.50592673119528e-05, "loss": 0.032, "step": 55670 }, { "epoch": 35.48757170172084, "grad_norm": 0.31374576687812805, "learning_rate": 9.505687813846363e-05, "loss": 0.0252, "step": 55680 }, { "epoch": 35.49394518801785, "grad_norm": 0.5982542037963867, "learning_rate": 9.505448841749085e-05, "loss": 0.0381, "step": 55690 }, { "epoch": 35.50031867431485, "grad_norm": 0.5191935896873474, "learning_rate": 9.505209814906351e-05, "loss": 0.0378, "step": 55700 }, { "epoch": 35.506692160611856, "grad_norm": 0.6178092956542969, "learning_rate": 9.504970733321062e-05, "loss": 0.027, "step": 55710 }, { "epoch": 35.513065646908856, "grad_norm": 0.3382032513618469, "learning_rate": 9.504731596996125e-05, "loss": 0.0292, "step": 55720 }, { "epoch": 35.519439133205864, "grad_norm": 0.4072098135948181, "learning_rate": 9.504492405934448e-05, "loss": 0.0207, "step": 55730 }, { "epoch": 35.525812619502865, "grad_norm": 0.5382463932037354, "learning_rate": 9.504253160138934e-05, "loss": 0.0239, "step": 55740 }, { "epoch": 35.53218610579987, "grad_norm": 0.3895913064479828, "learning_rate": 9.504013859612493e-05, "loss": 0.028, "step": 55750 }, { "epoch": 35.538559592096874, "grad_norm": 0.5287757515907288, "learning_rate": 9.503774504358029e-05, "loss": 0.0402, "step": 55760 }, { "epoch": 35.54493307839388, "grad_norm": 0.28353437781333923, "learning_rate": 9.503535094378453e-05, "loss": 0.021, "step": 55770 }, { "epoch": 35.55130656469088, "grad_norm": 0.2986859381198883, "learning_rate": 9.503295629676674e-05, "loss": 0.04, "step": 55780 }, { "epoch": 35.55768005098789, "grad_norm": 0.5009360909461975, "learning_rate": 9.503056110255601e-05, "loss": 0.022, "step": 55790 }, { "epoch": 35.56405353728489, "grad_norm": 0.41437825560569763, "learning_rate": 9.502816536118147e-05, "loss": 0.0378, "step": 55800 }, { "epoch": 35.5704270235819, "grad_norm": 0.384902685880661, "learning_rate": 9.502576907267218e-05, "loss": 0.0293, "step": 55810 }, { "epoch": 35.5768005098789, "grad_norm": 0.46506819128990173, "learning_rate": 9.502337223705729e-05, "loss": 0.0308, "step": 55820 }, { "epoch": 35.58317399617591, "grad_norm": 0.5410088896751404, "learning_rate": 9.502097485436591e-05, "loss": 0.0236, "step": 55830 }, { "epoch": 35.58954748247291, "grad_norm": 0.3385242819786072, "learning_rate": 9.501857692462719e-05, "loss": 0.0343, "step": 55840 }, { "epoch": 35.59592096876992, "grad_norm": 0.46601220965385437, "learning_rate": 9.501617844787025e-05, "loss": 0.0268, "step": 55850 }, { "epoch": 35.602294455066925, "grad_norm": 0.804152250289917, "learning_rate": 9.501377942412425e-05, "loss": 0.0273, "step": 55860 }, { "epoch": 35.608667941363926, "grad_norm": 0.40002262592315674, "learning_rate": 9.501137985341832e-05, "loss": 0.0167, "step": 55870 }, { "epoch": 35.615041427660934, "grad_norm": 0.5190397500991821, "learning_rate": 9.500897973578162e-05, "loss": 0.0314, "step": 55880 }, { "epoch": 35.621414913957935, "grad_norm": 0.2816859185695648, "learning_rate": 9.500657907124334e-05, "loss": 0.0268, "step": 55890 }, { "epoch": 35.62778840025494, "grad_norm": 0.33525821566581726, "learning_rate": 9.500417785983262e-05, "loss": 0.0267, "step": 55900 }, { "epoch": 35.634161886551944, "grad_norm": 0.5705966353416443, "learning_rate": 9.500177610157865e-05, "loss": 0.0236, "step": 55910 }, { "epoch": 35.64053537284895, "grad_norm": 0.7921620607376099, "learning_rate": 9.499937379651062e-05, "loss": 0.0314, "step": 55920 }, { "epoch": 35.64690885914595, "grad_norm": 0.3798553943634033, "learning_rate": 9.499697094465769e-05, "loss": 0.0266, "step": 55930 }, { "epoch": 35.65328234544296, "grad_norm": 0.580742359161377, "learning_rate": 9.499456754604909e-05, "loss": 0.0261, "step": 55940 }, { "epoch": 35.65965583173996, "grad_norm": 0.38662663102149963, "learning_rate": 9.499216360071402e-05, "loss": 0.0338, "step": 55950 }, { "epoch": 35.66602931803697, "grad_norm": 0.3803071677684784, "learning_rate": 9.498975910868168e-05, "loss": 0.0221, "step": 55960 }, { "epoch": 35.67240280433397, "grad_norm": 0.4569070041179657, "learning_rate": 9.498735406998128e-05, "loss": 0.0245, "step": 55970 }, { "epoch": 35.67877629063098, "grad_norm": 0.2580690085887909, "learning_rate": 9.498494848464206e-05, "loss": 0.0173, "step": 55980 }, { "epoch": 35.68514977692798, "grad_norm": 0.5265932083129883, "learning_rate": 9.498254235269325e-05, "loss": 0.0279, "step": 55990 }, { "epoch": 35.69152326322499, "grad_norm": 0.2920020520687103, "learning_rate": 9.498013567416407e-05, "loss": 0.03, "step": 56000 }, { "epoch": 35.69789674952199, "grad_norm": 0.3040582239627838, "learning_rate": 9.497772844908376e-05, "loss": 0.0193, "step": 56010 }, { "epoch": 35.704270235818996, "grad_norm": 0.4648805558681488, "learning_rate": 9.497532067748162e-05, "loss": 0.0209, "step": 56020 }, { "epoch": 35.710643722116, "grad_norm": 0.5383450388908386, "learning_rate": 9.497291235938685e-05, "loss": 0.0336, "step": 56030 }, { "epoch": 35.717017208413004, "grad_norm": 0.5212481617927551, "learning_rate": 9.497050349482872e-05, "loss": 0.0258, "step": 56040 }, { "epoch": 35.723390694710005, "grad_norm": 0.5252798199653625, "learning_rate": 9.496809408383654e-05, "loss": 0.032, "step": 56050 }, { "epoch": 35.72976418100701, "grad_norm": 0.2740311920642853, "learning_rate": 9.496568412643954e-05, "loss": 0.0213, "step": 56060 }, { "epoch": 35.736137667304014, "grad_norm": 0.718986988067627, "learning_rate": 9.496327362266702e-05, "loss": 0.0344, "step": 56070 }, { "epoch": 35.74251115360102, "grad_norm": 0.5655689239501953, "learning_rate": 9.496086257254829e-05, "loss": 0.023, "step": 56080 }, { "epoch": 35.74888463989802, "grad_norm": 0.34078076481819153, "learning_rate": 9.495845097611262e-05, "loss": 0.0358, "step": 56090 }, { "epoch": 35.75525812619503, "grad_norm": 0.6018723249435425, "learning_rate": 9.495603883338932e-05, "loss": 0.0304, "step": 56100 }, { "epoch": 35.76163161249203, "grad_norm": 0.6271097660064697, "learning_rate": 9.495362614440771e-05, "loss": 0.0243, "step": 56110 }, { "epoch": 35.76800509878904, "grad_norm": 0.5130691528320312, "learning_rate": 9.49512129091971e-05, "loss": 0.0262, "step": 56120 }, { "epoch": 35.77437858508604, "grad_norm": 0.5798001289367676, "learning_rate": 9.49487991277868e-05, "loss": 0.0295, "step": 56130 }, { "epoch": 35.78075207138305, "grad_norm": 0.4628540575504303, "learning_rate": 9.494638480020617e-05, "loss": 0.0202, "step": 56140 }, { "epoch": 35.78712555768005, "grad_norm": 0.2786974012851715, "learning_rate": 9.494396992648451e-05, "loss": 0.0324, "step": 56150 }, { "epoch": 35.79349904397706, "grad_norm": 0.8057183027267456, "learning_rate": 9.494155450665119e-05, "loss": 0.0324, "step": 56160 }, { "epoch": 35.79987253027406, "grad_norm": 0.6558979749679565, "learning_rate": 9.493913854073554e-05, "loss": 0.0263, "step": 56170 }, { "epoch": 35.806246016571066, "grad_norm": 0.44564953446388245, "learning_rate": 9.493672202876696e-05, "loss": 0.0217, "step": 56180 }, { "epoch": 35.81261950286807, "grad_norm": 0.37708356976509094, "learning_rate": 9.493430497077476e-05, "loss": 0.0222, "step": 56190 }, { "epoch": 35.818992989165075, "grad_norm": 0.461723268032074, "learning_rate": 9.493188736678831e-05, "loss": 0.0257, "step": 56200 }, { "epoch": 35.825366475462076, "grad_norm": 0.28750985860824585, "learning_rate": 9.492946921683703e-05, "loss": 0.0257, "step": 56210 }, { "epoch": 35.831739961759084, "grad_norm": 0.2941114008426666, "learning_rate": 9.492705052095028e-05, "loss": 0.0272, "step": 56220 }, { "epoch": 35.838113448056085, "grad_norm": 0.5967600345611572, "learning_rate": 9.492463127915742e-05, "loss": 0.0326, "step": 56230 }, { "epoch": 35.84448693435309, "grad_norm": 0.3645625710487366, "learning_rate": 9.49222114914879e-05, "loss": 0.0237, "step": 56240 }, { "epoch": 35.85086042065009, "grad_norm": 0.7042148113250732, "learning_rate": 9.491979115797109e-05, "loss": 0.03, "step": 56250 }, { "epoch": 35.8572339069471, "grad_norm": 0.33184170722961426, "learning_rate": 9.49173702786364e-05, "loss": 0.0271, "step": 56260 }, { "epoch": 35.8636073932441, "grad_norm": 0.4953864812850952, "learning_rate": 9.491494885351328e-05, "loss": 0.0255, "step": 56270 }, { "epoch": 35.86998087954111, "grad_norm": 0.40998318791389465, "learning_rate": 9.49125268826311e-05, "loss": 0.0429, "step": 56280 }, { "epoch": 35.87635436583811, "grad_norm": 0.5258964896202087, "learning_rate": 9.491010436601931e-05, "loss": 0.0268, "step": 56290 }, { "epoch": 35.88272785213512, "grad_norm": 0.47657454013824463, "learning_rate": 9.490768130370736e-05, "loss": 0.0398, "step": 56300 }, { "epoch": 35.88910133843212, "grad_norm": 0.5450493693351746, "learning_rate": 9.490525769572468e-05, "loss": 0.0208, "step": 56310 }, { "epoch": 35.89547482472913, "grad_norm": 0.4829743802547455, "learning_rate": 9.490283354210072e-05, "loss": 0.0261, "step": 56320 }, { "epoch": 35.90184831102613, "grad_norm": 0.7247356176376343, "learning_rate": 9.490040884286493e-05, "loss": 0.0234, "step": 56330 }, { "epoch": 35.90822179732314, "grad_norm": 0.3666912317276001, "learning_rate": 9.48979835980468e-05, "loss": 0.0202, "step": 56340 }, { "epoch": 35.91459528362014, "grad_norm": 0.3548961579799652, "learning_rate": 9.489555780767576e-05, "loss": 0.0198, "step": 56350 }, { "epoch": 35.920968769917145, "grad_norm": 0.6006926894187927, "learning_rate": 9.489313147178131e-05, "loss": 0.03, "step": 56360 }, { "epoch": 35.927342256214146, "grad_norm": 0.5787987112998962, "learning_rate": 9.489070459039293e-05, "loss": 0.0269, "step": 56370 }, { "epoch": 35.933715742511154, "grad_norm": 0.5361904501914978, "learning_rate": 9.48882771635401e-05, "loss": 0.0237, "step": 56380 }, { "epoch": 35.940089228808155, "grad_norm": 0.4886249303817749, "learning_rate": 9.488584919125232e-05, "loss": 0.0254, "step": 56390 }, { "epoch": 35.94646271510516, "grad_norm": 0.530318558216095, "learning_rate": 9.48834206735591e-05, "loss": 0.023, "step": 56400 }, { "epoch": 35.952836201402164, "grad_norm": 0.3418777883052826, "learning_rate": 9.488099161048994e-05, "loss": 0.0315, "step": 56410 }, { "epoch": 35.95920968769917, "grad_norm": 0.3609558939933777, "learning_rate": 9.487856200207436e-05, "loss": 0.0345, "step": 56420 }, { "epoch": 35.96558317399617, "grad_norm": 0.424207478761673, "learning_rate": 9.487613184834187e-05, "loss": 0.0256, "step": 56430 }, { "epoch": 35.97195666029318, "grad_norm": 0.26957613229751587, "learning_rate": 9.487370114932202e-05, "loss": 0.0173, "step": 56440 }, { "epoch": 35.97833014659018, "grad_norm": 0.2965833842754364, "learning_rate": 9.487126990504431e-05, "loss": 0.0249, "step": 56450 }, { "epoch": 35.98470363288719, "grad_norm": 0.32169368863105774, "learning_rate": 9.486883811553833e-05, "loss": 0.0302, "step": 56460 }, { "epoch": 35.99107711918419, "grad_norm": 0.7866684794425964, "learning_rate": 9.486640578083359e-05, "loss": 0.0257, "step": 56470 }, { "epoch": 35.9974506054812, "grad_norm": 0.5600200891494751, "learning_rate": 9.486397290095968e-05, "loss": 0.0239, "step": 56480 }, { "epoch": 36.003824091778206, "grad_norm": 0.2826383411884308, "learning_rate": 9.486153947594612e-05, "loss": 0.0227, "step": 56490 }, { "epoch": 36.01019757807521, "grad_norm": 0.49822118878364563, "learning_rate": 9.48591055058225e-05, "loss": 0.0229, "step": 56500 }, { "epoch": 36.016571064372215, "grad_norm": 0.599690318107605, "learning_rate": 9.485667099061839e-05, "loss": 0.0368, "step": 56510 }, { "epoch": 36.022944550669216, "grad_norm": 0.5329281091690063, "learning_rate": 9.485423593036339e-05, "loss": 0.0273, "step": 56520 }, { "epoch": 36.029318036966224, "grad_norm": 0.9028069376945496, "learning_rate": 9.485180032508708e-05, "loss": 0.025, "step": 56530 }, { "epoch": 36.035691523263225, "grad_norm": 0.3643564283847809, "learning_rate": 9.484936417481902e-05, "loss": 0.0332, "step": 56540 }, { "epoch": 36.04206500956023, "grad_norm": 0.5311803817749023, "learning_rate": 9.484692747958885e-05, "loss": 0.026, "step": 56550 }, { "epoch": 36.04843849585723, "grad_norm": 0.9790709018707275, "learning_rate": 9.484449023942619e-05, "loss": 0.0483, "step": 56560 }, { "epoch": 36.05481198215424, "grad_norm": 0.4993636906147003, "learning_rate": 9.484205245436062e-05, "loss": 0.0312, "step": 56570 }, { "epoch": 36.06118546845124, "grad_norm": 0.6763119697570801, "learning_rate": 9.483961412442177e-05, "loss": 0.0218, "step": 56580 }, { "epoch": 36.06755895474825, "grad_norm": 0.6167675256729126, "learning_rate": 9.483717524963926e-05, "loss": 0.0295, "step": 56590 }, { "epoch": 36.07393244104525, "grad_norm": 0.614103376865387, "learning_rate": 9.483473583004276e-05, "loss": 0.017, "step": 56600 }, { "epoch": 36.08030592734226, "grad_norm": 0.41841888427734375, "learning_rate": 9.483229586566189e-05, "loss": 0.0365, "step": 56610 }, { "epoch": 36.08667941363926, "grad_norm": 0.4662361145019531, "learning_rate": 9.482985535652629e-05, "loss": 0.0342, "step": 56620 }, { "epoch": 36.09305289993627, "grad_norm": 0.48006853461265564, "learning_rate": 9.482741430266561e-05, "loss": 0.0246, "step": 56630 }, { "epoch": 36.09942638623327, "grad_norm": 0.4246199429035187, "learning_rate": 9.482497270410953e-05, "loss": 0.0198, "step": 56640 }, { "epoch": 36.10579987253028, "grad_norm": 0.4166896939277649, "learning_rate": 9.482253056088773e-05, "loss": 0.0372, "step": 56650 }, { "epoch": 36.11217335882728, "grad_norm": 0.7195228338241577, "learning_rate": 9.482008787302983e-05, "loss": 0.0218, "step": 56660 }, { "epoch": 36.118546845124285, "grad_norm": 0.4906434118747711, "learning_rate": 9.481764464056556e-05, "loss": 0.0374, "step": 56670 }, { "epoch": 36.124920331421286, "grad_norm": 0.448713481426239, "learning_rate": 9.48152008635246e-05, "loss": 0.0178, "step": 56680 }, { "epoch": 36.131293817718294, "grad_norm": 0.29854315519332886, "learning_rate": 9.481275654193661e-05, "loss": 0.0338, "step": 56690 }, { "epoch": 36.137667304015295, "grad_norm": 0.4892030954360962, "learning_rate": 9.481031167583135e-05, "loss": 0.0234, "step": 56700 }, { "epoch": 36.1440407903123, "grad_norm": 0.7320384979248047, "learning_rate": 9.480786626523849e-05, "loss": 0.0369, "step": 56710 }, { "epoch": 36.150414276609304, "grad_norm": 0.4274139106273651, "learning_rate": 9.480542031018775e-05, "loss": 0.0186, "step": 56720 }, { "epoch": 36.15678776290631, "grad_norm": 0.34931451082229614, "learning_rate": 9.480297381070883e-05, "loss": 0.0262, "step": 56730 }, { "epoch": 36.16316124920331, "grad_norm": 0.4003312289714813, "learning_rate": 9.48005267668315e-05, "loss": 0.0196, "step": 56740 }, { "epoch": 36.16953473550032, "grad_norm": 0.5395488739013672, "learning_rate": 9.479807917858545e-05, "loss": 0.0301, "step": 56750 }, { "epoch": 36.17590822179732, "grad_norm": 0.457661509513855, "learning_rate": 9.479563104600046e-05, "loss": 0.0394, "step": 56760 }, { "epoch": 36.18228170809433, "grad_norm": 0.2268311232328415, "learning_rate": 9.479318236910625e-05, "loss": 0.0164, "step": 56770 }, { "epoch": 36.18865519439133, "grad_norm": 0.4309782385826111, "learning_rate": 9.479073314793259e-05, "loss": 0.0263, "step": 56780 }, { "epoch": 36.19502868068834, "grad_norm": 0.7037017941474915, "learning_rate": 9.478828338250922e-05, "loss": 0.0285, "step": 56790 }, { "epoch": 36.20140216698534, "grad_norm": 0.6677271127700806, "learning_rate": 9.478583307286592e-05, "loss": 0.0382, "step": 56800 }, { "epoch": 36.20777565328235, "grad_norm": 0.332619309425354, "learning_rate": 9.478338221903248e-05, "loss": 0.0167, "step": 56810 }, { "epoch": 36.21414913957935, "grad_norm": 0.4014318287372589, "learning_rate": 9.478093082103865e-05, "loss": 0.037, "step": 56820 }, { "epoch": 36.220522625876356, "grad_norm": 0.663598358631134, "learning_rate": 9.477847887891423e-05, "loss": 0.0286, "step": 56830 }, { "epoch": 36.22689611217336, "grad_norm": 0.7025123834609985, "learning_rate": 9.477602639268902e-05, "loss": 0.0284, "step": 56840 }, { "epoch": 36.233269598470365, "grad_norm": 0.5149930715560913, "learning_rate": 9.477357336239281e-05, "loss": 0.0205, "step": 56850 }, { "epoch": 36.239643084767366, "grad_norm": 0.47478237748146057, "learning_rate": 9.477111978805542e-05, "loss": 0.0344, "step": 56860 }, { "epoch": 36.246016571064374, "grad_norm": 0.32208940386772156, "learning_rate": 9.476866566970664e-05, "loss": 0.0451, "step": 56870 }, { "epoch": 36.252390057361374, "grad_norm": 0.3925299048423767, "learning_rate": 9.47662110073763e-05, "loss": 0.0354, "step": 56880 }, { "epoch": 36.25876354365838, "grad_norm": 0.2112465500831604, "learning_rate": 9.476375580109425e-05, "loss": 0.0187, "step": 56890 }, { "epoch": 36.26513702995538, "grad_norm": 0.5465281009674072, "learning_rate": 9.476130005089029e-05, "loss": 0.0351, "step": 56900 }, { "epoch": 36.27151051625239, "grad_norm": 0.46439221501350403, "learning_rate": 9.475884375679427e-05, "loss": 0.0317, "step": 56910 }, { "epoch": 36.27788400254939, "grad_norm": 0.4916335642337799, "learning_rate": 9.475638691883604e-05, "loss": 0.0352, "step": 56920 }, { "epoch": 36.2842574888464, "grad_norm": 0.4705188572406769, "learning_rate": 9.475392953704546e-05, "loss": 0.0265, "step": 56930 }, { "epoch": 36.2906309751434, "grad_norm": 0.5308962464332581, "learning_rate": 9.475147161145237e-05, "loss": 0.0277, "step": 56940 }, { "epoch": 36.29700446144041, "grad_norm": 0.4140111207962036, "learning_rate": 9.474901314208666e-05, "loss": 0.021, "step": 56950 }, { "epoch": 36.30337794773741, "grad_norm": 0.257232129573822, "learning_rate": 9.474655412897818e-05, "loss": 0.0204, "step": 56960 }, { "epoch": 36.30975143403442, "grad_norm": 0.4791267216205597, "learning_rate": 9.474409457215683e-05, "loss": 0.0245, "step": 56970 }, { "epoch": 36.31612492033142, "grad_norm": 0.47617340087890625, "learning_rate": 9.474163447165247e-05, "loss": 0.0294, "step": 56980 }, { "epoch": 36.322498406628426, "grad_norm": 0.4410161077976227, "learning_rate": 9.473917382749502e-05, "loss": 0.0263, "step": 56990 }, { "epoch": 36.32887189292543, "grad_norm": 0.5325472950935364, "learning_rate": 9.473671263971438e-05, "loss": 0.0277, "step": 57000 }, { "epoch": 36.335245379222435, "grad_norm": 0.4779888391494751, "learning_rate": 9.473425090834041e-05, "loss": 0.0218, "step": 57010 }, { "epoch": 36.341618865519436, "grad_norm": 0.8897963166236877, "learning_rate": 9.473178863340309e-05, "loss": 0.0502, "step": 57020 }, { "epoch": 36.347992351816444, "grad_norm": 0.45764678716659546, "learning_rate": 9.472932581493228e-05, "loss": 0.0277, "step": 57030 }, { "epoch": 36.354365838113445, "grad_norm": 0.5744030475616455, "learning_rate": 9.472686245295793e-05, "loss": 0.0233, "step": 57040 }, { "epoch": 36.36073932441045, "grad_norm": 0.2924349904060364, "learning_rate": 9.472439854750999e-05, "loss": 0.0251, "step": 57050 }, { "epoch": 36.367112810707454, "grad_norm": 0.43266424536705017, "learning_rate": 9.472193409861836e-05, "loss": 0.0391, "step": 57060 }, { "epoch": 36.37348629700446, "grad_norm": 0.6007348895072937, "learning_rate": 9.471946910631302e-05, "loss": 0.0259, "step": 57070 }, { "epoch": 36.37985978330146, "grad_norm": 0.4023219645023346, "learning_rate": 9.47170035706239e-05, "loss": 0.0148, "step": 57080 }, { "epoch": 36.38623326959847, "grad_norm": 0.4193379580974579, "learning_rate": 9.471453749158096e-05, "loss": 0.0303, "step": 57090 }, { "epoch": 36.39260675589547, "grad_norm": 0.5523746013641357, "learning_rate": 9.47120708692142e-05, "loss": 0.0215, "step": 57100 }, { "epoch": 36.39898024219248, "grad_norm": 0.5907568335533142, "learning_rate": 9.470960370355356e-05, "loss": 0.0269, "step": 57110 }, { "epoch": 36.40535372848949, "grad_norm": 0.43363630771636963, "learning_rate": 9.4707135994629e-05, "loss": 0.0294, "step": 57120 }, { "epoch": 36.41172721478649, "grad_norm": 0.6485674977302551, "learning_rate": 9.470466774247054e-05, "loss": 0.0273, "step": 57130 }, { "epoch": 36.418100701083496, "grad_norm": 0.4767267107963562, "learning_rate": 9.470219894710814e-05, "loss": 0.0184, "step": 57140 }, { "epoch": 36.4244741873805, "grad_norm": 0.5569348931312561, "learning_rate": 9.469972960857182e-05, "loss": 0.0222, "step": 57150 }, { "epoch": 36.430847673677505, "grad_norm": 0.35149672627449036, "learning_rate": 9.469725972689161e-05, "loss": 0.0355, "step": 57160 }, { "epoch": 36.437221159974506, "grad_norm": 0.3694823384284973, "learning_rate": 9.469478930209748e-05, "loss": 0.0258, "step": 57170 }, { "epoch": 36.443594646271514, "grad_norm": 0.3965859115123749, "learning_rate": 9.469231833421945e-05, "loss": 0.0226, "step": 57180 }, { "epoch": 36.449968132568515, "grad_norm": 0.5353180170059204, "learning_rate": 9.468984682328756e-05, "loss": 0.0275, "step": 57190 }, { "epoch": 36.45634161886552, "grad_norm": 0.5391090512275696, "learning_rate": 9.468737476933184e-05, "loss": 0.0326, "step": 57200 }, { "epoch": 36.46271510516252, "grad_norm": 0.36551183462142944, "learning_rate": 9.468490217238232e-05, "loss": 0.0306, "step": 57210 }, { "epoch": 36.46908859145953, "grad_norm": 0.3970765769481659, "learning_rate": 9.468242903246907e-05, "loss": 0.0218, "step": 57220 }, { "epoch": 36.47546207775653, "grad_norm": 0.3243199586868286, "learning_rate": 9.46799553496221e-05, "loss": 0.0205, "step": 57230 }, { "epoch": 36.48183556405354, "grad_norm": 0.44766804575920105, "learning_rate": 9.46774811238715e-05, "loss": 0.0241, "step": 57240 }, { "epoch": 36.48820905035054, "grad_norm": 0.5407687425613403, "learning_rate": 9.467500635524731e-05, "loss": 0.0214, "step": 57250 }, { "epoch": 36.49458253664755, "grad_norm": 0.5734948515892029, "learning_rate": 9.46725310437796e-05, "loss": 0.0256, "step": 57260 }, { "epoch": 36.50095602294455, "grad_norm": 0.5349106192588806, "learning_rate": 9.467005518949848e-05, "loss": 0.0358, "step": 57270 }, { "epoch": 36.50732950924156, "grad_norm": 0.646597146987915, "learning_rate": 9.466757879243402e-05, "loss": 0.0354, "step": 57280 }, { "epoch": 36.51370299553856, "grad_norm": 0.3602616488933563, "learning_rate": 9.46651018526163e-05, "loss": 0.0351, "step": 57290 }, { "epoch": 36.52007648183557, "grad_norm": 0.4239743947982788, "learning_rate": 9.46626243700754e-05, "loss": 0.029, "step": 57300 }, { "epoch": 36.52644996813257, "grad_norm": 0.2851369380950928, "learning_rate": 9.466014634484147e-05, "loss": 0.0335, "step": 57310 }, { "epoch": 36.532823454429575, "grad_norm": 0.5277607440948486, "learning_rate": 9.465766777694458e-05, "loss": 0.0334, "step": 57320 }, { "epoch": 36.539196940726576, "grad_norm": 0.6391661763191223, "learning_rate": 9.465518866641487e-05, "loss": 0.0339, "step": 57330 }, { "epoch": 36.545570427023584, "grad_norm": 0.6888791918754578, "learning_rate": 9.465270901328247e-05, "loss": 0.0287, "step": 57340 }, { "epoch": 36.551943913320585, "grad_norm": 0.3791259527206421, "learning_rate": 9.465022881757749e-05, "loss": 0.0259, "step": 57350 }, { "epoch": 36.55831739961759, "grad_norm": 0.5581915974617004, "learning_rate": 9.464774807933005e-05, "loss": 0.027, "step": 57360 }, { "epoch": 36.564690885914594, "grad_norm": 0.6255170106887817, "learning_rate": 9.464526679857034e-05, "loss": 0.0469, "step": 57370 }, { "epoch": 36.5710643722116, "grad_norm": 0.5116237998008728, "learning_rate": 9.464278497532848e-05, "loss": 0.0294, "step": 57380 }, { "epoch": 36.5774378585086, "grad_norm": 0.5179383158683777, "learning_rate": 9.464030260963462e-05, "loss": 0.0261, "step": 57390 }, { "epoch": 36.58381134480561, "grad_norm": 0.32617050409317017, "learning_rate": 9.463781970151895e-05, "loss": 0.0283, "step": 57400 }, { "epoch": 36.59018483110261, "grad_norm": 0.30168455839157104, "learning_rate": 9.463533625101162e-05, "loss": 0.0248, "step": 57410 }, { "epoch": 36.59655831739962, "grad_norm": 0.3762986958026886, "learning_rate": 9.463285225814283e-05, "loss": 0.0201, "step": 57420 }, { "epoch": 36.60293180369662, "grad_norm": 0.42218831181526184, "learning_rate": 9.463036772294272e-05, "loss": 0.0302, "step": 57430 }, { "epoch": 36.60930528999363, "grad_norm": 0.5094795227050781, "learning_rate": 9.462788264544152e-05, "loss": 0.0262, "step": 57440 }, { "epoch": 36.61567877629063, "grad_norm": 0.8828285932540894, "learning_rate": 9.462539702566939e-05, "loss": 0.0315, "step": 57450 }, { "epoch": 36.62205226258764, "grad_norm": 0.26420992612838745, "learning_rate": 9.462291086365657e-05, "loss": 0.0425, "step": 57460 }, { "epoch": 36.62842574888464, "grad_norm": 0.5642569065093994, "learning_rate": 9.462042415943324e-05, "loss": 0.022, "step": 57470 }, { "epoch": 36.634799235181646, "grad_norm": 0.5613830089569092, "learning_rate": 9.461793691302965e-05, "loss": 0.0286, "step": 57480 }, { "epoch": 36.64117272147865, "grad_norm": 0.38027775287628174, "learning_rate": 9.461544912447597e-05, "loss": 0.0301, "step": 57490 }, { "epoch": 36.647546207775655, "grad_norm": 0.36848384141921997, "learning_rate": 9.461296079380249e-05, "loss": 0.0292, "step": 57500 }, { "epoch": 36.653919694072655, "grad_norm": 0.3908831775188446, "learning_rate": 9.461047192103939e-05, "loss": 0.0274, "step": 57510 }, { "epoch": 36.66029318036966, "grad_norm": 0.8432488441467285, "learning_rate": 9.460798250621694e-05, "loss": 0.0366, "step": 57520 }, { "epoch": 36.666666666666664, "grad_norm": 0.8290928602218628, "learning_rate": 9.46054925493654e-05, "loss": 0.0282, "step": 57530 }, { "epoch": 36.67304015296367, "grad_norm": 0.5668666362762451, "learning_rate": 9.4603002050515e-05, "loss": 0.0316, "step": 57540 }, { "epoch": 36.67941363926067, "grad_norm": 0.5342413187026978, "learning_rate": 9.460051100969602e-05, "loss": 0.0345, "step": 57550 }, { "epoch": 36.68578712555768, "grad_norm": 0.3098469078540802, "learning_rate": 9.459801942693871e-05, "loss": 0.0329, "step": 57560 }, { "epoch": 36.69216061185468, "grad_norm": 0.3256721794605255, "learning_rate": 9.459552730227336e-05, "loss": 0.0156, "step": 57570 }, { "epoch": 36.69853409815169, "grad_norm": 0.6516679525375366, "learning_rate": 9.459303463573026e-05, "loss": 0.0268, "step": 57580 }, { "epoch": 36.70490758444869, "grad_norm": 0.33256176114082336, "learning_rate": 9.459054142733968e-05, "loss": 0.022, "step": 57590 }, { "epoch": 36.7112810707457, "grad_norm": 0.34782370924949646, "learning_rate": 9.458804767713192e-05, "loss": 0.0282, "step": 57600 }, { "epoch": 36.7176545570427, "grad_norm": 0.32653582096099854, "learning_rate": 9.458555338513728e-05, "loss": 0.0356, "step": 57610 }, { "epoch": 36.72402804333971, "grad_norm": 0.6436353325843811, "learning_rate": 9.458305855138607e-05, "loss": 0.0221, "step": 57620 }, { "epoch": 36.73040152963671, "grad_norm": 0.5634167790412903, "learning_rate": 9.458056317590862e-05, "loss": 0.0333, "step": 57630 }, { "epoch": 36.736775015933716, "grad_norm": 0.4234979450702667, "learning_rate": 9.457806725873522e-05, "loss": 0.0418, "step": 57640 }, { "epoch": 36.74314850223072, "grad_norm": 0.48141583800315857, "learning_rate": 9.457557079989622e-05, "loss": 0.0262, "step": 57650 }, { "epoch": 36.749521988527725, "grad_norm": 0.40675675868988037, "learning_rate": 9.457307379942195e-05, "loss": 0.0298, "step": 57660 }, { "epoch": 36.755895474824726, "grad_norm": 0.6099407076835632, "learning_rate": 9.457057625734273e-05, "loss": 0.0338, "step": 57670 }, { "epoch": 36.762268961121734, "grad_norm": 0.5035131573677063, "learning_rate": 9.456807817368895e-05, "loss": 0.0253, "step": 57680 }, { "epoch": 36.768642447418735, "grad_norm": 0.5587304830551147, "learning_rate": 9.456557954849094e-05, "loss": 0.0233, "step": 57690 }, { "epoch": 36.77501593371574, "grad_norm": 0.2693192958831787, "learning_rate": 9.456308038177906e-05, "loss": 0.0208, "step": 57700 }, { "epoch": 36.781389420012744, "grad_norm": 0.4463195204734802, "learning_rate": 9.456058067358367e-05, "loss": 0.0228, "step": 57710 }, { "epoch": 36.78776290630975, "grad_norm": 0.3880036175251007, "learning_rate": 9.455808042393516e-05, "loss": 0.0223, "step": 57720 }, { "epoch": 36.79413639260676, "grad_norm": 0.4854073226451874, "learning_rate": 9.45555796328639e-05, "loss": 0.0204, "step": 57730 }, { "epoch": 36.80050987890376, "grad_norm": 0.46327298879623413, "learning_rate": 9.455307830040028e-05, "loss": 0.0255, "step": 57740 }, { "epoch": 36.80688336520077, "grad_norm": 0.43582096695899963, "learning_rate": 9.455057642657469e-05, "loss": 0.0221, "step": 57750 }, { "epoch": 36.81325685149777, "grad_norm": 0.3683633804321289, "learning_rate": 9.454807401141754e-05, "loss": 0.03, "step": 57760 }, { "epoch": 36.81963033779478, "grad_norm": 0.38868263363838196, "learning_rate": 9.454557105495923e-05, "loss": 0.0179, "step": 57770 }, { "epoch": 36.82600382409178, "grad_norm": 0.47523003816604614, "learning_rate": 9.454306755723017e-05, "loss": 0.0314, "step": 57780 }, { "epoch": 36.832377310388786, "grad_norm": 0.6168904900550842, "learning_rate": 9.45405635182608e-05, "loss": 0.0303, "step": 57790 }, { "epoch": 36.83875079668579, "grad_norm": 0.4319334328174591, "learning_rate": 9.45380589380815e-05, "loss": 0.0189, "step": 57800 }, { "epoch": 36.845124282982795, "grad_norm": 0.7553130388259888, "learning_rate": 9.453555381672275e-05, "loss": 0.0532, "step": 57810 }, { "epoch": 36.851497769279796, "grad_norm": 0.3500904440879822, "learning_rate": 9.453304815421497e-05, "loss": 0.0298, "step": 57820 }, { "epoch": 36.8578712555768, "grad_norm": 0.40126967430114746, "learning_rate": 9.453054195058862e-05, "loss": 0.0204, "step": 57830 }, { "epoch": 36.864244741873804, "grad_norm": 0.5441057682037354, "learning_rate": 9.452803520587412e-05, "loss": 0.0256, "step": 57840 }, { "epoch": 36.87061822817081, "grad_norm": 0.4571659564971924, "learning_rate": 9.452552792010197e-05, "loss": 0.0304, "step": 57850 }, { "epoch": 36.87699171446781, "grad_norm": 0.442874014377594, "learning_rate": 9.452302009330261e-05, "loss": 0.0252, "step": 57860 }, { "epoch": 36.88336520076482, "grad_norm": 0.5466029047966003, "learning_rate": 9.452051172550652e-05, "loss": 0.0381, "step": 57870 }, { "epoch": 36.88973868706182, "grad_norm": 0.8083130717277527, "learning_rate": 9.451800281674417e-05, "loss": 0.0271, "step": 57880 }, { "epoch": 36.89611217335883, "grad_norm": 0.3117665648460388, "learning_rate": 9.451549336704606e-05, "loss": 0.0347, "step": 57890 }, { "epoch": 36.90248565965583, "grad_norm": 0.28044503927230835, "learning_rate": 9.451298337644267e-05, "loss": 0.035, "step": 57900 }, { "epoch": 36.90885914595284, "grad_norm": 0.5083986520767212, "learning_rate": 9.451047284496452e-05, "loss": 0.0298, "step": 57910 }, { "epoch": 36.91523263224984, "grad_norm": 0.5868697762489319, "learning_rate": 9.450796177264209e-05, "loss": 0.0222, "step": 57920 }, { "epoch": 36.92160611854685, "grad_norm": 0.8156291246414185, "learning_rate": 9.450545015950591e-05, "loss": 0.0455, "step": 57930 }, { "epoch": 36.92797960484385, "grad_norm": 0.8249725699424744, "learning_rate": 9.450293800558648e-05, "loss": 0.0342, "step": 57940 }, { "epoch": 36.934353091140856, "grad_norm": 0.3781333267688751, "learning_rate": 9.450042531091433e-05, "loss": 0.0297, "step": 57950 }, { "epoch": 36.94072657743786, "grad_norm": 0.40890398621559143, "learning_rate": 9.449791207552001e-05, "loss": 0.0294, "step": 57960 }, { "epoch": 36.947100063734865, "grad_norm": 0.33422401547431946, "learning_rate": 9.449539829943404e-05, "loss": 0.0239, "step": 57970 }, { "epoch": 36.953473550031866, "grad_norm": 0.4204844534397125, "learning_rate": 9.449288398268697e-05, "loss": 0.0289, "step": 57980 }, { "epoch": 36.959847036328874, "grad_norm": 0.5481045246124268, "learning_rate": 9.449036912530935e-05, "loss": 0.0292, "step": 57990 }, { "epoch": 36.966220522625875, "grad_norm": 0.2501024007797241, "learning_rate": 9.448785372733174e-05, "loss": 0.0182, "step": 58000 }, { "epoch": 36.97259400892288, "grad_norm": 0.43681061267852783, "learning_rate": 9.448533778878471e-05, "loss": 0.0218, "step": 58010 }, { "epoch": 36.978967495219884, "grad_norm": 0.6796426177024841, "learning_rate": 9.448282130969882e-05, "loss": 0.0228, "step": 58020 }, { "epoch": 36.98534098151689, "grad_norm": 0.40712788701057434, "learning_rate": 9.448030429010466e-05, "loss": 0.0411, "step": 58030 }, { "epoch": 36.99171446781389, "grad_norm": 0.6452886462211609, "learning_rate": 9.44777867300328e-05, "loss": 0.0242, "step": 58040 }, { "epoch": 36.9980879541109, "grad_norm": 0.4238446354866028, "learning_rate": 9.447526862951384e-05, "loss": 0.0361, "step": 58050 }, { "epoch": 37.0044614404079, "grad_norm": 0.5474826693534851, "learning_rate": 9.447274998857836e-05, "loss": 0.0163, "step": 58060 }, { "epoch": 37.01083492670491, "grad_norm": 0.4672966003417969, "learning_rate": 9.447023080725698e-05, "loss": 0.0234, "step": 58070 }, { "epoch": 37.01720841300191, "grad_norm": 0.6289032101631165, "learning_rate": 9.446771108558033e-05, "loss": 0.0282, "step": 58080 }, { "epoch": 37.02358189929892, "grad_norm": 0.8783146739006042, "learning_rate": 9.446519082357898e-05, "loss": 0.0209, "step": 58090 }, { "epoch": 37.02995538559592, "grad_norm": 0.3712984025478363, "learning_rate": 9.44626700212836e-05, "loss": 0.0352, "step": 58100 }, { "epoch": 37.03632887189293, "grad_norm": 0.48685890436172485, "learning_rate": 9.44601486787248e-05, "loss": 0.0257, "step": 58110 }, { "epoch": 37.04270235818993, "grad_norm": 0.3748556673526764, "learning_rate": 9.445762679593321e-05, "loss": 0.0211, "step": 58120 }, { "epoch": 37.049075844486936, "grad_norm": 0.3443540036678314, "learning_rate": 9.445510437293948e-05, "loss": 0.023, "step": 58130 }, { "epoch": 37.05544933078394, "grad_norm": 0.3426295816898346, "learning_rate": 9.445258140977426e-05, "loss": 0.0327, "step": 58140 }, { "epoch": 37.061822817080945, "grad_norm": 0.5828007459640503, "learning_rate": 9.445005790646819e-05, "loss": 0.0364, "step": 58150 }, { "epoch": 37.068196303377945, "grad_norm": 0.4944082796573639, "learning_rate": 9.444753386305196e-05, "loss": 0.0278, "step": 58160 }, { "epoch": 37.07456978967495, "grad_norm": 0.2439521700143814, "learning_rate": 9.444500927955624e-05, "loss": 0.0222, "step": 58170 }, { "epoch": 37.080943275971954, "grad_norm": 0.6633405089378357, "learning_rate": 9.444248415601168e-05, "loss": 0.0231, "step": 58180 }, { "epoch": 37.08731676226896, "grad_norm": 0.7136917114257812, "learning_rate": 9.443995849244899e-05, "loss": 0.0209, "step": 58190 }, { "epoch": 37.09369024856596, "grad_norm": 0.5466375350952148, "learning_rate": 9.443743228889883e-05, "loss": 0.0216, "step": 58200 }, { "epoch": 37.10006373486297, "grad_norm": 0.8081802725791931, "learning_rate": 9.443490554539191e-05, "loss": 0.0434, "step": 58210 }, { "epoch": 37.10643722115997, "grad_norm": 0.3328581750392914, "learning_rate": 9.443237826195897e-05, "loss": 0.0218, "step": 58220 }, { "epoch": 37.11281070745698, "grad_norm": 0.45257583260536194, "learning_rate": 9.442985043863064e-05, "loss": 0.031, "step": 58230 }, { "epoch": 37.11918419375398, "grad_norm": 0.5781674385070801, "learning_rate": 9.44273220754377e-05, "loss": 0.0232, "step": 58240 }, { "epoch": 37.12555768005099, "grad_norm": 0.7794846296310425, "learning_rate": 9.442479317241084e-05, "loss": 0.0332, "step": 58250 }, { "epoch": 37.13193116634799, "grad_norm": 0.43148553371429443, "learning_rate": 9.44222637295808e-05, "loss": 0.019, "step": 58260 }, { "epoch": 37.138304652645, "grad_norm": 0.45630818605422974, "learning_rate": 9.441973374697833e-05, "loss": 0.0263, "step": 58270 }, { "epoch": 37.144678138942, "grad_norm": 1.0317178964614868, "learning_rate": 9.441720322463414e-05, "loss": 0.0304, "step": 58280 }, { "epoch": 37.151051625239006, "grad_norm": 0.6345905065536499, "learning_rate": 9.4414672162579e-05, "loss": 0.0165, "step": 58290 }, { "epoch": 37.15742511153601, "grad_norm": 0.4350857138633728, "learning_rate": 9.441214056084365e-05, "loss": 0.0307, "step": 58300 }, { "epoch": 37.163798597833015, "grad_norm": 0.431850403547287, "learning_rate": 9.440960841945887e-05, "loss": 0.0392, "step": 58310 }, { "epoch": 37.170172084130016, "grad_norm": 0.6173940300941467, "learning_rate": 9.440707573845542e-05, "loss": 0.0203, "step": 58320 }, { "epoch": 37.176545570427024, "grad_norm": 0.6035770773887634, "learning_rate": 9.440454251786406e-05, "loss": 0.0261, "step": 58330 }, { "epoch": 37.182919056724025, "grad_norm": 0.46005141735076904, "learning_rate": 9.44020087577156e-05, "loss": 0.0243, "step": 58340 }, { "epoch": 37.18929254302103, "grad_norm": 1.101370930671692, "learning_rate": 9.439947445804081e-05, "loss": 0.0362, "step": 58350 }, { "epoch": 37.19566602931803, "grad_norm": 0.4578830301761627, "learning_rate": 9.439693961887046e-05, "loss": 0.0262, "step": 58360 }, { "epoch": 37.20203951561504, "grad_norm": 0.3971228003501892, "learning_rate": 9.43944042402354e-05, "loss": 0.0178, "step": 58370 }, { "epoch": 37.20841300191205, "grad_norm": 0.4319360852241516, "learning_rate": 9.43918683221664e-05, "loss": 0.0264, "step": 58380 }, { "epoch": 37.21478648820905, "grad_norm": 0.7146967053413391, "learning_rate": 9.43893318646943e-05, "loss": 0.0286, "step": 58390 }, { "epoch": 37.22115997450606, "grad_norm": 0.6462353467941284, "learning_rate": 9.43867948678499e-05, "loss": 0.03, "step": 58400 }, { "epoch": 37.22753346080306, "grad_norm": 0.876550555229187, "learning_rate": 9.438425733166404e-05, "loss": 0.0225, "step": 58410 }, { "epoch": 37.23390694710007, "grad_norm": 0.40169620513916016, "learning_rate": 9.438171925616754e-05, "loss": 0.0297, "step": 58420 }, { "epoch": 37.24028043339707, "grad_norm": 0.39487162232398987, "learning_rate": 9.437918064139126e-05, "loss": 0.0228, "step": 58430 }, { "epoch": 37.246653919694076, "grad_norm": 0.2838194966316223, "learning_rate": 9.437664148736603e-05, "loss": 0.0334, "step": 58440 }, { "epoch": 37.25302740599108, "grad_norm": 0.4718831777572632, "learning_rate": 9.437410179412269e-05, "loss": 0.0348, "step": 58450 }, { "epoch": 37.259400892288085, "grad_norm": 0.7466816902160645, "learning_rate": 9.437156156169213e-05, "loss": 0.0373, "step": 58460 }, { "epoch": 37.265774378585085, "grad_norm": 0.5572332143783569, "learning_rate": 9.436902079010521e-05, "loss": 0.0225, "step": 58470 }, { "epoch": 37.27214786488209, "grad_norm": 0.3881951570510864, "learning_rate": 9.436647947939279e-05, "loss": 0.0255, "step": 58480 }, { "epoch": 37.278521351179094, "grad_norm": 0.38517072796821594, "learning_rate": 9.436393762958575e-05, "loss": 0.0266, "step": 58490 }, { "epoch": 37.2848948374761, "grad_norm": 0.36928829550743103, "learning_rate": 9.4361395240715e-05, "loss": 0.0354, "step": 58500 }, { "epoch": 37.2912683237731, "grad_norm": 0.5534390807151794, "learning_rate": 9.435885231281139e-05, "loss": 0.0387, "step": 58510 }, { "epoch": 37.29764181007011, "grad_norm": 0.4806787073612213, "learning_rate": 9.435630884590586e-05, "loss": 0.0294, "step": 58520 }, { "epoch": 37.30401529636711, "grad_norm": 0.5615505576133728, "learning_rate": 9.435376484002928e-05, "loss": 0.031, "step": 58530 }, { "epoch": 37.31038878266412, "grad_norm": 0.5432212948799133, "learning_rate": 9.435122029521259e-05, "loss": 0.0266, "step": 58540 }, { "epoch": 37.31676226896112, "grad_norm": 0.3455148935317993, "learning_rate": 9.43486752114867e-05, "loss": 0.0337, "step": 58550 }, { "epoch": 37.32313575525813, "grad_norm": 0.9200009703636169, "learning_rate": 9.434612958888253e-05, "loss": 0.022, "step": 58560 }, { "epoch": 37.32950924155513, "grad_norm": 0.3465256094932556, "learning_rate": 9.434358342743103e-05, "loss": 0.0222, "step": 58570 }, { "epoch": 37.33588272785214, "grad_norm": 0.5053335428237915, "learning_rate": 9.434103672716311e-05, "loss": 0.0283, "step": 58580 }, { "epoch": 37.34225621414914, "grad_norm": 0.3810831308364868, "learning_rate": 9.433848948810973e-05, "loss": 0.0257, "step": 58590 }, { "epoch": 37.348629700446146, "grad_norm": 0.5449327230453491, "learning_rate": 9.433594171030184e-05, "loss": 0.0189, "step": 58600 }, { "epoch": 37.35500318674315, "grad_norm": 0.5504589676856995, "learning_rate": 9.43333933937704e-05, "loss": 0.0224, "step": 58610 }, { "epoch": 37.361376673040155, "grad_norm": 0.6452327966690063, "learning_rate": 9.433084453854638e-05, "loss": 0.0262, "step": 58620 }, { "epoch": 37.367750159337156, "grad_norm": 0.505751371383667, "learning_rate": 9.432829514466074e-05, "loss": 0.0241, "step": 58630 }, { "epoch": 37.374123645634164, "grad_norm": 0.727211594581604, "learning_rate": 9.432574521214445e-05, "loss": 0.0328, "step": 58640 }, { "epoch": 37.380497131931165, "grad_norm": 0.5360155701637268, "learning_rate": 9.432319474102852e-05, "loss": 0.0205, "step": 58650 }, { "epoch": 37.38687061822817, "grad_norm": 0.5676209926605225, "learning_rate": 9.432064373134392e-05, "loss": 0.0271, "step": 58660 }, { "epoch": 37.393244104525174, "grad_norm": 0.3800312578678131, "learning_rate": 9.431809218312166e-05, "loss": 0.023, "step": 58670 }, { "epoch": 37.39961759082218, "grad_norm": 0.4834483861923218, "learning_rate": 9.431554009639272e-05, "loss": 0.0268, "step": 58680 }, { "epoch": 37.40599107711918, "grad_norm": 0.35569557547569275, "learning_rate": 9.431298747118814e-05, "loss": 0.0273, "step": 58690 }, { "epoch": 37.41236456341619, "grad_norm": 0.2903531789779663, "learning_rate": 9.431043430753892e-05, "loss": 0.0249, "step": 58700 }, { "epoch": 37.41873804971319, "grad_norm": 0.5252476334571838, "learning_rate": 9.430788060547609e-05, "loss": 0.0279, "step": 58710 }, { "epoch": 37.4251115360102, "grad_norm": 0.6529276967048645, "learning_rate": 9.430532636503068e-05, "loss": 0.0355, "step": 58720 }, { "epoch": 37.4314850223072, "grad_norm": 0.3895089328289032, "learning_rate": 9.43027715862337e-05, "loss": 0.0277, "step": 58730 }, { "epoch": 37.43785850860421, "grad_norm": 0.36202213168144226, "learning_rate": 9.430021626911625e-05, "loss": 0.0335, "step": 58740 }, { "epoch": 37.44423199490121, "grad_norm": 0.5615801811218262, "learning_rate": 9.429766041370934e-05, "loss": 0.0183, "step": 58750 }, { "epoch": 37.45060548119822, "grad_norm": 0.3760916590690613, "learning_rate": 9.429510402004401e-05, "loss": 0.0257, "step": 58760 }, { "epoch": 37.45697896749522, "grad_norm": 0.622738778591156, "learning_rate": 9.429254708815136e-05, "loss": 0.0256, "step": 58770 }, { "epoch": 37.463352453792226, "grad_norm": 0.4987794756889343, "learning_rate": 9.428998961806244e-05, "loss": 0.0268, "step": 58780 }, { "epoch": 37.469725940089226, "grad_norm": 0.5066813230514526, "learning_rate": 9.428743160980833e-05, "loss": 0.0271, "step": 58790 }, { "epoch": 37.476099426386234, "grad_norm": 0.34712347388267517, "learning_rate": 9.428487306342011e-05, "loss": 0.0235, "step": 58800 }, { "epoch": 37.482472912683235, "grad_norm": 0.49838587641716003, "learning_rate": 9.428231397892888e-05, "loss": 0.0259, "step": 58810 }, { "epoch": 37.48884639898024, "grad_norm": 0.8514861464500427, "learning_rate": 9.427975435636571e-05, "loss": 0.044, "step": 58820 }, { "epoch": 37.495219885277244, "grad_norm": 0.3483809232711792, "learning_rate": 9.427719419576173e-05, "loss": 0.0397, "step": 58830 }, { "epoch": 37.50159337157425, "grad_norm": 0.5594362020492554, "learning_rate": 9.427463349714804e-05, "loss": 0.0352, "step": 58840 }, { "epoch": 37.50796685787125, "grad_norm": 0.41040951013565063, "learning_rate": 9.427207226055574e-05, "loss": 0.0331, "step": 58850 }, { "epoch": 37.51434034416826, "grad_norm": 0.37323230504989624, "learning_rate": 9.426951048601595e-05, "loss": 0.02, "step": 58860 }, { "epoch": 37.52071383046526, "grad_norm": 0.8023868799209595, "learning_rate": 9.426694817355983e-05, "loss": 0.028, "step": 58870 }, { "epoch": 37.52708731676227, "grad_norm": 0.5003507733345032, "learning_rate": 9.42643853232185e-05, "loss": 0.0366, "step": 58880 }, { "epoch": 37.53346080305927, "grad_norm": 0.5807331800460815, "learning_rate": 9.426182193502309e-05, "loss": 0.0409, "step": 58890 }, { "epoch": 37.53983428935628, "grad_norm": 0.4494272470474243, "learning_rate": 9.425925800900475e-05, "loss": 0.0311, "step": 58900 }, { "epoch": 37.54620777565328, "grad_norm": 0.3875611126422882, "learning_rate": 9.425669354519464e-05, "loss": 0.0264, "step": 58910 }, { "epoch": 37.55258126195029, "grad_norm": 0.4444333612918854, "learning_rate": 9.42541285436239e-05, "loss": 0.033, "step": 58920 }, { "epoch": 37.55895474824729, "grad_norm": 0.3254602551460266, "learning_rate": 9.425156300432373e-05, "loss": 0.0381, "step": 58930 }, { "epoch": 37.565328234544296, "grad_norm": 0.3972413241863251, "learning_rate": 9.424899692732529e-05, "loss": 0.0331, "step": 58940 }, { "epoch": 37.5717017208413, "grad_norm": 0.4516952335834503, "learning_rate": 9.424643031265976e-05, "loss": 0.0243, "step": 58950 }, { "epoch": 37.578075207138305, "grad_norm": 0.8825690150260925, "learning_rate": 9.424386316035833e-05, "loss": 0.0451, "step": 58960 }, { "epoch": 37.584448693435306, "grad_norm": 0.3919873535633087, "learning_rate": 9.424129547045218e-05, "loss": 0.0238, "step": 58970 }, { "epoch": 37.590822179732314, "grad_norm": 0.2833164930343628, "learning_rate": 9.423872724297252e-05, "loss": 0.0222, "step": 58980 }, { "epoch": 37.59719566602932, "grad_norm": 0.33000099658966064, "learning_rate": 9.423615847795057e-05, "loss": 0.0276, "step": 58990 }, { "epoch": 37.60356915232632, "grad_norm": 0.5424593687057495, "learning_rate": 9.423358917541751e-05, "loss": 0.0191, "step": 59000 }, { "epoch": 37.60994263862333, "grad_norm": 0.4948881268501282, "learning_rate": 9.423101933540459e-05, "loss": 0.0403, "step": 59010 }, { "epoch": 37.61631612492033, "grad_norm": 0.42425987124443054, "learning_rate": 9.422844895794301e-05, "loss": 0.0305, "step": 59020 }, { "epoch": 37.62268961121734, "grad_norm": 0.5077105164527893, "learning_rate": 9.422587804306404e-05, "loss": 0.0202, "step": 59030 }, { "epoch": 37.62906309751434, "grad_norm": 0.2441534847021103, "learning_rate": 9.422330659079887e-05, "loss": 0.0359, "step": 59040 }, { "epoch": 37.63543658381135, "grad_norm": 0.3992559313774109, "learning_rate": 9.42207346011788e-05, "loss": 0.0368, "step": 59050 }, { "epoch": 37.64181007010835, "grad_norm": 0.5145223736763, "learning_rate": 9.421816207423503e-05, "loss": 0.0318, "step": 59060 }, { "epoch": 37.64818355640536, "grad_norm": 0.4393545687198639, "learning_rate": 9.421558900999885e-05, "loss": 0.0318, "step": 59070 }, { "epoch": 37.65455704270236, "grad_norm": 0.6667615175247192, "learning_rate": 9.421301540850152e-05, "loss": 0.0298, "step": 59080 }, { "epoch": 37.660930528999366, "grad_norm": 0.5049098134040833, "learning_rate": 9.42104412697743e-05, "loss": 0.0382, "step": 59090 }, { "epoch": 37.66730401529637, "grad_norm": 0.5856661200523376, "learning_rate": 9.420786659384848e-05, "loss": 0.0408, "step": 59100 }, { "epoch": 37.673677501593374, "grad_norm": 0.4838673770427704, "learning_rate": 9.420529138075535e-05, "loss": 0.0333, "step": 59110 }, { "epoch": 37.680050987890375, "grad_norm": 0.6913846731185913, "learning_rate": 9.420271563052618e-05, "loss": 0.0326, "step": 59120 }, { "epoch": 37.68642447418738, "grad_norm": 0.5608755350112915, "learning_rate": 9.42001393431923e-05, "loss": 0.0266, "step": 59130 }, { "epoch": 37.692797960484384, "grad_norm": 0.617890477180481, "learning_rate": 9.419756251878498e-05, "loss": 0.0309, "step": 59140 }, { "epoch": 37.69917144678139, "grad_norm": 0.5727347731590271, "learning_rate": 9.419498515733554e-05, "loss": 0.0233, "step": 59150 }, { "epoch": 37.70554493307839, "grad_norm": 0.7098250389099121, "learning_rate": 9.41924072588753e-05, "loss": 0.0283, "step": 59160 }, { "epoch": 37.7119184193754, "grad_norm": 0.7657014727592468, "learning_rate": 9.418982882343561e-05, "loss": 0.0366, "step": 59170 }, { "epoch": 37.7182919056724, "grad_norm": 0.45679304003715515, "learning_rate": 9.418724985104777e-05, "loss": 0.0257, "step": 59180 }, { "epoch": 37.72466539196941, "grad_norm": 0.5330288410186768, "learning_rate": 9.418467034174312e-05, "loss": 0.0202, "step": 59190 }, { "epoch": 37.73103887826641, "grad_norm": 0.3801381587982178, "learning_rate": 9.4182090295553e-05, "loss": 0.0258, "step": 59200 }, { "epoch": 37.73741236456342, "grad_norm": 0.673989474773407, "learning_rate": 9.417950971250878e-05, "loss": 0.0173, "step": 59210 }, { "epoch": 37.74378585086042, "grad_norm": 0.6289236545562744, "learning_rate": 9.417692859264181e-05, "loss": 0.0275, "step": 59220 }, { "epoch": 37.75015933715743, "grad_norm": 0.6513596773147583, "learning_rate": 9.417434693598344e-05, "loss": 0.0222, "step": 59230 }, { "epoch": 37.75653282345443, "grad_norm": 0.5019389986991882, "learning_rate": 9.417176474256504e-05, "loss": 0.0275, "step": 59240 }, { "epoch": 37.762906309751436, "grad_norm": 0.39428678154945374, "learning_rate": 9.4169182012418e-05, "loss": 0.0198, "step": 59250 }, { "epoch": 37.76927979604844, "grad_norm": 0.3980458676815033, "learning_rate": 9.41665987455737e-05, "loss": 0.0275, "step": 59260 }, { "epoch": 37.775653282345445, "grad_norm": 0.3053942024707794, "learning_rate": 9.416401494206353e-05, "loss": 0.0244, "step": 59270 }, { "epoch": 37.782026768642446, "grad_norm": 0.3392542004585266, "learning_rate": 9.416143060191886e-05, "loss": 0.0207, "step": 59280 }, { "epoch": 37.788400254939454, "grad_norm": 0.43882572650909424, "learning_rate": 9.415884572517113e-05, "loss": 0.0204, "step": 59290 }, { "epoch": 37.794773741236455, "grad_norm": 0.40536171197891235, "learning_rate": 9.415626031185172e-05, "loss": 0.0239, "step": 59300 }, { "epoch": 37.80114722753346, "grad_norm": 0.42050808668136597, "learning_rate": 9.415367436199207e-05, "loss": 0.0255, "step": 59310 }, { "epoch": 37.80752071383046, "grad_norm": 0.7898723483085632, "learning_rate": 9.41510878756236e-05, "loss": 0.0336, "step": 59320 }, { "epoch": 37.81389420012747, "grad_norm": 0.2493704855442047, "learning_rate": 9.414850085277769e-05, "loss": 0.0296, "step": 59330 }, { "epoch": 37.82026768642447, "grad_norm": 0.4026285409927368, "learning_rate": 9.414591329348583e-05, "loss": 0.03, "step": 59340 }, { "epoch": 37.82664117272148, "grad_norm": 0.3279447853565216, "learning_rate": 9.414332519777944e-05, "loss": 0.0212, "step": 59350 }, { "epoch": 37.83301465901848, "grad_norm": 0.4612616300582886, "learning_rate": 9.414073656568998e-05, "loss": 0.0225, "step": 59360 }, { "epoch": 37.83938814531549, "grad_norm": 0.6776186227798462, "learning_rate": 9.413814739724889e-05, "loss": 0.0313, "step": 59370 }, { "epoch": 37.84576163161249, "grad_norm": 0.5331281423568726, "learning_rate": 9.413555769248763e-05, "loss": 0.0328, "step": 59380 }, { "epoch": 37.8521351179095, "grad_norm": 0.3487010896205902, "learning_rate": 9.413296745143769e-05, "loss": 0.0201, "step": 59390 }, { "epoch": 37.8585086042065, "grad_norm": 0.37419599294662476, "learning_rate": 9.413037667413051e-05, "loss": 0.0275, "step": 59400 }, { "epoch": 37.86488209050351, "grad_norm": 0.8000113368034363, "learning_rate": 9.412778536059759e-05, "loss": 0.0241, "step": 59410 }, { "epoch": 37.87125557680051, "grad_norm": 0.47382041811943054, "learning_rate": 9.412519351087041e-05, "loss": 0.0276, "step": 59420 }, { "epoch": 37.877629063097515, "grad_norm": 0.3366382420063019, "learning_rate": 9.412260112498049e-05, "loss": 0.0289, "step": 59430 }, { "epoch": 37.884002549394516, "grad_norm": 0.4883997142314911, "learning_rate": 9.412000820295928e-05, "loss": 0.0278, "step": 59440 }, { "epoch": 37.890376035691524, "grad_norm": 0.4499445855617523, "learning_rate": 9.411741474483833e-05, "loss": 0.0293, "step": 59450 }, { "epoch": 37.896749521988525, "grad_norm": 0.3948924243450165, "learning_rate": 9.411482075064913e-05, "loss": 0.0219, "step": 59460 }, { "epoch": 37.90312300828553, "grad_norm": 0.42939478158950806, "learning_rate": 9.411222622042322e-05, "loss": 0.0294, "step": 59470 }, { "epoch": 37.909496494582534, "grad_norm": 0.4224923849105835, "learning_rate": 9.410963115419209e-05, "loss": 0.0215, "step": 59480 }, { "epoch": 37.91586998087954, "grad_norm": 0.36642277240753174, "learning_rate": 9.410703555198731e-05, "loss": 0.0258, "step": 59490 }, { "epoch": 37.92224346717654, "grad_norm": 0.33882126212120056, "learning_rate": 9.410443941384039e-05, "loss": 0.0294, "step": 59500 }, { "epoch": 37.92861695347355, "grad_norm": 0.42448851466178894, "learning_rate": 9.41018427397829e-05, "loss": 0.0342, "step": 59510 }, { "epoch": 37.93499043977055, "grad_norm": 0.45555663108825684, "learning_rate": 9.40992455298464e-05, "loss": 0.0208, "step": 59520 }, { "epoch": 37.94136392606756, "grad_norm": 0.43489423394203186, "learning_rate": 9.40966477840624e-05, "loss": 0.023, "step": 59530 }, { "epoch": 37.94773741236456, "grad_norm": 0.6702353954315186, "learning_rate": 9.409404950246251e-05, "loss": 0.0325, "step": 59540 }, { "epoch": 37.95411089866157, "grad_norm": 0.41264376044273376, "learning_rate": 9.409145068507829e-05, "loss": 0.0263, "step": 59550 }, { "epoch": 37.96048438495857, "grad_norm": 0.48045140504837036, "learning_rate": 9.408885133194131e-05, "loss": 0.0297, "step": 59560 }, { "epoch": 37.96685787125558, "grad_norm": 0.37375152111053467, "learning_rate": 9.408625144308316e-05, "loss": 0.0389, "step": 59570 }, { "epoch": 37.97323135755258, "grad_norm": 0.4405043125152588, "learning_rate": 9.408365101853543e-05, "loss": 0.039, "step": 59580 }, { "epoch": 37.979604843849586, "grad_norm": 0.33477380871772766, "learning_rate": 9.408105005832972e-05, "loss": 0.0233, "step": 59590 }, { "epoch": 37.98597833014659, "grad_norm": 0.43110743165016174, "learning_rate": 9.407844856249763e-05, "loss": 0.0156, "step": 59600 }, { "epoch": 37.992351816443595, "grad_norm": 0.20999041199684143, "learning_rate": 9.407584653107078e-05, "loss": 0.0209, "step": 59610 }, { "epoch": 37.998725302740596, "grad_norm": 0.5987453460693359, "learning_rate": 9.407324396408077e-05, "loss": 0.0204, "step": 59620 }, { "epoch": 38.0050987890376, "grad_norm": 0.5122214555740356, "learning_rate": 9.407064086155924e-05, "loss": 0.0291, "step": 59630 }, { "epoch": 38.01147227533461, "grad_norm": 0.3935222029685974, "learning_rate": 9.40680372235378e-05, "loss": 0.0243, "step": 59640 }, { "epoch": 38.01784576163161, "grad_norm": 0.3285004198551178, "learning_rate": 9.406543305004812e-05, "loss": 0.0182, "step": 59650 }, { "epoch": 38.02421924792862, "grad_norm": 0.5359097719192505, "learning_rate": 9.406282834112182e-05, "loss": 0.031, "step": 59660 }, { "epoch": 38.03059273422562, "grad_norm": 0.25976693630218506, "learning_rate": 9.406022309679054e-05, "loss": 0.0389, "step": 59670 }, { "epoch": 38.03696622052263, "grad_norm": 0.4073411226272583, "learning_rate": 9.405761731708597e-05, "loss": 0.0399, "step": 59680 }, { "epoch": 38.04333970681963, "grad_norm": 0.43172791600227356, "learning_rate": 9.405501100203974e-05, "loss": 0.0283, "step": 59690 }, { "epoch": 38.04971319311664, "grad_norm": 0.6576762795448303, "learning_rate": 9.405240415168353e-05, "loss": 0.0442, "step": 59700 }, { "epoch": 38.05608667941364, "grad_norm": 0.5279595255851746, "learning_rate": 9.404979676604902e-05, "loss": 0.0363, "step": 59710 }, { "epoch": 38.06246016571065, "grad_norm": 0.33692070841789246, "learning_rate": 9.40471888451679e-05, "loss": 0.0295, "step": 59720 }, { "epoch": 38.06883365200765, "grad_norm": 0.3834635019302368, "learning_rate": 9.404458038907183e-05, "loss": 0.0277, "step": 59730 }, { "epoch": 38.075207138304656, "grad_norm": 0.444149374961853, "learning_rate": 9.404197139779253e-05, "loss": 0.0245, "step": 59740 }, { "epoch": 38.081580624601656, "grad_norm": 0.4036004841327667, "learning_rate": 9.403936187136169e-05, "loss": 0.0371, "step": 59750 }, { "epoch": 38.087954110898664, "grad_norm": 0.7655401229858398, "learning_rate": 9.403675180981104e-05, "loss": 0.0373, "step": 59760 }, { "epoch": 38.094327597195665, "grad_norm": 0.7658129334449768, "learning_rate": 9.403414121317226e-05, "loss": 0.0259, "step": 59770 }, { "epoch": 38.10070108349267, "grad_norm": 0.6385941505432129, "learning_rate": 9.403153008147709e-05, "loss": 0.0339, "step": 59780 }, { "epoch": 38.107074569789674, "grad_norm": 2.1804392337799072, "learning_rate": 9.402891841475728e-05, "loss": 0.0239, "step": 59790 }, { "epoch": 38.11344805608668, "grad_norm": 0.42595407366752625, "learning_rate": 9.402630621304452e-05, "loss": 0.0332, "step": 59800 }, { "epoch": 38.11982154238368, "grad_norm": 0.6152677536010742, "learning_rate": 9.402369347637057e-05, "loss": 0.0247, "step": 59810 }, { "epoch": 38.12619502868069, "grad_norm": 0.7131217122077942, "learning_rate": 9.402108020476718e-05, "loss": 0.0243, "step": 59820 }, { "epoch": 38.13256851497769, "grad_norm": 0.39470013976097107, "learning_rate": 9.401846639826612e-05, "loss": 0.0207, "step": 59830 }, { "epoch": 38.1389420012747, "grad_norm": 0.5159879326820374, "learning_rate": 9.401585205689912e-05, "loss": 0.0237, "step": 59840 }, { "epoch": 38.1453154875717, "grad_norm": 0.2219592183828354, "learning_rate": 9.401323718069796e-05, "loss": 0.0218, "step": 59850 }, { "epoch": 38.15168897386871, "grad_norm": 0.43782442808151245, "learning_rate": 9.401062176969442e-05, "loss": 0.0196, "step": 59860 }, { "epoch": 38.15806246016571, "grad_norm": 0.44834721088409424, "learning_rate": 9.400800582392027e-05, "loss": 0.0181, "step": 59870 }, { "epoch": 38.16443594646272, "grad_norm": 0.3449564576148987, "learning_rate": 9.400538934340728e-05, "loss": 0.0263, "step": 59880 }, { "epoch": 38.17080943275972, "grad_norm": 0.5342633724212646, "learning_rate": 9.400277232818728e-05, "loss": 0.0207, "step": 59890 }, { "epoch": 38.177182919056726, "grad_norm": 0.4120038151741028, "learning_rate": 9.400015477829205e-05, "loss": 0.0221, "step": 59900 }, { "epoch": 38.18355640535373, "grad_norm": 0.4158679246902466, "learning_rate": 9.399753669375339e-05, "loss": 0.0331, "step": 59910 }, { "epoch": 38.189929891650735, "grad_norm": 0.43505528569221497, "learning_rate": 9.399491807460312e-05, "loss": 0.0321, "step": 59920 }, { "epoch": 38.196303377947736, "grad_norm": 0.40071165561676025, "learning_rate": 9.399229892087306e-05, "loss": 0.0217, "step": 59930 }, { "epoch": 38.202676864244744, "grad_norm": 0.3870788514614105, "learning_rate": 9.398967923259503e-05, "loss": 0.0288, "step": 59940 }, { "epoch": 38.209050350541744, "grad_norm": 0.7408939003944397, "learning_rate": 9.398705900980088e-05, "loss": 0.0251, "step": 59950 }, { "epoch": 38.21542383683875, "grad_norm": 0.7131556868553162, "learning_rate": 9.398443825252241e-05, "loss": 0.0347, "step": 59960 }, { "epoch": 38.22179732313575, "grad_norm": 0.5342478156089783, "learning_rate": 9.398181696079149e-05, "loss": 0.028, "step": 59970 }, { "epoch": 38.22817080943276, "grad_norm": 0.5344670414924622, "learning_rate": 9.397919513463998e-05, "loss": 0.0263, "step": 59980 }, { "epoch": 38.23454429572976, "grad_norm": 0.33856165409088135, "learning_rate": 9.397657277409971e-05, "loss": 0.0215, "step": 59990 }, { "epoch": 38.24091778202677, "grad_norm": 0.5994165539741516, "learning_rate": 9.397394987920257e-05, "loss": 0.0366, "step": 60000 }, { "epoch": 38.24729126832377, "grad_norm": 0.491438090801239, "learning_rate": 9.397132644998042e-05, "loss": 0.0234, "step": 60010 }, { "epoch": 38.25366475462078, "grad_norm": 0.5649241209030151, "learning_rate": 9.396870248646515e-05, "loss": 0.0301, "step": 60020 }, { "epoch": 38.26003824091778, "grad_norm": 0.449146032333374, "learning_rate": 9.396607798868861e-05, "loss": 0.0196, "step": 60030 }, { "epoch": 38.26641172721479, "grad_norm": 0.3796714246273041, "learning_rate": 9.396345295668272e-05, "loss": 0.0219, "step": 60040 }, { "epoch": 38.27278521351179, "grad_norm": 0.8207438588142395, "learning_rate": 9.396082739047937e-05, "loss": 0.0242, "step": 60050 }, { "epoch": 38.2791586998088, "grad_norm": 0.3021965026855469, "learning_rate": 9.395820129011046e-05, "loss": 0.032, "step": 60060 }, { "epoch": 38.2855321861058, "grad_norm": 0.37927284836769104, "learning_rate": 9.39555746556079e-05, "loss": 0.0194, "step": 60070 }, { "epoch": 38.291905672402805, "grad_norm": 0.6051521301269531, "learning_rate": 9.395294748700361e-05, "loss": 0.0215, "step": 60080 }, { "epoch": 38.298279158699806, "grad_norm": 1.012482762336731, "learning_rate": 9.395031978432952e-05, "loss": 0.0348, "step": 60090 }, { "epoch": 38.304652644996814, "grad_norm": 0.5415579080581665, "learning_rate": 9.394769154761754e-05, "loss": 0.0177, "step": 60100 }, { "epoch": 38.311026131293815, "grad_norm": 0.5627006888389587, "learning_rate": 9.39450627768996e-05, "loss": 0.0371, "step": 60110 }, { "epoch": 38.31739961759082, "grad_norm": 0.6162716746330261, "learning_rate": 9.394243347220766e-05, "loss": 0.026, "step": 60120 }, { "epoch": 38.323773103887824, "grad_norm": 0.6367555260658264, "learning_rate": 9.393980363357368e-05, "loss": 0.031, "step": 60130 }, { "epoch": 38.33014659018483, "grad_norm": 0.520797848701477, "learning_rate": 9.393717326102959e-05, "loss": 0.0227, "step": 60140 }, { "epoch": 38.33652007648183, "grad_norm": 0.7482666969299316, "learning_rate": 9.393454235460736e-05, "loss": 0.0232, "step": 60150 }, { "epoch": 38.34289356277884, "grad_norm": 0.4850766956806183, "learning_rate": 9.393191091433896e-05, "loss": 0.0287, "step": 60160 }, { "epoch": 38.34926704907584, "grad_norm": 0.30845627188682556, "learning_rate": 9.392927894025637e-05, "loss": 0.0306, "step": 60170 }, { "epoch": 38.35564053537285, "grad_norm": 0.41747283935546875, "learning_rate": 9.392664643239155e-05, "loss": 0.0295, "step": 60180 }, { "epoch": 38.36201402166985, "grad_norm": 0.31774675846099854, "learning_rate": 9.392401339077652e-05, "loss": 0.0226, "step": 60190 }, { "epoch": 38.36838750796686, "grad_norm": 0.3680015504360199, "learning_rate": 9.392137981544325e-05, "loss": 0.0335, "step": 60200 }, { "epoch": 38.37476099426386, "grad_norm": 0.4650542438030243, "learning_rate": 9.391874570642375e-05, "loss": 0.0257, "step": 60210 }, { "epoch": 38.38113448056087, "grad_norm": 0.5423802137374878, "learning_rate": 9.391611106375002e-05, "loss": 0.0198, "step": 60220 }, { "epoch": 38.38750796685787, "grad_norm": 0.41541942954063416, "learning_rate": 9.391347588745407e-05, "loss": 0.0176, "step": 60230 }, { "epoch": 38.393881453154876, "grad_norm": 0.6082161068916321, "learning_rate": 9.391084017756794e-05, "loss": 0.0274, "step": 60240 }, { "epoch": 38.400254939451884, "grad_norm": 0.5188995599746704, "learning_rate": 9.390820393412363e-05, "loss": 0.0244, "step": 60250 }, { "epoch": 38.406628425748885, "grad_norm": 0.4696905016899109, "learning_rate": 9.39055671571532e-05, "loss": 0.031, "step": 60260 }, { "epoch": 38.41300191204589, "grad_norm": 0.4666711091995239, "learning_rate": 9.390292984668867e-05, "loss": 0.0278, "step": 60270 }, { "epoch": 38.41937539834289, "grad_norm": 0.3953295350074768, "learning_rate": 9.390029200276208e-05, "loss": 0.0273, "step": 60280 }, { "epoch": 38.4257488846399, "grad_norm": 0.4609638452529907, "learning_rate": 9.389765362540551e-05, "loss": 0.0218, "step": 60290 }, { "epoch": 38.4321223709369, "grad_norm": 0.6821250319480896, "learning_rate": 9.389501471465099e-05, "loss": 0.0269, "step": 60300 }, { "epoch": 38.43849585723391, "grad_norm": 0.46804070472717285, "learning_rate": 9.38923752705306e-05, "loss": 0.0241, "step": 60310 }, { "epoch": 38.44486934353091, "grad_norm": 0.46511170268058777, "learning_rate": 9.388973529307641e-05, "loss": 0.0194, "step": 60320 }, { "epoch": 38.45124282982792, "grad_norm": 0.35595694184303284, "learning_rate": 9.388709478232049e-05, "loss": 0.0253, "step": 60330 }, { "epoch": 38.45761631612492, "grad_norm": 0.5269393920898438, "learning_rate": 9.388445373829495e-05, "loss": 0.0221, "step": 60340 }, { "epoch": 38.46398980242193, "grad_norm": 0.7659209370613098, "learning_rate": 9.388181216103185e-05, "loss": 0.0344, "step": 60350 }, { "epoch": 38.47036328871893, "grad_norm": 0.709048330783844, "learning_rate": 9.38791700505633e-05, "loss": 0.0238, "step": 60360 }, { "epoch": 38.47673677501594, "grad_norm": 0.32942870259284973, "learning_rate": 9.38765274069214e-05, "loss": 0.0244, "step": 60370 }, { "epoch": 38.48311026131294, "grad_norm": 0.3576360046863556, "learning_rate": 9.38738842301383e-05, "loss": 0.0159, "step": 60380 }, { "epoch": 38.489483747609945, "grad_norm": 0.4606381356716156, "learning_rate": 9.387124052024605e-05, "loss": 0.0311, "step": 60390 }, { "epoch": 38.495857233906946, "grad_norm": 0.21920731663703918, "learning_rate": 9.386859627727682e-05, "loss": 0.0226, "step": 60400 }, { "epoch": 38.502230720203954, "grad_norm": 0.695374608039856, "learning_rate": 9.386595150126272e-05, "loss": 0.0239, "step": 60410 }, { "epoch": 38.508604206500955, "grad_norm": 0.5035502314567566, "learning_rate": 9.386330619223589e-05, "loss": 0.0264, "step": 60420 }, { "epoch": 38.51497769279796, "grad_norm": 0.3692714273929596, "learning_rate": 9.386066035022848e-05, "loss": 0.0173, "step": 60430 }, { "epoch": 38.521351179094964, "grad_norm": 0.698500394821167, "learning_rate": 9.385801397527263e-05, "loss": 0.0348, "step": 60440 }, { "epoch": 38.52772466539197, "grad_norm": 0.3472888171672821, "learning_rate": 9.385536706740051e-05, "loss": 0.0321, "step": 60450 }, { "epoch": 38.53409815168897, "grad_norm": 0.6455478668212891, "learning_rate": 9.385271962664427e-05, "loss": 0.0395, "step": 60460 }, { "epoch": 38.54047163798598, "grad_norm": 0.3985987603664398, "learning_rate": 9.385007165303609e-05, "loss": 0.0424, "step": 60470 }, { "epoch": 38.54684512428298, "grad_norm": 0.5377212762832642, "learning_rate": 9.384742314660812e-05, "loss": 0.0283, "step": 60480 }, { "epoch": 38.55321861057999, "grad_norm": 0.47219470143318176, "learning_rate": 9.384477410739257e-05, "loss": 0.0201, "step": 60490 }, { "epoch": 38.55959209687699, "grad_norm": 0.6544003486633301, "learning_rate": 9.384212453542162e-05, "loss": 0.0245, "step": 60500 }, { "epoch": 38.565965583174, "grad_norm": 0.36290815472602844, "learning_rate": 9.383947443072747e-05, "loss": 0.0366, "step": 60510 }, { "epoch": 38.572339069471, "grad_norm": 0.5753461122512817, "learning_rate": 9.383682379334231e-05, "loss": 0.0228, "step": 60520 }, { "epoch": 38.57871255576801, "grad_norm": 0.3562580645084381, "learning_rate": 9.383417262329834e-05, "loss": 0.0233, "step": 60530 }, { "epoch": 38.58508604206501, "grad_norm": 0.41233503818511963, "learning_rate": 9.38315209206278e-05, "loss": 0.0199, "step": 60540 }, { "epoch": 38.591459528362016, "grad_norm": 0.4829746186733246, "learning_rate": 9.38288686853629e-05, "loss": 0.0235, "step": 60550 }, { "epoch": 38.59783301465902, "grad_norm": 0.4323595464229584, "learning_rate": 9.382621591753584e-05, "loss": 0.0303, "step": 60560 }, { "epoch": 38.604206500956025, "grad_norm": 0.6684671640396118, "learning_rate": 9.38235626171789e-05, "loss": 0.0229, "step": 60570 }, { "epoch": 38.610579987253026, "grad_norm": 0.40282732248306274, "learning_rate": 9.38209087843243e-05, "loss": 0.0232, "step": 60580 }, { "epoch": 38.61695347355003, "grad_norm": 1.5885354280471802, "learning_rate": 9.381825441900427e-05, "loss": 0.0417, "step": 60590 }, { "epoch": 38.623326959847034, "grad_norm": 0.8120958805084229, "learning_rate": 9.381559952125108e-05, "loss": 0.0293, "step": 60600 }, { "epoch": 38.62970044614404, "grad_norm": 0.593139111995697, "learning_rate": 9.3812944091097e-05, "loss": 0.0214, "step": 60610 }, { "epoch": 38.63607393244104, "grad_norm": 0.32957184314727783, "learning_rate": 9.381028812857426e-05, "loss": 0.0215, "step": 60620 }, { "epoch": 38.64244741873805, "grad_norm": 0.5119672417640686, "learning_rate": 9.380763163371517e-05, "loss": 0.0216, "step": 60630 }, { "epoch": 38.64882090503505, "grad_norm": 0.42426764965057373, "learning_rate": 9.3804974606552e-05, "loss": 0.0211, "step": 60640 }, { "epoch": 38.65519439133206, "grad_norm": 0.48537662625312805, "learning_rate": 9.380231704711703e-05, "loss": 0.0238, "step": 60650 }, { "epoch": 38.66156787762906, "grad_norm": 0.4036189317703247, "learning_rate": 9.379965895544254e-05, "loss": 0.021, "step": 60660 }, { "epoch": 38.66794136392607, "grad_norm": 0.618143618106842, "learning_rate": 9.379700033156085e-05, "loss": 0.0205, "step": 60670 }, { "epoch": 38.67431485022307, "grad_norm": 0.675474226474762, "learning_rate": 9.379434117550425e-05, "loss": 0.0268, "step": 60680 }, { "epoch": 38.68068833652008, "grad_norm": 0.47942447662353516, "learning_rate": 9.379168148730506e-05, "loss": 0.0253, "step": 60690 }, { "epoch": 38.68706182281708, "grad_norm": 0.4218752384185791, "learning_rate": 9.378902126699558e-05, "loss": 0.0269, "step": 60700 }, { "epoch": 38.693435309114086, "grad_norm": 0.39426541328430176, "learning_rate": 9.378636051460816e-05, "loss": 0.0206, "step": 60710 }, { "epoch": 38.69980879541109, "grad_norm": 0.2343386709690094, "learning_rate": 9.378369923017511e-05, "loss": 0.0249, "step": 60720 }, { "epoch": 38.706182281708095, "grad_norm": 0.32120057940483093, "learning_rate": 9.378103741372879e-05, "loss": 0.0275, "step": 60730 }, { "epoch": 38.712555768005096, "grad_norm": 0.5255295038223267, "learning_rate": 9.377837506530151e-05, "loss": 0.0299, "step": 60740 }, { "epoch": 38.718929254302104, "grad_norm": 0.6949279308319092, "learning_rate": 9.377571218492567e-05, "loss": 0.0282, "step": 60750 }, { "epoch": 38.725302740599105, "grad_norm": 0.395681232213974, "learning_rate": 9.377304877263358e-05, "loss": 0.0191, "step": 60760 }, { "epoch": 38.73167622689611, "grad_norm": 0.48582664132118225, "learning_rate": 9.37703848284576e-05, "loss": 0.0327, "step": 60770 }, { "epoch": 38.738049713193114, "grad_norm": 1.597266674041748, "learning_rate": 9.376772035243013e-05, "loss": 0.0235, "step": 60780 }, { "epoch": 38.74442319949012, "grad_norm": 0.4729820489883423, "learning_rate": 9.376505534458355e-05, "loss": 0.0448, "step": 60790 }, { "epoch": 38.75079668578712, "grad_norm": 0.4519246220588684, "learning_rate": 9.37623898049502e-05, "loss": 0.0284, "step": 60800 }, { "epoch": 38.75717017208413, "grad_norm": 0.4270004630088806, "learning_rate": 9.375972373356252e-05, "loss": 0.0229, "step": 60810 }, { "epoch": 38.76354365838113, "grad_norm": 0.4639846682548523, "learning_rate": 9.375705713045287e-05, "loss": 0.0263, "step": 60820 }, { "epoch": 38.76991714467814, "grad_norm": 0.5524167418479919, "learning_rate": 9.375438999565366e-05, "loss": 0.0277, "step": 60830 }, { "epoch": 38.77629063097514, "grad_norm": 0.2680951654911041, "learning_rate": 9.375172232919731e-05, "loss": 0.0237, "step": 60840 }, { "epoch": 38.78266411727215, "grad_norm": 0.3372533321380615, "learning_rate": 9.374905413111622e-05, "loss": 0.026, "step": 60850 }, { "epoch": 38.78903760356915, "grad_norm": 0.5349087119102478, "learning_rate": 9.374638540144281e-05, "loss": 0.0304, "step": 60860 }, { "epoch": 38.79541108986616, "grad_norm": 0.5707732439041138, "learning_rate": 9.374371614020951e-05, "loss": 0.0275, "step": 60870 }, { "epoch": 38.80178457616316, "grad_norm": 0.4641343951225281, "learning_rate": 9.374104634744878e-05, "loss": 0.0258, "step": 60880 }, { "epoch": 38.808158062460166, "grad_norm": 0.44850394129753113, "learning_rate": 9.373837602319301e-05, "loss": 0.0194, "step": 60890 }, { "epoch": 38.814531548757174, "grad_norm": 0.4024102985858917, "learning_rate": 9.37357051674747e-05, "loss": 0.0221, "step": 60900 }, { "epoch": 38.820905035054174, "grad_norm": 0.5499832630157471, "learning_rate": 9.373303378032628e-05, "loss": 0.023, "step": 60910 }, { "epoch": 38.82727852135118, "grad_norm": 0.6478255987167358, "learning_rate": 9.373036186178019e-05, "loss": 0.0289, "step": 60920 }, { "epoch": 38.83365200764818, "grad_norm": 0.5420502424240112, "learning_rate": 9.372768941186891e-05, "loss": 0.0341, "step": 60930 }, { "epoch": 38.84002549394519, "grad_norm": 0.31860363483428955, "learning_rate": 9.372501643062494e-05, "loss": 0.0334, "step": 60940 }, { "epoch": 38.84639898024219, "grad_norm": 0.37389856576919556, "learning_rate": 9.372234291808073e-05, "loss": 0.0247, "step": 60950 }, { "epoch": 38.8527724665392, "grad_norm": 0.4862537682056427, "learning_rate": 9.371966887426877e-05, "loss": 0.0415, "step": 60960 }, { "epoch": 38.8591459528362, "grad_norm": 0.29928916692733765, "learning_rate": 9.371699429922157e-05, "loss": 0.0353, "step": 60970 }, { "epoch": 38.86551943913321, "grad_norm": 0.5489560961723328, "learning_rate": 9.371431919297158e-05, "loss": 0.0311, "step": 60980 }, { "epoch": 38.87189292543021, "grad_norm": 0.7020895481109619, "learning_rate": 9.371164355555135e-05, "loss": 0.0352, "step": 60990 }, { "epoch": 38.87826641172722, "grad_norm": 0.379985511302948, "learning_rate": 9.370896738699339e-05, "loss": 0.0264, "step": 61000 }, { "epoch": 38.88463989802422, "grad_norm": 0.36817467212677, "learning_rate": 9.370629068733022e-05, "loss": 0.0302, "step": 61010 }, { "epoch": 38.89101338432123, "grad_norm": 0.34796997904777527, "learning_rate": 9.370361345659433e-05, "loss": 0.0289, "step": 61020 }, { "epoch": 38.89738687061823, "grad_norm": 0.3631622791290283, "learning_rate": 9.370093569481829e-05, "loss": 0.0238, "step": 61030 }, { "epoch": 38.903760356915235, "grad_norm": 0.809110701084137, "learning_rate": 9.369825740203462e-05, "loss": 0.0251, "step": 61040 }, { "epoch": 38.910133843212236, "grad_norm": 0.5102279186248779, "learning_rate": 9.369557857827584e-05, "loss": 0.0356, "step": 61050 }, { "epoch": 38.916507329509244, "grad_norm": 0.6785749197006226, "learning_rate": 9.369289922357455e-05, "loss": 0.0321, "step": 61060 }, { "epoch": 38.922880815806245, "grad_norm": 0.5038653612136841, "learning_rate": 9.369021933796327e-05, "loss": 0.0249, "step": 61070 }, { "epoch": 38.92925430210325, "grad_norm": 0.5456666946411133, "learning_rate": 9.368753892147458e-05, "loss": 0.0359, "step": 61080 }, { "epoch": 38.935627788400254, "grad_norm": 0.6632387638092041, "learning_rate": 9.368485797414105e-05, "loss": 0.0303, "step": 61090 }, { "epoch": 38.94200127469726, "grad_norm": 0.3330095708370209, "learning_rate": 9.368217649599524e-05, "loss": 0.0225, "step": 61100 }, { "epoch": 38.94837476099426, "grad_norm": 0.32467252016067505, "learning_rate": 9.367949448706974e-05, "loss": 0.0284, "step": 61110 }, { "epoch": 38.95474824729127, "grad_norm": 0.4425029456615448, "learning_rate": 9.367681194739715e-05, "loss": 0.0212, "step": 61120 }, { "epoch": 38.96112173358827, "grad_norm": 0.5201866030693054, "learning_rate": 9.367412887701005e-05, "loss": 0.0283, "step": 61130 }, { "epoch": 38.96749521988528, "grad_norm": 0.45859822630882263, "learning_rate": 9.367144527594106e-05, "loss": 0.0276, "step": 61140 }, { "epoch": 38.97386870618228, "grad_norm": 0.4274231195449829, "learning_rate": 9.366876114422278e-05, "loss": 0.0224, "step": 61150 }, { "epoch": 38.98024219247929, "grad_norm": 0.5036986470222473, "learning_rate": 9.36660764818878e-05, "loss": 0.0315, "step": 61160 }, { "epoch": 38.98661567877629, "grad_norm": 0.3120930790901184, "learning_rate": 9.366339128896878e-05, "loss": 0.0251, "step": 61170 }, { "epoch": 38.9929891650733, "grad_norm": 0.25410112738609314, "learning_rate": 9.366070556549833e-05, "loss": 0.0384, "step": 61180 }, { "epoch": 38.9993626513703, "grad_norm": 0.8482842445373535, "learning_rate": 9.365801931150909e-05, "loss": 0.024, "step": 61190 }, { "epoch": 39.005736137667306, "grad_norm": 0.5147600173950195, "learning_rate": 9.365533252703368e-05, "loss": 0.0246, "step": 61200 }, { "epoch": 39.01210962396431, "grad_norm": 0.3764215111732483, "learning_rate": 9.365264521210478e-05, "loss": 0.0249, "step": 61210 }, { "epoch": 39.018483110261315, "grad_norm": 0.4297805726528168, "learning_rate": 9.3649957366755e-05, "loss": 0.0231, "step": 61220 }, { "epoch": 39.024856596558315, "grad_norm": 0.30081427097320557, "learning_rate": 9.364726899101703e-05, "loss": 0.0224, "step": 61230 }, { "epoch": 39.03123008285532, "grad_norm": 0.49720096588134766, "learning_rate": 9.364458008492356e-05, "loss": 0.0265, "step": 61240 }, { "epoch": 39.037603569152324, "grad_norm": 0.4793117046356201, "learning_rate": 9.364189064850721e-05, "loss": 0.0381, "step": 61250 }, { "epoch": 39.04397705544933, "grad_norm": 0.5174895524978638, "learning_rate": 9.36392006818007e-05, "loss": 0.0303, "step": 61260 }, { "epoch": 39.05035054174633, "grad_norm": 0.43084970116615295, "learning_rate": 9.363651018483667e-05, "loss": 0.0379, "step": 61270 }, { "epoch": 39.05672402804334, "grad_norm": 0.3522895574569702, "learning_rate": 9.363381915764788e-05, "loss": 0.0193, "step": 61280 }, { "epoch": 39.06309751434034, "grad_norm": 0.614811897277832, "learning_rate": 9.363112760026698e-05, "loss": 0.0396, "step": 61290 }, { "epoch": 39.06947100063735, "grad_norm": 0.6192757487297058, "learning_rate": 9.362843551272666e-05, "loss": 0.0377, "step": 61300 }, { "epoch": 39.07584448693435, "grad_norm": 0.7819889187812805, "learning_rate": 9.362574289505968e-05, "loss": 0.0377, "step": 61310 }, { "epoch": 39.08221797323136, "grad_norm": 0.4721345007419586, "learning_rate": 9.362304974729874e-05, "loss": 0.0328, "step": 61320 }, { "epoch": 39.08859145952836, "grad_norm": 0.3917187452316284, "learning_rate": 9.362035606947654e-05, "loss": 0.0292, "step": 61330 }, { "epoch": 39.09496494582537, "grad_norm": 0.3970746099948883, "learning_rate": 9.361766186162584e-05, "loss": 0.0322, "step": 61340 }, { "epoch": 39.10133843212237, "grad_norm": 0.7976200580596924, "learning_rate": 9.361496712377935e-05, "loss": 0.0241, "step": 61350 }, { "epoch": 39.107711918419376, "grad_norm": 0.5349223613739014, "learning_rate": 9.361227185596985e-05, "loss": 0.0224, "step": 61360 }, { "epoch": 39.11408540471638, "grad_norm": 0.4572047293186188, "learning_rate": 9.360957605823007e-05, "loss": 0.0232, "step": 61370 }, { "epoch": 39.120458891013385, "grad_norm": 0.4388844966888428, "learning_rate": 9.360687973059275e-05, "loss": 0.03, "step": 61380 }, { "epoch": 39.126832377310386, "grad_norm": 0.4839017391204834, "learning_rate": 9.360418287309067e-05, "loss": 0.0343, "step": 61390 }, { "epoch": 39.133205863607394, "grad_norm": 0.6416274309158325, "learning_rate": 9.360148548575662e-05, "loss": 0.031, "step": 61400 }, { "epoch": 39.139579349904395, "grad_norm": 0.624721884727478, "learning_rate": 9.359878756862333e-05, "loss": 0.0264, "step": 61410 }, { "epoch": 39.1459528362014, "grad_norm": 0.5643642544746399, "learning_rate": 9.359608912172362e-05, "loss": 0.0265, "step": 61420 }, { "epoch": 39.1523263224984, "grad_norm": 0.42867419123649597, "learning_rate": 9.359339014509028e-05, "loss": 0.0211, "step": 61430 }, { "epoch": 39.15869980879541, "grad_norm": 0.6147358417510986, "learning_rate": 9.359069063875607e-05, "loss": 0.0261, "step": 61440 }, { "epoch": 39.16507329509241, "grad_norm": 0.6237074732780457, "learning_rate": 9.358799060275382e-05, "loss": 0.0267, "step": 61450 }, { "epoch": 39.17144678138942, "grad_norm": 0.6123011112213135, "learning_rate": 9.358529003711632e-05, "loss": 0.0243, "step": 61460 }, { "epoch": 39.17782026768642, "grad_norm": 0.4978525638580322, "learning_rate": 9.35825889418764e-05, "loss": 0.0235, "step": 61470 }, { "epoch": 39.18419375398343, "grad_norm": 0.5927453637123108, "learning_rate": 9.357988731706689e-05, "loss": 0.0256, "step": 61480 }, { "epoch": 39.19056724028043, "grad_norm": 0.42637959122657776, "learning_rate": 9.357718516272058e-05, "loss": 0.0255, "step": 61490 }, { "epoch": 39.19694072657744, "grad_norm": 0.4307953715324402, "learning_rate": 9.357448247887036e-05, "loss": 0.0252, "step": 61500 }, { "epoch": 39.203314212874446, "grad_norm": 0.5068081617355347, "learning_rate": 9.3571779265549e-05, "loss": 0.0281, "step": 61510 }, { "epoch": 39.20968769917145, "grad_norm": 0.5475598573684692, "learning_rate": 9.35690755227894e-05, "loss": 0.0289, "step": 61520 }, { "epoch": 39.216061185468455, "grad_norm": 0.6064635515213013, "learning_rate": 9.356637125062439e-05, "loss": 0.025, "step": 61530 }, { "epoch": 39.222434671765456, "grad_norm": 0.3752914071083069, "learning_rate": 9.356366644908685e-05, "loss": 0.018, "step": 61540 }, { "epoch": 39.22880815806246, "grad_norm": 0.7047579884529114, "learning_rate": 9.356096111820963e-05, "loss": 0.0272, "step": 61550 }, { "epoch": 39.235181644359464, "grad_norm": 0.44205716252326965, "learning_rate": 9.355825525802559e-05, "loss": 0.0332, "step": 61560 }, { "epoch": 39.24155513065647, "grad_norm": 0.46331724524497986, "learning_rate": 9.355554886856762e-05, "loss": 0.0462, "step": 61570 }, { "epoch": 39.24792861695347, "grad_norm": 0.48321545124053955, "learning_rate": 9.355284194986862e-05, "loss": 0.0254, "step": 61580 }, { "epoch": 39.25430210325048, "grad_norm": 0.4457411766052246, "learning_rate": 9.355013450196146e-05, "loss": 0.0292, "step": 61590 }, { "epoch": 39.26067558954748, "grad_norm": 0.4657999575138092, "learning_rate": 9.354742652487907e-05, "loss": 0.0184, "step": 61600 }, { "epoch": 39.26704907584449, "grad_norm": 0.40327826142311096, "learning_rate": 9.35447180186543e-05, "loss": 0.0374, "step": 61610 }, { "epoch": 39.27342256214149, "grad_norm": 0.7968728542327881, "learning_rate": 9.354200898332012e-05, "loss": 0.0404, "step": 61620 }, { "epoch": 39.2797960484385, "grad_norm": 0.7786111235618591, "learning_rate": 9.35392994189094e-05, "loss": 0.0233, "step": 61630 }, { "epoch": 39.2861695347355, "grad_norm": 0.5833590626716614, "learning_rate": 9.35365893254551e-05, "loss": 0.0303, "step": 61640 }, { "epoch": 39.29254302103251, "grad_norm": 0.3441862165927887, "learning_rate": 9.353387870299013e-05, "loss": 0.0207, "step": 61650 }, { "epoch": 39.29891650732951, "grad_norm": 0.6298415660858154, "learning_rate": 9.353116755154742e-05, "loss": 0.0343, "step": 61660 }, { "epoch": 39.305289993626516, "grad_norm": 0.32209378480911255, "learning_rate": 9.352845587115994e-05, "loss": 0.047, "step": 61670 }, { "epoch": 39.31166347992352, "grad_norm": 0.44098782539367676, "learning_rate": 9.352574366186062e-05, "loss": 0.0229, "step": 61680 }, { "epoch": 39.318036966220525, "grad_norm": 0.4994329810142517, "learning_rate": 9.352303092368242e-05, "loss": 0.023, "step": 61690 }, { "epoch": 39.324410452517526, "grad_norm": 0.6623849868774414, "learning_rate": 9.35203176566583e-05, "loss": 0.0243, "step": 61700 }, { "epoch": 39.330783938814534, "grad_norm": 0.3024030029773712, "learning_rate": 9.351760386082123e-05, "loss": 0.0214, "step": 61710 }, { "epoch": 39.337157425111535, "grad_norm": 0.3235529065132141, "learning_rate": 9.351488953620418e-05, "loss": 0.0237, "step": 61720 }, { "epoch": 39.34353091140854, "grad_norm": 0.31255900859832764, "learning_rate": 9.351217468284014e-05, "loss": 0.0285, "step": 61730 }, { "epoch": 39.349904397705544, "grad_norm": 0.3639899790287018, "learning_rate": 9.35094593007621e-05, "loss": 0.0216, "step": 61740 }, { "epoch": 39.35627788400255, "grad_norm": 0.4521467387676239, "learning_rate": 9.350674339000303e-05, "loss": 0.0293, "step": 61750 }, { "epoch": 39.36265137029955, "grad_norm": 0.2472415268421173, "learning_rate": 9.350402695059598e-05, "loss": 0.0246, "step": 61760 }, { "epoch": 39.36902485659656, "grad_norm": 0.3704747259616852, "learning_rate": 9.350130998257391e-05, "loss": 0.0275, "step": 61770 }, { "epoch": 39.37539834289356, "grad_norm": 0.37372714281082153, "learning_rate": 9.349859248596986e-05, "loss": 0.0323, "step": 61780 }, { "epoch": 39.38177182919057, "grad_norm": 0.4915224015712738, "learning_rate": 9.349587446081684e-05, "loss": 0.0353, "step": 61790 }, { "epoch": 39.38814531548757, "grad_norm": 0.41226059198379517, "learning_rate": 9.349315590714786e-05, "loss": 0.0253, "step": 61800 }, { "epoch": 39.39451880178458, "grad_norm": 0.6475129723548889, "learning_rate": 9.3490436824996e-05, "loss": 0.0223, "step": 61810 }, { "epoch": 39.40089228808158, "grad_norm": 0.8030534982681274, "learning_rate": 9.348771721439426e-05, "loss": 0.0256, "step": 61820 }, { "epoch": 39.40726577437859, "grad_norm": 0.3132134974002838, "learning_rate": 9.34849970753757e-05, "loss": 0.0311, "step": 61830 }, { "epoch": 39.41363926067559, "grad_norm": 0.8907923102378845, "learning_rate": 9.348227640797336e-05, "loss": 0.0234, "step": 61840 }, { "epoch": 39.420012746972596, "grad_norm": 0.6327041387557983, "learning_rate": 9.347955521222032e-05, "loss": 0.0312, "step": 61850 }, { "epoch": 39.4263862332696, "grad_norm": 0.6635070443153381, "learning_rate": 9.347683348814962e-05, "loss": 0.0183, "step": 61860 }, { "epoch": 39.432759719566604, "grad_norm": 0.3856315612792969, "learning_rate": 9.347411123579435e-05, "loss": 0.0207, "step": 61870 }, { "epoch": 39.439133205863605, "grad_norm": 0.7494692802429199, "learning_rate": 9.347138845518757e-05, "loss": 0.0342, "step": 61880 }, { "epoch": 39.44550669216061, "grad_norm": 0.47494277358055115, "learning_rate": 9.34686651463624e-05, "loss": 0.03, "step": 61890 }, { "epoch": 39.451880178457614, "grad_norm": 0.602017343044281, "learning_rate": 9.346594130935189e-05, "loss": 0.0263, "step": 61900 }, { "epoch": 39.45825366475462, "grad_norm": 0.588264524936676, "learning_rate": 9.346321694418916e-05, "loss": 0.0213, "step": 61910 }, { "epoch": 39.46462715105162, "grad_norm": 0.7017486095428467, "learning_rate": 9.34604920509073e-05, "loss": 0.0276, "step": 61920 }, { "epoch": 39.47100063734863, "grad_norm": 0.6558878421783447, "learning_rate": 9.345776662953944e-05, "loss": 0.0236, "step": 61930 }, { "epoch": 39.47737412364563, "grad_norm": 0.6183075308799744, "learning_rate": 9.345504068011868e-05, "loss": 0.041, "step": 61940 }, { "epoch": 39.48374760994264, "grad_norm": 0.8397691249847412, "learning_rate": 9.345231420267814e-05, "loss": 0.0338, "step": 61950 }, { "epoch": 39.49012109623964, "grad_norm": 0.7653963565826416, "learning_rate": 9.344958719725098e-05, "loss": 0.0409, "step": 61960 }, { "epoch": 39.49649458253665, "grad_norm": 0.7158079743385315, "learning_rate": 9.34468596638703e-05, "loss": 0.03, "step": 61970 }, { "epoch": 39.50286806883365, "grad_norm": 0.3591374456882477, "learning_rate": 9.344413160256924e-05, "loss": 0.0279, "step": 61980 }, { "epoch": 39.50924155513066, "grad_norm": 0.45469602942466736, "learning_rate": 9.344140301338098e-05, "loss": 0.0186, "step": 61990 }, { "epoch": 39.51561504142766, "grad_norm": 0.44513916969299316, "learning_rate": 9.343867389633864e-05, "loss": 0.0313, "step": 62000 }, { "epoch": 39.521988527724666, "grad_norm": 0.5575897693634033, "learning_rate": 9.343594425147542e-05, "loss": 0.0249, "step": 62010 }, { "epoch": 39.52836201402167, "grad_norm": 0.4167698919773102, "learning_rate": 9.343321407882446e-05, "loss": 0.024, "step": 62020 }, { "epoch": 39.534735500318675, "grad_norm": 0.5615195035934448, "learning_rate": 9.343048337841893e-05, "loss": 0.024, "step": 62030 }, { "epoch": 39.541108986615676, "grad_norm": 0.34862884879112244, "learning_rate": 9.342775215029204e-05, "loss": 0.0166, "step": 62040 }, { "epoch": 39.547482472912684, "grad_norm": 0.3746243715286255, "learning_rate": 9.342502039447695e-05, "loss": 0.0244, "step": 62050 }, { "epoch": 39.553855959209685, "grad_norm": 0.649435818195343, "learning_rate": 9.342228811100685e-05, "loss": 0.0301, "step": 62060 }, { "epoch": 39.56022944550669, "grad_norm": 0.42625269293785095, "learning_rate": 9.341955529991497e-05, "loss": 0.0267, "step": 62070 }, { "epoch": 39.56660293180369, "grad_norm": 0.6124297976493835, "learning_rate": 9.34168219612345e-05, "loss": 0.0156, "step": 62080 }, { "epoch": 39.5729764181007, "grad_norm": 0.42768728733062744, "learning_rate": 9.341408809499863e-05, "loss": 0.0269, "step": 62090 }, { "epoch": 39.5793499043977, "grad_norm": 0.7170265316963196, "learning_rate": 9.341135370124062e-05, "loss": 0.0242, "step": 62100 }, { "epoch": 39.58572339069471, "grad_norm": 0.4428454041481018, "learning_rate": 9.340861877999367e-05, "loss": 0.0412, "step": 62110 }, { "epoch": 39.59209687699171, "grad_norm": 0.5291985869407654, "learning_rate": 9.340588333129101e-05, "loss": 0.0288, "step": 62120 }, { "epoch": 39.59847036328872, "grad_norm": 0.3868374526500702, "learning_rate": 9.340314735516588e-05, "loss": 0.0236, "step": 62130 }, { "epoch": 39.60484384958572, "grad_norm": 0.46146371960639954, "learning_rate": 9.340041085165156e-05, "loss": 0.0369, "step": 62140 }, { "epoch": 39.61121733588273, "grad_norm": 0.28697240352630615, "learning_rate": 9.339767382078125e-05, "loss": 0.0215, "step": 62150 }, { "epoch": 39.617590822179736, "grad_norm": 0.8732864856719971, "learning_rate": 9.339493626258824e-05, "loss": 0.0292, "step": 62160 }, { "epoch": 39.62396430847674, "grad_norm": 0.5142582654953003, "learning_rate": 9.339219817710576e-05, "loss": 0.0264, "step": 62170 }, { "epoch": 39.630337794773745, "grad_norm": 0.5248042345046997, "learning_rate": 9.338945956436712e-05, "loss": 0.0263, "step": 62180 }, { "epoch": 39.636711281070745, "grad_norm": 0.3663114905357361, "learning_rate": 9.338672042440559e-05, "loss": 0.0266, "step": 62190 }, { "epoch": 39.64308476736775, "grad_norm": 0.5371996760368347, "learning_rate": 9.338398075725442e-05, "loss": 0.0279, "step": 62200 }, { "epoch": 39.649458253664754, "grad_norm": 0.5203249454498291, "learning_rate": 9.338124056294694e-05, "loss": 0.0181, "step": 62210 }, { "epoch": 39.65583173996176, "grad_norm": 0.5507537722587585, "learning_rate": 9.337849984151642e-05, "loss": 0.0297, "step": 62220 }, { "epoch": 39.66220522625876, "grad_norm": 0.4123222231864929, "learning_rate": 9.337575859299617e-05, "loss": 0.0215, "step": 62230 }, { "epoch": 39.66857871255577, "grad_norm": 0.49891847372055054, "learning_rate": 9.33730168174195e-05, "loss": 0.0176, "step": 62240 }, { "epoch": 39.67495219885277, "grad_norm": 0.22477129101753235, "learning_rate": 9.337027451481973e-05, "loss": 0.0235, "step": 62250 }, { "epoch": 39.68132568514978, "grad_norm": 0.4633645713329315, "learning_rate": 9.336753168523017e-05, "loss": 0.0229, "step": 62260 }, { "epoch": 39.68769917144678, "grad_norm": 0.41032442450523376, "learning_rate": 9.336478832868416e-05, "loss": 0.0252, "step": 62270 }, { "epoch": 39.69407265774379, "grad_norm": 0.2654260993003845, "learning_rate": 9.336204444521502e-05, "loss": 0.0131, "step": 62280 }, { "epoch": 39.70044614404079, "grad_norm": 0.5611236095428467, "learning_rate": 9.335930003485612e-05, "loss": 0.0242, "step": 62290 }, { "epoch": 39.7068196303378, "grad_norm": 0.6043351888656616, "learning_rate": 9.335655509764076e-05, "loss": 0.0215, "step": 62300 }, { "epoch": 39.7131931166348, "grad_norm": 0.3117479085922241, "learning_rate": 9.335380963360233e-05, "loss": 0.0382, "step": 62310 }, { "epoch": 39.719566602931806, "grad_norm": 0.5316615700721741, "learning_rate": 9.335106364277419e-05, "loss": 0.0232, "step": 62320 }, { "epoch": 39.72594008922881, "grad_norm": 0.6081122159957886, "learning_rate": 9.334831712518969e-05, "loss": 0.0265, "step": 62330 }, { "epoch": 39.732313575525815, "grad_norm": 0.46377450227737427, "learning_rate": 9.334557008088221e-05, "loss": 0.0288, "step": 62340 }, { "epoch": 39.738687061822816, "grad_norm": 0.2957437336444855, "learning_rate": 9.334282250988511e-05, "loss": 0.0243, "step": 62350 }, { "epoch": 39.745060548119824, "grad_norm": 0.37826284766197205, "learning_rate": 9.334007441223183e-05, "loss": 0.0311, "step": 62360 }, { "epoch": 39.751434034416825, "grad_norm": 0.49150553345680237, "learning_rate": 9.333732578795571e-05, "loss": 0.0341, "step": 62370 }, { "epoch": 39.75780752071383, "grad_norm": 0.34751540422439575, "learning_rate": 9.333457663709015e-05, "loss": 0.0277, "step": 62380 }, { "epoch": 39.76418100701083, "grad_norm": 0.36657750606536865, "learning_rate": 9.333182695966858e-05, "loss": 0.0228, "step": 62390 }, { "epoch": 39.77055449330784, "grad_norm": 0.41618847846984863, "learning_rate": 9.33290767557244e-05, "loss": 0.0234, "step": 62400 }, { "epoch": 39.77692797960484, "grad_norm": 0.496868759393692, "learning_rate": 9.332632602529103e-05, "loss": 0.0253, "step": 62410 }, { "epoch": 39.78330146590185, "grad_norm": 0.40036311745643616, "learning_rate": 9.33235747684019e-05, "loss": 0.0339, "step": 62420 }, { "epoch": 39.78967495219885, "grad_norm": 0.5525063872337341, "learning_rate": 9.332082298509042e-05, "loss": 0.022, "step": 62430 }, { "epoch": 39.79604843849586, "grad_norm": 0.6169453263282776, "learning_rate": 9.331807067539005e-05, "loss": 0.0211, "step": 62440 }, { "epoch": 39.80242192479286, "grad_norm": 0.6020433306694031, "learning_rate": 9.331531783933421e-05, "loss": 0.0328, "step": 62450 }, { "epoch": 39.80879541108987, "grad_norm": 0.9072852730751038, "learning_rate": 9.331256447695635e-05, "loss": 0.0337, "step": 62460 }, { "epoch": 39.81516889738687, "grad_norm": 0.5409811735153198, "learning_rate": 9.330981058828996e-05, "loss": 0.0292, "step": 62470 }, { "epoch": 39.82154238368388, "grad_norm": 0.4983304440975189, "learning_rate": 9.330705617336848e-05, "loss": 0.0353, "step": 62480 }, { "epoch": 39.82791586998088, "grad_norm": 0.48281770944595337, "learning_rate": 9.330430123222537e-05, "loss": 0.0302, "step": 62490 }, { "epoch": 39.834289356277885, "grad_norm": 0.526093602180481, "learning_rate": 9.330154576489411e-05, "loss": 0.0225, "step": 62500 }, { "epoch": 39.840662842574886, "grad_norm": 0.568004310131073, "learning_rate": 9.32987897714082e-05, "loss": 0.0196, "step": 62510 }, { "epoch": 39.847036328871894, "grad_norm": 0.6009873747825623, "learning_rate": 9.32960332518011e-05, "loss": 0.0218, "step": 62520 }, { "epoch": 39.853409815168895, "grad_norm": 0.35112130641937256, "learning_rate": 9.329327620610633e-05, "loss": 0.0212, "step": 62530 }, { "epoch": 39.8597833014659, "grad_norm": 0.49526679515838623, "learning_rate": 9.329051863435738e-05, "loss": 0.0168, "step": 62540 }, { "epoch": 39.866156787762904, "grad_norm": 0.6273272037506104, "learning_rate": 9.328776053658774e-05, "loss": 0.0238, "step": 62550 }, { "epoch": 39.87253027405991, "grad_norm": 0.5736474394798279, "learning_rate": 9.328500191283096e-05, "loss": 0.0288, "step": 62560 }, { "epoch": 39.87890376035691, "grad_norm": 0.8673955798149109, "learning_rate": 9.328224276312052e-05, "loss": 0.0211, "step": 62570 }, { "epoch": 39.88527724665392, "grad_norm": 0.3162112534046173, "learning_rate": 9.327948308748997e-05, "loss": 0.0346, "step": 62580 }, { "epoch": 39.89165073295092, "grad_norm": 0.24026086926460266, "learning_rate": 9.327672288597285e-05, "loss": 0.0206, "step": 62590 }, { "epoch": 39.89802421924793, "grad_norm": 0.8014615774154663, "learning_rate": 9.327396215860267e-05, "loss": 0.044, "step": 62600 }, { "epoch": 39.90439770554493, "grad_norm": 0.38810208439826965, "learning_rate": 9.327120090541299e-05, "loss": 0.0269, "step": 62610 }, { "epoch": 39.91077119184194, "grad_norm": 0.34893250465393066, "learning_rate": 9.326843912643738e-05, "loss": 0.021, "step": 62620 }, { "epoch": 39.91714467813894, "grad_norm": 0.6214011311531067, "learning_rate": 9.326567682170936e-05, "loss": 0.0289, "step": 62630 }, { "epoch": 39.92351816443595, "grad_norm": 0.47269535064697266, "learning_rate": 9.326291399126255e-05, "loss": 0.016, "step": 62640 }, { "epoch": 39.92989165073295, "grad_norm": 0.5334193706512451, "learning_rate": 9.326015063513047e-05, "loss": 0.0355, "step": 62650 }, { "epoch": 39.936265137029956, "grad_norm": 0.6120707988739014, "learning_rate": 9.32573867533467e-05, "loss": 0.018, "step": 62660 }, { "epoch": 39.94263862332696, "grad_norm": 0.33862704038619995, "learning_rate": 9.325462234594486e-05, "loss": 0.0223, "step": 62670 }, { "epoch": 39.949012109623965, "grad_norm": 0.5869870185852051, "learning_rate": 9.325185741295852e-05, "loss": 0.0285, "step": 62680 }, { "epoch": 39.955385595920966, "grad_norm": 0.636843204498291, "learning_rate": 9.324909195442126e-05, "loss": 0.0246, "step": 62690 }, { "epoch": 39.961759082217974, "grad_norm": 0.5684895515441895, "learning_rate": 9.324632597036671e-05, "loss": 0.0296, "step": 62700 }, { "epoch": 39.968132568514974, "grad_norm": 0.2403639554977417, "learning_rate": 9.324355946082848e-05, "loss": 0.0166, "step": 62710 }, { "epoch": 39.97450605481198, "grad_norm": 0.2785743772983551, "learning_rate": 9.324079242584016e-05, "loss": 0.0314, "step": 62720 }, { "epoch": 39.98087954110898, "grad_norm": 0.3789971172809601, "learning_rate": 9.323802486543538e-05, "loss": 0.0148, "step": 62730 }, { "epoch": 39.98725302740599, "grad_norm": 0.5234839916229248, "learning_rate": 9.323525677964778e-05, "loss": 0.0301, "step": 62740 }, { "epoch": 39.99362651370299, "grad_norm": 0.4090384542942047, "learning_rate": 9.3232488168511e-05, "loss": 0.0157, "step": 62750 }, { "epoch": 40.0, "grad_norm": 0.5289058685302734, "learning_rate": 9.322971903205866e-05, "loss": 0.0295, "step": 62760 }, { "epoch": 40.00637348629701, "grad_norm": 0.3625575304031372, "learning_rate": 9.322694937032441e-05, "loss": 0.0412, "step": 62770 }, { "epoch": 40.01274697259401, "grad_norm": 0.3322247266769409, "learning_rate": 9.322417918334192e-05, "loss": 0.0196, "step": 62780 }, { "epoch": 40.01912045889102, "grad_norm": 0.33960163593292236, "learning_rate": 9.322140847114486e-05, "loss": 0.0202, "step": 62790 }, { "epoch": 40.02549394518802, "grad_norm": 0.8471006751060486, "learning_rate": 9.321863723376685e-05, "loss": 0.0252, "step": 62800 }, { "epoch": 40.031867431485026, "grad_norm": 0.37825411558151245, "learning_rate": 9.321586547124161e-05, "loss": 0.0203, "step": 62810 }, { "epoch": 40.038240917782026, "grad_norm": 0.36321723461151123, "learning_rate": 9.32130931836028e-05, "loss": 0.0276, "step": 62820 }, { "epoch": 40.044614404079034, "grad_norm": 0.4991523325443268, "learning_rate": 9.32103203708841e-05, "loss": 0.0347, "step": 62830 }, { "epoch": 40.050987890376035, "grad_norm": 0.6861948370933533, "learning_rate": 9.320754703311922e-05, "loss": 0.0229, "step": 62840 }, { "epoch": 40.05736137667304, "grad_norm": 0.6041613817214966, "learning_rate": 9.320477317034185e-05, "loss": 0.0253, "step": 62850 }, { "epoch": 40.063734862970044, "grad_norm": 0.3739321827888489, "learning_rate": 9.320199878258568e-05, "loss": 0.0238, "step": 62860 }, { "epoch": 40.07010834926705, "grad_norm": 0.40952154994010925, "learning_rate": 9.319922386988445e-05, "loss": 0.0262, "step": 62870 }, { "epoch": 40.07648183556405, "grad_norm": 0.31995755434036255, "learning_rate": 9.319644843227186e-05, "loss": 0.0266, "step": 62880 }, { "epoch": 40.08285532186106, "grad_norm": 0.5926610827445984, "learning_rate": 9.319367246978163e-05, "loss": 0.0265, "step": 62890 }, { "epoch": 40.08922880815806, "grad_norm": 0.3666624426841736, "learning_rate": 9.31908959824475e-05, "loss": 0.0244, "step": 62900 }, { "epoch": 40.09560229445507, "grad_norm": 0.4483577311038971, "learning_rate": 9.318811897030322e-05, "loss": 0.018, "step": 62910 }, { "epoch": 40.10197578075207, "grad_norm": 0.28789639472961426, "learning_rate": 9.318534143338249e-05, "loss": 0.0268, "step": 62920 }, { "epoch": 40.10834926704908, "grad_norm": 0.3110656142234802, "learning_rate": 9.318256337171911e-05, "loss": 0.0207, "step": 62930 }, { "epoch": 40.11472275334608, "grad_norm": 0.44846904277801514, "learning_rate": 9.317978478534681e-05, "loss": 0.0281, "step": 62940 }, { "epoch": 40.12109623964309, "grad_norm": 0.7120790481567383, "learning_rate": 9.317700567429935e-05, "loss": 0.0265, "step": 62950 }, { "epoch": 40.12746972594009, "grad_norm": 0.4908425807952881, "learning_rate": 9.31742260386105e-05, "loss": 0.0279, "step": 62960 }, { "epoch": 40.133843212237096, "grad_norm": 0.3664989173412323, "learning_rate": 9.317144587831405e-05, "loss": 0.0358, "step": 62970 }, { "epoch": 40.1402166985341, "grad_norm": 0.4467090368270874, "learning_rate": 9.316866519344378e-05, "loss": 0.0269, "step": 62980 }, { "epoch": 40.146590184831105, "grad_norm": 0.2641473114490509, "learning_rate": 9.316588398403345e-05, "loss": 0.024, "step": 62990 }, { "epoch": 40.152963671128106, "grad_norm": 0.712022066116333, "learning_rate": 9.316310225011689e-05, "loss": 0.0275, "step": 63000 }, { "epoch": 40.159337157425114, "grad_norm": 0.7244455218315125, "learning_rate": 9.316031999172788e-05, "loss": 0.0192, "step": 63010 }, { "epoch": 40.165710643722115, "grad_norm": 0.42831501364707947, "learning_rate": 9.315753720890024e-05, "loss": 0.0253, "step": 63020 }, { "epoch": 40.17208413001912, "grad_norm": 0.3438226282596588, "learning_rate": 9.315475390166777e-05, "loss": 0.0222, "step": 63030 }, { "epoch": 40.17845761631612, "grad_norm": 0.4604009985923767, "learning_rate": 9.315197007006429e-05, "loss": 0.0223, "step": 63040 }, { "epoch": 40.18483110261313, "grad_norm": 0.5738576650619507, "learning_rate": 9.314918571412363e-05, "loss": 0.0316, "step": 63050 }, { "epoch": 40.19120458891013, "grad_norm": 0.35562795400619507, "learning_rate": 9.314640083387963e-05, "loss": 0.0266, "step": 63060 }, { "epoch": 40.19757807520714, "grad_norm": 0.7249412536621094, "learning_rate": 9.314361542936613e-05, "loss": 0.0283, "step": 63070 }, { "epoch": 40.20395156150414, "grad_norm": 0.34809356927871704, "learning_rate": 9.314082950061695e-05, "loss": 0.0313, "step": 63080 }, { "epoch": 40.21032504780115, "grad_norm": 0.8902215361595154, "learning_rate": 9.313804304766598e-05, "loss": 0.0258, "step": 63090 }, { "epoch": 40.21669853409815, "grad_norm": 0.46959972381591797, "learning_rate": 9.313525607054705e-05, "loss": 0.0305, "step": 63100 }, { "epoch": 40.22307202039516, "grad_norm": 0.486623615026474, "learning_rate": 9.313246856929403e-05, "loss": 0.0303, "step": 63110 }, { "epoch": 40.22944550669216, "grad_norm": 0.4028576612472534, "learning_rate": 9.312968054394078e-05, "loss": 0.0285, "step": 63120 }, { "epoch": 40.23581899298917, "grad_norm": 0.4004812240600586, "learning_rate": 9.31268919945212e-05, "loss": 0.0261, "step": 63130 }, { "epoch": 40.24219247928617, "grad_norm": 0.5265500545501709, "learning_rate": 9.312410292106917e-05, "loss": 0.0301, "step": 63140 }, { "epoch": 40.248565965583175, "grad_norm": 0.9219235181808472, "learning_rate": 9.312131332361857e-05, "loss": 0.0224, "step": 63150 }, { "epoch": 40.254939451880176, "grad_norm": 0.5000272989273071, "learning_rate": 9.311852320220331e-05, "loss": 0.0261, "step": 63160 }, { "epoch": 40.261312938177184, "grad_norm": 0.5270541906356812, "learning_rate": 9.311573255685726e-05, "loss": 0.0336, "step": 63170 }, { "epoch": 40.267686424474185, "grad_norm": 0.4113840162754059, "learning_rate": 9.311294138761437e-05, "loss": 0.02, "step": 63180 }, { "epoch": 40.27405991077119, "grad_norm": 0.4303252696990967, "learning_rate": 9.311014969450853e-05, "loss": 0.026, "step": 63190 }, { "epoch": 40.280433397068194, "grad_norm": 0.40886190533638, "learning_rate": 9.310735747757367e-05, "loss": 0.022, "step": 63200 }, { "epoch": 40.2868068833652, "grad_norm": 0.6992054581642151, "learning_rate": 9.310456473684371e-05, "loss": 0.0365, "step": 63210 }, { "epoch": 40.2931803696622, "grad_norm": 0.5420033931732178, "learning_rate": 9.31017714723526e-05, "loss": 0.0274, "step": 63220 }, { "epoch": 40.29955385595921, "grad_norm": 0.48933863639831543, "learning_rate": 9.309897768413428e-05, "loss": 0.0308, "step": 63230 }, { "epoch": 40.30592734225621, "grad_norm": 0.3360384702682495, "learning_rate": 9.309618337222267e-05, "loss": 0.025, "step": 63240 }, { "epoch": 40.31230082855322, "grad_norm": 0.49482646584510803, "learning_rate": 9.309338853665174e-05, "loss": 0.0274, "step": 63250 }, { "epoch": 40.31867431485022, "grad_norm": 0.6703147292137146, "learning_rate": 9.309059317745547e-05, "loss": 0.0279, "step": 63260 }, { "epoch": 40.32504780114723, "grad_norm": 0.7898859977722168, "learning_rate": 9.308779729466781e-05, "loss": 0.0402, "step": 63270 }, { "epoch": 40.33142128744423, "grad_norm": 0.21609419584274292, "learning_rate": 9.308500088832272e-05, "loss": 0.0327, "step": 63280 }, { "epoch": 40.33779477374124, "grad_norm": 0.6299142241477966, "learning_rate": 9.30822039584542e-05, "loss": 0.0479, "step": 63290 }, { "epoch": 40.34416826003824, "grad_norm": 0.2556610703468323, "learning_rate": 9.307940650509621e-05, "loss": 0.0214, "step": 63300 }, { "epoch": 40.350541746335246, "grad_norm": 0.44743067026138306, "learning_rate": 9.307660852828276e-05, "loss": 0.0268, "step": 63310 }, { "epoch": 40.35691523263225, "grad_norm": 0.2864423394203186, "learning_rate": 9.307381002804785e-05, "loss": 0.0192, "step": 63320 }, { "epoch": 40.363288718929255, "grad_norm": 0.22436396777629852, "learning_rate": 9.307101100442548e-05, "loss": 0.0312, "step": 63330 }, { "epoch": 40.369662205226255, "grad_norm": 0.4444640874862671, "learning_rate": 9.306821145744967e-05, "loss": 0.025, "step": 63340 }, { "epoch": 40.37603569152326, "grad_norm": 0.3553168773651123, "learning_rate": 9.306541138715441e-05, "loss": 0.0197, "step": 63350 }, { "epoch": 40.382409177820264, "grad_norm": 0.2972046136856079, "learning_rate": 9.306261079357374e-05, "loss": 0.0299, "step": 63360 }, { "epoch": 40.38878266411727, "grad_norm": 0.4508771300315857, "learning_rate": 9.305980967674171e-05, "loss": 0.0203, "step": 63370 }, { "epoch": 40.39515615041427, "grad_norm": 0.5381671786308289, "learning_rate": 9.305700803669232e-05, "loss": 0.0399, "step": 63380 }, { "epoch": 40.40152963671128, "grad_norm": 0.5347617864608765, "learning_rate": 9.305420587345966e-05, "loss": 0.0258, "step": 63390 }, { "epoch": 40.40790312300829, "grad_norm": 0.45342910289764404, "learning_rate": 9.305140318707772e-05, "loss": 0.0238, "step": 63400 }, { "epoch": 40.41427660930529, "grad_norm": 0.5812029242515564, "learning_rate": 9.304859997758059e-05, "loss": 0.0274, "step": 63410 }, { "epoch": 40.4206500956023, "grad_norm": 0.3914230763912201, "learning_rate": 9.304579624500232e-05, "loss": 0.0181, "step": 63420 }, { "epoch": 40.4270235818993, "grad_norm": 0.6828516125679016, "learning_rate": 9.304299198937699e-05, "loss": 0.0254, "step": 63430 }, { "epoch": 40.43339706819631, "grad_norm": 0.36527177691459656, "learning_rate": 9.304018721073867e-05, "loss": 0.0268, "step": 63440 }, { "epoch": 40.43977055449331, "grad_norm": 0.6079586148262024, "learning_rate": 9.303738190912143e-05, "loss": 0.0351, "step": 63450 }, { "epoch": 40.446144040790315, "grad_norm": 0.23152729868888855, "learning_rate": 9.303457608455939e-05, "loss": 0.021, "step": 63460 }, { "epoch": 40.452517527087316, "grad_norm": 0.6136364936828613, "learning_rate": 9.30317697370866e-05, "loss": 0.0256, "step": 63470 }, { "epoch": 40.458891013384324, "grad_norm": 0.641066312789917, "learning_rate": 9.302896286673717e-05, "loss": 0.0306, "step": 63480 }, { "epoch": 40.465264499681325, "grad_norm": 0.48491403460502625, "learning_rate": 9.302615547354523e-05, "loss": 0.0219, "step": 63490 }, { "epoch": 40.47163798597833, "grad_norm": 0.8864009380340576, "learning_rate": 9.302334755754487e-05, "loss": 0.0379, "step": 63500 }, { "epoch": 40.478011472275334, "grad_norm": 0.3024265766143799, "learning_rate": 9.302053911877022e-05, "loss": 0.0302, "step": 63510 }, { "epoch": 40.48438495857234, "grad_norm": 0.3069208860397339, "learning_rate": 9.301773015725541e-05, "loss": 0.0301, "step": 63520 }, { "epoch": 40.49075844486934, "grad_norm": 0.3112746477127075, "learning_rate": 9.301492067303454e-05, "loss": 0.0377, "step": 63530 }, { "epoch": 40.49713193116635, "grad_norm": 0.3950579762458801, "learning_rate": 9.301211066614179e-05, "loss": 0.0183, "step": 63540 }, { "epoch": 40.50350541746335, "grad_norm": 0.5443411469459534, "learning_rate": 9.300930013661129e-05, "loss": 0.0257, "step": 63550 }, { "epoch": 40.50987890376036, "grad_norm": 0.3092659115791321, "learning_rate": 9.300648908447718e-05, "loss": 0.0205, "step": 63560 }, { "epoch": 40.51625239005736, "grad_norm": 0.33626043796539307, "learning_rate": 9.300367750977362e-05, "loss": 0.0238, "step": 63570 }, { "epoch": 40.52262587635437, "grad_norm": 0.6261965036392212, "learning_rate": 9.300086541253476e-05, "loss": 0.0394, "step": 63580 }, { "epoch": 40.52899936265137, "grad_norm": 0.8300643563270569, "learning_rate": 9.299805279279481e-05, "loss": 0.0298, "step": 63590 }, { "epoch": 40.53537284894838, "grad_norm": 0.40425753593444824, "learning_rate": 9.299523965058791e-05, "loss": 0.0343, "step": 63600 }, { "epoch": 40.54174633524538, "grad_norm": 0.4221994876861572, "learning_rate": 9.299242598594826e-05, "loss": 0.0225, "step": 63610 }, { "epoch": 40.548119821542386, "grad_norm": 0.4299912750720978, "learning_rate": 9.298961179891004e-05, "loss": 0.0285, "step": 63620 }, { "epoch": 40.55449330783939, "grad_norm": 0.6174660325050354, "learning_rate": 9.298679708950744e-05, "loss": 0.0374, "step": 63630 }, { "epoch": 40.560866794136395, "grad_norm": 0.5485066175460815, "learning_rate": 9.298398185777468e-05, "loss": 0.0274, "step": 63640 }, { "epoch": 40.567240280433396, "grad_norm": 0.36455947160720825, "learning_rate": 9.298116610374595e-05, "loss": 0.0238, "step": 63650 }, { "epoch": 40.5736137667304, "grad_norm": 0.7693423628807068, "learning_rate": 9.297834982745549e-05, "loss": 0.0261, "step": 63660 }, { "epoch": 40.579987253027404, "grad_norm": 0.4184761643409729, "learning_rate": 9.297553302893747e-05, "loss": 0.0276, "step": 63670 }, { "epoch": 40.58636073932441, "grad_norm": 0.3175077736377716, "learning_rate": 9.297271570822617e-05, "loss": 0.0182, "step": 63680 }, { "epoch": 40.59273422562141, "grad_norm": 0.6506162881851196, "learning_rate": 9.296989786535579e-05, "loss": 0.0291, "step": 63690 }, { "epoch": 40.59910771191842, "grad_norm": 0.515789806842804, "learning_rate": 9.296707950036059e-05, "loss": 0.0257, "step": 63700 }, { "epoch": 40.60548119821542, "grad_norm": 0.6423799395561218, "learning_rate": 9.296426061327479e-05, "loss": 0.0499, "step": 63710 }, { "epoch": 40.61185468451243, "grad_norm": 0.4790751039981842, "learning_rate": 9.296144120413267e-05, "loss": 0.0448, "step": 63720 }, { "epoch": 40.61822817080943, "grad_norm": 0.723077654838562, "learning_rate": 9.295862127296848e-05, "loss": 0.0266, "step": 63730 }, { "epoch": 40.62460165710644, "grad_norm": 0.33511605858802795, "learning_rate": 9.295580081981647e-05, "loss": 0.0229, "step": 63740 }, { "epoch": 40.63097514340344, "grad_norm": 0.3703979551792145, "learning_rate": 9.295297984471094e-05, "loss": 0.0208, "step": 63750 }, { "epoch": 40.63734862970045, "grad_norm": 0.31570249795913696, "learning_rate": 9.295015834768614e-05, "loss": 0.0165, "step": 63760 }, { "epoch": 40.64372211599745, "grad_norm": 0.3661767244338989, "learning_rate": 9.294733632877637e-05, "loss": 0.0211, "step": 63770 }, { "epoch": 40.650095602294456, "grad_norm": 0.2925328314304352, "learning_rate": 9.294451378801591e-05, "loss": 0.0198, "step": 63780 }, { "epoch": 40.65646908859146, "grad_norm": 0.35155346989631653, "learning_rate": 9.294169072543906e-05, "loss": 0.0263, "step": 63790 }, { "epoch": 40.662842574888465, "grad_norm": 0.4355350434780121, "learning_rate": 9.293886714108013e-05, "loss": 0.0298, "step": 63800 }, { "epoch": 40.669216061185466, "grad_norm": 0.5581322908401489, "learning_rate": 9.293604303497343e-05, "loss": 0.0266, "step": 63810 }, { "epoch": 40.675589547482474, "grad_norm": 0.5159093141555786, "learning_rate": 9.293321840715326e-05, "loss": 0.0361, "step": 63820 }, { "epoch": 40.681963033779475, "grad_norm": 0.3896491825580597, "learning_rate": 9.293039325765393e-05, "loss": 0.032, "step": 63830 }, { "epoch": 40.68833652007648, "grad_norm": 0.3469632863998413, "learning_rate": 9.292756758650983e-05, "loss": 0.037, "step": 63840 }, { "epoch": 40.694710006373484, "grad_norm": 0.8771584033966064, "learning_rate": 9.292474139375522e-05, "loss": 0.0369, "step": 63850 }, { "epoch": 40.70108349267049, "grad_norm": 0.9623972177505493, "learning_rate": 9.292191467942449e-05, "loss": 0.0325, "step": 63860 }, { "epoch": 40.70745697896749, "grad_norm": 0.5989234447479248, "learning_rate": 9.291908744355196e-05, "loss": 0.0237, "step": 63870 }, { "epoch": 40.7138304652645, "grad_norm": 0.48549893498420715, "learning_rate": 9.291625968617201e-05, "loss": 0.0212, "step": 63880 }, { "epoch": 40.7202039515615, "grad_norm": 0.7818636894226074, "learning_rate": 9.291343140731899e-05, "loss": 0.0239, "step": 63890 }, { "epoch": 40.72657743785851, "grad_norm": 0.6163133382797241, "learning_rate": 9.291060260702724e-05, "loss": 0.0365, "step": 63900 }, { "epoch": 40.73295092415551, "grad_norm": 0.32581809163093567, "learning_rate": 9.290777328533115e-05, "loss": 0.0213, "step": 63910 }, { "epoch": 40.73932441045252, "grad_norm": 0.33073508739471436, "learning_rate": 9.290494344226512e-05, "loss": 0.0241, "step": 63920 }, { "epoch": 40.74569789674952, "grad_norm": 0.41424766182899475, "learning_rate": 9.290211307786352e-05, "loss": 0.0278, "step": 63930 }, { "epoch": 40.75207138304653, "grad_norm": 0.4511762261390686, "learning_rate": 9.289928219216073e-05, "loss": 0.0339, "step": 63940 }, { "epoch": 40.75844486934353, "grad_norm": 0.2600332498550415, "learning_rate": 9.289645078519117e-05, "loss": 0.0246, "step": 63950 }, { "epoch": 40.764818355640536, "grad_norm": 0.44549131393432617, "learning_rate": 9.289361885698922e-05, "loss": 0.0228, "step": 63960 }, { "epoch": 40.77119184193754, "grad_norm": 0.3620595335960388, "learning_rate": 9.28907864075893e-05, "loss": 0.0245, "step": 63970 }, { "epoch": 40.777565328234544, "grad_norm": 0.6010586619377136, "learning_rate": 9.288795343702584e-05, "loss": 0.025, "step": 63980 }, { "epoch": 40.783938814531545, "grad_norm": 0.5105922222137451, "learning_rate": 9.288511994533325e-05, "loss": 0.0216, "step": 63990 }, { "epoch": 40.79031230082855, "grad_norm": 0.5473626255989075, "learning_rate": 9.288228593254597e-05, "loss": 0.0403, "step": 64000 }, { "epoch": 40.796685787125554, "grad_norm": 0.4037788212299347, "learning_rate": 9.287945139869843e-05, "loss": 0.0236, "step": 64010 }, { "epoch": 40.80305927342256, "grad_norm": 0.6704573631286621, "learning_rate": 9.287661634382505e-05, "loss": 0.0282, "step": 64020 }, { "epoch": 40.80943275971957, "grad_norm": 0.2981036603450775, "learning_rate": 9.287378076796031e-05, "loss": 0.0254, "step": 64030 }, { "epoch": 40.81580624601657, "grad_norm": 0.4433252215385437, "learning_rate": 9.287094467113865e-05, "loss": 0.02, "step": 64040 }, { "epoch": 40.82217973231358, "grad_norm": 0.4513382017612457, "learning_rate": 9.286810805339454e-05, "loss": 0.0245, "step": 64050 }, { "epoch": 40.82855321861058, "grad_norm": 0.4106238782405853, "learning_rate": 9.286527091476245e-05, "loss": 0.0233, "step": 64060 }, { "epoch": 40.83492670490759, "grad_norm": 0.36267948150634766, "learning_rate": 9.286243325527684e-05, "loss": 0.0181, "step": 64070 }, { "epoch": 40.84130019120459, "grad_norm": 0.3590107262134552, "learning_rate": 9.28595950749722e-05, "loss": 0.0201, "step": 64080 }, { "epoch": 40.8476736775016, "grad_norm": 0.4199380874633789, "learning_rate": 9.285675637388301e-05, "loss": 0.0232, "step": 64090 }, { "epoch": 40.8540471637986, "grad_norm": 0.24521631002426147, "learning_rate": 9.285391715204375e-05, "loss": 0.0213, "step": 64100 }, { "epoch": 40.860420650095605, "grad_norm": 0.5479651093482971, "learning_rate": 9.285107740948896e-05, "loss": 0.0272, "step": 64110 }, { "epoch": 40.866794136392606, "grad_norm": 0.6227468252182007, "learning_rate": 9.28482371462531e-05, "loss": 0.0263, "step": 64120 }, { "epoch": 40.873167622689614, "grad_norm": 0.5131120681762695, "learning_rate": 9.284539636237072e-05, "loss": 0.0241, "step": 64130 }, { "epoch": 40.879541108986615, "grad_norm": 0.693159818649292, "learning_rate": 9.284255505787632e-05, "loss": 0.0306, "step": 64140 }, { "epoch": 40.88591459528362, "grad_norm": 0.30935630202293396, "learning_rate": 9.283971323280442e-05, "loss": 0.0179, "step": 64150 }, { "epoch": 40.892288081580624, "grad_norm": 0.40742743015289307, "learning_rate": 9.283687088718957e-05, "loss": 0.0274, "step": 64160 }, { "epoch": 40.89866156787763, "grad_norm": 0.716603696346283, "learning_rate": 9.283402802106629e-05, "loss": 0.0334, "step": 64170 }, { "epoch": 40.90503505417463, "grad_norm": 0.4814022183418274, "learning_rate": 9.283118463446911e-05, "loss": 0.0216, "step": 64180 }, { "epoch": 40.91140854047164, "grad_norm": 0.6144996881484985, "learning_rate": 9.282834072743262e-05, "loss": 0.0347, "step": 64190 }, { "epoch": 40.91778202676864, "grad_norm": 0.48591917753219604, "learning_rate": 9.282549629999134e-05, "loss": 0.0242, "step": 64200 }, { "epoch": 40.92415551306565, "grad_norm": 0.43578240275382996, "learning_rate": 9.282265135217985e-05, "loss": 0.0348, "step": 64210 }, { "epoch": 40.93052899936265, "grad_norm": 0.8711801767349243, "learning_rate": 9.281980588403272e-05, "loss": 0.0238, "step": 64220 }, { "epoch": 40.93690248565966, "grad_norm": 0.53843754529953, "learning_rate": 9.281695989558451e-05, "loss": 0.0216, "step": 64230 }, { "epoch": 40.94327597195666, "grad_norm": 0.36928272247314453, "learning_rate": 9.281411338686982e-05, "loss": 0.0242, "step": 64240 }, { "epoch": 40.94964945825367, "grad_norm": 0.3841590881347656, "learning_rate": 9.281126635792322e-05, "loss": 0.0267, "step": 64250 }, { "epoch": 40.95602294455067, "grad_norm": 0.557015597820282, "learning_rate": 9.280841880877932e-05, "loss": 0.0183, "step": 64260 }, { "epoch": 40.962396430847676, "grad_norm": 0.6608373522758484, "learning_rate": 9.280557073947271e-05, "loss": 0.0234, "step": 64270 }, { "epoch": 40.96876991714468, "grad_norm": 0.3394624888896942, "learning_rate": 9.280272215003801e-05, "loss": 0.0179, "step": 64280 }, { "epoch": 40.975143403441685, "grad_norm": 0.41730642318725586, "learning_rate": 9.279987304050982e-05, "loss": 0.0308, "step": 64290 }, { "epoch": 40.981516889738685, "grad_norm": 0.7683445811271667, "learning_rate": 9.279702341092274e-05, "loss": 0.0417, "step": 64300 }, { "epoch": 40.98789037603569, "grad_norm": 0.3285045325756073, "learning_rate": 9.279417326131143e-05, "loss": 0.0354, "step": 64310 }, { "epoch": 40.994263862332694, "grad_norm": 0.5637664198875427, "learning_rate": 9.279132259171053e-05, "loss": 0.0273, "step": 64320 }, { "epoch": 41.0006373486297, "grad_norm": 0.5384771823883057, "learning_rate": 9.278847140215464e-05, "loss": 0.0303, "step": 64330 }, { "epoch": 41.0070108349267, "grad_norm": 0.3501203954219818, "learning_rate": 9.278561969267842e-05, "loss": 0.0258, "step": 64340 }, { "epoch": 41.01338432122371, "grad_norm": 0.37752172350883484, "learning_rate": 9.278276746331654e-05, "loss": 0.0229, "step": 64350 }, { "epoch": 41.01975780752071, "grad_norm": 0.2993519604206085, "learning_rate": 9.277991471410363e-05, "loss": 0.0384, "step": 64360 }, { "epoch": 41.02613129381772, "grad_norm": 0.3055190145969391, "learning_rate": 9.27770614450744e-05, "loss": 0.0166, "step": 64370 }, { "epoch": 41.03250478011472, "grad_norm": 0.4883407652378082, "learning_rate": 9.277420765626343e-05, "loss": 0.0233, "step": 64380 }, { "epoch": 41.03887826641173, "grad_norm": 0.7747842073440552, "learning_rate": 9.277135334770548e-05, "loss": 0.0321, "step": 64390 }, { "epoch": 41.04525175270873, "grad_norm": 0.42877596616744995, "learning_rate": 9.276849851943521e-05, "loss": 0.0291, "step": 64400 }, { "epoch": 41.05162523900574, "grad_norm": 0.41594821214675903, "learning_rate": 9.27656431714873e-05, "loss": 0.0218, "step": 64410 }, { "epoch": 41.05799872530274, "grad_norm": 0.41545432806015015, "learning_rate": 9.276278730389642e-05, "loss": 0.0182, "step": 64420 }, { "epoch": 41.064372211599746, "grad_norm": 0.6110652685165405, "learning_rate": 9.275993091669734e-05, "loss": 0.0335, "step": 64430 }, { "epoch": 41.07074569789675, "grad_norm": 0.3596121668815613, "learning_rate": 9.27570740099247e-05, "loss": 0.0159, "step": 64440 }, { "epoch": 41.077119184193755, "grad_norm": 0.6247264742851257, "learning_rate": 9.275421658361325e-05, "loss": 0.0231, "step": 64450 }, { "epoch": 41.083492670490756, "grad_norm": 0.8417141437530518, "learning_rate": 9.27513586377977e-05, "loss": 0.0276, "step": 64460 }, { "epoch": 41.089866156787764, "grad_norm": 0.3102891445159912, "learning_rate": 9.274850017251278e-05, "loss": 0.0234, "step": 64470 }, { "epoch": 41.096239643084765, "grad_norm": 0.4532087743282318, "learning_rate": 9.274564118779323e-05, "loss": 0.0283, "step": 64480 }, { "epoch": 41.10261312938177, "grad_norm": 0.7715798616409302, "learning_rate": 9.274278168367378e-05, "loss": 0.0369, "step": 64490 }, { "epoch": 41.10898661567877, "grad_norm": 0.5587839484214783, "learning_rate": 9.273992166018916e-05, "loss": 0.0241, "step": 64500 }, { "epoch": 41.11536010197578, "grad_norm": 0.5093590021133423, "learning_rate": 9.273706111737415e-05, "loss": 0.0239, "step": 64510 }, { "epoch": 41.12173358827278, "grad_norm": 0.6072325706481934, "learning_rate": 9.27342000552635e-05, "loss": 0.0251, "step": 64520 }, { "epoch": 41.12810707456979, "grad_norm": 0.21390171349048615, "learning_rate": 9.273133847389197e-05, "loss": 0.023, "step": 64530 }, { "epoch": 41.13448056086679, "grad_norm": 0.38769009709358215, "learning_rate": 9.272847637329431e-05, "loss": 0.0226, "step": 64540 }, { "epoch": 41.1408540471638, "grad_norm": 0.1946212202310562, "learning_rate": 9.272561375350535e-05, "loss": 0.026, "step": 64550 }, { "epoch": 41.1472275334608, "grad_norm": 0.4732208549976349, "learning_rate": 9.272275061455984e-05, "loss": 0.0364, "step": 64560 }, { "epoch": 41.15360101975781, "grad_norm": 0.5318770408630371, "learning_rate": 9.271988695649256e-05, "loss": 0.0289, "step": 64570 }, { "epoch": 41.15997450605481, "grad_norm": 0.4918212592601776, "learning_rate": 9.271702277933833e-05, "loss": 0.0257, "step": 64580 }, { "epoch": 41.16634799235182, "grad_norm": 0.5507063269615173, "learning_rate": 9.271415808313193e-05, "loss": 0.0422, "step": 64590 }, { "epoch": 41.17272147864882, "grad_norm": 0.4971425533294678, "learning_rate": 9.271129286790819e-05, "loss": 0.0175, "step": 64600 }, { "epoch": 41.179094964945826, "grad_norm": 0.4122069478034973, "learning_rate": 9.270842713370191e-05, "loss": 0.0222, "step": 64610 }, { "epoch": 41.185468451242826, "grad_norm": 0.5299763679504395, "learning_rate": 9.270556088054791e-05, "loss": 0.0318, "step": 64620 }, { "epoch": 41.191841937539834, "grad_norm": 0.2601962983608246, "learning_rate": 9.270269410848105e-05, "loss": 0.023, "step": 64630 }, { "epoch": 41.198215423836835, "grad_norm": 0.5905089378356934, "learning_rate": 9.269982681753612e-05, "loss": 0.0346, "step": 64640 }, { "epoch": 41.20458891013384, "grad_norm": 0.3155181407928467, "learning_rate": 9.269695900774798e-05, "loss": 0.0219, "step": 64650 }, { "epoch": 41.21096239643085, "grad_norm": 0.37132367491722107, "learning_rate": 9.269409067915147e-05, "loss": 0.0273, "step": 64660 }, { "epoch": 41.21733588272785, "grad_norm": 0.3521001636981964, "learning_rate": 9.269122183178146e-05, "loss": 0.023, "step": 64670 }, { "epoch": 41.22370936902486, "grad_norm": 0.3338051736354828, "learning_rate": 9.26883524656728e-05, "loss": 0.0226, "step": 64680 }, { "epoch": 41.23008285532186, "grad_norm": 0.3368802070617676, "learning_rate": 9.268548258086033e-05, "loss": 0.02, "step": 64690 }, { "epoch": 41.23645634161887, "grad_norm": 0.5896616578102112, "learning_rate": 9.268261217737897e-05, "loss": 0.0262, "step": 64700 }, { "epoch": 41.24282982791587, "grad_norm": 0.669954776763916, "learning_rate": 9.267974125526356e-05, "loss": 0.0475, "step": 64710 }, { "epoch": 41.24920331421288, "grad_norm": 0.32509857416152954, "learning_rate": 9.2676869814549e-05, "loss": 0.0276, "step": 64720 }, { "epoch": 41.25557680050988, "grad_norm": 0.3558478057384491, "learning_rate": 9.267399785527017e-05, "loss": 0.0192, "step": 64730 }, { "epoch": 41.261950286806886, "grad_norm": 0.6734727621078491, "learning_rate": 9.2671125377462e-05, "loss": 0.0325, "step": 64740 }, { "epoch": 41.26832377310389, "grad_norm": 0.6073101758956909, "learning_rate": 9.266825238115935e-05, "loss": 0.0324, "step": 64750 }, { "epoch": 41.274697259400895, "grad_norm": 0.4673481583595276, "learning_rate": 9.266537886639714e-05, "loss": 0.0277, "step": 64760 }, { "epoch": 41.281070745697896, "grad_norm": 0.39584973454475403, "learning_rate": 9.266250483321031e-05, "loss": 0.0234, "step": 64770 }, { "epoch": 41.287444231994904, "grad_norm": 0.5112573504447937, "learning_rate": 9.265963028163376e-05, "loss": 0.0215, "step": 64780 }, { "epoch": 41.293817718291905, "grad_norm": 0.5653212070465088, "learning_rate": 9.265675521170243e-05, "loss": 0.0209, "step": 64790 }, { "epoch": 41.30019120458891, "grad_norm": 0.5304442644119263, "learning_rate": 9.265387962345125e-05, "loss": 0.034, "step": 64800 }, { "epoch": 41.306564690885914, "grad_norm": 0.24975214898586273, "learning_rate": 9.265100351691516e-05, "loss": 0.0349, "step": 64810 }, { "epoch": 41.31293817718292, "grad_norm": 0.33544230461120605, "learning_rate": 9.26481268921291e-05, "loss": 0.0205, "step": 64820 }, { "epoch": 41.31931166347992, "grad_norm": 0.4971437454223633, "learning_rate": 9.264524974912803e-05, "loss": 0.0274, "step": 64830 }, { "epoch": 41.32568514977693, "grad_norm": 0.44031015038490295, "learning_rate": 9.264237208794692e-05, "loss": 0.0177, "step": 64840 }, { "epoch": 41.33205863607393, "grad_norm": 0.4225476086139679, "learning_rate": 9.263949390862073e-05, "loss": 0.0391, "step": 64850 }, { "epoch": 41.33843212237094, "grad_norm": 0.592457115650177, "learning_rate": 9.263661521118442e-05, "loss": 0.0203, "step": 64860 }, { "epoch": 41.34480560866794, "grad_norm": 0.7723691463470459, "learning_rate": 9.263373599567298e-05, "loss": 0.0218, "step": 64870 }, { "epoch": 41.35117909496495, "grad_norm": 0.6781921982765198, "learning_rate": 9.26308562621214e-05, "loss": 0.0384, "step": 64880 }, { "epoch": 41.35755258126195, "grad_norm": 0.3004898130893707, "learning_rate": 9.262797601056468e-05, "loss": 0.0176, "step": 64890 }, { "epoch": 41.36392606755896, "grad_norm": 0.902631402015686, "learning_rate": 9.262509524103778e-05, "loss": 0.0276, "step": 64900 }, { "epoch": 41.37029955385596, "grad_norm": 0.5635977387428284, "learning_rate": 9.262221395357574e-05, "loss": 0.0182, "step": 64910 }, { "epoch": 41.376673040152966, "grad_norm": 0.7531383633613586, "learning_rate": 9.261933214821357e-05, "loss": 0.0336, "step": 64920 }, { "epoch": 41.38304652644997, "grad_norm": 0.32179540395736694, "learning_rate": 9.261644982498626e-05, "loss": 0.0317, "step": 64930 }, { "epoch": 41.389420012746974, "grad_norm": 0.3562071919441223, "learning_rate": 9.261356698392884e-05, "loss": 0.0225, "step": 64940 }, { "epoch": 41.395793499043975, "grad_norm": 0.5225604772567749, "learning_rate": 9.261068362507636e-05, "loss": 0.0227, "step": 64950 }, { "epoch": 41.40216698534098, "grad_norm": 0.5845931172370911, "learning_rate": 9.260779974846383e-05, "loss": 0.03, "step": 64960 }, { "epoch": 41.408540471637984, "grad_norm": 0.5382035374641418, "learning_rate": 9.260491535412631e-05, "loss": 0.0305, "step": 64970 }, { "epoch": 41.41491395793499, "grad_norm": 0.3906625807285309, "learning_rate": 9.260203044209885e-05, "loss": 0.0401, "step": 64980 }, { "epoch": 41.42128744423199, "grad_norm": 0.4308006763458252, "learning_rate": 9.259914501241652e-05, "loss": 0.0248, "step": 64990 }, { "epoch": 41.427660930529, "grad_norm": 0.5970350503921509, "learning_rate": 9.259625906511431e-05, "loss": 0.0293, "step": 65000 }, { "epoch": 41.434034416826, "grad_norm": 0.3551969528198242, "learning_rate": 9.259337260022736e-05, "loss": 0.0158, "step": 65010 }, { "epoch": 41.44040790312301, "grad_norm": 0.4995523691177368, "learning_rate": 9.259048561779071e-05, "loss": 0.0362, "step": 65020 }, { "epoch": 41.44678138942001, "grad_norm": 0.6094891428947449, "learning_rate": 9.258759811783946e-05, "loss": 0.0245, "step": 65030 }, { "epoch": 41.45315487571702, "grad_norm": 0.5610607862472534, "learning_rate": 9.258471010040868e-05, "loss": 0.0286, "step": 65040 }, { "epoch": 41.45952836201402, "grad_norm": 0.6460846066474915, "learning_rate": 9.258182156553346e-05, "loss": 0.0372, "step": 65050 }, { "epoch": 41.46590184831103, "grad_norm": 0.46597814559936523, "learning_rate": 9.257893251324892e-05, "loss": 0.0313, "step": 65060 }, { "epoch": 41.47227533460803, "grad_norm": 0.5409643650054932, "learning_rate": 9.257604294359012e-05, "loss": 0.0208, "step": 65070 }, { "epoch": 41.478648820905036, "grad_norm": 0.7999165058135986, "learning_rate": 9.257315285659222e-05, "loss": 0.0296, "step": 65080 }, { "epoch": 41.48502230720204, "grad_norm": 0.9311299324035645, "learning_rate": 9.257026225229031e-05, "loss": 0.0169, "step": 65090 }, { "epoch": 41.491395793499045, "grad_norm": 0.5974019169807434, "learning_rate": 9.256737113071951e-05, "loss": 0.0233, "step": 65100 }, { "epoch": 41.497769279796046, "grad_norm": 0.5159413814544678, "learning_rate": 9.256447949191498e-05, "loss": 0.0229, "step": 65110 }, { "epoch": 41.504142766093054, "grad_norm": 0.3097835183143616, "learning_rate": 9.256158733591184e-05, "loss": 0.0214, "step": 65120 }, { "epoch": 41.510516252390055, "grad_norm": 0.31960171461105347, "learning_rate": 9.255869466274522e-05, "loss": 0.0238, "step": 65130 }, { "epoch": 41.51688973868706, "grad_norm": 0.3472208082675934, "learning_rate": 9.255580147245027e-05, "loss": 0.0298, "step": 65140 }, { "epoch": 41.52326322498406, "grad_norm": 0.5966345071792603, "learning_rate": 9.255290776506215e-05, "loss": 0.0417, "step": 65150 }, { "epoch": 41.52963671128107, "grad_norm": 0.3870909810066223, "learning_rate": 9.255001354061604e-05, "loss": 0.0218, "step": 65160 }, { "epoch": 41.53601019757807, "grad_norm": 0.3882139027118683, "learning_rate": 9.25471187991471e-05, "loss": 0.0184, "step": 65170 }, { "epoch": 41.54238368387508, "grad_norm": 0.38385850191116333, "learning_rate": 9.254422354069048e-05, "loss": 0.0157, "step": 65180 }, { "epoch": 41.54875717017208, "grad_norm": 0.6244174242019653, "learning_rate": 9.254132776528137e-05, "loss": 0.0237, "step": 65190 }, { "epoch": 41.55513065646909, "grad_norm": 0.5376093983650208, "learning_rate": 9.253843147295497e-05, "loss": 0.0245, "step": 65200 }, { "epoch": 41.56150414276609, "grad_norm": 0.7808029651641846, "learning_rate": 9.253553466374646e-05, "loss": 0.0274, "step": 65210 }, { "epoch": 41.5678776290631, "grad_norm": 0.45590806007385254, "learning_rate": 9.253263733769105e-05, "loss": 0.0239, "step": 65220 }, { "epoch": 41.5742511153601, "grad_norm": 0.46953079104423523, "learning_rate": 9.252973949482393e-05, "loss": 0.0236, "step": 65230 }, { "epoch": 41.58062460165711, "grad_norm": 0.35137003660202026, "learning_rate": 9.252684113518033e-05, "loss": 0.0287, "step": 65240 }, { "epoch": 41.58699808795411, "grad_norm": 0.32690155506134033, "learning_rate": 9.252394225879545e-05, "loss": 0.0216, "step": 65250 }, { "epoch": 41.593371574251115, "grad_norm": 0.40403011441230774, "learning_rate": 9.252104286570453e-05, "loss": 0.0235, "step": 65260 }, { "epoch": 41.599745060548116, "grad_norm": 0.3692602217197418, "learning_rate": 9.251814295594279e-05, "loss": 0.0266, "step": 65270 }, { "epoch": 41.606118546845124, "grad_norm": 0.35846632719039917, "learning_rate": 9.251524252954547e-05, "loss": 0.0305, "step": 65280 }, { "epoch": 41.61249203314213, "grad_norm": 0.4267905652523041, "learning_rate": 9.25123415865478e-05, "loss": 0.0196, "step": 65290 }, { "epoch": 41.61886551943913, "grad_norm": 0.4155419170856476, "learning_rate": 9.250944012698506e-05, "loss": 0.0282, "step": 65300 }, { "epoch": 41.62523900573614, "grad_norm": 0.6177881956100464, "learning_rate": 9.250653815089247e-05, "loss": 0.0211, "step": 65310 }, { "epoch": 41.63161249203314, "grad_norm": 0.6805815696716309, "learning_rate": 9.250363565830532e-05, "loss": 0.0293, "step": 65320 }, { "epoch": 41.63798597833015, "grad_norm": 0.2599300444126129, "learning_rate": 9.250073264925887e-05, "loss": 0.0296, "step": 65330 }, { "epoch": 41.64435946462715, "grad_norm": 0.6139668822288513, "learning_rate": 9.249782912378838e-05, "loss": 0.03, "step": 65340 }, { "epoch": 41.65073295092416, "grad_norm": 0.4658694863319397, "learning_rate": 9.249492508192914e-05, "loss": 0.0316, "step": 65350 }, { "epoch": 41.65710643722116, "grad_norm": 0.3623603582382202, "learning_rate": 9.249202052371646e-05, "loss": 0.0196, "step": 65360 }, { "epoch": 41.66347992351817, "grad_norm": 0.5396273136138916, "learning_rate": 9.248911544918559e-05, "loss": 0.0211, "step": 65370 }, { "epoch": 41.66985340981517, "grad_norm": 0.37033963203430176, "learning_rate": 9.248620985837186e-05, "loss": 0.017, "step": 65380 }, { "epoch": 41.676226896112176, "grad_norm": 0.6679165959358215, "learning_rate": 9.248330375131055e-05, "loss": 0.0181, "step": 65390 }, { "epoch": 41.68260038240918, "grad_norm": 0.26984667778015137, "learning_rate": 9.248039712803701e-05, "loss": 0.0303, "step": 65400 }, { "epoch": 41.688973868706185, "grad_norm": 0.7891584634780884, "learning_rate": 9.247748998858653e-05, "loss": 0.0221, "step": 65410 }, { "epoch": 41.695347355003186, "grad_norm": 0.5628881454467773, "learning_rate": 9.247458233299444e-05, "loss": 0.0239, "step": 65420 }, { "epoch": 41.701720841300194, "grad_norm": 0.3488200008869171, "learning_rate": 9.247167416129607e-05, "loss": 0.0327, "step": 65430 }, { "epoch": 41.708094327597195, "grad_norm": 0.6590137481689453, "learning_rate": 9.246876547352676e-05, "loss": 0.027, "step": 65440 }, { "epoch": 41.7144678138942, "grad_norm": 0.6593069434165955, "learning_rate": 9.246585626972186e-05, "loss": 0.027, "step": 65450 }, { "epoch": 41.7208413001912, "grad_norm": 0.4028383195400238, "learning_rate": 9.24629465499167e-05, "loss": 0.0227, "step": 65460 }, { "epoch": 41.72721478648821, "grad_norm": 0.39276933670043945, "learning_rate": 9.246003631414666e-05, "loss": 0.0203, "step": 65470 }, { "epoch": 41.73358827278521, "grad_norm": 0.7703338265419006, "learning_rate": 9.245712556244708e-05, "loss": 0.0277, "step": 65480 }, { "epoch": 41.73996175908222, "grad_norm": 0.5041046738624573, "learning_rate": 9.245421429485335e-05, "loss": 0.0189, "step": 65490 }, { "epoch": 41.74633524537922, "grad_norm": 0.6143819093704224, "learning_rate": 9.245130251140082e-05, "loss": 0.0292, "step": 65500 }, { "epoch": 41.75270873167623, "grad_norm": 0.6391715407371521, "learning_rate": 9.244839021212488e-05, "loss": 0.023, "step": 65510 }, { "epoch": 41.75908221797323, "grad_norm": 0.32556939125061035, "learning_rate": 9.244547739706093e-05, "loss": 0.0311, "step": 65520 }, { "epoch": 41.76545570427024, "grad_norm": 0.3450542092323303, "learning_rate": 9.244256406624436e-05, "loss": 0.042, "step": 65530 }, { "epoch": 41.77182919056724, "grad_norm": 0.8907830119132996, "learning_rate": 9.243965021971057e-05, "loss": 0.0268, "step": 65540 }, { "epoch": 41.77820267686425, "grad_norm": 0.33248305320739746, "learning_rate": 9.243673585749494e-05, "loss": 0.0231, "step": 65550 }, { "epoch": 41.78457616316125, "grad_norm": 0.3961670994758606, "learning_rate": 9.243382097963291e-05, "loss": 0.0179, "step": 65560 }, { "epoch": 41.790949649458256, "grad_norm": 0.17833109200000763, "learning_rate": 9.243090558615989e-05, "loss": 0.0283, "step": 65570 }, { "epoch": 41.797323135755256, "grad_norm": 0.35153061151504517, "learning_rate": 9.24279896771113e-05, "loss": 0.0232, "step": 65580 }, { "epoch": 41.803696622052264, "grad_norm": 0.3453674614429474, "learning_rate": 9.242507325252259e-05, "loss": 0.0363, "step": 65590 }, { "epoch": 41.810070108349265, "grad_norm": 0.5939841866493225, "learning_rate": 9.242215631242919e-05, "loss": 0.0253, "step": 65600 }, { "epoch": 41.81644359464627, "grad_norm": 0.3945414423942566, "learning_rate": 9.241923885686652e-05, "loss": 0.0222, "step": 65610 }, { "epoch": 41.822817080943274, "grad_norm": 0.6010554432868958, "learning_rate": 9.241632088587004e-05, "loss": 0.0347, "step": 65620 }, { "epoch": 41.82919056724028, "grad_norm": 0.5998820662498474, "learning_rate": 9.241340239947523e-05, "loss": 0.0249, "step": 65630 }, { "epoch": 41.83556405353728, "grad_norm": 0.4723265469074249, "learning_rate": 9.241048339771754e-05, "loss": 0.047, "step": 65640 }, { "epoch": 41.84193753983429, "grad_norm": 0.5665018558502197, "learning_rate": 9.240756388063242e-05, "loss": 0.0292, "step": 65650 }, { "epoch": 41.84831102613129, "grad_norm": 0.8665750622749329, "learning_rate": 9.240464384825538e-05, "loss": 0.0233, "step": 65660 }, { "epoch": 41.8546845124283, "grad_norm": 0.642570972442627, "learning_rate": 9.240172330062186e-05, "loss": 0.0242, "step": 65670 }, { "epoch": 41.8610579987253, "grad_norm": 0.43970707058906555, "learning_rate": 9.239880223776738e-05, "loss": 0.0239, "step": 65680 }, { "epoch": 41.86743148502231, "grad_norm": 0.6904773116111755, "learning_rate": 9.239588065972742e-05, "loss": 0.0288, "step": 65690 }, { "epoch": 41.87380497131931, "grad_norm": 0.3054540455341339, "learning_rate": 9.239295856653747e-05, "loss": 0.0205, "step": 65700 }, { "epoch": 41.88017845761632, "grad_norm": 0.45055291056632996, "learning_rate": 9.239003595823306e-05, "loss": 0.0181, "step": 65710 }, { "epoch": 41.88655194391332, "grad_norm": 1.852647304534912, "learning_rate": 9.238711283484968e-05, "loss": 0.0327, "step": 65720 }, { "epoch": 41.892925430210326, "grad_norm": 0.48864656686782837, "learning_rate": 9.238418919642287e-05, "loss": 0.0176, "step": 65730 }, { "epoch": 41.89929891650733, "grad_norm": 0.3609170913696289, "learning_rate": 9.238126504298814e-05, "loss": 0.0157, "step": 65740 }, { "epoch": 41.905672402804335, "grad_norm": 0.6786608099937439, "learning_rate": 9.237834037458102e-05, "loss": 0.0294, "step": 65750 }, { "epoch": 41.912045889101336, "grad_norm": 0.4320574104785919, "learning_rate": 9.237541519123707e-05, "loss": 0.0246, "step": 65760 }, { "epoch": 41.918419375398344, "grad_norm": 0.4893729090690613, "learning_rate": 9.23724894929918e-05, "loss": 0.0262, "step": 65770 }, { "epoch": 41.924792861695344, "grad_norm": 0.5465174913406372, "learning_rate": 9.236956327988078e-05, "loss": 0.0182, "step": 65780 }, { "epoch": 41.93116634799235, "grad_norm": 0.3505243957042694, "learning_rate": 9.236663655193955e-05, "loss": 0.0195, "step": 65790 }, { "epoch": 41.93753983428935, "grad_norm": 1.6846966743469238, "learning_rate": 9.236370930920371e-05, "loss": 0.033, "step": 65800 }, { "epoch": 41.94391332058636, "grad_norm": 0.51729816198349, "learning_rate": 9.23607815517088e-05, "loss": 0.0147, "step": 65810 }, { "epoch": 41.95028680688336, "grad_norm": 0.4280257225036621, "learning_rate": 9.235785327949038e-05, "loss": 0.0203, "step": 65820 }, { "epoch": 41.95666029318037, "grad_norm": 0.2931711673736572, "learning_rate": 9.235492449258407e-05, "loss": 0.0199, "step": 65830 }, { "epoch": 41.96303377947737, "grad_norm": 0.32455819845199585, "learning_rate": 9.235199519102542e-05, "loss": 0.0178, "step": 65840 }, { "epoch": 41.96940726577438, "grad_norm": 0.44798725843429565, "learning_rate": 9.234906537485006e-05, "loss": 0.0244, "step": 65850 }, { "epoch": 41.97578075207138, "grad_norm": 0.5284509062767029, "learning_rate": 9.234613504409356e-05, "loss": 0.0328, "step": 65860 }, { "epoch": 41.98215423836839, "grad_norm": 0.3808888792991638, "learning_rate": 9.234320419879154e-05, "loss": 0.0226, "step": 65870 }, { "epoch": 41.98852772466539, "grad_norm": 0.5114511847496033, "learning_rate": 9.234027283897961e-05, "loss": 0.0318, "step": 65880 }, { "epoch": 41.9949012109624, "grad_norm": 0.433860719203949, "learning_rate": 9.233734096469339e-05, "loss": 0.0238, "step": 65890 }, { "epoch": 42.001274697259404, "grad_norm": 0.49082469940185547, "learning_rate": 9.23344085759685e-05, "loss": 0.0243, "step": 65900 }, { "epoch": 42.007648183556405, "grad_norm": 0.6280624270439148, "learning_rate": 9.233147567284057e-05, "loss": 0.0252, "step": 65910 }, { "epoch": 42.01402166985341, "grad_norm": 0.47607970237731934, "learning_rate": 9.232854225534527e-05, "loss": 0.0285, "step": 65920 }, { "epoch": 42.020395156150414, "grad_norm": 0.35698407888412476, "learning_rate": 9.23256083235182e-05, "loss": 0.0231, "step": 65930 }, { "epoch": 42.02676864244742, "grad_norm": 0.4576590061187744, "learning_rate": 9.232267387739502e-05, "loss": 0.0315, "step": 65940 }, { "epoch": 42.03314212874442, "grad_norm": 0.3582839071750641, "learning_rate": 9.23197389170114e-05, "loss": 0.0185, "step": 65950 }, { "epoch": 42.03951561504143, "grad_norm": 0.4310932159423828, "learning_rate": 9.231680344240298e-05, "loss": 0.0345, "step": 65960 }, { "epoch": 42.04588910133843, "grad_norm": 0.5213659405708313, "learning_rate": 9.231386745360546e-05, "loss": 0.0269, "step": 65970 }, { "epoch": 42.05226258763544, "grad_norm": 0.414416640996933, "learning_rate": 9.231093095065451e-05, "loss": 0.029, "step": 65980 }, { "epoch": 42.05863607393244, "grad_norm": 0.4381742477416992, "learning_rate": 9.23079939335858e-05, "loss": 0.0244, "step": 65990 }, { "epoch": 42.06500956022945, "grad_norm": 0.5713891983032227, "learning_rate": 9.2305056402435e-05, "loss": 0.0272, "step": 66000 }, { "epoch": 42.07138304652645, "grad_norm": 0.6084326505661011, "learning_rate": 9.230211835723783e-05, "loss": 0.0204, "step": 66010 }, { "epoch": 42.07775653282346, "grad_norm": 0.6460614800453186, "learning_rate": 9.229917979802997e-05, "loss": 0.025, "step": 66020 }, { "epoch": 42.08413001912046, "grad_norm": 0.31622523069381714, "learning_rate": 9.229624072484716e-05, "loss": 0.022, "step": 66030 }, { "epoch": 42.090503505417466, "grad_norm": 0.355973482131958, "learning_rate": 9.229330113772507e-05, "loss": 0.0181, "step": 66040 }, { "epoch": 42.09687699171447, "grad_norm": 0.4074205458164215, "learning_rate": 9.229036103669946e-05, "loss": 0.0253, "step": 66050 }, { "epoch": 42.103250478011475, "grad_norm": 0.5366103053092957, "learning_rate": 9.228742042180603e-05, "loss": 0.0278, "step": 66060 }, { "epoch": 42.109623964308476, "grad_norm": 0.39441221952438354, "learning_rate": 9.228447929308051e-05, "loss": 0.0268, "step": 66070 }, { "epoch": 42.115997450605484, "grad_norm": 0.5194296836853027, "learning_rate": 9.228153765055864e-05, "loss": 0.0229, "step": 66080 }, { "epoch": 42.122370936902485, "grad_norm": 0.43909651041030884, "learning_rate": 9.227859549427618e-05, "loss": 0.0237, "step": 66090 }, { "epoch": 42.12874442319949, "grad_norm": 0.40062376856803894, "learning_rate": 9.227565282426887e-05, "loss": 0.0266, "step": 66100 }, { "epoch": 42.13511790949649, "grad_norm": 0.5049161314964294, "learning_rate": 9.227270964057245e-05, "loss": 0.0309, "step": 66110 }, { "epoch": 42.1414913957935, "grad_norm": 0.29995375871658325, "learning_rate": 9.22697659432227e-05, "loss": 0.0167, "step": 66120 }, { "epoch": 42.1478648820905, "grad_norm": 0.5678418874740601, "learning_rate": 9.226682173225537e-05, "loss": 0.0207, "step": 66130 }, { "epoch": 42.15423836838751, "grad_norm": 0.3566882312297821, "learning_rate": 9.226387700770626e-05, "loss": 0.0143, "step": 66140 }, { "epoch": 42.16061185468451, "grad_norm": 0.5927200317382812, "learning_rate": 9.226093176961115e-05, "loss": 0.0182, "step": 66150 }, { "epoch": 42.16698534098152, "grad_norm": 0.33880048990249634, "learning_rate": 9.225798601800582e-05, "loss": 0.0249, "step": 66160 }, { "epoch": 42.17335882727852, "grad_norm": 0.44698598980903625, "learning_rate": 9.225503975292604e-05, "loss": 0.0232, "step": 66170 }, { "epoch": 42.17973231357553, "grad_norm": 0.4112945795059204, "learning_rate": 9.225209297440764e-05, "loss": 0.0248, "step": 66180 }, { "epoch": 42.18610579987253, "grad_norm": 0.46955057978630066, "learning_rate": 9.224914568248644e-05, "loss": 0.0387, "step": 66190 }, { "epoch": 42.19247928616954, "grad_norm": 0.6163523197174072, "learning_rate": 9.22461978771982e-05, "loss": 0.0165, "step": 66200 }, { "epoch": 42.19885277246654, "grad_norm": 0.40762218832969666, "learning_rate": 9.224324955857879e-05, "loss": 0.0206, "step": 66210 }, { "epoch": 42.205226258763545, "grad_norm": 0.2340753972530365, "learning_rate": 9.224030072666401e-05, "loss": 0.018, "step": 66220 }, { "epoch": 42.211599745060546, "grad_norm": 0.25114962458610535, "learning_rate": 9.223735138148968e-05, "loss": 0.0246, "step": 66230 }, { "epoch": 42.217973231357554, "grad_norm": 0.4387468993663788, "learning_rate": 9.223440152309166e-05, "loss": 0.0235, "step": 66240 }, { "epoch": 42.224346717654555, "grad_norm": 0.37101712822914124, "learning_rate": 9.22314511515058e-05, "loss": 0.021, "step": 66250 }, { "epoch": 42.23072020395156, "grad_norm": 0.3785618543624878, "learning_rate": 9.222850026676791e-05, "loss": 0.0275, "step": 66260 }, { "epoch": 42.237093690248564, "grad_norm": 0.4702027440071106, "learning_rate": 9.222554886891389e-05, "loss": 0.0241, "step": 66270 }, { "epoch": 42.24346717654557, "grad_norm": 0.33600151538848877, "learning_rate": 9.222259695797958e-05, "loss": 0.0224, "step": 66280 }, { "epoch": 42.24984066284257, "grad_norm": 0.47428736090660095, "learning_rate": 9.221964453400084e-05, "loss": 0.0267, "step": 66290 }, { "epoch": 42.25621414913958, "grad_norm": 0.660992443561554, "learning_rate": 9.221669159701358e-05, "loss": 0.0322, "step": 66300 }, { "epoch": 42.26258763543658, "grad_norm": 0.336344450712204, "learning_rate": 9.221373814705365e-05, "loss": 0.0254, "step": 66310 }, { "epoch": 42.26896112173359, "grad_norm": 0.47817614674568176, "learning_rate": 9.221078418415693e-05, "loss": 0.0269, "step": 66320 }, { "epoch": 42.27533460803059, "grad_norm": 0.6063815355300903, "learning_rate": 9.220782970835934e-05, "loss": 0.0244, "step": 66330 }, { "epoch": 42.2817080943276, "grad_norm": 0.44965827465057373, "learning_rate": 9.220487471969677e-05, "loss": 0.0296, "step": 66340 }, { "epoch": 42.2880815806246, "grad_norm": 0.455697238445282, "learning_rate": 9.220191921820511e-05, "loss": 0.0261, "step": 66350 }, { "epoch": 42.29445506692161, "grad_norm": 0.4224787652492523, "learning_rate": 9.219896320392031e-05, "loss": 0.0188, "step": 66360 }, { "epoch": 42.30082855321861, "grad_norm": 0.42280644178390503, "learning_rate": 9.219600667687824e-05, "loss": 0.0191, "step": 66370 }, { "epoch": 42.307202039515616, "grad_norm": 0.6243449449539185, "learning_rate": 9.219304963711487e-05, "loss": 0.0292, "step": 66380 }, { "epoch": 42.31357552581262, "grad_norm": 0.5976722836494446, "learning_rate": 9.21900920846661e-05, "loss": 0.0279, "step": 66390 }, { "epoch": 42.319949012109625, "grad_norm": 0.6923816204071045, "learning_rate": 9.218713401956788e-05, "loss": 0.0312, "step": 66400 }, { "epoch": 42.326322498406626, "grad_norm": 0.37685689330101013, "learning_rate": 9.218417544185615e-05, "loss": 0.0238, "step": 66410 }, { "epoch": 42.33269598470363, "grad_norm": 0.31143316626548767, "learning_rate": 9.218121635156686e-05, "loss": 0.021, "step": 66420 }, { "epoch": 42.339069471000634, "grad_norm": 0.7185904383659363, "learning_rate": 9.217825674873597e-05, "loss": 0.0335, "step": 66430 }, { "epoch": 42.34544295729764, "grad_norm": 0.30323293805122375, "learning_rate": 9.217529663339942e-05, "loss": 0.0205, "step": 66440 }, { "epoch": 42.35181644359464, "grad_norm": 0.4969540238380432, "learning_rate": 9.21723360055932e-05, "loss": 0.0245, "step": 66450 }, { "epoch": 42.35818992989165, "grad_norm": 0.48340144753456116, "learning_rate": 9.216937486535331e-05, "loss": 0.0251, "step": 66460 }, { "epoch": 42.36456341618865, "grad_norm": 0.3917814791202545, "learning_rate": 9.216641321271567e-05, "loss": 0.0205, "step": 66470 }, { "epoch": 42.37093690248566, "grad_norm": 0.6298745274543762, "learning_rate": 9.216345104771631e-05, "loss": 0.0307, "step": 66480 }, { "epoch": 42.37731038878266, "grad_norm": 0.4075070321559906, "learning_rate": 9.21604883703912e-05, "loss": 0.0218, "step": 66490 }, { "epoch": 42.38368387507967, "grad_norm": 0.44286903738975525, "learning_rate": 9.215752518077635e-05, "loss": 0.0298, "step": 66500 }, { "epoch": 42.39005736137667, "grad_norm": 0.4755772352218628, "learning_rate": 9.215456147890779e-05, "loss": 0.0311, "step": 66510 }, { "epoch": 42.39643084767368, "grad_norm": 0.5091001391410828, "learning_rate": 9.215159726482148e-05, "loss": 0.0169, "step": 66520 }, { "epoch": 42.40280433397068, "grad_norm": 0.4735628068447113, "learning_rate": 9.214863253855348e-05, "loss": 0.0214, "step": 66530 }, { "epoch": 42.409177820267686, "grad_norm": 0.42597267031669617, "learning_rate": 9.214566730013981e-05, "loss": 0.0291, "step": 66540 }, { "epoch": 42.415551306564694, "grad_norm": 0.8328556418418884, "learning_rate": 9.214270154961647e-05, "loss": 0.0214, "step": 66550 }, { "epoch": 42.421924792861695, "grad_norm": 0.4224740266799927, "learning_rate": 9.213973528701952e-05, "loss": 0.0208, "step": 66560 }, { "epoch": 42.4282982791587, "grad_norm": 0.5940009355545044, "learning_rate": 9.213676851238501e-05, "loss": 0.0201, "step": 66570 }, { "epoch": 42.434671765455704, "grad_norm": 0.3207134008407593, "learning_rate": 9.213380122574898e-05, "loss": 0.0259, "step": 66580 }, { "epoch": 42.44104525175271, "grad_norm": 0.1958329826593399, "learning_rate": 9.213083342714747e-05, "loss": 0.025, "step": 66590 }, { "epoch": 42.44741873804971, "grad_norm": 0.668378472328186, "learning_rate": 9.212786511661656e-05, "loss": 0.0269, "step": 66600 }, { "epoch": 42.45379222434672, "grad_norm": 0.4601955711841583, "learning_rate": 9.212489629419232e-05, "loss": 0.0323, "step": 66610 }, { "epoch": 42.46016571064372, "grad_norm": 0.3183055520057678, "learning_rate": 9.212192695991082e-05, "loss": 0.028, "step": 66620 }, { "epoch": 42.46653919694073, "grad_norm": 0.23550714552402496, "learning_rate": 9.211895711380812e-05, "loss": 0.0236, "step": 66630 }, { "epoch": 42.47291268323773, "grad_norm": 0.6197898983955383, "learning_rate": 9.211598675592034e-05, "loss": 0.0316, "step": 66640 }, { "epoch": 42.47928616953474, "grad_norm": 0.3372287154197693, "learning_rate": 9.211301588628354e-05, "loss": 0.016, "step": 66650 }, { "epoch": 42.48565965583174, "grad_norm": 0.438781201839447, "learning_rate": 9.211004450493386e-05, "loss": 0.0186, "step": 66660 }, { "epoch": 42.49203314212875, "grad_norm": 0.5597378611564636, "learning_rate": 9.210707261190736e-05, "loss": 0.0255, "step": 66670 }, { "epoch": 42.49840662842575, "grad_norm": 0.4239688515663147, "learning_rate": 9.210410020724018e-05, "loss": 0.0225, "step": 66680 }, { "epoch": 42.504780114722756, "grad_norm": 0.3857112228870392, "learning_rate": 9.210112729096843e-05, "loss": 0.0245, "step": 66690 }, { "epoch": 42.51115360101976, "grad_norm": 0.5746787190437317, "learning_rate": 9.209815386312824e-05, "loss": 0.0335, "step": 66700 }, { "epoch": 42.517527087316765, "grad_norm": 0.5966720581054688, "learning_rate": 9.209517992375572e-05, "loss": 0.0208, "step": 66710 }, { "epoch": 42.523900573613766, "grad_norm": 0.5027617812156677, "learning_rate": 9.209220547288704e-05, "loss": 0.0249, "step": 66720 }, { "epoch": 42.530274059910774, "grad_norm": 0.8104925751686096, "learning_rate": 9.20892305105583e-05, "loss": 0.0224, "step": 66730 }, { "epoch": 42.536647546207774, "grad_norm": 1.0852450132369995, "learning_rate": 9.208625503680568e-05, "loss": 0.0225, "step": 66740 }, { "epoch": 42.54302103250478, "grad_norm": 0.3320201635360718, "learning_rate": 9.208327905166533e-05, "loss": 0.0253, "step": 66750 }, { "epoch": 42.54939451880178, "grad_norm": 0.3694443702697754, "learning_rate": 9.208030255517339e-05, "loss": 0.0309, "step": 66760 }, { "epoch": 42.55576800509879, "grad_norm": 0.37521687150001526, "learning_rate": 9.207732554736606e-05, "loss": 0.024, "step": 66770 }, { "epoch": 42.56214149139579, "grad_norm": 0.4098055958747864, "learning_rate": 9.20743480282795e-05, "loss": 0.0301, "step": 66780 }, { "epoch": 42.5685149776928, "grad_norm": 0.7972168922424316, "learning_rate": 9.207136999794987e-05, "loss": 0.0252, "step": 66790 }, { "epoch": 42.5748884639898, "grad_norm": 0.6641728281974792, "learning_rate": 9.206839145641339e-05, "loss": 0.0264, "step": 66800 }, { "epoch": 42.58126195028681, "grad_norm": 0.35140904784202576, "learning_rate": 9.20654124037062e-05, "loss": 0.0365, "step": 66810 }, { "epoch": 42.58763543658381, "grad_norm": 0.49575087428092957, "learning_rate": 9.206243283986458e-05, "loss": 0.0422, "step": 66820 }, { "epoch": 42.59400892288082, "grad_norm": 0.9227663278579712, "learning_rate": 9.205945276492465e-05, "loss": 0.0269, "step": 66830 }, { "epoch": 42.60038240917782, "grad_norm": 0.42800629138946533, "learning_rate": 9.205647217892268e-05, "loss": 0.0233, "step": 66840 }, { "epoch": 42.606755895474826, "grad_norm": 0.305634468793869, "learning_rate": 9.205349108189486e-05, "loss": 0.0224, "step": 66850 }, { "epoch": 42.61312938177183, "grad_norm": 0.5572590827941895, "learning_rate": 9.205050947387741e-05, "loss": 0.0168, "step": 66860 }, { "epoch": 42.619502868068835, "grad_norm": 0.788881242275238, "learning_rate": 9.204752735490658e-05, "loss": 0.0219, "step": 66870 }, { "epoch": 42.625876354365836, "grad_norm": 0.5477340221405029, "learning_rate": 9.204454472501859e-05, "loss": 0.022, "step": 66880 }, { "epoch": 42.632249840662844, "grad_norm": 0.4275846779346466, "learning_rate": 9.204156158424968e-05, "loss": 0.0431, "step": 66890 }, { "epoch": 42.638623326959845, "grad_norm": 0.2976972162723541, "learning_rate": 9.20385779326361e-05, "loss": 0.0174, "step": 66900 }, { "epoch": 42.64499681325685, "grad_norm": 0.31384608149528503, "learning_rate": 9.203559377021412e-05, "loss": 0.0383, "step": 66910 }, { "epoch": 42.651370299553854, "grad_norm": 0.5117910504341125, "learning_rate": 9.203260909701999e-05, "loss": 0.0197, "step": 66920 }, { "epoch": 42.65774378585086, "grad_norm": 0.42648959159851074, "learning_rate": 9.202962391308996e-05, "loss": 0.0194, "step": 66930 }, { "epoch": 42.66411727214786, "grad_norm": 0.5145031809806824, "learning_rate": 9.202663821846032e-05, "loss": 0.0227, "step": 66940 }, { "epoch": 42.67049075844487, "grad_norm": 0.5373769998550415, "learning_rate": 9.202365201316734e-05, "loss": 0.029, "step": 66950 }, { "epoch": 42.67686424474187, "grad_norm": 0.9795006513595581, "learning_rate": 9.202066529724734e-05, "loss": 0.026, "step": 66960 }, { "epoch": 42.68323773103888, "grad_norm": 0.3242267370223999, "learning_rate": 9.201767807073656e-05, "loss": 0.019, "step": 66970 }, { "epoch": 42.68961121733588, "grad_norm": 0.1857665479183197, "learning_rate": 9.201469033367132e-05, "loss": 0.0223, "step": 66980 }, { "epoch": 42.69598470363289, "grad_norm": 0.6057352423667908, "learning_rate": 9.201170208608791e-05, "loss": 0.0279, "step": 66990 }, { "epoch": 42.70235818992989, "grad_norm": 0.5301851630210876, "learning_rate": 9.200871332802267e-05, "loss": 0.0191, "step": 67000 }, { "epoch": 42.7087316762269, "grad_norm": 0.5958722233772278, "learning_rate": 9.200572405951192e-05, "loss": 0.0281, "step": 67010 }, { "epoch": 42.7151051625239, "grad_norm": 0.7547402381896973, "learning_rate": 9.200273428059193e-05, "loss": 0.0307, "step": 67020 }, { "epoch": 42.721478648820906, "grad_norm": 0.718107283115387, "learning_rate": 9.199974399129907e-05, "loss": 0.0269, "step": 67030 }, { "epoch": 42.72785213511791, "grad_norm": 0.6762949824333191, "learning_rate": 9.199675319166967e-05, "loss": 0.0312, "step": 67040 }, { "epoch": 42.734225621414915, "grad_norm": 0.4817274212837219, "learning_rate": 9.199376188174007e-05, "loss": 0.0238, "step": 67050 }, { "epoch": 42.740599107711915, "grad_norm": 0.5245202779769897, "learning_rate": 9.199077006154661e-05, "loss": 0.0272, "step": 67060 }, { "epoch": 42.74697259400892, "grad_norm": 0.9268798232078552, "learning_rate": 9.198777773112566e-05, "loss": 0.0399, "step": 67070 }, { "epoch": 42.753346080305924, "grad_norm": 0.5143446922302246, "learning_rate": 9.198478489051354e-05, "loss": 0.0299, "step": 67080 }, { "epoch": 42.75971956660293, "grad_norm": 0.525508463382721, "learning_rate": 9.198179153974667e-05, "loss": 0.0236, "step": 67090 }, { "epoch": 42.76609305289993, "grad_norm": 0.4677829444408417, "learning_rate": 9.197879767886138e-05, "loss": 0.0147, "step": 67100 }, { "epoch": 42.77246653919694, "grad_norm": 0.6096376180648804, "learning_rate": 9.197580330789407e-05, "loss": 0.0216, "step": 67110 }, { "epoch": 42.77884002549394, "grad_norm": 0.7969074249267578, "learning_rate": 9.197280842688112e-05, "loss": 0.0289, "step": 67120 }, { "epoch": 42.78521351179095, "grad_norm": 0.3099363446235657, "learning_rate": 9.196981303585892e-05, "loss": 0.0276, "step": 67130 }, { "epoch": 42.79158699808795, "grad_norm": 0.4763953685760498, "learning_rate": 9.196681713486386e-05, "loss": 0.032, "step": 67140 }, { "epoch": 42.79796048438496, "grad_norm": 0.21379682421684265, "learning_rate": 9.196382072393235e-05, "loss": 0.0199, "step": 67150 }, { "epoch": 42.80433397068197, "grad_norm": 0.3612785339355469, "learning_rate": 9.196082380310079e-05, "loss": 0.0315, "step": 67160 }, { "epoch": 42.81070745697897, "grad_norm": 0.5285971164703369, "learning_rate": 9.195782637240562e-05, "loss": 0.0249, "step": 67170 }, { "epoch": 42.817080943275975, "grad_norm": 0.5027537941932678, "learning_rate": 9.195482843188323e-05, "loss": 0.0213, "step": 67180 }, { "epoch": 42.823454429572976, "grad_norm": 0.43655627965927124, "learning_rate": 9.195182998157007e-05, "loss": 0.0236, "step": 67190 }, { "epoch": 42.829827915869984, "grad_norm": 0.28319913148880005, "learning_rate": 9.194883102150256e-05, "loss": 0.0371, "step": 67200 }, { "epoch": 42.836201402166985, "grad_norm": 0.25379395484924316, "learning_rate": 9.194583155171716e-05, "loss": 0.0218, "step": 67210 }, { "epoch": 42.84257488846399, "grad_norm": 0.36270880699157715, "learning_rate": 9.19428315722503e-05, "loss": 0.0281, "step": 67220 }, { "epoch": 42.848948374760994, "grad_norm": 0.4661734700202942, "learning_rate": 9.193983108313841e-05, "loss": 0.0264, "step": 67230 }, { "epoch": 42.855321861058, "grad_norm": 0.5283739566802979, "learning_rate": 9.193683008441801e-05, "loss": 0.0368, "step": 67240 }, { "epoch": 42.861695347355, "grad_norm": 0.3699325919151306, "learning_rate": 9.193382857612552e-05, "loss": 0.0211, "step": 67250 }, { "epoch": 42.86806883365201, "grad_norm": 0.6959346532821655, "learning_rate": 9.193082655829742e-05, "loss": 0.0295, "step": 67260 }, { "epoch": 42.87444231994901, "grad_norm": 0.6285856366157532, "learning_rate": 9.192782403097018e-05, "loss": 0.0304, "step": 67270 }, { "epoch": 42.88081580624602, "grad_norm": 0.271664023399353, "learning_rate": 9.19248209941803e-05, "loss": 0.0184, "step": 67280 }, { "epoch": 42.88718929254302, "grad_norm": 0.2527194321155548, "learning_rate": 9.192181744796425e-05, "loss": 0.0255, "step": 67290 }, { "epoch": 42.89356277884003, "grad_norm": 0.5781341195106506, "learning_rate": 9.191881339235855e-05, "loss": 0.026, "step": 67300 }, { "epoch": 42.89993626513703, "grad_norm": 0.47590798139572144, "learning_rate": 9.191580882739967e-05, "loss": 0.0165, "step": 67310 }, { "epoch": 42.90630975143404, "grad_norm": 0.7032646536827087, "learning_rate": 9.191280375312417e-05, "loss": 0.0314, "step": 67320 }, { "epoch": 42.91268323773104, "grad_norm": 0.49506446719169617, "learning_rate": 9.19097981695685e-05, "loss": 0.0203, "step": 67330 }, { "epoch": 42.919056724028046, "grad_norm": 0.5686636567115784, "learning_rate": 9.190679207676922e-05, "loss": 0.0346, "step": 67340 }, { "epoch": 42.92543021032505, "grad_norm": 0.5911718010902405, "learning_rate": 9.190378547476286e-05, "loss": 0.0241, "step": 67350 }, { "epoch": 42.931803696622055, "grad_norm": 0.42223984003067017, "learning_rate": 9.190077836358594e-05, "loss": 0.0204, "step": 67360 }, { "epoch": 42.938177182919055, "grad_norm": 0.49546900391578674, "learning_rate": 9.189777074327501e-05, "loss": 0.0173, "step": 67370 }, { "epoch": 42.94455066921606, "grad_norm": 0.32483208179473877, "learning_rate": 9.189476261386659e-05, "loss": 0.0245, "step": 67380 }, { "epoch": 42.950924155513064, "grad_norm": 0.3610183894634247, "learning_rate": 9.189175397539727e-05, "loss": 0.0249, "step": 67390 }, { "epoch": 42.95729764181007, "grad_norm": 0.6013220548629761, "learning_rate": 9.188874482790358e-05, "loss": 0.0307, "step": 67400 }, { "epoch": 42.96367112810707, "grad_norm": 0.4735114872455597, "learning_rate": 9.188573517142208e-05, "loss": 0.0198, "step": 67410 }, { "epoch": 42.97004461440408, "grad_norm": 0.5498377680778503, "learning_rate": 9.188272500598935e-05, "loss": 0.0296, "step": 67420 }, { "epoch": 42.97641810070108, "grad_norm": 0.34324944019317627, "learning_rate": 9.187971433164198e-05, "loss": 0.0257, "step": 67430 }, { "epoch": 42.98279158699809, "grad_norm": 0.40846747159957886, "learning_rate": 9.187670314841653e-05, "loss": 0.0207, "step": 67440 }, { "epoch": 42.98916507329509, "grad_norm": 0.26030123233795166, "learning_rate": 9.187369145634958e-05, "loss": 0.0296, "step": 67450 }, { "epoch": 42.9955385595921, "grad_norm": 0.6081669330596924, "learning_rate": 9.187067925547779e-05, "loss": 0.0225, "step": 67460 }, { "epoch": 43.0019120458891, "grad_norm": 0.400501549243927, "learning_rate": 9.186766654583767e-05, "loss": 0.0234, "step": 67470 }, { "epoch": 43.00828553218611, "grad_norm": 0.3854467570781708, "learning_rate": 9.18646533274659e-05, "loss": 0.0209, "step": 67480 }, { "epoch": 43.01465901848311, "grad_norm": 0.727696418762207, "learning_rate": 9.186163960039906e-05, "loss": 0.0284, "step": 67490 }, { "epoch": 43.021032504780116, "grad_norm": 0.5858085751533508, "learning_rate": 9.185862536467377e-05, "loss": 0.0314, "step": 67500 }, { "epoch": 43.02740599107712, "grad_norm": 0.3960177004337311, "learning_rate": 9.185561062032666e-05, "loss": 0.0246, "step": 67510 }, { "epoch": 43.033779477374125, "grad_norm": 0.4704405963420868, "learning_rate": 9.185259536739437e-05, "loss": 0.0279, "step": 67520 }, { "epoch": 43.040152963671126, "grad_norm": 0.3718126714229584, "learning_rate": 9.184957960591351e-05, "loss": 0.0202, "step": 67530 }, { "epoch": 43.046526449968134, "grad_norm": 0.6491569876670837, "learning_rate": 9.184656333592076e-05, "loss": 0.0261, "step": 67540 }, { "epoch": 43.052899936265135, "grad_norm": 0.38035815954208374, "learning_rate": 9.184354655745277e-05, "loss": 0.0288, "step": 67550 }, { "epoch": 43.05927342256214, "grad_norm": 0.5793561935424805, "learning_rate": 9.184052927054616e-05, "loss": 0.0369, "step": 67560 }, { "epoch": 43.065646908859144, "grad_norm": 0.45132583379745483, "learning_rate": 9.183751147523762e-05, "loss": 0.0255, "step": 67570 }, { "epoch": 43.07202039515615, "grad_norm": 0.3775477409362793, "learning_rate": 9.183449317156382e-05, "loss": 0.0342, "step": 67580 }, { "epoch": 43.07839388145315, "grad_norm": 0.4891328513622284, "learning_rate": 9.183147435956145e-05, "loss": 0.0398, "step": 67590 }, { "epoch": 43.08476736775016, "grad_norm": 0.4321242570877075, "learning_rate": 9.182845503926713e-05, "loss": 0.0364, "step": 67600 }, { "epoch": 43.09114085404716, "grad_norm": 0.6292339563369751, "learning_rate": 9.182543521071763e-05, "loss": 0.0239, "step": 67610 }, { "epoch": 43.09751434034417, "grad_norm": 0.5853735208511353, "learning_rate": 9.182241487394958e-05, "loss": 0.0439, "step": 67620 }, { "epoch": 43.10388782664117, "grad_norm": 0.5192533731460571, "learning_rate": 9.181939402899969e-05, "loss": 0.0251, "step": 67630 }, { "epoch": 43.11026131293818, "grad_norm": 0.5969426035881042, "learning_rate": 9.18163726759047e-05, "loss": 0.0361, "step": 67640 }, { "epoch": 43.11663479923518, "grad_norm": 0.7025969624519348, "learning_rate": 9.181335081470129e-05, "loss": 0.0251, "step": 67650 }, { "epoch": 43.12300828553219, "grad_norm": 0.7367541790008545, "learning_rate": 9.181032844542619e-05, "loss": 0.0193, "step": 67660 }, { "epoch": 43.12938177182919, "grad_norm": 0.45818081498146057, "learning_rate": 9.180730556811612e-05, "loss": 0.0181, "step": 67670 }, { "epoch": 43.135755258126196, "grad_norm": 0.45922696590423584, "learning_rate": 9.180428218280783e-05, "loss": 0.0232, "step": 67680 }, { "epoch": 43.1421287444232, "grad_norm": 0.4780043959617615, "learning_rate": 9.180125828953803e-05, "loss": 0.0361, "step": 67690 }, { "epoch": 43.148502230720204, "grad_norm": 0.37901586294174194, "learning_rate": 9.179823388834347e-05, "loss": 0.019, "step": 67700 }, { "epoch": 43.154875717017205, "grad_norm": 0.6662536859512329, "learning_rate": 9.179520897926092e-05, "loss": 0.0201, "step": 67710 }, { "epoch": 43.16124920331421, "grad_norm": 0.369272917509079, "learning_rate": 9.17921835623271e-05, "loss": 0.0204, "step": 67720 }, { "epoch": 43.167622689611214, "grad_norm": 0.6434751749038696, "learning_rate": 9.178915763757881e-05, "loss": 0.0224, "step": 67730 }, { "epoch": 43.17399617590822, "grad_norm": 0.6363507509231567, "learning_rate": 9.17861312050528e-05, "loss": 0.0288, "step": 67740 }, { "epoch": 43.18036966220522, "grad_norm": 0.5389083623886108, "learning_rate": 9.178310426478582e-05, "loss": 0.0196, "step": 67750 }, { "epoch": 43.18674314850223, "grad_norm": 0.915726363658905, "learning_rate": 9.17800768168147e-05, "loss": 0.0245, "step": 67760 }, { "epoch": 43.19311663479923, "grad_norm": 0.6710572242736816, "learning_rate": 9.177704886117619e-05, "loss": 0.0341, "step": 67770 }, { "epoch": 43.19949012109624, "grad_norm": 0.31386756896972656, "learning_rate": 9.177402039790708e-05, "loss": 0.023, "step": 67780 }, { "epoch": 43.20586360739325, "grad_norm": 0.4639987349510193, "learning_rate": 9.177099142704419e-05, "loss": 0.0142, "step": 67790 }, { "epoch": 43.21223709369025, "grad_norm": 0.1893891543149948, "learning_rate": 9.176796194862433e-05, "loss": 0.0177, "step": 67800 }, { "epoch": 43.218610579987256, "grad_norm": 0.5431473255157471, "learning_rate": 9.176493196268427e-05, "loss": 0.0326, "step": 67810 }, { "epoch": 43.22498406628426, "grad_norm": 0.5981893539428711, "learning_rate": 9.176190146926087e-05, "loss": 0.0259, "step": 67820 }, { "epoch": 43.231357552581265, "grad_norm": 0.32504671812057495, "learning_rate": 9.175887046839093e-05, "loss": 0.0204, "step": 67830 }, { "epoch": 43.237731038878266, "grad_norm": 0.4664250910282135, "learning_rate": 9.175583896011131e-05, "loss": 0.0392, "step": 67840 }, { "epoch": 43.244104525175274, "grad_norm": 0.2956750988960266, "learning_rate": 9.17528069444588e-05, "loss": 0.0244, "step": 67850 }, { "epoch": 43.250478011472275, "grad_norm": 0.5154459476470947, "learning_rate": 9.174977442147027e-05, "loss": 0.0211, "step": 67860 }, { "epoch": 43.25685149776928, "grad_norm": 0.3835418224334717, "learning_rate": 9.174674139118256e-05, "loss": 0.0376, "step": 67870 }, { "epoch": 43.263224984066284, "grad_norm": 0.260303258895874, "learning_rate": 9.174370785363253e-05, "loss": 0.0201, "step": 67880 }, { "epoch": 43.26959847036329, "grad_norm": 0.296442449092865, "learning_rate": 9.174067380885703e-05, "loss": 0.0206, "step": 67890 }, { "epoch": 43.27597195666029, "grad_norm": 0.8064464330673218, "learning_rate": 9.173763925689294e-05, "loss": 0.0193, "step": 67900 }, { "epoch": 43.2823454429573, "grad_norm": 0.5235943794250488, "learning_rate": 9.173460419777714e-05, "loss": 0.0209, "step": 67910 }, { "epoch": 43.2887189292543, "grad_norm": 0.5060768723487854, "learning_rate": 9.173156863154646e-05, "loss": 0.02, "step": 67920 }, { "epoch": 43.29509241555131, "grad_norm": 0.24220284819602966, "learning_rate": 9.172853255823785e-05, "loss": 0.0239, "step": 67930 }, { "epoch": 43.30146590184831, "grad_norm": 0.3179489076137543, "learning_rate": 9.172549597788815e-05, "loss": 0.0245, "step": 67940 }, { "epoch": 43.30783938814532, "grad_norm": 0.49112677574157715, "learning_rate": 9.17224588905343e-05, "loss": 0.0259, "step": 67950 }, { "epoch": 43.31421287444232, "grad_norm": 0.8284446597099304, "learning_rate": 9.171942129621317e-05, "loss": 0.0189, "step": 67960 }, { "epoch": 43.32058636073933, "grad_norm": 0.3642578721046448, "learning_rate": 9.171638319496167e-05, "loss": 0.0416, "step": 67970 }, { "epoch": 43.32695984703633, "grad_norm": 0.414824515581131, "learning_rate": 9.171334458681676e-05, "loss": 0.0264, "step": 67980 }, { "epoch": 43.333333333333336, "grad_norm": 0.45178139209747314, "learning_rate": 9.17103054718153e-05, "loss": 0.0254, "step": 67990 }, { "epoch": 43.33970681963034, "grad_norm": 0.4547591209411621, "learning_rate": 9.170726584999425e-05, "loss": 0.0248, "step": 68000 }, { "epoch": 43.346080305927345, "grad_norm": 0.6854702234268188, "learning_rate": 9.170422572139055e-05, "loss": 0.038, "step": 68010 }, { "epoch": 43.352453792224345, "grad_norm": 0.5485163927078247, "learning_rate": 9.170118508604115e-05, "loss": 0.0357, "step": 68020 }, { "epoch": 43.35882727852135, "grad_norm": 0.5728838443756104, "learning_rate": 9.169814394398296e-05, "loss": 0.0259, "step": 68030 }, { "epoch": 43.365200764818354, "grad_norm": 0.3205910623073578, "learning_rate": 9.169510229525294e-05, "loss": 0.0216, "step": 68040 }, { "epoch": 43.37157425111536, "grad_norm": 0.48292049765586853, "learning_rate": 9.169206013988807e-05, "loss": 0.0282, "step": 68050 }, { "epoch": 43.37794773741236, "grad_norm": 0.48375338315963745, "learning_rate": 9.168901747792531e-05, "loss": 0.0299, "step": 68060 }, { "epoch": 43.38432122370937, "grad_norm": 0.8962668776512146, "learning_rate": 9.168597430940162e-05, "loss": 0.0289, "step": 68070 }, { "epoch": 43.39069471000637, "grad_norm": 0.7977877855300903, "learning_rate": 9.1682930634354e-05, "loss": 0.0238, "step": 68080 }, { "epoch": 43.39706819630338, "grad_norm": 0.568190336227417, "learning_rate": 9.16798864528194e-05, "loss": 0.023, "step": 68090 }, { "epoch": 43.40344168260038, "grad_norm": 0.5441506505012512, "learning_rate": 9.167684176483483e-05, "loss": 0.0228, "step": 68100 }, { "epoch": 43.40981516889739, "grad_norm": 0.4669262766838074, "learning_rate": 9.167379657043728e-05, "loss": 0.0255, "step": 68110 }, { "epoch": 43.41618865519439, "grad_norm": 0.4530961811542511, "learning_rate": 9.167075086966376e-05, "loss": 0.0279, "step": 68120 }, { "epoch": 43.4225621414914, "grad_norm": 0.3778327405452728, "learning_rate": 9.166770466255128e-05, "loss": 0.0182, "step": 68130 }, { "epoch": 43.4289356277884, "grad_norm": 0.37270620465278625, "learning_rate": 9.166465794913685e-05, "loss": 0.0278, "step": 68140 }, { "epoch": 43.435309114085406, "grad_norm": 0.8757895231246948, "learning_rate": 9.166161072945747e-05, "loss": 0.0208, "step": 68150 }, { "epoch": 43.44168260038241, "grad_norm": 0.3630264699459076, "learning_rate": 9.16585630035502e-05, "loss": 0.0217, "step": 68160 }, { "epoch": 43.448056086679415, "grad_norm": 0.47920316457748413, "learning_rate": 9.165551477145204e-05, "loss": 0.0255, "step": 68170 }, { "epoch": 43.454429572976416, "grad_norm": 0.4455852508544922, "learning_rate": 9.165246603320006e-05, "loss": 0.0234, "step": 68180 }, { "epoch": 43.460803059273424, "grad_norm": 0.7605883479118347, "learning_rate": 9.16494167888313e-05, "loss": 0.0288, "step": 68190 }, { "epoch": 43.467176545570425, "grad_norm": 0.48727014660835266, "learning_rate": 9.164636703838278e-05, "loss": 0.0135, "step": 68200 }, { "epoch": 43.47355003186743, "grad_norm": 0.386100709438324, "learning_rate": 9.16433167818916e-05, "loss": 0.0285, "step": 68210 }, { "epoch": 43.47992351816443, "grad_norm": 0.2492709457874298, "learning_rate": 9.16402660193948e-05, "loss": 0.0171, "step": 68220 }, { "epoch": 43.48629700446144, "grad_norm": 0.5817500352859497, "learning_rate": 9.163721475092945e-05, "loss": 0.0254, "step": 68230 }, { "epoch": 43.49267049075844, "grad_norm": 0.44650399684906006, "learning_rate": 9.163416297653262e-05, "loss": 0.0227, "step": 68240 }, { "epoch": 43.49904397705545, "grad_norm": 0.6451097726821899, "learning_rate": 9.16311106962414e-05, "loss": 0.0243, "step": 68250 }, { "epoch": 43.50541746335245, "grad_norm": 0.34349969029426575, "learning_rate": 9.16280579100929e-05, "loss": 0.0177, "step": 68260 }, { "epoch": 43.51179094964946, "grad_norm": 0.14573559165000916, "learning_rate": 9.162500461812418e-05, "loss": 0.0234, "step": 68270 }, { "epoch": 43.51816443594646, "grad_norm": 0.3621068596839905, "learning_rate": 9.162195082037234e-05, "loss": 0.0243, "step": 68280 }, { "epoch": 43.52453792224347, "grad_norm": 0.4514976739883423, "learning_rate": 9.161889651687452e-05, "loss": 0.0361, "step": 68290 }, { "epoch": 43.53091140854047, "grad_norm": 0.42492860555648804, "learning_rate": 9.16158417076678e-05, "loss": 0.0426, "step": 68300 }, { "epoch": 43.53728489483748, "grad_norm": 0.49247828125953674, "learning_rate": 9.161278639278932e-05, "loss": 0.0276, "step": 68310 }, { "epoch": 43.54365838113448, "grad_norm": 0.6026017665863037, "learning_rate": 9.160973057227619e-05, "loss": 0.0298, "step": 68320 }, { "epoch": 43.550031867431485, "grad_norm": 0.49606752395629883, "learning_rate": 9.160667424616555e-05, "loss": 0.0263, "step": 68330 }, { "epoch": 43.556405353728486, "grad_norm": 0.6589988470077515, "learning_rate": 9.160361741449453e-05, "loss": 0.0287, "step": 68340 }, { "epoch": 43.562778840025494, "grad_norm": 0.7434877157211304, "learning_rate": 9.16005600773003e-05, "loss": 0.0217, "step": 68350 }, { "epoch": 43.569152326322495, "grad_norm": 0.4063299894332886, "learning_rate": 9.159750223461996e-05, "loss": 0.0159, "step": 68360 }, { "epoch": 43.5755258126195, "grad_norm": 0.2699035406112671, "learning_rate": 9.15944438864907e-05, "loss": 0.016, "step": 68370 }, { "epoch": 43.581899298916504, "grad_norm": 0.5909351110458374, "learning_rate": 9.159138503294968e-05, "loss": 0.023, "step": 68380 }, { "epoch": 43.58827278521351, "grad_norm": 0.41453972458839417, "learning_rate": 9.158832567403406e-05, "loss": 0.0324, "step": 68390 }, { "epoch": 43.59464627151051, "grad_norm": 0.28453704714775085, "learning_rate": 9.158526580978102e-05, "loss": 0.0284, "step": 68400 }, { "epoch": 43.60101975780752, "grad_norm": 0.23097805678844452, "learning_rate": 9.158220544022773e-05, "loss": 0.0171, "step": 68410 }, { "epoch": 43.60739324410453, "grad_norm": 0.4289282560348511, "learning_rate": 9.157914456541138e-05, "loss": 0.0282, "step": 68420 }, { "epoch": 43.61376673040153, "grad_norm": 0.5962466597557068, "learning_rate": 9.157608318536919e-05, "loss": 0.0221, "step": 68430 }, { "epoch": 43.62014021669854, "grad_norm": 0.22112102806568146, "learning_rate": 9.157302130013831e-05, "loss": 0.0228, "step": 68440 }, { "epoch": 43.62651370299554, "grad_norm": 0.3450978696346283, "learning_rate": 9.156995890975599e-05, "loss": 0.0235, "step": 68450 }, { "epoch": 43.632887189292546, "grad_norm": 0.6185300946235657, "learning_rate": 9.156689601425942e-05, "loss": 0.0459, "step": 68460 }, { "epoch": 43.63926067558955, "grad_norm": 0.23976293206214905, "learning_rate": 9.15638326136858e-05, "loss": 0.0229, "step": 68470 }, { "epoch": 43.645634161886555, "grad_norm": 0.7175559401512146, "learning_rate": 9.156076870807239e-05, "loss": 0.0328, "step": 68480 }, { "epoch": 43.652007648183556, "grad_norm": 0.2574958801269531, "learning_rate": 9.155770429745638e-05, "loss": 0.0264, "step": 68490 }, { "epoch": 43.658381134480564, "grad_norm": 0.6846250891685486, "learning_rate": 9.155463938187505e-05, "loss": 0.0283, "step": 68500 }, { "epoch": 43.664754620777565, "grad_norm": 0.4479522109031677, "learning_rate": 9.15515739613656e-05, "loss": 0.031, "step": 68510 }, { "epoch": 43.67112810707457, "grad_norm": 0.3911721408367157, "learning_rate": 9.154850803596529e-05, "loss": 0.0258, "step": 68520 }, { "epoch": 43.677501593371574, "grad_norm": 0.49033698439598083, "learning_rate": 9.15454416057114e-05, "loss": 0.0329, "step": 68530 }, { "epoch": 43.68387507966858, "grad_norm": 0.5527986884117126, "learning_rate": 9.154237467064115e-05, "loss": 0.027, "step": 68540 }, { "epoch": 43.69024856596558, "grad_norm": 0.8870930075645447, "learning_rate": 9.153930723079184e-05, "loss": 0.0372, "step": 68550 }, { "epoch": 43.69662205226259, "grad_norm": 0.3780331015586853, "learning_rate": 9.153623928620072e-05, "loss": 0.0375, "step": 68560 }, { "epoch": 43.70299553855959, "grad_norm": 0.3388501703739166, "learning_rate": 9.153317083690507e-05, "loss": 0.0177, "step": 68570 }, { "epoch": 43.7093690248566, "grad_norm": 0.5259823203086853, "learning_rate": 9.153010188294219e-05, "loss": 0.0166, "step": 68580 }, { "epoch": 43.7157425111536, "grad_norm": 0.6734403371810913, "learning_rate": 9.152703242434936e-05, "loss": 0.029, "step": 68590 }, { "epoch": 43.72211599745061, "grad_norm": 0.5611023306846619, "learning_rate": 9.152396246116387e-05, "loss": 0.027, "step": 68600 }, { "epoch": 43.72848948374761, "grad_norm": 0.447003573179245, "learning_rate": 9.152089199342304e-05, "loss": 0.0274, "step": 68610 }, { "epoch": 43.73486297004462, "grad_norm": 0.6926752328872681, "learning_rate": 9.151782102116417e-05, "loss": 0.036, "step": 68620 }, { "epoch": 43.74123645634162, "grad_norm": 0.5731717944145203, "learning_rate": 9.151474954442457e-05, "loss": 0.0253, "step": 68630 }, { "epoch": 43.747609942638626, "grad_norm": 0.3329644203186035, "learning_rate": 9.151167756324157e-05, "loss": 0.0196, "step": 68640 }, { "epoch": 43.753983428935626, "grad_norm": 0.5053603649139404, "learning_rate": 9.150860507765248e-05, "loss": 0.0339, "step": 68650 }, { "epoch": 43.760356915232634, "grad_norm": 0.6382958292961121, "learning_rate": 9.150553208769466e-05, "loss": 0.0284, "step": 68660 }, { "epoch": 43.766730401529635, "grad_norm": 0.4499886631965637, "learning_rate": 9.150245859340544e-05, "loss": 0.04, "step": 68670 }, { "epoch": 43.77310388782664, "grad_norm": 0.27924874424934387, "learning_rate": 9.149938459482218e-05, "loss": 0.0177, "step": 68680 }, { "epoch": 43.779477374123644, "grad_norm": 0.3499974012374878, "learning_rate": 9.149631009198218e-05, "loss": 0.0289, "step": 68690 }, { "epoch": 43.78585086042065, "grad_norm": 0.6004745960235596, "learning_rate": 9.149323508492286e-05, "loss": 0.0185, "step": 68700 }, { "epoch": 43.79222434671765, "grad_norm": 0.44475454092025757, "learning_rate": 9.149015957368155e-05, "loss": 0.0275, "step": 68710 }, { "epoch": 43.79859783301466, "grad_norm": 0.7249120473861694, "learning_rate": 9.148708355829562e-05, "loss": 0.024, "step": 68720 }, { "epoch": 43.80497131931166, "grad_norm": 0.5135484933853149, "learning_rate": 9.148400703880246e-05, "loss": 0.0299, "step": 68730 }, { "epoch": 43.81134480560867, "grad_norm": 0.6492828726768494, "learning_rate": 9.148093001523945e-05, "loss": 0.0275, "step": 68740 }, { "epoch": 43.81771829190567, "grad_norm": 0.3619154989719391, "learning_rate": 9.147785248764397e-05, "loss": 0.0245, "step": 68750 }, { "epoch": 43.82409177820268, "grad_norm": 0.4305607080459595, "learning_rate": 9.147477445605343e-05, "loss": 0.026, "step": 68760 }, { "epoch": 43.83046526449968, "grad_norm": 0.34129786491394043, "learning_rate": 9.147169592050522e-05, "loss": 0.0277, "step": 68770 }, { "epoch": 43.83683875079669, "grad_norm": 0.4091038703918457, "learning_rate": 9.146861688103674e-05, "loss": 0.0167, "step": 68780 }, { "epoch": 43.84321223709369, "grad_norm": 0.5101672410964966, "learning_rate": 9.146553733768542e-05, "loss": 0.0254, "step": 68790 }, { "epoch": 43.849585723390696, "grad_norm": 0.5365066528320312, "learning_rate": 9.146245729048865e-05, "loss": 0.0305, "step": 68800 }, { "epoch": 43.8559592096877, "grad_norm": 0.3254113495349884, "learning_rate": 9.145937673948391e-05, "loss": 0.0285, "step": 68810 }, { "epoch": 43.862332695984705, "grad_norm": 0.5276579260826111, "learning_rate": 9.145629568470857e-05, "loss": 0.0393, "step": 68820 }, { "epoch": 43.868706182281706, "grad_norm": 0.5106366872787476, "learning_rate": 9.145321412620011e-05, "loss": 0.0201, "step": 68830 }, { "epoch": 43.875079668578714, "grad_norm": 0.4448765814304352, "learning_rate": 9.145013206399597e-05, "loss": 0.033, "step": 68840 }, { "epoch": 43.881453154875715, "grad_norm": 0.4537672996520996, "learning_rate": 9.144704949813357e-05, "loss": 0.0212, "step": 68850 }, { "epoch": 43.88782664117272, "grad_norm": 0.39413580298423767, "learning_rate": 9.14439664286504e-05, "loss": 0.024, "step": 68860 }, { "epoch": 43.89420012746972, "grad_norm": 0.4654833972454071, "learning_rate": 9.144088285558389e-05, "loss": 0.0208, "step": 68870 }, { "epoch": 43.90057361376673, "grad_norm": 0.7494217157363892, "learning_rate": 9.143779877897155e-05, "loss": 0.0291, "step": 68880 }, { "epoch": 43.90694710006373, "grad_norm": 0.3948776423931122, "learning_rate": 9.143471419885082e-05, "loss": 0.0203, "step": 68890 }, { "epoch": 43.91332058636074, "grad_norm": 1.3631755113601685, "learning_rate": 9.14316291152592e-05, "loss": 0.029, "step": 68900 }, { "epoch": 43.91969407265774, "grad_norm": 0.3103019893169403, "learning_rate": 9.142854352823416e-05, "loss": 0.028, "step": 68910 }, { "epoch": 43.92606755895475, "grad_norm": 0.3118044137954712, "learning_rate": 9.14254574378132e-05, "loss": 0.0186, "step": 68920 }, { "epoch": 43.93244104525175, "grad_norm": 0.3544737696647644, "learning_rate": 9.142237084403382e-05, "loss": 0.0185, "step": 68930 }, { "epoch": 43.93881453154876, "grad_norm": 0.6615563631057739, "learning_rate": 9.141928374693353e-05, "loss": 0.028, "step": 68940 }, { "epoch": 43.94518801784576, "grad_norm": 0.3993005156517029, "learning_rate": 9.141619614654984e-05, "loss": 0.0235, "step": 68950 }, { "epoch": 43.95156150414277, "grad_norm": 0.4049597382545471, "learning_rate": 9.141310804292025e-05, "loss": 0.0257, "step": 68960 }, { "epoch": 43.95793499043977, "grad_norm": 0.43174174427986145, "learning_rate": 9.14100194360823e-05, "loss": 0.0244, "step": 68970 }, { "epoch": 43.964308476736775, "grad_norm": 0.45717301964759827, "learning_rate": 9.140693032607353e-05, "loss": 0.024, "step": 68980 }, { "epoch": 43.970681963033776, "grad_norm": 0.43140095472335815, "learning_rate": 9.140384071293145e-05, "loss": 0.0406, "step": 68990 }, { "epoch": 43.977055449330784, "grad_norm": 0.35039833188056946, "learning_rate": 9.140075059669362e-05, "loss": 0.0216, "step": 69000 }, { "epoch": 43.983428935627785, "grad_norm": 0.5256003141403198, "learning_rate": 9.139765997739758e-05, "loss": 0.0236, "step": 69010 }, { "epoch": 43.98980242192479, "grad_norm": 0.5736014246940613, "learning_rate": 9.139456885508086e-05, "loss": 0.0218, "step": 69020 }, { "epoch": 43.9961759082218, "grad_norm": 0.47632429003715515, "learning_rate": 9.139147722978108e-05, "loss": 0.0382, "step": 69030 }, { "epoch": 44.0025493945188, "grad_norm": 0.7445272207260132, "learning_rate": 9.138838510153576e-05, "loss": 0.0237, "step": 69040 }, { "epoch": 44.00892288081581, "grad_norm": 0.5885515809059143, "learning_rate": 9.138529247038247e-05, "loss": 0.0259, "step": 69050 }, { "epoch": 44.01529636711281, "grad_norm": 0.5254964828491211, "learning_rate": 9.138219933635883e-05, "loss": 0.0287, "step": 69060 }, { "epoch": 44.02166985340982, "grad_norm": 0.7452909350395203, "learning_rate": 9.13791056995024e-05, "loss": 0.0246, "step": 69070 }, { "epoch": 44.02804333970682, "grad_norm": 0.41195812821388245, "learning_rate": 9.137601155985073e-05, "loss": 0.0174, "step": 69080 }, { "epoch": 44.03441682600383, "grad_norm": 0.46608859300613403, "learning_rate": 9.137291691744147e-05, "loss": 0.0226, "step": 69090 }, { "epoch": 44.04079031230083, "grad_norm": 0.36798718571662903, "learning_rate": 9.136982177231221e-05, "loss": 0.0307, "step": 69100 }, { "epoch": 44.047163798597836, "grad_norm": 0.3875144422054291, "learning_rate": 9.136672612450057e-05, "loss": 0.0166, "step": 69110 }, { "epoch": 44.05353728489484, "grad_norm": 0.23577532172203064, "learning_rate": 9.136362997404413e-05, "loss": 0.0169, "step": 69120 }, { "epoch": 44.059910771191845, "grad_norm": 0.47972291707992554, "learning_rate": 9.136053332098054e-05, "loss": 0.0294, "step": 69130 }, { "epoch": 44.066284257488846, "grad_norm": 0.6593591570854187, "learning_rate": 9.135743616534742e-05, "loss": 0.043, "step": 69140 }, { "epoch": 44.072657743785854, "grad_norm": 0.33547520637512207, "learning_rate": 9.13543385071824e-05, "loss": 0.0308, "step": 69150 }, { "epoch": 44.079031230082855, "grad_norm": 0.6319129467010498, "learning_rate": 9.135124034652313e-05, "loss": 0.0243, "step": 69160 }, { "epoch": 44.08540471637986, "grad_norm": 0.5081955194473267, "learning_rate": 9.134814168340724e-05, "loss": 0.0204, "step": 69170 }, { "epoch": 44.09177820267686, "grad_norm": 0.5048558115959167, "learning_rate": 9.134504251787239e-05, "loss": 0.0265, "step": 69180 }, { "epoch": 44.09815168897387, "grad_norm": 0.4899298846721649, "learning_rate": 9.134194284995624e-05, "loss": 0.0319, "step": 69190 }, { "epoch": 44.10452517527087, "grad_norm": 1.1364482641220093, "learning_rate": 9.133884267969646e-05, "loss": 0.0318, "step": 69200 }, { "epoch": 44.11089866156788, "grad_norm": 0.5241957902908325, "learning_rate": 9.133574200713068e-05, "loss": 0.029, "step": 69210 }, { "epoch": 44.11727214786488, "grad_norm": 0.3964727520942688, "learning_rate": 9.133264083229665e-05, "loss": 0.0222, "step": 69220 }, { "epoch": 44.12364563416189, "grad_norm": 0.4636227786540985, "learning_rate": 9.132953915523198e-05, "loss": 0.0188, "step": 69230 }, { "epoch": 44.13001912045889, "grad_norm": 0.6302806735038757, "learning_rate": 9.132643697597439e-05, "loss": 0.0425, "step": 69240 }, { "epoch": 44.1363926067559, "grad_norm": 0.6307168006896973, "learning_rate": 9.132333429456158e-05, "loss": 0.0176, "step": 69250 }, { "epoch": 44.1427660930529, "grad_norm": 0.788242757320404, "learning_rate": 9.132023111103125e-05, "loss": 0.0414, "step": 69260 }, { "epoch": 44.14913957934991, "grad_norm": 0.3888011872768402, "learning_rate": 9.131712742542108e-05, "loss": 0.0223, "step": 69270 }, { "epoch": 44.15551306564691, "grad_norm": 0.20786574482917786, "learning_rate": 9.13140232377688e-05, "loss": 0.0197, "step": 69280 }, { "epoch": 44.161886551943915, "grad_norm": 0.34875741600990295, "learning_rate": 9.131091854811216e-05, "loss": 0.0236, "step": 69290 }, { "epoch": 44.168260038240916, "grad_norm": 0.5778881311416626, "learning_rate": 9.130781335648883e-05, "loss": 0.0215, "step": 69300 }, { "epoch": 44.174633524537924, "grad_norm": 0.5591598749160767, "learning_rate": 9.130470766293657e-05, "loss": 0.0193, "step": 69310 }, { "epoch": 44.181007010834925, "grad_norm": 0.5050633549690247, "learning_rate": 9.130160146749312e-05, "loss": 0.0297, "step": 69320 }, { "epoch": 44.18738049713193, "grad_norm": 0.6095276474952698, "learning_rate": 9.129849477019621e-05, "loss": 0.034, "step": 69330 }, { "epoch": 44.193753983428934, "grad_norm": 0.6065836548805237, "learning_rate": 9.12953875710836e-05, "loss": 0.0409, "step": 69340 }, { "epoch": 44.20012746972594, "grad_norm": 0.8167550563812256, "learning_rate": 9.129227987019304e-05, "loss": 0.0273, "step": 69350 }, { "epoch": 44.20650095602294, "grad_norm": 0.5086143016815186, "learning_rate": 9.12891716675623e-05, "loss": 0.0287, "step": 69360 }, { "epoch": 44.21287444231995, "grad_norm": 1.3218098878860474, "learning_rate": 9.128606296322915e-05, "loss": 0.0192, "step": 69370 }, { "epoch": 44.21924792861695, "grad_norm": 0.2761237323284149, "learning_rate": 9.128295375723133e-05, "loss": 0.0186, "step": 69380 }, { "epoch": 44.22562141491396, "grad_norm": 0.3746655583381653, "learning_rate": 9.127984404960666e-05, "loss": 0.0452, "step": 69390 }, { "epoch": 44.23199490121096, "grad_norm": 0.47848236560821533, "learning_rate": 9.12767338403929e-05, "loss": 0.0354, "step": 69400 }, { "epoch": 44.23836838750797, "grad_norm": 0.4695889949798584, "learning_rate": 9.127362312962786e-05, "loss": 0.0264, "step": 69410 }, { "epoch": 44.24474187380497, "grad_norm": 0.5995346307754517, "learning_rate": 9.127051191734933e-05, "loss": 0.0312, "step": 69420 }, { "epoch": 44.25111536010198, "grad_norm": 0.5833046436309814, "learning_rate": 9.126740020359512e-05, "loss": 0.031, "step": 69430 }, { "epoch": 44.25748884639898, "grad_norm": 0.3650643229484558, "learning_rate": 9.126428798840302e-05, "loss": 0.0386, "step": 69440 }, { "epoch": 44.263862332695986, "grad_norm": 0.7167417407035828, "learning_rate": 9.126117527181086e-05, "loss": 0.0178, "step": 69450 }, { "epoch": 44.27023581899299, "grad_norm": 0.3195766508579254, "learning_rate": 9.125806205385648e-05, "loss": 0.0178, "step": 69460 }, { "epoch": 44.276609305289995, "grad_norm": 0.3028687834739685, "learning_rate": 9.125494833457769e-05, "loss": 0.0293, "step": 69470 }, { "epoch": 44.282982791586996, "grad_norm": 0.3424360156059265, "learning_rate": 9.125183411401233e-05, "loss": 0.018, "step": 69480 }, { "epoch": 44.289356277884, "grad_norm": 0.40720537304878235, "learning_rate": 9.124871939219821e-05, "loss": 0.0205, "step": 69490 }, { "epoch": 44.295729764181004, "grad_norm": 0.35510939359664917, "learning_rate": 9.124560416917323e-05, "loss": 0.024, "step": 69500 }, { "epoch": 44.30210325047801, "grad_norm": 0.45656684041023254, "learning_rate": 9.12424884449752e-05, "loss": 0.0173, "step": 69510 }, { "epoch": 44.30847673677501, "grad_norm": 0.357785701751709, "learning_rate": 9.1239372219642e-05, "loss": 0.0187, "step": 69520 }, { "epoch": 44.31485022307202, "grad_norm": 0.48116904497146606, "learning_rate": 9.123625549321151e-05, "loss": 0.0258, "step": 69530 }, { "epoch": 44.32122370936902, "grad_norm": 0.8287373781204224, "learning_rate": 9.123313826572156e-05, "loss": 0.0185, "step": 69540 }, { "epoch": 44.32759719566603, "grad_norm": 0.6212223768234253, "learning_rate": 9.123002053721005e-05, "loss": 0.0303, "step": 69550 }, { "epoch": 44.33397068196303, "grad_norm": 0.3443302512168884, "learning_rate": 9.122690230771488e-05, "loss": 0.0184, "step": 69560 }, { "epoch": 44.34034416826004, "grad_norm": 0.38666781783103943, "learning_rate": 9.122378357727393e-05, "loss": 0.0151, "step": 69570 }, { "epoch": 44.34671765455704, "grad_norm": 1.0148160457611084, "learning_rate": 9.122066434592506e-05, "loss": 0.0269, "step": 69580 }, { "epoch": 44.35309114085405, "grad_norm": 0.4918820858001709, "learning_rate": 9.12175446137062e-05, "loss": 0.0198, "step": 69590 }, { "epoch": 44.35946462715105, "grad_norm": 0.4328464865684509, "learning_rate": 9.121442438065528e-05, "loss": 0.0259, "step": 69600 }, { "epoch": 44.365838113448056, "grad_norm": 0.6793479323387146, "learning_rate": 9.121130364681016e-05, "loss": 0.0208, "step": 69610 }, { "epoch": 44.37221159974506, "grad_norm": 0.5032913684844971, "learning_rate": 9.120818241220882e-05, "loss": 0.0202, "step": 69620 }, { "epoch": 44.378585086042065, "grad_norm": 0.5774743556976318, "learning_rate": 9.120506067688914e-05, "loss": 0.0237, "step": 69630 }, { "epoch": 44.384958572339066, "grad_norm": 0.3497629761695862, "learning_rate": 9.120193844088908e-05, "loss": 0.0198, "step": 69640 }, { "epoch": 44.391332058636074, "grad_norm": 0.3199697732925415, "learning_rate": 9.119881570424656e-05, "loss": 0.0232, "step": 69650 }, { "epoch": 44.397705544933075, "grad_norm": 0.30733513832092285, "learning_rate": 9.119569246699952e-05, "loss": 0.0239, "step": 69660 }, { "epoch": 44.40407903123008, "grad_norm": 0.47568878531455994, "learning_rate": 9.119256872918594e-05, "loss": 0.0285, "step": 69670 }, { "epoch": 44.41045251752709, "grad_norm": 0.2987971901893616, "learning_rate": 9.118944449084377e-05, "loss": 0.0302, "step": 69680 }, { "epoch": 44.41682600382409, "grad_norm": 0.45306047797203064, "learning_rate": 9.118631975201092e-05, "loss": 0.0287, "step": 69690 }, { "epoch": 44.4231994901211, "grad_norm": 0.3489025831222534, "learning_rate": 9.118319451272544e-05, "loss": 0.0259, "step": 69700 }, { "epoch": 44.4295729764181, "grad_norm": 0.32040104269981384, "learning_rate": 9.118006877302524e-05, "loss": 0.0285, "step": 69710 }, { "epoch": 44.43594646271511, "grad_norm": 0.3904116451740265, "learning_rate": 9.117694253294833e-05, "loss": 0.0246, "step": 69720 }, { "epoch": 44.44231994901211, "grad_norm": 0.6685142517089844, "learning_rate": 9.117381579253268e-05, "loss": 0.0375, "step": 69730 }, { "epoch": 44.44869343530912, "grad_norm": 0.2558314800262451, "learning_rate": 9.11706885518163e-05, "loss": 0.0239, "step": 69740 }, { "epoch": 44.45506692160612, "grad_norm": 0.7195074558258057, "learning_rate": 9.11675608108372e-05, "loss": 0.022, "step": 69750 }, { "epoch": 44.461440407903126, "grad_norm": 0.6453682780265808, "learning_rate": 9.116443256963333e-05, "loss": 0.0263, "step": 69760 }, { "epoch": 44.46781389420013, "grad_norm": 0.26266562938690186, "learning_rate": 9.116130382824277e-05, "loss": 0.0232, "step": 69770 }, { "epoch": 44.474187380497135, "grad_norm": 0.29991114139556885, "learning_rate": 9.11581745867035e-05, "loss": 0.0211, "step": 69780 }, { "epoch": 44.480560866794136, "grad_norm": 0.619339644908905, "learning_rate": 9.115504484505355e-05, "loss": 0.0195, "step": 69790 }, { "epoch": 44.486934353091144, "grad_norm": 0.3754052221775055, "learning_rate": 9.115191460333095e-05, "loss": 0.0326, "step": 69800 }, { "epoch": 44.493307839388144, "grad_norm": 0.6344118118286133, "learning_rate": 9.114878386157372e-05, "loss": 0.0256, "step": 69810 }, { "epoch": 44.49968132568515, "grad_norm": 0.4744247496128082, "learning_rate": 9.114565261981992e-05, "loss": 0.0198, "step": 69820 }, { "epoch": 44.50605481198215, "grad_norm": 0.30229365825653076, "learning_rate": 9.114252087810759e-05, "loss": 0.023, "step": 69830 }, { "epoch": 44.51242829827916, "grad_norm": 0.5179222822189331, "learning_rate": 9.11393886364748e-05, "loss": 0.0294, "step": 69840 }, { "epoch": 44.51880178457616, "grad_norm": 0.40976962447166443, "learning_rate": 9.113625589495958e-05, "loss": 0.0274, "step": 69850 }, { "epoch": 44.52517527087317, "grad_norm": 0.36087876558303833, "learning_rate": 9.113312265360003e-05, "loss": 0.0243, "step": 69860 }, { "epoch": 44.53154875717017, "grad_norm": 0.3029163181781769, "learning_rate": 9.112998891243419e-05, "loss": 0.0234, "step": 69870 }, { "epoch": 44.53792224346718, "grad_norm": 0.7869796752929688, "learning_rate": 9.112685467150015e-05, "loss": 0.0235, "step": 69880 }, { "epoch": 44.54429572976418, "grad_norm": 0.7458316683769226, "learning_rate": 9.1123719930836e-05, "loss": 0.0261, "step": 69890 }, { "epoch": 44.55066921606119, "grad_norm": 0.4367789328098297, "learning_rate": 9.112058469047982e-05, "loss": 0.038, "step": 69900 }, { "epoch": 44.55704270235819, "grad_norm": 0.45899897813796997, "learning_rate": 9.111744895046972e-05, "loss": 0.0182, "step": 69910 }, { "epoch": 44.5634161886552, "grad_norm": 0.29348084330558777, "learning_rate": 9.11143127108438e-05, "loss": 0.0265, "step": 69920 }, { "epoch": 44.5697896749522, "grad_norm": 0.4629710018634796, "learning_rate": 9.111117597164015e-05, "loss": 0.0295, "step": 69930 }, { "epoch": 44.576163161249205, "grad_norm": 0.4618318974971771, "learning_rate": 9.11080387328969e-05, "loss": 0.0307, "step": 69940 }, { "epoch": 44.582536647546206, "grad_norm": 0.5685538649559021, "learning_rate": 9.110490099465216e-05, "loss": 0.0284, "step": 69950 }, { "epoch": 44.588910133843214, "grad_norm": 1.0322154760360718, "learning_rate": 9.11017627569441e-05, "loss": 0.0235, "step": 69960 }, { "epoch": 44.595283620140215, "grad_norm": 0.5729140639305115, "learning_rate": 9.109862401981077e-05, "loss": 0.0241, "step": 69970 }, { "epoch": 44.60165710643722, "grad_norm": 0.32351556420326233, "learning_rate": 9.109548478329037e-05, "loss": 0.0255, "step": 69980 }, { "epoch": 44.608030592734224, "grad_norm": 0.31515541672706604, "learning_rate": 9.109234504742103e-05, "loss": 0.0194, "step": 69990 }, { "epoch": 44.61440407903123, "grad_norm": 0.6408838033676147, "learning_rate": 9.10892048122409e-05, "loss": 0.0162, "step": 70000 }, { "epoch": 44.62077756532823, "grad_norm": 0.3009513020515442, "learning_rate": 9.108606407778815e-05, "loss": 0.0225, "step": 70010 }, { "epoch": 44.62715105162524, "grad_norm": 1.4135018587112427, "learning_rate": 9.108292284410091e-05, "loss": 0.0166, "step": 70020 }, { "epoch": 44.63352453792224, "grad_norm": 0.5567973852157593, "learning_rate": 9.107978111121737e-05, "loss": 0.0244, "step": 70030 }, { "epoch": 44.63989802421925, "grad_norm": 0.44984617829322815, "learning_rate": 9.107663887917571e-05, "loss": 0.0214, "step": 70040 }, { "epoch": 44.64627151051625, "grad_norm": 0.3634715974330902, "learning_rate": 9.10734961480141e-05, "loss": 0.0208, "step": 70050 }, { "epoch": 44.65264499681326, "grad_norm": 0.4572800397872925, "learning_rate": 9.107035291777075e-05, "loss": 0.0185, "step": 70060 }, { "epoch": 44.65901848311026, "grad_norm": 0.4640272557735443, "learning_rate": 9.106720918848381e-05, "loss": 0.0321, "step": 70070 }, { "epoch": 44.66539196940727, "grad_norm": 0.3801213502883911, "learning_rate": 9.106406496019152e-05, "loss": 0.026, "step": 70080 }, { "epoch": 44.67176545570427, "grad_norm": 0.3513534665107727, "learning_rate": 9.106092023293206e-05, "loss": 0.0237, "step": 70090 }, { "epoch": 44.678138942001276, "grad_norm": 0.5299761891365051, "learning_rate": 9.105777500674366e-05, "loss": 0.0318, "step": 70100 }, { "epoch": 44.68451242829828, "grad_norm": 0.5048236846923828, "learning_rate": 9.105462928166452e-05, "loss": 0.0228, "step": 70110 }, { "epoch": 44.690885914595285, "grad_norm": 0.34549281001091003, "learning_rate": 9.10514830577329e-05, "loss": 0.021, "step": 70120 }, { "epoch": 44.697259400892285, "grad_norm": 0.5473931431770325, "learning_rate": 9.104833633498697e-05, "loss": 0.0301, "step": 70130 }, { "epoch": 44.70363288718929, "grad_norm": 0.7582613229751587, "learning_rate": 9.104518911346501e-05, "loss": 0.0327, "step": 70140 }, { "epoch": 44.710006373486294, "grad_norm": 0.5976056456565857, "learning_rate": 9.104204139320524e-05, "loss": 0.0313, "step": 70150 }, { "epoch": 44.7163798597833, "grad_norm": 0.7555539011955261, "learning_rate": 9.103889317424593e-05, "loss": 0.0452, "step": 70160 }, { "epoch": 44.7227533460803, "grad_norm": 0.40818914771080017, "learning_rate": 9.103574445662532e-05, "loss": 0.0282, "step": 70170 }, { "epoch": 44.72912683237731, "grad_norm": 0.3522981107234955, "learning_rate": 9.103259524038166e-05, "loss": 0.0274, "step": 70180 }, { "epoch": 44.73550031867431, "grad_norm": 0.4655044376850128, "learning_rate": 9.102944552555323e-05, "loss": 0.033, "step": 70190 }, { "epoch": 44.74187380497132, "grad_norm": 0.5036574602127075, "learning_rate": 9.10262953121783e-05, "loss": 0.0291, "step": 70200 }, { "epoch": 44.74824729126832, "grad_norm": 0.6618916392326355, "learning_rate": 9.102314460029516e-05, "loss": 0.0185, "step": 70210 }, { "epoch": 44.75462077756533, "grad_norm": 0.5666359663009644, "learning_rate": 9.101999338994207e-05, "loss": 0.0256, "step": 70220 }, { "epoch": 44.76099426386233, "grad_norm": 0.3985896110534668, "learning_rate": 9.101684168115732e-05, "loss": 0.0224, "step": 70230 }, { "epoch": 44.76736775015934, "grad_norm": 0.6574935913085938, "learning_rate": 9.101368947397923e-05, "loss": 0.0259, "step": 70240 }, { "epoch": 44.77374123645634, "grad_norm": 0.3646835684776306, "learning_rate": 9.101053676844608e-05, "loss": 0.0298, "step": 70250 }, { "epoch": 44.780114722753346, "grad_norm": 0.3888828754425049, "learning_rate": 9.10073835645962e-05, "loss": 0.0276, "step": 70260 }, { "epoch": 44.78648820905035, "grad_norm": 0.5211093425750732, "learning_rate": 9.100422986246788e-05, "loss": 0.0224, "step": 70270 }, { "epoch": 44.792861695347355, "grad_norm": 0.4625350832939148, "learning_rate": 9.100107566209946e-05, "loss": 0.0361, "step": 70280 }, { "epoch": 44.79923518164436, "grad_norm": 0.40586015582084656, "learning_rate": 9.099792096352926e-05, "loss": 0.0318, "step": 70290 }, { "epoch": 44.805608667941364, "grad_norm": 0.5788999795913696, "learning_rate": 9.099476576679563e-05, "loss": 0.0354, "step": 70300 }, { "epoch": 44.81198215423837, "grad_norm": 0.6112302541732788, "learning_rate": 9.099161007193687e-05, "loss": 0.0292, "step": 70310 }, { "epoch": 44.81835564053537, "grad_norm": 0.31626275181770325, "learning_rate": 9.098845387899133e-05, "loss": 0.0215, "step": 70320 }, { "epoch": 44.82472912683238, "grad_norm": 0.8335644602775574, "learning_rate": 9.09852971879974e-05, "loss": 0.0314, "step": 70330 }, { "epoch": 44.83110261312938, "grad_norm": 0.392109215259552, "learning_rate": 9.09821399989934e-05, "loss": 0.0277, "step": 70340 }, { "epoch": 44.83747609942639, "grad_norm": 0.5769377946853638, "learning_rate": 9.09789823120177e-05, "loss": 0.0218, "step": 70350 }, { "epoch": 44.84384958572339, "grad_norm": 0.4938219487667084, "learning_rate": 9.09758241271087e-05, "loss": 0.0271, "step": 70360 }, { "epoch": 44.8502230720204, "grad_norm": 0.6601452231407166, "learning_rate": 9.097266544430472e-05, "loss": 0.0204, "step": 70370 }, { "epoch": 44.8565965583174, "grad_norm": 0.4282577335834503, "learning_rate": 9.096950626364418e-05, "loss": 0.0193, "step": 70380 }, { "epoch": 44.86297004461441, "grad_norm": 0.5425766110420227, "learning_rate": 9.096634658516547e-05, "loss": 0.0192, "step": 70390 }, { "epoch": 44.86934353091141, "grad_norm": 0.2954298257827759, "learning_rate": 9.096318640890695e-05, "loss": 0.0209, "step": 70400 }, { "epoch": 44.875717017208416, "grad_norm": 0.2802054286003113, "learning_rate": 9.096002573490707e-05, "loss": 0.0238, "step": 70410 }, { "epoch": 44.88209050350542, "grad_norm": 0.3437669277191162, "learning_rate": 9.095686456320418e-05, "loss": 0.017, "step": 70420 }, { "epoch": 44.888463989802425, "grad_norm": 0.4398162066936493, "learning_rate": 9.095370289383672e-05, "loss": 0.0167, "step": 70430 }, { "epoch": 44.894837476099426, "grad_norm": 0.34611621499061584, "learning_rate": 9.09505407268431e-05, "loss": 0.0182, "step": 70440 }, { "epoch": 44.90121096239643, "grad_norm": 0.318350613117218, "learning_rate": 9.094737806226176e-05, "loss": 0.0284, "step": 70450 }, { "epoch": 44.907584448693434, "grad_norm": 0.8930920362472534, "learning_rate": 9.09442149001311e-05, "loss": 0.0191, "step": 70460 }, { "epoch": 44.91395793499044, "grad_norm": 0.8092803955078125, "learning_rate": 9.094105124048958e-05, "loss": 0.0187, "step": 70470 }, { "epoch": 44.92033142128744, "grad_norm": 0.35930967330932617, "learning_rate": 9.093788708337563e-05, "loss": 0.0277, "step": 70480 }, { "epoch": 44.92670490758445, "grad_norm": 0.6216056942939758, "learning_rate": 9.09347224288277e-05, "loss": 0.0189, "step": 70490 }, { "epoch": 44.93307839388145, "grad_norm": 0.22093836963176727, "learning_rate": 9.093155727688425e-05, "loss": 0.0222, "step": 70500 }, { "epoch": 44.93945188017846, "grad_norm": 0.5702199339866638, "learning_rate": 9.092839162758372e-05, "loss": 0.0308, "step": 70510 }, { "epoch": 44.94582536647546, "grad_norm": 0.2813703119754791, "learning_rate": 9.09252254809646e-05, "loss": 0.0216, "step": 70520 }, { "epoch": 44.95219885277247, "grad_norm": 0.3856235146522522, "learning_rate": 9.092205883706535e-05, "loss": 0.0172, "step": 70530 }, { "epoch": 44.95857233906947, "grad_norm": 0.5444739460945129, "learning_rate": 9.091889169592444e-05, "loss": 0.0234, "step": 70540 }, { "epoch": 44.96494582536648, "grad_norm": 0.5959101915359497, "learning_rate": 9.091572405758038e-05, "loss": 0.0285, "step": 70550 }, { "epoch": 44.97131931166348, "grad_norm": 0.47324520349502563, "learning_rate": 9.091255592207161e-05, "loss": 0.0283, "step": 70560 }, { "epoch": 44.977692797960486, "grad_norm": 0.5032578110694885, "learning_rate": 9.090938728943669e-05, "loss": 0.028, "step": 70570 }, { "epoch": 44.98406628425749, "grad_norm": 0.35291022062301636, "learning_rate": 9.090621815971408e-05, "loss": 0.0219, "step": 70580 }, { "epoch": 44.990439770554495, "grad_norm": 0.34902822971343994, "learning_rate": 9.09030485329423e-05, "loss": 0.0205, "step": 70590 }, { "epoch": 44.996813256851496, "grad_norm": 0.492389053106308, "learning_rate": 9.089987840915985e-05, "loss": 0.0204, "step": 70600 }, { "epoch": 45.003186743148504, "grad_norm": 0.374727725982666, "learning_rate": 9.089670778840527e-05, "loss": 0.0237, "step": 70610 }, { "epoch": 45.009560229445505, "grad_norm": 0.5217189192771912, "learning_rate": 9.089353667071707e-05, "loss": 0.0218, "step": 70620 }, { "epoch": 45.01593371574251, "grad_norm": 0.8215938210487366, "learning_rate": 9.089036505613379e-05, "loss": 0.0298, "step": 70630 }, { "epoch": 45.022307202039514, "grad_norm": 0.6175441741943359, "learning_rate": 9.088719294469397e-05, "loss": 0.0335, "step": 70640 }, { "epoch": 45.02868068833652, "grad_norm": 0.2065260410308838, "learning_rate": 9.088402033643615e-05, "loss": 0.0175, "step": 70650 }, { "epoch": 45.03505417463352, "grad_norm": 0.4584850072860718, "learning_rate": 9.088084723139889e-05, "loss": 0.0327, "step": 70660 }, { "epoch": 45.04142766093053, "grad_norm": 0.6576526761054993, "learning_rate": 9.087767362962074e-05, "loss": 0.0405, "step": 70670 }, { "epoch": 45.04780114722753, "grad_norm": 0.5055668354034424, "learning_rate": 9.087449953114025e-05, "loss": 0.0305, "step": 70680 }, { "epoch": 45.05417463352454, "grad_norm": 0.501064121723175, "learning_rate": 9.0871324935996e-05, "loss": 0.029, "step": 70690 }, { "epoch": 45.06054811982154, "grad_norm": 0.6340673565864563, "learning_rate": 9.086814984422657e-05, "loss": 0.0192, "step": 70700 }, { "epoch": 45.06692160611855, "grad_norm": 1.3853297233581543, "learning_rate": 9.086497425587053e-05, "loss": 0.0223, "step": 70710 }, { "epoch": 45.07329509241555, "grad_norm": 0.5656533241271973, "learning_rate": 9.086179817096648e-05, "loss": 0.0251, "step": 70720 }, { "epoch": 45.07966857871256, "grad_norm": 0.40715739130973816, "learning_rate": 9.085862158955299e-05, "loss": 0.0274, "step": 70730 }, { "epoch": 45.08604206500956, "grad_norm": 0.2692093849182129, "learning_rate": 9.085544451166867e-05, "loss": 0.0188, "step": 70740 }, { "epoch": 45.092415551306566, "grad_norm": 0.4276832342147827, "learning_rate": 9.085226693735211e-05, "loss": 0.0267, "step": 70750 }, { "epoch": 45.09878903760357, "grad_norm": 0.4307619631290436, "learning_rate": 9.084908886664196e-05, "loss": 0.0287, "step": 70760 }, { "epoch": 45.105162523900574, "grad_norm": 0.4087086617946625, "learning_rate": 9.084591029957681e-05, "loss": 0.0194, "step": 70770 }, { "epoch": 45.111536010197575, "grad_norm": 0.4947662949562073, "learning_rate": 9.084273123619529e-05, "loss": 0.0287, "step": 70780 }, { "epoch": 45.11790949649458, "grad_norm": 0.33628049492836, "learning_rate": 9.0839551676536e-05, "loss": 0.0298, "step": 70790 }, { "epoch": 45.124282982791584, "grad_norm": 0.5575718879699707, "learning_rate": 9.083637162063762e-05, "loss": 0.0407, "step": 70800 }, { "epoch": 45.13065646908859, "grad_norm": 0.5913023948669434, "learning_rate": 9.083319106853876e-05, "loss": 0.0372, "step": 70810 }, { "epoch": 45.13702995538559, "grad_norm": 0.6538296937942505, "learning_rate": 9.083001002027807e-05, "loss": 0.0289, "step": 70820 }, { "epoch": 45.1434034416826, "grad_norm": 0.5777148008346558, "learning_rate": 9.082682847589423e-05, "loss": 0.0236, "step": 70830 }, { "epoch": 45.1497769279796, "grad_norm": 0.363638311624527, "learning_rate": 9.082364643542585e-05, "loss": 0.0297, "step": 70840 }, { "epoch": 45.15615041427661, "grad_norm": 0.38163599371910095, "learning_rate": 9.082046389891164e-05, "loss": 0.0194, "step": 70850 }, { "epoch": 45.16252390057361, "grad_norm": 0.6348438858985901, "learning_rate": 9.081728086639025e-05, "loss": 0.0239, "step": 70860 }, { "epoch": 45.16889738687062, "grad_norm": 0.6211579442024231, "learning_rate": 9.081409733790036e-05, "loss": 0.0157, "step": 70870 }, { "epoch": 45.17527087316762, "grad_norm": 0.38792866468429565, "learning_rate": 9.081091331348064e-05, "loss": 0.037, "step": 70880 }, { "epoch": 45.18164435946463, "grad_norm": 0.41175323724746704, "learning_rate": 9.080772879316982e-05, "loss": 0.0261, "step": 70890 }, { "epoch": 45.18801784576163, "grad_norm": 0.3493616282939911, "learning_rate": 9.080454377700653e-05, "loss": 0.0297, "step": 70900 }, { "epoch": 45.194391332058636, "grad_norm": 0.4021454453468323, "learning_rate": 9.080135826502953e-05, "loss": 0.0295, "step": 70910 }, { "epoch": 45.20076481835564, "grad_norm": 0.36986207962036133, "learning_rate": 9.079817225727749e-05, "loss": 0.0248, "step": 70920 }, { "epoch": 45.207138304652645, "grad_norm": 1.6000581979751587, "learning_rate": 9.079498575378915e-05, "loss": 0.0539, "step": 70930 }, { "epoch": 45.21351179094965, "grad_norm": 0.46225082874298096, "learning_rate": 9.079179875460322e-05, "loss": 0.0236, "step": 70940 }, { "epoch": 45.219885277246654, "grad_norm": 0.35660475492477417, "learning_rate": 9.07886112597584e-05, "loss": 0.0264, "step": 70950 }, { "epoch": 45.22625876354366, "grad_norm": 0.28614985942840576, "learning_rate": 9.078542326929347e-05, "loss": 0.0273, "step": 70960 }, { "epoch": 45.23263224984066, "grad_norm": 0.39445197582244873, "learning_rate": 9.078223478324712e-05, "loss": 0.0231, "step": 70970 }, { "epoch": 45.23900573613767, "grad_norm": 0.36632266640663147, "learning_rate": 9.077904580165811e-05, "loss": 0.036, "step": 70980 }, { "epoch": 45.24537922243467, "grad_norm": 0.46131616830825806, "learning_rate": 9.07758563245652e-05, "loss": 0.0297, "step": 70990 }, { "epoch": 45.25175270873168, "grad_norm": 0.36648306250572205, "learning_rate": 9.077266635200715e-05, "loss": 0.0159, "step": 71000 }, { "epoch": 45.25812619502868, "grad_norm": 0.28990280628204346, "learning_rate": 9.076947588402268e-05, "loss": 0.0213, "step": 71010 }, { "epoch": 45.26449968132569, "grad_norm": 0.38565969467163086, "learning_rate": 9.076628492065061e-05, "loss": 0.0203, "step": 71020 }, { "epoch": 45.27087316762269, "grad_norm": 2.693587064743042, "learning_rate": 9.076309346192967e-05, "loss": 0.0243, "step": 71030 }, { "epoch": 45.2772466539197, "grad_norm": 0.37950462102890015, "learning_rate": 9.075990150789866e-05, "loss": 0.0257, "step": 71040 }, { "epoch": 45.2836201402167, "grad_norm": 0.6521598696708679, "learning_rate": 9.075670905859637e-05, "loss": 0.0313, "step": 71050 }, { "epoch": 45.289993626513706, "grad_norm": 0.6091110706329346, "learning_rate": 9.07535161140616e-05, "loss": 0.0331, "step": 71060 }, { "epoch": 45.29636711281071, "grad_norm": 0.3839631676673889, "learning_rate": 9.07503226743331e-05, "loss": 0.0467, "step": 71070 }, { "epoch": 45.302740599107715, "grad_norm": 0.34343329071998596, "learning_rate": 9.074712873944971e-05, "loss": 0.0191, "step": 71080 }, { "epoch": 45.309114085404715, "grad_norm": 0.3329848647117615, "learning_rate": 9.074393430945025e-05, "loss": 0.0203, "step": 71090 }, { "epoch": 45.31548757170172, "grad_norm": 0.5943746566772461, "learning_rate": 9.074073938437351e-05, "loss": 0.0242, "step": 71100 }, { "epoch": 45.321861057998724, "grad_norm": 0.6341114640235901, "learning_rate": 9.073754396425833e-05, "loss": 0.0199, "step": 71110 }, { "epoch": 45.32823454429573, "grad_norm": 0.390470027923584, "learning_rate": 9.073434804914352e-05, "loss": 0.0246, "step": 71120 }, { "epoch": 45.33460803059273, "grad_norm": 0.391428142786026, "learning_rate": 9.073115163906793e-05, "loss": 0.0398, "step": 71130 }, { "epoch": 45.34098151688974, "grad_norm": 0.27678635716438293, "learning_rate": 9.072795473407038e-05, "loss": 0.0314, "step": 71140 }, { "epoch": 45.34735500318674, "grad_norm": 0.3356377184391022, "learning_rate": 9.072475733418973e-05, "loss": 0.0195, "step": 71150 }, { "epoch": 45.35372848948375, "grad_norm": 0.3680795729160309, "learning_rate": 9.072155943946483e-05, "loss": 0.0209, "step": 71160 }, { "epoch": 45.36010197578075, "grad_norm": 0.4007362127304077, "learning_rate": 9.071836104993453e-05, "loss": 0.0146, "step": 71170 }, { "epoch": 45.36647546207776, "grad_norm": 0.6581327319145203, "learning_rate": 9.07151621656377e-05, "loss": 0.0233, "step": 71180 }, { "epoch": 45.37284894837476, "grad_norm": 0.33917173743247986, "learning_rate": 9.071196278661322e-05, "loss": 0.0144, "step": 71190 }, { "epoch": 45.37922243467177, "grad_norm": 0.5440648198127747, "learning_rate": 9.070876291289994e-05, "loss": 0.0374, "step": 71200 }, { "epoch": 45.38559592096877, "grad_norm": 0.636914074420929, "learning_rate": 9.070556254453676e-05, "loss": 0.0466, "step": 71210 }, { "epoch": 45.391969407265776, "grad_norm": 0.2487654834985733, "learning_rate": 9.070236168156256e-05, "loss": 0.0229, "step": 71220 }, { "epoch": 45.39834289356278, "grad_norm": 0.7482965588569641, "learning_rate": 9.069916032401624e-05, "loss": 0.0378, "step": 71230 }, { "epoch": 45.404716379859785, "grad_norm": 0.34821584820747375, "learning_rate": 9.069595847193669e-05, "loss": 0.0212, "step": 71240 }, { "epoch": 45.411089866156786, "grad_norm": 0.36341798305511475, "learning_rate": 9.069275612536283e-05, "loss": 0.0171, "step": 71250 }, { "epoch": 45.417463352453794, "grad_norm": 0.5606079697608948, "learning_rate": 9.068955328433355e-05, "loss": 0.0281, "step": 71260 }, { "epoch": 45.423836838750795, "grad_norm": 0.4267158806324005, "learning_rate": 9.06863499488878e-05, "loss": 0.0214, "step": 71270 }, { "epoch": 45.4302103250478, "grad_norm": 0.3178037703037262, "learning_rate": 9.068314611906447e-05, "loss": 0.0184, "step": 71280 }, { "epoch": 45.4365838113448, "grad_norm": 0.5893178582191467, "learning_rate": 9.067994179490251e-05, "loss": 0.0192, "step": 71290 }, { "epoch": 45.44295729764181, "grad_norm": 0.42788901925086975, "learning_rate": 9.067673697644084e-05, "loss": 0.0285, "step": 71300 }, { "epoch": 45.44933078393881, "grad_norm": 0.5455334186553955, "learning_rate": 9.067353166371841e-05, "loss": 0.0273, "step": 71310 }, { "epoch": 45.45570427023582, "grad_norm": 0.5173894166946411, "learning_rate": 9.067032585677418e-05, "loss": 0.0195, "step": 71320 }, { "epoch": 45.46207775653282, "grad_norm": 0.42463773488998413, "learning_rate": 9.066711955564708e-05, "loss": 0.0258, "step": 71330 }, { "epoch": 45.46845124282983, "grad_norm": 0.738262951374054, "learning_rate": 9.06639127603761e-05, "loss": 0.0198, "step": 71340 }, { "epoch": 45.47482472912683, "grad_norm": 0.6661819219589233, "learning_rate": 9.066070547100016e-05, "loss": 0.0362, "step": 71350 }, { "epoch": 45.48119821542384, "grad_norm": 0.641676664352417, "learning_rate": 9.065749768755826e-05, "loss": 0.0227, "step": 71360 }, { "epoch": 45.48757170172084, "grad_norm": 0.4392363727092743, "learning_rate": 9.06542894100894e-05, "loss": 0.0323, "step": 71370 }, { "epoch": 45.49394518801785, "grad_norm": 0.3640446066856384, "learning_rate": 9.065108063863253e-05, "loss": 0.0222, "step": 71380 }, { "epoch": 45.50031867431485, "grad_norm": 0.9032019972801208, "learning_rate": 9.064787137322664e-05, "loss": 0.0336, "step": 71390 }, { "epoch": 45.506692160611856, "grad_norm": 0.5808727145195007, "learning_rate": 9.064466161391075e-05, "loss": 0.0468, "step": 71400 }, { "epoch": 45.513065646908856, "grad_norm": 0.37410789728164673, "learning_rate": 9.064145136072384e-05, "loss": 0.0339, "step": 71410 }, { "epoch": 45.519439133205864, "grad_norm": 0.5197312235832214, "learning_rate": 9.063824061370491e-05, "loss": 0.03, "step": 71420 }, { "epoch": 45.525812619502865, "grad_norm": 0.5331740379333496, "learning_rate": 9.063502937289301e-05, "loss": 0.0321, "step": 71430 }, { "epoch": 45.53218610579987, "grad_norm": 0.3824875056743622, "learning_rate": 9.063181763832713e-05, "loss": 0.0257, "step": 71440 }, { "epoch": 45.538559592096874, "grad_norm": 0.4603721797466278, "learning_rate": 9.06286054100463e-05, "loss": 0.0155, "step": 71450 }, { "epoch": 45.54493307839388, "grad_norm": 0.4769715666770935, "learning_rate": 9.062539268808955e-05, "loss": 0.0267, "step": 71460 }, { "epoch": 45.55130656469088, "grad_norm": 0.3128666877746582, "learning_rate": 9.062217947249592e-05, "loss": 0.0251, "step": 71470 }, { "epoch": 45.55768005098789, "grad_norm": 0.3092348575592041, "learning_rate": 9.061896576330447e-05, "loss": 0.0217, "step": 71480 }, { "epoch": 45.56405353728489, "grad_norm": 0.7825839519500732, "learning_rate": 9.061575156055421e-05, "loss": 0.0184, "step": 71490 }, { "epoch": 45.5704270235819, "grad_norm": 0.5529515147209167, "learning_rate": 9.061253686428426e-05, "loss": 0.0279, "step": 71500 }, { "epoch": 45.5768005098789, "grad_norm": 0.4902040958404541, "learning_rate": 9.060932167453362e-05, "loss": 0.0252, "step": 71510 }, { "epoch": 45.58317399617591, "grad_norm": 0.6579200029373169, "learning_rate": 9.060610599134137e-05, "loss": 0.037, "step": 71520 }, { "epoch": 45.58954748247291, "grad_norm": 0.48239102959632874, "learning_rate": 9.060288981474661e-05, "loss": 0.0261, "step": 71530 }, { "epoch": 45.59592096876992, "grad_norm": 0.35501357913017273, "learning_rate": 9.059967314478838e-05, "loss": 0.0398, "step": 71540 }, { "epoch": 45.602294455066925, "grad_norm": 0.34689027070999146, "learning_rate": 9.059645598150582e-05, "loss": 0.0254, "step": 71550 }, { "epoch": 45.608667941363926, "grad_norm": 0.3267618417739868, "learning_rate": 9.059323832493798e-05, "loss": 0.03, "step": 71560 }, { "epoch": 45.615041427660934, "grad_norm": 0.38036423921585083, "learning_rate": 9.059002017512397e-05, "loss": 0.021, "step": 71570 }, { "epoch": 45.621414913957935, "grad_norm": 0.5270289182662964, "learning_rate": 9.058680153210289e-05, "loss": 0.0259, "step": 71580 }, { "epoch": 45.62778840025494, "grad_norm": 0.6625394225120544, "learning_rate": 9.058358239591384e-05, "loss": 0.0256, "step": 71590 }, { "epoch": 45.634161886551944, "grad_norm": 0.479209303855896, "learning_rate": 9.058036276659595e-05, "loss": 0.0317, "step": 71600 }, { "epoch": 45.64053537284895, "grad_norm": 0.25597789883613586, "learning_rate": 9.057714264418836e-05, "loss": 0.0186, "step": 71610 }, { "epoch": 45.64690885914595, "grad_norm": 0.4768673777580261, "learning_rate": 9.057392202873014e-05, "loss": 0.0237, "step": 71620 }, { "epoch": 45.65328234544296, "grad_norm": 0.27209198474884033, "learning_rate": 9.057070092026048e-05, "loss": 0.0226, "step": 71630 }, { "epoch": 45.65965583173996, "grad_norm": 0.402756929397583, "learning_rate": 9.056747931881851e-05, "loss": 0.0202, "step": 71640 }, { "epoch": 45.66602931803697, "grad_norm": 0.26985570788383484, "learning_rate": 9.056425722444335e-05, "loss": 0.0218, "step": 71650 }, { "epoch": 45.67240280433397, "grad_norm": 0.3981522023677826, "learning_rate": 9.056103463717415e-05, "loss": 0.0305, "step": 71660 }, { "epoch": 45.67877629063098, "grad_norm": 0.2474077343940735, "learning_rate": 9.055781155705011e-05, "loss": 0.0187, "step": 71670 }, { "epoch": 45.68514977692798, "grad_norm": 0.47016680240631104, "learning_rate": 9.055458798411035e-05, "loss": 0.0383, "step": 71680 }, { "epoch": 45.69152326322499, "grad_norm": 0.29472893476486206, "learning_rate": 9.055136391839407e-05, "loss": 0.0363, "step": 71690 }, { "epoch": 45.69789674952199, "grad_norm": 0.4056607782840729, "learning_rate": 9.054813935994042e-05, "loss": 0.0277, "step": 71700 }, { "epoch": 45.704270235818996, "grad_norm": 0.6031860709190369, "learning_rate": 9.05449143087886e-05, "loss": 0.0327, "step": 71710 }, { "epoch": 45.710643722116, "grad_norm": 0.467719703912735, "learning_rate": 9.054168876497778e-05, "loss": 0.0201, "step": 71720 }, { "epoch": 45.717017208413004, "grad_norm": 0.5735359787940979, "learning_rate": 9.053846272854718e-05, "loss": 0.0228, "step": 71730 }, { "epoch": 45.723390694710005, "grad_norm": 0.3167482018470764, "learning_rate": 9.053523619953596e-05, "loss": 0.0268, "step": 71740 }, { "epoch": 45.72976418100701, "grad_norm": 0.6318914294242859, "learning_rate": 9.053200917798337e-05, "loss": 0.0375, "step": 71750 }, { "epoch": 45.736137667304014, "grad_norm": 0.8060351610183716, "learning_rate": 9.05287816639286e-05, "loss": 0.0273, "step": 71760 }, { "epoch": 45.74251115360102, "grad_norm": 0.5795696973800659, "learning_rate": 9.052555365741086e-05, "loss": 0.0251, "step": 71770 }, { "epoch": 45.74888463989802, "grad_norm": 0.28991326689720154, "learning_rate": 9.052232515846936e-05, "loss": 0.0249, "step": 71780 }, { "epoch": 45.75525812619503, "grad_norm": 0.357660710811615, "learning_rate": 9.051909616714339e-05, "loss": 0.0262, "step": 71790 }, { "epoch": 45.76163161249203, "grad_norm": 0.2897481620311737, "learning_rate": 9.051586668347211e-05, "loss": 0.0202, "step": 71800 }, { "epoch": 45.76800509878904, "grad_norm": 0.25315046310424805, "learning_rate": 9.051263670749482e-05, "loss": 0.0244, "step": 71810 }, { "epoch": 45.77437858508604, "grad_norm": 0.6223523616790771, "learning_rate": 9.050940623925073e-05, "loss": 0.0227, "step": 71820 }, { "epoch": 45.78075207138305, "grad_norm": 0.3144519031047821, "learning_rate": 9.050617527877911e-05, "loss": 0.0303, "step": 71830 }, { "epoch": 45.78712555768005, "grad_norm": 0.5086790919303894, "learning_rate": 9.050294382611924e-05, "loss": 0.0298, "step": 71840 }, { "epoch": 45.79349904397706, "grad_norm": 0.3743646442890167, "learning_rate": 9.049971188131032e-05, "loss": 0.0176, "step": 71850 }, { "epoch": 45.79987253027406, "grad_norm": 0.5054063200950623, "learning_rate": 9.04964794443917e-05, "loss": 0.0376, "step": 71860 }, { "epoch": 45.806246016571066, "grad_norm": 0.5286169052124023, "learning_rate": 9.04932465154026e-05, "loss": 0.0285, "step": 71870 }, { "epoch": 45.81261950286807, "grad_norm": 0.477058082818985, "learning_rate": 9.049001309438232e-05, "loss": 0.0259, "step": 71880 }, { "epoch": 45.818992989165075, "grad_norm": 0.368804007768631, "learning_rate": 9.048677918137017e-05, "loss": 0.032, "step": 71890 }, { "epoch": 45.825366475462076, "grad_norm": 0.5150417685508728, "learning_rate": 9.04835447764054e-05, "loss": 0.0378, "step": 71900 }, { "epoch": 45.831739961759084, "grad_norm": 0.3546220362186432, "learning_rate": 9.048030987952735e-05, "loss": 0.0188, "step": 71910 }, { "epoch": 45.838113448056085, "grad_norm": 0.5040913820266724, "learning_rate": 9.047707449077533e-05, "loss": 0.0244, "step": 71920 }, { "epoch": 45.84448693435309, "grad_norm": 0.40532317757606506, "learning_rate": 9.047383861018862e-05, "loss": 0.0248, "step": 71930 }, { "epoch": 45.85086042065009, "grad_norm": 0.3688770532608032, "learning_rate": 9.047060223780657e-05, "loss": 0.0189, "step": 71940 }, { "epoch": 45.8572339069471, "grad_norm": 0.7196228504180908, "learning_rate": 9.046736537366848e-05, "loss": 0.0338, "step": 71950 }, { "epoch": 45.8636073932441, "grad_norm": 0.5405769944190979, "learning_rate": 9.04641280178137e-05, "loss": 0.0298, "step": 71960 }, { "epoch": 45.86998087954111, "grad_norm": 0.28096744418144226, "learning_rate": 9.046089017028156e-05, "loss": 0.0328, "step": 71970 }, { "epoch": 45.87635436583811, "grad_norm": 0.5554543137550354, "learning_rate": 9.04576518311114e-05, "loss": 0.027, "step": 71980 }, { "epoch": 45.88272785213512, "grad_norm": 0.6111116409301758, "learning_rate": 9.045441300034257e-05, "loss": 0.027, "step": 71990 }, { "epoch": 45.88910133843212, "grad_norm": 0.4526421129703522, "learning_rate": 9.045117367801444e-05, "loss": 0.0198, "step": 72000 }, { "epoch": 45.89547482472913, "grad_norm": 0.6276025176048279, "learning_rate": 9.044793386416635e-05, "loss": 0.0257, "step": 72010 }, { "epoch": 45.90184831102613, "grad_norm": 0.46438947319984436, "learning_rate": 9.044469355883767e-05, "loss": 0.0362, "step": 72020 }, { "epoch": 45.90822179732314, "grad_norm": 0.5982154011726379, "learning_rate": 9.044145276206778e-05, "loss": 0.0321, "step": 72030 }, { "epoch": 45.91459528362014, "grad_norm": 0.5239140391349792, "learning_rate": 9.043821147389606e-05, "loss": 0.0257, "step": 72040 }, { "epoch": 45.920968769917145, "grad_norm": 0.3900355100631714, "learning_rate": 9.043496969436188e-05, "loss": 0.0224, "step": 72050 }, { "epoch": 45.927342256214146, "grad_norm": 0.6715278029441833, "learning_rate": 9.043172742350465e-05, "loss": 0.0423, "step": 72060 }, { "epoch": 45.933715742511154, "grad_norm": 0.5185863971710205, "learning_rate": 9.042848466136376e-05, "loss": 0.0308, "step": 72070 }, { "epoch": 45.940089228808155, "grad_norm": 0.296200156211853, "learning_rate": 9.04252414079786e-05, "loss": 0.0223, "step": 72080 }, { "epoch": 45.94646271510516, "grad_norm": 0.618979811668396, "learning_rate": 9.04219976633886e-05, "loss": 0.0189, "step": 72090 }, { "epoch": 45.952836201402164, "grad_norm": 0.5431245565414429, "learning_rate": 9.041875342763315e-05, "loss": 0.0246, "step": 72100 }, { "epoch": 45.95920968769917, "grad_norm": 0.4358237087726593, "learning_rate": 9.041550870075171e-05, "loss": 0.0358, "step": 72110 }, { "epoch": 45.96558317399617, "grad_norm": 0.6650050282478333, "learning_rate": 9.041226348278365e-05, "loss": 0.0344, "step": 72120 }, { "epoch": 45.97195666029318, "grad_norm": 0.4044601321220398, "learning_rate": 9.040901777376844e-05, "loss": 0.0322, "step": 72130 }, { "epoch": 45.97833014659018, "grad_norm": 0.30342942476272583, "learning_rate": 9.040577157374553e-05, "loss": 0.0232, "step": 72140 }, { "epoch": 45.98470363288719, "grad_norm": 0.5635536313056946, "learning_rate": 9.040252488275433e-05, "loss": 0.0277, "step": 72150 }, { "epoch": 45.99107711918419, "grad_norm": 0.35416656732559204, "learning_rate": 9.039927770083432e-05, "loss": 0.0232, "step": 72160 }, { "epoch": 45.9974506054812, "grad_norm": 0.4141038656234741, "learning_rate": 9.039603002802493e-05, "loss": 0.0319, "step": 72170 }, { "epoch": 46.003824091778206, "grad_norm": 0.7099750638008118, "learning_rate": 9.039278186436563e-05, "loss": 0.023, "step": 72180 }, { "epoch": 46.01019757807521, "grad_norm": 0.33698150515556335, "learning_rate": 9.038953320989592e-05, "loss": 0.025, "step": 72190 }, { "epoch": 46.016571064372215, "grad_norm": 0.5433964133262634, "learning_rate": 9.038628406465522e-05, "loss": 0.031, "step": 72200 }, { "epoch": 46.022944550669216, "grad_norm": 0.48015496134757996, "learning_rate": 9.038303442868304e-05, "loss": 0.0342, "step": 72210 }, { "epoch": 46.029318036966224, "grad_norm": 0.5014916658401489, "learning_rate": 9.037978430201888e-05, "loss": 0.0276, "step": 72220 }, { "epoch": 46.035691523263225, "grad_norm": 0.4410669505596161, "learning_rate": 9.03765336847022e-05, "loss": 0.0177, "step": 72230 }, { "epoch": 46.04206500956023, "grad_norm": 0.6270255446434021, "learning_rate": 9.037328257677252e-05, "loss": 0.0373, "step": 72240 }, { "epoch": 46.04843849585723, "grad_norm": 0.411275178194046, "learning_rate": 9.037003097826935e-05, "loss": 0.0276, "step": 72250 }, { "epoch": 46.05481198215424, "grad_norm": 0.5081455111503601, "learning_rate": 9.036677888923217e-05, "loss": 0.0216, "step": 72260 }, { "epoch": 46.06118546845124, "grad_norm": 0.5085729360580444, "learning_rate": 9.036352630970052e-05, "loss": 0.0186, "step": 72270 }, { "epoch": 46.06755895474825, "grad_norm": 0.6005397439002991, "learning_rate": 9.036027323971392e-05, "loss": 0.023, "step": 72280 }, { "epoch": 46.07393244104525, "grad_norm": 0.3970995843410492, "learning_rate": 9.035701967931188e-05, "loss": 0.0207, "step": 72290 }, { "epoch": 46.08030592734226, "grad_norm": 0.2823580205440521, "learning_rate": 9.035376562853397e-05, "loss": 0.0207, "step": 72300 }, { "epoch": 46.08667941363926, "grad_norm": 0.5190545320510864, "learning_rate": 9.035051108741969e-05, "loss": 0.0198, "step": 72310 }, { "epoch": 46.09305289993627, "grad_norm": 0.29820096492767334, "learning_rate": 9.03472560560086e-05, "loss": 0.0252, "step": 72320 }, { "epoch": 46.09942638623327, "grad_norm": 0.25740161538124084, "learning_rate": 9.034400053434027e-05, "loss": 0.0245, "step": 72330 }, { "epoch": 46.10579987253028, "grad_norm": 0.5008454918861389, "learning_rate": 9.034074452245423e-05, "loss": 0.0297, "step": 72340 }, { "epoch": 46.11217335882728, "grad_norm": 0.3892051875591278, "learning_rate": 9.033748802039006e-05, "loss": 0.0295, "step": 72350 }, { "epoch": 46.118546845124285, "grad_norm": 0.44915884733200073, "learning_rate": 9.033423102818734e-05, "loss": 0.0151, "step": 72360 }, { "epoch": 46.124920331421286, "grad_norm": 0.46310245990753174, "learning_rate": 9.033097354588561e-05, "loss": 0.0167, "step": 72370 }, { "epoch": 46.131293817718294, "grad_norm": 0.48191672563552856, "learning_rate": 9.032771557352448e-05, "loss": 0.0484, "step": 72380 }, { "epoch": 46.137667304015295, "grad_norm": 0.32372158765792847, "learning_rate": 9.032445711114353e-05, "loss": 0.0157, "step": 72390 }, { "epoch": 46.1440407903123, "grad_norm": 0.331342488527298, "learning_rate": 9.032119815878236e-05, "loss": 0.0165, "step": 72400 }, { "epoch": 46.150414276609304, "grad_norm": 0.2668777406215668, "learning_rate": 9.031793871648056e-05, "loss": 0.0193, "step": 72410 }, { "epoch": 46.15678776290631, "grad_norm": 0.3436919152736664, "learning_rate": 9.031467878427773e-05, "loss": 0.0239, "step": 72420 }, { "epoch": 46.16316124920331, "grad_norm": 0.3849315941333771, "learning_rate": 9.03114183622135e-05, "loss": 0.0238, "step": 72430 }, { "epoch": 46.16953473550032, "grad_norm": 0.34982162714004517, "learning_rate": 9.030815745032746e-05, "loss": 0.021, "step": 72440 }, { "epoch": 46.17590822179732, "grad_norm": 0.4643505811691284, "learning_rate": 9.030489604865927e-05, "loss": 0.0213, "step": 72450 }, { "epoch": 46.18228170809433, "grad_norm": 0.46418124437332153, "learning_rate": 9.030163415724853e-05, "loss": 0.024, "step": 72460 }, { "epoch": 46.18865519439133, "grad_norm": 0.31765952706336975, "learning_rate": 9.02983717761349e-05, "loss": 0.0141, "step": 72470 }, { "epoch": 46.19502868068834, "grad_norm": 0.4196479022502899, "learning_rate": 9.029510890535797e-05, "loss": 0.0214, "step": 72480 }, { "epoch": 46.20140216698534, "grad_norm": 0.6092623472213745, "learning_rate": 9.029184554495745e-05, "loss": 0.0339, "step": 72490 }, { "epoch": 46.20777565328235, "grad_norm": 0.35390105843544006, "learning_rate": 9.028858169497295e-05, "loss": 0.0174, "step": 72500 }, { "epoch": 46.21414913957935, "grad_norm": 0.3529788851737976, "learning_rate": 9.028531735544415e-05, "loss": 0.0151, "step": 72510 }, { "epoch": 46.220522625876356, "grad_norm": 1.114708423614502, "learning_rate": 9.028205252641071e-05, "loss": 0.02, "step": 72520 }, { "epoch": 46.22689611217336, "grad_norm": 0.17391476035118103, "learning_rate": 9.027878720791229e-05, "loss": 0.0238, "step": 72530 }, { "epoch": 46.233269598470365, "grad_norm": 0.49776044487953186, "learning_rate": 9.02755213999886e-05, "loss": 0.018, "step": 72540 }, { "epoch": 46.239643084767366, "grad_norm": 0.4329456388950348, "learning_rate": 9.027225510267928e-05, "loss": 0.0263, "step": 72550 }, { "epoch": 46.246016571064374, "grad_norm": 0.4147505462169647, "learning_rate": 9.026898831602404e-05, "loss": 0.025, "step": 72560 }, { "epoch": 46.252390057361374, "grad_norm": 0.3755078911781311, "learning_rate": 9.026572104006257e-05, "loss": 0.0207, "step": 72570 }, { "epoch": 46.25876354365838, "grad_norm": 0.43202298879623413, "learning_rate": 9.026245327483456e-05, "loss": 0.0335, "step": 72580 }, { "epoch": 46.26513702995538, "grad_norm": 0.6159024238586426, "learning_rate": 9.025918502037975e-05, "loss": 0.0215, "step": 72590 }, { "epoch": 46.27151051625239, "grad_norm": 0.8211655616760254, "learning_rate": 9.025591627673782e-05, "loss": 0.0241, "step": 72600 }, { "epoch": 46.27788400254939, "grad_norm": 0.27816665172576904, "learning_rate": 9.025264704394851e-05, "loss": 0.0131, "step": 72610 }, { "epoch": 46.2842574888464, "grad_norm": 0.3577656149864197, "learning_rate": 9.024937732205152e-05, "loss": 0.0228, "step": 72620 }, { "epoch": 46.2906309751434, "grad_norm": 0.520428478717804, "learning_rate": 9.024610711108659e-05, "loss": 0.0254, "step": 72630 }, { "epoch": 46.29700446144041, "grad_norm": 0.2858421802520752, "learning_rate": 9.024283641109348e-05, "loss": 0.0296, "step": 72640 }, { "epoch": 46.30337794773741, "grad_norm": 0.5425296425819397, "learning_rate": 9.02395652221119e-05, "loss": 0.0314, "step": 72650 }, { "epoch": 46.30975143403442, "grad_norm": 0.37338337302207947, "learning_rate": 9.023629354418162e-05, "loss": 0.028, "step": 72660 }, { "epoch": 46.31612492033142, "grad_norm": 0.5902696251869202, "learning_rate": 9.023302137734238e-05, "loss": 0.0196, "step": 72670 }, { "epoch": 46.322498406628426, "grad_norm": 2.0285696983337402, "learning_rate": 9.022974872163392e-05, "loss": 0.0223, "step": 72680 }, { "epoch": 46.32887189292543, "grad_norm": 0.36439642310142517, "learning_rate": 9.022647557709604e-05, "loss": 0.0292, "step": 72690 }, { "epoch": 46.335245379222435, "grad_norm": 0.39056140184402466, "learning_rate": 9.022320194376853e-05, "loss": 0.0164, "step": 72700 }, { "epoch": 46.341618865519436, "grad_norm": 0.6784292459487915, "learning_rate": 9.021992782169112e-05, "loss": 0.0287, "step": 72710 }, { "epoch": 46.347992351816444, "grad_norm": 0.7538209557533264, "learning_rate": 9.021665321090361e-05, "loss": 0.0331, "step": 72720 }, { "epoch": 46.354365838113445, "grad_norm": 0.6099622249603271, "learning_rate": 9.02133781114458e-05, "loss": 0.0214, "step": 72730 }, { "epoch": 46.36073932441045, "grad_norm": 0.4085283577442169, "learning_rate": 9.021010252335747e-05, "loss": 0.026, "step": 72740 }, { "epoch": 46.367112810707454, "grad_norm": 0.461270272731781, "learning_rate": 9.020682644667844e-05, "loss": 0.0303, "step": 72750 }, { "epoch": 46.37348629700446, "grad_norm": 0.4061415195465088, "learning_rate": 9.02035498814485e-05, "loss": 0.0372, "step": 72760 }, { "epoch": 46.37985978330146, "grad_norm": 0.3742014467716217, "learning_rate": 9.020027282770747e-05, "loss": 0.0224, "step": 72770 }, { "epoch": 46.38623326959847, "grad_norm": 1.1309475898742676, "learning_rate": 9.019699528549518e-05, "loss": 0.0326, "step": 72780 }, { "epoch": 46.39260675589547, "grad_norm": 0.3065764009952545, "learning_rate": 9.019371725485144e-05, "loss": 0.0166, "step": 72790 }, { "epoch": 46.39898024219248, "grad_norm": 0.5502302646636963, "learning_rate": 9.019043873581607e-05, "loss": 0.0212, "step": 72800 }, { "epoch": 46.40535372848949, "grad_norm": 0.4120645523071289, "learning_rate": 9.018715972842895e-05, "loss": 0.0186, "step": 72810 }, { "epoch": 46.41172721478649, "grad_norm": 0.3674103915691376, "learning_rate": 9.018388023272987e-05, "loss": 0.0168, "step": 72820 }, { "epoch": 46.418100701083496, "grad_norm": 0.4591769874095917, "learning_rate": 9.018060024875872e-05, "loss": 0.0156, "step": 72830 }, { "epoch": 46.4244741873805, "grad_norm": 0.34724441170692444, "learning_rate": 9.017731977655535e-05, "loss": 0.0143, "step": 72840 }, { "epoch": 46.430847673677505, "grad_norm": 0.2619026005268097, "learning_rate": 9.017403881615959e-05, "loss": 0.0211, "step": 72850 }, { "epoch": 46.437221159974506, "grad_norm": 0.5359959602355957, "learning_rate": 9.017075736761133e-05, "loss": 0.0222, "step": 72860 }, { "epoch": 46.443594646271514, "grad_norm": 0.3027917146682739, "learning_rate": 9.016747543095046e-05, "loss": 0.0235, "step": 72870 }, { "epoch": 46.449968132568515, "grad_norm": 0.4403434097766876, "learning_rate": 9.016419300621683e-05, "loss": 0.0201, "step": 72880 }, { "epoch": 46.45634161886552, "grad_norm": 0.30479562282562256, "learning_rate": 9.016091009345033e-05, "loss": 0.0264, "step": 72890 }, { "epoch": 46.46271510516252, "grad_norm": 0.3653997480869293, "learning_rate": 9.015762669269085e-05, "loss": 0.0283, "step": 72900 }, { "epoch": 46.46908859145953, "grad_norm": 0.33493807911872864, "learning_rate": 9.01543428039783e-05, "loss": 0.0193, "step": 72910 }, { "epoch": 46.47546207775653, "grad_norm": 0.6176809072494507, "learning_rate": 9.015105842735257e-05, "loss": 0.031, "step": 72920 }, { "epoch": 46.48183556405354, "grad_norm": 0.4378930628299713, "learning_rate": 9.014777356285356e-05, "loss": 0.0336, "step": 72930 }, { "epoch": 46.48820905035054, "grad_norm": 0.5343567728996277, "learning_rate": 9.01444882105212e-05, "loss": 0.0213, "step": 72940 }, { "epoch": 46.49458253664755, "grad_norm": 0.25798654556274414, "learning_rate": 9.01412023703954e-05, "loss": 0.0214, "step": 72950 }, { "epoch": 46.50095602294455, "grad_norm": 0.5395998954772949, "learning_rate": 9.013791604251611e-05, "loss": 0.0312, "step": 72960 }, { "epoch": 46.50732950924156, "grad_norm": 0.2777687907218933, "learning_rate": 9.013462922692324e-05, "loss": 0.0202, "step": 72970 }, { "epoch": 46.51370299553856, "grad_norm": 0.5118763446807861, "learning_rate": 9.013134192365673e-05, "loss": 0.028, "step": 72980 }, { "epoch": 46.52007648183557, "grad_norm": 0.5927591919898987, "learning_rate": 9.01280541327565e-05, "loss": 0.0354, "step": 72990 }, { "epoch": 46.52644996813257, "grad_norm": 0.4125881493091583, "learning_rate": 9.012476585426257e-05, "loss": 0.0235, "step": 73000 }, { "epoch": 46.532823454429575, "grad_norm": 0.4646165370941162, "learning_rate": 9.012147708821483e-05, "loss": 0.0275, "step": 73010 }, { "epoch": 46.539196940726576, "grad_norm": 0.28710806369781494, "learning_rate": 9.011818783465326e-05, "loss": 0.0343, "step": 73020 }, { "epoch": 46.545570427023584, "grad_norm": 0.23979824781417847, "learning_rate": 9.011489809361782e-05, "loss": 0.0205, "step": 73030 }, { "epoch": 46.551943913320585, "grad_norm": 0.3615192770957947, "learning_rate": 9.011160786514852e-05, "loss": 0.0273, "step": 73040 }, { "epoch": 46.55831739961759, "grad_norm": 0.7982115745544434, "learning_rate": 9.01083171492853e-05, "loss": 0.0183, "step": 73050 }, { "epoch": 46.564690885914594, "grad_norm": 0.5820039510726929, "learning_rate": 9.010502594606815e-05, "loss": 0.0147, "step": 73060 }, { "epoch": 46.5710643722116, "grad_norm": 0.314201295375824, "learning_rate": 9.010173425553708e-05, "loss": 0.0279, "step": 73070 }, { "epoch": 46.5774378585086, "grad_norm": 0.4855760335922241, "learning_rate": 9.009844207773206e-05, "loss": 0.0384, "step": 73080 }, { "epoch": 46.58381134480561, "grad_norm": 0.3538670837879181, "learning_rate": 9.009514941269313e-05, "loss": 0.0195, "step": 73090 }, { "epoch": 46.59018483110261, "grad_norm": 0.5878674983978271, "learning_rate": 9.009185626046026e-05, "loss": 0.0246, "step": 73100 }, { "epoch": 46.59655831739962, "grad_norm": 0.4360893964767456, "learning_rate": 9.008856262107349e-05, "loss": 0.0267, "step": 73110 }, { "epoch": 46.60293180369662, "grad_norm": 0.538303017616272, "learning_rate": 9.008526849457284e-05, "loss": 0.0285, "step": 73120 }, { "epoch": 46.60930528999363, "grad_norm": 0.3162635862827301, "learning_rate": 9.008197388099833e-05, "loss": 0.0131, "step": 73130 }, { "epoch": 46.61567877629063, "grad_norm": 0.4767117500305176, "learning_rate": 9.007867878038998e-05, "loss": 0.0197, "step": 73140 }, { "epoch": 46.62205226258764, "grad_norm": 0.6441197991371155, "learning_rate": 9.007538319278785e-05, "loss": 0.0348, "step": 73150 }, { "epoch": 46.62842574888464, "grad_norm": 0.5303916335105896, "learning_rate": 9.007208711823198e-05, "loss": 0.0249, "step": 73160 }, { "epoch": 46.634799235181646, "grad_norm": 0.4124069809913635, "learning_rate": 9.006879055676243e-05, "loss": 0.0254, "step": 73170 }, { "epoch": 46.64117272147865, "grad_norm": 0.7156108021736145, "learning_rate": 9.006549350841923e-05, "loss": 0.029, "step": 73180 }, { "epoch": 46.647546207775655, "grad_norm": 0.27211278676986694, "learning_rate": 9.006219597324245e-05, "loss": 0.0235, "step": 73190 }, { "epoch": 46.653919694072655, "grad_norm": 0.32286497950553894, "learning_rate": 9.005889795127219e-05, "loss": 0.0278, "step": 73200 }, { "epoch": 46.66029318036966, "grad_norm": 0.5292196273803711, "learning_rate": 9.005559944254846e-05, "loss": 0.0205, "step": 73210 }, { "epoch": 46.666666666666664, "grad_norm": 0.40824341773986816, "learning_rate": 9.00523004471114e-05, "loss": 0.0306, "step": 73220 }, { "epoch": 46.67304015296367, "grad_norm": 0.5618685483932495, "learning_rate": 9.004900096500107e-05, "loss": 0.0186, "step": 73230 }, { "epoch": 46.67941363926067, "grad_norm": 0.42674723267555237, "learning_rate": 9.004570099625758e-05, "loss": 0.0274, "step": 73240 }, { "epoch": 46.68578712555768, "grad_norm": 0.9148685336112976, "learning_rate": 9.004240054092099e-05, "loss": 0.0227, "step": 73250 }, { "epoch": 46.69216061185468, "grad_norm": 0.6640389561653137, "learning_rate": 9.003909959903144e-05, "loss": 0.03, "step": 73260 }, { "epoch": 46.69853409815169, "grad_norm": 0.7129845023155212, "learning_rate": 9.003579817062901e-05, "loss": 0.0194, "step": 73270 }, { "epoch": 46.70490758444869, "grad_norm": 0.270382821559906, "learning_rate": 9.003249625575385e-05, "loss": 0.0173, "step": 73280 }, { "epoch": 46.7112810707457, "grad_norm": 0.33116793632507324, "learning_rate": 9.002919385444607e-05, "loss": 0.0237, "step": 73290 }, { "epoch": 46.7176545570427, "grad_norm": 0.5110591053962708, "learning_rate": 9.002589096674578e-05, "loss": 0.017, "step": 73300 }, { "epoch": 46.72402804333971, "grad_norm": 0.5841027498245239, "learning_rate": 9.002258759269312e-05, "loss": 0.0234, "step": 73310 }, { "epoch": 46.73040152963671, "grad_norm": 0.29364582896232605, "learning_rate": 9.001928373232825e-05, "loss": 0.0204, "step": 73320 }, { "epoch": 46.736775015933716, "grad_norm": 0.49264290928840637, "learning_rate": 9.001597938569128e-05, "loss": 0.0207, "step": 73330 }, { "epoch": 46.74314850223072, "grad_norm": 0.24101972579956055, "learning_rate": 9.001267455282239e-05, "loss": 0.0271, "step": 73340 }, { "epoch": 46.749521988527725, "grad_norm": 0.4318997263908386, "learning_rate": 9.000936923376171e-05, "loss": 0.0201, "step": 73350 }, { "epoch": 46.755895474824726, "grad_norm": 0.47395551204681396, "learning_rate": 9.000606342854942e-05, "loss": 0.0214, "step": 73360 }, { "epoch": 46.762268961121734, "grad_norm": 0.3183608949184418, "learning_rate": 9.00027571372257e-05, "loss": 0.0261, "step": 73370 }, { "epoch": 46.768642447418735, "grad_norm": 0.4427333474159241, "learning_rate": 8.999945035983072e-05, "loss": 0.0163, "step": 73380 }, { "epoch": 46.77501593371574, "grad_norm": 0.43970638513565063, "learning_rate": 8.999614309640464e-05, "loss": 0.0278, "step": 73390 }, { "epoch": 46.781389420012744, "grad_norm": 0.38535189628601074, "learning_rate": 8.999283534698766e-05, "loss": 0.025, "step": 73400 }, { "epoch": 46.78776290630975, "grad_norm": 0.24869051575660706, "learning_rate": 8.998952711161996e-05, "loss": 0.0194, "step": 73410 }, { "epoch": 46.79413639260676, "grad_norm": 0.5560080409049988, "learning_rate": 8.998621839034177e-05, "loss": 0.0184, "step": 73420 }, { "epoch": 46.80050987890376, "grad_norm": 0.35762229561805725, "learning_rate": 8.998290918319327e-05, "loss": 0.0246, "step": 73430 }, { "epoch": 46.80688336520077, "grad_norm": 0.3314529061317444, "learning_rate": 8.997959949021466e-05, "loss": 0.0218, "step": 73440 }, { "epoch": 46.81325685149777, "grad_norm": 0.6552419066429138, "learning_rate": 8.997628931144619e-05, "loss": 0.0376, "step": 73450 }, { "epoch": 46.81963033779478, "grad_norm": 0.5220251679420471, "learning_rate": 8.997297864692803e-05, "loss": 0.0313, "step": 73460 }, { "epoch": 46.82600382409178, "grad_norm": 0.3886709213256836, "learning_rate": 8.996966749670046e-05, "loss": 0.0232, "step": 73470 }, { "epoch": 46.832377310388786, "grad_norm": 0.462576299905777, "learning_rate": 8.996635586080369e-05, "loss": 0.021, "step": 73480 }, { "epoch": 46.83875079668579, "grad_norm": 0.43018224835395813, "learning_rate": 8.996304373927795e-05, "loss": 0.025, "step": 73490 }, { "epoch": 46.845124282982795, "grad_norm": 0.3941669762134552, "learning_rate": 8.995973113216352e-05, "loss": 0.0368, "step": 73500 }, { "epoch": 46.851497769279796, "grad_norm": 0.5081268548965454, "learning_rate": 8.995641803950062e-05, "loss": 0.0173, "step": 73510 }, { "epoch": 46.8578712555768, "grad_norm": 0.3971559405326843, "learning_rate": 8.99531044613295e-05, "loss": 0.026, "step": 73520 }, { "epoch": 46.864244741873804, "grad_norm": 0.3755420446395874, "learning_rate": 8.994979039769046e-05, "loss": 0.025, "step": 73530 }, { "epoch": 46.87061822817081, "grad_norm": 0.3128087520599365, "learning_rate": 8.994647584862374e-05, "loss": 0.0288, "step": 73540 }, { "epoch": 46.87699171446781, "grad_norm": 0.4109993577003479, "learning_rate": 8.994316081416961e-05, "loss": 0.0206, "step": 73550 }, { "epoch": 46.88336520076482, "grad_norm": 0.6940052509307861, "learning_rate": 8.993984529436839e-05, "loss": 0.0225, "step": 73560 }, { "epoch": 46.88973868706182, "grad_norm": 0.5909662246704102, "learning_rate": 8.99365292892603e-05, "loss": 0.0202, "step": 73570 }, { "epoch": 46.89611217335883, "grad_norm": 0.358241468667984, "learning_rate": 8.993321279888571e-05, "loss": 0.0206, "step": 73580 }, { "epoch": 46.90248565965583, "grad_norm": 0.42378413677215576, "learning_rate": 8.992989582328485e-05, "loss": 0.0473, "step": 73590 }, { "epoch": 46.90885914595284, "grad_norm": 0.4851742386817932, "learning_rate": 8.992657836249809e-05, "loss": 0.0234, "step": 73600 }, { "epoch": 46.91523263224984, "grad_norm": 0.42154306173324585, "learning_rate": 8.992326041656568e-05, "loss": 0.019, "step": 73610 }, { "epoch": 46.92160611854685, "grad_norm": 0.426662802696228, "learning_rate": 8.991994198552798e-05, "loss": 0.0354, "step": 73620 }, { "epoch": 46.92797960484385, "grad_norm": 0.3295361399650574, "learning_rate": 8.991662306942529e-05, "loss": 0.0229, "step": 73630 }, { "epoch": 46.934353091140856, "grad_norm": 0.3389422297477722, "learning_rate": 8.991330366829794e-05, "loss": 0.0229, "step": 73640 }, { "epoch": 46.94072657743786, "grad_norm": 0.6542728543281555, "learning_rate": 8.990998378218627e-05, "loss": 0.0348, "step": 73650 }, { "epoch": 46.947100063734865, "grad_norm": 0.6729058027267456, "learning_rate": 8.990666341113061e-05, "loss": 0.0239, "step": 73660 }, { "epoch": 46.953473550031866, "grad_norm": 0.9905174970626831, "learning_rate": 8.99033425551713e-05, "loss": 0.0304, "step": 73670 }, { "epoch": 46.959847036328874, "grad_norm": 0.33427032828330994, "learning_rate": 8.990002121434874e-05, "loss": 0.0286, "step": 73680 }, { "epoch": 46.966220522625875, "grad_norm": 0.5040664076805115, "learning_rate": 8.989669938870321e-05, "loss": 0.0321, "step": 73690 }, { "epoch": 46.97259400892288, "grad_norm": 0.49050062894821167, "learning_rate": 8.989337707827513e-05, "loss": 0.0169, "step": 73700 }, { "epoch": 46.978967495219884, "grad_norm": 0.28033921122550964, "learning_rate": 8.989005428310483e-05, "loss": 0.028, "step": 73710 }, { "epoch": 46.98534098151689, "grad_norm": 0.56724613904953, "learning_rate": 8.988673100323274e-05, "loss": 0.03, "step": 73720 }, { "epoch": 46.99171446781389, "grad_norm": 0.3086795508861542, "learning_rate": 8.988340723869922e-05, "loss": 0.0205, "step": 73730 }, { "epoch": 46.9980879541109, "grad_norm": 0.30637165904045105, "learning_rate": 8.988008298954461e-05, "loss": 0.0235, "step": 73740 }, { "epoch": 47.0044614404079, "grad_norm": 0.5054007768630981, "learning_rate": 8.987675825580936e-05, "loss": 0.0222, "step": 73750 }, { "epoch": 47.01083492670491, "grad_norm": 0.7684495449066162, "learning_rate": 8.987343303753384e-05, "loss": 0.0199, "step": 73760 }, { "epoch": 47.01720841300191, "grad_norm": 0.3937392830848694, "learning_rate": 8.987010733475846e-05, "loss": 0.0239, "step": 73770 }, { "epoch": 47.02358189929892, "grad_norm": 0.3889522850513458, "learning_rate": 8.986678114752363e-05, "loss": 0.0357, "step": 73780 }, { "epoch": 47.02995538559592, "grad_norm": 0.7441005706787109, "learning_rate": 8.986345447586978e-05, "loss": 0.0234, "step": 73790 }, { "epoch": 47.03632887189293, "grad_norm": 0.3977629542350769, "learning_rate": 8.98601273198373e-05, "loss": 0.0267, "step": 73800 }, { "epoch": 47.04270235818993, "grad_norm": 0.453766793012619, "learning_rate": 8.985679967946666e-05, "loss": 0.019, "step": 73810 }, { "epoch": 47.049075844486936, "grad_norm": 0.5155267715454102, "learning_rate": 8.985347155479827e-05, "loss": 0.0257, "step": 73820 }, { "epoch": 47.05544933078394, "grad_norm": 0.4805154800415039, "learning_rate": 8.985014294587258e-05, "loss": 0.0259, "step": 73830 }, { "epoch": 47.061822817080945, "grad_norm": 0.4487517476081848, "learning_rate": 8.984681385273001e-05, "loss": 0.0195, "step": 73840 }, { "epoch": 47.068196303377945, "grad_norm": 0.4629252254962921, "learning_rate": 8.984348427541103e-05, "loss": 0.0268, "step": 73850 }, { "epoch": 47.07456978967495, "grad_norm": 0.44507139921188354, "learning_rate": 8.98401542139561e-05, "loss": 0.0216, "step": 73860 }, { "epoch": 47.080943275971954, "grad_norm": 0.6372135281562805, "learning_rate": 8.98368236684057e-05, "loss": 0.0259, "step": 73870 }, { "epoch": 47.08731676226896, "grad_norm": 0.854559063911438, "learning_rate": 8.983349263880026e-05, "loss": 0.0266, "step": 73880 }, { "epoch": 47.09369024856596, "grad_norm": 0.6928684711456299, "learning_rate": 8.983016112518028e-05, "loss": 0.0198, "step": 73890 }, { "epoch": 47.10006373486297, "grad_norm": 0.24849557876586914, "learning_rate": 8.982682912758624e-05, "loss": 0.02, "step": 73900 }, { "epoch": 47.10643722115997, "grad_norm": 0.32001641392707825, "learning_rate": 8.982349664605863e-05, "loss": 0.0146, "step": 73910 }, { "epoch": 47.11281070745698, "grad_norm": 0.39378491044044495, "learning_rate": 8.982016368063792e-05, "loss": 0.0176, "step": 73920 }, { "epoch": 47.11918419375398, "grad_norm": 0.3528071343898773, "learning_rate": 8.981683023136462e-05, "loss": 0.0193, "step": 73930 }, { "epoch": 47.12555768005099, "grad_norm": 0.3132566809654236, "learning_rate": 8.981349629827926e-05, "loss": 0.0275, "step": 73940 }, { "epoch": 47.13193116634799, "grad_norm": 0.49145981669425964, "learning_rate": 8.981016188142233e-05, "loss": 0.0259, "step": 73950 }, { "epoch": 47.138304652645, "grad_norm": 0.4064086377620697, "learning_rate": 8.980682698083433e-05, "loss": 0.0282, "step": 73960 }, { "epoch": 47.144678138942, "grad_norm": 0.7351500391960144, "learning_rate": 8.980349159655581e-05, "loss": 0.0153, "step": 73970 }, { "epoch": 47.151051625239006, "grad_norm": 0.5894414782524109, "learning_rate": 8.980015572862727e-05, "loss": 0.0253, "step": 73980 }, { "epoch": 47.15742511153601, "grad_norm": 0.280489057302475, "learning_rate": 8.979681937708928e-05, "loss": 0.0207, "step": 73990 }, { "epoch": 47.163798597833015, "grad_norm": 0.42340391874313354, "learning_rate": 8.979348254198235e-05, "loss": 0.0257, "step": 74000 }, { "epoch": 47.170172084130016, "grad_norm": 0.3818224370479584, "learning_rate": 8.979014522334704e-05, "loss": 0.0225, "step": 74010 }, { "epoch": 47.176545570427024, "grad_norm": 0.43937116861343384, "learning_rate": 8.97868074212239e-05, "loss": 0.0207, "step": 74020 }, { "epoch": 47.182919056724025, "grad_norm": 0.3366369307041168, "learning_rate": 8.978346913565348e-05, "loss": 0.023, "step": 74030 }, { "epoch": 47.18929254302103, "grad_norm": 0.6811960339546204, "learning_rate": 8.978013036667634e-05, "loss": 0.0307, "step": 74040 }, { "epoch": 47.19566602931803, "grad_norm": 0.2967533767223358, "learning_rate": 8.977679111433308e-05, "loss": 0.0168, "step": 74050 }, { "epoch": 47.20203951561504, "grad_norm": 0.4854520261287689, "learning_rate": 8.977345137866422e-05, "loss": 0.0433, "step": 74060 }, { "epoch": 47.20841300191205, "grad_norm": 0.31170758605003357, "learning_rate": 8.97701111597104e-05, "loss": 0.0214, "step": 74070 }, { "epoch": 47.21478648820905, "grad_norm": 0.8534892797470093, "learning_rate": 8.976677045751217e-05, "loss": 0.0259, "step": 74080 }, { "epoch": 47.22115997450606, "grad_norm": 0.5970094203948975, "learning_rate": 8.976342927211014e-05, "loss": 0.0275, "step": 74090 }, { "epoch": 47.22753346080306, "grad_norm": 0.7391400337219238, "learning_rate": 8.976008760354488e-05, "loss": 0.0322, "step": 74100 }, { "epoch": 47.23390694710007, "grad_norm": 0.38938653469085693, "learning_rate": 8.975674545185703e-05, "loss": 0.0242, "step": 74110 }, { "epoch": 47.24028043339707, "grad_norm": 0.5006521344184875, "learning_rate": 8.97534028170872e-05, "loss": 0.0191, "step": 74120 }, { "epoch": 47.246653919694076, "grad_norm": 0.40289443731307983, "learning_rate": 8.975005969927599e-05, "loss": 0.028, "step": 74130 }, { "epoch": 47.25302740599108, "grad_norm": 0.513492226600647, "learning_rate": 8.974671609846402e-05, "loss": 0.031, "step": 74140 }, { "epoch": 47.259400892288085, "grad_norm": 0.6376273036003113, "learning_rate": 8.97433720146919e-05, "loss": 0.0431, "step": 74150 }, { "epoch": 47.265774378585085, "grad_norm": 0.5154752731323242, "learning_rate": 8.974002744800033e-05, "loss": 0.0347, "step": 74160 }, { "epoch": 47.27214786488209, "grad_norm": 0.5322593450546265, "learning_rate": 8.973668239842986e-05, "loss": 0.0194, "step": 74170 }, { "epoch": 47.278521351179094, "grad_norm": 0.7278199195861816, "learning_rate": 8.97333368660212e-05, "loss": 0.0302, "step": 74180 }, { "epoch": 47.2848948374761, "grad_norm": 0.4009280502796173, "learning_rate": 8.972999085081498e-05, "loss": 0.0223, "step": 74190 }, { "epoch": 47.2912683237731, "grad_norm": 0.8206236362457275, "learning_rate": 8.972664435285188e-05, "loss": 0.0391, "step": 74200 }, { "epoch": 47.29764181007011, "grad_norm": 0.5483067631721497, "learning_rate": 8.972329737217251e-05, "loss": 0.0162, "step": 74210 }, { "epoch": 47.30401529636711, "grad_norm": 0.3799150884151459, "learning_rate": 8.971994990881759e-05, "loss": 0.0167, "step": 74220 }, { "epoch": 47.31038878266412, "grad_norm": 0.5094002485275269, "learning_rate": 8.971660196282779e-05, "loss": 0.0418, "step": 74230 }, { "epoch": 47.31676226896112, "grad_norm": 0.4540728032588959, "learning_rate": 8.971325353424374e-05, "loss": 0.0175, "step": 74240 }, { "epoch": 47.32313575525813, "grad_norm": 0.7567378878593445, "learning_rate": 8.97099046231062e-05, "loss": 0.0348, "step": 74250 }, { "epoch": 47.32950924155513, "grad_norm": 0.5790677070617676, "learning_rate": 8.970655522945581e-05, "loss": 0.0153, "step": 74260 }, { "epoch": 47.33588272785214, "grad_norm": 0.6918450593948364, "learning_rate": 8.970320535333327e-05, "loss": 0.0224, "step": 74270 }, { "epoch": 47.34225621414914, "grad_norm": 0.5672913193702698, "learning_rate": 8.969985499477933e-05, "loss": 0.0208, "step": 74280 }, { "epoch": 47.348629700446146, "grad_norm": 0.36000218987464905, "learning_rate": 8.969650415383466e-05, "loss": 0.0304, "step": 74290 }, { "epoch": 47.35500318674315, "grad_norm": 0.4054889976978302, "learning_rate": 8.969315283053997e-05, "loss": 0.0175, "step": 74300 }, { "epoch": 47.361376673040155, "grad_norm": 0.22036977112293243, "learning_rate": 8.9689801024936e-05, "loss": 0.015, "step": 74310 }, { "epoch": 47.367750159337156, "grad_norm": 0.48379793763160706, "learning_rate": 8.96864487370635e-05, "loss": 0.02, "step": 74320 }, { "epoch": 47.374123645634164, "grad_norm": 0.27999424934387207, "learning_rate": 8.968309596696315e-05, "loss": 0.0282, "step": 74330 }, { "epoch": 47.380497131931165, "grad_norm": 0.43151018023490906, "learning_rate": 8.967974271467572e-05, "loss": 0.0283, "step": 74340 }, { "epoch": 47.38687061822817, "grad_norm": 0.5295888781547546, "learning_rate": 8.967638898024195e-05, "loss": 0.0218, "step": 74350 }, { "epoch": 47.393244104525174, "grad_norm": 0.426802396774292, "learning_rate": 8.967303476370261e-05, "loss": 0.0218, "step": 74360 }, { "epoch": 47.39961759082218, "grad_norm": 0.47229161858558655, "learning_rate": 8.966968006509844e-05, "loss": 0.0299, "step": 74370 }, { "epoch": 47.40599107711918, "grad_norm": 0.4737984538078308, "learning_rate": 8.966632488447018e-05, "loss": 0.028, "step": 74380 }, { "epoch": 47.41236456341619, "grad_norm": 1.129098653793335, "learning_rate": 8.966296922185864e-05, "loss": 0.0389, "step": 74390 }, { "epoch": 47.41873804971319, "grad_norm": 0.3622661530971527, "learning_rate": 8.965961307730457e-05, "loss": 0.0242, "step": 74400 }, { "epoch": 47.4251115360102, "grad_norm": 0.5072131156921387, "learning_rate": 8.965625645084877e-05, "loss": 0.0213, "step": 74410 }, { "epoch": 47.4314850223072, "grad_norm": 0.6081326603889465, "learning_rate": 8.965289934253198e-05, "loss": 0.0362, "step": 74420 }, { "epoch": 47.43785850860421, "grad_norm": 0.5486676096916199, "learning_rate": 8.964954175239506e-05, "loss": 0.023, "step": 74430 }, { "epoch": 47.44423199490121, "grad_norm": 0.5025041699409485, "learning_rate": 8.964618368047875e-05, "loss": 0.0224, "step": 74440 }, { "epoch": 47.45060548119822, "grad_norm": 0.43800249695777893, "learning_rate": 8.964282512682387e-05, "loss": 0.0258, "step": 74450 }, { "epoch": 47.45697896749522, "grad_norm": 0.5994106531143188, "learning_rate": 8.963946609147126e-05, "loss": 0.0234, "step": 74460 }, { "epoch": 47.463352453792226, "grad_norm": 0.45363786816596985, "learning_rate": 8.963610657446169e-05, "loss": 0.0246, "step": 74470 }, { "epoch": 47.469725940089226, "grad_norm": 0.35016995668411255, "learning_rate": 8.963274657583602e-05, "loss": 0.0166, "step": 74480 }, { "epoch": 47.476099426386234, "grad_norm": 0.5277679562568665, "learning_rate": 8.962938609563504e-05, "loss": 0.0216, "step": 74490 }, { "epoch": 47.482472912683235, "grad_norm": 0.44823870062828064, "learning_rate": 8.962602513389961e-05, "loss": 0.0216, "step": 74500 }, { "epoch": 47.48884639898024, "grad_norm": 0.650191605091095, "learning_rate": 8.962266369067058e-05, "loss": 0.0155, "step": 74510 }, { "epoch": 47.495219885277244, "grad_norm": 0.6429213285446167, "learning_rate": 8.961930176598877e-05, "loss": 0.033, "step": 74520 }, { "epoch": 47.50159337157425, "grad_norm": 0.46865591406822205, "learning_rate": 8.961593935989502e-05, "loss": 0.0233, "step": 74530 }, { "epoch": 47.50796685787125, "grad_norm": 0.37498950958251953, "learning_rate": 8.961257647243021e-05, "loss": 0.0162, "step": 74540 }, { "epoch": 47.51434034416826, "grad_norm": 0.3605881929397583, "learning_rate": 8.960921310363519e-05, "loss": 0.0223, "step": 74550 }, { "epoch": 47.52071383046526, "grad_norm": 0.322897344827652, "learning_rate": 8.960584925355085e-05, "loss": 0.018, "step": 74560 }, { "epoch": 47.52708731676227, "grad_norm": 0.3989594280719757, "learning_rate": 8.960248492221804e-05, "loss": 0.0317, "step": 74570 }, { "epoch": 47.53346080305927, "grad_norm": 0.41885626316070557, "learning_rate": 8.959912010967764e-05, "loss": 0.0225, "step": 74580 }, { "epoch": 47.53983428935628, "grad_norm": 0.3031647205352783, "learning_rate": 8.959575481597055e-05, "loss": 0.0216, "step": 74590 }, { "epoch": 47.54620777565328, "grad_norm": 0.313166081905365, "learning_rate": 8.959238904113765e-05, "loss": 0.0245, "step": 74600 }, { "epoch": 47.55258126195029, "grad_norm": 0.4249531030654907, "learning_rate": 8.958902278521984e-05, "loss": 0.0212, "step": 74610 }, { "epoch": 47.55895474824729, "grad_norm": 0.27459752559661865, "learning_rate": 8.958565604825803e-05, "loss": 0.016, "step": 74620 }, { "epoch": 47.565328234544296, "grad_norm": 0.23955778777599335, "learning_rate": 8.958228883029312e-05, "loss": 0.0243, "step": 74630 }, { "epoch": 47.5717017208413, "grad_norm": 0.5925716757774353, "learning_rate": 8.957892113136604e-05, "loss": 0.0301, "step": 74640 }, { "epoch": 47.578075207138305, "grad_norm": 0.4620061218738556, "learning_rate": 8.95755529515177e-05, "loss": 0.025, "step": 74650 }, { "epoch": 47.584448693435306, "grad_norm": 0.40409213304519653, "learning_rate": 8.9572184290789e-05, "loss": 0.0253, "step": 74660 }, { "epoch": 47.590822179732314, "grad_norm": 0.5958356261253357, "learning_rate": 8.956881514922092e-05, "loss": 0.0392, "step": 74670 }, { "epoch": 47.59719566602932, "grad_norm": 0.23627549409866333, "learning_rate": 8.956544552685437e-05, "loss": 0.0284, "step": 74680 }, { "epoch": 47.60356915232632, "grad_norm": 0.5844635963439941, "learning_rate": 8.95620754237303e-05, "loss": 0.0445, "step": 74690 }, { "epoch": 47.60994263862333, "grad_norm": 0.6666755080223083, "learning_rate": 8.955870483988966e-05, "loss": 0.0253, "step": 74700 }, { "epoch": 47.61631612492033, "grad_norm": 0.7627792954444885, "learning_rate": 8.955533377537342e-05, "loss": 0.0289, "step": 74710 }, { "epoch": 47.62268961121734, "grad_norm": 0.5118746757507324, "learning_rate": 8.955196223022251e-05, "loss": 0.03, "step": 74720 }, { "epoch": 47.62906309751434, "grad_norm": 0.46787917613983154, "learning_rate": 8.954859020447793e-05, "loss": 0.0253, "step": 74730 }, { "epoch": 47.63543658381135, "grad_norm": 0.4045558273792267, "learning_rate": 8.954521769818063e-05, "loss": 0.0213, "step": 74740 }, { "epoch": 47.64181007010835, "grad_norm": 0.6612014174461365, "learning_rate": 8.954184471137161e-05, "loss": 0.03, "step": 74750 }, { "epoch": 47.64818355640536, "grad_norm": 0.5688436627388, "learning_rate": 8.953847124409183e-05, "loss": 0.0247, "step": 74760 }, { "epoch": 47.65455704270236, "grad_norm": 0.5196619033813477, "learning_rate": 8.95350972963823e-05, "loss": 0.0298, "step": 74770 }, { "epoch": 47.660930528999366, "grad_norm": 0.5087519884109497, "learning_rate": 8.953172286828401e-05, "loss": 0.0303, "step": 74780 }, { "epoch": 47.66730401529637, "grad_norm": 0.5044441223144531, "learning_rate": 8.952834795983796e-05, "loss": 0.0235, "step": 74790 }, { "epoch": 47.673677501593374, "grad_norm": 0.28789567947387695, "learning_rate": 8.952497257108517e-05, "loss": 0.0295, "step": 74800 }, { "epoch": 47.680050987890375, "grad_norm": 0.623721182346344, "learning_rate": 8.952159670206664e-05, "loss": 0.0443, "step": 74810 }, { "epoch": 47.68642447418738, "grad_norm": 0.5846380591392517, "learning_rate": 8.95182203528234e-05, "loss": 0.0386, "step": 74820 }, { "epoch": 47.692797960484384, "grad_norm": 0.44565868377685547, "learning_rate": 8.951484352339645e-05, "loss": 0.0308, "step": 74830 }, { "epoch": 47.69917144678139, "grad_norm": 1.4342551231384277, "learning_rate": 8.951146621382687e-05, "loss": 0.0256, "step": 74840 }, { "epoch": 47.70554493307839, "grad_norm": 0.25130876898765564, "learning_rate": 8.950808842415565e-05, "loss": 0.0218, "step": 74850 }, { "epoch": 47.7119184193754, "grad_norm": 0.6003896594047546, "learning_rate": 8.950471015442386e-05, "loss": 0.0193, "step": 74860 }, { "epoch": 47.7182919056724, "grad_norm": 0.38983699679374695, "learning_rate": 8.950133140467255e-05, "loss": 0.0276, "step": 74870 }, { "epoch": 47.72466539196941, "grad_norm": 0.29974237084388733, "learning_rate": 8.949795217494276e-05, "loss": 0.0302, "step": 74880 }, { "epoch": 47.73103887826641, "grad_norm": 0.4310964345932007, "learning_rate": 8.949457246527558e-05, "loss": 0.0168, "step": 74890 }, { "epoch": 47.73741236456342, "grad_norm": 0.7860645055770874, "learning_rate": 8.949119227571204e-05, "loss": 0.0264, "step": 74900 }, { "epoch": 47.74378585086042, "grad_norm": 0.3847382068634033, "learning_rate": 8.948781160629322e-05, "loss": 0.0157, "step": 74910 }, { "epoch": 47.75015933715743, "grad_norm": 0.41175761818885803, "learning_rate": 8.948443045706021e-05, "loss": 0.0205, "step": 74920 }, { "epoch": 47.75653282345443, "grad_norm": 0.30427005887031555, "learning_rate": 8.948104882805409e-05, "loss": 0.0213, "step": 74930 }, { "epoch": 47.762906309751436, "grad_norm": 0.2957451641559601, "learning_rate": 8.947766671931596e-05, "loss": 0.0247, "step": 74940 }, { "epoch": 47.76927979604844, "grad_norm": 0.290268212556839, "learning_rate": 8.94742841308869e-05, "loss": 0.0321, "step": 74950 }, { "epoch": 47.775653282345445, "grad_norm": 0.547484278678894, "learning_rate": 8.947090106280802e-05, "loss": 0.0276, "step": 74960 }, { "epoch": 47.782026768642446, "grad_norm": 0.476398766040802, "learning_rate": 8.946751751512041e-05, "loss": 0.0195, "step": 74970 }, { "epoch": 47.788400254939454, "grad_norm": 0.5073164105415344, "learning_rate": 8.946413348786522e-05, "loss": 0.024, "step": 74980 }, { "epoch": 47.794773741236455, "grad_norm": 0.2715703845024109, "learning_rate": 8.946074898108355e-05, "loss": 0.0183, "step": 74990 }, { "epoch": 47.80114722753346, "grad_norm": 0.49503853917121887, "learning_rate": 8.94573639948165e-05, "loss": 0.0215, "step": 75000 }, { "epoch": 47.80752071383046, "grad_norm": 0.3227577805519104, "learning_rate": 8.945397852910524e-05, "loss": 0.0238, "step": 75010 }, { "epoch": 47.81389420012747, "grad_norm": 0.7431999444961548, "learning_rate": 8.945059258399089e-05, "loss": 0.0162, "step": 75020 }, { "epoch": 47.82026768642447, "grad_norm": 0.48027366399765015, "learning_rate": 8.944720615951459e-05, "loss": 0.0227, "step": 75030 }, { "epoch": 47.82664117272148, "grad_norm": 0.4501822292804718, "learning_rate": 8.944381925571748e-05, "loss": 0.0221, "step": 75040 }, { "epoch": 47.83301465901848, "grad_norm": 0.5772968530654907, "learning_rate": 8.944043187264073e-05, "loss": 0.022, "step": 75050 }, { "epoch": 47.83938814531549, "grad_norm": 0.40132200717926025, "learning_rate": 8.943704401032551e-05, "loss": 0.0226, "step": 75060 }, { "epoch": 47.84576163161249, "grad_norm": 0.5347130298614502, "learning_rate": 8.943365566881295e-05, "loss": 0.0187, "step": 75070 }, { "epoch": 47.8521351179095, "grad_norm": 0.41869693994522095, "learning_rate": 8.943026684814426e-05, "loss": 0.0262, "step": 75080 }, { "epoch": 47.8585086042065, "grad_norm": 0.2702358663082123, "learning_rate": 8.942687754836059e-05, "loss": 0.0279, "step": 75090 }, { "epoch": 47.86488209050351, "grad_norm": 0.4160465598106384, "learning_rate": 8.942348776950313e-05, "loss": 0.0271, "step": 75100 }, { "epoch": 47.87125557680051, "grad_norm": 0.2977065443992615, "learning_rate": 8.942009751161308e-05, "loss": 0.0164, "step": 75110 }, { "epoch": 47.877629063097515, "grad_norm": 0.2532102167606354, "learning_rate": 8.941670677473162e-05, "loss": 0.0356, "step": 75120 }, { "epoch": 47.884002549394516, "grad_norm": 0.4975897967815399, "learning_rate": 8.941331555889995e-05, "loss": 0.0228, "step": 75130 }, { "epoch": 47.890376035691524, "grad_norm": 0.3004886209964752, "learning_rate": 8.94099238641593e-05, "loss": 0.0306, "step": 75140 }, { "epoch": 47.896749521988525, "grad_norm": 0.3163139820098877, "learning_rate": 8.940653169055087e-05, "loss": 0.0225, "step": 75150 }, { "epoch": 47.90312300828553, "grad_norm": 0.5864865183830261, "learning_rate": 8.940313903811586e-05, "loss": 0.0235, "step": 75160 }, { "epoch": 47.909496494582534, "grad_norm": 0.6935971975326538, "learning_rate": 8.939974590689551e-05, "loss": 0.0267, "step": 75170 }, { "epoch": 47.91586998087954, "grad_norm": 0.5338389873504639, "learning_rate": 8.939635229693104e-05, "loss": 0.026, "step": 75180 }, { "epoch": 47.92224346717654, "grad_norm": 0.35723260045051575, "learning_rate": 8.93929582082637e-05, "loss": 0.0347, "step": 75190 }, { "epoch": 47.92861695347355, "grad_norm": 0.3422093689441681, "learning_rate": 8.938956364093474e-05, "loss": 0.0188, "step": 75200 }, { "epoch": 47.93499043977055, "grad_norm": 0.25266242027282715, "learning_rate": 8.938616859498536e-05, "loss": 0.0157, "step": 75210 }, { "epoch": 47.94136392606756, "grad_norm": 0.9404997229576111, "learning_rate": 8.938277307045687e-05, "loss": 0.032, "step": 75220 }, { "epoch": 47.94773741236456, "grad_norm": 0.5825783610343933, "learning_rate": 8.937937706739051e-05, "loss": 0.0328, "step": 75230 }, { "epoch": 47.95411089866157, "grad_norm": 0.4462839961051941, "learning_rate": 8.937598058582751e-05, "loss": 0.0283, "step": 75240 }, { "epoch": 47.96048438495857, "grad_norm": 0.4645649790763855, "learning_rate": 8.937258362580918e-05, "loss": 0.0287, "step": 75250 }, { "epoch": 47.96685787125558, "grad_norm": 0.42328014969825745, "learning_rate": 8.93691861873768e-05, "loss": 0.0234, "step": 75260 }, { "epoch": 47.97323135755258, "grad_norm": 0.48635467886924744, "learning_rate": 8.936578827057163e-05, "loss": 0.0235, "step": 75270 }, { "epoch": 47.979604843849586, "grad_norm": 0.3894951641559601, "learning_rate": 8.936238987543496e-05, "loss": 0.022, "step": 75280 }, { "epoch": 47.98597833014659, "grad_norm": 0.5891890525817871, "learning_rate": 8.935899100200808e-05, "loss": 0.0345, "step": 75290 }, { "epoch": 47.992351816443595, "grad_norm": 0.4908401370048523, "learning_rate": 8.935559165033231e-05, "loss": 0.0233, "step": 75300 }, { "epoch": 47.998725302740596, "grad_norm": 0.5012348294258118, "learning_rate": 8.935219182044894e-05, "loss": 0.0252, "step": 75310 }, { "epoch": 48.0050987890376, "grad_norm": 1.4121203422546387, "learning_rate": 8.93487915123993e-05, "loss": 0.0244, "step": 75320 }, { "epoch": 48.01147227533461, "grad_norm": 0.45563772320747375, "learning_rate": 8.934539072622467e-05, "loss": 0.0204, "step": 75330 }, { "epoch": 48.01784576163161, "grad_norm": 0.5013715028762817, "learning_rate": 8.93419894619664e-05, "loss": 0.0217, "step": 75340 }, { "epoch": 48.02421924792862, "grad_norm": 0.5847480893135071, "learning_rate": 8.933858771966581e-05, "loss": 0.0174, "step": 75350 }, { "epoch": 48.03059273422562, "grad_norm": 0.547662615776062, "learning_rate": 8.933518549936427e-05, "loss": 0.025, "step": 75360 }, { "epoch": 48.03696622052263, "grad_norm": 0.2706519067287445, "learning_rate": 8.933178280110305e-05, "loss": 0.0193, "step": 75370 }, { "epoch": 48.04333970681963, "grad_norm": 0.5265756249427795, "learning_rate": 8.932837962492355e-05, "loss": 0.0207, "step": 75380 }, { "epoch": 48.04971319311664, "grad_norm": 0.7416671514511108, "learning_rate": 8.932497597086708e-05, "loss": 0.0231, "step": 75390 }, { "epoch": 48.05608667941364, "grad_norm": 0.52782142162323, "learning_rate": 8.932157183897505e-05, "loss": 0.0334, "step": 75400 }, { "epoch": 48.06246016571065, "grad_norm": 0.7380556464195251, "learning_rate": 8.931816722928878e-05, "loss": 0.0263, "step": 75410 }, { "epoch": 48.06883365200765, "grad_norm": 0.5533484220504761, "learning_rate": 8.931476214184968e-05, "loss": 0.0232, "step": 75420 }, { "epoch": 48.075207138304656, "grad_norm": 0.5061846375465393, "learning_rate": 8.931135657669909e-05, "loss": 0.0198, "step": 75430 }, { "epoch": 48.081580624601656, "grad_norm": 0.3923509120941162, "learning_rate": 8.930795053387839e-05, "loss": 0.0262, "step": 75440 }, { "epoch": 48.087954110898664, "grad_norm": 0.4133387804031372, "learning_rate": 8.930454401342899e-05, "loss": 0.0188, "step": 75450 }, { "epoch": 48.094327597195665, "grad_norm": 0.32283154129981995, "learning_rate": 8.930113701539227e-05, "loss": 0.0243, "step": 75460 }, { "epoch": 48.10070108349267, "grad_norm": 0.5459981560707092, "learning_rate": 8.929772953980962e-05, "loss": 0.0238, "step": 75470 }, { "epoch": 48.107074569789674, "grad_norm": 1.0120607614517212, "learning_rate": 8.929432158672245e-05, "loss": 0.0185, "step": 75480 }, { "epoch": 48.11344805608668, "grad_norm": 0.4208924472332001, "learning_rate": 8.929091315617219e-05, "loss": 0.0245, "step": 75490 }, { "epoch": 48.11982154238368, "grad_norm": 0.6432639956474304, "learning_rate": 8.928750424820023e-05, "loss": 0.0272, "step": 75500 }, { "epoch": 48.12619502868069, "grad_norm": 0.49686241149902344, "learning_rate": 8.928409486284799e-05, "loss": 0.0169, "step": 75510 }, { "epoch": 48.13256851497769, "grad_norm": 0.5811657309532166, "learning_rate": 8.928068500015691e-05, "loss": 0.0201, "step": 75520 }, { "epoch": 48.1389420012747, "grad_norm": 0.43043413758277893, "learning_rate": 8.927727466016843e-05, "loss": 0.0294, "step": 75530 }, { "epoch": 48.1453154875717, "grad_norm": 0.596208930015564, "learning_rate": 8.927386384292397e-05, "loss": 0.0338, "step": 75540 }, { "epoch": 48.15168897386871, "grad_norm": 0.5121863484382629, "learning_rate": 8.9270452548465e-05, "loss": 0.0195, "step": 75550 }, { "epoch": 48.15806246016571, "grad_norm": 0.597659170627594, "learning_rate": 8.926704077683294e-05, "loss": 0.0216, "step": 75560 }, { "epoch": 48.16443594646272, "grad_norm": 0.23879288136959076, "learning_rate": 8.926362852806926e-05, "loss": 0.0198, "step": 75570 }, { "epoch": 48.17080943275972, "grad_norm": 0.35727983713150024, "learning_rate": 8.926021580221542e-05, "loss": 0.0156, "step": 75580 }, { "epoch": 48.177182919056726, "grad_norm": 0.46775326132774353, "learning_rate": 8.925680259931291e-05, "loss": 0.0165, "step": 75590 }, { "epoch": 48.18355640535373, "grad_norm": 0.511959969997406, "learning_rate": 8.925338891940316e-05, "loss": 0.0167, "step": 75600 }, { "epoch": 48.189929891650735, "grad_norm": 0.3565732538700104, "learning_rate": 8.924997476252769e-05, "loss": 0.0296, "step": 75610 }, { "epoch": 48.196303377947736, "grad_norm": 0.4447449743747711, "learning_rate": 8.924656012872797e-05, "loss": 0.0213, "step": 75620 }, { "epoch": 48.202676864244744, "grad_norm": 0.6001524329185486, "learning_rate": 8.924314501804548e-05, "loss": 0.0232, "step": 75630 }, { "epoch": 48.209050350541744, "grad_norm": 0.41357290744781494, "learning_rate": 8.923972943052171e-05, "loss": 0.0209, "step": 75640 }, { "epoch": 48.21542383683875, "grad_norm": 0.2810899019241333, "learning_rate": 8.92363133661982e-05, "loss": 0.0138, "step": 75650 }, { "epoch": 48.22179732313575, "grad_norm": 0.39860740303993225, "learning_rate": 8.923289682511644e-05, "loss": 0.0234, "step": 75660 }, { "epoch": 48.22817080943276, "grad_norm": 0.3649909198284149, "learning_rate": 8.922947980731792e-05, "loss": 0.0309, "step": 75670 }, { "epoch": 48.23454429572976, "grad_norm": 0.8226469159126282, "learning_rate": 8.92260623128442e-05, "loss": 0.0224, "step": 75680 }, { "epoch": 48.24091778202677, "grad_norm": 0.6315596103668213, "learning_rate": 8.92226443417368e-05, "loss": 0.0274, "step": 75690 }, { "epoch": 48.24729126832377, "grad_norm": 0.47064143419265747, "learning_rate": 8.921922589403721e-05, "loss": 0.0327, "step": 75700 }, { "epoch": 48.25366475462078, "grad_norm": 0.5965138077735901, "learning_rate": 8.921580696978702e-05, "loss": 0.026, "step": 75710 }, { "epoch": 48.26003824091778, "grad_norm": 0.3230724632740021, "learning_rate": 8.921238756902773e-05, "loss": 0.0276, "step": 75720 }, { "epoch": 48.26641172721479, "grad_norm": 0.44348838925361633, "learning_rate": 8.920896769180092e-05, "loss": 0.0153, "step": 75730 }, { "epoch": 48.27278521351179, "grad_norm": 0.48899027705192566, "learning_rate": 8.920554733814812e-05, "loss": 0.0278, "step": 75740 }, { "epoch": 48.2791586998088, "grad_norm": 0.35626277327537537, "learning_rate": 8.920212650811093e-05, "loss": 0.0225, "step": 75750 }, { "epoch": 48.2855321861058, "grad_norm": 0.5385358929634094, "learning_rate": 8.919870520173085e-05, "loss": 0.0264, "step": 75760 }, { "epoch": 48.291905672402805, "grad_norm": 0.27742716670036316, "learning_rate": 8.919528341904953e-05, "loss": 0.0263, "step": 75770 }, { "epoch": 48.298279158699806, "grad_norm": 0.3483431935310364, "learning_rate": 8.91918611601085e-05, "loss": 0.0153, "step": 75780 }, { "epoch": 48.304652644996814, "grad_norm": 0.21175138652324677, "learning_rate": 8.918843842494935e-05, "loss": 0.0242, "step": 75790 }, { "epoch": 48.311026131293815, "grad_norm": 0.3293132483959198, "learning_rate": 8.918501521361366e-05, "loss": 0.0199, "step": 75800 }, { "epoch": 48.31739961759082, "grad_norm": 0.36684665083885193, "learning_rate": 8.918159152614306e-05, "loss": 0.0176, "step": 75810 }, { "epoch": 48.323773103887824, "grad_norm": 0.5040516257286072, "learning_rate": 8.917816736257911e-05, "loss": 0.0188, "step": 75820 }, { "epoch": 48.33014659018483, "grad_norm": 0.4116906225681305, "learning_rate": 8.917474272296345e-05, "loss": 0.0252, "step": 75830 }, { "epoch": 48.33652007648183, "grad_norm": 1.108932614326477, "learning_rate": 8.917131760733768e-05, "loss": 0.0428, "step": 75840 }, { "epoch": 48.34289356277884, "grad_norm": 0.6733537316322327, "learning_rate": 8.916789201574341e-05, "loss": 0.021, "step": 75850 }, { "epoch": 48.34926704907584, "grad_norm": 0.5461486577987671, "learning_rate": 8.916446594822228e-05, "loss": 0.0248, "step": 75860 }, { "epoch": 48.35564053537285, "grad_norm": 0.40634480118751526, "learning_rate": 8.91610394048159e-05, "loss": 0.0305, "step": 75870 }, { "epoch": 48.36201402166985, "grad_norm": 0.38543078303337097, "learning_rate": 8.915761238556592e-05, "loss": 0.0306, "step": 75880 }, { "epoch": 48.36838750796686, "grad_norm": 0.47182533144950867, "learning_rate": 8.9154184890514e-05, "loss": 0.0236, "step": 75890 }, { "epoch": 48.37476099426386, "grad_norm": 0.300396591424942, "learning_rate": 8.915075691970173e-05, "loss": 0.0145, "step": 75900 }, { "epoch": 48.38113448056087, "grad_norm": 0.5201929211616516, "learning_rate": 8.914732847317081e-05, "loss": 0.0322, "step": 75910 }, { "epoch": 48.38750796685787, "grad_norm": 0.3313845098018646, "learning_rate": 8.91438995509629e-05, "loss": 0.0178, "step": 75920 }, { "epoch": 48.393881453154876, "grad_norm": 0.24322479963302612, "learning_rate": 8.914047015311966e-05, "loss": 0.0229, "step": 75930 }, { "epoch": 48.400254939451884, "grad_norm": 0.5620152354240417, "learning_rate": 8.913704027968273e-05, "loss": 0.0208, "step": 75940 }, { "epoch": 48.406628425748885, "grad_norm": 0.4299810826778412, "learning_rate": 8.913360993069382e-05, "loss": 0.0162, "step": 75950 }, { "epoch": 48.41300191204589, "grad_norm": 0.4556967318058014, "learning_rate": 8.91301791061946e-05, "loss": 0.0217, "step": 75960 }, { "epoch": 48.41937539834289, "grad_norm": 0.29739418625831604, "learning_rate": 8.912674780622676e-05, "loss": 0.0156, "step": 75970 }, { "epoch": 48.4257488846399, "grad_norm": 0.27627599239349365, "learning_rate": 8.912331603083199e-05, "loss": 0.0157, "step": 75980 }, { "epoch": 48.4321223709369, "grad_norm": 0.6144986748695374, "learning_rate": 8.911988378005199e-05, "loss": 0.0234, "step": 75990 }, { "epoch": 48.43849585723391, "grad_norm": 0.3971882462501526, "learning_rate": 8.911645105392848e-05, "loss": 0.0291, "step": 76000 }, { "epoch": 48.44486934353091, "grad_norm": 0.5654749870300293, "learning_rate": 8.911301785250314e-05, "loss": 0.0163, "step": 76010 }, { "epoch": 48.45124282982792, "grad_norm": 0.6646308302879333, "learning_rate": 8.910958417581772e-05, "loss": 0.0163, "step": 76020 }, { "epoch": 48.45761631612492, "grad_norm": 0.28484082221984863, "learning_rate": 8.910615002391392e-05, "loss": 0.0225, "step": 76030 }, { "epoch": 48.46398980242193, "grad_norm": 0.4190088212490082, "learning_rate": 8.910271539683346e-05, "loss": 0.0197, "step": 76040 }, { "epoch": 48.47036328871893, "grad_norm": 0.7240654230117798, "learning_rate": 8.90992802946181e-05, "loss": 0.0243, "step": 76050 }, { "epoch": 48.47673677501594, "grad_norm": 0.4968497157096863, "learning_rate": 8.909584471730957e-05, "loss": 0.032, "step": 76060 }, { "epoch": 48.48311026131294, "grad_norm": 0.9180871248245239, "learning_rate": 8.909240866494961e-05, "loss": 0.0309, "step": 76070 }, { "epoch": 48.489483747609945, "grad_norm": 0.6101062297821045, "learning_rate": 8.908897213757999e-05, "loss": 0.0252, "step": 76080 }, { "epoch": 48.495857233906946, "grad_norm": 0.47214585542678833, "learning_rate": 8.908553513524245e-05, "loss": 0.0177, "step": 76090 }, { "epoch": 48.502230720203954, "grad_norm": 0.4123691916465759, "learning_rate": 8.908209765797873e-05, "loss": 0.0257, "step": 76100 }, { "epoch": 48.508604206500955, "grad_norm": 0.2346596121788025, "learning_rate": 8.907865970583064e-05, "loss": 0.0357, "step": 76110 }, { "epoch": 48.51497769279796, "grad_norm": 0.26319506764411926, "learning_rate": 8.907522127883995e-05, "loss": 0.0219, "step": 76120 }, { "epoch": 48.521351179094964, "grad_norm": 0.7729846239089966, "learning_rate": 8.907178237704842e-05, "loss": 0.0189, "step": 76130 }, { "epoch": 48.52772466539197, "grad_norm": 0.5811246037483215, "learning_rate": 8.906834300049785e-05, "loss": 0.0268, "step": 76140 }, { "epoch": 48.53409815168897, "grad_norm": 0.47579431533813477, "learning_rate": 8.906490314923e-05, "loss": 0.0212, "step": 76150 }, { "epoch": 48.54047163798598, "grad_norm": 0.49991121888160706, "learning_rate": 8.906146282328671e-05, "loss": 0.0256, "step": 76160 }, { "epoch": 48.54684512428298, "grad_norm": 0.41751834750175476, "learning_rate": 8.905802202270977e-05, "loss": 0.0231, "step": 76170 }, { "epoch": 48.55321861057999, "grad_norm": 0.7454197406768799, "learning_rate": 8.905458074754098e-05, "loss": 0.0273, "step": 76180 }, { "epoch": 48.55959209687699, "grad_norm": 0.4572162330150604, "learning_rate": 8.905113899782218e-05, "loss": 0.0192, "step": 76190 }, { "epoch": 48.565965583174, "grad_norm": 0.5171675682067871, "learning_rate": 8.904769677359515e-05, "loss": 0.0262, "step": 76200 }, { "epoch": 48.572339069471, "grad_norm": 0.6909810304641724, "learning_rate": 8.904425407490174e-05, "loss": 0.0222, "step": 76210 }, { "epoch": 48.57871255576801, "grad_norm": 0.5221519470214844, "learning_rate": 8.904081090178378e-05, "loss": 0.0188, "step": 76220 }, { "epoch": 48.58508604206501, "grad_norm": 0.4986597001552582, "learning_rate": 8.90373672542831e-05, "loss": 0.0167, "step": 76230 }, { "epoch": 48.591459528362016, "grad_norm": 0.9982235431671143, "learning_rate": 8.903392313244157e-05, "loss": 0.0207, "step": 76240 }, { "epoch": 48.59783301465902, "grad_norm": 0.3603552579879761, "learning_rate": 8.9030478536301e-05, "loss": 0.0196, "step": 76250 }, { "epoch": 48.604206500956025, "grad_norm": 0.45641520619392395, "learning_rate": 8.902703346590327e-05, "loss": 0.018, "step": 76260 }, { "epoch": 48.610579987253026, "grad_norm": 0.3008120059967041, "learning_rate": 8.902358792129024e-05, "loss": 0.0175, "step": 76270 }, { "epoch": 48.61695347355003, "grad_norm": 0.3910423219203949, "learning_rate": 8.902014190250377e-05, "loss": 0.0236, "step": 76280 }, { "epoch": 48.623326959847034, "grad_norm": 0.2935914099216461, "learning_rate": 8.901669540958574e-05, "loss": 0.0254, "step": 76290 }, { "epoch": 48.62970044614404, "grad_norm": 0.28027310967445374, "learning_rate": 8.901324844257802e-05, "loss": 0.0223, "step": 76300 }, { "epoch": 48.63607393244104, "grad_norm": 0.3035241961479187, "learning_rate": 8.90098010015225e-05, "loss": 0.025, "step": 76310 }, { "epoch": 48.64244741873805, "grad_norm": 0.24401028454303741, "learning_rate": 8.900635308646106e-05, "loss": 0.0221, "step": 76320 }, { "epoch": 48.64882090503505, "grad_norm": 0.30767902731895447, "learning_rate": 8.900290469743559e-05, "loss": 0.0234, "step": 76330 }, { "epoch": 48.65519439133206, "grad_norm": 0.3489046096801758, "learning_rate": 8.899945583448803e-05, "loss": 0.0267, "step": 76340 }, { "epoch": 48.66156787762906, "grad_norm": 0.6133447289466858, "learning_rate": 8.899600649766024e-05, "loss": 0.0245, "step": 76350 }, { "epoch": 48.66794136392607, "grad_norm": 0.779419481754303, "learning_rate": 8.899255668699415e-05, "loss": 0.0241, "step": 76360 }, { "epoch": 48.67431485022307, "grad_norm": 0.9319865107536316, "learning_rate": 8.898910640253171e-05, "loss": 0.0244, "step": 76370 }, { "epoch": 48.68068833652008, "grad_norm": 0.4049406349658966, "learning_rate": 8.898565564431478e-05, "loss": 0.0246, "step": 76380 }, { "epoch": 48.68706182281708, "grad_norm": 0.959450364112854, "learning_rate": 8.898220441238534e-05, "loss": 0.0302, "step": 76390 }, { "epoch": 48.693435309114086, "grad_norm": 0.4261667728424072, "learning_rate": 8.897875270678532e-05, "loss": 0.0186, "step": 76400 }, { "epoch": 48.69980879541109, "grad_norm": 0.6010489463806152, "learning_rate": 8.897530052755665e-05, "loss": 0.0241, "step": 76410 }, { "epoch": 48.706182281708095, "grad_norm": 0.3465341329574585, "learning_rate": 8.897184787474127e-05, "loss": 0.0348, "step": 76420 }, { "epoch": 48.712555768005096, "grad_norm": 0.35874900221824646, "learning_rate": 8.896839474838115e-05, "loss": 0.0249, "step": 76430 }, { "epoch": 48.718929254302104, "grad_norm": 0.4277724623680115, "learning_rate": 8.896494114851823e-05, "loss": 0.0239, "step": 76440 }, { "epoch": 48.725302740599105, "grad_norm": 0.41374707221984863, "learning_rate": 8.89614870751945e-05, "loss": 0.0194, "step": 76450 }, { "epoch": 48.73167622689611, "grad_norm": 0.3206664025783539, "learning_rate": 8.895803252845192e-05, "loss": 0.0249, "step": 76460 }, { "epoch": 48.738049713193114, "grad_norm": 0.3847193419933319, "learning_rate": 8.895457750833244e-05, "loss": 0.0191, "step": 76470 }, { "epoch": 48.74442319949012, "grad_norm": 0.236180379986763, "learning_rate": 8.895112201487808e-05, "loss": 0.0274, "step": 76480 }, { "epoch": 48.75079668578712, "grad_norm": 0.3643247187137604, "learning_rate": 8.89476660481308e-05, "loss": 0.0211, "step": 76490 }, { "epoch": 48.75717017208413, "grad_norm": 0.2318602204322815, "learning_rate": 8.894420960813263e-05, "loss": 0.0158, "step": 76500 }, { "epoch": 48.76354365838113, "grad_norm": 0.4744431972503662, "learning_rate": 8.894075269492552e-05, "loss": 0.0273, "step": 76510 }, { "epoch": 48.76991714467814, "grad_norm": 0.3607683777809143, "learning_rate": 8.89372953085515e-05, "loss": 0.0197, "step": 76520 }, { "epoch": 48.77629063097514, "grad_norm": 0.39541012048721313, "learning_rate": 8.893383744905259e-05, "loss": 0.0233, "step": 76530 }, { "epoch": 48.78266411727215, "grad_norm": 0.8160480856895447, "learning_rate": 8.893037911647079e-05, "loss": 0.0268, "step": 76540 }, { "epoch": 48.78903760356915, "grad_norm": 0.30967459082603455, "learning_rate": 8.892692031084814e-05, "loss": 0.028, "step": 76550 }, { "epoch": 48.79541108986616, "grad_norm": 0.700473964214325, "learning_rate": 8.892346103222664e-05, "loss": 0.029, "step": 76560 }, { "epoch": 48.80178457616316, "grad_norm": 0.3504991829395294, "learning_rate": 8.892000128064834e-05, "loss": 0.0228, "step": 76570 }, { "epoch": 48.808158062460166, "grad_norm": 0.31831562519073486, "learning_rate": 8.891654105615528e-05, "loss": 0.0154, "step": 76580 }, { "epoch": 48.814531548757174, "grad_norm": 0.44795316457748413, "learning_rate": 8.891308035878951e-05, "loss": 0.0265, "step": 76590 }, { "epoch": 48.820905035054174, "grad_norm": 0.35997816920280457, "learning_rate": 8.890961918859306e-05, "loss": 0.0243, "step": 76600 }, { "epoch": 48.82727852135118, "grad_norm": 0.5554684400558472, "learning_rate": 8.890615754560801e-05, "loss": 0.0303, "step": 76610 }, { "epoch": 48.83365200764818, "grad_norm": 0.5572734475135803, "learning_rate": 8.890269542987642e-05, "loss": 0.0223, "step": 76620 }, { "epoch": 48.84002549394519, "grad_norm": 0.4752501845359802, "learning_rate": 8.889923284144033e-05, "loss": 0.034, "step": 76630 }, { "epoch": 48.84639898024219, "grad_norm": 0.26858168840408325, "learning_rate": 8.889576978034184e-05, "loss": 0.0188, "step": 76640 }, { "epoch": 48.8527724665392, "grad_norm": 0.3426305055618286, "learning_rate": 8.889230624662303e-05, "loss": 0.0169, "step": 76650 }, { "epoch": 48.8591459528362, "grad_norm": 0.4153650403022766, "learning_rate": 8.888884224032598e-05, "loss": 0.0329, "step": 76660 }, { "epoch": 48.86551943913321, "grad_norm": 0.30574050545692444, "learning_rate": 8.888537776149277e-05, "loss": 0.0271, "step": 76670 }, { "epoch": 48.87189292543021, "grad_norm": 0.4339156746864319, "learning_rate": 8.888191281016551e-05, "loss": 0.0158, "step": 76680 }, { "epoch": 48.87826641172722, "grad_norm": 0.3017025589942932, "learning_rate": 8.88784473863863e-05, "loss": 0.0293, "step": 76690 }, { "epoch": 48.88463989802422, "grad_norm": 0.40968942642211914, "learning_rate": 8.887498149019725e-05, "loss": 0.0253, "step": 76700 }, { "epoch": 48.89101338432123, "grad_norm": 0.883637011051178, "learning_rate": 8.887151512164048e-05, "loss": 0.0212, "step": 76710 }, { "epoch": 48.89738687061823, "grad_norm": 0.4192010462284088, "learning_rate": 8.886804828075807e-05, "loss": 0.0275, "step": 76720 }, { "epoch": 48.903760356915235, "grad_norm": 0.3322286605834961, "learning_rate": 8.88645809675922e-05, "loss": 0.0227, "step": 76730 }, { "epoch": 48.910133843212236, "grad_norm": 0.3261466920375824, "learning_rate": 8.886111318218498e-05, "loss": 0.0191, "step": 76740 }, { "epoch": 48.916507329509244, "grad_norm": 0.3612256944179535, "learning_rate": 8.885764492457854e-05, "loss": 0.0218, "step": 76750 }, { "epoch": 48.922880815806245, "grad_norm": 0.6655124425888062, "learning_rate": 8.8854176194815e-05, "loss": 0.0264, "step": 76760 }, { "epoch": 48.92925430210325, "grad_norm": 0.29641762375831604, "learning_rate": 8.885070699293656e-05, "loss": 0.0199, "step": 76770 }, { "epoch": 48.935627788400254, "grad_norm": 0.5378060340881348, "learning_rate": 8.884723731898536e-05, "loss": 0.0371, "step": 76780 }, { "epoch": 48.94200127469726, "grad_norm": 0.46489793062210083, "learning_rate": 8.884376717300354e-05, "loss": 0.0196, "step": 76790 }, { "epoch": 48.94837476099426, "grad_norm": 0.37712621688842773, "learning_rate": 8.884029655503325e-05, "loss": 0.0262, "step": 76800 }, { "epoch": 48.95474824729127, "grad_norm": 0.43126267194747925, "learning_rate": 8.883682546511671e-05, "loss": 0.0143, "step": 76810 }, { "epoch": 48.96112173358827, "grad_norm": 0.22369031608104706, "learning_rate": 8.883335390329607e-05, "loss": 0.0166, "step": 76820 }, { "epoch": 48.96749521988528, "grad_norm": 0.6358847618103027, "learning_rate": 8.88298818696135e-05, "loss": 0.0153, "step": 76830 }, { "epoch": 48.97386870618228, "grad_norm": 0.35620570182800293, "learning_rate": 8.882640936411122e-05, "loss": 0.0203, "step": 76840 }, { "epoch": 48.98024219247929, "grad_norm": 0.27171969413757324, "learning_rate": 8.882293638683141e-05, "loss": 0.0312, "step": 76850 }, { "epoch": 48.98661567877629, "grad_norm": 0.47724440693855286, "learning_rate": 8.881946293781626e-05, "loss": 0.018, "step": 76860 }, { "epoch": 48.9929891650733, "grad_norm": 0.6344947814941406, "learning_rate": 8.881598901710798e-05, "loss": 0.0366, "step": 76870 }, { "epoch": 48.9993626513703, "grad_norm": 0.3690052628517151, "learning_rate": 8.88125146247488e-05, "loss": 0.0218, "step": 76880 }, { "epoch": 49.005736137667306, "grad_norm": 0.3196150064468384, "learning_rate": 8.88090397607809e-05, "loss": 0.02, "step": 76890 }, { "epoch": 49.01210962396431, "grad_norm": 0.2323322743177414, "learning_rate": 8.880556442524655e-05, "loss": 0.018, "step": 76900 }, { "epoch": 49.018483110261315, "grad_norm": 0.5877742171287537, "learning_rate": 8.880208861818792e-05, "loss": 0.0312, "step": 76910 }, { "epoch": 49.024856596558315, "grad_norm": 0.46226146817207336, "learning_rate": 8.879861233964732e-05, "loss": 0.0227, "step": 76920 }, { "epoch": 49.03123008285532, "grad_norm": 0.770878791809082, "learning_rate": 8.879513558966693e-05, "loss": 0.0215, "step": 76930 }, { "epoch": 49.037603569152324, "grad_norm": 0.6134500503540039, "learning_rate": 8.8791658368289e-05, "loss": 0.029, "step": 76940 }, { "epoch": 49.04397705544933, "grad_norm": 0.2950141727924347, "learning_rate": 8.87881806755558e-05, "loss": 0.0182, "step": 76950 }, { "epoch": 49.05035054174633, "grad_norm": 0.47258442640304565, "learning_rate": 8.878470251150959e-05, "loss": 0.0142, "step": 76960 }, { "epoch": 49.05672402804334, "grad_norm": 0.536088764667511, "learning_rate": 8.878122387619264e-05, "loss": 0.0293, "step": 76970 }, { "epoch": 49.06309751434034, "grad_norm": 0.4390946924686432, "learning_rate": 8.877774476964718e-05, "loss": 0.0311, "step": 76980 }, { "epoch": 49.06947100063735, "grad_norm": 0.5162464380264282, "learning_rate": 8.877426519191552e-05, "loss": 0.0187, "step": 76990 }, { "epoch": 49.07584448693435, "grad_norm": 0.6299289464950562, "learning_rate": 8.877078514303993e-05, "loss": 0.0364, "step": 77000 }, { "epoch": 49.08221797323136, "grad_norm": 0.5679315328598022, "learning_rate": 8.876730462306268e-05, "loss": 0.0266, "step": 77010 }, { "epoch": 49.08859145952836, "grad_norm": 0.29127684235572815, "learning_rate": 8.87638236320261e-05, "loss": 0.024, "step": 77020 }, { "epoch": 49.09496494582537, "grad_norm": 0.3191298842430115, "learning_rate": 8.876034216997245e-05, "loss": 0.0259, "step": 77030 }, { "epoch": 49.10133843212237, "grad_norm": 0.20821388065814972, "learning_rate": 8.875686023694404e-05, "loss": 0.0231, "step": 77040 }, { "epoch": 49.107711918419376, "grad_norm": 0.4768683612346649, "learning_rate": 8.875337783298321e-05, "loss": 0.0268, "step": 77050 }, { "epoch": 49.11408540471638, "grad_norm": 0.4767087399959564, "learning_rate": 8.874989495813223e-05, "loss": 0.0199, "step": 77060 }, { "epoch": 49.120458891013385, "grad_norm": 0.21784226596355438, "learning_rate": 8.874641161243345e-05, "loss": 0.0181, "step": 77070 }, { "epoch": 49.126832377310386, "grad_norm": 0.647822916507721, "learning_rate": 8.874292779592918e-05, "loss": 0.0171, "step": 77080 }, { "epoch": 49.133205863607394, "grad_norm": 0.6063865423202515, "learning_rate": 8.873944350866175e-05, "loss": 0.0197, "step": 77090 }, { "epoch": 49.139579349904395, "grad_norm": 0.4788207709789276, "learning_rate": 8.873595875067352e-05, "loss": 0.0304, "step": 77100 }, { "epoch": 49.1459528362014, "grad_norm": 0.1634960174560547, "learning_rate": 8.87324735220068e-05, "loss": 0.0202, "step": 77110 }, { "epoch": 49.1523263224984, "grad_norm": 0.6162344217300415, "learning_rate": 8.872898782270397e-05, "loss": 0.0251, "step": 77120 }, { "epoch": 49.15869980879541, "grad_norm": 0.5550015568733215, "learning_rate": 8.872550165280738e-05, "loss": 0.0218, "step": 77130 }, { "epoch": 49.16507329509241, "grad_norm": 0.6461644172668457, "learning_rate": 8.872201501235937e-05, "loss": 0.0279, "step": 77140 }, { "epoch": 49.17144678138942, "grad_norm": 0.38927990198135376, "learning_rate": 8.871852790140233e-05, "loss": 0.0329, "step": 77150 }, { "epoch": 49.17782026768642, "grad_norm": 0.2511272430419922, "learning_rate": 8.871504031997862e-05, "loss": 0.0194, "step": 77160 }, { "epoch": 49.18419375398343, "grad_norm": 0.5065391659736633, "learning_rate": 8.87115522681306e-05, "loss": 0.0173, "step": 77170 }, { "epoch": 49.19056724028043, "grad_norm": 0.44868627190589905, "learning_rate": 8.870806374590068e-05, "loss": 0.0262, "step": 77180 }, { "epoch": 49.19694072657744, "grad_norm": 0.7002864480018616, "learning_rate": 8.870457475333123e-05, "loss": 0.0203, "step": 77190 }, { "epoch": 49.203314212874446, "grad_norm": 0.36071133613586426, "learning_rate": 8.870108529046466e-05, "loss": 0.0209, "step": 77200 }, { "epoch": 49.20968769917145, "grad_norm": 0.2789326310157776, "learning_rate": 8.869759535734336e-05, "loss": 0.0167, "step": 77210 }, { "epoch": 49.216061185468455, "grad_norm": 0.43922707438468933, "learning_rate": 8.869410495400976e-05, "loss": 0.0179, "step": 77220 }, { "epoch": 49.222434671765456, "grad_norm": 0.23701466619968414, "learning_rate": 8.869061408050623e-05, "loss": 0.0187, "step": 77230 }, { "epoch": 49.22880815806246, "grad_norm": 0.5033986568450928, "learning_rate": 8.868712273687521e-05, "loss": 0.0248, "step": 77240 }, { "epoch": 49.235181644359464, "grad_norm": 0.4085116386413574, "learning_rate": 8.868363092315914e-05, "loss": 0.0168, "step": 77250 }, { "epoch": 49.24155513065647, "grad_norm": 0.9206916689872742, "learning_rate": 8.868013863940041e-05, "loss": 0.0211, "step": 77260 }, { "epoch": 49.24792861695347, "grad_norm": 0.262408584356308, "learning_rate": 8.867664588564148e-05, "loss": 0.0209, "step": 77270 }, { "epoch": 49.25430210325048, "grad_norm": 0.2950029671192169, "learning_rate": 8.867315266192479e-05, "loss": 0.0167, "step": 77280 }, { "epoch": 49.26067558954748, "grad_norm": 0.5051262974739075, "learning_rate": 8.866965896829278e-05, "loss": 0.0231, "step": 77290 }, { "epoch": 49.26704907584449, "grad_norm": 0.47850650548934937, "learning_rate": 8.866616480478791e-05, "loss": 0.0192, "step": 77300 }, { "epoch": 49.27342256214149, "grad_norm": 0.23890072107315063, "learning_rate": 8.866267017145262e-05, "loss": 0.0134, "step": 77310 }, { "epoch": 49.2797960484385, "grad_norm": 0.24545453488826752, "learning_rate": 8.865917506832939e-05, "loss": 0.028, "step": 77320 }, { "epoch": 49.2861695347355, "grad_norm": 0.2781742513179779, "learning_rate": 8.865567949546068e-05, "loss": 0.0259, "step": 77330 }, { "epoch": 49.29254302103251, "grad_norm": 0.6485303640365601, "learning_rate": 8.865218345288897e-05, "loss": 0.0174, "step": 77340 }, { "epoch": 49.29891650732951, "grad_norm": 0.7008166909217834, "learning_rate": 8.864868694065674e-05, "loss": 0.0267, "step": 77350 }, { "epoch": 49.305289993626516, "grad_norm": 0.48867708444595337, "learning_rate": 8.864518995880646e-05, "loss": 0.0324, "step": 77360 }, { "epoch": 49.31166347992352, "grad_norm": 0.3192881643772125, "learning_rate": 8.864169250738064e-05, "loss": 0.0214, "step": 77370 }, { "epoch": 49.318036966220525, "grad_norm": 0.3369359076023102, "learning_rate": 8.863819458642179e-05, "loss": 0.0142, "step": 77380 }, { "epoch": 49.324410452517526, "grad_norm": 0.41567790508270264, "learning_rate": 8.863469619597238e-05, "loss": 0.022, "step": 77390 }, { "epoch": 49.330783938814534, "grad_norm": 0.578463077545166, "learning_rate": 8.863119733607493e-05, "loss": 0.0185, "step": 77400 }, { "epoch": 49.337157425111535, "grad_norm": 0.4225253164768219, "learning_rate": 8.862769800677197e-05, "loss": 0.0283, "step": 77410 }, { "epoch": 49.34353091140854, "grad_norm": 0.2911081612110138, "learning_rate": 8.8624198208106e-05, "loss": 0.0307, "step": 77420 }, { "epoch": 49.349904397705544, "grad_norm": 0.37398651242256165, "learning_rate": 8.862069794011956e-05, "loss": 0.0164, "step": 77430 }, { "epoch": 49.35627788400255, "grad_norm": 0.541105329990387, "learning_rate": 8.861719720285517e-05, "loss": 0.0233, "step": 77440 }, { "epoch": 49.36265137029955, "grad_norm": 0.4153154790401459, "learning_rate": 8.861369599635538e-05, "loss": 0.0168, "step": 77450 }, { "epoch": 49.36902485659656, "grad_norm": 0.7136523127555847, "learning_rate": 8.861019432066272e-05, "loss": 0.0177, "step": 77460 }, { "epoch": 49.37539834289356, "grad_norm": 0.4418129622936249, "learning_rate": 8.860669217581975e-05, "loss": 0.0231, "step": 77470 }, { "epoch": 49.38177182919057, "grad_norm": 0.30205056071281433, "learning_rate": 8.860318956186901e-05, "loss": 0.0163, "step": 77480 }, { "epoch": 49.38814531548757, "grad_norm": 0.40903952717781067, "learning_rate": 8.859968647885309e-05, "loss": 0.0229, "step": 77490 }, { "epoch": 49.39451880178458, "grad_norm": 0.5170226097106934, "learning_rate": 8.859618292681451e-05, "loss": 0.0264, "step": 77500 }, { "epoch": 49.40089228808158, "grad_norm": 0.42853203415870667, "learning_rate": 8.859267890579587e-05, "loss": 0.0208, "step": 77510 }, { "epoch": 49.40726577437859, "grad_norm": 0.22943678498268127, "learning_rate": 8.858917441583976e-05, "loss": 0.0272, "step": 77520 }, { "epoch": 49.41363926067559, "grad_norm": 0.27444449067115784, "learning_rate": 8.858566945698874e-05, "loss": 0.0261, "step": 77530 }, { "epoch": 49.420012746972596, "grad_norm": 0.4407459795475006, "learning_rate": 8.858216402928541e-05, "loss": 0.0189, "step": 77540 }, { "epoch": 49.4263862332696, "grad_norm": 0.5925467014312744, "learning_rate": 8.857865813277235e-05, "loss": 0.019, "step": 77550 }, { "epoch": 49.432759719566604, "grad_norm": 0.4955221116542816, "learning_rate": 8.857515176749217e-05, "loss": 0.0262, "step": 77560 }, { "epoch": 49.439133205863605, "grad_norm": 0.4943479299545288, "learning_rate": 8.857164493348749e-05, "loss": 0.0372, "step": 77570 }, { "epoch": 49.44550669216061, "grad_norm": 0.4927414655685425, "learning_rate": 8.856813763080089e-05, "loss": 0.0266, "step": 77580 }, { "epoch": 49.451880178457614, "grad_norm": 0.669206976890564, "learning_rate": 8.8564629859475e-05, "loss": 0.0195, "step": 77590 }, { "epoch": 49.45825366475462, "grad_norm": 0.30219265818595886, "learning_rate": 8.856112161955245e-05, "loss": 0.0254, "step": 77600 }, { "epoch": 49.46462715105162, "grad_norm": 0.3279782831668854, "learning_rate": 8.855761291107587e-05, "loss": 0.0187, "step": 77610 }, { "epoch": 49.47100063734863, "grad_norm": 0.5357823371887207, "learning_rate": 8.85541037340879e-05, "loss": 0.023, "step": 77620 }, { "epoch": 49.47737412364563, "grad_norm": 0.9298370480537415, "learning_rate": 8.855059408863114e-05, "loss": 0.0257, "step": 77630 }, { "epoch": 49.48374760994264, "grad_norm": 0.4157021939754486, "learning_rate": 8.854708397474827e-05, "loss": 0.0224, "step": 77640 }, { "epoch": 49.49012109623964, "grad_norm": 0.453798770904541, "learning_rate": 8.854357339248196e-05, "loss": 0.034, "step": 77650 }, { "epoch": 49.49649458253665, "grad_norm": 0.5312669277191162, "learning_rate": 8.854006234187482e-05, "loss": 0.0231, "step": 77660 }, { "epoch": 49.50286806883365, "grad_norm": 0.7626932263374329, "learning_rate": 8.853655082296955e-05, "loss": 0.0251, "step": 77670 }, { "epoch": 49.50924155513066, "grad_norm": 0.4287428855895996, "learning_rate": 8.85330388358088e-05, "loss": 0.0228, "step": 77680 }, { "epoch": 49.51561504142766, "grad_norm": 0.5255755186080933, "learning_rate": 8.852952638043523e-05, "loss": 0.018, "step": 77690 }, { "epoch": 49.521988527724666, "grad_norm": 0.372689425945282, "learning_rate": 8.852601345689155e-05, "loss": 0.026, "step": 77700 }, { "epoch": 49.52836201402167, "grad_norm": 0.3128395974636078, "learning_rate": 8.852250006522042e-05, "loss": 0.0217, "step": 77710 }, { "epoch": 49.534735500318675, "grad_norm": 0.30768460035324097, "learning_rate": 8.851898620546456e-05, "loss": 0.0227, "step": 77720 }, { "epoch": 49.541108986615676, "grad_norm": 0.3918452262878418, "learning_rate": 8.851547187766665e-05, "loss": 0.0241, "step": 77730 }, { "epoch": 49.547482472912684, "grad_norm": 0.300589919090271, "learning_rate": 8.851195708186938e-05, "loss": 0.0167, "step": 77740 }, { "epoch": 49.553855959209685, "grad_norm": 0.35560616850852966, "learning_rate": 8.850844181811547e-05, "loss": 0.0198, "step": 77750 }, { "epoch": 49.56022944550669, "grad_norm": 0.403573215007782, "learning_rate": 8.850492608644763e-05, "loss": 0.0224, "step": 77760 }, { "epoch": 49.56660293180369, "grad_norm": 0.22804976999759674, "learning_rate": 8.85014098869086e-05, "loss": 0.0283, "step": 77770 }, { "epoch": 49.5729764181007, "grad_norm": 0.9392104148864746, "learning_rate": 8.849789321954109e-05, "loss": 0.0332, "step": 77780 }, { "epoch": 49.5793499043977, "grad_norm": 0.4760918915271759, "learning_rate": 8.849437608438782e-05, "loss": 0.0297, "step": 77790 }, { "epoch": 49.58572339069471, "grad_norm": 0.36500418186187744, "learning_rate": 8.849085848149152e-05, "loss": 0.0265, "step": 77800 }, { "epoch": 49.59209687699171, "grad_norm": 0.2690048813819885, "learning_rate": 8.848734041089497e-05, "loss": 0.0199, "step": 77810 }, { "epoch": 49.59847036328872, "grad_norm": 0.6772710680961609, "learning_rate": 8.84838218726409e-05, "loss": 0.027, "step": 77820 }, { "epoch": 49.60484384958572, "grad_norm": 0.5453587770462036, "learning_rate": 8.848030286677205e-05, "loss": 0.0208, "step": 77830 }, { "epoch": 49.61121733588273, "grad_norm": 1.0597739219665527, "learning_rate": 8.847678339333119e-05, "loss": 0.0169, "step": 77840 }, { "epoch": 49.617590822179736, "grad_norm": 0.5157111883163452, "learning_rate": 8.847326345236106e-05, "loss": 0.0203, "step": 77850 }, { "epoch": 49.62396430847674, "grad_norm": 0.5466475486755371, "learning_rate": 8.846974304390449e-05, "loss": 0.0214, "step": 77860 }, { "epoch": 49.630337794773745, "grad_norm": 0.6174289584159851, "learning_rate": 8.84662221680042e-05, "loss": 0.0249, "step": 77870 }, { "epoch": 49.636711281070745, "grad_norm": 0.5043948292732239, "learning_rate": 8.8462700824703e-05, "loss": 0.0302, "step": 77880 }, { "epoch": 49.64308476736775, "grad_norm": 0.551700234413147, "learning_rate": 8.845917901404368e-05, "loss": 0.027, "step": 77890 }, { "epoch": 49.649458253664754, "grad_norm": 0.5174895524978638, "learning_rate": 8.8455656736069e-05, "loss": 0.0217, "step": 77900 }, { "epoch": 49.65583173996176, "grad_norm": 0.5218281149864197, "learning_rate": 8.84521339908218e-05, "loss": 0.0314, "step": 77910 }, { "epoch": 49.66220522625876, "grad_norm": 0.72174072265625, "learning_rate": 8.844861077834486e-05, "loss": 0.0438, "step": 77920 }, { "epoch": 49.66857871255577, "grad_norm": 0.2980845272541046, "learning_rate": 8.8445087098681e-05, "loss": 0.0289, "step": 77930 }, { "epoch": 49.67495219885277, "grad_norm": 0.6927303671836853, "learning_rate": 8.844156295187303e-05, "loss": 0.0244, "step": 77940 }, { "epoch": 49.68132568514978, "grad_norm": 0.6235808730125427, "learning_rate": 8.843803833796378e-05, "loss": 0.0377, "step": 77950 }, { "epoch": 49.68769917144678, "grad_norm": 0.5893200635910034, "learning_rate": 8.843451325699608e-05, "loss": 0.0316, "step": 77960 }, { "epoch": 49.69407265774379, "grad_norm": 0.5723037719726562, "learning_rate": 8.843098770901274e-05, "loss": 0.0166, "step": 77970 }, { "epoch": 49.70044614404079, "grad_norm": 0.6646707653999329, "learning_rate": 8.842746169405663e-05, "loss": 0.0264, "step": 77980 }, { "epoch": 49.7068196303378, "grad_norm": 0.3388667404651642, "learning_rate": 8.842393521217057e-05, "loss": 0.0189, "step": 77990 }, { "epoch": 49.7131931166348, "grad_norm": 0.5286176204681396, "learning_rate": 8.842040826339741e-05, "loss": 0.0172, "step": 78000 }, { "epoch": 49.719566602931806, "grad_norm": 0.46457430720329285, "learning_rate": 8.841688084778001e-05, "loss": 0.0177, "step": 78010 }, { "epoch": 49.72594008922881, "grad_norm": 0.3973242938518524, "learning_rate": 8.841335296536127e-05, "loss": 0.0246, "step": 78020 }, { "epoch": 49.732313575525815, "grad_norm": 0.4042084217071533, "learning_rate": 8.8409824616184e-05, "loss": 0.0232, "step": 78030 }, { "epoch": 49.738687061822816, "grad_norm": 0.4817942976951599, "learning_rate": 8.840629580029109e-05, "loss": 0.0211, "step": 78040 }, { "epoch": 49.745060548119824, "grad_norm": 0.22444985806941986, "learning_rate": 8.840276651772544e-05, "loss": 0.0174, "step": 78050 }, { "epoch": 49.751434034416825, "grad_norm": 0.41225841641426086, "learning_rate": 8.839923676852992e-05, "loss": 0.0326, "step": 78060 }, { "epoch": 49.75780752071383, "grad_norm": 0.5309574604034424, "learning_rate": 8.83957065527474e-05, "loss": 0.0247, "step": 78070 }, { "epoch": 49.76418100701083, "grad_norm": 0.24984565377235413, "learning_rate": 8.839217587042081e-05, "loss": 0.0199, "step": 78080 }, { "epoch": 49.77055449330784, "grad_norm": 0.5951162576675415, "learning_rate": 8.838864472159303e-05, "loss": 0.0245, "step": 78090 }, { "epoch": 49.77692797960484, "grad_norm": 0.2755894362926483, "learning_rate": 8.838511310630697e-05, "loss": 0.0132, "step": 78100 }, { "epoch": 49.78330146590185, "grad_norm": 0.795556902885437, "learning_rate": 8.838158102460554e-05, "loss": 0.0362, "step": 78110 }, { "epoch": 49.78967495219885, "grad_norm": 0.3600795865058899, "learning_rate": 8.837804847653168e-05, "loss": 0.0262, "step": 78120 }, { "epoch": 49.79604843849586, "grad_norm": 0.3459025025367737, "learning_rate": 8.837451546212827e-05, "loss": 0.0308, "step": 78130 }, { "epoch": 49.80242192479286, "grad_norm": 0.28365886211395264, "learning_rate": 8.83709819814383e-05, "loss": 0.0241, "step": 78140 }, { "epoch": 49.80879541108987, "grad_norm": 0.317590594291687, "learning_rate": 8.836744803450464e-05, "loss": 0.0197, "step": 78150 }, { "epoch": 49.81516889738687, "grad_norm": 0.5156533718109131, "learning_rate": 8.836391362137027e-05, "loss": 0.0274, "step": 78160 }, { "epoch": 49.82154238368388, "grad_norm": 0.5411946773529053, "learning_rate": 8.836037874207813e-05, "loss": 0.0197, "step": 78170 }, { "epoch": 49.82791586998088, "grad_norm": 0.5274532437324524, "learning_rate": 8.835684339667117e-05, "loss": 0.0288, "step": 78180 }, { "epoch": 49.834289356277885, "grad_norm": 0.31607282161712646, "learning_rate": 8.835330758519236e-05, "loss": 0.0243, "step": 78190 }, { "epoch": 49.840662842574886, "grad_norm": 0.5689270496368408, "learning_rate": 8.834977130768462e-05, "loss": 0.0273, "step": 78200 }, { "epoch": 49.847036328871894, "grad_norm": 0.6697741150856018, "learning_rate": 8.834623456419097e-05, "loss": 0.0395, "step": 78210 }, { "epoch": 49.853409815168895, "grad_norm": 0.47497016191482544, "learning_rate": 8.834269735475438e-05, "loss": 0.0242, "step": 78220 }, { "epoch": 49.8597833014659, "grad_norm": 0.35743266344070435, "learning_rate": 8.83391596794178e-05, "loss": 0.0174, "step": 78230 }, { "epoch": 49.866156787762904, "grad_norm": 0.4587726294994354, "learning_rate": 8.833562153822422e-05, "loss": 0.0288, "step": 78240 }, { "epoch": 49.87253027405991, "grad_norm": 0.43858957290649414, "learning_rate": 8.833208293121667e-05, "loss": 0.0252, "step": 78250 }, { "epoch": 49.87890376035691, "grad_norm": 0.36850830912590027, "learning_rate": 8.83285438584381e-05, "loss": 0.0217, "step": 78260 }, { "epoch": 49.88527724665392, "grad_norm": 0.3949533700942993, "learning_rate": 8.832500431993153e-05, "loss": 0.0323, "step": 78270 }, { "epoch": 49.89165073295092, "grad_norm": 0.5647585391998291, "learning_rate": 8.832146431573999e-05, "loss": 0.0309, "step": 78280 }, { "epoch": 49.89802421924793, "grad_norm": 0.5529283881187439, "learning_rate": 8.831792384590647e-05, "loss": 0.031, "step": 78290 }, { "epoch": 49.90439770554493, "grad_norm": 0.3740663230419159, "learning_rate": 8.831438291047401e-05, "loss": 0.0189, "step": 78300 }, { "epoch": 49.91077119184194, "grad_norm": 0.5241754055023193, "learning_rate": 8.83108415094856e-05, "loss": 0.0196, "step": 78310 }, { "epoch": 49.91714467813894, "grad_norm": 0.34823673963546753, "learning_rate": 8.83072996429843e-05, "loss": 0.0248, "step": 78320 }, { "epoch": 49.92351816443595, "grad_norm": 0.4125213623046875, "learning_rate": 8.830375731101315e-05, "loss": 0.0245, "step": 78330 }, { "epoch": 49.92989165073295, "grad_norm": 0.4358454942703247, "learning_rate": 8.830021451361518e-05, "loss": 0.0242, "step": 78340 }, { "epoch": 49.936265137029956, "grad_norm": 0.4816979467868805, "learning_rate": 8.829667125083344e-05, "loss": 0.0148, "step": 78350 }, { "epoch": 49.94263862332696, "grad_norm": 0.5312575697898865, "learning_rate": 8.8293127522711e-05, "loss": 0.0134, "step": 78360 }, { "epoch": 49.949012109623965, "grad_norm": 1.1517137289047241, "learning_rate": 8.828958332929088e-05, "loss": 0.0288, "step": 78370 }, { "epoch": 49.955385595920966, "grad_norm": 0.6083505749702454, "learning_rate": 8.828603867061619e-05, "loss": 0.0261, "step": 78380 }, { "epoch": 49.961759082217974, "grad_norm": 0.3646330237388611, "learning_rate": 8.828249354672997e-05, "loss": 0.0167, "step": 78390 }, { "epoch": 49.968132568514974, "grad_norm": 0.5330575704574585, "learning_rate": 8.827894795767533e-05, "loss": 0.0206, "step": 78400 }, { "epoch": 49.97450605481198, "grad_norm": 0.36092036962509155, "learning_rate": 8.82754019034953e-05, "loss": 0.0201, "step": 78410 }, { "epoch": 49.98087954110898, "grad_norm": 0.3796204924583435, "learning_rate": 8.827185538423302e-05, "loss": 0.0223, "step": 78420 }, { "epoch": 49.98725302740599, "grad_norm": 0.5716899633407593, "learning_rate": 8.826830839993157e-05, "loss": 0.0233, "step": 78430 }, { "epoch": 49.99362651370299, "grad_norm": 0.45647770166397095, "learning_rate": 8.826476095063403e-05, "loss": 0.0172, "step": 78440 }, { "epoch": 50.0, "grad_norm": 0.6249266266822815, "learning_rate": 8.826121303638351e-05, "loss": 0.0096, "step": 78450 }, { "epoch": 50.00637348629701, "grad_norm": 0.3381504416465759, "learning_rate": 8.825766465722313e-05, "loss": 0.0174, "step": 78460 }, { "epoch": 50.01274697259401, "grad_norm": 0.3201063573360443, "learning_rate": 8.825411581319601e-05, "loss": 0.0233, "step": 78470 }, { "epoch": 50.01912045889102, "grad_norm": 0.48725444078445435, "learning_rate": 8.825056650434527e-05, "loss": 0.0166, "step": 78480 }, { "epoch": 50.02549394518802, "grad_norm": 0.5825637578964233, "learning_rate": 8.824701673071402e-05, "loss": 0.0216, "step": 78490 }, { "epoch": 50.031867431485026, "grad_norm": 0.433925598859787, "learning_rate": 8.82434664923454e-05, "loss": 0.0291, "step": 78500 }, { "epoch": 50.038240917782026, "grad_norm": 0.4949580132961273, "learning_rate": 8.823991578928256e-05, "loss": 0.0252, "step": 78510 }, { "epoch": 50.044614404079034, "grad_norm": 0.680374026298523, "learning_rate": 8.823636462156866e-05, "loss": 0.0177, "step": 78520 }, { "epoch": 50.050987890376035, "grad_norm": 0.41085904836654663, "learning_rate": 8.823281298924681e-05, "loss": 0.021, "step": 78530 }, { "epoch": 50.05736137667304, "grad_norm": 0.31770628690719604, "learning_rate": 8.822926089236019e-05, "loss": 0.0198, "step": 78540 }, { "epoch": 50.063734862970044, "grad_norm": 0.42102426290512085, "learning_rate": 8.822570833095193e-05, "loss": 0.0236, "step": 78550 }, { "epoch": 50.07010834926705, "grad_norm": 0.270768404006958, "learning_rate": 8.822215530506526e-05, "loss": 0.013, "step": 78560 }, { "epoch": 50.07648183556405, "grad_norm": 0.7138969302177429, "learning_rate": 8.82186018147433e-05, "loss": 0.0148, "step": 78570 }, { "epoch": 50.08285532186106, "grad_norm": 0.2616616487503052, "learning_rate": 8.821504786002925e-05, "loss": 0.0142, "step": 78580 }, { "epoch": 50.08922880815806, "grad_norm": 0.3224712908267975, "learning_rate": 8.821149344096628e-05, "loss": 0.0183, "step": 78590 }, { "epoch": 50.09560229445507, "grad_norm": 0.49666711688041687, "learning_rate": 8.820793855759758e-05, "loss": 0.0175, "step": 78600 }, { "epoch": 50.10197578075207, "grad_norm": 0.6033188104629517, "learning_rate": 8.820438320996637e-05, "loss": 0.0182, "step": 78610 }, { "epoch": 50.10834926704908, "grad_norm": 0.6989789009094238, "learning_rate": 8.820082739811582e-05, "loss": 0.025, "step": 78620 }, { "epoch": 50.11472275334608, "grad_norm": 0.3877696990966797, "learning_rate": 8.819727112208916e-05, "loss": 0.0353, "step": 78630 }, { "epoch": 50.12109623964309, "grad_norm": 0.3612739145755768, "learning_rate": 8.819371438192958e-05, "loss": 0.0185, "step": 78640 }, { "epoch": 50.12746972594009, "grad_norm": 0.4553041458129883, "learning_rate": 8.819015717768032e-05, "loss": 0.0231, "step": 78650 }, { "epoch": 50.133843212237096, "grad_norm": 0.25178971886634827, "learning_rate": 8.818659950938457e-05, "loss": 0.0253, "step": 78660 }, { "epoch": 50.1402166985341, "grad_norm": 0.5894650220870972, "learning_rate": 8.81830413770856e-05, "loss": 0.0293, "step": 78670 }, { "epoch": 50.146590184831105, "grad_norm": 0.3403888940811157, "learning_rate": 8.817948278082663e-05, "loss": 0.0256, "step": 78680 }, { "epoch": 50.152963671128106, "grad_norm": 0.652225136756897, "learning_rate": 8.817592372065089e-05, "loss": 0.0215, "step": 78690 }, { "epoch": 50.159337157425114, "grad_norm": 0.531622588634491, "learning_rate": 8.817236419660164e-05, "loss": 0.0188, "step": 78700 }, { "epoch": 50.165710643722115, "grad_norm": 0.2072288691997528, "learning_rate": 8.816880420872212e-05, "loss": 0.0161, "step": 78710 }, { "epoch": 50.17208413001912, "grad_norm": 0.4642229378223419, "learning_rate": 8.816524375705559e-05, "loss": 0.0217, "step": 78720 }, { "epoch": 50.17845761631612, "grad_norm": 0.2581069767475128, "learning_rate": 8.816168284164531e-05, "loss": 0.0209, "step": 78730 }, { "epoch": 50.18483110261313, "grad_norm": 0.3714772164821625, "learning_rate": 8.815812146253456e-05, "loss": 0.0153, "step": 78740 }, { "epoch": 50.19120458891013, "grad_norm": 0.5910926461219788, "learning_rate": 8.815455961976661e-05, "loss": 0.0199, "step": 78750 }, { "epoch": 50.19757807520714, "grad_norm": 0.4600505828857422, "learning_rate": 8.815099731338472e-05, "loss": 0.0221, "step": 78760 }, { "epoch": 50.20395156150414, "grad_norm": 0.4295461177825928, "learning_rate": 8.814743454343221e-05, "loss": 0.0211, "step": 78770 }, { "epoch": 50.21032504780115, "grad_norm": 0.698404848575592, "learning_rate": 8.814387130995234e-05, "loss": 0.0231, "step": 78780 }, { "epoch": 50.21669853409815, "grad_norm": 0.2891003489494324, "learning_rate": 8.814030761298843e-05, "loss": 0.0326, "step": 78790 }, { "epoch": 50.22307202039516, "grad_norm": 0.43265002965927124, "learning_rate": 8.813674345258378e-05, "loss": 0.0217, "step": 78800 }, { "epoch": 50.22944550669216, "grad_norm": 0.40825155377388, "learning_rate": 8.813317882878168e-05, "loss": 0.0164, "step": 78810 }, { "epoch": 50.23581899298917, "grad_norm": 0.3375880718231201, "learning_rate": 8.812961374162546e-05, "loss": 0.0299, "step": 78820 }, { "epoch": 50.24219247928617, "grad_norm": 0.5209476947784424, "learning_rate": 8.812604819115842e-05, "loss": 0.0215, "step": 78830 }, { "epoch": 50.248565965583175, "grad_norm": 0.6618402600288391, "learning_rate": 8.812248217742389e-05, "loss": 0.019, "step": 78840 }, { "epoch": 50.254939451880176, "grad_norm": 0.36991608142852783, "learning_rate": 8.811891570046523e-05, "loss": 0.0277, "step": 78850 }, { "epoch": 50.261312938177184, "grad_norm": 0.4550458490848541, "learning_rate": 8.811534876032575e-05, "loss": 0.0194, "step": 78860 }, { "epoch": 50.267686424474185, "grad_norm": 0.49720388650894165, "learning_rate": 8.811178135704879e-05, "loss": 0.0341, "step": 78870 }, { "epoch": 50.27405991077119, "grad_norm": 0.43963623046875, "learning_rate": 8.81082134906777e-05, "loss": 0.0199, "step": 78880 }, { "epoch": 50.280433397068194, "grad_norm": 0.7907137274742126, "learning_rate": 8.810464516125582e-05, "loss": 0.0189, "step": 78890 }, { "epoch": 50.2868068833652, "grad_norm": 0.3150918483734131, "learning_rate": 8.810107636882656e-05, "loss": 0.0221, "step": 78900 }, { "epoch": 50.2931803696622, "grad_norm": 0.4583601653575897, "learning_rate": 8.809750711343323e-05, "loss": 0.0161, "step": 78910 }, { "epoch": 50.29955385595921, "grad_norm": 0.2740141451358795, "learning_rate": 8.809393739511922e-05, "loss": 0.0432, "step": 78920 }, { "epoch": 50.30592734225621, "grad_norm": 0.4254174828529358, "learning_rate": 8.80903672139279e-05, "loss": 0.0207, "step": 78930 }, { "epoch": 50.31230082855322, "grad_norm": 0.3557502031326294, "learning_rate": 8.808679656990266e-05, "loss": 0.0169, "step": 78940 }, { "epoch": 50.31867431485022, "grad_norm": 0.4067841172218323, "learning_rate": 8.80832254630869e-05, "loss": 0.031, "step": 78950 }, { "epoch": 50.32504780114723, "grad_norm": 0.6919642686843872, "learning_rate": 8.807965389352397e-05, "loss": 0.0178, "step": 78960 }, { "epoch": 50.33142128744423, "grad_norm": 0.4111323654651642, "learning_rate": 8.807608186125729e-05, "loss": 0.021, "step": 78970 }, { "epoch": 50.33779477374124, "grad_norm": 0.6904878616333008, "learning_rate": 8.807250936633026e-05, "loss": 0.0275, "step": 78980 }, { "epoch": 50.34416826003824, "grad_norm": 0.9003934264183044, "learning_rate": 8.806893640878632e-05, "loss": 0.0256, "step": 78990 }, { "epoch": 50.350541746335246, "grad_norm": 0.4523234963417053, "learning_rate": 8.806536298866886e-05, "loss": 0.0344, "step": 79000 }, { "epoch": 50.35691523263225, "grad_norm": 0.5343559980392456, "learning_rate": 8.806178910602127e-05, "loss": 0.0231, "step": 79010 }, { "epoch": 50.363288718929255, "grad_norm": 0.3758242130279541, "learning_rate": 8.805821476088704e-05, "loss": 0.0242, "step": 79020 }, { "epoch": 50.369662205226255, "grad_norm": 0.3731200098991394, "learning_rate": 8.805463995330955e-05, "loss": 0.0236, "step": 79030 }, { "epoch": 50.37603569152326, "grad_norm": 0.3420630693435669, "learning_rate": 8.805106468333225e-05, "loss": 0.0253, "step": 79040 }, { "epoch": 50.382409177820264, "grad_norm": 0.5027868151664734, "learning_rate": 8.804748895099859e-05, "loss": 0.0168, "step": 79050 }, { "epoch": 50.38878266411727, "grad_norm": 0.4659039378166199, "learning_rate": 8.804391275635202e-05, "loss": 0.0312, "step": 79060 }, { "epoch": 50.39515615041427, "grad_norm": 0.35768458247184753, "learning_rate": 8.8040336099436e-05, "loss": 0.0207, "step": 79070 }, { "epoch": 50.40152963671128, "grad_norm": 0.7950899004936218, "learning_rate": 8.803675898029397e-05, "loss": 0.0292, "step": 79080 }, { "epoch": 50.40790312300829, "grad_norm": 0.23811520636081696, "learning_rate": 8.803318139896942e-05, "loss": 0.0245, "step": 79090 }, { "epoch": 50.41427660930529, "grad_norm": 0.43506231904029846, "learning_rate": 8.802960335550578e-05, "loss": 0.0252, "step": 79100 }, { "epoch": 50.4206500956023, "grad_norm": 0.22509349882602692, "learning_rate": 8.802602484994658e-05, "loss": 0.0289, "step": 79110 }, { "epoch": 50.4270235818993, "grad_norm": 0.34668299555778503, "learning_rate": 8.802244588233526e-05, "loss": 0.0235, "step": 79120 }, { "epoch": 50.43339706819631, "grad_norm": 0.43714094161987305, "learning_rate": 8.801886645271533e-05, "loss": 0.0278, "step": 79130 }, { "epoch": 50.43977055449331, "grad_norm": 0.40661999583244324, "learning_rate": 8.801528656113027e-05, "loss": 0.0161, "step": 79140 }, { "epoch": 50.446144040790315, "grad_norm": 0.3066137433052063, "learning_rate": 8.801170620762358e-05, "loss": 0.0178, "step": 79150 }, { "epoch": 50.452517527087316, "grad_norm": 0.4663435220718384, "learning_rate": 8.800812539223879e-05, "loss": 0.03, "step": 79160 }, { "epoch": 50.458891013384324, "grad_norm": 0.3341331481933594, "learning_rate": 8.800454411501938e-05, "loss": 0.0289, "step": 79170 }, { "epoch": 50.465264499681325, "grad_norm": 0.5391457676887512, "learning_rate": 8.800096237600888e-05, "loss": 0.0277, "step": 79180 }, { "epoch": 50.47163798597833, "grad_norm": 0.7094258069992065, "learning_rate": 8.799738017525081e-05, "loss": 0.0269, "step": 79190 }, { "epoch": 50.478011472275334, "grad_norm": 0.5689120888710022, "learning_rate": 8.799379751278869e-05, "loss": 0.0191, "step": 79200 }, { "epoch": 50.48438495857234, "grad_norm": 0.5518429279327393, "learning_rate": 8.799021438866606e-05, "loss": 0.0265, "step": 79210 }, { "epoch": 50.49075844486934, "grad_norm": 0.558884859085083, "learning_rate": 8.798663080292646e-05, "loss": 0.0217, "step": 79220 }, { "epoch": 50.49713193116635, "grad_norm": 0.760184645652771, "learning_rate": 8.798304675561342e-05, "loss": 0.0322, "step": 79230 }, { "epoch": 50.50350541746335, "grad_norm": 0.49008604884147644, "learning_rate": 8.797946224677052e-05, "loss": 0.033, "step": 79240 }, { "epoch": 50.50987890376036, "grad_norm": 0.5365548133850098, "learning_rate": 8.797587727644128e-05, "loss": 0.023, "step": 79250 }, { "epoch": 50.51625239005736, "grad_norm": 0.4222807288169861, "learning_rate": 8.797229184466927e-05, "loss": 0.0213, "step": 79260 }, { "epoch": 50.52262587635437, "grad_norm": 0.7180867791175842, "learning_rate": 8.796870595149806e-05, "loss": 0.0224, "step": 79270 }, { "epoch": 50.52899936265137, "grad_norm": 0.3896630108356476, "learning_rate": 8.796511959697124e-05, "loss": 0.016, "step": 79280 }, { "epoch": 50.53537284894838, "grad_norm": 0.4448973834514618, "learning_rate": 8.796153278113237e-05, "loss": 0.0189, "step": 79290 }, { "epoch": 50.54174633524538, "grad_norm": 0.27137893438339233, "learning_rate": 8.795794550402503e-05, "loss": 0.0232, "step": 79300 }, { "epoch": 50.548119821542386, "grad_norm": 0.6930257678031921, "learning_rate": 8.795435776569282e-05, "loss": 0.04, "step": 79310 }, { "epoch": 50.55449330783939, "grad_norm": 0.6881492137908936, "learning_rate": 8.795076956617932e-05, "loss": 0.0313, "step": 79320 }, { "epoch": 50.560866794136395, "grad_norm": 0.611892580986023, "learning_rate": 8.794718090552815e-05, "loss": 0.0224, "step": 79330 }, { "epoch": 50.567240280433396, "grad_norm": 0.39108553528785706, "learning_rate": 8.794359178378289e-05, "loss": 0.0163, "step": 79340 }, { "epoch": 50.5736137667304, "grad_norm": 0.6892337203025818, "learning_rate": 8.794000220098717e-05, "loss": 0.032, "step": 79350 }, { "epoch": 50.579987253027404, "grad_norm": 0.24506813287734985, "learning_rate": 8.79364121571846e-05, "loss": 0.0221, "step": 79360 }, { "epoch": 50.58636073932441, "grad_norm": 0.3120301365852356, "learning_rate": 8.79328216524188e-05, "loss": 0.0398, "step": 79370 }, { "epoch": 50.59273422562141, "grad_norm": 0.446262001991272, "learning_rate": 8.792923068673342e-05, "loss": 0.0249, "step": 79380 }, { "epoch": 50.59910771191842, "grad_norm": 0.5298762321472168, "learning_rate": 8.792563926017207e-05, "loss": 0.0246, "step": 79390 }, { "epoch": 50.60548119821542, "grad_norm": 0.5015619993209839, "learning_rate": 8.792204737277838e-05, "loss": 0.0333, "step": 79400 }, { "epoch": 50.61185468451243, "grad_norm": 0.6664139628410339, "learning_rate": 8.791845502459604e-05, "loss": 0.0422, "step": 79410 }, { "epoch": 50.61822817080943, "grad_norm": 0.5999925136566162, "learning_rate": 8.791486221566865e-05, "loss": 0.036, "step": 79420 }, { "epoch": 50.62460165710644, "grad_norm": 0.6568979620933533, "learning_rate": 8.791126894603987e-05, "loss": 0.0344, "step": 79430 }, { "epoch": 50.63097514340344, "grad_norm": 0.7174100279808044, "learning_rate": 8.790767521575341e-05, "loss": 0.0247, "step": 79440 }, { "epoch": 50.63734862970045, "grad_norm": 0.8572925925254822, "learning_rate": 8.790408102485287e-05, "loss": 0.026, "step": 79450 }, { "epoch": 50.64372211599745, "grad_norm": 0.31181514263153076, "learning_rate": 8.7900486373382e-05, "loss": 0.0276, "step": 79460 }, { "epoch": 50.650095602294456, "grad_norm": 0.39411067962646484, "learning_rate": 8.789689126138441e-05, "loss": 0.0231, "step": 79470 }, { "epoch": 50.65646908859146, "grad_norm": 0.5730124711990356, "learning_rate": 8.789329568890382e-05, "loss": 0.0238, "step": 79480 }, { "epoch": 50.662842574888465, "grad_norm": 0.4170571565628052, "learning_rate": 8.78896996559839e-05, "loss": 0.0315, "step": 79490 }, { "epoch": 50.669216061185466, "grad_norm": 0.3468725085258484, "learning_rate": 8.788610316266837e-05, "loss": 0.0311, "step": 79500 }, { "epoch": 50.675589547482474, "grad_norm": 0.4995209574699402, "learning_rate": 8.788250620900091e-05, "loss": 0.0174, "step": 79510 }, { "epoch": 50.681963033779475, "grad_norm": 0.366370290517807, "learning_rate": 8.787890879502522e-05, "loss": 0.0231, "step": 79520 }, { "epoch": 50.68833652007648, "grad_norm": 0.473082959651947, "learning_rate": 8.787531092078504e-05, "loss": 0.0169, "step": 79530 }, { "epoch": 50.694710006373484, "grad_norm": 0.3006952702999115, "learning_rate": 8.787171258632407e-05, "loss": 0.0174, "step": 79540 }, { "epoch": 50.70108349267049, "grad_norm": 0.4600659906864166, "learning_rate": 8.786811379168604e-05, "loss": 0.0443, "step": 79550 }, { "epoch": 50.70745697896749, "grad_norm": 0.32899153232574463, "learning_rate": 8.786451453691466e-05, "loss": 0.0151, "step": 79560 }, { "epoch": 50.7138304652645, "grad_norm": 0.5269507169723511, "learning_rate": 8.78609148220537e-05, "loss": 0.0242, "step": 79570 }, { "epoch": 50.7202039515615, "grad_norm": 0.5984697937965393, "learning_rate": 8.785731464714686e-05, "loss": 0.0324, "step": 79580 }, { "epoch": 50.72657743785851, "grad_norm": 0.558231770992279, "learning_rate": 8.78537140122379e-05, "loss": 0.0164, "step": 79590 }, { "epoch": 50.73295092415551, "grad_norm": 0.6354507207870483, "learning_rate": 8.785011291737058e-05, "loss": 0.0208, "step": 79600 }, { "epoch": 50.73932441045252, "grad_norm": 0.5819699168205261, "learning_rate": 8.784651136258865e-05, "loss": 0.0282, "step": 79610 }, { "epoch": 50.74569789674952, "grad_norm": 0.3994032144546509, "learning_rate": 8.784290934793589e-05, "loss": 0.0218, "step": 79620 }, { "epoch": 50.75207138304653, "grad_norm": 0.4292277693748474, "learning_rate": 8.783930687345603e-05, "loss": 0.024, "step": 79630 }, { "epoch": 50.75844486934353, "grad_norm": 1.104720950126648, "learning_rate": 8.783570393919288e-05, "loss": 0.0282, "step": 79640 }, { "epoch": 50.764818355640536, "grad_norm": 0.41390469670295715, "learning_rate": 8.78321005451902e-05, "loss": 0.0261, "step": 79650 }, { "epoch": 50.77119184193754, "grad_norm": 0.5971600413322449, "learning_rate": 8.782849669149178e-05, "loss": 0.0166, "step": 79660 }, { "epoch": 50.777565328234544, "grad_norm": 0.38011690974235535, "learning_rate": 8.78248923781414e-05, "loss": 0.0192, "step": 79670 }, { "epoch": 50.783938814531545, "grad_norm": 1.0127087831497192, "learning_rate": 8.782128760518287e-05, "loss": 0.0255, "step": 79680 }, { "epoch": 50.79031230082855, "grad_norm": 0.6223517060279846, "learning_rate": 8.781768237266e-05, "loss": 0.0237, "step": 79690 }, { "epoch": 50.796685787125554, "grad_norm": 0.43792593479156494, "learning_rate": 8.781407668061656e-05, "loss": 0.0244, "step": 79700 }, { "epoch": 50.80305927342256, "grad_norm": 0.4628196060657501, "learning_rate": 8.781047052909641e-05, "loss": 0.021, "step": 79710 }, { "epoch": 50.80943275971957, "grad_norm": 0.4288725256919861, "learning_rate": 8.780686391814333e-05, "loss": 0.0235, "step": 79720 }, { "epoch": 50.81580624601657, "grad_norm": 0.5100308656692505, "learning_rate": 8.780325684780115e-05, "loss": 0.0198, "step": 79730 }, { "epoch": 50.82217973231358, "grad_norm": 0.34135833382606506, "learning_rate": 8.779964931811373e-05, "loss": 0.0226, "step": 79740 }, { "epoch": 50.82855321861058, "grad_norm": 0.45776090025901794, "learning_rate": 8.779604132912487e-05, "loss": 0.0179, "step": 79750 }, { "epoch": 50.83492670490759, "grad_norm": 0.5911512970924377, "learning_rate": 8.779243288087842e-05, "loss": 0.0194, "step": 79760 }, { "epoch": 50.84130019120459, "grad_norm": 0.7046938538551331, "learning_rate": 8.778882397341823e-05, "loss": 0.0268, "step": 79770 }, { "epoch": 50.8476736775016, "grad_norm": 0.659308910369873, "learning_rate": 8.778521460678815e-05, "loss": 0.0232, "step": 79780 }, { "epoch": 50.8540471637986, "grad_norm": 0.30571916699409485, "learning_rate": 8.778160478103204e-05, "loss": 0.0194, "step": 79790 }, { "epoch": 50.860420650095605, "grad_norm": 0.2518063485622406, "learning_rate": 8.777799449619375e-05, "loss": 0.0185, "step": 79800 }, { "epoch": 50.866794136392606, "grad_norm": 0.6893629431724548, "learning_rate": 8.777438375231717e-05, "loss": 0.0279, "step": 79810 }, { "epoch": 50.873167622689614, "grad_norm": 0.45029497146606445, "learning_rate": 8.777077254944616e-05, "loss": 0.0276, "step": 79820 }, { "epoch": 50.879541108986615, "grad_norm": 0.4098087549209595, "learning_rate": 8.77671608876246e-05, "loss": 0.0209, "step": 79830 }, { "epoch": 50.88591459528362, "grad_norm": 0.4715461730957031, "learning_rate": 8.776354876689638e-05, "loss": 0.0196, "step": 79840 }, { "epoch": 50.892288081580624, "grad_norm": 0.7570445537567139, "learning_rate": 8.775993618730538e-05, "loss": 0.0244, "step": 79850 }, { "epoch": 50.89866156787763, "grad_norm": 0.44060954451560974, "learning_rate": 8.775632314889552e-05, "loss": 0.0252, "step": 79860 }, { "epoch": 50.90503505417463, "grad_norm": 0.4500226676464081, "learning_rate": 8.775270965171067e-05, "loss": 0.0249, "step": 79870 }, { "epoch": 50.91140854047164, "grad_norm": 0.5883831977844238, "learning_rate": 8.774909569579476e-05, "loss": 0.0238, "step": 79880 }, { "epoch": 50.91778202676864, "grad_norm": 0.9299945831298828, "learning_rate": 8.774548128119169e-05, "loss": 0.033, "step": 79890 }, { "epoch": 50.92415551306565, "grad_norm": 0.5143661499023438, "learning_rate": 8.77418664079454e-05, "loss": 0.0236, "step": 79900 }, { "epoch": 50.93052899936265, "grad_norm": 0.4556948244571686, "learning_rate": 8.773825107609979e-05, "loss": 0.0176, "step": 79910 }, { "epoch": 50.93690248565966, "grad_norm": 0.2951599061489105, "learning_rate": 8.773463528569879e-05, "loss": 0.016, "step": 79920 }, { "epoch": 50.94327597195666, "grad_norm": 0.4766428470611572, "learning_rate": 8.773101903678635e-05, "loss": 0.0218, "step": 79930 }, { "epoch": 50.94964945825367, "grad_norm": 0.23115359246730804, "learning_rate": 8.77274023294064e-05, "loss": 0.0235, "step": 79940 }, { "epoch": 50.95602294455067, "grad_norm": 0.4037020802497864, "learning_rate": 8.77237851636029e-05, "loss": 0.0204, "step": 79950 }, { "epoch": 50.962396430847676, "grad_norm": 0.3885551989078522, "learning_rate": 8.772016753941979e-05, "loss": 0.0255, "step": 79960 }, { "epoch": 50.96876991714468, "grad_norm": 0.6576778888702393, "learning_rate": 8.771654945690104e-05, "loss": 0.026, "step": 79970 }, { "epoch": 50.975143403441685, "grad_norm": 0.3049602210521698, "learning_rate": 8.771293091609058e-05, "loss": 0.0285, "step": 79980 }, { "epoch": 50.981516889738685, "grad_norm": 0.3631782531738281, "learning_rate": 8.770931191703243e-05, "loss": 0.014, "step": 79990 }, { "epoch": 50.98789037603569, "grad_norm": 0.7584400773048401, "learning_rate": 8.770569245977051e-05, "loss": 0.0211, "step": 80000 }, { "epoch": 50.994263862332694, "grad_norm": 0.8856493234634399, "learning_rate": 8.770207254434886e-05, "loss": 0.0248, "step": 80010 }, { "epoch": 51.0006373486297, "grad_norm": 0.3686481714248657, "learning_rate": 8.769845217081141e-05, "loss": 0.0178, "step": 80020 }, { "epoch": 51.0070108349267, "grad_norm": 0.5448732376098633, "learning_rate": 8.769483133920218e-05, "loss": 0.022, "step": 80030 }, { "epoch": 51.01338432122371, "grad_norm": 0.21480068564414978, "learning_rate": 8.769121004956516e-05, "loss": 0.0183, "step": 80040 }, { "epoch": 51.01975780752071, "grad_norm": 0.34940415620803833, "learning_rate": 8.768758830194434e-05, "loss": 0.0244, "step": 80050 }, { "epoch": 51.02613129381772, "grad_norm": 0.47757336497306824, "learning_rate": 8.768396609638374e-05, "loss": 0.0208, "step": 80060 }, { "epoch": 51.03250478011472, "grad_norm": 0.4913415014743805, "learning_rate": 8.768034343292737e-05, "loss": 0.0185, "step": 80070 }, { "epoch": 51.03887826641173, "grad_norm": 0.4642592966556549, "learning_rate": 8.767672031161926e-05, "loss": 0.0197, "step": 80080 }, { "epoch": 51.04525175270873, "grad_norm": 0.6955509781837463, "learning_rate": 8.767309673250344e-05, "loss": 0.0183, "step": 80090 }, { "epoch": 51.05162523900574, "grad_norm": 0.5535367727279663, "learning_rate": 8.76694726956239e-05, "loss": 0.0205, "step": 80100 }, { "epoch": 51.05799872530274, "grad_norm": 0.24094238877296448, "learning_rate": 8.76658482010247e-05, "loss": 0.0235, "step": 80110 }, { "epoch": 51.064372211599746, "grad_norm": 0.3038710057735443, "learning_rate": 8.766222324874989e-05, "loss": 0.0214, "step": 80120 }, { "epoch": 51.07074569789675, "grad_norm": 0.49538812041282654, "learning_rate": 8.765859783884351e-05, "loss": 0.0249, "step": 80130 }, { "epoch": 51.077119184193755, "grad_norm": 0.5099695324897766, "learning_rate": 8.765497197134962e-05, "loss": 0.0253, "step": 80140 }, { "epoch": 51.083492670490756, "grad_norm": 0.43623775243759155, "learning_rate": 8.765134564631225e-05, "loss": 0.0201, "step": 80150 }, { "epoch": 51.089866156787764, "grad_norm": 0.7550836801528931, "learning_rate": 8.76477188637755e-05, "loss": 0.0157, "step": 80160 }, { "epoch": 51.096239643084765, "grad_norm": 0.6480455994606018, "learning_rate": 8.764409162378342e-05, "loss": 0.0238, "step": 80170 }, { "epoch": 51.10261312938177, "grad_norm": 0.7898861169815063, "learning_rate": 8.764046392638008e-05, "loss": 0.028, "step": 80180 }, { "epoch": 51.10898661567877, "grad_norm": 0.6691673398017883, "learning_rate": 8.763683577160955e-05, "loss": 0.0294, "step": 80190 }, { "epoch": 51.11536010197578, "grad_norm": 0.5142627954483032, "learning_rate": 8.763320715951595e-05, "loss": 0.0143, "step": 80200 }, { "epoch": 51.12173358827278, "grad_norm": 0.34042656421661377, "learning_rate": 8.762957809014334e-05, "loss": 0.0256, "step": 80210 }, { "epoch": 51.12810707456979, "grad_norm": 0.6345179080963135, "learning_rate": 8.762594856353583e-05, "loss": 0.0278, "step": 80220 }, { "epoch": 51.13448056086679, "grad_norm": 0.29842260479927063, "learning_rate": 8.762231857973752e-05, "loss": 0.0209, "step": 80230 }, { "epoch": 51.1408540471638, "grad_norm": 0.5378405451774597, "learning_rate": 8.761868813879253e-05, "loss": 0.0272, "step": 80240 }, { "epoch": 51.1472275334608, "grad_norm": 0.534474790096283, "learning_rate": 8.761505724074495e-05, "loss": 0.0225, "step": 80250 }, { "epoch": 51.15360101975781, "grad_norm": 0.4871113896369934, "learning_rate": 8.761142588563891e-05, "loss": 0.0178, "step": 80260 }, { "epoch": 51.15997450605481, "grad_norm": 0.3796258270740509, "learning_rate": 8.760779407351853e-05, "loss": 0.0283, "step": 80270 }, { "epoch": 51.16634799235182, "grad_norm": 0.2343824803829193, "learning_rate": 8.760416180442795e-05, "loss": 0.0209, "step": 80280 }, { "epoch": 51.17272147864882, "grad_norm": 0.4615975320339203, "learning_rate": 8.760052907841129e-05, "loss": 0.0232, "step": 80290 }, { "epoch": 51.179094964945826, "grad_norm": 0.4010646343231201, "learning_rate": 8.759689589551272e-05, "loss": 0.0221, "step": 80300 }, { "epoch": 51.185468451242826, "grad_norm": 0.5545438528060913, "learning_rate": 8.759326225577635e-05, "loss": 0.0246, "step": 80310 }, { "epoch": 51.191841937539834, "grad_norm": 0.6877827048301697, "learning_rate": 8.758962815924636e-05, "loss": 0.0189, "step": 80320 }, { "epoch": 51.198215423836835, "grad_norm": 0.4778510332107544, "learning_rate": 8.75859936059669e-05, "loss": 0.015, "step": 80330 }, { "epoch": 51.20458891013384, "grad_norm": 0.751682460308075, "learning_rate": 8.758235859598212e-05, "loss": 0.0196, "step": 80340 }, { "epoch": 51.21096239643085, "grad_norm": 0.3829878568649292, "learning_rate": 8.757872312933622e-05, "loss": 0.0172, "step": 80350 }, { "epoch": 51.21733588272785, "grad_norm": 0.5323362946510315, "learning_rate": 8.757508720607332e-05, "loss": 0.0233, "step": 80360 }, { "epoch": 51.22370936902486, "grad_norm": 0.7504252195358276, "learning_rate": 8.757145082623766e-05, "loss": 0.0323, "step": 80370 }, { "epoch": 51.23008285532186, "grad_norm": 0.42253565788269043, "learning_rate": 8.75678139898734e-05, "loss": 0.0171, "step": 80380 }, { "epoch": 51.23645634161887, "grad_norm": 0.5518625378608704, "learning_rate": 8.756417669702472e-05, "loss": 0.0264, "step": 80390 }, { "epoch": 51.24282982791587, "grad_norm": 0.3475361764431, "learning_rate": 8.756053894773583e-05, "loss": 0.0264, "step": 80400 }, { "epoch": 51.24920331421288, "grad_norm": 0.35568803548812866, "learning_rate": 8.755690074205092e-05, "loss": 0.0152, "step": 80410 }, { "epoch": 51.25557680050988, "grad_norm": 0.40979093313217163, "learning_rate": 8.755326208001422e-05, "loss": 0.0188, "step": 80420 }, { "epoch": 51.261950286806886, "grad_norm": 0.5376774668693542, "learning_rate": 8.754962296166992e-05, "loss": 0.03, "step": 80430 }, { "epoch": 51.26832377310389, "grad_norm": 0.7888513207435608, "learning_rate": 8.754598338706224e-05, "loss": 0.0172, "step": 80440 }, { "epoch": 51.274697259400895, "grad_norm": 0.38959643244743347, "learning_rate": 8.754234335623544e-05, "loss": 0.0214, "step": 80450 }, { "epoch": 51.281070745697896, "grad_norm": 0.29616448283195496, "learning_rate": 8.75387028692337e-05, "loss": 0.0173, "step": 80460 }, { "epoch": 51.287444231994904, "grad_norm": 0.4908255338668823, "learning_rate": 8.753506192610129e-05, "loss": 0.0207, "step": 80470 }, { "epoch": 51.293817718291905, "grad_norm": 0.57209312915802, "learning_rate": 8.753142052688243e-05, "loss": 0.0158, "step": 80480 }, { "epoch": 51.30019120458891, "grad_norm": 0.3021024465560913, "learning_rate": 8.752777867162137e-05, "loss": 0.0226, "step": 80490 }, { "epoch": 51.306564690885914, "grad_norm": 0.21389898657798767, "learning_rate": 8.752413636036237e-05, "loss": 0.0136, "step": 80500 }, { "epoch": 51.31293817718292, "grad_norm": 0.3558087944984436, "learning_rate": 8.752049359314969e-05, "loss": 0.011, "step": 80510 }, { "epoch": 51.31931166347992, "grad_norm": 0.38619810342788696, "learning_rate": 8.751685037002759e-05, "loss": 0.0183, "step": 80520 }, { "epoch": 51.32568514977693, "grad_norm": 0.31868284940719604, "learning_rate": 8.751320669104034e-05, "loss": 0.0157, "step": 80530 }, { "epoch": 51.33205863607393, "grad_norm": 0.5700975656509399, "learning_rate": 8.75095625562322e-05, "loss": 0.0231, "step": 80540 }, { "epoch": 51.33843212237094, "grad_norm": 0.3430424630641937, "learning_rate": 8.750591796564746e-05, "loss": 0.0224, "step": 80550 }, { "epoch": 51.34480560866794, "grad_norm": 0.6331754922866821, "learning_rate": 8.750227291933042e-05, "loss": 0.0266, "step": 80560 }, { "epoch": 51.35117909496495, "grad_norm": 0.4514833092689514, "learning_rate": 8.749862741732532e-05, "loss": 0.0191, "step": 80570 }, { "epoch": 51.35755258126195, "grad_norm": 0.47303831577301025, "learning_rate": 8.749498145967654e-05, "loss": 0.0301, "step": 80580 }, { "epoch": 51.36392606755896, "grad_norm": 0.46819037199020386, "learning_rate": 8.74913350464283e-05, "loss": 0.023, "step": 80590 }, { "epoch": 51.37029955385596, "grad_norm": 0.45215895771980286, "learning_rate": 8.748768817762496e-05, "loss": 0.0311, "step": 80600 }, { "epoch": 51.376673040152966, "grad_norm": 0.4320497214794159, "learning_rate": 8.748404085331081e-05, "loss": 0.0353, "step": 80610 }, { "epoch": 51.38304652644997, "grad_norm": 0.7134445905685425, "learning_rate": 8.748039307353017e-05, "loss": 0.0198, "step": 80620 }, { "epoch": 51.389420012746974, "grad_norm": 0.5475758910179138, "learning_rate": 8.747674483832737e-05, "loss": 0.0215, "step": 80630 }, { "epoch": 51.395793499043975, "grad_norm": 0.42564475536346436, "learning_rate": 8.747309614774673e-05, "loss": 0.0355, "step": 80640 }, { "epoch": 51.40216698534098, "grad_norm": 0.32626304030418396, "learning_rate": 8.746944700183259e-05, "loss": 0.0216, "step": 80650 }, { "epoch": 51.408540471637984, "grad_norm": 0.22941851615905762, "learning_rate": 8.74657974006293e-05, "loss": 0.0239, "step": 80660 }, { "epoch": 51.41491395793499, "grad_norm": 0.4409743845462799, "learning_rate": 8.746214734418119e-05, "loss": 0.0219, "step": 80670 }, { "epoch": 51.42128744423199, "grad_norm": 0.3632958233356476, "learning_rate": 8.745849683253262e-05, "loss": 0.0191, "step": 80680 }, { "epoch": 51.427660930529, "grad_norm": 0.18139545619487762, "learning_rate": 8.745484586572796e-05, "loss": 0.0271, "step": 80690 }, { "epoch": 51.434034416826, "grad_norm": 0.528667151927948, "learning_rate": 8.745119444381154e-05, "loss": 0.0305, "step": 80700 }, { "epoch": 51.44040790312301, "grad_norm": 0.31671062111854553, "learning_rate": 8.744754256682776e-05, "loss": 0.0149, "step": 80710 }, { "epoch": 51.44678138942001, "grad_norm": 0.490806519985199, "learning_rate": 8.744389023482098e-05, "loss": 0.0382, "step": 80720 }, { "epoch": 51.45315487571702, "grad_norm": 0.3629381060600281, "learning_rate": 8.744023744783558e-05, "loss": 0.0303, "step": 80730 }, { "epoch": 51.45952836201402, "grad_norm": 0.4655114710330963, "learning_rate": 8.743658420591595e-05, "loss": 0.0279, "step": 80740 }, { "epoch": 51.46590184831103, "grad_norm": 0.6693554520606995, "learning_rate": 8.743293050910647e-05, "loss": 0.0248, "step": 80750 }, { "epoch": 51.47227533460803, "grad_norm": 0.35395073890686035, "learning_rate": 8.742927635745155e-05, "loss": 0.0205, "step": 80760 }, { "epoch": 51.478648820905036, "grad_norm": 0.38700437545776367, "learning_rate": 8.742562175099558e-05, "loss": 0.0311, "step": 80770 }, { "epoch": 51.48502230720204, "grad_norm": 0.39843520522117615, "learning_rate": 8.742196668978297e-05, "loss": 0.0233, "step": 80780 }, { "epoch": 51.491395793499045, "grad_norm": 0.5317781567573547, "learning_rate": 8.741831117385812e-05, "loss": 0.0124, "step": 80790 }, { "epoch": 51.497769279796046, "grad_norm": 0.3286365270614624, "learning_rate": 8.741465520326548e-05, "loss": 0.0132, "step": 80800 }, { "epoch": 51.504142766093054, "grad_norm": 0.4258129596710205, "learning_rate": 8.741099877804943e-05, "loss": 0.0513, "step": 80810 }, { "epoch": 51.510516252390055, "grad_norm": 0.3511577844619751, "learning_rate": 8.740734189825443e-05, "loss": 0.0251, "step": 80820 }, { "epoch": 51.51688973868706, "grad_norm": 0.32276976108551025, "learning_rate": 8.740368456392493e-05, "loss": 0.0261, "step": 80830 }, { "epoch": 51.52326322498406, "grad_norm": 0.3797406256198883, "learning_rate": 8.740002677510532e-05, "loss": 0.022, "step": 80840 }, { "epoch": 51.52963671128107, "grad_norm": 0.3691493272781372, "learning_rate": 8.739636853184008e-05, "loss": 0.021, "step": 80850 }, { "epoch": 51.53601019757807, "grad_norm": 0.4678703248500824, "learning_rate": 8.739270983417366e-05, "loss": 0.0187, "step": 80860 }, { "epoch": 51.54238368387508, "grad_norm": 0.7122126221656799, "learning_rate": 8.738905068215052e-05, "loss": 0.0239, "step": 80870 }, { "epoch": 51.54875717017208, "grad_norm": 0.34224560856819153, "learning_rate": 8.73853910758151e-05, "loss": 0.0213, "step": 80880 }, { "epoch": 51.55513065646909, "grad_norm": 0.6818312406539917, "learning_rate": 8.738173101521189e-05, "loss": 0.0314, "step": 80890 }, { "epoch": 51.56150414276609, "grad_norm": 0.48439544439315796, "learning_rate": 8.737807050038534e-05, "loss": 0.0357, "step": 80900 }, { "epoch": 51.5678776290631, "grad_norm": 0.49195408821105957, "learning_rate": 8.737440953137997e-05, "loss": 0.0351, "step": 80910 }, { "epoch": 51.5742511153601, "grad_norm": 0.5004852414131165, "learning_rate": 8.737074810824019e-05, "loss": 0.0146, "step": 80920 }, { "epoch": 51.58062460165711, "grad_norm": 0.7326125502586365, "learning_rate": 8.736708623101058e-05, "loss": 0.0212, "step": 80930 }, { "epoch": 51.58699808795411, "grad_norm": 0.31113579869270325, "learning_rate": 8.736342389973557e-05, "loss": 0.0139, "step": 80940 }, { "epoch": 51.593371574251115, "grad_norm": 0.5144744515419006, "learning_rate": 8.73597611144597e-05, "loss": 0.0182, "step": 80950 }, { "epoch": 51.599745060548116, "grad_norm": 0.4252970218658447, "learning_rate": 8.735609787522745e-05, "loss": 0.0228, "step": 80960 }, { "epoch": 51.606118546845124, "grad_norm": 0.26974132657051086, "learning_rate": 8.735243418208333e-05, "loss": 0.0234, "step": 80970 }, { "epoch": 51.61249203314213, "grad_norm": 0.3215683698654175, "learning_rate": 8.734877003507188e-05, "loss": 0.0296, "step": 80980 }, { "epoch": 51.61886551943913, "grad_norm": 0.3875557780265808, "learning_rate": 8.73451054342376e-05, "loss": 0.0248, "step": 80990 }, { "epoch": 51.62523900573614, "grad_norm": 0.40797123312950134, "learning_rate": 8.734144037962504e-05, "loss": 0.0192, "step": 81000 }, { "epoch": 51.63161249203314, "grad_norm": 0.49092209339141846, "learning_rate": 8.733777487127872e-05, "loss": 0.0162, "step": 81010 }, { "epoch": 51.63798597833015, "grad_norm": 0.19122110307216644, "learning_rate": 8.733410890924318e-05, "loss": 0.0148, "step": 81020 }, { "epoch": 51.64435946462715, "grad_norm": 0.4005959630012512, "learning_rate": 8.733044249356298e-05, "loss": 0.0224, "step": 81030 }, { "epoch": 51.65073295092416, "grad_norm": 0.4438916742801666, "learning_rate": 8.732677562428264e-05, "loss": 0.0267, "step": 81040 }, { "epoch": 51.65710643722116, "grad_norm": 0.515895426273346, "learning_rate": 8.732310830144675e-05, "loss": 0.0218, "step": 81050 }, { "epoch": 51.66347992351817, "grad_norm": 0.3400057256221771, "learning_rate": 8.731944052509985e-05, "loss": 0.0144, "step": 81060 }, { "epoch": 51.66985340981517, "grad_norm": 0.3960772454738617, "learning_rate": 8.731577229528652e-05, "loss": 0.0242, "step": 81070 }, { "epoch": 51.676226896112176, "grad_norm": 0.37429627776145935, "learning_rate": 8.73121036120513e-05, "loss": 0.0265, "step": 81080 }, { "epoch": 51.68260038240918, "grad_norm": 0.3340900242328644, "learning_rate": 8.730843447543882e-05, "loss": 0.0244, "step": 81090 }, { "epoch": 51.688973868706185, "grad_norm": 0.386396586894989, "learning_rate": 8.730476488549362e-05, "loss": 0.0175, "step": 81100 }, { "epoch": 51.695347355003186, "grad_norm": 0.511213481426239, "learning_rate": 8.73010948422603e-05, "loss": 0.0235, "step": 81110 }, { "epoch": 51.701720841300194, "grad_norm": 0.48626789450645447, "learning_rate": 8.729742434578348e-05, "loss": 0.0139, "step": 81120 }, { "epoch": 51.708094327597195, "grad_norm": 0.45288586616516113, "learning_rate": 8.729375339610773e-05, "loss": 0.0177, "step": 81130 }, { "epoch": 51.7144678138942, "grad_norm": 0.19390925765037537, "learning_rate": 8.729008199327766e-05, "loss": 0.0226, "step": 81140 }, { "epoch": 51.7208413001912, "grad_norm": 0.5896400213241577, "learning_rate": 8.728641013733789e-05, "loss": 0.0154, "step": 81150 }, { "epoch": 51.72721478648821, "grad_norm": 0.504985511302948, "learning_rate": 8.728273782833303e-05, "loss": 0.0147, "step": 81160 }, { "epoch": 51.73358827278521, "grad_norm": 0.7524563670158386, "learning_rate": 8.727906506630769e-05, "loss": 0.0209, "step": 81170 }, { "epoch": 51.73996175908222, "grad_norm": 0.22519922256469727, "learning_rate": 8.727539185130654e-05, "loss": 0.0248, "step": 81180 }, { "epoch": 51.74633524537922, "grad_norm": 0.4732171893119812, "learning_rate": 8.727171818337417e-05, "loss": 0.0205, "step": 81190 }, { "epoch": 51.75270873167623, "grad_norm": 0.5426949858665466, "learning_rate": 8.726804406255525e-05, "loss": 0.0294, "step": 81200 }, { "epoch": 51.75908221797323, "grad_norm": 0.46127068996429443, "learning_rate": 8.726436948889438e-05, "loss": 0.0241, "step": 81210 }, { "epoch": 51.76545570427024, "grad_norm": 0.40696483850479126, "learning_rate": 8.726069446243625e-05, "loss": 0.0128, "step": 81220 }, { "epoch": 51.77182919056724, "grad_norm": 0.35702329874038696, "learning_rate": 8.725701898322551e-05, "loss": 0.0174, "step": 81230 }, { "epoch": 51.77820267686425, "grad_norm": 0.28448814153671265, "learning_rate": 8.725334305130682e-05, "loss": 0.0173, "step": 81240 }, { "epoch": 51.78457616316125, "grad_norm": 0.3182666599750519, "learning_rate": 8.724966666672481e-05, "loss": 0.0157, "step": 81250 }, { "epoch": 51.790949649458256, "grad_norm": 0.3836197555065155, "learning_rate": 8.724598982952421e-05, "loss": 0.0196, "step": 81260 }, { "epoch": 51.797323135755256, "grad_norm": 0.5858541131019592, "learning_rate": 8.724231253974964e-05, "loss": 0.0233, "step": 81270 }, { "epoch": 51.803696622052264, "grad_norm": 0.41204991936683655, "learning_rate": 8.723863479744583e-05, "loss": 0.0215, "step": 81280 }, { "epoch": 51.810070108349265, "grad_norm": 0.4540404975414276, "learning_rate": 8.723495660265745e-05, "loss": 0.0141, "step": 81290 }, { "epoch": 51.81644359464627, "grad_norm": 0.5977151989936829, "learning_rate": 8.723127795542919e-05, "loss": 0.0183, "step": 81300 }, { "epoch": 51.822817080943274, "grad_norm": 0.3443578779697418, "learning_rate": 8.722759885580573e-05, "loss": 0.0212, "step": 81310 }, { "epoch": 51.82919056724028, "grad_norm": 0.4912019371986389, "learning_rate": 8.722391930383182e-05, "loss": 0.0321, "step": 81320 }, { "epoch": 51.83556405353728, "grad_norm": 0.3972175419330597, "learning_rate": 8.722023929955213e-05, "loss": 0.0219, "step": 81330 }, { "epoch": 51.84193753983429, "grad_norm": 0.4738822281360626, "learning_rate": 8.72165588430114e-05, "loss": 0.0233, "step": 81340 }, { "epoch": 51.84831102613129, "grad_norm": 0.6504071354866028, "learning_rate": 8.721287793425434e-05, "loss": 0.0305, "step": 81350 }, { "epoch": 51.8546845124283, "grad_norm": 0.6047699451446533, "learning_rate": 8.720919657332566e-05, "loss": 0.0319, "step": 81360 }, { "epoch": 51.8610579987253, "grad_norm": 0.47295376658439636, "learning_rate": 8.720551476027015e-05, "loss": 0.0347, "step": 81370 }, { "epoch": 51.86743148502231, "grad_norm": 0.5706363916397095, "learning_rate": 8.720183249513248e-05, "loss": 0.0418, "step": 81380 }, { "epoch": 51.87380497131931, "grad_norm": 0.40852758288383484, "learning_rate": 8.719814977795741e-05, "loss": 0.0174, "step": 81390 }, { "epoch": 51.88017845761632, "grad_norm": 0.48534706234931946, "learning_rate": 8.719446660878973e-05, "loss": 0.0223, "step": 81400 }, { "epoch": 51.88655194391332, "grad_norm": 0.6410535573959351, "learning_rate": 8.719078298767415e-05, "loss": 0.0184, "step": 81410 }, { "epoch": 51.892925430210326, "grad_norm": 0.4877389967441559, "learning_rate": 8.718709891465544e-05, "loss": 0.024, "step": 81420 }, { "epoch": 51.89929891650733, "grad_norm": 0.5058755874633789, "learning_rate": 8.718341438977837e-05, "loss": 0.0194, "step": 81430 }, { "epoch": 51.905672402804335, "grad_norm": 0.31806501746177673, "learning_rate": 8.71797294130877e-05, "loss": 0.035, "step": 81440 }, { "epoch": 51.912045889101336, "grad_norm": 0.3037961721420288, "learning_rate": 8.717604398462822e-05, "loss": 0.0179, "step": 81450 }, { "epoch": 51.918419375398344, "grad_norm": 0.5935440063476562, "learning_rate": 8.717235810444471e-05, "loss": 0.0216, "step": 81460 }, { "epoch": 51.924792861695344, "grad_norm": 0.5404843091964722, "learning_rate": 8.716867177258194e-05, "loss": 0.0142, "step": 81470 }, { "epoch": 51.93116634799235, "grad_norm": 0.5413027405738831, "learning_rate": 8.716498498908473e-05, "loss": 0.0224, "step": 81480 }, { "epoch": 51.93753983428935, "grad_norm": 0.3893929421901703, "learning_rate": 8.716129775399786e-05, "loss": 0.0206, "step": 81490 }, { "epoch": 51.94391332058636, "grad_norm": 0.3883205056190491, "learning_rate": 8.715761006736614e-05, "loss": 0.0208, "step": 81500 }, { "epoch": 51.95028680688336, "grad_norm": 0.38589707016944885, "learning_rate": 8.715392192923437e-05, "loss": 0.0243, "step": 81510 }, { "epoch": 51.95666029318037, "grad_norm": 0.1656983643770218, "learning_rate": 8.715023333964736e-05, "loss": 0.0225, "step": 81520 }, { "epoch": 51.96303377947737, "grad_norm": 0.5262556076049805, "learning_rate": 8.714654429864994e-05, "loss": 0.0148, "step": 81530 }, { "epoch": 51.96940726577438, "grad_norm": 0.308219850063324, "learning_rate": 8.714285480628694e-05, "loss": 0.0243, "step": 81540 }, { "epoch": 51.97578075207138, "grad_norm": 0.6634693145751953, "learning_rate": 8.713916486260319e-05, "loss": 0.0235, "step": 81550 }, { "epoch": 51.98215423836839, "grad_norm": 1.0083328485488892, "learning_rate": 8.713547446764352e-05, "loss": 0.0326, "step": 81560 }, { "epoch": 51.98852772466539, "grad_norm": 0.46199601888656616, "learning_rate": 8.713178362145278e-05, "loss": 0.0208, "step": 81570 }, { "epoch": 51.9949012109624, "grad_norm": 0.5431836843490601, "learning_rate": 8.712809232407579e-05, "loss": 0.0211, "step": 81580 }, { "epoch": 52.001274697259404, "grad_norm": 0.9592965841293335, "learning_rate": 8.712440057555744e-05, "loss": 0.0247, "step": 81590 }, { "epoch": 52.007648183556405, "grad_norm": 0.3221082389354706, "learning_rate": 8.712070837594257e-05, "loss": 0.016, "step": 81600 }, { "epoch": 52.01402166985341, "grad_norm": 0.5910373330116272, "learning_rate": 8.711701572527605e-05, "loss": 0.0214, "step": 81610 }, { "epoch": 52.020395156150414, "grad_norm": 0.2582823634147644, "learning_rate": 8.711332262360274e-05, "loss": 0.0159, "step": 81620 }, { "epoch": 52.02676864244742, "grad_norm": 0.3554055988788605, "learning_rate": 8.710962907096751e-05, "loss": 0.0187, "step": 81630 }, { "epoch": 52.03314212874442, "grad_norm": 0.3477265536785126, "learning_rate": 8.710593506741526e-05, "loss": 0.0209, "step": 81640 }, { "epoch": 52.03951561504143, "grad_norm": 0.3982335925102234, "learning_rate": 8.710224061299086e-05, "loss": 0.0254, "step": 81650 }, { "epoch": 52.04588910133843, "grad_norm": 0.3781326413154602, "learning_rate": 8.709854570773921e-05, "loss": 0.0348, "step": 81660 }, { "epoch": 52.05226258763544, "grad_norm": 0.5378394722938538, "learning_rate": 8.70948503517052e-05, "loss": 0.0174, "step": 81670 }, { "epoch": 52.05863607393244, "grad_norm": 0.8060970902442932, "learning_rate": 8.709115454493375e-05, "loss": 0.0227, "step": 81680 }, { "epoch": 52.06500956022945, "grad_norm": 0.4416046738624573, "learning_rate": 8.708745828746973e-05, "loss": 0.0186, "step": 81690 }, { "epoch": 52.07138304652645, "grad_norm": 0.6488023996353149, "learning_rate": 8.708376157935808e-05, "loss": 0.019, "step": 81700 }, { "epoch": 52.07775653282346, "grad_norm": 0.3122628927230835, "learning_rate": 8.708006442064372e-05, "loss": 0.032, "step": 81710 }, { "epoch": 52.08413001912046, "grad_norm": 0.5989399552345276, "learning_rate": 8.707636681137156e-05, "loss": 0.0189, "step": 81720 }, { "epoch": 52.090503505417466, "grad_norm": 0.5266742706298828, "learning_rate": 8.707266875158655e-05, "loss": 0.0168, "step": 81730 }, { "epoch": 52.09687699171447, "grad_norm": 0.3032207489013672, "learning_rate": 8.70689702413336e-05, "loss": 0.0275, "step": 81740 }, { "epoch": 52.103250478011475, "grad_norm": 0.5930258631706238, "learning_rate": 8.706527128065767e-05, "loss": 0.0333, "step": 81750 }, { "epoch": 52.109623964308476, "grad_norm": 0.5194653868675232, "learning_rate": 8.706157186960371e-05, "loss": 0.0199, "step": 81760 }, { "epoch": 52.115997450605484, "grad_norm": 0.705303966999054, "learning_rate": 8.705787200821664e-05, "loss": 0.0206, "step": 81770 }, { "epoch": 52.122370936902485, "grad_norm": 0.29638293385505676, "learning_rate": 8.705417169654145e-05, "loss": 0.0323, "step": 81780 }, { "epoch": 52.12874442319949, "grad_norm": 0.21216413378715515, "learning_rate": 8.705047093462307e-05, "loss": 0.0181, "step": 81790 }, { "epoch": 52.13511790949649, "grad_norm": 0.5359810590744019, "learning_rate": 8.704676972250651e-05, "loss": 0.0151, "step": 81800 }, { "epoch": 52.1414913957935, "grad_norm": 0.2751602232456207, "learning_rate": 8.70430680602367e-05, "loss": 0.0323, "step": 81810 }, { "epoch": 52.1478648820905, "grad_norm": 0.4097042679786682, "learning_rate": 8.703936594785865e-05, "loss": 0.0192, "step": 81820 }, { "epoch": 52.15423836838751, "grad_norm": 0.37493911385536194, "learning_rate": 8.703566338541733e-05, "loss": 0.0166, "step": 81830 }, { "epoch": 52.16061185468451, "grad_norm": 0.6698934435844421, "learning_rate": 8.703196037295773e-05, "loss": 0.026, "step": 81840 }, { "epoch": 52.16698534098152, "grad_norm": 0.36260277032852173, "learning_rate": 8.702825691052486e-05, "loss": 0.0309, "step": 81850 }, { "epoch": 52.17335882727852, "grad_norm": 0.1413009911775589, "learning_rate": 8.70245529981637e-05, "loss": 0.0101, "step": 81860 }, { "epoch": 52.17973231357553, "grad_norm": 0.4117658734321594, "learning_rate": 8.702084863591926e-05, "loss": 0.0273, "step": 81870 }, { "epoch": 52.18610579987253, "grad_norm": 0.47552376985549927, "learning_rate": 8.701714382383655e-05, "loss": 0.0185, "step": 81880 }, { "epoch": 52.19247928616954, "grad_norm": 0.6598080992698669, "learning_rate": 8.701343856196061e-05, "loss": 0.021, "step": 81890 }, { "epoch": 52.19885277246654, "grad_norm": 0.3817600607872009, "learning_rate": 8.700973285033643e-05, "loss": 0.0289, "step": 81900 }, { "epoch": 52.205226258763545, "grad_norm": 0.5875312089920044, "learning_rate": 8.700602668900906e-05, "loss": 0.0217, "step": 81910 }, { "epoch": 52.211599745060546, "grad_norm": 0.37648648023605347, "learning_rate": 8.700232007802351e-05, "loss": 0.0226, "step": 81920 }, { "epoch": 52.217973231357554, "grad_norm": 0.5557661056518555, "learning_rate": 8.699861301742486e-05, "loss": 0.0194, "step": 81930 }, { "epoch": 52.224346717654555, "grad_norm": 0.7103478312492371, "learning_rate": 8.699490550725811e-05, "loss": 0.0197, "step": 81940 }, { "epoch": 52.23072020395156, "grad_norm": 0.3458067774772644, "learning_rate": 8.699119754756835e-05, "loss": 0.0196, "step": 81950 }, { "epoch": 52.237093690248564, "grad_norm": 0.5022388100624084, "learning_rate": 8.69874891384006e-05, "loss": 0.0181, "step": 81960 }, { "epoch": 52.24346717654557, "grad_norm": 0.4194103479385376, "learning_rate": 8.698378027979994e-05, "loss": 0.0202, "step": 81970 }, { "epoch": 52.24984066284257, "grad_norm": 0.3981282114982605, "learning_rate": 8.698007097181142e-05, "loss": 0.0306, "step": 81980 }, { "epoch": 52.25621414913958, "grad_norm": 0.5014811158180237, "learning_rate": 8.697636121448014e-05, "loss": 0.0189, "step": 81990 }, { "epoch": 52.26258763543658, "grad_norm": 0.3814089596271515, "learning_rate": 8.697265100785115e-05, "loss": 0.0254, "step": 82000 }, { "epoch": 52.26896112173359, "grad_norm": 0.4989509880542755, "learning_rate": 8.696894035196953e-05, "loss": 0.0155, "step": 82010 }, { "epoch": 52.27533460803059, "grad_norm": 0.18871840834617615, "learning_rate": 8.696522924688041e-05, "loss": 0.0206, "step": 82020 }, { "epoch": 52.2817080943276, "grad_norm": 0.2247210443019867, "learning_rate": 8.696151769262883e-05, "loss": 0.027, "step": 82030 }, { "epoch": 52.2880815806246, "grad_norm": 0.3103176951408386, "learning_rate": 8.695780568925992e-05, "loss": 0.0327, "step": 82040 }, { "epoch": 52.29445506692161, "grad_norm": 0.6794428825378418, "learning_rate": 8.695409323681877e-05, "loss": 0.0204, "step": 82050 }, { "epoch": 52.30082855321861, "grad_norm": 0.3984158933162689, "learning_rate": 8.695038033535052e-05, "loss": 0.0168, "step": 82060 }, { "epoch": 52.307202039515616, "grad_norm": 0.33990049362182617, "learning_rate": 8.694666698490024e-05, "loss": 0.0217, "step": 82070 }, { "epoch": 52.31357552581262, "grad_norm": 0.3129803538322449, "learning_rate": 8.694295318551309e-05, "loss": 0.0202, "step": 82080 }, { "epoch": 52.319949012109625, "grad_norm": 0.21619196236133575, "learning_rate": 8.693923893723416e-05, "loss": 0.0222, "step": 82090 }, { "epoch": 52.326322498406626, "grad_norm": 0.24451348185539246, "learning_rate": 8.69355242401086e-05, "loss": 0.0229, "step": 82100 }, { "epoch": 52.33269598470363, "grad_norm": 0.28829383850097656, "learning_rate": 8.693180909418156e-05, "loss": 0.0209, "step": 82110 }, { "epoch": 52.339069471000634, "grad_norm": 0.6541208028793335, "learning_rate": 8.692809349949816e-05, "loss": 0.0183, "step": 82120 }, { "epoch": 52.34544295729764, "grad_norm": 0.5769757032394409, "learning_rate": 8.692437745610357e-05, "loss": 0.0213, "step": 82130 }, { "epoch": 52.35181644359464, "grad_norm": 0.40215086936950684, "learning_rate": 8.692066096404293e-05, "loss": 0.0151, "step": 82140 }, { "epoch": 52.35818992989165, "grad_norm": 0.4064716696739197, "learning_rate": 8.691694402336139e-05, "loss": 0.0358, "step": 82150 }, { "epoch": 52.36456341618865, "grad_norm": 0.5725083351135254, "learning_rate": 8.691322663410413e-05, "loss": 0.0183, "step": 82160 }, { "epoch": 52.37093690248566, "grad_norm": 0.5562969446182251, "learning_rate": 8.690950879631632e-05, "loss": 0.0228, "step": 82170 }, { "epoch": 52.37731038878266, "grad_norm": 0.4773479998111725, "learning_rate": 8.690579051004312e-05, "loss": 0.0212, "step": 82180 }, { "epoch": 52.38368387507967, "grad_norm": 0.36255648732185364, "learning_rate": 8.690207177532971e-05, "loss": 0.0239, "step": 82190 }, { "epoch": 52.39005736137667, "grad_norm": 0.25858187675476074, "learning_rate": 8.689835259222132e-05, "loss": 0.0125, "step": 82200 }, { "epoch": 52.39643084767368, "grad_norm": 0.41720646619796753, "learning_rate": 8.689463296076308e-05, "loss": 0.0382, "step": 82210 }, { "epoch": 52.40280433397068, "grad_norm": 0.35158684849739075, "learning_rate": 8.689091288100025e-05, "loss": 0.0183, "step": 82220 }, { "epoch": 52.409177820267686, "grad_norm": 0.47380325198173523, "learning_rate": 8.688719235297796e-05, "loss": 0.0183, "step": 82230 }, { "epoch": 52.415551306564694, "grad_norm": 0.42066654562950134, "learning_rate": 8.688347137674146e-05, "loss": 0.0255, "step": 82240 }, { "epoch": 52.421924792861695, "grad_norm": 0.6499806642532349, "learning_rate": 8.687974995233597e-05, "loss": 0.0325, "step": 82250 }, { "epoch": 52.4282982791587, "grad_norm": 0.33543363213539124, "learning_rate": 8.68760280798067e-05, "loss": 0.0188, "step": 82260 }, { "epoch": 52.434671765455704, "grad_norm": 0.25506868958473206, "learning_rate": 8.687230575919886e-05, "loss": 0.0277, "step": 82270 }, { "epoch": 52.44104525175271, "grad_norm": 0.347199410200119, "learning_rate": 8.68685829905577e-05, "loss": 0.0201, "step": 82280 }, { "epoch": 52.44741873804971, "grad_norm": 0.5167614221572876, "learning_rate": 8.686485977392844e-05, "loss": 0.0169, "step": 82290 }, { "epoch": 52.45379222434672, "grad_norm": 0.5606223940849304, "learning_rate": 8.686113610935634e-05, "loss": 0.0391, "step": 82300 }, { "epoch": 52.46016571064372, "grad_norm": 0.38621944189071655, "learning_rate": 8.685741199688662e-05, "loss": 0.0271, "step": 82310 }, { "epoch": 52.46653919694073, "grad_norm": 0.4576461911201477, "learning_rate": 8.685368743656456e-05, "loss": 0.0151, "step": 82320 }, { "epoch": 52.47291268323773, "grad_norm": 0.493071973323822, "learning_rate": 8.684996242843539e-05, "loss": 0.0232, "step": 82330 }, { "epoch": 52.47928616953474, "grad_norm": 0.346667617559433, "learning_rate": 8.684623697254439e-05, "loss": 0.0276, "step": 82340 }, { "epoch": 52.48565965583174, "grad_norm": 0.4078003764152527, "learning_rate": 8.684251106893681e-05, "loss": 0.0229, "step": 82350 }, { "epoch": 52.49203314212875, "grad_norm": 0.6541142463684082, "learning_rate": 8.683878471765796e-05, "loss": 0.0139, "step": 82360 }, { "epoch": 52.49840662842575, "grad_norm": 0.6681216359138489, "learning_rate": 8.683505791875308e-05, "loss": 0.0222, "step": 82370 }, { "epoch": 52.504780114722756, "grad_norm": 0.3047804832458496, "learning_rate": 8.683133067226748e-05, "loss": 0.0181, "step": 82380 }, { "epoch": 52.51115360101976, "grad_norm": 0.24645927548408508, "learning_rate": 8.682760297824641e-05, "loss": 0.0192, "step": 82390 }, { "epoch": 52.517527087316765, "grad_norm": 0.4493977427482605, "learning_rate": 8.682387483673522e-05, "loss": 0.0237, "step": 82400 }, { "epoch": 52.523900573613766, "grad_norm": 0.4152500331401825, "learning_rate": 8.682014624777917e-05, "loss": 0.0236, "step": 82410 }, { "epoch": 52.530274059910774, "grad_norm": 0.4852845072746277, "learning_rate": 8.681641721142358e-05, "loss": 0.0321, "step": 82420 }, { "epoch": 52.536647546207774, "grad_norm": 0.7115806341171265, "learning_rate": 8.681268772771376e-05, "loss": 0.0306, "step": 82430 }, { "epoch": 52.54302103250478, "grad_norm": 0.4402046501636505, "learning_rate": 8.680895779669502e-05, "loss": 0.0205, "step": 82440 }, { "epoch": 52.54939451880178, "grad_norm": 0.24096879363059998, "learning_rate": 8.680522741841271e-05, "loss": 0.0212, "step": 82450 }, { "epoch": 52.55576800509879, "grad_norm": 0.3535328209400177, "learning_rate": 8.680149659291212e-05, "loss": 0.0247, "step": 82460 }, { "epoch": 52.56214149139579, "grad_norm": 0.6806064248085022, "learning_rate": 8.679776532023861e-05, "loss": 0.0214, "step": 82470 }, { "epoch": 52.5685149776928, "grad_norm": 0.27704957127571106, "learning_rate": 8.679403360043749e-05, "loss": 0.0159, "step": 82480 }, { "epoch": 52.5748884639898, "grad_norm": 0.4703468680381775, "learning_rate": 8.679030143355416e-05, "loss": 0.0215, "step": 82490 }, { "epoch": 52.58126195028681, "grad_norm": 0.3734597861766815, "learning_rate": 8.67865688196339e-05, "loss": 0.0177, "step": 82500 }, { "epoch": 52.58763543658381, "grad_norm": 0.37866517901420593, "learning_rate": 8.67828357587221e-05, "loss": 0.0195, "step": 82510 }, { "epoch": 52.59400892288082, "grad_norm": 0.18547266721725464, "learning_rate": 8.677910225086412e-05, "loss": 0.0174, "step": 82520 }, { "epoch": 52.60038240917782, "grad_norm": 0.6179351210594177, "learning_rate": 8.677536829610533e-05, "loss": 0.0231, "step": 82530 }, { "epoch": 52.606755895474826, "grad_norm": 0.28108468651771545, "learning_rate": 8.677163389449108e-05, "loss": 0.0274, "step": 82540 }, { "epoch": 52.61312938177183, "grad_norm": 0.48772165179252625, "learning_rate": 8.676789904606676e-05, "loss": 0.0262, "step": 82550 }, { "epoch": 52.619502868068835, "grad_norm": 0.271126925945282, "learning_rate": 8.676416375087776e-05, "loss": 0.0207, "step": 82560 }, { "epoch": 52.625876354365836, "grad_norm": 0.2954702079296112, "learning_rate": 8.676042800896947e-05, "loss": 0.0163, "step": 82570 }, { "epoch": 52.632249840662844, "grad_norm": 0.3048221170902252, "learning_rate": 8.675669182038725e-05, "loss": 0.0449, "step": 82580 }, { "epoch": 52.638623326959845, "grad_norm": 0.4199792742729187, "learning_rate": 8.675295518517655e-05, "loss": 0.0228, "step": 82590 }, { "epoch": 52.64499681325685, "grad_norm": 0.2763124108314514, "learning_rate": 8.674921810338272e-05, "loss": 0.015, "step": 82600 }, { "epoch": 52.651370299553854, "grad_norm": 0.6224765181541443, "learning_rate": 8.674548057505119e-05, "loss": 0.0261, "step": 82610 }, { "epoch": 52.65774378585086, "grad_norm": 0.3449157178401947, "learning_rate": 8.674174260022738e-05, "loss": 0.0249, "step": 82620 }, { "epoch": 52.66411727214786, "grad_norm": 0.2401883900165558, "learning_rate": 8.673800417895672e-05, "loss": 0.0291, "step": 82630 }, { "epoch": 52.67049075844487, "grad_norm": 0.35523858666419983, "learning_rate": 8.67342653112846e-05, "loss": 0.0272, "step": 82640 }, { "epoch": 52.67686424474187, "grad_norm": 0.5049643516540527, "learning_rate": 8.673052599725651e-05, "loss": 0.0239, "step": 82650 }, { "epoch": 52.68323773103888, "grad_norm": 0.46251335740089417, "learning_rate": 8.672678623691783e-05, "loss": 0.0299, "step": 82660 }, { "epoch": 52.68961121733588, "grad_norm": 0.5315220952033997, "learning_rate": 8.672304603031402e-05, "loss": 0.0272, "step": 82670 }, { "epoch": 52.69598470363289, "grad_norm": 0.6401378512382507, "learning_rate": 8.671930537749053e-05, "loss": 0.0187, "step": 82680 }, { "epoch": 52.70235818992989, "grad_norm": 0.6892049312591553, "learning_rate": 8.67155642784928e-05, "loss": 0.0248, "step": 82690 }, { "epoch": 52.7087316762269, "grad_norm": 0.5373736023902893, "learning_rate": 8.671182273336632e-05, "loss": 0.0204, "step": 82700 }, { "epoch": 52.7151051625239, "grad_norm": 0.45758479833602905, "learning_rate": 8.67080807421565e-05, "loss": 0.0222, "step": 82710 }, { "epoch": 52.721478648820906, "grad_norm": 0.26840701699256897, "learning_rate": 8.670433830490887e-05, "loss": 0.022, "step": 82720 }, { "epoch": 52.72785213511791, "grad_norm": 0.6195713877677917, "learning_rate": 8.670059542166887e-05, "loss": 0.0182, "step": 82730 }, { "epoch": 52.734225621414915, "grad_norm": 0.4386936128139496, "learning_rate": 8.669685209248198e-05, "loss": 0.0192, "step": 82740 }, { "epoch": 52.740599107711915, "grad_norm": 0.23305873572826385, "learning_rate": 8.669310831739369e-05, "loss": 0.0172, "step": 82750 }, { "epoch": 52.74697259400892, "grad_norm": 0.43298399448394775, "learning_rate": 8.66893640964495e-05, "loss": 0.0237, "step": 82760 }, { "epoch": 52.753346080305924, "grad_norm": 0.6129809617996216, "learning_rate": 8.668561942969488e-05, "loss": 0.0235, "step": 82770 }, { "epoch": 52.75971956660293, "grad_norm": 0.4076831042766571, "learning_rate": 8.668187431717535e-05, "loss": 0.016, "step": 82780 }, { "epoch": 52.76609305289993, "grad_norm": 0.558911919593811, "learning_rate": 8.667812875893641e-05, "loss": 0.0197, "step": 82790 }, { "epoch": 52.77246653919694, "grad_norm": 0.540917694568634, "learning_rate": 8.66743827550236e-05, "loss": 0.0239, "step": 82800 }, { "epoch": 52.77884002549394, "grad_norm": 0.2936572730541229, "learning_rate": 8.66706363054824e-05, "loss": 0.0186, "step": 82810 }, { "epoch": 52.78521351179095, "grad_norm": 0.46870535612106323, "learning_rate": 8.666688941035832e-05, "loss": 0.0274, "step": 82820 }, { "epoch": 52.79158699808795, "grad_norm": 0.3715057969093323, "learning_rate": 8.666314206969694e-05, "loss": 0.0182, "step": 82830 }, { "epoch": 52.79796048438496, "grad_norm": 0.5497382879257202, "learning_rate": 8.665939428354377e-05, "loss": 0.0181, "step": 82840 }, { "epoch": 52.80433397068197, "grad_norm": 0.19983555376529694, "learning_rate": 8.665564605194434e-05, "loss": 0.0189, "step": 82850 }, { "epoch": 52.81070745697897, "grad_norm": 0.6503185033798218, "learning_rate": 8.66518973749442e-05, "loss": 0.0302, "step": 82860 }, { "epoch": 52.817080943275975, "grad_norm": 0.4079761505126953, "learning_rate": 8.664814825258892e-05, "loss": 0.0186, "step": 82870 }, { "epoch": 52.823454429572976, "grad_norm": 0.6760707497596741, "learning_rate": 8.664439868492402e-05, "loss": 0.03, "step": 82880 }, { "epoch": 52.829827915869984, "grad_norm": 0.5822170972824097, "learning_rate": 8.664064867199508e-05, "loss": 0.03, "step": 82890 }, { "epoch": 52.836201402166985, "grad_norm": 0.21028010547161102, "learning_rate": 8.663689821384766e-05, "loss": 0.0175, "step": 82900 }, { "epoch": 52.84257488846399, "grad_norm": 1.0702921152114868, "learning_rate": 8.663314731052734e-05, "loss": 0.0365, "step": 82910 }, { "epoch": 52.848948374760994, "grad_norm": 0.6527359485626221, "learning_rate": 8.662939596207968e-05, "loss": 0.0246, "step": 82920 }, { "epoch": 52.855321861058, "grad_norm": 0.8144825100898743, "learning_rate": 8.66256441685503e-05, "loss": 0.0306, "step": 82930 }, { "epoch": 52.861695347355, "grad_norm": 0.3406814932823181, "learning_rate": 8.662189192998474e-05, "loss": 0.0278, "step": 82940 }, { "epoch": 52.86806883365201, "grad_norm": 0.3905585706233978, "learning_rate": 8.661813924642863e-05, "loss": 0.0207, "step": 82950 }, { "epoch": 52.87444231994901, "grad_norm": 0.24959905445575714, "learning_rate": 8.661438611792756e-05, "loss": 0.0174, "step": 82960 }, { "epoch": 52.88081580624602, "grad_norm": 0.9980722665786743, "learning_rate": 8.661063254452712e-05, "loss": 0.032, "step": 82970 }, { "epoch": 52.88718929254302, "grad_norm": 0.4151591360569, "learning_rate": 8.660687852627292e-05, "loss": 0.019, "step": 82980 }, { "epoch": 52.89356277884003, "grad_norm": 1.07515549659729, "learning_rate": 8.660312406321059e-05, "loss": 0.0188, "step": 82990 }, { "epoch": 52.89993626513703, "grad_norm": 0.6609443426132202, "learning_rate": 8.659936915538576e-05, "loss": 0.0414, "step": 83000 }, { "epoch": 52.90630975143404, "grad_norm": 0.4041118919849396, "learning_rate": 8.659561380284402e-05, "loss": 0.0175, "step": 83010 }, { "epoch": 52.91268323773104, "grad_norm": 0.5503021478652954, "learning_rate": 8.659185800563102e-05, "loss": 0.022, "step": 83020 }, { "epoch": 52.919056724028046, "grad_norm": 0.5025366544723511, "learning_rate": 8.658810176379241e-05, "loss": 0.027, "step": 83030 }, { "epoch": 52.92543021032505, "grad_norm": 0.4865161180496216, "learning_rate": 8.65843450773738e-05, "loss": 0.0164, "step": 83040 }, { "epoch": 52.931803696622055, "grad_norm": 0.30988237261772156, "learning_rate": 8.658058794642088e-05, "loss": 0.0262, "step": 83050 }, { "epoch": 52.938177182919055, "grad_norm": 0.44865894317626953, "learning_rate": 8.657683037097926e-05, "loss": 0.0185, "step": 83060 }, { "epoch": 52.94455066921606, "grad_norm": 0.5063388347625732, "learning_rate": 8.657307235109462e-05, "loss": 0.0284, "step": 83070 }, { "epoch": 52.950924155513064, "grad_norm": 0.3172318935394287, "learning_rate": 8.656931388681262e-05, "loss": 0.0172, "step": 83080 }, { "epoch": 52.95729764181007, "grad_norm": 0.5125734806060791, "learning_rate": 8.656555497817893e-05, "loss": 0.0212, "step": 83090 }, { "epoch": 52.96367112810707, "grad_norm": 0.694328784942627, "learning_rate": 8.656179562523922e-05, "loss": 0.0194, "step": 83100 }, { "epoch": 52.97004461440408, "grad_norm": 0.551508903503418, "learning_rate": 8.655803582803919e-05, "loss": 0.025, "step": 83110 }, { "epoch": 52.97641810070108, "grad_norm": 0.4099421799182892, "learning_rate": 8.65542755866245e-05, "loss": 0.026, "step": 83120 }, { "epoch": 52.98279158699809, "grad_norm": 0.22747386991977692, "learning_rate": 8.655051490104085e-05, "loss": 0.0267, "step": 83130 }, { "epoch": 52.98916507329509, "grad_norm": 0.2044578194618225, "learning_rate": 8.654675377133392e-05, "loss": 0.0217, "step": 83140 }, { "epoch": 52.9955385595921, "grad_norm": 0.6027621030807495, "learning_rate": 8.654299219754945e-05, "loss": 0.0279, "step": 83150 }, { "epoch": 53.0019120458891, "grad_norm": 0.4147590100765228, "learning_rate": 8.65392301797331e-05, "loss": 0.0244, "step": 83160 }, { "epoch": 53.00828553218611, "grad_norm": 0.2936416566371918, "learning_rate": 8.653546771793061e-05, "loss": 0.0378, "step": 83170 }, { "epoch": 53.01465901848311, "grad_norm": 0.48172932863235474, "learning_rate": 8.653170481218769e-05, "loss": 0.0227, "step": 83180 }, { "epoch": 53.021032504780116, "grad_norm": 0.19088655710220337, "learning_rate": 8.652794146255008e-05, "loss": 0.0353, "step": 83190 }, { "epoch": 53.02740599107712, "grad_norm": 0.35283172130584717, "learning_rate": 8.652417766906348e-05, "loss": 0.0234, "step": 83200 }, { "epoch": 53.033779477374125, "grad_norm": 0.353640615940094, "learning_rate": 8.652041343177364e-05, "loss": 0.0249, "step": 83210 }, { "epoch": 53.040152963671126, "grad_norm": 0.38593441247940063, "learning_rate": 8.651664875072629e-05, "loss": 0.0282, "step": 83220 }, { "epoch": 53.046526449968134, "grad_norm": 0.43414387106895447, "learning_rate": 8.651288362596718e-05, "loss": 0.0166, "step": 83230 }, { "epoch": 53.052899936265135, "grad_norm": 0.3523552119731903, "learning_rate": 8.650911805754208e-05, "loss": 0.0172, "step": 83240 }, { "epoch": 53.05927342256214, "grad_norm": 0.4244237542152405, "learning_rate": 8.65053520454967e-05, "loss": 0.0278, "step": 83250 }, { "epoch": 53.065646908859144, "grad_norm": 0.22153964638710022, "learning_rate": 8.650158558987685e-05, "loss": 0.0272, "step": 83260 }, { "epoch": 53.07202039515615, "grad_norm": 0.3531704545021057, "learning_rate": 8.649781869072826e-05, "loss": 0.0235, "step": 83270 }, { "epoch": 53.07839388145315, "grad_norm": 0.2647213339805603, "learning_rate": 8.64940513480967e-05, "loss": 0.0175, "step": 83280 }, { "epoch": 53.08476736775016, "grad_norm": 0.32380354404449463, "learning_rate": 8.649028356202798e-05, "loss": 0.0241, "step": 83290 }, { "epoch": 53.09114085404716, "grad_norm": 0.4932224452495575, "learning_rate": 8.648651533256785e-05, "loss": 0.0308, "step": 83300 }, { "epoch": 53.09751434034417, "grad_norm": 0.2854040861129761, "learning_rate": 8.648274665976211e-05, "loss": 0.0131, "step": 83310 }, { "epoch": 53.10388782664117, "grad_norm": 0.34878554940223694, "learning_rate": 8.647897754365656e-05, "loss": 0.0214, "step": 83320 }, { "epoch": 53.11026131293818, "grad_norm": 0.40313050150871277, "learning_rate": 8.647520798429698e-05, "loss": 0.0159, "step": 83330 }, { "epoch": 53.11663479923518, "grad_norm": 0.6901426911354065, "learning_rate": 8.647143798172919e-05, "loss": 0.0302, "step": 83340 }, { "epoch": 53.12300828553219, "grad_norm": 0.32066458463668823, "learning_rate": 8.6467667535999e-05, "loss": 0.0234, "step": 83350 }, { "epoch": 53.12938177182919, "grad_norm": 0.6121717095375061, "learning_rate": 8.646389664715223e-05, "loss": 0.0231, "step": 83360 }, { "epoch": 53.135755258126196, "grad_norm": 0.3708300292491913, "learning_rate": 8.646012531523467e-05, "loss": 0.028, "step": 83370 }, { "epoch": 53.1421287444232, "grad_norm": 0.46055150032043457, "learning_rate": 8.645635354029216e-05, "loss": 0.0269, "step": 83380 }, { "epoch": 53.148502230720204, "grad_norm": 0.5853471755981445, "learning_rate": 8.645258132237054e-05, "loss": 0.0262, "step": 83390 }, { "epoch": 53.154875717017205, "grad_norm": 0.6062769889831543, "learning_rate": 8.644880866151564e-05, "loss": 0.0201, "step": 83400 }, { "epoch": 53.16124920331421, "grad_norm": 0.510599672794342, "learning_rate": 8.64450355577733e-05, "loss": 0.024, "step": 83410 }, { "epoch": 53.167622689611214, "grad_norm": 0.322971373796463, "learning_rate": 8.644126201118936e-05, "loss": 0.018, "step": 83420 }, { "epoch": 53.17399617590822, "grad_norm": 0.46966373920440674, "learning_rate": 8.64374880218097e-05, "loss": 0.0226, "step": 83430 }, { "epoch": 53.18036966220522, "grad_norm": 0.6031379103660583, "learning_rate": 8.643371358968015e-05, "loss": 0.0241, "step": 83440 }, { "epoch": 53.18674314850223, "grad_norm": 0.3685668110847473, "learning_rate": 8.642993871484657e-05, "loss": 0.0146, "step": 83450 }, { "epoch": 53.19311663479923, "grad_norm": 0.5559024810791016, "learning_rate": 8.642616339735485e-05, "loss": 0.015, "step": 83460 }, { "epoch": 53.19949012109624, "grad_norm": 0.3526475727558136, "learning_rate": 8.642238763725085e-05, "loss": 0.0178, "step": 83470 }, { "epoch": 53.20586360739325, "grad_norm": 0.2626326382160187, "learning_rate": 8.641861143458045e-05, "loss": 0.0234, "step": 83480 }, { "epoch": 53.21223709369025, "grad_norm": 0.7630456686019897, "learning_rate": 8.641483478938954e-05, "loss": 0.0207, "step": 83490 }, { "epoch": 53.218610579987256, "grad_norm": 0.5011307001113892, "learning_rate": 8.6411057701724e-05, "loss": 0.017, "step": 83500 }, { "epoch": 53.22498406628426, "grad_norm": 0.3697773814201355, "learning_rate": 8.640728017162974e-05, "loss": 0.0144, "step": 83510 }, { "epoch": 53.231357552581265, "grad_norm": 0.30675068497657776, "learning_rate": 8.640350219915266e-05, "loss": 0.0203, "step": 83520 }, { "epoch": 53.237731038878266, "grad_norm": 0.5024585723876953, "learning_rate": 8.639972378433863e-05, "loss": 0.0229, "step": 83530 }, { "epoch": 53.244104525175274, "grad_norm": 0.5399240255355835, "learning_rate": 8.63959449272336e-05, "loss": 0.0234, "step": 83540 }, { "epoch": 53.250478011472275, "grad_norm": 0.5785936713218689, "learning_rate": 8.639216562788349e-05, "loss": 0.0158, "step": 83550 }, { "epoch": 53.25685149776928, "grad_norm": 0.6639803051948547, "learning_rate": 8.638838588633419e-05, "loss": 0.0133, "step": 83560 }, { "epoch": 53.263224984066284, "grad_norm": 0.6767406463623047, "learning_rate": 8.638460570263165e-05, "loss": 0.0172, "step": 83570 }, { "epoch": 53.26959847036329, "grad_norm": 0.33116331696510315, "learning_rate": 8.63808250768218e-05, "loss": 0.0256, "step": 83580 }, { "epoch": 53.27597195666029, "grad_norm": 0.24956843256950378, "learning_rate": 8.637704400895058e-05, "loss": 0.0253, "step": 83590 }, { "epoch": 53.2823454429573, "grad_norm": 0.3254590928554535, "learning_rate": 8.637326249906392e-05, "loss": 0.0218, "step": 83600 }, { "epoch": 53.2887189292543, "grad_norm": 0.30613407492637634, "learning_rate": 8.636948054720779e-05, "loss": 0.0188, "step": 83610 }, { "epoch": 53.29509241555131, "grad_norm": 0.37428730726242065, "learning_rate": 8.636569815342813e-05, "loss": 0.0213, "step": 83620 }, { "epoch": 53.30146590184831, "grad_norm": 0.24845999479293823, "learning_rate": 8.636191531777089e-05, "loss": 0.021, "step": 83630 }, { "epoch": 53.30783938814532, "grad_norm": 0.7117323279380798, "learning_rate": 8.635813204028207e-05, "loss": 0.0294, "step": 83640 }, { "epoch": 53.31421287444232, "grad_norm": 0.3941926062107086, "learning_rate": 8.63543483210076e-05, "loss": 0.0178, "step": 83650 }, { "epoch": 53.32058636073933, "grad_norm": 0.27202844619750977, "learning_rate": 8.635056415999347e-05, "loss": 0.0212, "step": 83660 }, { "epoch": 53.32695984703633, "grad_norm": 0.33432695269584656, "learning_rate": 8.634677955728567e-05, "loss": 0.0221, "step": 83670 }, { "epoch": 53.333333333333336, "grad_norm": 0.40294480323791504, "learning_rate": 8.63429945129302e-05, "loss": 0.0233, "step": 83680 }, { "epoch": 53.33970681963034, "grad_norm": 0.5967270135879517, "learning_rate": 8.6339209026973e-05, "loss": 0.0293, "step": 83690 }, { "epoch": 53.346080305927345, "grad_norm": 0.4534532129764557, "learning_rate": 8.633542309946014e-05, "loss": 0.0272, "step": 83700 }, { "epoch": 53.352453792224345, "grad_norm": 0.6571350693702698, "learning_rate": 8.633163673043756e-05, "loss": 0.0269, "step": 83710 }, { "epoch": 53.35882727852135, "grad_norm": 0.5004278421401978, "learning_rate": 8.632784991995129e-05, "loss": 0.0206, "step": 83720 }, { "epoch": 53.365200764818354, "grad_norm": 0.2790112793445587, "learning_rate": 8.632406266804733e-05, "loss": 0.0131, "step": 83730 }, { "epoch": 53.37157425111536, "grad_norm": 0.4587191343307495, "learning_rate": 8.632027497477174e-05, "loss": 0.031, "step": 83740 }, { "epoch": 53.37794773741236, "grad_norm": 0.27395695447921753, "learning_rate": 8.63164868401705e-05, "loss": 0.0195, "step": 83750 }, { "epoch": 53.38432122370937, "grad_norm": 0.2730388641357422, "learning_rate": 8.631269826428967e-05, "loss": 0.0188, "step": 83760 }, { "epoch": 53.39069471000637, "grad_norm": 0.5676767230033875, "learning_rate": 8.630890924717527e-05, "loss": 0.0278, "step": 83770 }, { "epoch": 53.39706819630338, "grad_norm": 0.4027407765388489, "learning_rate": 8.630511978887334e-05, "loss": 0.0196, "step": 83780 }, { "epoch": 53.40344168260038, "grad_norm": 0.42916440963745117, "learning_rate": 8.630132988942991e-05, "loss": 0.0216, "step": 83790 }, { "epoch": 53.40981516889739, "grad_norm": 0.45874109864234924, "learning_rate": 8.629753954889107e-05, "loss": 0.0181, "step": 83800 }, { "epoch": 53.41618865519439, "grad_norm": 0.6286768317222595, "learning_rate": 8.629374876730284e-05, "loss": 0.0235, "step": 83810 }, { "epoch": 53.4225621414914, "grad_norm": 0.35537102818489075, "learning_rate": 8.62899575447113e-05, "loss": 0.0232, "step": 83820 }, { "epoch": 53.4289356277884, "grad_norm": 0.3062276840209961, "learning_rate": 8.628616588116252e-05, "loss": 0.0226, "step": 83830 }, { "epoch": 53.435309114085406, "grad_norm": 0.5648360252380371, "learning_rate": 8.628237377670257e-05, "loss": 0.032, "step": 83840 }, { "epoch": 53.44168260038241, "grad_norm": 0.45258596539497375, "learning_rate": 8.627858123137748e-05, "loss": 0.0212, "step": 83850 }, { "epoch": 53.448056086679415, "grad_norm": 0.63763427734375, "learning_rate": 8.627478824523342e-05, "loss": 0.0227, "step": 83860 }, { "epoch": 53.454429572976416, "grad_norm": 0.5500563383102417, "learning_rate": 8.627099481831641e-05, "loss": 0.0252, "step": 83870 }, { "epoch": 53.460803059273424, "grad_norm": 0.35673612356185913, "learning_rate": 8.626720095067257e-05, "loss": 0.0328, "step": 83880 }, { "epoch": 53.467176545570425, "grad_norm": 0.5232762694358826, "learning_rate": 8.626340664234802e-05, "loss": 0.027, "step": 83890 }, { "epoch": 53.47355003186743, "grad_norm": 0.33171963691711426, "learning_rate": 8.625961189338882e-05, "loss": 0.0215, "step": 83900 }, { "epoch": 53.47992351816443, "grad_norm": 0.5208948254585266, "learning_rate": 8.62558167038411e-05, "loss": 0.0346, "step": 83910 }, { "epoch": 53.48629700446144, "grad_norm": 0.844214141368866, "learning_rate": 8.6252021073751e-05, "loss": 0.0213, "step": 83920 }, { "epoch": 53.49267049075844, "grad_norm": 0.44236111640930176, "learning_rate": 8.624822500316458e-05, "loss": 0.029, "step": 83930 }, { "epoch": 53.49904397705545, "grad_norm": 0.4460776746273041, "learning_rate": 8.624442849212802e-05, "loss": 0.0203, "step": 83940 }, { "epoch": 53.50541746335245, "grad_norm": 0.329765647649765, "learning_rate": 8.624063154068744e-05, "loss": 0.0254, "step": 83950 }, { "epoch": 53.51179094964946, "grad_norm": 0.39859193563461304, "learning_rate": 8.623683414888895e-05, "loss": 0.0312, "step": 83960 }, { "epoch": 53.51816443594646, "grad_norm": 0.47122275829315186, "learning_rate": 8.623303631677873e-05, "loss": 0.0222, "step": 83970 }, { "epoch": 53.52453792224347, "grad_norm": 0.4777257442474365, "learning_rate": 8.62292380444029e-05, "loss": 0.0205, "step": 83980 }, { "epoch": 53.53091140854047, "grad_norm": 0.6008886098861694, "learning_rate": 8.622543933180762e-05, "loss": 0.0156, "step": 83990 }, { "epoch": 53.53728489483748, "grad_norm": 1.3685667514801025, "learning_rate": 8.622164017903907e-05, "loss": 0.043, "step": 84000 }, { "epoch": 53.54365838113448, "grad_norm": 0.4811289310455322, "learning_rate": 8.621784058614337e-05, "loss": 0.0302, "step": 84010 }, { "epoch": 53.550031867431485, "grad_norm": 0.4748082160949707, "learning_rate": 8.621404055316673e-05, "loss": 0.0208, "step": 84020 }, { "epoch": 53.556405353728486, "grad_norm": 0.29795607924461365, "learning_rate": 8.621024008015528e-05, "loss": 0.0165, "step": 84030 }, { "epoch": 53.562778840025494, "grad_norm": 0.43697237968444824, "learning_rate": 8.620643916715525e-05, "loss": 0.0212, "step": 84040 }, { "epoch": 53.569152326322495, "grad_norm": 0.40295693278312683, "learning_rate": 8.620263781421278e-05, "loss": 0.0147, "step": 84050 }, { "epoch": 53.5755258126195, "grad_norm": 0.6573575139045715, "learning_rate": 8.619883602137408e-05, "loss": 0.0276, "step": 84060 }, { "epoch": 53.581899298916504, "grad_norm": 0.2965167164802551, "learning_rate": 8.619503378868536e-05, "loss": 0.018, "step": 84070 }, { "epoch": 53.58827278521351, "grad_norm": 0.3197731375694275, "learning_rate": 8.619123111619281e-05, "loss": 0.0192, "step": 84080 }, { "epoch": 53.59464627151051, "grad_norm": 0.41930368542671204, "learning_rate": 8.618742800394261e-05, "loss": 0.0186, "step": 84090 }, { "epoch": 53.60101975780752, "grad_norm": 0.338462233543396, "learning_rate": 8.618362445198101e-05, "loss": 0.0188, "step": 84100 }, { "epoch": 53.60739324410453, "grad_norm": 0.3289066553115845, "learning_rate": 8.61798204603542e-05, "loss": 0.031, "step": 84110 }, { "epoch": 53.61376673040153, "grad_norm": 0.4314407706260681, "learning_rate": 8.617601602910842e-05, "loss": 0.0214, "step": 84120 }, { "epoch": 53.62014021669854, "grad_norm": 0.3921735882759094, "learning_rate": 8.617221115828988e-05, "loss": 0.0149, "step": 84130 }, { "epoch": 53.62651370299554, "grad_norm": 0.3524188995361328, "learning_rate": 8.616840584794483e-05, "loss": 0.0119, "step": 84140 }, { "epoch": 53.632887189292546, "grad_norm": 0.8854230642318726, "learning_rate": 8.61646000981195e-05, "loss": 0.0212, "step": 84150 }, { "epoch": 53.63926067558955, "grad_norm": 0.2818962037563324, "learning_rate": 8.616079390886014e-05, "loss": 0.0161, "step": 84160 }, { "epoch": 53.645634161886555, "grad_norm": 0.34340497851371765, "learning_rate": 8.615698728021298e-05, "loss": 0.0183, "step": 84170 }, { "epoch": 53.652007648183556, "grad_norm": 0.5003302097320557, "learning_rate": 8.61531802122243e-05, "loss": 0.0199, "step": 84180 }, { "epoch": 53.658381134480564, "grad_norm": 0.5449504852294922, "learning_rate": 8.614937270494032e-05, "loss": 0.0235, "step": 84190 }, { "epoch": 53.664754620777565, "grad_norm": 0.48928606510162354, "learning_rate": 8.614556475840735e-05, "loss": 0.0232, "step": 84200 }, { "epoch": 53.67112810707457, "grad_norm": 0.38711628317832947, "learning_rate": 8.614175637267165e-05, "loss": 0.02, "step": 84210 }, { "epoch": 53.677501593371574, "grad_norm": 0.35434842109680176, "learning_rate": 8.613794754777946e-05, "loss": 0.0248, "step": 84220 }, { "epoch": 53.68387507966858, "grad_norm": 0.4495094418525696, "learning_rate": 8.613413828377711e-05, "loss": 0.026, "step": 84230 }, { "epoch": 53.69024856596558, "grad_norm": 0.3084827661514282, "learning_rate": 8.613032858071086e-05, "loss": 0.0251, "step": 84240 }, { "epoch": 53.69662205226259, "grad_norm": 0.35163170099258423, "learning_rate": 8.6126518438627e-05, "loss": 0.0317, "step": 84250 }, { "epoch": 53.70299553855959, "grad_norm": 0.41589632630348206, "learning_rate": 8.612270785757183e-05, "loss": 0.0276, "step": 84260 }, { "epoch": 53.7093690248566, "grad_norm": 0.3462205231189728, "learning_rate": 8.611889683759166e-05, "loss": 0.0184, "step": 84270 }, { "epoch": 53.7157425111536, "grad_norm": 0.34634271264076233, "learning_rate": 8.61150853787328e-05, "loss": 0.0199, "step": 84280 }, { "epoch": 53.72211599745061, "grad_norm": 0.2034803032875061, "learning_rate": 8.611127348104153e-05, "loss": 0.0162, "step": 84290 }, { "epoch": 53.72848948374761, "grad_norm": 0.40219178795814514, "learning_rate": 8.61074611445642e-05, "loss": 0.0199, "step": 84300 }, { "epoch": 53.73486297004462, "grad_norm": 0.28426632285118103, "learning_rate": 8.610364836934714e-05, "loss": 0.0283, "step": 84310 }, { "epoch": 53.74123645634162, "grad_norm": 0.3836933374404907, "learning_rate": 8.609983515543667e-05, "loss": 0.0345, "step": 84320 }, { "epoch": 53.747609942638626, "grad_norm": 0.4653191864490509, "learning_rate": 8.60960215028791e-05, "loss": 0.0331, "step": 84330 }, { "epoch": 53.753983428935626, "grad_norm": 0.4005078375339508, "learning_rate": 8.609220741172079e-05, "loss": 0.0195, "step": 84340 }, { "epoch": 53.760356915232634, "grad_norm": 0.5394477248191833, "learning_rate": 8.60883928820081e-05, "loss": 0.0255, "step": 84350 }, { "epoch": 53.766730401529635, "grad_norm": 0.29108813405036926, "learning_rate": 8.608457791378734e-05, "loss": 0.0222, "step": 84360 }, { "epoch": 53.77310388782664, "grad_norm": 0.7882189154624939, "learning_rate": 8.60807625071049e-05, "loss": 0.0187, "step": 84370 }, { "epoch": 53.779477374123644, "grad_norm": 0.6536107659339905, "learning_rate": 8.607694666200715e-05, "loss": 0.0196, "step": 84380 }, { "epoch": 53.78585086042065, "grad_norm": 0.39699968695640564, "learning_rate": 8.60731303785404e-05, "loss": 0.0323, "step": 84390 }, { "epoch": 53.79222434671765, "grad_norm": 0.376953125, "learning_rate": 8.606931365675108e-05, "loss": 0.0197, "step": 84400 }, { "epoch": 53.79859783301466, "grad_norm": 0.4711611270904541, "learning_rate": 8.606549649668553e-05, "loss": 0.0273, "step": 84410 }, { "epoch": 53.80497131931166, "grad_norm": 0.5453488230705261, "learning_rate": 8.606167889839016e-05, "loss": 0.0309, "step": 84420 }, { "epoch": 53.81134480560867, "grad_norm": 0.7160468101501465, "learning_rate": 8.605786086191132e-05, "loss": 0.0287, "step": 84430 }, { "epoch": 53.81771829190567, "grad_norm": 0.9054581522941589, "learning_rate": 8.605404238729544e-05, "loss": 0.0195, "step": 84440 }, { "epoch": 53.82409177820268, "grad_norm": 0.5454513430595398, "learning_rate": 8.605022347458891e-05, "loss": 0.0229, "step": 84450 }, { "epoch": 53.83046526449968, "grad_norm": 0.4035387337207794, "learning_rate": 8.604640412383814e-05, "loss": 0.0178, "step": 84460 }, { "epoch": 53.83683875079669, "grad_norm": 0.354356586933136, "learning_rate": 8.60425843350895e-05, "loss": 0.0209, "step": 84470 }, { "epoch": 53.84321223709369, "grad_norm": 0.616362988948822, "learning_rate": 8.603876410838944e-05, "loss": 0.0241, "step": 84480 }, { "epoch": 53.849585723390696, "grad_norm": 0.4769623875617981, "learning_rate": 8.603494344378438e-05, "loss": 0.0187, "step": 84490 }, { "epoch": 53.8559592096877, "grad_norm": 0.3293713629245758, "learning_rate": 8.603112234132073e-05, "loss": 0.0165, "step": 84500 }, { "epoch": 53.862332695984705, "grad_norm": 0.5047313570976257, "learning_rate": 8.60273008010449e-05, "loss": 0.0168, "step": 84510 }, { "epoch": 53.868706182281706, "grad_norm": 0.6762397885322571, "learning_rate": 8.60234788230034e-05, "loss": 0.0189, "step": 84520 }, { "epoch": 53.875079668578714, "grad_norm": 0.7818517088890076, "learning_rate": 8.601965640724257e-05, "loss": 0.0218, "step": 84530 }, { "epoch": 53.881453154875715, "grad_norm": 0.4257866144180298, "learning_rate": 8.601583355380894e-05, "loss": 0.0208, "step": 84540 }, { "epoch": 53.88782664117272, "grad_norm": 0.3315708339214325, "learning_rate": 8.60120102627489e-05, "loss": 0.0291, "step": 84550 }, { "epoch": 53.89420012746972, "grad_norm": 0.3500559329986572, "learning_rate": 8.600818653410895e-05, "loss": 0.0219, "step": 84560 }, { "epoch": 53.90057361376673, "grad_norm": 0.48855504393577576, "learning_rate": 8.600436236793554e-05, "loss": 0.0251, "step": 84570 }, { "epoch": 53.90694710006373, "grad_norm": 0.4539107382297516, "learning_rate": 8.600053776427513e-05, "loss": 0.0167, "step": 84580 }, { "epoch": 53.91332058636074, "grad_norm": 0.2650492787361145, "learning_rate": 8.59967127231742e-05, "loss": 0.0211, "step": 84590 }, { "epoch": 53.91969407265774, "grad_norm": 0.2661218047142029, "learning_rate": 8.599288724467921e-05, "loss": 0.0175, "step": 84600 }, { "epoch": 53.92606755895475, "grad_norm": 0.37709036469459534, "learning_rate": 8.598906132883667e-05, "loss": 0.0246, "step": 84610 }, { "epoch": 53.93244104525175, "grad_norm": 0.4451025724411011, "learning_rate": 8.598523497569305e-05, "loss": 0.0295, "step": 84620 }, { "epoch": 53.93881453154876, "grad_norm": 0.5607625842094421, "learning_rate": 8.598140818529485e-05, "loss": 0.0244, "step": 84630 }, { "epoch": 53.94518801784576, "grad_norm": 0.6989381909370422, "learning_rate": 8.597758095768858e-05, "loss": 0.0224, "step": 84640 }, { "epoch": 53.95156150414277, "grad_norm": 0.3150734603404999, "learning_rate": 8.597375329292073e-05, "loss": 0.016, "step": 84650 }, { "epoch": 53.95793499043977, "grad_norm": 0.36495548486709595, "learning_rate": 8.59699251910378e-05, "loss": 0.0155, "step": 84660 }, { "epoch": 53.964308476736775, "grad_norm": 0.5768551826477051, "learning_rate": 8.596609665208631e-05, "loss": 0.0254, "step": 84670 }, { "epoch": 53.970681963033776, "grad_norm": 0.5511453747749329, "learning_rate": 8.59622676761128e-05, "loss": 0.0207, "step": 84680 }, { "epoch": 53.977055449330784, "grad_norm": 0.4071067273616791, "learning_rate": 8.595843826316379e-05, "loss": 0.0262, "step": 84690 }, { "epoch": 53.983428935627785, "grad_norm": 0.6202547550201416, "learning_rate": 8.59546084132858e-05, "loss": 0.0216, "step": 84700 }, { "epoch": 53.98980242192479, "grad_norm": 0.32500600814819336, "learning_rate": 8.595077812652536e-05, "loss": 0.0195, "step": 84710 }, { "epoch": 53.9961759082218, "grad_norm": 0.2570093870162964, "learning_rate": 8.594694740292905e-05, "loss": 0.0178, "step": 84720 }, { "epoch": 54.0025493945188, "grad_norm": 0.5094832181930542, "learning_rate": 8.594311624254337e-05, "loss": 0.0242, "step": 84730 }, { "epoch": 54.00892288081581, "grad_norm": 0.18848976492881775, "learning_rate": 8.59392846454149e-05, "loss": 0.0191, "step": 84740 }, { "epoch": 54.01529636711281, "grad_norm": 0.3133816719055176, "learning_rate": 8.593545261159017e-05, "loss": 0.0123, "step": 84750 }, { "epoch": 54.02166985340982, "grad_norm": 0.23563435673713684, "learning_rate": 8.593162014111577e-05, "loss": 0.0147, "step": 84760 }, { "epoch": 54.02804333970682, "grad_norm": 0.43539753556251526, "learning_rate": 8.592778723403827e-05, "loss": 0.0167, "step": 84770 }, { "epoch": 54.03441682600383, "grad_norm": 0.4799574911594391, "learning_rate": 8.592395389040424e-05, "loss": 0.0241, "step": 84780 }, { "epoch": 54.04079031230083, "grad_norm": 0.2808975875377655, "learning_rate": 8.592012011026023e-05, "loss": 0.0196, "step": 84790 }, { "epoch": 54.047163798597836, "grad_norm": 0.5799266695976257, "learning_rate": 8.591628589365286e-05, "loss": 0.0274, "step": 84800 }, { "epoch": 54.05353728489484, "grad_norm": 0.39807814359664917, "learning_rate": 8.59124512406287e-05, "loss": 0.011, "step": 84810 }, { "epoch": 54.059910771191845, "grad_norm": 0.32962679862976074, "learning_rate": 8.590861615123435e-05, "loss": 0.0307, "step": 84820 }, { "epoch": 54.066284257488846, "grad_norm": 0.31680044531822205, "learning_rate": 8.590478062551643e-05, "loss": 0.0189, "step": 84830 }, { "epoch": 54.072657743785854, "grad_norm": 0.3967878818511963, "learning_rate": 8.59009446635215e-05, "loss": 0.0212, "step": 84840 }, { "epoch": 54.079031230082855, "grad_norm": 0.37752580642700195, "learning_rate": 8.589710826529622e-05, "loss": 0.0207, "step": 84850 }, { "epoch": 54.08540471637986, "grad_norm": 0.5211849808692932, "learning_rate": 8.589327143088715e-05, "loss": 0.0188, "step": 84860 }, { "epoch": 54.09177820267686, "grad_norm": 0.44455671310424805, "learning_rate": 8.588943416034097e-05, "loss": 0.017, "step": 84870 }, { "epoch": 54.09815168897387, "grad_norm": 0.26508474349975586, "learning_rate": 8.588559645370428e-05, "loss": 0.0291, "step": 84880 }, { "epoch": 54.10452517527087, "grad_norm": 0.521621584892273, "learning_rate": 8.58817583110237e-05, "loss": 0.0241, "step": 84890 }, { "epoch": 54.11089866156788, "grad_norm": 0.6650809645652771, "learning_rate": 8.587791973234588e-05, "loss": 0.0171, "step": 84900 }, { "epoch": 54.11727214786488, "grad_norm": 0.44355618953704834, "learning_rate": 8.587408071771745e-05, "loss": 0.0194, "step": 84910 }, { "epoch": 54.12364563416189, "grad_norm": 0.4106016755104065, "learning_rate": 8.58702412671851e-05, "loss": 0.0251, "step": 84920 }, { "epoch": 54.13001912045889, "grad_norm": 0.26764363050460815, "learning_rate": 8.586640138079543e-05, "loss": 0.0306, "step": 84930 }, { "epoch": 54.1363926067559, "grad_norm": 0.3683658242225647, "learning_rate": 8.586256105859512e-05, "loss": 0.0263, "step": 84940 }, { "epoch": 54.1427660930529, "grad_norm": 0.35346463322639465, "learning_rate": 8.585872030063082e-05, "loss": 0.033, "step": 84950 }, { "epoch": 54.14913957934991, "grad_norm": 0.42187610268592834, "learning_rate": 8.585487910694925e-05, "loss": 0.0198, "step": 84960 }, { "epoch": 54.15551306564691, "grad_norm": 0.4995281994342804, "learning_rate": 8.585103747759702e-05, "loss": 0.0194, "step": 84970 }, { "epoch": 54.161886551943915, "grad_norm": 0.4050264358520508, "learning_rate": 8.584719541262082e-05, "loss": 0.0201, "step": 84980 }, { "epoch": 54.168260038240916, "grad_norm": 0.44719401001930237, "learning_rate": 8.584335291206739e-05, "loss": 0.0285, "step": 84990 }, { "epoch": 54.174633524537924, "grad_norm": 0.4625746011734009, "learning_rate": 8.583950997598336e-05, "loss": 0.0127, "step": 85000 }, { "epoch": 54.181007010834925, "grad_norm": 0.4877275228500366, "learning_rate": 8.583566660441545e-05, "loss": 0.0198, "step": 85010 }, { "epoch": 54.18738049713193, "grad_norm": 0.5660766959190369, "learning_rate": 8.583182279741035e-05, "loss": 0.0222, "step": 85020 }, { "epoch": 54.193753983428934, "grad_norm": 0.25721949338912964, "learning_rate": 8.582797855501476e-05, "loss": 0.0154, "step": 85030 }, { "epoch": 54.20012746972594, "grad_norm": 0.6928658485412598, "learning_rate": 8.582413387727542e-05, "loss": 0.0262, "step": 85040 }, { "epoch": 54.20650095602294, "grad_norm": 0.4689923822879791, "learning_rate": 8.582028876423904e-05, "loss": 0.0279, "step": 85050 }, { "epoch": 54.21287444231995, "grad_norm": 0.32293230295181274, "learning_rate": 8.581644321595231e-05, "loss": 0.0186, "step": 85060 }, { "epoch": 54.21924792861695, "grad_norm": 0.45877373218536377, "learning_rate": 8.5812597232462e-05, "loss": 0.0227, "step": 85070 }, { "epoch": 54.22562141491396, "grad_norm": 0.4153297245502472, "learning_rate": 8.580875081381481e-05, "loss": 0.032, "step": 85080 }, { "epoch": 54.23199490121096, "grad_norm": 0.3995732069015503, "learning_rate": 8.580490396005746e-05, "loss": 0.0232, "step": 85090 }, { "epoch": 54.23836838750797, "grad_norm": 0.351686954498291, "learning_rate": 8.580105667123677e-05, "loss": 0.0287, "step": 85100 }, { "epoch": 54.24474187380497, "grad_norm": 0.37945473194122314, "learning_rate": 8.579720894739943e-05, "loss": 0.0171, "step": 85110 }, { "epoch": 54.25111536010198, "grad_norm": 0.4682920277118683, "learning_rate": 8.579336078859217e-05, "loss": 0.0278, "step": 85120 }, { "epoch": 54.25748884639898, "grad_norm": 0.4148190915584564, "learning_rate": 8.578951219486182e-05, "loss": 0.0238, "step": 85130 }, { "epoch": 54.263862332695986, "grad_norm": 0.31452199816703796, "learning_rate": 8.578566316625507e-05, "loss": 0.0135, "step": 85140 }, { "epoch": 54.27023581899299, "grad_norm": 0.5568601489067078, "learning_rate": 8.578181370281875e-05, "loss": 0.0225, "step": 85150 }, { "epoch": 54.276609305289995, "grad_norm": 0.5864519476890564, "learning_rate": 8.577796380459961e-05, "loss": 0.0177, "step": 85160 }, { "epoch": 54.282982791586996, "grad_norm": 0.5937750339508057, "learning_rate": 8.577411347164442e-05, "loss": 0.0225, "step": 85170 }, { "epoch": 54.289356277884, "grad_norm": 0.3964175581932068, "learning_rate": 8.577026270399997e-05, "loss": 0.0291, "step": 85180 }, { "epoch": 54.295729764181004, "grad_norm": 0.5911518931388855, "learning_rate": 8.576641150171307e-05, "loss": 0.0238, "step": 85190 }, { "epoch": 54.30210325047801, "grad_norm": 0.8719543218612671, "learning_rate": 8.57625598648305e-05, "loss": 0.0305, "step": 85200 }, { "epoch": 54.30847673677501, "grad_norm": 0.2648394703865051, "learning_rate": 8.575870779339905e-05, "loss": 0.0303, "step": 85210 }, { "epoch": 54.31485022307202, "grad_norm": 0.5186700820922852, "learning_rate": 8.575485528746556e-05, "loss": 0.0251, "step": 85220 }, { "epoch": 54.32122370936902, "grad_norm": 0.47704845666885376, "learning_rate": 8.57510023470768e-05, "loss": 0.0173, "step": 85230 }, { "epoch": 54.32759719566603, "grad_norm": 0.3839057683944702, "learning_rate": 8.574714897227963e-05, "loss": 0.0118, "step": 85240 }, { "epoch": 54.33397068196303, "grad_norm": 0.70790696144104, "learning_rate": 8.574329516312081e-05, "loss": 0.0207, "step": 85250 }, { "epoch": 54.34034416826004, "grad_norm": 0.33637383580207825, "learning_rate": 8.573944091964723e-05, "loss": 0.0225, "step": 85260 }, { "epoch": 54.34671765455704, "grad_norm": 0.404680460691452, "learning_rate": 8.573558624190569e-05, "loss": 0.0187, "step": 85270 }, { "epoch": 54.35309114085405, "grad_norm": 0.473209410905838, "learning_rate": 8.573173112994306e-05, "loss": 0.0253, "step": 85280 }, { "epoch": 54.35946462715105, "grad_norm": 0.44637513160705566, "learning_rate": 8.572787558380614e-05, "loss": 0.024, "step": 85290 }, { "epoch": 54.365838113448056, "grad_norm": 0.890125572681427, "learning_rate": 8.572401960354182e-05, "loss": 0.0283, "step": 85300 }, { "epoch": 54.37221159974506, "grad_norm": 0.4701524078845978, "learning_rate": 8.572016318919691e-05, "loss": 0.0204, "step": 85310 }, { "epoch": 54.378585086042065, "grad_norm": 0.2882644236087799, "learning_rate": 8.57163063408183e-05, "loss": 0.0162, "step": 85320 }, { "epoch": 54.384958572339066, "grad_norm": 0.38176703453063965, "learning_rate": 8.571244905845284e-05, "loss": 0.0127, "step": 85330 }, { "epoch": 54.391332058636074, "grad_norm": 0.4173128604888916, "learning_rate": 8.570859134214741e-05, "loss": 0.0225, "step": 85340 }, { "epoch": 54.397705544933075, "grad_norm": 0.7550672888755798, "learning_rate": 8.570473319194889e-05, "loss": 0.0233, "step": 85350 }, { "epoch": 54.40407903123008, "grad_norm": 0.30151137709617615, "learning_rate": 8.570087460790415e-05, "loss": 0.0189, "step": 85360 }, { "epoch": 54.41045251752709, "grad_norm": 0.77811199426651, "learning_rate": 8.569701559006008e-05, "loss": 0.0261, "step": 85370 }, { "epoch": 54.41682600382409, "grad_norm": 0.6901721954345703, "learning_rate": 8.569315613846354e-05, "loss": 0.0169, "step": 85380 }, { "epoch": 54.4231994901211, "grad_norm": 0.3597675859928131, "learning_rate": 8.568929625316146e-05, "loss": 0.0179, "step": 85390 }, { "epoch": 54.4295729764181, "grad_norm": 0.4583025872707367, "learning_rate": 8.568543593420076e-05, "loss": 0.0216, "step": 85400 }, { "epoch": 54.43594646271511, "grad_norm": 0.41436609625816345, "learning_rate": 8.568157518162829e-05, "loss": 0.024, "step": 85410 }, { "epoch": 54.44231994901211, "grad_norm": 0.2920903265476227, "learning_rate": 8.5677713995491e-05, "loss": 0.011, "step": 85420 }, { "epoch": 54.44869343530912, "grad_norm": 0.36646533012390137, "learning_rate": 8.567385237583581e-05, "loss": 0.0246, "step": 85430 }, { "epoch": 54.45506692160612, "grad_norm": 0.5572330355644226, "learning_rate": 8.566999032270964e-05, "loss": 0.0327, "step": 85440 }, { "epoch": 54.461440407903126, "grad_norm": 0.6074442267417908, "learning_rate": 8.56661278361594e-05, "loss": 0.0312, "step": 85450 }, { "epoch": 54.46781389420013, "grad_norm": 0.5684938430786133, "learning_rate": 8.566226491623203e-05, "loss": 0.0181, "step": 85460 }, { "epoch": 54.474187380497135, "grad_norm": 0.3168732523918152, "learning_rate": 8.565840156297447e-05, "loss": 0.0245, "step": 85470 }, { "epoch": 54.480560866794136, "grad_norm": 0.5349665284156799, "learning_rate": 8.565453777643366e-05, "loss": 0.0238, "step": 85480 }, { "epoch": 54.486934353091144, "grad_norm": 0.3802516758441925, "learning_rate": 8.565067355665656e-05, "loss": 0.0197, "step": 85490 }, { "epoch": 54.493307839388144, "grad_norm": 0.3255811333656311, "learning_rate": 8.564680890369011e-05, "loss": 0.025, "step": 85500 }, { "epoch": 54.49968132568515, "grad_norm": 0.5623638033866882, "learning_rate": 8.564294381758128e-05, "loss": 0.0205, "step": 85510 }, { "epoch": 54.50605481198215, "grad_norm": 0.6858962774276733, "learning_rate": 8.563907829837703e-05, "loss": 0.0241, "step": 85520 }, { "epoch": 54.51242829827916, "grad_norm": 0.46164995431900024, "learning_rate": 8.563521234612433e-05, "loss": 0.019, "step": 85530 }, { "epoch": 54.51880178457616, "grad_norm": 0.36765310168266296, "learning_rate": 8.563134596087015e-05, "loss": 0.027, "step": 85540 }, { "epoch": 54.52517527087317, "grad_norm": 0.35005107522010803, "learning_rate": 8.562747914266149e-05, "loss": 0.0253, "step": 85550 }, { "epoch": 54.53154875717017, "grad_norm": 0.7256959080696106, "learning_rate": 8.56236118915453e-05, "loss": 0.0192, "step": 85560 }, { "epoch": 54.53792224346718, "grad_norm": 0.8625956773757935, "learning_rate": 8.56197442075686e-05, "loss": 0.0157, "step": 85570 }, { "epoch": 54.54429572976418, "grad_norm": 0.33602285385131836, "learning_rate": 8.561587609077837e-05, "loss": 0.0228, "step": 85580 }, { "epoch": 54.55066921606119, "grad_norm": 0.498575896024704, "learning_rate": 8.561200754122163e-05, "loss": 0.0278, "step": 85590 }, { "epoch": 54.55704270235819, "grad_norm": 0.5766869783401489, "learning_rate": 8.560813855894539e-05, "loss": 0.0297, "step": 85600 }, { "epoch": 54.5634161886552, "grad_norm": 0.7085694670677185, "learning_rate": 8.560426914399662e-05, "loss": 0.023, "step": 85610 }, { "epoch": 54.5697896749522, "grad_norm": 0.6201806664466858, "learning_rate": 8.560039929642236e-05, "loss": 0.0206, "step": 85620 }, { "epoch": 54.576163161249205, "grad_norm": 0.42374396324157715, "learning_rate": 8.559652901626966e-05, "loss": 0.0174, "step": 85630 }, { "epoch": 54.582536647546206, "grad_norm": 0.4321466386318207, "learning_rate": 8.559265830358552e-05, "loss": 0.0288, "step": 85640 }, { "epoch": 54.588910133843214, "grad_norm": 0.5083127617835999, "learning_rate": 8.558878715841696e-05, "loss": 0.0166, "step": 85650 }, { "epoch": 54.595283620140215, "grad_norm": 0.5091572999954224, "learning_rate": 8.558491558081104e-05, "loss": 0.0259, "step": 85660 }, { "epoch": 54.60165710643722, "grad_norm": 0.5440375804901123, "learning_rate": 8.55810435708148e-05, "loss": 0.0222, "step": 85670 }, { "epoch": 54.608030592734224, "grad_norm": 0.4063795208930969, "learning_rate": 8.557717112847529e-05, "loss": 0.0306, "step": 85680 }, { "epoch": 54.61440407903123, "grad_norm": 0.4484158456325531, "learning_rate": 8.557329825383956e-05, "loss": 0.0174, "step": 85690 }, { "epoch": 54.62077756532823, "grad_norm": 0.5302240252494812, "learning_rate": 8.556942494695467e-05, "loss": 0.0249, "step": 85700 }, { "epoch": 54.62715105162524, "grad_norm": 0.3472258448600769, "learning_rate": 8.556555120786769e-05, "loss": 0.0191, "step": 85710 }, { "epoch": 54.63352453792224, "grad_norm": 0.35999906063079834, "learning_rate": 8.556167703662566e-05, "loss": 0.0309, "step": 85720 }, { "epoch": 54.63989802421925, "grad_norm": 0.7518576979637146, "learning_rate": 8.555780243327569e-05, "loss": 0.0247, "step": 85730 }, { "epoch": 54.64627151051625, "grad_norm": 0.5930815935134888, "learning_rate": 8.555392739786485e-05, "loss": 0.0272, "step": 85740 }, { "epoch": 54.65264499681326, "grad_norm": 0.3392670452594757, "learning_rate": 8.555005193044023e-05, "loss": 0.0186, "step": 85750 }, { "epoch": 54.65901848311026, "grad_norm": 0.6263693571090698, "learning_rate": 8.55461760310489e-05, "loss": 0.0272, "step": 85760 }, { "epoch": 54.66539196940727, "grad_norm": 0.43870407342910767, "learning_rate": 8.5542299699738e-05, "loss": 0.0306, "step": 85770 }, { "epoch": 54.67176545570427, "grad_norm": 0.4818042516708374, "learning_rate": 8.553842293655459e-05, "loss": 0.0227, "step": 85780 }, { "epoch": 54.678138942001276, "grad_norm": 0.28415727615356445, "learning_rate": 8.553454574154577e-05, "loss": 0.0228, "step": 85790 }, { "epoch": 54.68451242829828, "grad_norm": 1.2254512310028076, "learning_rate": 8.553066811475867e-05, "loss": 0.0328, "step": 85800 }, { "epoch": 54.690885914595285, "grad_norm": 0.5818784832954407, "learning_rate": 8.552679005624041e-05, "loss": 0.0247, "step": 85810 }, { "epoch": 54.697259400892285, "grad_norm": 0.4756927192211151, "learning_rate": 8.55229115660381e-05, "loss": 0.032, "step": 85820 }, { "epoch": 54.70363288718929, "grad_norm": 0.25622543692588806, "learning_rate": 8.55190326441989e-05, "loss": 0.0172, "step": 85830 }, { "epoch": 54.710006373486294, "grad_norm": 0.3521977663040161, "learning_rate": 8.551515329076991e-05, "loss": 0.0225, "step": 85840 }, { "epoch": 54.7163798597833, "grad_norm": 0.5041458010673523, "learning_rate": 8.551127350579827e-05, "loss": 0.0287, "step": 85850 }, { "epoch": 54.7227533460803, "grad_norm": 0.3370957374572754, "learning_rate": 8.550739328933112e-05, "loss": 0.0169, "step": 85860 }, { "epoch": 54.72912683237731, "grad_norm": 0.333385169506073, "learning_rate": 8.550351264141563e-05, "loss": 0.0251, "step": 85870 }, { "epoch": 54.73550031867431, "grad_norm": 0.3394824266433716, "learning_rate": 8.549963156209894e-05, "loss": 0.0168, "step": 85880 }, { "epoch": 54.74187380497132, "grad_norm": 2.013967752456665, "learning_rate": 8.54957500514282e-05, "loss": 0.0321, "step": 85890 }, { "epoch": 54.74824729126832, "grad_norm": 0.4192601442337036, "learning_rate": 8.549186810945059e-05, "loss": 0.0277, "step": 85900 }, { "epoch": 54.75462077756533, "grad_norm": 0.4651796817779541, "learning_rate": 8.548798573621327e-05, "loss": 0.0173, "step": 85910 }, { "epoch": 54.76099426386233, "grad_norm": 0.49102360010147095, "learning_rate": 8.548410293176341e-05, "loss": 0.021, "step": 85920 }, { "epoch": 54.76736775015934, "grad_norm": 0.7323653697967529, "learning_rate": 8.548021969614821e-05, "loss": 0.0323, "step": 85930 }, { "epoch": 54.77374123645634, "grad_norm": 0.4036193788051605, "learning_rate": 8.547633602941484e-05, "loss": 0.0208, "step": 85940 }, { "epoch": 54.780114722753346, "grad_norm": 0.35377243161201477, "learning_rate": 8.547245193161049e-05, "loss": 0.0213, "step": 85950 }, { "epoch": 54.78648820905035, "grad_norm": 0.45204293727874756, "learning_rate": 8.546856740278235e-05, "loss": 0.0193, "step": 85960 }, { "epoch": 54.792861695347355, "grad_norm": 0.5348866581916809, "learning_rate": 8.546468244297765e-05, "loss": 0.0236, "step": 85970 }, { "epoch": 54.79923518164436, "grad_norm": 0.6440789699554443, "learning_rate": 8.546079705224354e-05, "loss": 0.025, "step": 85980 }, { "epoch": 54.805608667941364, "grad_norm": 0.48172369599342346, "learning_rate": 8.54569112306273e-05, "loss": 0.0235, "step": 85990 }, { "epoch": 54.81198215423837, "grad_norm": 0.7067767381668091, "learning_rate": 8.545302497817609e-05, "loss": 0.0351, "step": 86000 }, { "epoch": 54.81835564053537, "grad_norm": 0.4462307095527649, "learning_rate": 8.544913829493718e-05, "loss": 0.0203, "step": 86010 }, { "epoch": 54.82472912683238, "grad_norm": 0.35432422161102295, "learning_rate": 8.544525118095774e-05, "loss": 0.0195, "step": 86020 }, { "epoch": 54.83110261312938, "grad_norm": 0.7455963492393494, "learning_rate": 8.544136363628503e-05, "loss": 0.0232, "step": 86030 }, { "epoch": 54.83747609942639, "grad_norm": 0.28518256545066833, "learning_rate": 8.54374756609663e-05, "loss": 0.0273, "step": 86040 }, { "epoch": 54.84384958572339, "grad_norm": 0.5440710783004761, "learning_rate": 8.54335872550488e-05, "loss": 0.0216, "step": 86050 }, { "epoch": 54.8502230720204, "grad_norm": 0.583558976650238, "learning_rate": 8.542969841857972e-05, "loss": 0.0168, "step": 86060 }, { "epoch": 54.8565965583174, "grad_norm": 0.4299478530883789, "learning_rate": 8.542580915160638e-05, "loss": 0.0275, "step": 86070 }, { "epoch": 54.86297004461441, "grad_norm": 0.26359421014785767, "learning_rate": 8.542191945417601e-05, "loss": 0.0172, "step": 86080 }, { "epoch": 54.86934353091141, "grad_norm": 0.4829738736152649, "learning_rate": 8.541802932633588e-05, "loss": 0.0281, "step": 86090 }, { "epoch": 54.875717017208416, "grad_norm": 0.4335806667804718, "learning_rate": 8.541413876813325e-05, "loss": 0.0191, "step": 86100 }, { "epoch": 54.88209050350542, "grad_norm": 0.43728089332580566, "learning_rate": 8.54102477796154e-05, "loss": 0.0186, "step": 86110 }, { "epoch": 54.888463989802425, "grad_norm": 0.4246975779533386, "learning_rate": 8.540635636082959e-05, "loss": 0.019, "step": 86120 }, { "epoch": 54.894837476099426, "grad_norm": 0.5081120729446411, "learning_rate": 8.540246451182314e-05, "loss": 0.0238, "step": 86130 }, { "epoch": 54.90121096239643, "grad_norm": 0.6616365313529968, "learning_rate": 8.539857223264331e-05, "loss": 0.0173, "step": 86140 }, { "epoch": 54.907584448693434, "grad_norm": 0.31998535990715027, "learning_rate": 8.53946795233374e-05, "loss": 0.0205, "step": 86150 }, { "epoch": 54.91395793499044, "grad_norm": 0.48805829882621765, "learning_rate": 8.539078638395273e-05, "loss": 0.0291, "step": 86160 }, { "epoch": 54.92033142128744, "grad_norm": 0.5109148621559143, "learning_rate": 8.538689281453657e-05, "loss": 0.0163, "step": 86170 }, { "epoch": 54.92670490758445, "grad_norm": 0.3015347421169281, "learning_rate": 8.538299881513628e-05, "loss": 0.0188, "step": 86180 }, { "epoch": 54.93307839388145, "grad_norm": 0.5622946619987488, "learning_rate": 8.537910438579913e-05, "loss": 0.0181, "step": 86190 }, { "epoch": 54.93945188017846, "grad_norm": 0.3560231029987335, "learning_rate": 8.537520952657246e-05, "loss": 0.0139, "step": 86200 }, { "epoch": 54.94582536647546, "grad_norm": 0.39006224274635315, "learning_rate": 8.537131423750359e-05, "loss": 0.0159, "step": 86210 }, { "epoch": 54.95219885277247, "grad_norm": 0.6075239181518555, "learning_rate": 8.536741851863988e-05, "loss": 0.0219, "step": 86220 }, { "epoch": 54.95857233906947, "grad_norm": 0.2855777144432068, "learning_rate": 8.536352237002861e-05, "loss": 0.0185, "step": 86230 }, { "epoch": 54.96494582536648, "grad_norm": 0.5669342875480652, "learning_rate": 8.535962579171717e-05, "loss": 0.0176, "step": 86240 }, { "epoch": 54.97131931166348, "grad_norm": 0.45056989789009094, "learning_rate": 8.535572878375288e-05, "loss": 0.0221, "step": 86250 }, { "epoch": 54.977692797960486, "grad_norm": 0.446866899728775, "learning_rate": 8.535183134618311e-05, "loss": 0.0178, "step": 86260 }, { "epoch": 54.98406628425749, "grad_norm": 0.44455230236053467, "learning_rate": 8.534793347905522e-05, "loss": 0.0212, "step": 86270 }, { "epoch": 54.990439770554495, "grad_norm": 0.45189395546913147, "learning_rate": 8.534403518241657e-05, "loss": 0.0265, "step": 86280 }, { "epoch": 54.996813256851496, "grad_norm": 0.40467384457588196, "learning_rate": 8.534013645631452e-05, "loss": 0.0168, "step": 86290 }, { "epoch": 55.003186743148504, "grad_norm": 0.5539917945861816, "learning_rate": 8.533623730079643e-05, "loss": 0.0155, "step": 86300 }, { "epoch": 55.009560229445505, "grad_norm": 0.5893488526344299, "learning_rate": 8.533233771590971e-05, "loss": 0.0254, "step": 86310 }, { "epoch": 55.01593371574251, "grad_norm": 0.5297513008117676, "learning_rate": 8.532843770170171e-05, "loss": 0.0159, "step": 86320 }, { "epoch": 55.022307202039514, "grad_norm": 0.3981828987598419, "learning_rate": 8.532453725821984e-05, "loss": 0.0155, "step": 86330 }, { "epoch": 55.02868068833652, "grad_norm": 0.4786800742149353, "learning_rate": 8.532063638551151e-05, "loss": 0.0221, "step": 86340 }, { "epoch": 55.03505417463352, "grad_norm": 0.7285570502281189, "learning_rate": 8.531673508362408e-05, "loss": 0.0244, "step": 86350 }, { "epoch": 55.04142766093053, "grad_norm": 0.402475506067276, "learning_rate": 8.531283335260498e-05, "loss": 0.0359, "step": 86360 }, { "epoch": 55.04780114722753, "grad_norm": 0.4064553678035736, "learning_rate": 8.530893119250161e-05, "loss": 0.0245, "step": 86370 }, { "epoch": 55.05417463352454, "grad_norm": 0.6936973929405212, "learning_rate": 8.53050286033614e-05, "loss": 0.0236, "step": 86380 }, { "epoch": 55.06054811982154, "grad_norm": 0.4192628562450409, "learning_rate": 8.530112558523175e-05, "loss": 0.0198, "step": 86390 }, { "epoch": 55.06692160611855, "grad_norm": 0.7908431887626648, "learning_rate": 8.529722213816009e-05, "loss": 0.0224, "step": 86400 }, { "epoch": 55.07329509241555, "grad_norm": 0.6014593839645386, "learning_rate": 8.529331826219387e-05, "loss": 0.0163, "step": 86410 }, { "epoch": 55.07966857871256, "grad_norm": 0.6179099678993225, "learning_rate": 8.52894139573805e-05, "loss": 0.0201, "step": 86420 }, { "epoch": 55.08604206500956, "grad_norm": 0.7082214951515198, "learning_rate": 8.528550922376744e-05, "loss": 0.0367, "step": 86430 }, { "epoch": 55.092415551306566, "grad_norm": 0.5604935884475708, "learning_rate": 8.528160406140211e-05, "loss": 0.0196, "step": 86440 }, { "epoch": 55.09878903760357, "grad_norm": 0.4169777035713196, "learning_rate": 8.5277698470332e-05, "loss": 0.0238, "step": 86450 }, { "epoch": 55.105162523900574, "grad_norm": 0.6768692135810852, "learning_rate": 8.527379245060453e-05, "loss": 0.0348, "step": 86460 }, { "epoch": 55.111536010197575, "grad_norm": 0.5148084759712219, "learning_rate": 8.52698860022672e-05, "loss": 0.0353, "step": 86470 }, { "epoch": 55.11790949649458, "grad_norm": 0.5138073563575745, "learning_rate": 8.526597912536743e-05, "loss": 0.0201, "step": 86480 }, { "epoch": 55.124282982791584, "grad_norm": 0.4618106782436371, "learning_rate": 8.526207181995272e-05, "loss": 0.0147, "step": 86490 }, { "epoch": 55.13065646908859, "grad_norm": 0.3576715588569641, "learning_rate": 8.525816408607055e-05, "loss": 0.0304, "step": 86500 }, { "epoch": 55.13702995538559, "grad_norm": 0.5648069977760315, "learning_rate": 8.52542559237684e-05, "loss": 0.0266, "step": 86510 }, { "epoch": 55.1434034416826, "grad_norm": 0.4206564724445343, "learning_rate": 8.525034733309376e-05, "loss": 0.029, "step": 86520 }, { "epoch": 55.1497769279796, "grad_norm": 0.45659393072128296, "learning_rate": 8.524643831409412e-05, "loss": 0.0161, "step": 86530 }, { "epoch": 55.15615041427661, "grad_norm": 0.39672553539276123, "learning_rate": 8.524252886681697e-05, "loss": 0.0254, "step": 86540 }, { "epoch": 55.16252390057361, "grad_norm": 0.23461787402629852, "learning_rate": 8.52386189913098e-05, "loss": 0.0237, "step": 86550 }, { "epoch": 55.16889738687062, "grad_norm": 0.3028956651687622, "learning_rate": 8.523470868762016e-05, "loss": 0.0228, "step": 86560 }, { "epoch": 55.17527087316762, "grad_norm": 0.6393052935600281, "learning_rate": 8.523079795579554e-05, "loss": 0.0199, "step": 86570 }, { "epoch": 55.18164435946463, "grad_norm": 0.5493573546409607, "learning_rate": 8.522688679588346e-05, "loss": 0.0257, "step": 86580 }, { "epoch": 55.18801784576163, "grad_norm": 0.46611249446868896, "learning_rate": 8.522297520793144e-05, "loss": 0.0147, "step": 86590 }, { "epoch": 55.194391332058636, "grad_norm": 0.4382973611354828, "learning_rate": 8.521906319198703e-05, "loss": 0.0183, "step": 86600 }, { "epoch": 55.20076481835564, "grad_norm": 0.500516951084137, "learning_rate": 8.521515074809772e-05, "loss": 0.0196, "step": 86610 }, { "epoch": 55.207138304652645, "grad_norm": 0.6276620030403137, "learning_rate": 8.521123787631112e-05, "loss": 0.0241, "step": 86620 }, { "epoch": 55.21351179094965, "grad_norm": 0.3739742338657379, "learning_rate": 8.520732457667471e-05, "loss": 0.0204, "step": 86630 }, { "epoch": 55.219885277246654, "grad_norm": 0.6306214928627014, "learning_rate": 8.520341084923607e-05, "loss": 0.0195, "step": 86640 }, { "epoch": 55.22625876354366, "grad_norm": 0.43090885877609253, "learning_rate": 8.519949669404274e-05, "loss": 0.0237, "step": 86650 }, { "epoch": 55.23263224984066, "grad_norm": 0.36384761333465576, "learning_rate": 8.519558211114232e-05, "loss": 0.0208, "step": 86660 }, { "epoch": 55.23900573613767, "grad_norm": 0.34764453768730164, "learning_rate": 8.519166710058231e-05, "loss": 0.013, "step": 86670 }, { "epoch": 55.24537922243467, "grad_norm": 0.4838601052761078, "learning_rate": 8.518775166241034e-05, "loss": 0.0216, "step": 86680 }, { "epoch": 55.25175270873168, "grad_norm": 0.4001941680908203, "learning_rate": 8.518383579667396e-05, "loss": 0.0243, "step": 86690 }, { "epoch": 55.25812619502868, "grad_norm": 0.7490586638450623, "learning_rate": 8.517991950342077e-05, "loss": 0.0282, "step": 86700 }, { "epoch": 55.26449968132569, "grad_norm": 0.5147513747215271, "learning_rate": 8.517600278269833e-05, "loss": 0.0249, "step": 86710 }, { "epoch": 55.27087316762269, "grad_norm": 0.2288786917924881, "learning_rate": 8.517208563455423e-05, "loss": 0.0234, "step": 86720 }, { "epoch": 55.2772466539197, "grad_norm": 0.18899662792682648, "learning_rate": 8.516816805903611e-05, "loss": 0.0198, "step": 86730 }, { "epoch": 55.2836201402167, "grad_norm": 0.3168834447860718, "learning_rate": 8.516425005619153e-05, "loss": 0.0169, "step": 86740 }, { "epoch": 55.289993626513706, "grad_norm": 0.5765627026557922, "learning_rate": 8.516033162606811e-05, "loss": 0.0322, "step": 86750 }, { "epoch": 55.29636711281071, "grad_norm": 0.6220970153808594, "learning_rate": 8.515641276871346e-05, "loss": 0.032, "step": 86760 }, { "epoch": 55.302740599107715, "grad_norm": 0.4723188877105713, "learning_rate": 8.515249348417522e-05, "loss": 0.0218, "step": 86770 }, { "epoch": 55.309114085404715, "grad_norm": 0.3541519045829773, "learning_rate": 8.514857377250098e-05, "loss": 0.015, "step": 86780 }, { "epoch": 55.31548757170172, "grad_norm": 0.4373086988925934, "learning_rate": 8.514465363373839e-05, "loss": 0.023, "step": 86790 }, { "epoch": 55.321861057998724, "grad_norm": 0.4673744738101959, "learning_rate": 8.514073306793506e-05, "loss": 0.0311, "step": 86800 }, { "epoch": 55.32823454429573, "grad_norm": 0.5443738698959351, "learning_rate": 8.513681207513867e-05, "loss": 0.0174, "step": 86810 }, { "epoch": 55.33460803059273, "grad_norm": 0.48015719652175903, "learning_rate": 8.513289065539683e-05, "loss": 0.0302, "step": 86820 }, { "epoch": 55.34098151688974, "grad_norm": 0.2767954170703888, "learning_rate": 8.512896880875718e-05, "loss": 0.0198, "step": 86830 }, { "epoch": 55.34735500318674, "grad_norm": 0.45080387592315674, "learning_rate": 8.51250465352674e-05, "loss": 0.0233, "step": 86840 }, { "epoch": 55.35372848948375, "grad_norm": 0.35147109627723694, "learning_rate": 8.512112383497515e-05, "loss": 0.0152, "step": 86850 }, { "epoch": 55.36010197578075, "grad_norm": 0.4502158463001251, "learning_rate": 8.511720070792808e-05, "loss": 0.0214, "step": 86860 }, { "epoch": 55.36647546207776, "grad_norm": 0.5294784307479858, "learning_rate": 8.511327715417388e-05, "loss": 0.0294, "step": 86870 }, { "epoch": 55.37284894837476, "grad_norm": 0.33427414298057556, "learning_rate": 8.510935317376019e-05, "loss": 0.0267, "step": 86880 }, { "epoch": 55.37922243467177, "grad_norm": 0.21311397850513458, "learning_rate": 8.51054287667347e-05, "loss": 0.0132, "step": 86890 }, { "epoch": 55.38559592096877, "grad_norm": 1.059682846069336, "learning_rate": 8.510150393314513e-05, "loss": 0.0212, "step": 86900 }, { "epoch": 55.391969407265776, "grad_norm": 0.2483416497707367, "learning_rate": 8.509757867303914e-05, "loss": 0.0187, "step": 86910 }, { "epoch": 55.39834289356278, "grad_norm": 0.474150151014328, "learning_rate": 8.509365298646444e-05, "loss": 0.0265, "step": 86920 }, { "epoch": 55.404716379859785, "grad_norm": 0.24896462261676788, "learning_rate": 8.50897268734687e-05, "loss": 0.015, "step": 86930 }, { "epoch": 55.411089866156786, "grad_norm": 0.508091390132904, "learning_rate": 8.508580033409967e-05, "loss": 0.019, "step": 86940 }, { "epoch": 55.417463352453794, "grad_norm": 0.2760290503501892, "learning_rate": 8.508187336840502e-05, "loss": 0.0295, "step": 86950 }, { "epoch": 55.423836838750795, "grad_norm": 0.26464876532554626, "learning_rate": 8.50779459764325e-05, "loss": 0.0167, "step": 86960 }, { "epoch": 55.4302103250478, "grad_norm": 0.4598115384578705, "learning_rate": 8.50740181582298e-05, "loss": 0.019, "step": 86970 }, { "epoch": 55.4365838113448, "grad_norm": 0.28399914503097534, "learning_rate": 8.507008991384469e-05, "loss": 0.0184, "step": 86980 }, { "epoch": 55.44295729764181, "grad_norm": 0.5353851914405823, "learning_rate": 8.506616124332485e-05, "loss": 0.0184, "step": 86990 }, { "epoch": 55.44933078393881, "grad_norm": 0.2473720759153366, "learning_rate": 8.506223214671807e-05, "loss": 0.0285, "step": 87000 }, { "epoch": 55.45570427023582, "grad_norm": 0.2210370898246765, "learning_rate": 8.505830262407204e-05, "loss": 0.0312, "step": 87010 }, { "epoch": 55.46207775653282, "grad_norm": 1.7098239660263062, "learning_rate": 8.505437267543454e-05, "loss": 0.0242, "step": 87020 }, { "epoch": 55.46845124282983, "grad_norm": 0.3326781094074249, "learning_rate": 8.505044230085333e-05, "loss": 0.0321, "step": 87030 }, { "epoch": 55.47482472912683, "grad_norm": 0.5609923005104065, "learning_rate": 8.504651150037614e-05, "loss": 0.0225, "step": 87040 }, { "epoch": 55.48119821542384, "grad_norm": 0.5638129711151123, "learning_rate": 8.504258027405074e-05, "loss": 0.021, "step": 87050 }, { "epoch": 55.48757170172084, "grad_norm": 0.507244884967804, "learning_rate": 8.503864862192491e-05, "loss": 0.0215, "step": 87060 }, { "epoch": 55.49394518801785, "grad_norm": 0.37580406665802, "learning_rate": 8.503471654404643e-05, "loss": 0.0204, "step": 87070 }, { "epoch": 55.50031867431485, "grad_norm": 0.39622536301612854, "learning_rate": 8.503078404046305e-05, "loss": 0.0231, "step": 87080 }, { "epoch": 55.506692160611856, "grad_norm": 0.2698856592178345, "learning_rate": 8.502685111122258e-05, "loss": 0.0192, "step": 87090 }, { "epoch": 55.513065646908856, "grad_norm": 0.43262979388237, "learning_rate": 8.502291775637278e-05, "loss": 0.0126, "step": 87100 }, { "epoch": 55.519439133205864, "grad_norm": 0.2894722819328308, "learning_rate": 8.501898397596148e-05, "loss": 0.0161, "step": 87110 }, { "epoch": 55.525812619502865, "grad_norm": 0.6817557215690613, "learning_rate": 8.501504977003645e-05, "loss": 0.0217, "step": 87120 }, { "epoch": 55.53218610579987, "grad_norm": 0.421288400888443, "learning_rate": 8.501111513864551e-05, "loss": 0.0192, "step": 87130 }, { "epoch": 55.538559592096874, "grad_norm": 0.23434068262577057, "learning_rate": 8.500718008183648e-05, "loss": 0.0251, "step": 87140 }, { "epoch": 55.54493307839388, "grad_norm": 0.49615976214408875, "learning_rate": 8.500324459965714e-05, "loss": 0.0333, "step": 87150 }, { "epoch": 55.55130656469088, "grad_norm": 0.3419979214668274, "learning_rate": 8.499930869215534e-05, "loss": 0.0165, "step": 87160 }, { "epoch": 55.55768005098789, "grad_norm": 0.5330997109413147, "learning_rate": 8.499537235937888e-05, "loss": 0.022, "step": 87170 }, { "epoch": 55.56405353728489, "grad_norm": 0.6238991618156433, "learning_rate": 8.499143560137561e-05, "loss": 0.022, "step": 87180 }, { "epoch": 55.5704270235819, "grad_norm": 0.3130239248275757, "learning_rate": 8.498749841819337e-05, "loss": 0.0173, "step": 87190 }, { "epoch": 55.5768005098789, "grad_norm": 0.5819651484489441, "learning_rate": 8.498356080987999e-05, "loss": 0.0167, "step": 87200 }, { "epoch": 55.58317399617591, "grad_norm": 0.4238913655281067, "learning_rate": 8.49796227764833e-05, "loss": 0.0215, "step": 87210 }, { "epoch": 55.58954748247291, "grad_norm": 0.46811988949775696, "learning_rate": 8.497568431805119e-05, "loss": 0.0206, "step": 87220 }, { "epoch": 55.59592096876992, "grad_norm": 0.5295667052268982, "learning_rate": 8.497174543463147e-05, "loss": 0.014, "step": 87230 }, { "epoch": 55.602294455066925, "grad_norm": 0.22549764811992645, "learning_rate": 8.496780612627205e-05, "loss": 0.0268, "step": 87240 }, { "epoch": 55.608667941363926, "grad_norm": 0.4042595624923706, "learning_rate": 8.496386639302076e-05, "loss": 0.019, "step": 87250 }, { "epoch": 55.615041427660934, "grad_norm": 0.6040352582931519, "learning_rate": 8.495992623492547e-05, "loss": 0.0444, "step": 87260 }, { "epoch": 55.621414913957935, "grad_norm": 0.7099024057388306, "learning_rate": 8.495598565203407e-05, "loss": 0.0234, "step": 87270 }, { "epoch": 55.62778840025494, "grad_norm": 0.35963934659957886, "learning_rate": 8.495204464439446e-05, "loss": 0.0186, "step": 87280 }, { "epoch": 55.634161886551944, "grad_norm": 0.3433648943901062, "learning_rate": 8.494810321205449e-05, "loss": 0.0161, "step": 87290 }, { "epoch": 55.64053537284895, "grad_norm": 0.34662947058677673, "learning_rate": 8.494416135506208e-05, "loss": 0.0255, "step": 87300 }, { "epoch": 55.64690885914595, "grad_norm": 0.6556997299194336, "learning_rate": 8.49402190734651e-05, "loss": 0.0265, "step": 87310 }, { "epoch": 55.65328234544296, "grad_norm": 0.34683218598365784, "learning_rate": 8.493627636731147e-05, "loss": 0.0237, "step": 87320 }, { "epoch": 55.65965583173996, "grad_norm": 0.38685324788093567, "learning_rate": 8.493233323664911e-05, "loss": 0.0257, "step": 87330 }, { "epoch": 55.66602931803697, "grad_norm": 0.3549465835094452, "learning_rate": 8.492838968152592e-05, "loss": 0.028, "step": 87340 }, { "epoch": 55.67240280433397, "grad_norm": 0.29948002099990845, "learning_rate": 8.492444570198982e-05, "loss": 0.0159, "step": 87350 }, { "epoch": 55.67877629063098, "grad_norm": 0.40896064043045044, "learning_rate": 8.492050129808871e-05, "loss": 0.0179, "step": 87360 }, { "epoch": 55.68514977692798, "grad_norm": 0.3966725766658783, "learning_rate": 8.491655646987056e-05, "loss": 0.0142, "step": 87370 }, { "epoch": 55.69152326322499, "grad_norm": 0.34421515464782715, "learning_rate": 8.491261121738328e-05, "loss": 0.0215, "step": 87380 }, { "epoch": 55.69789674952199, "grad_norm": 0.5204769372940063, "learning_rate": 8.49086655406748e-05, "loss": 0.0268, "step": 87390 }, { "epoch": 55.704270235818996, "grad_norm": 0.2608213722705841, "learning_rate": 8.490471943979308e-05, "loss": 0.0224, "step": 87400 }, { "epoch": 55.710643722116, "grad_norm": 0.3870161175727844, "learning_rate": 8.490077291478606e-05, "loss": 0.0213, "step": 87410 }, { "epoch": 55.717017208413004, "grad_norm": 0.488418847322464, "learning_rate": 8.489682596570171e-05, "loss": 0.0324, "step": 87420 }, { "epoch": 55.723390694710005, "grad_norm": 0.5423234105110168, "learning_rate": 8.489287859258796e-05, "loss": 0.0172, "step": 87430 }, { "epoch": 55.72976418100701, "grad_norm": 0.4633110463619232, "learning_rate": 8.488893079549281e-05, "loss": 0.0175, "step": 87440 }, { "epoch": 55.736137667304014, "grad_norm": 0.2761323153972626, "learning_rate": 8.48849825744642e-05, "loss": 0.024, "step": 87450 }, { "epoch": 55.74251115360102, "grad_norm": 0.5709750056266785, "learning_rate": 8.488103392955012e-05, "loss": 0.0244, "step": 87460 }, { "epoch": 55.74888463989802, "grad_norm": 0.4144880473613739, "learning_rate": 8.487708486079853e-05, "loss": 0.0198, "step": 87470 }, { "epoch": 55.75525812619503, "grad_norm": 0.41861483454704285, "learning_rate": 8.487313536825744e-05, "loss": 0.0152, "step": 87480 }, { "epoch": 55.76163161249203, "grad_norm": 0.45658978819847107, "learning_rate": 8.486918545197483e-05, "loss": 0.0134, "step": 87490 }, { "epoch": 55.76800509878904, "grad_norm": 0.42662158608436584, "learning_rate": 8.486523511199869e-05, "loss": 0.0223, "step": 87500 }, { "epoch": 55.77437858508604, "grad_norm": 0.38616234064102173, "learning_rate": 8.486128434837704e-05, "loss": 0.0224, "step": 87510 }, { "epoch": 55.78075207138305, "grad_norm": 0.46418219804763794, "learning_rate": 8.485733316115785e-05, "loss": 0.0177, "step": 87520 }, { "epoch": 55.78712555768005, "grad_norm": 0.4100997745990753, "learning_rate": 8.485338155038915e-05, "loss": 0.021, "step": 87530 }, { "epoch": 55.79349904397706, "grad_norm": 0.32803383469581604, "learning_rate": 8.484942951611896e-05, "loss": 0.0197, "step": 87540 }, { "epoch": 55.79987253027406, "grad_norm": 0.40545889735221863, "learning_rate": 8.484547705839532e-05, "loss": 0.0173, "step": 87550 }, { "epoch": 55.806246016571066, "grad_norm": 0.3022302985191345, "learning_rate": 8.48415241772662e-05, "loss": 0.0235, "step": 87560 }, { "epoch": 55.81261950286807, "grad_norm": 0.5484441518783569, "learning_rate": 8.483757087277969e-05, "loss": 0.0161, "step": 87570 }, { "epoch": 55.818992989165075, "grad_norm": 0.31769126653671265, "learning_rate": 8.483361714498379e-05, "loss": 0.0205, "step": 87580 }, { "epoch": 55.825366475462076, "grad_norm": 0.6617003679275513, "learning_rate": 8.482966299392656e-05, "loss": 0.0243, "step": 87590 }, { "epoch": 55.831739961759084, "grad_norm": 0.6437682509422302, "learning_rate": 8.482570841965604e-05, "loss": 0.0159, "step": 87600 }, { "epoch": 55.838113448056085, "grad_norm": 0.5488786101341248, "learning_rate": 8.482175342222029e-05, "loss": 0.0319, "step": 87610 }, { "epoch": 55.84448693435309, "grad_norm": 0.40729546546936035, "learning_rate": 8.481779800166735e-05, "loss": 0.0215, "step": 87620 }, { "epoch": 55.85086042065009, "grad_norm": 0.38643932342529297, "learning_rate": 8.481384215804529e-05, "loss": 0.0219, "step": 87630 }, { "epoch": 55.8572339069471, "grad_norm": 0.5844265818595886, "learning_rate": 8.480988589140217e-05, "loss": 0.0357, "step": 87640 }, { "epoch": 55.8636073932441, "grad_norm": 0.3872585594654083, "learning_rate": 8.480592920178608e-05, "loss": 0.0176, "step": 87650 }, { "epoch": 55.86998087954111, "grad_norm": 0.2649463415145874, "learning_rate": 8.48019720892451e-05, "loss": 0.0247, "step": 87660 }, { "epoch": 55.87635436583811, "grad_norm": 0.47644591331481934, "learning_rate": 8.479801455382727e-05, "loss": 0.02, "step": 87670 }, { "epoch": 55.88272785213512, "grad_norm": 0.32054460048675537, "learning_rate": 8.479405659558072e-05, "loss": 0.01, "step": 87680 }, { "epoch": 55.88910133843212, "grad_norm": 0.2663976550102234, "learning_rate": 8.479009821455355e-05, "loss": 0.0193, "step": 87690 }, { "epoch": 55.89547482472913, "grad_norm": 0.30819013714790344, "learning_rate": 8.478613941079383e-05, "loss": 0.0139, "step": 87700 }, { "epoch": 55.90184831102613, "grad_norm": 0.49763163924217224, "learning_rate": 8.478218018434968e-05, "loss": 0.0298, "step": 87710 }, { "epoch": 55.90822179732314, "grad_norm": 0.4986898601055145, "learning_rate": 8.477822053526917e-05, "loss": 0.0219, "step": 87720 }, { "epoch": 55.91459528362014, "grad_norm": 0.4842838943004608, "learning_rate": 8.477426046360046e-05, "loss": 0.026, "step": 87730 }, { "epoch": 55.920968769917145, "grad_norm": 0.719170093536377, "learning_rate": 8.477029996939167e-05, "loss": 0.0214, "step": 87740 }, { "epoch": 55.927342256214146, "grad_norm": 0.7870925664901733, "learning_rate": 8.476633905269089e-05, "loss": 0.0293, "step": 87750 }, { "epoch": 55.933715742511154, "grad_norm": 0.4626970887184143, "learning_rate": 8.476237771354627e-05, "loss": 0.0204, "step": 87760 }, { "epoch": 55.940089228808155, "grad_norm": 0.617261528968811, "learning_rate": 8.475841595200593e-05, "loss": 0.0253, "step": 87770 }, { "epoch": 55.94646271510516, "grad_norm": 0.6017148494720459, "learning_rate": 8.475445376811803e-05, "loss": 0.0206, "step": 87780 }, { "epoch": 55.952836201402164, "grad_norm": 0.4775473177433014, "learning_rate": 8.475049116193071e-05, "loss": 0.0189, "step": 87790 }, { "epoch": 55.95920968769917, "grad_norm": 0.23234955966472626, "learning_rate": 8.47465281334921e-05, "loss": 0.0269, "step": 87800 }, { "epoch": 55.96558317399617, "grad_norm": 0.5189412236213684, "learning_rate": 8.474256468285037e-05, "loss": 0.0266, "step": 87810 }, { "epoch": 55.97195666029318, "grad_norm": 0.44525763392448425, "learning_rate": 8.473860081005369e-05, "loss": 0.019, "step": 87820 }, { "epoch": 55.97833014659018, "grad_norm": 0.30188095569610596, "learning_rate": 8.473463651515019e-05, "loss": 0.0186, "step": 87830 }, { "epoch": 55.98470363288719, "grad_norm": 0.8558187484741211, "learning_rate": 8.473067179818808e-05, "loss": 0.0313, "step": 87840 }, { "epoch": 55.99107711918419, "grad_norm": 0.2960856258869171, "learning_rate": 8.47267066592155e-05, "loss": 0.0161, "step": 87850 }, { "epoch": 55.9974506054812, "grad_norm": 0.7070679068565369, "learning_rate": 8.472274109828065e-05, "loss": 0.0221, "step": 87860 }, { "epoch": 56.003824091778206, "grad_norm": 0.44926151633262634, "learning_rate": 8.471877511543172e-05, "loss": 0.0148, "step": 87870 }, { "epoch": 56.01019757807521, "grad_norm": 0.42732515931129456, "learning_rate": 8.471480871071689e-05, "loss": 0.0157, "step": 87880 }, { "epoch": 56.016571064372215, "grad_norm": 0.5049242973327637, "learning_rate": 8.471084188418436e-05, "loss": 0.0217, "step": 87890 }, { "epoch": 56.022944550669216, "grad_norm": 0.4717085659503937, "learning_rate": 8.470687463588231e-05, "loss": 0.0161, "step": 87900 }, { "epoch": 56.029318036966224, "grad_norm": 0.5494597554206848, "learning_rate": 8.470290696585899e-05, "loss": 0.0271, "step": 87910 }, { "epoch": 56.035691523263225, "grad_norm": 0.41327694058418274, "learning_rate": 8.469893887416259e-05, "loss": 0.0302, "step": 87920 }, { "epoch": 56.04206500956023, "grad_norm": 0.6733253598213196, "learning_rate": 8.469497036084128e-05, "loss": 0.0463, "step": 87930 }, { "epoch": 56.04843849585723, "grad_norm": 0.6372947692871094, "learning_rate": 8.469100142594337e-05, "loss": 0.0233, "step": 87940 }, { "epoch": 56.05481198215424, "grad_norm": 0.46489372849464417, "learning_rate": 8.4687032069517e-05, "loss": 0.0177, "step": 87950 }, { "epoch": 56.06118546845124, "grad_norm": 0.5751542448997498, "learning_rate": 8.468306229161047e-05, "loss": 0.0263, "step": 87960 }, { "epoch": 56.06755895474825, "grad_norm": 0.8481424450874329, "learning_rate": 8.467909209227197e-05, "loss": 0.0188, "step": 87970 }, { "epoch": 56.07393244104525, "grad_norm": 0.3661065697669983, "learning_rate": 8.467512147154976e-05, "loss": 0.019, "step": 87980 }, { "epoch": 56.08030592734226, "grad_norm": 0.4513249099254608, "learning_rate": 8.46711504294921e-05, "loss": 0.0152, "step": 87990 }, { "epoch": 56.08667941363926, "grad_norm": 0.4545053541660309, "learning_rate": 8.46671789661472e-05, "loss": 0.0282, "step": 88000 }, { "epoch": 56.09305289993627, "grad_norm": 0.6917222142219543, "learning_rate": 8.466320708156336e-05, "loss": 0.0234, "step": 88010 }, { "epoch": 56.09942638623327, "grad_norm": 0.9094108939170837, "learning_rate": 8.465923477578883e-05, "loss": 0.0219, "step": 88020 }, { "epoch": 56.10579987253028, "grad_norm": 0.5615320205688477, "learning_rate": 8.465526204887187e-05, "loss": 0.0237, "step": 88030 }, { "epoch": 56.11217335882728, "grad_norm": 0.5545759201049805, "learning_rate": 8.465128890086075e-05, "loss": 0.0239, "step": 88040 }, { "epoch": 56.118546845124285, "grad_norm": 0.3962620198726654, "learning_rate": 8.464731533180376e-05, "loss": 0.0175, "step": 88050 }, { "epoch": 56.124920331421286, "grad_norm": 0.4922754764556885, "learning_rate": 8.464334134174919e-05, "loss": 0.0239, "step": 88060 }, { "epoch": 56.131293817718294, "grad_norm": 0.4500688910484314, "learning_rate": 8.46393669307453e-05, "loss": 0.0199, "step": 88070 }, { "epoch": 56.137667304015295, "grad_norm": 0.44794538617134094, "learning_rate": 8.46353920988404e-05, "loss": 0.0332, "step": 88080 }, { "epoch": 56.1440407903123, "grad_norm": 0.3919735848903656, "learning_rate": 8.463141684608278e-05, "loss": 0.0183, "step": 88090 }, { "epoch": 56.150414276609304, "grad_norm": 0.491375207901001, "learning_rate": 8.462744117252074e-05, "loss": 0.0307, "step": 88100 }, { "epoch": 56.15678776290631, "grad_norm": 0.535897970199585, "learning_rate": 8.46234650782026e-05, "loss": 0.0304, "step": 88110 }, { "epoch": 56.16316124920331, "grad_norm": 0.5247924327850342, "learning_rate": 8.461948856317669e-05, "loss": 0.0195, "step": 88120 }, { "epoch": 56.16953473550032, "grad_norm": 0.2518024742603302, "learning_rate": 8.461551162749128e-05, "loss": 0.0183, "step": 88130 }, { "epoch": 56.17590822179732, "grad_norm": 0.22354301810264587, "learning_rate": 8.461153427119474e-05, "loss": 0.0233, "step": 88140 }, { "epoch": 56.18228170809433, "grad_norm": 0.2721497714519501, "learning_rate": 8.460755649433538e-05, "loss": 0.02, "step": 88150 }, { "epoch": 56.18865519439133, "grad_norm": 0.4538951814174652, "learning_rate": 8.460357829696152e-05, "loss": 0.0171, "step": 88160 }, { "epoch": 56.19502868068834, "grad_norm": 0.7810394167900085, "learning_rate": 8.459959967912153e-05, "loss": 0.0252, "step": 88170 }, { "epoch": 56.20140216698534, "grad_norm": 0.38765275478363037, "learning_rate": 8.459562064086372e-05, "loss": 0.0352, "step": 88180 }, { "epoch": 56.20777565328235, "grad_norm": 0.5936357975006104, "learning_rate": 8.459164118223646e-05, "loss": 0.0247, "step": 88190 }, { "epoch": 56.21414913957935, "grad_norm": 0.5780360102653503, "learning_rate": 8.45876613032881e-05, "loss": 0.0168, "step": 88200 }, { "epoch": 56.220522625876356, "grad_norm": 0.3825396001338959, "learning_rate": 8.458368100406701e-05, "loss": 0.0166, "step": 88210 }, { "epoch": 56.22689611217336, "grad_norm": 0.36209428310394287, "learning_rate": 8.457970028462153e-05, "loss": 0.0303, "step": 88220 }, { "epoch": 56.233269598470365, "grad_norm": 0.5339956879615784, "learning_rate": 8.457571914500004e-05, "loss": 0.0257, "step": 88230 }, { "epoch": 56.239643084767366, "grad_norm": 0.5305065512657166, "learning_rate": 8.457173758525095e-05, "loss": 0.0251, "step": 88240 }, { "epoch": 56.246016571064374, "grad_norm": 0.5893843770027161, "learning_rate": 8.456775560542256e-05, "loss": 0.0205, "step": 88250 }, { "epoch": 56.252390057361374, "grad_norm": 0.25814586877822876, "learning_rate": 8.456377320556332e-05, "loss": 0.0122, "step": 88260 }, { "epoch": 56.25876354365838, "grad_norm": 0.3158847987651825, "learning_rate": 8.455979038572162e-05, "loss": 0.0139, "step": 88270 }, { "epoch": 56.26513702995538, "grad_norm": 0.4592151939868927, "learning_rate": 8.455580714594582e-05, "loss": 0.0147, "step": 88280 }, { "epoch": 56.27151051625239, "grad_norm": 0.4977256953716278, "learning_rate": 8.455182348628434e-05, "loss": 0.0301, "step": 88290 }, { "epoch": 56.27788400254939, "grad_norm": 0.402782142162323, "learning_rate": 8.454783940678559e-05, "loss": 0.0195, "step": 88300 }, { "epoch": 56.2842574888464, "grad_norm": 0.6122016310691833, "learning_rate": 8.454385490749795e-05, "loss": 0.0161, "step": 88310 }, { "epoch": 56.2906309751434, "grad_norm": 0.6580690741539001, "learning_rate": 8.453986998846988e-05, "loss": 0.0353, "step": 88320 }, { "epoch": 56.29700446144041, "grad_norm": 0.2752344608306885, "learning_rate": 8.453588464974978e-05, "loss": 0.0136, "step": 88330 }, { "epoch": 56.30337794773741, "grad_norm": 0.37171950936317444, "learning_rate": 8.453189889138607e-05, "loss": 0.0288, "step": 88340 }, { "epoch": 56.30975143403442, "grad_norm": 4.347799301147461, "learning_rate": 8.452791271342717e-05, "loss": 0.03, "step": 88350 }, { "epoch": 56.31612492033142, "grad_norm": 0.6892208456993103, "learning_rate": 8.452392611592153e-05, "loss": 0.0239, "step": 88360 }, { "epoch": 56.322498406628426, "grad_norm": 0.5005861520767212, "learning_rate": 8.451993909891762e-05, "loss": 0.0202, "step": 88370 }, { "epoch": 56.32887189292543, "grad_norm": 0.2772069275379181, "learning_rate": 8.451595166246384e-05, "loss": 0.0213, "step": 88380 }, { "epoch": 56.335245379222435, "grad_norm": 0.6833258271217346, "learning_rate": 8.451196380660865e-05, "loss": 0.0144, "step": 88390 }, { "epoch": 56.341618865519436, "grad_norm": 0.789817750453949, "learning_rate": 8.450797553140053e-05, "loss": 0.0289, "step": 88400 }, { "epoch": 56.347992351816444, "grad_norm": 0.33028924465179443, "learning_rate": 8.450398683688792e-05, "loss": 0.0336, "step": 88410 }, { "epoch": 56.354365838113445, "grad_norm": 0.6567124128341675, "learning_rate": 8.44999977231193e-05, "loss": 0.0184, "step": 88420 }, { "epoch": 56.36073932441045, "grad_norm": 0.27108368277549744, "learning_rate": 8.449600819014312e-05, "loss": 0.0227, "step": 88430 }, { "epoch": 56.367112810707454, "grad_norm": 0.44227170944213867, "learning_rate": 8.449201823800787e-05, "loss": 0.0197, "step": 88440 }, { "epoch": 56.37348629700446, "grad_norm": 0.29102951288223267, "learning_rate": 8.448802786676205e-05, "loss": 0.0215, "step": 88450 }, { "epoch": 56.37985978330146, "grad_norm": 0.45583412051200867, "learning_rate": 8.448403707645412e-05, "loss": 0.0385, "step": 88460 }, { "epoch": 56.38623326959847, "grad_norm": 0.4818269610404968, "learning_rate": 8.448004586713258e-05, "loss": 0.0218, "step": 88470 }, { "epoch": 56.39260675589547, "grad_norm": 0.38694676756858826, "learning_rate": 8.447605423884593e-05, "loss": 0.0148, "step": 88480 }, { "epoch": 56.39898024219248, "grad_norm": 0.46213018894195557, "learning_rate": 8.447206219164268e-05, "loss": 0.022, "step": 88490 }, { "epoch": 56.40535372848949, "grad_norm": 0.37967929244041443, "learning_rate": 8.446806972557131e-05, "loss": 0.0272, "step": 88500 }, { "epoch": 56.41172721478649, "grad_norm": 0.7696411609649658, "learning_rate": 8.446407684068037e-05, "loss": 0.0271, "step": 88510 }, { "epoch": 56.418100701083496, "grad_norm": 0.6426995396614075, "learning_rate": 8.446008353701833e-05, "loss": 0.0217, "step": 88520 }, { "epoch": 56.4244741873805, "grad_norm": 0.5220997333526611, "learning_rate": 8.445608981463375e-05, "loss": 0.0161, "step": 88530 }, { "epoch": 56.430847673677505, "grad_norm": 0.37450170516967773, "learning_rate": 8.445209567357515e-05, "loss": 0.0117, "step": 88540 }, { "epoch": 56.437221159974506, "grad_norm": 0.38822221755981445, "learning_rate": 8.444810111389108e-05, "loss": 0.0193, "step": 88550 }, { "epoch": 56.443594646271514, "grad_norm": 0.27645304799079895, "learning_rate": 8.444410613563004e-05, "loss": 0.0212, "step": 88560 }, { "epoch": 56.449968132568515, "grad_norm": 0.46493765711784363, "learning_rate": 8.444011073884057e-05, "loss": 0.0245, "step": 88570 }, { "epoch": 56.45634161886552, "grad_norm": 0.6677665114402771, "learning_rate": 8.443611492357126e-05, "loss": 0.0245, "step": 88580 }, { "epoch": 56.46271510516252, "grad_norm": 0.3798067569732666, "learning_rate": 8.443211868987063e-05, "loss": 0.0224, "step": 88590 }, { "epoch": 56.46908859145953, "grad_norm": 0.3099798262119293, "learning_rate": 8.442812203778726e-05, "loss": 0.0134, "step": 88600 }, { "epoch": 56.47546207775653, "grad_norm": 0.4550483524799347, "learning_rate": 8.44241249673697e-05, "loss": 0.0201, "step": 88610 }, { "epoch": 56.48183556405354, "grad_norm": 0.37614384293556213, "learning_rate": 8.44201274786665e-05, "loss": 0.0236, "step": 88620 }, { "epoch": 56.48820905035054, "grad_norm": 0.4918666183948517, "learning_rate": 8.441612957172626e-05, "loss": 0.0233, "step": 88630 }, { "epoch": 56.49458253664755, "grad_norm": 0.365041047334671, "learning_rate": 8.441213124659757e-05, "loss": 0.0217, "step": 88640 }, { "epoch": 56.50095602294455, "grad_norm": 0.31699469685554504, "learning_rate": 8.4408132503329e-05, "loss": 0.0212, "step": 88650 }, { "epoch": 56.50732950924156, "grad_norm": 0.371976375579834, "learning_rate": 8.44041333419691e-05, "loss": 0.0207, "step": 88660 }, { "epoch": 56.51370299553856, "grad_norm": 0.3743080496788025, "learning_rate": 8.440013376256653e-05, "loss": 0.0145, "step": 88670 }, { "epoch": 56.52007648183557, "grad_norm": 0.6124394536018372, "learning_rate": 8.439613376516984e-05, "loss": 0.0143, "step": 88680 }, { "epoch": 56.52644996813257, "grad_norm": 0.42020294070243835, "learning_rate": 8.439213334982765e-05, "loss": 0.0261, "step": 88690 }, { "epoch": 56.532823454429575, "grad_norm": 0.31843769550323486, "learning_rate": 8.438813251658857e-05, "loss": 0.0146, "step": 88700 }, { "epoch": 56.539196940726576, "grad_norm": 0.6491100788116455, "learning_rate": 8.438413126550119e-05, "loss": 0.0255, "step": 88710 }, { "epoch": 56.545570427023584, "grad_norm": 0.2174498736858368, "learning_rate": 8.438012959661419e-05, "loss": 0.0137, "step": 88720 }, { "epoch": 56.551943913320585, "grad_norm": 0.40119218826293945, "learning_rate": 8.437612750997614e-05, "loss": 0.0222, "step": 88730 }, { "epoch": 56.55831739961759, "grad_norm": 0.40927138924598694, "learning_rate": 8.437212500563567e-05, "loss": 0.0175, "step": 88740 }, { "epoch": 56.564690885914594, "grad_norm": 0.605706512928009, "learning_rate": 8.436812208364144e-05, "loss": 0.03, "step": 88750 }, { "epoch": 56.5710643722116, "grad_norm": 0.4160982072353363, "learning_rate": 8.436411874404209e-05, "loss": 0.0247, "step": 88760 }, { "epoch": 56.5774378585086, "grad_norm": 0.3002631962299347, "learning_rate": 8.436011498688624e-05, "loss": 0.0209, "step": 88770 }, { "epoch": 56.58381134480561, "grad_norm": 0.22086432576179504, "learning_rate": 8.435611081222255e-05, "loss": 0.0142, "step": 88780 }, { "epoch": 56.59018483110261, "grad_norm": 0.30750617384910583, "learning_rate": 8.435210622009967e-05, "loss": 0.0208, "step": 88790 }, { "epoch": 56.59655831739962, "grad_norm": 0.6891700029373169, "learning_rate": 8.434810121056627e-05, "loss": 0.0339, "step": 88800 }, { "epoch": 56.60293180369662, "grad_norm": 0.4078996479511261, "learning_rate": 8.434409578367102e-05, "loss": 0.0197, "step": 88810 }, { "epoch": 56.60930528999363, "grad_norm": 0.6693564057350159, "learning_rate": 8.434008993946256e-05, "loss": 0.0276, "step": 88820 }, { "epoch": 56.61567877629063, "grad_norm": 0.3512659966945648, "learning_rate": 8.433608367798961e-05, "loss": 0.0122, "step": 88830 }, { "epoch": 56.62205226258764, "grad_norm": 0.21837028861045837, "learning_rate": 8.433207699930081e-05, "loss": 0.0124, "step": 88840 }, { "epoch": 56.62842574888464, "grad_norm": 0.4864201545715332, "learning_rate": 8.432806990344486e-05, "loss": 0.0203, "step": 88850 }, { "epoch": 56.634799235181646, "grad_norm": 0.25383928418159485, "learning_rate": 8.432406239047046e-05, "loss": 0.02, "step": 88860 }, { "epoch": 56.64117272147865, "grad_norm": 0.502954363822937, "learning_rate": 8.432005446042627e-05, "loss": 0.0319, "step": 88870 }, { "epoch": 56.647546207775655, "grad_norm": 0.7558793425559998, "learning_rate": 8.431604611336105e-05, "loss": 0.0213, "step": 88880 }, { "epoch": 56.653919694072655, "grad_norm": 0.30104249715805054, "learning_rate": 8.431203734932345e-05, "loss": 0.0193, "step": 88890 }, { "epoch": 56.66029318036966, "grad_norm": 0.4844406843185425, "learning_rate": 8.430802816836221e-05, "loss": 0.0221, "step": 88900 }, { "epoch": 56.666666666666664, "grad_norm": 0.3854568898677826, "learning_rate": 8.430401857052601e-05, "loss": 0.0156, "step": 88910 }, { "epoch": 56.67304015296367, "grad_norm": 0.5751428008079529, "learning_rate": 8.430000855586362e-05, "loss": 0.035, "step": 88920 }, { "epoch": 56.67941363926067, "grad_norm": 0.5251092314720154, "learning_rate": 8.429599812442372e-05, "loss": 0.0184, "step": 88930 }, { "epoch": 56.68578712555768, "grad_norm": 0.5491990447044373, "learning_rate": 8.429198727625509e-05, "loss": 0.0232, "step": 88940 }, { "epoch": 56.69216061185468, "grad_norm": 0.4346757233142853, "learning_rate": 8.428797601140642e-05, "loss": 0.0214, "step": 88950 }, { "epoch": 56.69853409815169, "grad_norm": 0.24910549819469452, "learning_rate": 8.428396432992648e-05, "loss": 0.0127, "step": 88960 }, { "epoch": 56.70490758444869, "grad_norm": 0.4894624948501587, "learning_rate": 8.4279952231864e-05, "loss": 0.0258, "step": 88970 }, { "epoch": 56.7112810707457, "grad_norm": 0.47326216101646423, "learning_rate": 8.427593971726772e-05, "loss": 0.0207, "step": 88980 }, { "epoch": 56.7176545570427, "grad_norm": 0.2441207468509674, "learning_rate": 8.427192678618643e-05, "loss": 0.0266, "step": 88990 }, { "epoch": 56.72402804333971, "grad_norm": 0.4305455982685089, "learning_rate": 8.426791343866887e-05, "loss": 0.0163, "step": 89000 }, { "epoch": 56.73040152963671, "grad_norm": 0.4559779167175293, "learning_rate": 8.426389967476381e-05, "loss": 0.0176, "step": 89010 }, { "epoch": 56.736775015933716, "grad_norm": 0.4351821541786194, "learning_rate": 8.425988549452e-05, "loss": 0.027, "step": 89020 }, { "epoch": 56.74314850223072, "grad_norm": 0.3989706039428711, "learning_rate": 8.425587089798624e-05, "loss": 0.017, "step": 89030 }, { "epoch": 56.749521988527725, "grad_norm": 0.45699042081832886, "learning_rate": 8.425185588521131e-05, "loss": 0.0401, "step": 89040 }, { "epoch": 56.755895474824726, "grad_norm": 0.3551419973373413, "learning_rate": 8.424784045624398e-05, "loss": 0.018, "step": 89050 }, { "epoch": 56.762268961121734, "grad_norm": 0.26561206579208374, "learning_rate": 8.424382461113307e-05, "loss": 0.0193, "step": 89060 }, { "epoch": 56.768642447418735, "grad_norm": 0.3921823799610138, "learning_rate": 8.423980834992736e-05, "loss": 0.0138, "step": 89070 }, { "epoch": 56.77501593371574, "grad_norm": 0.6603374481201172, "learning_rate": 8.423579167267564e-05, "loss": 0.0235, "step": 89080 }, { "epoch": 56.781389420012744, "grad_norm": 0.425673246383667, "learning_rate": 8.423177457942674e-05, "loss": 0.0281, "step": 89090 }, { "epoch": 56.78776290630975, "grad_norm": 0.5114652514457703, "learning_rate": 8.422775707022946e-05, "loss": 0.0212, "step": 89100 }, { "epoch": 56.79413639260676, "grad_norm": 0.4565389156341553, "learning_rate": 8.42237391451326e-05, "loss": 0.029, "step": 89110 }, { "epoch": 56.80050987890376, "grad_norm": 0.31695836782455444, "learning_rate": 8.4219720804185e-05, "loss": 0.0163, "step": 89120 }, { "epoch": 56.80688336520077, "grad_norm": 0.3238644301891327, "learning_rate": 8.421570204743549e-05, "loss": 0.0195, "step": 89130 }, { "epoch": 56.81325685149777, "grad_norm": 0.5549646019935608, "learning_rate": 8.421168287493288e-05, "loss": 0.0177, "step": 89140 }, { "epoch": 56.81963033779478, "grad_norm": 0.785845160484314, "learning_rate": 8.420766328672604e-05, "loss": 0.0144, "step": 89150 }, { "epoch": 56.82600382409178, "grad_norm": 0.6765007376670837, "learning_rate": 8.420364328286379e-05, "loss": 0.0161, "step": 89160 }, { "epoch": 56.832377310388786, "grad_norm": 0.29179131984710693, "learning_rate": 8.419962286339497e-05, "loss": 0.0129, "step": 89170 }, { "epoch": 56.83875079668579, "grad_norm": 0.24377037584781647, "learning_rate": 8.419560202836844e-05, "loss": 0.0205, "step": 89180 }, { "epoch": 56.845124282982795, "grad_norm": 0.5405519604682922, "learning_rate": 8.419158077783309e-05, "loss": 0.0214, "step": 89190 }, { "epoch": 56.851497769279796, "grad_norm": 0.16153636574745178, "learning_rate": 8.418755911183773e-05, "loss": 0.0143, "step": 89200 }, { "epoch": 56.8578712555768, "grad_norm": 0.3196083903312683, "learning_rate": 8.418353703043125e-05, "loss": 0.0139, "step": 89210 }, { "epoch": 56.864244741873804, "grad_norm": 0.8624732494354248, "learning_rate": 8.417951453366251e-05, "loss": 0.0183, "step": 89220 }, { "epoch": 56.87061822817081, "grad_norm": 0.6127678751945496, "learning_rate": 8.41754916215804e-05, "loss": 0.0181, "step": 89230 }, { "epoch": 56.87699171446781, "grad_norm": 0.38532915711402893, "learning_rate": 8.417146829423383e-05, "loss": 0.0243, "step": 89240 }, { "epoch": 56.88336520076482, "grad_norm": 0.5425240397453308, "learning_rate": 8.416744455167162e-05, "loss": 0.0188, "step": 89250 }, { "epoch": 56.88973868706182, "grad_norm": 0.4425295889377594, "learning_rate": 8.416342039394272e-05, "loss": 0.0227, "step": 89260 }, { "epoch": 56.89611217335883, "grad_norm": 0.3195371925830841, "learning_rate": 8.4159395821096e-05, "loss": 0.02, "step": 89270 }, { "epoch": 56.90248565965583, "grad_norm": 0.7239754796028137, "learning_rate": 8.415537083318038e-05, "loss": 0.0216, "step": 89280 }, { "epoch": 56.90885914595284, "grad_norm": 0.2908647060394287, "learning_rate": 8.415134543024474e-05, "loss": 0.0227, "step": 89290 }, { "epoch": 56.91523263224984, "grad_norm": 0.5402952432632446, "learning_rate": 8.4147319612338e-05, "loss": 0.0345, "step": 89300 }, { "epoch": 56.92160611854685, "grad_norm": 0.4413334131240845, "learning_rate": 8.41432933795091e-05, "loss": 0.0193, "step": 89310 }, { "epoch": 56.92797960484385, "grad_norm": 0.4403780400753021, "learning_rate": 8.413926673180696e-05, "loss": 0.0189, "step": 89320 }, { "epoch": 56.934353091140856, "grad_norm": 0.39796262979507446, "learning_rate": 8.413523966928048e-05, "loss": 0.026, "step": 89330 }, { "epoch": 56.94072657743786, "grad_norm": 0.41868263483047485, "learning_rate": 8.413121219197861e-05, "loss": 0.0178, "step": 89340 }, { "epoch": 56.947100063734865, "grad_norm": 0.671564519405365, "learning_rate": 8.41271842999503e-05, "loss": 0.0219, "step": 89350 }, { "epoch": 56.953473550031866, "grad_norm": 0.4594668745994568, "learning_rate": 8.412315599324448e-05, "loss": 0.0211, "step": 89360 }, { "epoch": 56.959847036328874, "grad_norm": 0.316904753446579, "learning_rate": 8.411912727191006e-05, "loss": 0.0212, "step": 89370 }, { "epoch": 56.966220522625875, "grad_norm": 0.3670274615287781, "learning_rate": 8.411509813599608e-05, "loss": 0.0192, "step": 89380 }, { "epoch": 56.97259400892288, "grad_norm": 0.3125549256801605, "learning_rate": 8.411106858555142e-05, "loss": 0.0189, "step": 89390 }, { "epoch": 56.978967495219884, "grad_norm": 0.30466461181640625, "learning_rate": 8.410703862062508e-05, "loss": 0.0246, "step": 89400 }, { "epoch": 56.98534098151689, "grad_norm": 0.30217137932777405, "learning_rate": 8.4103008241266e-05, "loss": 0.0207, "step": 89410 }, { "epoch": 56.99171446781389, "grad_norm": 0.45970234274864197, "learning_rate": 8.40989774475232e-05, "loss": 0.0157, "step": 89420 }, { "epoch": 56.9980879541109, "grad_norm": 0.272697389125824, "learning_rate": 8.40949462394456e-05, "loss": 0.0173, "step": 89430 }, { "epoch": 57.0044614404079, "grad_norm": 0.2503385841846466, "learning_rate": 8.409091461708224e-05, "loss": 0.0173, "step": 89440 }, { "epoch": 57.01083492670491, "grad_norm": 0.31914013624191284, "learning_rate": 8.408688258048207e-05, "loss": 0.0177, "step": 89450 }, { "epoch": 57.01720841300191, "grad_norm": 0.4664101302623749, "learning_rate": 8.408285012969408e-05, "loss": 0.0233, "step": 89460 }, { "epoch": 57.02358189929892, "grad_norm": 0.4527615010738373, "learning_rate": 8.40788172647673e-05, "loss": 0.0169, "step": 89470 }, { "epoch": 57.02995538559592, "grad_norm": 0.1816396415233612, "learning_rate": 8.40747839857507e-05, "loss": 0.0245, "step": 89480 }, { "epoch": 57.03632887189293, "grad_norm": 0.5130508542060852, "learning_rate": 8.407075029269332e-05, "loss": 0.0286, "step": 89490 }, { "epoch": 57.04270235818993, "grad_norm": 0.5528929829597473, "learning_rate": 8.406671618564413e-05, "loss": 0.02, "step": 89500 }, { "epoch": 57.049075844486936, "grad_norm": 0.6410553455352783, "learning_rate": 8.406268166465218e-05, "loss": 0.0285, "step": 89510 }, { "epoch": 57.05544933078394, "grad_norm": 0.5955047011375427, "learning_rate": 8.405864672976652e-05, "loss": 0.031, "step": 89520 }, { "epoch": 57.061822817080945, "grad_norm": 0.35587820410728455, "learning_rate": 8.405461138103612e-05, "loss": 0.0222, "step": 89530 }, { "epoch": 57.068196303377945, "grad_norm": 0.49654316902160645, "learning_rate": 8.405057561851005e-05, "loss": 0.0191, "step": 89540 }, { "epoch": 57.07456978967495, "grad_norm": 0.2847869098186493, "learning_rate": 8.40465394422373e-05, "loss": 0.0239, "step": 89550 }, { "epoch": 57.080943275971954, "grad_norm": 0.29756835103034973, "learning_rate": 8.4042502852267e-05, "loss": 0.0165, "step": 89560 }, { "epoch": 57.08731676226896, "grad_norm": 0.31888434290885925, "learning_rate": 8.403846584864812e-05, "loss": 0.0171, "step": 89570 }, { "epoch": 57.09369024856596, "grad_norm": 0.6209361553192139, "learning_rate": 8.403442843142974e-05, "loss": 0.0198, "step": 89580 }, { "epoch": 57.10006373486297, "grad_norm": 0.24462798237800598, "learning_rate": 8.403039060066094e-05, "loss": 0.0146, "step": 89590 }, { "epoch": 57.10643722115997, "grad_norm": 0.30650609731674194, "learning_rate": 8.402635235639075e-05, "loss": 0.0311, "step": 89600 }, { "epoch": 57.11281070745698, "grad_norm": 0.411437451839447, "learning_rate": 8.402231369866827e-05, "loss": 0.0133, "step": 89610 }, { "epoch": 57.11918419375398, "grad_norm": 0.2467217892408371, "learning_rate": 8.401827462754253e-05, "loss": 0.0247, "step": 89620 }, { "epoch": 57.12555768005099, "grad_norm": 0.3094038665294647, "learning_rate": 8.401423514306265e-05, "loss": 0.0173, "step": 89630 }, { "epoch": 57.13193116634799, "grad_norm": 0.7158586978912354, "learning_rate": 8.401019524527769e-05, "loss": 0.0268, "step": 89640 }, { "epoch": 57.138304652645, "grad_norm": 0.42708465456962585, "learning_rate": 8.400615493423675e-05, "loss": 0.0298, "step": 89650 }, { "epoch": 57.144678138942, "grad_norm": 0.40915343165397644, "learning_rate": 8.400211420998893e-05, "loss": 0.0187, "step": 89660 }, { "epoch": 57.151051625239006, "grad_norm": 0.2856992781162262, "learning_rate": 8.399807307258332e-05, "loss": 0.017, "step": 89670 }, { "epoch": 57.15742511153601, "grad_norm": 0.5486305356025696, "learning_rate": 8.399403152206901e-05, "loss": 0.0231, "step": 89680 }, { "epoch": 57.163798597833015, "grad_norm": 0.4538962244987488, "learning_rate": 8.398998955849513e-05, "loss": 0.0261, "step": 89690 }, { "epoch": 57.170172084130016, "grad_norm": 0.40554606914520264, "learning_rate": 8.398594718191079e-05, "loss": 0.0143, "step": 89700 }, { "epoch": 57.176545570427024, "grad_norm": 0.31516873836517334, "learning_rate": 8.398190439236508e-05, "loss": 0.0182, "step": 89710 }, { "epoch": 57.182919056724025, "grad_norm": 0.4766589403152466, "learning_rate": 8.397786118990717e-05, "loss": 0.0264, "step": 89720 }, { "epoch": 57.18929254302103, "grad_norm": 0.5214773416519165, "learning_rate": 8.397381757458616e-05, "loss": 0.0253, "step": 89730 }, { "epoch": 57.19566602931803, "grad_norm": 0.45385462045669556, "learning_rate": 8.396977354645119e-05, "loss": 0.0241, "step": 89740 }, { "epoch": 57.20203951561504, "grad_norm": 0.30090734362602234, "learning_rate": 8.396572910555138e-05, "loss": 0.0247, "step": 89750 }, { "epoch": 57.20841300191205, "grad_norm": 0.35574355721473694, "learning_rate": 8.396168425193592e-05, "loss": 0.0183, "step": 89760 }, { "epoch": 57.21478648820905, "grad_norm": 0.7694368958473206, "learning_rate": 8.395763898565392e-05, "loss": 0.0185, "step": 89770 }, { "epoch": 57.22115997450606, "grad_norm": 0.5543959140777588, "learning_rate": 8.395359330675454e-05, "loss": 0.0153, "step": 89780 }, { "epoch": 57.22753346080306, "grad_norm": 0.3384433388710022, "learning_rate": 8.394954721528696e-05, "loss": 0.0209, "step": 89790 }, { "epoch": 57.23390694710007, "grad_norm": 0.699780285358429, "learning_rate": 8.394550071130031e-05, "loss": 0.0398, "step": 89800 }, { "epoch": 57.24028043339707, "grad_norm": 0.32391348481178284, "learning_rate": 8.39414537948438e-05, "loss": 0.0289, "step": 89810 }, { "epoch": 57.246653919694076, "grad_norm": 0.4683291018009186, "learning_rate": 8.393740646596655e-05, "loss": 0.035, "step": 89820 }, { "epoch": 57.25302740599108, "grad_norm": 0.6120004653930664, "learning_rate": 8.393335872471777e-05, "loss": 0.0275, "step": 89830 }, { "epoch": 57.259400892288085, "grad_norm": 0.419186532497406, "learning_rate": 8.392931057114667e-05, "loss": 0.0257, "step": 89840 }, { "epoch": 57.265774378585085, "grad_norm": 0.5148308277130127, "learning_rate": 8.39252620053024e-05, "loss": 0.0195, "step": 89850 }, { "epoch": 57.27214786488209, "grad_norm": 0.5679914355278015, "learning_rate": 8.392121302723417e-05, "loss": 0.0289, "step": 89860 }, { "epoch": 57.278521351179094, "grad_norm": 0.4626162648200989, "learning_rate": 8.391716363699115e-05, "loss": 0.0232, "step": 89870 }, { "epoch": 57.2848948374761, "grad_norm": 0.31322181224823, "learning_rate": 8.391311383462259e-05, "loss": 0.015, "step": 89880 }, { "epoch": 57.2912683237731, "grad_norm": 0.3781730830669403, "learning_rate": 8.390906362017767e-05, "loss": 0.0182, "step": 89890 }, { "epoch": 57.29764181007011, "grad_norm": 0.4961112439632416, "learning_rate": 8.390501299370563e-05, "loss": 0.0212, "step": 89900 }, { "epoch": 57.30401529636711, "grad_norm": 0.6286005973815918, "learning_rate": 8.390096195525565e-05, "loss": 0.0166, "step": 89910 }, { "epoch": 57.31038878266412, "grad_norm": 0.5444197058677673, "learning_rate": 8.389691050487699e-05, "loss": 0.0223, "step": 89920 }, { "epoch": 57.31676226896112, "grad_norm": 0.2742018401622772, "learning_rate": 8.389285864261887e-05, "loss": 0.0176, "step": 89930 }, { "epoch": 57.32313575525813, "grad_norm": 0.49162429571151733, "learning_rate": 8.38888063685305e-05, "loss": 0.0239, "step": 89940 }, { "epoch": 57.32950924155513, "grad_norm": 0.5490507483482361, "learning_rate": 8.388475368266113e-05, "loss": 0.014, "step": 89950 }, { "epoch": 57.33588272785214, "grad_norm": 0.3573302626609802, "learning_rate": 8.388070058506001e-05, "loss": 0.0151, "step": 89960 }, { "epoch": 57.34225621414914, "grad_norm": 0.4907649755477905, "learning_rate": 8.38766470757764e-05, "loss": 0.0243, "step": 89970 }, { "epoch": 57.348629700446146, "grad_norm": 0.7088059782981873, "learning_rate": 8.387259315485954e-05, "loss": 0.0348, "step": 89980 }, { "epoch": 57.35500318674315, "grad_norm": 0.36850523948669434, "learning_rate": 8.38685388223587e-05, "loss": 0.0184, "step": 89990 }, { "epoch": 57.361376673040155, "grad_norm": 0.3041794002056122, "learning_rate": 8.386448407832313e-05, "loss": 0.0147, "step": 90000 }, { "epoch": 57.367750159337156, "grad_norm": 0.3795422315597534, "learning_rate": 8.38604289228021e-05, "loss": 0.02, "step": 90010 }, { "epoch": 57.374123645634164, "grad_norm": 0.33706802129745483, "learning_rate": 8.38563733558449e-05, "loss": 0.0219, "step": 90020 }, { "epoch": 57.380497131931165, "grad_norm": 0.47220587730407715, "learning_rate": 8.385231737750078e-05, "loss": 0.0203, "step": 90030 }, { "epoch": 57.38687061822817, "grad_norm": 0.2156868427991867, "learning_rate": 8.384826098781906e-05, "loss": 0.0134, "step": 90040 }, { "epoch": 57.393244104525174, "grad_norm": 0.3545942008495331, "learning_rate": 8.384420418684901e-05, "loss": 0.0164, "step": 90050 }, { "epoch": 57.39961759082218, "grad_norm": 0.3200230002403259, "learning_rate": 8.384014697463992e-05, "loss": 0.0251, "step": 90060 }, { "epoch": 57.40599107711918, "grad_norm": 0.34192800521850586, "learning_rate": 8.383608935124109e-05, "loss": 0.0163, "step": 90070 }, { "epoch": 57.41236456341619, "grad_norm": 0.7476663589477539, "learning_rate": 8.383203131670183e-05, "loss": 0.0285, "step": 90080 }, { "epoch": 57.41873804971319, "grad_norm": 0.5134480595588684, "learning_rate": 8.382797287107145e-05, "loss": 0.0211, "step": 90090 }, { "epoch": 57.4251115360102, "grad_norm": 0.40954017639160156, "learning_rate": 8.382391401439927e-05, "loss": 0.0227, "step": 90100 }, { "epoch": 57.4314850223072, "grad_norm": 0.6072366237640381, "learning_rate": 8.381985474673458e-05, "loss": 0.0193, "step": 90110 }, { "epoch": 57.43785850860421, "grad_norm": 0.4288114011287689, "learning_rate": 8.381579506812673e-05, "loss": 0.0195, "step": 90120 }, { "epoch": 57.44423199490121, "grad_norm": 0.27396345138549805, "learning_rate": 8.381173497862504e-05, "loss": 0.0176, "step": 90130 }, { "epoch": 57.45060548119822, "grad_norm": 0.6597847938537598, "learning_rate": 8.380767447827886e-05, "loss": 0.023, "step": 90140 }, { "epoch": 57.45697896749522, "grad_norm": 0.2794700562953949, "learning_rate": 8.380361356713751e-05, "loss": 0.0164, "step": 90150 }, { "epoch": 57.463352453792226, "grad_norm": 0.277604877948761, "learning_rate": 8.379955224525032e-05, "loss": 0.0136, "step": 90160 }, { "epoch": 57.469725940089226, "grad_norm": 0.3635536730289459, "learning_rate": 8.379549051266669e-05, "loss": 0.0171, "step": 90170 }, { "epoch": 57.476099426386234, "grad_norm": 0.2913624048233032, "learning_rate": 8.379142836943592e-05, "loss": 0.0128, "step": 90180 }, { "epoch": 57.482472912683235, "grad_norm": 0.6907758116722107, "learning_rate": 8.37873658156074e-05, "loss": 0.0262, "step": 90190 }, { "epoch": 57.48884639898024, "grad_norm": 0.2994072139263153, "learning_rate": 8.378330285123048e-05, "loss": 0.019, "step": 90200 }, { "epoch": 57.495219885277244, "grad_norm": 0.5289950966835022, "learning_rate": 8.377923947635453e-05, "loss": 0.015, "step": 90210 }, { "epoch": 57.50159337157425, "grad_norm": 0.26163166761398315, "learning_rate": 8.377517569102894e-05, "loss": 0.0136, "step": 90220 }, { "epoch": 57.50796685787125, "grad_norm": 0.3765043616294861, "learning_rate": 8.377111149530306e-05, "loss": 0.0186, "step": 90230 }, { "epoch": 57.51434034416826, "grad_norm": 0.6472104787826538, "learning_rate": 8.376704688922629e-05, "loss": 0.0147, "step": 90240 }, { "epoch": 57.52071383046526, "grad_norm": 0.34595000743865967, "learning_rate": 8.376298187284803e-05, "loss": 0.0228, "step": 90250 }, { "epoch": 57.52708731676227, "grad_norm": 0.6250975728034973, "learning_rate": 8.375891644621767e-05, "loss": 0.023, "step": 90260 }, { "epoch": 57.53346080305927, "grad_norm": 0.4491671919822693, "learning_rate": 8.375485060938459e-05, "loss": 0.0145, "step": 90270 }, { "epoch": 57.53983428935628, "grad_norm": 0.5883461833000183, "learning_rate": 8.375078436239821e-05, "loss": 0.0223, "step": 90280 }, { "epoch": 57.54620777565328, "grad_norm": 0.3602447211742401, "learning_rate": 8.374671770530793e-05, "loss": 0.0222, "step": 90290 }, { "epoch": 57.55258126195029, "grad_norm": 0.43664276599884033, "learning_rate": 8.374265063816316e-05, "loss": 0.0262, "step": 90300 }, { "epoch": 57.55895474824729, "grad_norm": 0.3652399778366089, "learning_rate": 8.373858316101335e-05, "loss": 0.0183, "step": 90310 }, { "epoch": 57.565328234544296, "grad_norm": 0.342951238155365, "learning_rate": 8.373451527390787e-05, "loss": 0.0158, "step": 90320 }, { "epoch": 57.5717017208413, "grad_norm": 0.2996302545070648, "learning_rate": 8.373044697689621e-05, "loss": 0.0207, "step": 90330 }, { "epoch": 57.578075207138305, "grad_norm": 0.470356822013855, "learning_rate": 8.372637827002776e-05, "loss": 0.0229, "step": 90340 }, { "epoch": 57.584448693435306, "grad_norm": 0.48482009768486023, "learning_rate": 8.372230915335195e-05, "loss": 0.0187, "step": 90350 }, { "epoch": 57.590822179732314, "grad_norm": 0.6500172019004822, "learning_rate": 8.371823962691827e-05, "loss": 0.0178, "step": 90360 }, { "epoch": 57.59719566602932, "grad_norm": 0.3064844608306885, "learning_rate": 8.371416969077614e-05, "loss": 0.0165, "step": 90370 }, { "epoch": 57.60356915232632, "grad_norm": 0.49580442905426025, "learning_rate": 8.371009934497502e-05, "loss": 0.0201, "step": 90380 }, { "epoch": 57.60994263862333, "grad_norm": 0.3165915608406067, "learning_rate": 8.370602858956434e-05, "loss": 0.0202, "step": 90390 }, { "epoch": 57.61631612492033, "grad_norm": 0.3350183665752411, "learning_rate": 8.370195742459362e-05, "loss": 0.0224, "step": 90400 }, { "epoch": 57.62268961121734, "grad_norm": 0.9619952440261841, "learning_rate": 8.36978858501123e-05, "loss": 0.0296, "step": 90410 }, { "epoch": 57.62906309751434, "grad_norm": 0.39266541600227356, "learning_rate": 8.369381386616981e-05, "loss": 0.0174, "step": 90420 }, { "epoch": 57.63543658381135, "grad_norm": 0.43098047375679016, "learning_rate": 8.36897414728157e-05, "loss": 0.0167, "step": 90430 }, { "epoch": 57.64181007010835, "grad_norm": 0.6389949321746826, "learning_rate": 8.368566867009942e-05, "loss": 0.0166, "step": 90440 }, { "epoch": 57.64818355640536, "grad_norm": 0.2967068552970886, "learning_rate": 8.368159545807047e-05, "loss": 0.0213, "step": 90450 }, { "epoch": 57.65455704270236, "grad_norm": 0.4471738636493683, "learning_rate": 8.36775218367783e-05, "loss": 0.0207, "step": 90460 }, { "epoch": 57.660930528999366, "grad_norm": 0.24095755815505981, "learning_rate": 8.367344780627246e-05, "loss": 0.0192, "step": 90470 }, { "epoch": 57.66730401529637, "grad_norm": 0.28147220611572266, "learning_rate": 8.366937336660246e-05, "loss": 0.0213, "step": 90480 }, { "epoch": 57.673677501593374, "grad_norm": 0.380973756313324, "learning_rate": 8.366529851781776e-05, "loss": 0.0217, "step": 90490 }, { "epoch": 57.680050987890375, "grad_norm": 0.27357548475265503, "learning_rate": 8.36612232599679e-05, "loss": 0.0202, "step": 90500 }, { "epoch": 57.68642447418738, "grad_norm": 0.5490698218345642, "learning_rate": 8.36571475931024e-05, "loss": 0.0209, "step": 90510 }, { "epoch": 57.692797960484384, "grad_norm": 0.40168046951293945, "learning_rate": 8.365307151727078e-05, "loss": 0.0173, "step": 90520 }, { "epoch": 57.69917144678139, "grad_norm": 0.4302065670490265, "learning_rate": 8.364899503252257e-05, "loss": 0.0216, "step": 90530 }, { "epoch": 57.70554493307839, "grad_norm": 0.5215052962303162, "learning_rate": 8.36449181389073e-05, "loss": 0.0156, "step": 90540 }, { "epoch": 57.7119184193754, "grad_norm": 0.6719648838043213, "learning_rate": 8.36408408364745e-05, "loss": 0.0287, "step": 90550 }, { "epoch": 57.7182919056724, "grad_norm": 0.41499289870262146, "learning_rate": 8.363676312527372e-05, "loss": 0.024, "step": 90560 }, { "epoch": 57.72466539196941, "grad_norm": 0.37071362137794495, "learning_rate": 8.363268500535452e-05, "loss": 0.0201, "step": 90570 }, { "epoch": 57.73103887826641, "grad_norm": 0.49417492747306824, "learning_rate": 8.362860647676644e-05, "loss": 0.0162, "step": 90580 }, { "epoch": 57.73741236456342, "grad_norm": 0.520778238773346, "learning_rate": 8.362452753955904e-05, "loss": 0.0211, "step": 90590 }, { "epoch": 57.74378585086042, "grad_norm": 0.3248370587825775, "learning_rate": 8.362044819378188e-05, "loss": 0.0238, "step": 90600 }, { "epoch": 57.75015933715743, "grad_norm": 0.45907577872276306, "learning_rate": 8.361636843948454e-05, "loss": 0.0139, "step": 90610 }, { "epoch": 57.75653282345443, "grad_norm": 0.39414939284324646, "learning_rate": 8.361228827671657e-05, "loss": 0.0154, "step": 90620 }, { "epoch": 57.762906309751436, "grad_norm": 0.5685986876487732, "learning_rate": 8.360820770552757e-05, "loss": 0.0256, "step": 90630 }, { "epoch": 57.76927979604844, "grad_norm": 0.4703851342201233, "learning_rate": 8.360412672596712e-05, "loss": 0.0108, "step": 90640 }, { "epoch": 57.775653282345445, "grad_norm": 0.4228808879852295, "learning_rate": 8.36000453380848e-05, "loss": 0.0126, "step": 90650 }, { "epoch": 57.782026768642446, "grad_norm": 0.29923856258392334, "learning_rate": 8.35959635419302e-05, "loss": 0.0179, "step": 90660 }, { "epoch": 57.788400254939454, "grad_norm": 0.7327203154563904, "learning_rate": 8.359188133755294e-05, "loss": 0.0289, "step": 90670 }, { "epoch": 57.794773741236455, "grad_norm": 0.4385051131248474, "learning_rate": 8.358779872500259e-05, "loss": 0.0172, "step": 90680 }, { "epoch": 57.80114722753346, "grad_norm": 0.4585539400577545, "learning_rate": 8.358371570432877e-05, "loss": 0.0179, "step": 90690 }, { "epoch": 57.80752071383046, "grad_norm": 0.5885264873504639, "learning_rate": 8.35796322755811e-05, "loss": 0.0196, "step": 90700 }, { "epoch": 57.81389420012747, "grad_norm": 0.25167813897132874, "learning_rate": 8.357554843880919e-05, "loss": 0.0182, "step": 90710 }, { "epoch": 57.82026768642447, "grad_norm": 0.45056578516960144, "learning_rate": 8.357146419406268e-05, "loss": 0.0292, "step": 90720 }, { "epoch": 57.82664117272148, "grad_norm": 0.8499084115028381, "learning_rate": 8.356737954139118e-05, "loss": 0.0327, "step": 90730 }, { "epoch": 57.83301465901848, "grad_norm": 0.6569238305091858, "learning_rate": 8.356329448084432e-05, "loss": 0.0181, "step": 90740 }, { "epoch": 57.83938814531549, "grad_norm": 0.3645811676979065, "learning_rate": 8.355920901247174e-05, "loss": 0.0199, "step": 90750 }, { "epoch": 57.84576163161249, "grad_norm": 0.45569077134132385, "learning_rate": 8.355512313632309e-05, "loss": 0.0189, "step": 90760 }, { "epoch": 57.8521351179095, "grad_norm": 0.28670087456703186, "learning_rate": 8.355103685244801e-05, "loss": 0.0197, "step": 90770 }, { "epoch": 57.8585086042065, "grad_norm": 0.24539832770824432, "learning_rate": 8.354695016089616e-05, "loss": 0.0178, "step": 90780 }, { "epoch": 57.86488209050351, "grad_norm": 0.23846940696239471, "learning_rate": 8.35428630617172e-05, "loss": 0.0234, "step": 90790 }, { "epoch": 57.87125557680051, "grad_norm": 0.519631564617157, "learning_rate": 8.353877555496076e-05, "loss": 0.0182, "step": 90800 }, { "epoch": 57.877629063097515, "grad_norm": 0.6478245854377747, "learning_rate": 8.353468764067654e-05, "loss": 0.033, "step": 90810 }, { "epoch": 57.884002549394516, "grad_norm": 0.3000773787498474, "learning_rate": 8.353059931891422e-05, "loss": 0.0236, "step": 90820 }, { "epoch": 57.890376035691524, "grad_norm": 0.2870498597621918, "learning_rate": 8.352651058972344e-05, "loss": 0.024, "step": 90830 }, { "epoch": 57.896749521988525, "grad_norm": 0.5736768841743469, "learning_rate": 8.352242145315392e-05, "loss": 0.0234, "step": 90840 }, { "epoch": 57.90312300828553, "grad_norm": 0.2978558838367462, "learning_rate": 8.351833190925531e-05, "loss": 0.0143, "step": 90850 }, { "epoch": 57.909496494582534, "grad_norm": 0.23081423342227936, "learning_rate": 8.351424195807733e-05, "loss": 0.0191, "step": 90860 }, { "epoch": 57.91586998087954, "grad_norm": 0.5447297692298889, "learning_rate": 8.351015159966966e-05, "loss": 0.0159, "step": 90870 }, { "epoch": 57.92224346717654, "grad_norm": 0.3341819643974304, "learning_rate": 8.350606083408202e-05, "loss": 0.0215, "step": 90880 }, { "epoch": 57.92861695347355, "grad_norm": 0.36266979575157166, "learning_rate": 8.35019696613641e-05, "loss": 0.0179, "step": 90890 }, { "epoch": 57.93499043977055, "grad_norm": 0.2404865026473999, "learning_rate": 8.349787808156562e-05, "loss": 0.0194, "step": 90900 }, { "epoch": 57.94136392606756, "grad_norm": 0.3146337568759918, "learning_rate": 8.34937860947363e-05, "loss": 0.0232, "step": 90910 }, { "epoch": 57.94773741236456, "grad_norm": 0.42501819133758545, "learning_rate": 8.348969370092584e-05, "loss": 0.0312, "step": 90920 }, { "epoch": 57.95411089866157, "grad_norm": 0.3568401634693146, "learning_rate": 8.348560090018399e-05, "loss": 0.0129, "step": 90930 }, { "epoch": 57.96048438495857, "grad_norm": 0.5907946825027466, "learning_rate": 8.348150769256047e-05, "loss": 0.0222, "step": 90940 }, { "epoch": 57.96685787125558, "grad_norm": 0.41706475615501404, "learning_rate": 8.347741407810503e-05, "loss": 0.026, "step": 90950 }, { "epoch": 57.97323135755258, "grad_norm": 0.39935413002967834, "learning_rate": 8.34733200568674e-05, "loss": 0.0369, "step": 90960 }, { "epoch": 57.979604843849586, "grad_norm": 0.3769091069698334, "learning_rate": 8.346922562889732e-05, "loss": 0.0251, "step": 90970 }, { "epoch": 57.98597833014659, "grad_norm": 0.38300979137420654, "learning_rate": 8.346513079424455e-05, "loss": 0.0193, "step": 90980 }, { "epoch": 57.992351816443595, "grad_norm": 0.4498318135738373, "learning_rate": 8.346103555295883e-05, "loss": 0.0197, "step": 90990 }, { "epoch": 57.998725302740596, "grad_norm": 0.455406129360199, "learning_rate": 8.345693990508996e-05, "loss": 0.0227, "step": 91000 }, { "epoch": 58.0050987890376, "grad_norm": 0.5942676067352295, "learning_rate": 8.345284385068768e-05, "loss": 0.021, "step": 91010 }, { "epoch": 58.01147227533461, "grad_norm": 0.3158631920814514, "learning_rate": 8.344874738980174e-05, "loss": 0.0244, "step": 91020 }, { "epoch": 58.01784576163161, "grad_norm": 0.713674008846283, "learning_rate": 8.344465052248196e-05, "loss": 0.027, "step": 91030 }, { "epoch": 58.02421924792862, "grad_norm": 0.3068496584892273, "learning_rate": 8.34405532487781e-05, "loss": 0.0181, "step": 91040 }, { "epoch": 58.03059273422562, "grad_norm": 0.437433123588562, "learning_rate": 8.343645556873994e-05, "loss": 0.0132, "step": 91050 }, { "epoch": 58.03696622052263, "grad_norm": 0.5695892572402954, "learning_rate": 8.343235748241727e-05, "loss": 0.0277, "step": 91060 }, { "epoch": 58.04333970681963, "grad_norm": 0.4153307378292084, "learning_rate": 8.342825898985989e-05, "loss": 0.0277, "step": 91070 }, { "epoch": 58.04971319311664, "grad_norm": 0.4828537702560425, "learning_rate": 8.34241600911176e-05, "loss": 0.0169, "step": 91080 }, { "epoch": 58.05608667941364, "grad_norm": 0.5518837571144104, "learning_rate": 8.342006078624023e-05, "loss": 0.021, "step": 91090 }, { "epoch": 58.06246016571065, "grad_norm": 0.6673458814620972, "learning_rate": 8.341596107527753e-05, "loss": 0.0238, "step": 91100 }, { "epoch": 58.06883365200765, "grad_norm": 0.3064559996128082, "learning_rate": 8.341186095827939e-05, "loss": 0.024, "step": 91110 }, { "epoch": 58.075207138304656, "grad_norm": 0.6941390037536621, "learning_rate": 8.340776043529557e-05, "loss": 0.015, "step": 91120 }, { "epoch": 58.081580624601656, "grad_norm": 0.5841116905212402, "learning_rate": 8.340365950637592e-05, "loss": 0.0171, "step": 91130 }, { "epoch": 58.087954110898664, "grad_norm": 0.48552754521369934, "learning_rate": 8.339955817157028e-05, "loss": 0.0259, "step": 91140 }, { "epoch": 58.094327597195665, "grad_norm": 0.5929617285728455, "learning_rate": 8.339545643092848e-05, "loss": 0.03, "step": 91150 }, { "epoch": 58.10070108349267, "grad_norm": 0.49149858951568604, "learning_rate": 8.339135428450034e-05, "loss": 0.0371, "step": 91160 }, { "epoch": 58.107074569789674, "grad_norm": 0.4183204174041748, "learning_rate": 8.338725173233571e-05, "loss": 0.0172, "step": 91170 }, { "epoch": 58.11344805608668, "grad_norm": 0.27595144510269165, "learning_rate": 8.338314877448447e-05, "loss": 0.0189, "step": 91180 }, { "epoch": 58.11982154238368, "grad_norm": 0.7357737421989441, "learning_rate": 8.337904541099643e-05, "loss": 0.0199, "step": 91190 }, { "epoch": 58.12619502868069, "grad_norm": 0.40764376521110535, "learning_rate": 8.337494164192149e-05, "loss": 0.0294, "step": 91200 }, { "epoch": 58.13256851497769, "grad_norm": 0.3002212345600128, "learning_rate": 8.33708374673095e-05, "loss": 0.0204, "step": 91210 }, { "epoch": 58.1389420012747, "grad_norm": 0.28194454312324524, "learning_rate": 8.336673288721031e-05, "loss": 0.0213, "step": 91220 }, { "epoch": 58.1453154875717, "grad_norm": 0.35231664776802063, "learning_rate": 8.336262790167383e-05, "loss": 0.0237, "step": 91230 }, { "epoch": 58.15168897386871, "grad_norm": 0.48193374276161194, "learning_rate": 8.335852251074992e-05, "loss": 0.0218, "step": 91240 }, { "epoch": 58.15806246016571, "grad_norm": 0.3032952547073364, "learning_rate": 8.335441671448845e-05, "loss": 0.0151, "step": 91250 }, { "epoch": 58.16443594646272, "grad_norm": 0.7593832612037659, "learning_rate": 8.335031051293933e-05, "loss": 0.0346, "step": 91260 }, { "epoch": 58.17080943275972, "grad_norm": 0.2837711572647095, "learning_rate": 8.334620390615246e-05, "loss": 0.0188, "step": 91270 }, { "epoch": 58.177182919056726, "grad_norm": 0.272061824798584, "learning_rate": 8.334209689417773e-05, "loss": 0.0192, "step": 91280 }, { "epoch": 58.18355640535373, "grad_norm": 0.4461846351623535, "learning_rate": 8.333798947706504e-05, "loss": 0.022, "step": 91290 }, { "epoch": 58.189929891650735, "grad_norm": 0.6074658036231995, "learning_rate": 8.333388165486429e-05, "loss": 0.0239, "step": 91300 }, { "epoch": 58.196303377947736, "grad_norm": 0.43046948313713074, "learning_rate": 8.332977342762541e-05, "loss": 0.0206, "step": 91310 }, { "epoch": 58.202676864244744, "grad_norm": 0.19798019528388977, "learning_rate": 8.332566479539833e-05, "loss": 0.0127, "step": 91320 }, { "epoch": 58.209050350541744, "grad_norm": 0.3855135440826416, "learning_rate": 8.332155575823294e-05, "loss": 0.0193, "step": 91330 }, { "epoch": 58.21542383683875, "grad_norm": 0.32505425810813904, "learning_rate": 8.33174463161792e-05, "loss": 0.0181, "step": 91340 }, { "epoch": 58.22179732313575, "grad_norm": 0.46706852316856384, "learning_rate": 8.331333646928701e-05, "loss": 0.0196, "step": 91350 }, { "epoch": 58.22817080943276, "grad_norm": 0.3934280574321747, "learning_rate": 8.330922621760635e-05, "loss": 0.0152, "step": 91360 }, { "epoch": 58.23454429572976, "grad_norm": 0.3870113790035248, "learning_rate": 8.330511556118713e-05, "loss": 0.0173, "step": 91370 }, { "epoch": 58.24091778202677, "grad_norm": 0.42795488238334656, "learning_rate": 8.330100450007931e-05, "loss": 0.0246, "step": 91380 }, { "epoch": 58.24729126832377, "grad_norm": 0.379820853471756, "learning_rate": 8.329689303433285e-05, "loss": 0.0201, "step": 91390 }, { "epoch": 58.25366475462078, "grad_norm": 0.6734418869018555, "learning_rate": 8.32927811639977e-05, "loss": 0.0159, "step": 91400 }, { "epoch": 58.26003824091778, "grad_norm": 0.3046272397041321, "learning_rate": 8.328866888912384e-05, "loss": 0.0191, "step": 91410 }, { "epoch": 58.26641172721479, "grad_norm": 0.2194511443376541, "learning_rate": 8.32845562097612e-05, "loss": 0.0141, "step": 91420 }, { "epoch": 58.27278521351179, "grad_norm": 0.516089916229248, "learning_rate": 8.32804431259598e-05, "loss": 0.0178, "step": 91430 }, { "epoch": 58.2791586998088, "grad_norm": 0.49367237091064453, "learning_rate": 8.327632963776956e-05, "loss": 0.0154, "step": 91440 }, { "epoch": 58.2855321861058, "grad_norm": 0.23633047938346863, "learning_rate": 8.327221574524053e-05, "loss": 0.0247, "step": 91450 }, { "epoch": 58.291905672402805, "grad_norm": 0.46073925495147705, "learning_rate": 8.326810144842265e-05, "loss": 0.0183, "step": 91460 }, { "epoch": 58.298279158699806, "grad_norm": 0.3954058587551117, "learning_rate": 8.326398674736592e-05, "loss": 0.0147, "step": 91470 }, { "epoch": 58.304652644996814, "grad_norm": 0.8969352841377258, "learning_rate": 8.325987164212036e-05, "loss": 0.0198, "step": 91480 }, { "epoch": 58.311026131293815, "grad_norm": 0.38300052285194397, "learning_rate": 8.325575613273595e-05, "loss": 0.0166, "step": 91490 }, { "epoch": 58.31739961759082, "grad_norm": 0.3057568073272705, "learning_rate": 8.32516402192627e-05, "loss": 0.0188, "step": 91500 }, { "epoch": 58.323773103887824, "grad_norm": 0.1950339376926422, "learning_rate": 8.324752390175063e-05, "loss": 0.0263, "step": 91510 }, { "epoch": 58.33014659018483, "grad_norm": 0.5426521897315979, "learning_rate": 8.324340718024977e-05, "loss": 0.0177, "step": 91520 }, { "epoch": 58.33652007648183, "grad_norm": 0.37104398012161255, "learning_rate": 8.32392900548101e-05, "loss": 0.0202, "step": 91530 }, { "epoch": 58.34289356277884, "grad_norm": 0.3837517499923706, "learning_rate": 8.323517252548169e-05, "loss": 0.0184, "step": 91540 }, { "epoch": 58.34926704907584, "grad_norm": 0.4697130620479584, "learning_rate": 8.323105459231455e-05, "loss": 0.0128, "step": 91550 }, { "epoch": 58.35564053537285, "grad_norm": 0.46409332752227783, "learning_rate": 8.322693625535872e-05, "loss": 0.0206, "step": 91560 }, { "epoch": 58.36201402166985, "grad_norm": 0.30239129066467285, "learning_rate": 8.322281751466424e-05, "loss": 0.0316, "step": 91570 }, { "epoch": 58.36838750796686, "grad_norm": 0.6283408403396606, "learning_rate": 8.321869837028116e-05, "loss": 0.0232, "step": 91580 }, { "epoch": 58.37476099426386, "grad_norm": 0.601788341999054, "learning_rate": 8.321457882225952e-05, "loss": 0.0167, "step": 91590 }, { "epoch": 58.38113448056087, "grad_norm": 0.5789155960083008, "learning_rate": 8.32104588706494e-05, "loss": 0.023, "step": 91600 }, { "epoch": 58.38750796685787, "grad_norm": 0.4448530673980713, "learning_rate": 8.320633851550086e-05, "loss": 0.0204, "step": 91610 }, { "epoch": 58.393881453154876, "grad_norm": 0.6346665620803833, "learning_rate": 8.320221775686394e-05, "loss": 0.0155, "step": 91620 }, { "epoch": 58.400254939451884, "grad_norm": 0.5220521092414856, "learning_rate": 8.319809659478873e-05, "loss": 0.0264, "step": 91630 }, { "epoch": 58.406628425748885, "grad_norm": 0.2869974970817566, "learning_rate": 8.31939750293253e-05, "loss": 0.0216, "step": 91640 }, { "epoch": 58.41300191204589, "grad_norm": 0.3271142244338989, "learning_rate": 8.318985306052374e-05, "loss": 0.0255, "step": 91650 }, { "epoch": 58.41937539834289, "grad_norm": 0.3671657145023346, "learning_rate": 8.318573068843413e-05, "loss": 0.0257, "step": 91660 }, { "epoch": 58.4257488846399, "grad_norm": 0.2706374228000641, "learning_rate": 8.318160791310653e-05, "loss": 0.024, "step": 91670 }, { "epoch": 58.4321223709369, "grad_norm": 0.4009813666343689, "learning_rate": 8.317748473459111e-05, "loss": 0.0206, "step": 91680 }, { "epoch": 58.43849585723391, "grad_norm": 0.4837583899497986, "learning_rate": 8.31733611529379e-05, "loss": 0.0157, "step": 91690 }, { "epoch": 58.44486934353091, "grad_norm": 0.42084696888923645, "learning_rate": 8.316923716819704e-05, "loss": 0.0163, "step": 91700 }, { "epoch": 58.45124282982792, "grad_norm": 0.21271800994873047, "learning_rate": 8.316511278041862e-05, "loss": 0.0201, "step": 91710 }, { "epoch": 58.45761631612492, "grad_norm": 0.5939106941223145, "learning_rate": 8.316098798965279e-05, "loss": 0.0192, "step": 91720 }, { "epoch": 58.46398980242193, "grad_norm": 0.26054078340530396, "learning_rate": 8.315686279594964e-05, "loss": 0.0158, "step": 91730 }, { "epoch": 58.47036328871893, "grad_norm": 0.664340615272522, "learning_rate": 8.315273719935928e-05, "loss": 0.0216, "step": 91740 }, { "epoch": 58.47673677501594, "grad_norm": 0.6056423783302307, "learning_rate": 8.314861119993186e-05, "loss": 0.0214, "step": 91750 }, { "epoch": 58.48311026131294, "grad_norm": 0.42953068017959595, "learning_rate": 8.314448479771754e-05, "loss": 0.0194, "step": 91760 }, { "epoch": 58.489483747609945, "grad_norm": 0.5340674519538879, "learning_rate": 8.314035799276643e-05, "loss": 0.0172, "step": 91770 }, { "epoch": 58.495857233906946, "grad_norm": 0.26035353541374207, "learning_rate": 8.313623078512869e-05, "loss": 0.0131, "step": 91780 }, { "epoch": 58.502230720203954, "grad_norm": 0.4818899631500244, "learning_rate": 8.313210317485443e-05, "loss": 0.0184, "step": 91790 }, { "epoch": 58.508604206500955, "grad_norm": 0.586942195892334, "learning_rate": 8.312797516199387e-05, "loss": 0.0214, "step": 91800 }, { "epoch": 58.51497769279796, "grad_norm": 0.5176035165786743, "learning_rate": 8.31238467465971e-05, "loss": 0.0247, "step": 91810 }, { "epoch": 58.521351179094964, "grad_norm": 0.60249924659729, "learning_rate": 8.311971792871435e-05, "loss": 0.0157, "step": 91820 }, { "epoch": 58.52772466539197, "grad_norm": 0.45866042375564575, "learning_rate": 8.311558870839573e-05, "loss": 0.0284, "step": 91830 }, { "epoch": 58.53409815168897, "grad_norm": 0.5293552875518799, "learning_rate": 8.311145908569147e-05, "loss": 0.0156, "step": 91840 }, { "epoch": 58.54047163798598, "grad_norm": 0.4334186911582947, "learning_rate": 8.310732906065168e-05, "loss": 0.023, "step": 91850 }, { "epoch": 58.54684512428298, "grad_norm": 0.2653294801712036, "learning_rate": 8.310319863332662e-05, "loss": 0.0179, "step": 91860 }, { "epoch": 58.55321861057999, "grad_norm": 0.34615960717201233, "learning_rate": 8.309906780376642e-05, "loss": 0.0142, "step": 91870 }, { "epoch": 58.55959209687699, "grad_norm": 0.3851594924926758, "learning_rate": 8.309493657202131e-05, "loss": 0.0192, "step": 91880 }, { "epoch": 58.565965583174, "grad_norm": 0.3321509063243866, "learning_rate": 8.309080493814147e-05, "loss": 0.0181, "step": 91890 }, { "epoch": 58.572339069471, "grad_norm": 0.270266592502594, "learning_rate": 8.30866729021771e-05, "loss": 0.0215, "step": 91900 }, { "epoch": 58.57871255576801, "grad_norm": 0.42378073930740356, "learning_rate": 8.308254046417843e-05, "loss": 0.0215, "step": 91910 }, { "epoch": 58.58508604206501, "grad_norm": 0.6370362639427185, "learning_rate": 8.307840762419563e-05, "loss": 0.0271, "step": 91920 }, { "epoch": 58.591459528362016, "grad_norm": 0.5408931970596313, "learning_rate": 8.307427438227897e-05, "loss": 0.0211, "step": 91930 }, { "epoch": 58.59783301465902, "grad_norm": 0.35163483023643494, "learning_rate": 8.307014073847864e-05, "loss": 0.0255, "step": 91940 }, { "epoch": 58.604206500956025, "grad_norm": 0.2837541401386261, "learning_rate": 8.306600669284488e-05, "loss": 0.0174, "step": 91950 }, { "epoch": 58.610579987253026, "grad_norm": 0.28027379512786865, "learning_rate": 8.306187224542791e-05, "loss": 0.0138, "step": 91960 }, { "epoch": 58.61695347355003, "grad_norm": 0.35531988739967346, "learning_rate": 8.3057737396278e-05, "loss": 0.0257, "step": 91970 }, { "epoch": 58.623326959847034, "grad_norm": 0.47617557644844055, "learning_rate": 8.305360214544535e-05, "loss": 0.0162, "step": 91980 }, { "epoch": 58.62970044614404, "grad_norm": 0.41942328214645386, "learning_rate": 8.304946649298023e-05, "loss": 0.0132, "step": 91990 }, { "epoch": 58.63607393244104, "grad_norm": 0.33593735098838806, "learning_rate": 8.304533043893288e-05, "loss": 0.027, "step": 92000 }, { "epoch": 58.64244741873805, "grad_norm": 0.1853705793619156, "learning_rate": 8.304119398335357e-05, "loss": 0.0176, "step": 92010 }, { "epoch": 58.64882090503505, "grad_norm": 0.750265896320343, "learning_rate": 8.303705712629257e-05, "loss": 0.0208, "step": 92020 }, { "epoch": 58.65519439133206, "grad_norm": 0.4242609739303589, "learning_rate": 8.303291986780014e-05, "loss": 0.0176, "step": 92030 }, { "epoch": 58.66156787762906, "grad_norm": 0.4339333176612854, "learning_rate": 8.302878220792651e-05, "loss": 0.0248, "step": 92040 }, { "epoch": 58.66794136392607, "grad_norm": 0.38782384991645813, "learning_rate": 8.302464414672202e-05, "loss": 0.0296, "step": 92050 }, { "epoch": 58.67431485022307, "grad_norm": 0.648065447807312, "learning_rate": 8.302050568423692e-05, "loss": 0.02, "step": 92060 }, { "epoch": 58.68068833652008, "grad_norm": 0.532376766204834, "learning_rate": 8.301636682052149e-05, "loss": 0.0225, "step": 92070 }, { "epoch": 58.68706182281708, "grad_norm": 0.32371726632118225, "learning_rate": 8.301222755562603e-05, "loss": 0.0153, "step": 92080 }, { "epoch": 58.693435309114086, "grad_norm": 0.6140322089195251, "learning_rate": 8.300808788960083e-05, "loss": 0.0237, "step": 92090 }, { "epoch": 58.69980879541109, "grad_norm": 0.2801460921764374, "learning_rate": 8.300394782249621e-05, "loss": 0.0174, "step": 92100 }, { "epoch": 58.706182281708095, "grad_norm": 0.42450451850891113, "learning_rate": 8.299980735436248e-05, "loss": 0.0219, "step": 92110 }, { "epoch": 58.712555768005096, "grad_norm": 0.607071578502655, "learning_rate": 8.299566648524991e-05, "loss": 0.0201, "step": 92120 }, { "epoch": 58.718929254302104, "grad_norm": 0.5148131251335144, "learning_rate": 8.299152521520884e-05, "loss": 0.0262, "step": 92130 }, { "epoch": 58.725302740599105, "grad_norm": 0.3308785557746887, "learning_rate": 8.29873835442896e-05, "loss": 0.0263, "step": 92140 }, { "epoch": 58.73167622689611, "grad_norm": 0.7380141019821167, "learning_rate": 8.29832414725425e-05, "loss": 0.0126, "step": 92150 }, { "epoch": 58.738049713193114, "grad_norm": 0.3335932195186615, "learning_rate": 8.297909900001787e-05, "loss": 0.0165, "step": 92160 }, { "epoch": 58.74442319949012, "grad_norm": 0.27494731545448303, "learning_rate": 8.297495612676604e-05, "loss": 0.0314, "step": 92170 }, { "epoch": 58.75079668578712, "grad_norm": 0.3603711724281311, "learning_rate": 8.297081285283738e-05, "loss": 0.0217, "step": 92180 }, { "epoch": 58.75717017208413, "grad_norm": 0.3022996485233307, "learning_rate": 8.296666917828222e-05, "loss": 0.0264, "step": 92190 }, { "epoch": 58.76354365838113, "grad_norm": 0.3028657138347626, "learning_rate": 8.296252510315089e-05, "loss": 0.0246, "step": 92200 }, { "epoch": 58.76991714467814, "grad_norm": 0.7696290612220764, "learning_rate": 8.295838062749374e-05, "loss": 0.0245, "step": 92210 }, { "epoch": 58.77629063097514, "grad_norm": 0.3597107231616974, "learning_rate": 8.295423575136117e-05, "loss": 0.0446, "step": 92220 }, { "epoch": 58.78266411727215, "grad_norm": 0.7335372567176819, "learning_rate": 8.295009047480354e-05, "loss": 0.018, "step": 92230 }, { "epoch": 58.78903760356915, "grad_norm": 0.31487876176834106, "learning_rate": 8.294594479787116e-05, "loss": 0.0183, "step": 92240 }, { "epoch": 58.79541108986616, "grad_norm": 0.47309112548828125, "learning_rate": 8.294179872061447e-05, "loss": 0.0318, "step": 92250 }, { "epoch": 58.80178457616316, "grad_norm": 0.29446467757225037, "learning_rate": 8.293765224308381e-05, "loss": 0.0251, "step": 92260 }, { "epoch": 58.808158062460166, "grad_norm": 0.2540798783302307, "learning_rate": 8.29335053653296e-05, "loss": 0.0227, "step": 92270 }, { "epoch": 58.814531548757174, "grad_norm": 0.35325634479522705, "learning_rate": 8.292935808740218e-05, "loss": 0.0203, "step": 92280 }, { "epoch": 58.820905035054174, "grad_norm": 0.6192834973335266, "learning_rate": 8.292521040935197e-05, "loss": 0.0339, "step": 92290 }, { "epoch": 58.82727852135118, "grad_norm": 0.6190416812896729, "learning_rate": 8.292106233122939e-05, "loss": 0.0203, "step": 92300 }, { "epoch": 58.83365200764818, "grad_norm": 0.6363612413406372, "learning_rate": 8.29169138530848e-05, "loss": 0.0348, "step": 92310 }, { "epoch": 58.84002549394519, "grad_norm": 0.444632351398468, "learning_rate": 8.291276497496863e-05, "loss": 0.0178, "step": 92320 }, { "epoch": 58.84639898024219, "grad_norm": 0.779443085193634, "learning_rate": 8.290861569693129e-05, "loss": 0.028, "step": 92330 }, { "epoch": 58.8527724665392, "grad_norm": 0.6048526167869568, "learning_rate": 8.29044660190232e-05, "loss": 0.0277, "step": 92340 }, { "epoch": 58.8591459528362, "grad_norm": 0.6131560802459717, "learning_rate": 8.29003159412948e-05, "loss": 0.0212, "step": 92350 }, { "epoch": 58.86551943913321, "grad_norm": 0.5680294632911682, "learning_rate": 8.289616546379646e-05, "loss": 0.0187, "step": 92360 }, { "epoch": 58.87189292543021, "grad_norm": 0.5247848033905029, "learning_rate": 8.289201458657868e-05, "loss": 0.0245, "step": 92370 }, { "epoch": 58.87826641172722, "grad_norm": 0.48301443457603455, "learning_rate": 8.288786330969185e-05, "loss": 0.0238, "step": 92380 }, { "epoch": 58.88463989802422, "grad_norm": 0.6292096376419067, "learning_rate": 8.288371163318645e-05, "loss": 0.0221, "step": 92390 }, { "epoch": 58.89101338432123, "grad_norm": 0.32192355394363403, "learning_rate": 8.287955955711289e-05, "loss": 0.0156, "step": 92400 }, { "epoch": 58.89738687061823, "grad_norm": 0.8315854668617249, "learning_rate": 8.287540708152164e-05, "loss": 0.0359, "step": 92410 }, { "epoch": 58.903760356915235, "grad_norm": 0.6144301891326904, "learning_rate": 8.287125420646316e-05, "loss": 0.0268, "step": 92420 }, { "epoch": 58.910133843212236, "grad_norm": 0.4382031559944153, "learning_rate": 8.286710093198791e-05, "loss": 0.0191, "step": 92430 }, { "epoch": 58.916507329509244, "grad_norm": 0.3596999943256378, "learning_rate": 8.286294725814635e-05, "loss": 0.0235, "step": 92440 }, { "epoch": 58.922880815806245, "grad_norm": 0.4521946310997009, "learning_rate": 8.285879318498893e-05, "loss": 0.0202, "step": 92450 }, { "epoch": 58.92925430210325, "grad_norm": 0.20184192061424255, "learning_rate": 8.285463871256618e-05, "loss": 0.0174, "step": 92460 }, { "epoch": 58.935627788400254, "grad_norm": 0.2752222418785095, "learning_rate": 8.285048384092853e-05, "loss": 0.0261, "step": 92470 }, { "epoch": 58.94200127469726, "grad_norm": 0.38707879185676575, "learning_rate": 8.28463285701265e-05, "loss": 0.0176, "step": 92480 }, { "epoch": 58.94837476099426, "grad_norm": 1.079452633857727, "learning_rate": 8.284217290021055e-05, "loss": 0.0302, "step": 92490 }, { "epoch": 58.95474824729127, "grad_norm": 0.48543980717658997, "learning_rate": 8.28380168312312e-05, "loss": 0.0213, "step": 92500 }, { "epoch": 58.96112173358827, "grad_norm": 0.4242631494998932, "learning_rate": 8.283386036323894e-05, "loss": 0.018, "step": 92510 }, { "epoch": 58.96749521988528, "grad_norm": 0.3445139229297638, "learning_rate": 8.282970349628427e-05, "loss": 0.0142, "step": 92520 }, { "epoch": 58.97386870618228, "grad_norm": 0.1800592988729477, "learning_rate": 8.282554623041773e-05, "loss": 0.0181, "step": 92530 }, { "epoch": 58.98024219247929, "grad_norm": 0.5218713879585266, "learning_rate": 8.282138856568978e-05, "loss": 0.0254, "step": 92540 }, { "epoch": 58.98661567877629, "grad_norm": 0.5554972887039185, "learning_rate": 8.2817230502151e-05, "loss": 0.0182, "step": 92550 }, { "epoch": 58.9929891650733, "grad_norm": 0.3502090275287628, "learning_rate": 8.281307203985186e-05, "loss": 0.0198, "step": 92560 }, { "epoch": 58.9993626513703, "grad_norm": 0.34401580691337585, "learning_rate": 8.280891317884291e-05, "loss": 0.0207, "step": 92570 }, { "epoch": 59.005736137667306, "grad_norm": 0.3515547215938568, "learning_rate": 8.28047539191747e-05, "loss": 0.0266, "step": 92580 }, { "epoch": 59.01210962396431, "grad_norm": 0.7415618896484375, "learning_rate": 8.280059426089777e-05, "loss": 0.0276, "step": 92590 }, { "epoch": 59.018483110261315, "grad_norm": 0.4392441511154175, "learning_rate": 8.279643420406265e-05, "loss": 0.0197, "step": 92600 }, { "epoch": 59.024856596558315, "grad_norm": 0.4125823676586151, "learning_rate": 8.279227374871988e-05, "loss": 0.0219, "step": 92610 }, { "epoch": 59.03123008285532, "grad_norm": 0.2168254852294922, "learning_rate": 8.278811289492001e-05, "loss": 0.0163, "step": 92620 }, { "epoch": 59.037603569152324, "grad_norm": 0.5477132201194763, "learning_rate": 8.278395164271362e-05, "loss": 0.0287, "step": 92630 }, { "epoch": 59.04397705544933, "grad_norm": 0.5849064588546753, "learning_rate": 8.277978999215127e-05, "loss": 0.0297, "step": 92640 }, { "epoch": 59.05035054174633, "grad_norm": 0.34799671173095703, "learning_rate": 8.277562794328352e-05, "loss": 0.0212, "step": 92650 }, { "epoch": 59.05672402804334, "grad_norm": 0.21220850944519043, "learning_rate": 8.277146549616094e-05, "loss": 0.0225, "step": 92660 }, { "epoch": 59.06309751434034, "grad_norm": 0.34040048718452454, "learning_rate": 8.276730265083412e-05, "loss": 0.0235, "step": 92670 }, { "epoch": 59.06947100063735, "grad_norm": 0.4934118688106537, "learning_rate": 8.276313940735363e-05, "loss": 0.0213, "step": 92680 }, { "epoch": 59.07584448693435, "grad_norm": 0.6165369749069214, "learning_rate": 8.275897576577007e-05, "loss": 0.0237, "step": 92690 }, { "epoch": 59.08221797323136, "grad_norm": 0.5506377220153809, "learning_rate": 8.275481172613402e-05, "loss": 0.0158, "step": 92700 }, { "epoch": 59.08859145952836, "grad_norm": 0.4537414312362671, "learning_rate": 8.275064728849608e-05, "loss": 0.0205, "step": 92710 }, { "epoch": 59.09496494582537, "grad_norm": 0.38948196172714233, "learning_rate": 8.274648245290686e-05, "loss": 0.0168, "step": 92720 }, { "epoch": 59.10133843212237, "grad_norm": 0.26395899057388306, "learning_rate": 8.274231721941695e-05, "loss": 0.0272, "step": 92730 }, { "epoch": 59.107711918419376, "grad_norm": 0.41868460178375244, "learning_rate": 8.273815158807699e-05, "loss": 0.0225, "step": 92740 }, { "epoch": 59.11408540471638, "grad_norm": 0.6008462905883789, "learning_rate": 8.273398555893756e-05, "loss": 0.0235, "step": 92750 }, { "epoch": 59.120458891013385, "grad_norm": 0.5641804337501526, "learning_rate": 8.272981913204931e-05, "loss": 0.0249, "step": 92760 }, { "epoch": 59.126832377310386, "grad_norm": 0.43208417296409607, "learning_rate": 8.272565230746284e-05, "loss": 0.0145, "step": 92770 }, { "epoch": 59.133205863607394, "grad_norm": 0.24256716668605804, "learning_rate": 8.272148508522882e-05, "loss": 0.0112, "step": 92780 }, { "epoch": 59.139579349904395, "grad_norm": 0.5895251035690308, "learning_rate": 8.271731746539784e-05, "loss": 0.0206, "step": 92790 }, { "epoch": 59.1459528362014, "grad_norm": 0.46897920966148376, "learning_rate": 8.271314944802057e-05, "loss": 0.0236, "step": 92800 }, { "epoch": 59.1523263224984, "grad_norm": 0.47891977429389954, "learning_rate": 8.270898103314764e-05, "loss": 0.0237, "step": 92810 }, { "epoch": 59.15869980879541, "grad_norm": 0.5131921768188477, "learning_rate": 8.270481222082974e-05, "loss": 0.0201, "step": 92820 }, { "epoch": 59.16507329509241, "grad_norm": 0.2485666424036026, "learning_rate": 8.270064301111746e-05, "loss": 0.0203, "step": 92830 }, { "epoch": 59.17144678138942, "grad_norm": 0.31107446551322937, "learning_rate": 8.26964734040615e-05, "loss": 0.0275, "step": 92840 }, { "epoch": 59.17782026768642, "grad_norm": 0.8494387865066528, "learning_rate": 8.269230339971253e-05, "loss": 0.0294, "step": 92850 }, { "epoch": 59.18419375398343, "grad_norm": 0.30487751960754395, "learning_rate": 8.268813299812119e-05, "loss": 0.0295, "step": 92860 }, { "epoch": 59.19056724028043, "grad_norm": 0.46762609481811523, "learning_rate": 8.268396219933819e-05, "loss": 0.0214, "step": 92870 }, { "epoch": 59.19694072657744, "grad_norm": 0.33592015504837036, "learning_rate": 8.267979100341418e-05, "loss": 0.0236, "step": 92880 }, { "epoch": 59.203314212874446, "grad_norm": 0.41084024310112, "learning_rate": 8.267561941039985e-05, "loss": 0.0275, "step": 92890 }, { "epoch": 59.20968769917145, "grad_norm": 0.5496814250946045, "learning_rate": 8.26714474203459e-05, "loss": 0.0249, "step": 92900 }, { "epoch": 59.216061185468455, "grad_norm": 0.4792845845222473, "learning_rate": 8.266727503330302e-05, "loss": 0.0409, "step": 92910 }, { "epoch": 59.222434671765456, "grad_norm": 0.23644034564495087, "learning_rate": 8.26631022493219e-05, "loss": 0.0327, "step": 92920 }, { "epoch": 59.22880815806246, "grad_norm": 0.5033069252967834, "learning_rate": 8.265892906845325e-05, "loss": 0.0254, "step": 92930 }, { "epoch": 59.235181644359464, "grad_norm": 0.4442286491394043, "learning_rate": 8.265475549074777e-05, "loss": 0.03, "step": 92940 }, { "epoch": 59.24155513065647, "grad_norm": 0.3033333420753479, "learning_rate": 8.265058151625617e-05, "loss": 0.0321, "step": 92950 }, { "epoch": 59.24792861695347, "grad_norm": 0.40133604407310486, "learning_rate": 8.26464071450292e-05, "loss": 0.014, "step": 92960 }, { "epoch": 59.25430210325048, "grad_norm": 0.37757736444473267, "learning_rate": 8.264223237711756e-05, "loss": 0.0168, "step": 92970 }, { "epoch": 59.26067558954748, "grad_norm": 0.35297250747680664, "learning_rate": 8.263805721257198e-05, "loss": 0.022, "step": 92980 }, { "epoch": 59.26704907584449, "grad_norm": 0.3134089708328247, "learning_rate": 8.263388165144317e-05, "loss": 0.0258, "step": 92990 }, { "epoch": 59.27342256214149, "grad_norm": 0.4771571755409241, "learning_rate": 8.26297056937819e-05, "loss": 0.0268, "step": 93000 }, { "epoch": 59.2797960484385, "grad_norm": 0.6517699956893921, "learning_rate": 8.26255293396389e-05, "loss": 0.0221, "step": 93010 }, { "epoch": 59.2861695347355, "grad_norm": 0.6683975458145142, "learning_rate": 8.26213525890649e-05, "loss": 0.0334, "step": 93020 }, { "epoch": 59.29254302103251, "grad_norm": 0.599269688129425, "learning_rate": 8.261717544211067e-05, "loss": 0.021, "step": 93030 }, { "epoch": 59.29891650732951, "grad_norm": 0.32951679825782776, "learning_rate": 8.261299789882697e-05, "loss": 0.0149, "step": 93040 }, { "epoch": 59.305289993626516, "grad_norm": 0.4724511206150055, "learning_rate": 8.260881995926455e-05, "loss": 0.0319, "step": 93050 }, { "epoch": 59.31166347992352, "grad_norm": 0.7388535737991333, "learning_rate": 8.260464162347418e-05, "loss": 0.0304, "step": 93060 }, { "epoch": 59.318036966220525, "grad_norm": 0.37618768215179443, "learning_rate": 8.260046289150662e-05, "loss": 0.0234, "step": 93070 }, { "epoch": 59.324410452517526, "grad_norm": 0.6019701361656189, "learning_rate": 8.259628376341269e-05, "loss": 0.0216, "step": 93080 }, { "epoch": 59.330783938814534, "grad_norm": 0.8116775751113892, "learning_rate": 8.25921042392431e-05, "loss": 0.0241, "step": 93090 }, { "epoch": 59.337157425111535, "grad_norm": 0.2904490530490875, "learning_rate": 8.25879243190487e-05, "loss": 0.017, "step": 93100 }, { "epoch": 59.34353091140854, "grad_norm": 0.298493891954422, "learning_rate": 8.258374400288022e-05, "loss": 0.0197, "step": 93110 }, { "epoch": 59.349904397705544, "grad_norm": 0.5321019887924194, "learning_rate": 8.257956329078851e-05, "loss": 0.0321, "step": 93120 }, { "epoch": 59.35627788400255, "grad_norm": 0.37851524353027344, "learning_rate": 8.257538218282433e-05, "loss": 0.0158, "step": 93130 }, { "epoch": 59.36265137029955, "grad_norm": 0.4570450484752655, "learning_rate": 8.257120067903852e-05, "loss": 0.0284, "step": 93140 }, { "epoch": 59.36902485659656, "grad_norm": 0.44990336894989014, "learning_rate": 8.256701877948185e-05, "loss": 0.0339, "step": 93150 }, { "epoch": 59.37539834289356, "grad_norm": 0.621115505695343, "learning_rate": 8.256283648420516e-05, "loss": 0.0234, "step": 93160 }, { "epoch": 59.38177182919057, "grad_norm": 0.5486621856689453, "learning_rate": 8.255865379325927e-05, "loss": 0.0177, "step": 93170 }, { "epoch": 59.38814531548757, "grad_norm": 0.3188325762748718, "learning_rate": 8.2554470706695e-05, "loss": 0.0238, "step": 93180 }, { "epoch": 59.39451880178458, "grad_norm": 0.4178983271121979, "learning_rate": 8.255028722456314e-05, "loss": 0.0197, "step": 93190 }, { "epoch": 59.40089228808158, "grad_norm": 0.3587867021560669, "learning_rate": 8.25461033469146e-05, "loss": 0.0194, "step": 93200 }, { "epoch": 59.40726577437859, "grad_norm": 0.35675036907196045, "learning_rate": 8.254191907380016e-05, "loss": 0.0261, "step": 93210 }, { "epoch": 59.41363926067559, "grad_norm": 0.49231988191604614, "learning_rate": 8.253773440527066e-05, "loss": 0.0198, "step": 93220 }, { "epoch": 59.420012746972596, "grad_norm": 0.6578166484832764, "learning_rate": 8.2533549341377e-05, "loss": 0.0267, "step": 93230 }, { "epoch": 59.4263862332696, "grad_norm": 0.5459911823272705, "learning_rate": 8.252936388216998e-05, "loss": 0.0156, "step": 93240 }, { "epoch": 59.432759719566604, "grad_norm": 0.5205587148666382, "learning_rate": 8.252517802770048e-05, "loss": 0.0268, "step": 93250 }, { "epoch": 59.439133205863605, "grad_norm": 0.1915075182914734, "learning_rate": 8.252099177801935e-05, "loss": 0.0154, "step": 93260 }, { "epoch": 59.44550669216061, "grad_norm": 0.2727292478084564, "learning_rate": 8.251680513317746e-05, "loss": 0.0233, "step": 93270 }, { "epoch": 59.451880178457614, "grad_norm": 0.3489912450313568, "learning_rate": 8.25126180932257e-05, "loss": 0.0237, "step": 93280 }, { "epoch": 59.45825366475462, "grad_norm": 0.7254497408866882, "learning_rate": 8.250843065821494e-05, "loss": 0.0192, "step": 93290 }, { "epoch": 59.46462715105162, "grad_norm": 0.4784906208515167, "learning_rate": 8.250424282819605e-05, "loss": 0.0166, "step": 93300 }, { "epoch": 59.47100063734863, "grad_norm": 0.4103676974773407, "learning_rate": 8.25000546032199e-05, "loss": 0.0212, "step": 93310 }, { "epoch": 59.47737412364563, "grad_norm": 0.46184688806533813, "learning_rate": 8.249586598333742e-05, "loss": 0.0295, "step": 93320 }, { "epoch": 59.48374760994264, "grad_norm": 0.28949835896492004, "learning_rate": 8.24916769685995e-05, "loss": 0.0208, "step": 93330 }, { "epoch": 59.49012109623964, "grad_norm": 0.3224061131477356, "learning_rate": 8.2487487559057e-05, "loss": 0.0206, "step": 93340 }, { "epoch": 59.49649458253665, "grad_norm": 0.24052144587039948, "learning_rate": 8.248329775476087e-05, "loss": 0.0213, "step": 93350 }, { "epoch": 59.50286806883365, "grad_norm": 0.45123717188835144, "learning_rate": 8.2479107555762e-05, "loss": 0.0211, "step": 93360 }, { "epoch": 59.50924155513066, "grad_norm": 0.2407354861497879, "learning_rate": 8.247491696211129e-05, "loss": 0.0134, "step": 93370 }, { "epoch": 59.51561504142766, "grad_norm": 0.5685559511184692, "learning_rate": 8.24707259738597e-05, "loss": 0.0304, "step": 93380 }, { "epoch": 59.521988527724666, "grad_norm": 0.5383599996566772, "learning_rate": 8.246653459105813e-05, "loss": 0.0156, "step": 93390 }, { "epoch": 59.52836201402167, "grad_norm": 0.6316070556640625, "learning_rate": 8.246234281375751e-05, "loss": 0.0185, "step": 93400 }, { "epoch": 59.534735500318675, "grad_norm": 0.5343974828720093, "learning_rate": 8.245815064200877e-05, "loss": 0.022, "step": 93410 }, { "epoch": 59.541108986615676, "grad_norm": 0.6939537525177002, "learning_rate": 8.245395807586284e-05, "loss": 0.0265, "step": 93420 }, { "epoch": 59.547482472912684, "grad_norm": 0.3878512978553772, "learning_rate": 8.24497651153707e-05, "loss": 0.0145, "step": 93430 }, { "epoch": 59.553855959209685, "grad_norm": 0.33297035098075867, "learning_rate": 8.244557176058327e-05, "loss": 0.0177, "step": 93440 }, { "epoch": 59.56022944550669, "grad_norm": 0.3212027847766876, "learning_rate": 8.244137801155151e-05, "loss": 0.0156, "step": 93450 }, { "epoch": 59.56660293180369, "grad_norm": 0.28421279788017273, "learning_rate": 8.243718386832638e-05, "loss": 0.0104, "step": 93460 }, { "epoch": 59.5729764181007, "grad_norm": 0.34876295924186707, "learning_rate": 8.243298933095883e-05, "loss": 0.0198, "step": 93470 }, { "epoch": 59.5793499043977, "grad_norm": 0.439009428024292, "learning_rate": 8.242879439949984e-05, "loss": 0.0204, "step": 93480 }, { "epoch": 59.58572339069471, "grad_norm": 0.37273940443992615, "learning_rate": 8.242459907400039e-05, "loss": 0.0174, "step": 93490 }, { "epoch": 59.59209687699171, "grad_norm": 0.5547595620155334, "learning_rate": 8.242040335451144e-05, "loss": 0.0185, "step": 93500 }, { "epoch": 59.59847036328872, "grad_norm": 0.5012189149856567, "learning_rate": 8.241620724108399e-05, "loss": 0.0236, "step": 93510 }, { "epoch": 59.60484384958572, "grad_norm": 0.24134746193885803, "learning_rate": 8.2412010733769e-05, "loss": 0.0225, "step": 93520 }, { "epoch": 59.61121733588273, "grad_norm": 0.27147793769836426, "learning_rate": 8.240781383261749e-05, "loss": 0.0161, "step": 93530 }, { "epoch": 59.617590822179736, "grad_norm": 0.3964003026485443, "learning_rate": 8.240361653768042e-05, "loss": 0.0201, "step": 93540 }, { "epoch": 59.62396430847674, "grad_norm": 0.4895612597465515, "learning_rate": 8.239941884900884e-05, "loss": 0.0275, "step": 93550 }, { "epoch": 59.630337794773745, "grad_norm": 0.6507510542869568, "learning_rate": 8.239522076665374e-05, "loss": 0.0217, "step": 93560 }, { "epoch": 59.636711281070745, "grad_norm": 0.3782055675983429, "learning_rate": 8.23910222906661e-05, "loss": 0.0147, "step": 93570 }, { "epoch": 59.64308476736775, "grad_norm": 0.5713593363761902, "learning_rate": 8.238682342109695e-05, "loss": 0.0446, "step": 93580 }, { "epoch": 59.649458253664754, "grad_norm": 0.27633485198020935, "learning_rate": 8.238262415799734e-05, "loss": 0.0302, "step": 93590 }, { "epoch": 59.65583173996176, "grad_norm": 0.42386385798454285, "learning_rate": 8.237842450141826e-05, "loss": 0.0146, "step": 93600 }, { "epoch": 59.66220522625876, "grad_norm": 0.42912089824676514, "learning_rate": 8.237422445141075e-05, "loss": 0.0241, "step": 93610 }, { "epoch": 59.66857871255577, "grad_norm": 0.42248407006263733, "learning_rate": 8.237002400802586e-05, "loss": 0.0156, "step": 93620 }, { "epoch": 59.67495219885277, "grad_norm": 0.7035057544708252, "learning_rate": 8.23658231713146e-05, "loss": 0.0319, "step": 93630 }, { "epoch": 59.68132568514978, "grad_norm": 0.3953835368156433, "learning_rate": 8.236162194132803e-05, "loss": 0.023, "step": 93640 }, { "epoch": 59.68769917144678, "grad_norm": 0.3582005798816681, "learning_rate": 8.235742031811722e-05, "loss": 0.0207, "step": 93650 }, { "epoch": 59.69407265774379, "grad_norm": 0.5828365683555603, "learning_rate": 8.235321830173318e-05, "loss": 0.0181, "step": 93660 }, { "epoch": 59.70044614404079, "grad_norm": 0.2936023771762848, "learning_rate": 8.234901589222701e-05, "loss": 0.0238, "step": 93670 }, { "epoch": 59.7068196303378, "grad_norm": 0.2770211398601532, "learning_rate": 8.234481308964974e-05, "loss": 0.028, "step": 93680 }, { "epoch": 59.7131931166348, "grad_norm": 0.4537712335586548, "learning_rate": 8.234060989405246e-05, "loss": 0.0154, "step": 93690 }, { "epoch": 59.719566602931806, "grad_norm": 0.32808563113212585, "learning_rate": 8.233640630548622e-05, "loss": 0.0164, "step": 93700 }, { "epoch": 59.72594008922881, "grad_norm": 0.2303311675786972, "learning_rate": 8.233220232400214e-05, "loss": 0.0232, "step": 93710 }, { "epoch": 59.732313575525815, "grad_norm": 0.7458415031433105, "learning_rate": 8.232799794965125e-05, "loss": 0.0251, "step": 93720 }, { "epoch": 59.738687061822816, "grad_norm": 0.5359320640563965, "learning_rate": 8.232379318248466e-05, "loss": 0.0189, "step": 93730 }, { "epoch": 59.745060548119824, "grad_norm": 0.6634984016418457, "learning_rate": 8.231958802255348e-05, "loss": 0.0158, "step": 93740 }, { "epoch": 59.751434034416825, "grad_norm": 0.3584100306034088, "learning_rate": 8.23153824699088e-05, "loss": 0.0161, "step": 93750 }, { "epoch": 59.75780752071383, "grad_norm": 0.3296360373497009, "learning_rate": 8.231117652460169e-05, "loss": 0.0231, "step": 93760 }, { "epoch": 59.76418100701083, "grad_norm": 0.26982828974723816, "learning_rate": 8.230697018668329e-05, "loss": 0.0299, "step": 93770 }, { "epoch": 59.77055449330784, "grad_norm": 0.7162748575210571, "learning_rate": 8.23027634562047e-05, "loss": 0.0246, "step": 93780 }, { "epoch": 59.77692797960484, "grad_norm": 0.81397545337677, "learning_rate": 8.229855633321702e-05, "loss": 0.0352, "step": 93790 }, { "epoch": 59.78330146590185, "grad_norm": 0.6027111411094666, "learning_rate": 8.22943488177714e-05, "loss": 0.0218, "step": 93800 }, { "epoch": 59.78967495219885, "grad_norm": 0.3139319121837616, "learning_rate": 8.229014090991896e-05, "loss": 0.0207, "step": 93810 }, { "epoch": 59.79604843849586, "grad_norm": 0.43289288878440857, "learning_rate": 8.22859326097108e-05, "loss": 0.02, "step": 93820 }, { "epoch": 59.80242192479286, "grad_norm": 0.6288232207298279, "learning_rate": 8.228172391719809e-05, "loss": 0.0169, "step": 93830 }, { "epoch": 59.80879541108987, "grad_norm": 0.24853788316249847, "learning_rate": 8.227751483243195e-05, "loss": 0.0247, "step": 93840 }, { "epoch": 59.81516889738687, "grad_norm": 0.7029234766960144, "learning_rate": 8.227330535546353e-05, "loss": 0.0209, "step": 93850 }, { "epoch": 59.82154238368388, "grad_norm": 0.4341537058353424, "learning_rate": 8.226909548634398e-05, "loss": 0.0175, "step": 93860 }, { "epoch": 59.82791586998088, "grad_norm": 0.51505446434021, "learning_rate": 8.226488522512446e-05, "loss": 0.0181, "step": 93870 }, { "epoch": 59.834289356277885, "grad_norm": 0.2237633466720581, "learning_rate": 8.226067457185611e-05, "loss": 0.0193, "step": 93880 }, { "epoch": 59.840662842574886, "grad_norm": 0.5321182012557983, "learning_rate": 8.225646352659011e-05, "loss": 0.0201, "step": 93890 }, { "epoch": 59.847036328871894, "grad_norm": 0.446595162153244, "learning_rate": 8.225225208937761e-05, "loss": 0.0255, "step": 93900 }, { "epoch": 59.853409815168895, "grad_norm": 1.4309738874435425, "learning_rate": 8.224804026026981e-05, "loss": 0.0312, "step": 93910 }, { "epoch": 59.8597833014659, "grad_norm": 0.28338631987571716, "learning_rate": 8.224382803931788e-05, "loss": 0.0174, "step": 93920 }, { "epoch": 59.866156787762904, "grad_norm": 0.46626850962638855, "learning_rate": 8.223961542657297e-05, "loss": 0.0216, "step": 93930 }, { "epoch": 59.87253027405991, "grad_norm": 0.23753514885902405, "learning_rate": 8.223540242208631e-05, "loss": 0.0303, "step": 93940 }, { "epoch": 59.87890376035691, "grad_norm": 0.5911844372749329, "learning_rate": 8.223118902590908e-05, "loss": 0.022, "step": 93950 }, { "epoch": 59.88527724665392, "grad_norm": 0.6595186591148376, "learning_rate": 8.222697523809248e-05, "loss": 0.022, "step": 93960 }, { "epoch": 59.89165073295092, "grad_norm": 0.5199986696243286, "learning_rate": 8.22227610586877e-05, "loss": 0.0182, "step": 93970 }, { "epoch": 59.89802421924793, "grad_norm": 0.3174617290496826, "learning_rate": 8.221854648774592e-05, "loss": 0.0238, "step": 93980 }, { "epoch": 59.90439770554493, "grad_norm": 0.48729005455970764, "learning_rate": 8.221433152531841e-05, "loss": 0.0247, "step": 93990 }, { "epoch": 59.91077119184194, "grad_norm": 0.5597407817840576, "learning_rate": 8.221011617145635e-05, "loss": 0.0201, "step": 94000 }, { "epoch": 59.91714467813894, "grad_norm": 0.31119054555892944, "learning_rate": 8.220590042621097e-05, "loss": 0.0264, "step": 94010 }, { "epoch": 59.92351816443595, "grad_norm": 0.33432674407958984, "learning_rate": 8.220168428963348e-05, "loss": 0.0239, "step": 94020 }, { "epoch": 59.92989165073295, "grad_norm": 0.4152015447616577, "learning_rate": 8.219746776177513e-05, "loss": 0.0137, "step": 94030 }, { "epoch": 59.936265137029956, "grad_norm": 0.6544189453125, "learning_rate": 8.219325084268715e-05, "loss": 0.0247, "step": 94040 }, { "epoch": 59.94263862332696, "grad_norm": 0.4218325912952423, "learning_rate": 8.218903353242077e-05, "loss": 0.0158, "step": 94050 }, { "epoch": 59.949012109623965, "grad_norm": 0.4683205485343933, "learning_rate": 8.218481583102726e-05, "loss": 0.0165, "step": 94060 }, { "epoch": 59.955385595920966, "grad_norm": 0.4502445161342621, "learning_rate": 8.218059773855782e-05, "loss": 0.0192, "step": 94070 }, { "epoch": 59.961759082217974, "grad_norm": 0.4613311290740967, "learning_rate": 8.217637925506376e-05, "loss": 0.0283, "step": 94080 }, { "epoch": 59.968132568514974, "grad_norm": 0.39782169461250305, "learning_rate": 8.21721603805963e-05, "loss": 0.0239, "step": 94090 }, { "epoch": 59.97450605481198, "grad_norm": 0.574532151222229, "learning_rate": 8.216794111520671e-05, "loss": 0.0304, "step": 94100 }, { "epoch": 59.98087954110898, "grad_norm": 0.37155768275260925, "learning_rate": 8.216372145894627e-05, "loss": 0.0212, "step": 94110 }, { "epoch": 59.98725302740599, "grad_norm": 0.5361500382423401, "learning_rate": 8.215950141186626e-05, "loss": 0.0178, "step": 94120 }, { "epoch": 59.99362651370299, "grad_norm": 0.4008428752422333, "learning_rate": 8.215528097401793e-05, "loss": 0.0172, "step": 94130 }, { "epoch": 60.0, "grad_norm": 0.740408182144165, "learning_rate": 8.215106014545258e-05, "loss": 0.0118, "step": 94140 }, { "epoch": 60.00637348629701, "grad_norm": 0.36434316635131836, "learning_rate": 8.21468389262215e-05, "loss": 0.0196, "step": 94150 }, { "epoch": 60.01274697259401, "grad_norm": 0.2639552652835846, "learning_rate": 8.214261731637596e-05, "loss": 0.0246, "step": 94160 }, { "epoch": 60.01912045889102, "grad_norm": 0.33472633361816406, "learning_rate": 8.213839531596728e-05, "loss": 0.0197, "step": 94170 }, { "epoch": 60.02549394518802, "grad_norm": 0.6816985607147217, "learning_rate": 8.213417292504674e-05, "loss": 0.0255, "step": 94180 }, { "epoch": 60.031867431485026, "grad_norm": 0.5622140169143677, "learning_rate": 8.212995014366569e-05, "loss": 0.02, "step": 94190 }, { "epoch": 60.038240917782026, "grad_norm": 0.30110636353492737, "learning_rate": 8.212572697187539e-05, "loss": 0.0179, "step": 94200 }, { "epoch": 60.044614404079034, "grad_norm": 0.471817284822464, "learning_rate": 8.212150340972717e-05, "loss": 0.0159, "step": 94210 }, { "epoch": 60.050987890376035, "grad_norm": 0.5833938121795654, "learning_rate": 8.211727945727237e-05, "loss": 0.0132, "step": 94220 }, { "epoch": 60.05736137667304, "grad_norm": 0.7653937339782715, "learning_rate": 8.211305511456229e-05, "loss": 0.0212, "step": 94230 }, { "epoch": 60.063734862970044, "grad_norm": 0.3829895555973053, "learning_rate": 8.210883038164828e-05, "loss": 0.0199, "step": 94240 }, { "epoch": 60.07010834926705, "grad_norm": 0.21789944171905518, "learning_rate": 8.210460525858166e-05, "loss": 0.0219, "step": 94250 }, { "epoch": 60.07648183556405, "grad_norm": 0.47638124227523804, "learning_rate": 8.210037974541377e-05, "loss": 0.0187, "step": 94260 }, { "epoch": 60.08285532186106, "grad_norm": 0.469731867313385, "learning_rate": 8.209615384219596e-05, "loss": 0.023, "step": 94270 }, { "epoch": 60.08922880815806, "grad_norm": 0.6022607088088989, "learning_rate": 8.209192754897958e-05, "loss": 0.017, "step": 94280 }, { "epoch": 60.09560229445507, "grad_norm": 0.1777062714099884, "learning_rate": 8.2087700865816e-05, "loss": 0.0122, "step": 94290 }, { "epoch": 60.10197578075207, "grad_norm": 0.2868332862854004, "learning_rate": 8.208347379275654e-05, "loss": 0.0154, "step": 94300 }, { "epoch": 60.10834926704908, "grad_norm": 0.5057234168052673, "learning_rate": 8.207924632985258e-05, "loss": 0.0281, "step": 94310 }, { "epoch": 60.11472275334608, "grad_norm": 0.406100332736969, "learning_rate": 8.207501847715549e-05, "loss": 0.017, "step": 94320 }, { "epoch": 60.12109623964309, "grad_norm": 0.33227604627609253, "learning_rate": 8.207079023471664e-05, "loss": 0.0136, "step": 94330 }, { "epoch": 60.12746972594009, "grad_norm": 0.5410879254341125, "learning_rate": 8.206656160258742e-05, "loss": 0.0251, "step": 94340 }, { "epoch": 60.133843212237096, "grad_norm": 0.4204516112804413, "learning_rate": 8.206233258081918e-05, "loss": 0.0176, "step": 94350 }, { "epoch": 60.1402166985341, "grad_norm": 0.3921091556549072, "learning_rate": 8.205810316946334e-05, "loss": 0.0155, "step": 94360 }, { "epoch": 60.146590184831105, "grad_norm": 0.49312520027160645, "learning_rate": 8.205387336857128e-05, "loss": 0.0218, "step": 94370 }, { "epoch": 60.152963671128106, "grad_norm": 0.7163779735565186, "learning_rate": 8.204964317819439e-05, "loss": 0.0276, "step": 94380 }, { "epoch": 60.159337157425114, "grad_norm": 0.6908515095710754, "learning_rate": 8.204541259838408e-05, "loss": 0.0254, "step": 94390 }, { "epoch": 60.165710643722115, "grad_norm": 0.27434495091438293, "learning_rate": 8.204118162919174e-05, "loss": 0.0177, "step": 94400 }, { "epoch": 60.17208413001912, "grad_norm": 0.5171014666557312, "learning_rate": 8.20369502706688e-05, "loss": 0.0194, "step": 94410 }, { "epoch": 60.17845761631612, "grad_norm": 0.4232417047023773, "learning_rate": 8.203271852286667e-05, "loss": 0.0189, "step": 94420 }, { "epoch": 60.18483110261313, "grad_norm": 0.37033259868621826, "learning_rate": 8.202848638583675e-05, "loss": 0.0171, "step": 94430 }, { "epoch": 60.19120458891013, "grad_norm": 0.3788917660713196, "learning_rate": 8.20242538596305e-05, "loss": 0.0151, "step": 94440 }, { "epoch": 60.19757807520714, "grad_norm": 0.5438482165336609, "learning_rate": 8.202002094429933e-05, "loss": 0.0212, "step": 94450 }, { "epoch": 60.20395156150414, "grad_norm": 0.5093549489974976, "learning_rate": 8.201578763989466e-05, "loss": 0.0238, "step": 94460 }, { "epoch": 60.21032504780115, "grad_norm": 0.15039405226707458, "learning_rate": 8.201155394646796e-05, "loss": 0.0188, "step": 94470 }, { "epoch": 60.21669853409815, "grad_norm": 0.2884507477283478, "learning_rate": 8.200731986407064e-05, "loss": 0.0177, "step": 94480 }, { "epoch": 60.22307202039516, "grad_norm": 0.3501623570919037, "learning_rate": 8.200308539275417e-05, "loss": 0.0257, "step": 94490 }, { "epoch": 60.22944550669216, "grad_norm": 0.3756864666938782, "learning_rate": 8.199885053256999e-05, "loss": 0.0157, "step": 94500 }, { "epoch": 60.23581899298917, "grad_norm": 0.46834036707878113, "learning_rate": 8.199461528356958e-05, "loss": 0.0312, "step": 94510 }, { "epoch": 60.24219247928617, "grad_norm": 0.600739598274231, "learning_rate": 8.199037964580436e-05, "loss": 0.0217, "step": 94520 }, { "epoch": 60.248565965583175, "grad_norm": 0.4493928849697113, "learning_rate": 8.198614361932583e-05, "loss": 0.0144, "step": 94530 }, { "epoch": 60.254939451880176, "grad_norm": 0.38744163513183594, "learning_rate": 8.198190720418549e-05, "loss": 0.0223, "step": 94540 }, { "epoch": 60.261312938177184, "grad_norm": 0.47216832637786865, "learning_rate": 8.197767040043474e-05, "loss": 0.0169, "step": 94550 }, { "epoch": 60.267686424474185, "grad_norm": 0.538172721862793, "learning_rate": 8.197343320812512e-05, "loss": 0.0226, "step": 94560 }, { "epoch": 60.27405991077119, "grad_norm": 0.35447004437446594, "learning_rate": 8.19691956273081e-05, "loss": 0.0337, "step": 94570 }, { "epoch": 60.280433397068194, "grad_norm": 0.9600151777267456, "learning_rate": 8.196495765803515e-05, "loss": 0.034, "step": 94580 }, { "epoch": 60.2868068833652, "grad_norm": 0.297158420085907, "learning_rate": 8.19607193003578e-05, "loss": 0.0268, "step": 94590 }, { "epoch": 60.2931803696622, "grad_norm": 0.3200402557849884, "learning_rate": 8.195648055432755e-05, "loss": 0.0139, "step": 94600 }, { "epoch": 60.29955385595921, "grad_norm": 0.42138639092445374, "learning_rate": 8.195224141999587e-05, "loss": 0.0271, "step": 94610 }, { "epoch": 60.30592734225621, "grad_norm": 0.6901991963386536, "learning_rate": 8.19480018974143e-05, "loss": 0.0278, "step": 94620 }, { "epoch": 60.31230082855322, "grad_norm": 0.7307824492454529, "learning_rate": 8.194376198663434e-05, "loss": 0.0184, "step": 94630 }, { "epoch": 60.31867431485022, "grad_norm": 0.4220949411392212, "learning_rate": 8.19395216877075e-05, "loss": 0.0305, "step": 94640 }, { "epoch": 60.32504780114723, "grad_norm": 0.3536379933357239, "learning_rate": 8.193528100068532e-05, "loss": 0.0228, "step": 94650 }, { "epoch": 60.33142128744423, "grad_norm": 0.42400944232940674, "learning_rate": 8.193103992561933e-05, "loss": 0.0189, "step": 94660 }, { "epoch": 60.33779477374124, "grad_norm": 0.32926443219184875, "learning_rate": 8.192679846256106e-05, "loss": 0.0217, "step": 94670 }, { "epoch": 60.34416826003824, "grad_norm": 0.5759711861610413, "learning_rate": 8.192255661156202e-05, "loss": 0.026, "step": 94680 }, { "epoch": 60.350541746335246, "grad_norm": 0.679394006729126, "learning_rate": 8.19183143726738e-05, "loss": 0.0265, "step": 94690 }, { "epoch": 60.35691523263225, "grad_norm": 0.2350999414920807, "learning_rate": 8.191407174594794e-05, "loss": 0.0235, "step": 94700 }, { "epoch": 60.363288718929255, "grad_norm": 0.5690944790840149, "learning_rate": 8.190982873143594e-05, "loss": 0.0142, "step": 94710 }, { "epoch": 60.369662205226255, "grad_norm": 0.3832121193408966, "learning_rate": 8.190558532918941e-05, "loss": 0.023, "step": 94720 }, { "epoch": 60.37603569152326, "grad_norm": 0.8401877284049988, "learning_rate": 8.19013415392599e-05, "loss": 0.0252, "step": 94730 }, { "epoch": 60.382409177820264, "grad_norm": 0.7321261763572693, "learning_rate": 8.189709736169898e-05, "loss": 0.0229, "step": 94740 }, { "epoch": 60.38878266411727, "grad_norm": 0.38541319966316223, "learning_rate": 8.189285279655818e-05, "loss": 0.0292, "step": 94750 }, { "epoch": 60.39515615041427, "grad_norm": 0.3211003243923187, "learning_rate": 8.188860784388913e-05, "loss": 0.0222, "step": 94760 }, { "epoch": 60.40152963671128, "grad_norm": 0.41377875208854675, "learning_rate": 8.188436250374339e-05, "loss": 0.0205, "step": 94770 }, { "epoch": 60.40790312300829, "grad_norm": 0.6658848524093628, "learning_rate": 8.188011677617251e-05, "loss": 0.0272, "step": 94780 }, { "epoch": 60.41427660930529, "grad_norm": 0.46615052223205566, "learning_rate": 8.187587066122814e-05, "loss": 0.0324, "step": 94790 }, { "epoch": 60.4206500956023, "grad_norm": 0.6167742609977722, "learning_rate": 8.187162415896184e-05, "loss": 0.0189, "step": 94800 }, { "epoch": 60.4270235818993, "grad_norm": 0.2914963662624359, "learning_rate": 8.186737726942522e-05, "loss": 0.025, "step": 94810 }, { "epoch": 60.43339706819631, "grad_norm": 0.5335922837257385, "learning_rate": 8.186312999266987e-05, "loss": 0.0235, "step": 94820 }, { "epoch": 60.43977055449331, "grad_norm": 0.33137714862823486, "learning_rate": 8.18588823287474e-05, "loss": 0.0152, "step": 94830 }, { "epoch": 60.446144040790315, "grad_norm": 0.35201752185821533, "learning_rate": 8.185463427770944e-05, "loss": 0.0193, "step": 94840 }, { "epoch": 60.452517527087316, "grad_norm": 0.3196587562561035, "learning_rate": 8.18503858396076e-05, "loss": 0.0211, "step": 94850 }, { "epoch": 60.458891013384324, "grad_norm": 0.45294442772865295, "learning_rate": 8.184613701449349e-05, "loss": 0.0178, "step": 94860 }, { "epoch": 60.465264499681325, "grad_norm": 0.38281112909317017, "learning_rate": 8.184188780241876e-05, "loss": 0.0245, "step": 94870 }, { "epoch": 60.47163798597833, "grad_norm": 0.5284104943275452, "learning_rate": 8.183763820343501e-05, "loss": 0.0225, "step": 94880 }, { "epoch": 60.478011472275334, "grad_norm": 0.21307097375392914, "learning_rate": 8.183338821759392e-05, "loss": 0.03, "step": 94890 }, { "epoch": 60.48438495857234, "grad_norm": 0.3982771039009094, "learning_rate": 8.18291378449471e-05, "loss": 0.0224, "step": 94900 }, { "epoch": 60.49075844486934, "grad_norm": 0.26546910405158997, "learning_rate": 8.18248870855462e-05, "loss": 0.0385, "step": 94910 }, { "epoch": 60.49713193116635, "grad_norm": 0.44611799716949463, "learning_rate": 8.182063593944287e-05, "loss": 0.0232, "step": 94920 }, { "epoch": 60.50350541746335, "grad_norm": 0.3830655515193939, "learning_rate": 8.181638440668879e-05, "loss": 0.0224, "step": 94930 }, { "epoch": 60.50987890376036, "grad_norm": 0.31376492977142334, "learning_rate": 8.181213248733556e-05, "loss": 0.0161, "step": 94940 }, { "epoch": 60.51625239005736, "grad_norm": 0.3416409194469452, "learning_rate": 8.180788018143492e-05, "loss": 0.0198, "step": 94950 }, { "epoch": 60.52262587635437, "grad_norm": 0.30884602665901184, "learning_rate": 8.180362748903849e-05, "loss": 0.0134, "step": 94960 }, { "epoch": 60.52899936265137, "grad_norm": 0.3671185076236725, "learning_rate": 8.179937441019798e-05, "loss": 0.0161, "step": 94970 }, { "epoch": 60.53537284894838, "grad_norm": 0.3416841924190521, "learning_rate": 8.179512094496502e-05, "loss": 0.0193, "step": 94980 }, { "epoch": 60.54174633524538, "grad_norm": 0.4086098074913025, "learning_rate": 8.179086709339132e-05, "loss": 0.0108, "step": 94990 }, { "epoch": 60.548119821542386, "grad_norm": 0.5163726210594177, "learning_rate": 8.178661285552858e-05, "loss": 0.0162, "step": 95000 }, { "epoch": 60.55449330783939, "grad_norm": 0.7988197803497314, "learning_rate": 8.178235823142849e-05, "loss": 0.0258, "step": 95010 }, { "epoch": 60.560866794136395, "grad_norm": 0.5301536917686462, "learning_rate": 8.177810322114274e-05, "loss": 0.0298, "step": 95020 }, { "epoch": 60.567240280433396, "grad_norm": 0.4332931339740753, "learning_rate": 8.177384782472302e-05, "loss": 0.0248, "step": 95030 }, { "epoch": 60.5736137667304, "grad_norm": 0.3519963026046753, "learning_rate": 8.176959204222106e-05, "loss": 0.0177, "step": 95040 }, { "epoch": 60.579987253027404, "grad_norm": 0.4729333817958832, "learning_rate": 8.176533587368854e-05, "loss": 0.0268, "step": 95050 }, { "epoch": 60.58636073932441, "grad_norm": 0.27976396679878235, "learning_rate": 8.176107931917721e-05, "loss": 0.0241, "step": 95060 }, { "epoch": 60.59273422562141, "grad_norm": 0.2705378234386444, "learning_rate": 8.175682237873878e-05, "loss": 0.0124, "step": 95070 }, { "epoch": 60.59910771191842, "grad_norm": 0.4462827444076538, "learning_rate": 8.175256505242497e-05, "loss": 0.0239, "step": 95080 }, { "epoch": 60.60548119821542, "grad_norm": 0.3681188225746155, "learning_rate": 8.174830734028753e-05, "loss": 0.0267, "step": 95090 }, { "epoch": 60.61185468451243, "grad_norm": 0.4259365200996399, "learning_rate": 8.174404924237817e-05, "loss": 0.0162, "step": 95100 }, { "epoch": 60.61822817080943, "grad_norm": 0.30780887603759766, "learning_rate": 8.173979075874864e-05, "loss": 0.0214, "step": 95110 }, { "epoch": 60.62460165710644, "grad_norm": 1.293228268623352, "learning_rate": 8.173553188945069e-05, "loss": 0.016, "step": 95120 }, { "epoch": 60.63097514340344, "grad_norm": 0.38159847259521484, "learning_rate": 8.173127263453606e-05, "loss": 0.0331, "step": 95130 }, { "epoch": 60.63734862970045, "grad_norm": 0.4378938376903534, "learning_rate": 8.172701299405651e-05, "loss": 0.0287, "step": 95140 }, { "epoch": 60.64372211599745, "grad_norm": 0.3534691631793976, "learning_rate": 8.17227529680638e-05, "loss": 0.0193, "step": 95150 }, { "epoch": 60.650095602294456, "grad_norm": 0.5505616068840027, "learning_rate": 8.171849255660968e-05, "loss": 0.0343, "step": 95160 }, { "epoch": 60.65646908859146, "grad_norm": 0.1939314603805542, "learning_rate": 8.171423175974594e-05, "loss": 0.026, "step": 95170 }, { "epoch": 60.662842574888465, "grad_norm": 0.2661811411380768, "learning_rate": 8.170997057752434e-05, "loss": 0.0196, "step": 95180 }, { "epoch": 60.669216061185466, "grad_norm": 0.5986185669898987, "learning_rate": 8.170570900999665e-05, "loss": 0.0191, "step": 95190 }, { "epoch": 60.675589547482474, "grad_norm": 0.5243088006973267, "learning_rate": 8.170144705721465e-05, "loss": 0.0256, "step": 95200 }, { "epoch": 60.681963033779475, "grad_norm": 0.5933504104614258, "learning_rate": 8.169718471923016e-05, "loss": 0.0178, "step": 95210 }, { "epoch": 60.68833652007648, "grad_norm": 0.27748236060142517, "learning_rate": 8.169292199609494e-05, "loss": 0.0196, "step": 95220 }, { "epoch": 60.694710006373484, "grad_norm": 0.38703417778015137, "learning_rate": 8.16886588878608e-05, "loss": 0.0271, "step": 95230 }, { "epoch": 60.70108349267049, "grad_norm": 0.28786715865135193, "learning_rate": 8.168439539457953e-05, "loss": 0.0092, "step": 95240 }, { "epoch": 60.70745697896749, "grad_norm": 0.28245866298675537, "learning_rate": 8.168013151630294e-05, "loss": 0.0208, "step": 95250 }, { "epoch": 60.7138304652645, "grad_norm": 0.44695496559143066, "learning_rate": 8.167586725308283e-05, "loss": 0.0216, "step": 95260 }, { "epoch": 60.7202039515615, "grad_norm": 0.2618371844291687, "learning_rate": 8.167160260497105e-05, "loss": 0.0264, "step": 95270 }, { "epoch": 60.72657743785851, "grad_norm": 0.4371306598186493, "learning_rate": 8.166733757201937e-05, "loss": 0.0216, "step": 95280 }, { "epoch": 60.73295092415551, "grad_norm": 0.2725467383861542, "learning_rate": 8.166307215427967e-05, "loss": 0.0174, "step": 95290 }, { "epoch": 60.73932441045252, "grad_norm": 0.4796711802482605, "learning_rate": 8.165880635180372e-05, "loss": 0.0188, "step": 95300 }, { "epoch": 60.74569789674952, "grad_norm": 0.45764002203941345, "learning_rate": 8.16545401646434e-05, "loss": 0.0207, "step": 95310 }, { "epoch": 60.75207138304653, "grad_norm": 0.49266281723976135, "learning_rate": 8.165027359285052e-05, "loss": 0.0243, "step": 95320 }, { "epoch": 60.75844486934353, "grad_norm": 0.25585174560546875, "learning_rate": 8.164600663647693e-05, "loss": 0.0276, "step": 95330 }, { "epoch": 60.764818355640536, "grad_norm": 0.2526433765888214, "learning_rate": 8.164173929557449e-05, "loss": 0.0217, "step": 95340 }, { "epoch": 60.77119184193754, "grad_norm": 0.275379478931427, "learning_rate": 8.163747157019503e-05, "loss": 0.0248, "step": 95350 }, { "epoch": 60.777565328234544, "grad_norm": 0.39631468057632446, "learning_rate": 8.163320346039044e-05, "loss": 0.0214, "step": 95360 }, { "epoch": 60.783938814531545, "grad_norm": 0.31717073917388916, "learning_rate": 8.162893496621255e-05, "loss": 0.0138, "step": 95370 }, { "epoch": 60.79031230082855, "grad_norm": 0.23715782165527344, "learning_rate": 8.162466608771324e-05, "loss": 0.0127, "step": 95380 }, { "epoch": 60.796685787125554, "grad_norm": 0.42857521772384644, "learning_rate": 8.162039682494437e-05, "loss": 0.0321, "step": 95390 }, { "epoch": 60.80305927342256, "grad_norm": 0.6073705554008484, "learning_rate": 8.161612717795783e-05, "loss": 0.033, "step": 95400 }, { "epoch": 60.80943275971957, "grad_norm": 0.3969240188598633, "learning_rate": 8.161185714680551e-05, "loss": 0.0154, "step": 95410 }, { "epoch": 60.81580624601657, "grad_norm": 0.544790506362915, "learning_rate": 8.160758673153926e-05, "loss": 0.0267, "step": 95420 }, { "epoch": 60.82217973231358, "grad_norm": 0.622099757194519, "learning_rate": 8.160331593221099e-05, "loss": 0.0199, "step": 95430 }, { "epoch": 60.82855321861058, "grad_norm": 0.9002111554145813, "learning_rate": 8.159904474887259e-05, "loss": 0.0263, "step": 95440 }, { "epoch": 60.83492670490759, "grad_norm": 0.2974507808685303, "learning_rate": 8.159477318157598e-05, "loss": 0.0199, "step": 95450 }, { "epoch": 60.84130019120459, "grad_norm": 0.5293838381767273, "learning_rate": 8.159050123037301e-05, "loss": 0.0232, "step": 95460 }, { "epoch": 60.8476736775016, "grad_norm": 0.42628180980682373, "learning_rate": 8.158622889531567e-05, "loss": 0.0199, "step": 95470 }, { "epoch": 60.8540471637986, "grad_norm": 0.40943604707717896, "learning_rate": 8.158195617645579e-05, "loss": 0.0205, "step": 95480 }, { "epoch": 60.860420650095605, "grad_norm": 0.23170246183872223, "learning_rate": 8.157768307384535e-05, "loss": 0.0189, "step": 95490 }, { "epoch": 60.866794136392606, "grad_norm": 0.6541046500205994, "learning_rate": 8.157340958753624e-05, "loss": 0.0248, "step": 95500 }, { "epoch": 60.873167622689614, "grad_norm": 0.2681419253349304, "learning_rate": 8.156913571758037e-05, "loss": 0.027, "step": 95510 }, { "epoch": 60.879541108986615, "grad_norm": 0.6248667240142822, "learning_rate": 8.156486146402972e-05, "loss": 0.0228, "step": 95520 }, { "epoch": 60.88591459528362, "grad_norm": 0.4736101031303406, "learning_rate": 8.156058682693619e-05, "loss": 0.0173, "step": 95530 }, { "epoch": 60.892288081580624, "grad_norm": 0.58992999792099, "learning_rate": 8.155631180635172e-05, "loss": 0.0377, "step": 95540 }, { "epoch": 60.89866156787763, "grad_norm": 0.40210843086242676, "learning_rate": 8.155203640232829e-05, "loss": 0.0219, "step": 95550 }, { "epoch": 60.90503505417463, "grad_norm": 0.3092411160469055, "learning_rate": 8.154776061491782e-05, "loss": 0.0268, "step": 95560 }, { "epoch": 60.91140854047164, "grad_norm": 0.5102787613868713, "learning_rate": 8.154348444417227e-05, "loss": 0.0146, "step": 95570 }, { "epoch": 60.91778202676864, "grad_norm": 0.3227364122867584, "learning_rate": 8.15392078901436e-05, "loss": 0.0173, "step": 95580 }, { "epoch": 60.92415551306565, "grad_norm": 0.8906571865081787, "learning_rate": 8.153493095288379e-05, "loss": 0.0195, "step": 95590 }, { "epoch": 60.93052899936265, "grad_norm": 0.659977912902832, "learning_rate": 8.153065363244479e-05, "loss": 0.0151, "step": 95600 }, { "epoch": 60.93690248565966, "grad_norm": 0.4790312349796295, "learning_rate": 8.152637592887857e-05, "loss": 0.0181, "step": 95610 }, { "epoch": 60.94327597195666, "grad_norm": 0.5095083713531494, "learning_rate": 8.152209784223712e-05, "loss": 0.0203, "step": 95620 }, { "epoch": 60.94964945825367, "grad_norm": 0.29002413153648376, "learning_rate": 8.151781937257241e-05, "loss": 0.0219, "step": 95630 }, { "epoch": 60.95602294455067, "grad_norm": 0.5270630717277527, "learning_rate": 8.151354051993644e-05, "loss": 0.0148, "step": 95640 }, { "epoch": 60.962396430847676, "grad_norm": 0.9076797962188721, "learning_rate": 8.150926128438122e-05, "loss": 0.0198, "step": 95650 }, { "epoch": 60.96876991714468, "grad_norm": 0.3433191776275635, "learning_rate": 8.150498166595871e-05, "loss": 0.019, "step": 95660 }, { "epoch": 60.975143403441685, "grad_norm": 0.4772210717201233, "learning_rate": 8.150070166472093e-05, "loss": 0.0321, "step": 95670 }, { "epoch": 60.981516889738685, "grad_norm": 0.4406152367591858, "learning_rate": 8.149642128071988e-05, "loss": 0.0209, "step": 95680 }, { "epoch": 60.98789037603569, "grad_norm": 0.4365846514701843, "learning_rate": 8.149214051400757e-05, "loss": 0.0246, "step": 95690 }, { "epoch": 60.994263862332694, "grad_norm": 0.4946298897266388, "learning_rate": 8.148785936463603e-05, "loss": 0.0443, "step": 95700 }, { "epoch": 61.0006373486297, "grad_norm": 0.39964190125465393, "learning_rate": 8.148357783265727e-05, "loss": 0.0154, "step": 95710 }, { "epoch": 61.0070108349267, "grad_norm": 0.3821636140346527, "learning_rate": 8.147929591812331e-05, "loss": 0.0276, "step": 95720 }, { "epoch": 61.01338432122371, "grad_norm": 0.4688991904258728, "learning_rate": 8.147501362108617e-05, "loss": 0.0181, "step": 95730 }, { "epoch": 61.01975780752071, "grad_norm": 0.5446555018424988, "learning_rate": 8.147073094159792e-05, "loss": 0.026, "step": 95740 }, { "epoch": 61.02613129381772, "grad_norm": 0.2936679720878601, "learning_rate": 8.146644787971056e-05, "loss": 0.0228, "step": 95750 }, { "epoch": 61.03250478011472, "grad_norm": 0.26948612928390503, "learning_rate": 8.146216443547615e-05, "loss": 0.0166, "step": 95760 }, { "epoch": 61.03887826641173, "grad_norm": 0.4455488324165344, "learning_rate": 8.145788060894674e-05, "loss": 0.0205, "step": 95770 }, { "epoch": 61.04525175270873, "grad_norm": 0.36903873085975647, "learning_rate": 8.145359640017438e-05, "loss": 0.0213, "step": 95780 }, { "epoch": 61.05162523900574, "grad_norm": 0.4006199538707733, "learning_rate": 8.144931180921114e-05, "loss": 0.0128, "step": 95790 }, { "epoch": 61.05799872530274, "grad_norm": 0.4446595311164856, "learning_rate": 8.144502683610904e-05, "loss": 0.0197, "step": 95800 }, { "epoch": 61.064372211599746, "grad_norm": 0.33677929639816284, "learning_rate": 8.144074148092021e-05, "loss": 0.0223, "step": 95810 }, { "epoch": 61.07074569789675, "grad_norm": 0.3379022777080536, "learning_rate": 8.143645574369664e-05, "loss": 0.0229, "step": 95820 }, { "epoch": 61.077119184193755, "grad_norm": 0.4703719913959503, "learning_rate": 8.14321696244905e-05, "loss": 0.0165, "step": 95830 }, { "epoch": 61.083492670490756, "grad_norm": 0.3277742862701416, "learning_rate": 8.142788312335379e-05, "loss": 0.0152, "step": 95840 }, { "epoch": 61.089866156787764, "grad_norm": 0.6366003751754761, "learning_rate": 8.142359624033865e-05, "loss": 0.0293, "step": 95850 }, { "epoch": 61.096239643084765, "grad_norm": 0.25846773386001587, "learning_rate": 8.141930897549713e-05, "loss": 0.0182, "step": 95860 }, { "epoch": 61.10261312938177, "grad_norm": 0.32974714040756226, "learning_rate": 8.141502132888132e-05, "loss": 0.0213, "step": 95870 }, { "epoch": 61.10898661567877, "grad_norm": 0.6804724931716919, "learning_rate": 8.141073330054338e-05, "loss": 0.0265, "step": 95880 }, { "epoch": 61.11536010197578, "grad_norm": 0.2344520390033722, "learning_rate": 8.140644489053535e-05, "loss": 0.0201, "step": 95890 }, { "epoch": 61.12173358827278, "grad_norm": 0.3404674530029297, "learning_rate": 8.140215609890938e-05, "loss": 0.0144, "step": 95900 }, { "epoch": 61.12810707456979, "grad_norm": 0.4693468511104584, "learning_rate": 8.139786692571755e-05, "loss": 0.027, "step": 95910 }, { "epoch": 61.13448056086679, "grad_norm": 0.3311028480529785, "learning_rate": 8.139357737101201e-05, "loss": 0.0146, "step": 95920 }, { "epoch": 61.1408540471638, "grad_norm": 0.4409269094467163, "learning_rate": 8.138928743484484e-05, "loss": 0.0212, "step": 95930 }, { "epoch": 61.1472275334608, "grad_norm": 0.4106735587120056, "learning_rate": 8.138499711726821e-05, "loss": 0.02, "step": 95940 }, { "epoch": 61.15360101975781, "grad_norm": 0.43851763010025024, "learning_rate": 8.138070641833421e-05, "loss": 0.0216, "step": 95950 }, { "epoch": 61.15997450605481, "grad_norm": 0.8622010946273804, "learning_rate": 8.137641533809504e-05, "loss": 0.0175, "step": 95960 }, { "epoch": 61.16634799235182, "grad_norm": 0.3043321371078491, "learning_rate": 8.137212387660276e-05, "loss": 0.0422, "step": 95970 }, { "epoch": 61.17272147864882, "grad_norm": 0.36679545044898987, "learning_rate": 8.136783203390958e-05, "loss": 0.0204, "step": 95980 }, { "epoch": 61.179094964945826, "grad_norm": 0.33673644065856934, "learning_rate": 8.13635398100676e-05, "loss": 0.0195, "step": 95990 }, { "epoch": 61.185468451242826, "grad_norm": 0.42075634002685547, "learning_rate": 8.135924720512903e-05, "loss": 0.0265, "step": 96000 }, { "epoch": 61.191841937539834, "grad_norm": 0.3706428110599518, "learning_rate": 8.135495421914597e-05, "loss": 0.0179, "step": 96010 }, { "epoch": 61.198215423836835, "grad_norm": 0.9598209261894226, "learning_rate": 8.135066085217062e-05, "loss": 0.0294, "step": 96020 }, { "epoch": 61.20458891013384, "grad_norm": 0.3607963025569916, "learning_rate": 8.134636710425513e-05, "loss": 0.0183, "step": 96030 }, { "epoch": 61.21096239643085, "grad_norm": 0.40037673711776733, "learning_rate": 8.134207297545171e-05, "loss": 0.0241, "step": 96040 }, { "epoch": 61.21733588272785, "grad_norm": 0.3431549668312073, "learning_rate": 8.13377784658125e-05, "loss": 0.0204, "step": 96050 }, { "epoch": 61.22370936902486, "grad_norm": 0.3988195061683655, "learning_rate": 8.133348357538971e-05, "loss": 0.0201, "step": 96060 }, { "epoch": 61.23008285532186, "grad_norm": 0.29358696937561035, "learning_rate": 8.132918830423548e-05, "loss": 0.0178, "step": 96070 }, { "epoch": 61.23645634161887, "grad_norm": 0.2851429581642151, "learning_rate": 8.132489265240205e-05, "loss": 0.02, "step": 96080 }, { "epoch": 61.24282982791587, "grad_norm": 0.32349932193756104, "learning_rate": 8.13205966199416e-05, "loss": 0.0187, "step": 96090 }, { "epoch": 61.24920331421288, "grad_norm": 0.3963437080383301, "learning_rate": 8.131630020690632e-05, "loss": 0.0305, "step": 96100 }, { "epoch": 61.25557680050988, "grad_norm": 0.2823528051376343, "learning_rate": 8.131200341334843e-05, "loss": 0.0169, "step": 96110 }, { "epoch": 61.261950286806886, "grad_norm": 0.6109834909439087, "learning_rate": 8.130770623932013e-05, "loss": 0.0311, "step": 96120 }, { "epoch": 61.26832377310389, "grad_norm": 0.37428396940231323, "learning_rate": 8.130340868487365e-05, "loss": 0.0171, "step": 96130 }, { "epoch": 61.274697259400895, "grad_norm": 0.38158851861953735, "learning_rate": 8.12991107500612e-05, "loss": 0.0146, "step": 96140 }, { "epoch": 61.281070745697896, "grad_norm": 0.2713254392147064, "learning_rate": 8.1294812434935e-05, "loss": 0.0155, "step": 96150 }, { "epoch": 61.287444231994904, "grad_norm": 0.41676440834999084, "learning_rate": 8.129051373954728e-05, "loss": 0.0341, "step": 96160 }, { "epoch": 61.293817718291905, "grad_norm": 0.5367199182510376, "learning_rate": 8.128621466395028e-05, "loss": 0.0178, "step": 96170 }, { "epoch": 61.30019120458891, "grad_norm": 0.48037564754486084, "learning_rate": 8.128191520819621e-05, "loss": 0.0165, "step": 96180 }, { "epoch": 61.306564690885914, "grad_norm": 0.7687711119651794, "learning_rate": 8.127761537233736e-05, "loss": 0.0245, "step": 96190 }, { "epoch": 61.31293817718292, "grad_norm": 0.6043808460235596, "learning_rate": 8.127331515642595e-05, "loss": 0.0292, "step": 96200 }, { "epoch": 61.31931166347992, "grad_norm": 0.40612417459487915, "learning_rate": 8.126901456051421e-05, "loss": 0.0151, "step": 96210 }, { "epoch": 61.32568514977693, "grad_norm": 0.31163692474365234, "learning_rate": 8.126471358465445e-05, "loss": 0.016, "step": 96220 }, { "epoch": 61.33205863607393, "grad_norm": 0.5293201208114624, "learning_rate": 8.126041222889889e-05, "loss": 0.0235, "step": 96230 }, { "epoch": 61.33843212237094, "grad_norm": 0.4634445309638977, "learning_rate": 8.125611049329981e-05, "loss": 0.0243, "step": 96240 }, { "epoch": 61.34480560866794, "grad_norm": 0.49156299233436584, "learning_rate": 8.125180837790946e-05, "loss": 0.0178, "step": 96250 }, { "epoch": 61.35117909496495, "grad_norm": 0.38550353050231934, "learning_rate": 8.124750588278017e-05, "loss": 0.0186, "step": 96260 }, { "epoch": 61.35755258126195, "grad_norm": 0.3207261860370636, "learning_rate": 8.124320300796414e-05, "loss": 0.0138, "step": 96270 }, { "epoch": 61.36392606755896, "grad_norm": 0.2582796812057495, "learning_rate": 8.12388997535137e-05, "loss": 0.0128, "step": 96280 }, { "epoch": 61.37029955385596, "grad_norm": 0.8275569677352905, "learning_rate": 8.123459611948116e-05, "loss": 0.0201, "step": 96290 }, { "epoch": 61.376673040152966, "grad_norm": 0.46774613857269287, "learning_rate": 8.123029210591876e-05, "loss": 0.0345, "step": 96300 }, { "epoch": 61.38304652644997, "grad_norm": 0.1818963587284088, "learning_rate": 8.122598771287883e-05, "loss": 0.0216, "step": 96310 }, { "epoch": 61.389420012746974, "grad_norm": 0.5210548043251038, "learning_rate": 8.122168294041368e-05, "loss": 0.0197, "step": 96320 }, { "epoch": 61.395793499043975, "grad_norm": 0.306632936000824, "learning_rate": 8.12173777885756e-05, "loss": 0.0142, "step": 96330 }, { "epoch": 61.40216698534098, "grad_norm": 0.2799775302410126, "learning_rate": 8.12130722574169e-05, "loss": 0.0165, "step": 96340 }, { "epoch": 61.408540471637984, "grad_norm": 0.5510163307189941, "learning_rate": 8.120876634698991e-05, "loss": 0.0201, "step": 96350 }, { "epoch": 61.41491395793499, "grad_norm": 0.3885762393474579, "learning_rate": 8.120446005734694e-05, "loss": 0.041, "step": 96360 }, { "epoch": 61.42128744423199, "grad_norm": 0.3218502104282379, "learning_rate": 8.12001533885403e-05, "loss": 0.0202, "step": 96370 }, { "epoch": 61.427660930529, "grad_norm": 0.42520588636398315, "learning_rate": 8.119584634062237e-05, "loss": 0.0242, "step": 96380 }, { "epoch": 61.434034416826, "grad_norm": 0.3278224468231201, "learning_rate": 8.119153891364544e-05, "loss": 0.0167, "step": 96390 }, { "epoch": 61.44040790312301, "grad_norm": 0.40820711851119995, "learning_rate": 8.118723110766185e-05, "loss": 0.0179, "step": 96400 }, { "epoch": 61.44678138942001, "grad_norm": 0.441184937953949, "learning_rate": 8.118292292272399e-05, "loss": 0.0163, "step": 96410 }, { "epoch": 61.45315487571702, "grad_norm": 0.356674462556839, "learning_rate": 8.117861435888413e-05, "loss": 0.0137, "step": 96420 }, { "epoch": 61.45952836201402, "grad_norm": 0.6369248628616333, "learning_rate": 8.11743054161947e-05, "loss": 0.0276, "step": 96430 }, { "epoch": 61.46590184831103, "grad_norm": 0.42223960161209106, "learning_rate": 8.116999609470801e-05, "loss": 0.0137, "step": 96440 }, { "epoch": 61.47227533460803, "grad_norm": 0.3947093188762665, "learning_rate": 8.116568639447644e-05, "loss": 0.0202, "step": 96450 }, { "epoch": 61.478648820905036, "grad_norm": 0.3592786490917206, "learning_rate": 8.116137631555236e-05, "loss": 0.0216, "step": 96460 }, { "epoch": 61.48502230720204, "grad_norm": 0.3977213203907013, "learning_rate": 8.115706585798814e-05, "loss": 0.0188, "step": 96470 }, { "epoch": 61.491395793499045, "grad_norm": 0.3117183744907379, "learning_rate": 8.115275502183615e-05, "loss": 0.0137, "step": 96480 }, { "epoch": 61.497769279796046, "grad_norm": 0.5651962757110596, "learning_rate": 8.114844380714876e-05, "loss": 0.0336, "step": 96490 }, { "epoch": 61.504142766093054, "grad_norm": 0.286560595035553, "learning_rate": 8.114413221397838e-05, "loss": 0.0166, "step": 96500 }, { "epoch": 61.510516252390055, "grad_norm": 0.40250521898269653, "learning_rate": 8.11398202423774e-05, "loss": 0.0205, "step": 96510 }, { "epoch": 61.51688973868706, "grad_norm": 0.3812200129032135, "learning_rate": 8.113550789239818e-05, "loss": 0.0174, "step": 96520 }, { "epoch": 61.52326322498406, "grad_norm": 0.36307859420776367, "learning_rate": 8.113119516409316e-05, "loss": 0.0127, "step": 96530 }, { "epoch": 61.52963671128107, "grad_norm": 0.39085614681243896, "learning_rate": 8.112688205751472e-05, "loss": 0.0155, "step": 96540 }, { "epoch": 61.53601019757807, "grad_norm": 0.4594487249851227, "learning_rate": 8.112256857271528e-05, "loss": 0.0176, "step": 96550 }, { "epoch": 61.54238368387508, "grad_norm": 0.37602975964546204, "learning_rate": 8.111825470974724e-05, "loss": 0.016, "step": 96560 }, { "epoch": 61.54875717017208, "grad_norm": 0.2828028202056885, "learning_rate": 8.111394046866302e-05, "loss": 0.0161, "step": 96570 }, { "epoch": 61.55513065646909, "grad_norm": 0.32374995946884155, "learning_rate": 8.110962584951506e-05, "loss": 0.0196, "step": 96580 }, { "epoch": 61.56150414276609, "grad_norm": 0.607700765132904, "learning_rate": 8.110531085235578e-05, "loss": 0.0173, "step": 96590 }, { "epoch": 61.5678776290631, "grad_norm": 0.4178611636161804, "learning_rate": 8.110099547723757e-05, "loss": 0.0196, "step": 96600 }, { "epoch": 61.5742511153601, "grad_norm": 0.4291205108165741, "learning_rate": 8.109667972421293e-05, "loss": 0.0134, "step": 96610 }, { "epoch": 61.58062460165711, "grad_norm": 0.4629915654659271, "learning_rate": 8.109236359333426e-05, "loss": 0.0198, "step": 96620 }, { "epoch": 61.58699808795411, "grad_norm": 0.5891873836517334, "learning_rate": 8.108804708465403e-05, "loss": 0.0232, "step": 96630 }, { "epoch": 61.593371574251115, "grad_norm": 0.3979177474975586, "learning_rate": 8.108373019822466e-05, "loss": 0.0203, "step": 96640 }, { "epoch": 61.599745060548116, "grad_norm": 0.2961391508579254, "learning_rate": 8.107941293409863e-05, "loss": 0.0217, "step": 96650 }, { "epoch": 61.606118546845124, "grad_norm": 0.4202664792537689, "learning_rate": 8.107509529232838e-05, "loss": 0.0216, "step": 96660 }, { "epoch": 61.61249203314213, "grad_norm": 0.29675528407096863, "learning_rate": 8.107077727296639e-05, "loss": 0.0156, "step": 96670 }, { "epoch": 61.61886551943913, "grad_norm": 0.7197170257568359, "learning_rate": 8.106645887606512e-05, "loss": 0.0155, "step": 96680 }, { "epoch": 61.62523900573614, "grad_norm": 0.41031673550605774, "learning_rate": 8.106214010167704e-05, "loss": 0.0176, "step": 96690 }, { "epoch": 61.63161249203314, "grad_norm": 0.6684808135032654, "learning_rate": 8.105782094985464e-05, "loss": 0.0198, "step": 96700 }, { "epoch": 61.63798597833015, "grad_norm": 0.40795400738716125, "learning_rate": 8.105350142065037e-05, "loss": 0.0252, "step": 96710 }, { "epoch": 61.64435946462715, "grad_norm": 0.47454220056533813, "learning_rate": 8.104918151411676e-05, "loss": 0.0188, "step": 96720 }, { "epoch": 61.65073295092416, "grad_norm": 0.5199906229972839, "learning_rate": 8.104486123030627e-05, "loss": 0.0179, "step": 96730 }, { "epoch": 61.65710643722116, "grad_norm": 0.3327793478965759, "learning_rate": 8.104054056927139e-05, "loss": 0.0166, "step": 96740 }, { "epoch": 61.66347992351817, "grad_norm": 0.3099733293056488, "learning_rate": 8.103621953106466e-05, "loss": 0.0239, "step": 96750 }, { "epoch": 61.66985340981517, "grad_norm": 0.18611934781074524, "learning_rate": 8.103189811573855e-05, "loss": 0.0227, "step": 96760 }, { "epoch": 61.676226896112176, "grad_norm": 0.4122558534145355, "learning_rate": 8.102757632334558e-05, "loss": 0.0232, "step": 96770 }, { "epoch": 61.68260038240918, "grad_norm": 0.5746716260910034, "learning_rate": 8.102325415393826e-05, "loss": 0.0251, "step": 96780 }, { "epoch": 61.688973868706185, "grad_norm": 0.5040014982223511, "learning_rate": 8.101893160756913e-05, "loss": 0.0191, "step": 96790 }, { "epoch": 61.695347355003186, "grad_norm": 0.48117077350616455, "learning_rate": 8.101460868429067e-05, "loss": 0.0241, "step": 96800 }, { "epoch": 61.701720841300194, "grad_norm": 0.40665462613105774, "learning_rate": 8.101028538415543e-05, "loss": 0.0187, "step": 96810 }, { "epoch": 61.708094327597195, "grad_norm": 0.6190661191940308, "learning_rate": 8.100596170721597e-05, "loss": 0.0184, "step": 96820 }, { "epoch": 61.7144678138942, "grad_norm": 0.3660863935947418, "learning_rate": 8.100163765352478e-05, "loss": 0.0246, "step": 96830 }, { "epoch": 61.7208413001912, "grad_norm": 0.25920796394348145, "learning_rate": 8.099731322313442e-05, "loss": 0.0143, "step": 96840 }, { "epoch": 61.72721478648821, "grad_norm": 0.9363122582435608, "learning_rate": 8.099298841609745e-05, "loss": 0.0198, "step": 96850 }, { "epoch": 61.73358827278521, "grad_norm": 0.6316894292831421, "learning_rate": 8.098866323246642e-05, "loss": 0.023, "step": 96860 }, { "epoch": 61.73996175908222, "grad_norm": 0.603313148021698, "learning_rate": 8.098433767229384e-05, "loss": 0.0217, "step": 96870 }, { "epoch": 61.74633524537922, "grad_norm": 0.6552619934082031, "learning_rate": 8.098001173563232e-05, "loss": 0.0146, "step": 96880 }, { "epoch": 61.75270873167623, "grad_norm": 0.37630438804626465, "learning_rate": 8.097568542253441e-05, "loss": 0.0241, "step": 96890 }, { "epoch": 61.75908221797323, "grad_norm": 0.33536455035209656, "learning_rate": 8.097135873305267e-05, "loss": 0.016, "step": 96900 }, { "epoch": 61.76545570427024, "grad_norm": 0.4042932689189911, "learning_rate": 8.096703166723968e-05, "loss": 0.0192, "step": 96910 }, { "epoch": 61.77182919056724, "grad_norm": 0.6125134229660034, "learning_rate": 8.096270422514802e-05, "loss": 0.016, "step": 96920 }, { "epoch": 61.77820267686425, "grad_norm": 0.46857771277427673, "learning_rate": 8.095837640683029e-05, "loss": 0.0237, "step": 96930 }, { "epoch": 61.78457616316125, "grad_norm": 0.5416263341903687, "learning_rate": 8.095404821233903e-05, "loss": 0.0216, "step": 96940 }, { "epoch": 61.790949649458256, "grad_norm": 0.28832900524139404, "learning_rate": 8.094971964172687e-05, "loss": 0.0185, "step": 96950 }, { "epoch": 61.797323135755256, "grad_norm": 0.3597073554992676, "learning_rate": 8.094539069504639e-05, "loss": 0.0192, "step": 96960 }, { "epoch": 61.803696622052264, "grad_norm": 0.3857431411743164, "learning_rate": 8.094106137235019e-05, "loss": 0.0274, "step": 96970 }, { "epoch": 61.810070108349265, "grad_norm": 0.4957766830921173, "learning_rate": 8.093673167369089e-05, "loss": 0.0199, "step": 96980 }, { "epoch": 61.81644359464627, "grad_norm": 0.33757513761520386, "learning_rate": 8.093240159912108e-05, "loss": 0.0241, "step": 96990 }, { "epoch": 61.822817080943274, "grad_norm": 0.42400628328323364, "learning_rate": 8.09280711486934e-05, "loss": 0.0176, "step": 97000 }, { "epoch": 61.82919056724028, "grad_norm": 0.33330148458480835, "learning_rate": 8.092374032246045e-05, "loss": 0.0156, "step": 97010 }, { "epoch": 61.83556405353728, "grad_norm": 0.38017845153808594, "learning_rate": 8.091940912047486e-05, "loss": 0.0186, "step": 97020 }, { "epoch": 61.84193753983429, "grad_norm": 0.472946435213089, "learning_rate": 8.091507754278927e-05, "loss": 0.0275, "step": 97030 }, { "epoch": 61.84831102613129, "grad_norm": 0.1467377245426178, "learning_rate": 8.091074558945626e-05, "loss": 0.0155, "step": 97040 }, { "epoch": 61.8546845124283, "grad_norm": 0.758780837059021, "learning_rate": 8.090641326052854e-05, "loss": 0.0159, "step": 97050 }, { "epoch": 61.8610579987253, "grad_norm": 0.573049008846283, "learning_rate": 8.09020805560587e-05, "loss": 0.0263, "step": 97060 }, { "epoch": 61.86743148502231, "grad_norm": 0.38568615913391113, "learning_rate": 8.089774747609942e-05, "loss": 0.0118, "step": 97070 }, { "epoch": 61.87380497131931, "grad_norm": 0.2276342213153839, "learning_rate": 8.089341402070331e-05, "loss": 0.0124, "step": 97080 }, { "epoch": 61.88017845761632, "grad_norm": 0.8106245398521423, "learning_rate": 8.088908018992309e-05, "loss": 0.0238, "step": 97090 }, { "epoch": 61.88655194391332, "grad_norm": 0.39384767413139343, "learning_rate": 8.088474598381135e-05, "loss": 0.0239, "step": 97100 }, { "epoch": 61.892925430210326, "grad_norm": 0.428750216960907, "learning_rate": 8.088041140242081e-05, "loss": 0.016, "step": 97110 }, { "epoch": 61.89929891650733, "grad_norm": 0.5590060353279114, "learning_rate": 8.08760764458041e-05, "loss": 0.0223, "step": 97120 }, { "epoch": 61.905672402804335, "grad_norm": 0.3726309537887573, "learning_rate": 8.087174111401393e-05, "loss": 0.0233, "step": 97130 }, { "epoch": 61.912045889101336, "grad_norm": 0.5455565452575684, "learning_rate": 8.086740540710293e-05, "loss": 0.0222, "step": 97140 }, { "epoch": 61.918419375398344, "grad_norm": 0.5680257081985474, "learning_rate": 8.086306932512382e-05, "loss": 0.0167, "step": 97150 }, { "epoch": 61.924792861695344, "grad_norm": 0.4302905201911926, "learning_rate": 8.085873286812928e-05, "loss": 0.016, "step": 97160 }, { "epoch": 61.93116634799235, "grad_norm": 0.5440407991409302, "learning_rate": 8.085439603617199e-05, "loss": 0.0339, "step": 97170 }, { "epoch": 61.93753983428935, "grad_norm": 0.6203112602233887, "learning_rate": 8.085005882930468e-05, "loss": 0.0149, "step": 97180 }, { "epoch": 61.94391332058636, "grad_norm": 0.38591358065605164, "learning_rate": 8.084572124758001e-05, "loss": 0.0167, "step": 97190 }, { "epoch": 61.95028680688336, "grad_norm": 0.38126569986343384, "learning_rate": 8.084138329105071e-05, "loss": 0.0208, "step": 97200 }, { "epoch": 61.95666029318037, "grad_norm": 0.25042882561683655, "learning_rate": 8.08370449597695e-05, "loss": 0.0169, "step": 97210 }, { "epoch": 61.96303377947737, "grad_norm": 0.4394133687019348, "learning_rate": 8.083270625378905e-05, "loss": 0.0243, "step": 97220 }, { "epoch": 61.96940726577438, "grad_norm": 0.6699545979499817, "learning_rate": 8.08283671731621e-05, "loss": 0.0195, "step": 97230 }, { "epoch": 61.97578075207138, "grad_norm": 0.2697429060935974, "learning_rate": 8.082402771794141e-05, "loss": 0.0136, "step": 97240 }, { "epoch": 61.98215423836839, "grad_norm": 0.5341203808784485, "learning_rate": 8.081968788817967e-05, "loss": 0.023, "step": 97250 }, { "epoch": 61.98852772466539, "grad_norm": 0.29507946968078613, "learning_rate": 8.081534768392963e-05, "loss": 0.0167, "step": 97260 }, { "epoch": 61.9949012109624, "grad_norm": 0.3525209426879883, "learning_rate": 8.081100710524401e-05, "loss": 0.0208, "step": 97270 }, { "epoch": 62.001274697259404, "grad_norm": 0.5168564915657043, "learning_rate": 8.080666615217557e-05, "loss": 0.0248, "step": 97280 }, { "epoch": 62.007648183556405, "grad_norm": 0.5288978219032288, "learning_rate": 8.080232482477705e-05, "loss": 0.0354, "step": 97290 }, { "epoch": 62.01402166985341, "grad_norm": 0.38537314534187317, "learning_rate": 8.07979831231012e-05, "loss": 0.0167, "step": 97300 }, { "epoch": 62.020395156150414, "grad_norm": 0.38581278920173645, "learning_rate": 8.079364104720078e-05, "loss": 0.0182, "step": 97310 }, { "epoch": 62.02676864244742, "grad_norm": 0.3052176833152771, "learning_rate": 8.078929859712855e-05, "loss": 0.0297, "step": 97320 }, { "epoch": 62.03314212874442, "grad_norm": 0.3487243056297302, "learning_rate": 8.078495577293727e-05, "loss": 0.025, "step": 97330 }, { "epoch": 62.03951561504143, "grad_norm": 0.6181469559669495, "learning_rate": 8.07806125746797e-05, "loss": 0.0265, "step": 97340 }, { "epoch": 62.04588910133843, "grad_norm": 0.468288391828537, "learning_rate": 8.077626900240865e-05, "loss": 0.0239, "step": 97350 }, { "epoch": 62.05226258763544, "grad_norm": 0.6747252345085144, "learning_rate": 8.077192505617684e-05, "loss": 0.0163, "step": 97360 }, { "epoch": 62.05863607393244, "grad_norm": 0.26907429099082947, "learning_rate": 8.076758073603712e-05, "loss": 0.02, "step": 97370 }, { "epoch": 62.06500956022945, "grad_norm": 0.43044260144233704, "learning_rate": 8.076323604204222e-05, "loss": 0.0153, "step": 97380 }, { "epoch": 62.07138304652645, "grad_norm": 0.20085649192333221, "learning_rate": 8.075889097424496e-05, "loss": 0.0169, "step": 97390 }, { "epoch": 62.07775653282346, "grad_norm": 0.5214897990226746, "learning_rate": 8.075454553269815e-05, "loss": 0.0159, "step": 97400 }, { "epoch": 62.08413001912046, "grad_norm": 0.31792503595352173, "learning_rate": 8.075019971745457e-05, "loss": 0.0371, "step": 97410 }, { "epoch": 62.090503505417466, "grad_norm": 0.27119511365890503, "learning_rate": 8.074585352856701e-05, "loss": 0.0196, "step": 97420 }, { "epoch": 62.09687699171447, "grad_norm": 0.38794219493865967, "learning_rate": 8.074150696608832e-05, "loss": 0.0172, "step": 97430 }, { "epoch": 62.103250478011475, "grad_norm": 0.4796970784664154, "learning_rate": 8.07371600300713e-05, "loss": 0.0295, "step": 97440 }, { "epoch": 62.109623964308476, "grad_norm": 0.5864056348800659, "learning_rate": 8.073281272056874e-05, "loss": 0.0178, "step": 97450 }, { "epoch": 62.115997450605484, "grad_norm": 0.7065044045448303, "learning_rate": 8.072846503763351e-05, "loss": 0.0283, "step": 97460 }, { "epoch": 62.122370936902485, "grad_norm": 0.4096756875514984, "learning_rate": 8.072411698131841e-05, "loss": 0.0217, "step": 97470 }, { "epoch": 62.12874442319949, "grad_norm": 0.8006265163421631, "learning_rate": 8.071976855167629e-05, "loss": 0.0207, "step": 97480 }, { "epoch": 62.13511790949649, "grad_norm": 0.5989569425582886, "learning_rate": 8.071541974875996e-05, "loss": 0.0184, "step": 97490 }, { "epoch": 62.1414913957935, "grad_norm": 0.7030155062675476, "learning_rate": 8.071107057262231e-05, "loss": 0.0187, "step": 97500 }, { "epoch": 62.1478648820905, "grad_norm": 0.8609601259231567, "learning_rate": 8.070672102331614e-05, "loss": 0.0272, "step": 97510 }, { "epoch": 62.15423836838751, "grad_norm": 0.6227999925613403, "learning_rate": 8.07023711008943e-05, "loss": 0.0137, "step": 97520 }, { "epoch": 62.16061185468451, "grad_norm": 0.42242833971977234, "learning_rate": 8.069802080540969e-05, "loss": 0.0229, "step": 97530 }, { "epoch": 62.16698534098152, "grad_norm": 0.42195412516593933, "learning_rate": 8.069367013691512e-05, "loss": 0.0196, "step": 97540 }, { "epoch": 62.17335882727852, "grad_norm": 0.44220268726348877, "learning_rate": 8.068931909546349e-05, "loss": 0.0236, "step": 97550 }, { "epoch": 62.17973231357553, "grad_norm": 0.402554988861084, "learning_rate": 8.068496768110766e-05, "loss": 0.0186, "step": 97560 }, { "epoch": 62.18610579987253, "grad_norm": 0.467043399810791, "learning_rate": 8.06806158939005e-05, "loss": 0.0398, "step": 97570 }, { "epoch": 62.19247928616954, "grad_norm": 0.502907395362854, "learning_rate": 8.067626373389488e-05, "loss": 0.0282, "step": 97580 }, { "epoch": 62.19885277246654, "grad_norm": 0.4417386054992676, "learning_rate": 8.06719112011437e-05, "loss": 0.0159, "step": 97590 }, { "epoch": 62.205226258763545, "grad_norm": 0.6036456823348999, "learning_rate": 8.066755829569985e-05, "loss": 0.0267, "step": 97600 }, { "epoch": 62.211599745060546, "grad_norm": 0.7202828526496887, "learning_rate": 8.066320501761619e-05, "loss": 0.0262, "step": 97610 }, { "epoch": 62.217973231357554, "grad_norm": 0.3794562816619873, "learning_rate": 8.065885136694566e-05, "loss": 0.015, "step": 97620 }, { "epoch": 62.224346717654555, "grad_norm": 0.47899314761161804, "learning_rate": 8.065449734374113e-05, "loss": 0.0196, "step": 97630 }, { "epoch": 62.23072020395156, "grad_norm": 0.4210587739944458, "learning_rate": 8.065014294805551e-05, "loss": 0.0132, "step": 97640 }, { "epoch": 62.237093690248564, "grad_norm": 0.6893230080604553, "learning_rate": 8.064578817994172e-05, "loss": 0.017, "step": 97650 }, { "epoch": 62.24346717654557, "grad_norm": 0.4412880837917328, "learning_rate": 8.064143303945266e-05, "loss": 0.0178, "step": 97660 }, { "epoch": 62.24984066284257, "grad_norm": 0.32984164357185364, "learning_rate": 8.063707752664127e-05, "loss": 0.0191, "step": 97670 }, { "epoch": 62.25621414913958, "grad_norm": 0.20432505011558533, "learning_rate": 8.063272164156046e-05, "loss": 0.0245, "step": 97680 }, { "epoch": 62.26258763543658, "grad_norm": 0.5770242214202881, "learning_rate": 8.062836538426316e-05, "loss": 0.0224, "step": 97690 }, { "epoch": 62.26896112173359, "grad_norm": 0.3613213300704956, "learning_rate": 8.06240087548023e-05, "loss": 0.0351, "step": 97700 }, { "epoch": 62.27533460803059, "grad_norm": 0.3641296923160553, "learning_rate": 8.061965175323083e-05, "loss": 0.0317, "step": 97710 }, { "epoch": 62.2817080943276, "grad_norm": 0.3477620482444763, "learning_rate": 8.061529437960167e-05, "loss": 0.0228, "step": 97720 }, { "epoch": 62.2880815806246, "grad_norm": 0.587247371673584, "learning_rate": 8.061093663396778e-05, "loss": 0.016, "step": 97730 }, { "epoch": 62.29445506692161, "grad_norm": 0.3307460844516754, "learning_rate": 8.060657851638209e-05, "loss": 0.0137, "step": 97740 }, { "epoch": 62.30082855321861, "grad_norm": 0.33146101236343384, "learning_rate": 8.06022200268976e-05, "loss": 0.0194, "step": 97750 }, { "epoch": 62.307202039515616, "grad_norm": 0.348911315202713, "learning_rate": 8.059786116556724e-05, "loss": 0.0156, "step": 97760 }, { "epoch": 62.31357552581262, "grad_norm": 0.2472749948501587, "learning_rate": 8.059350193244396e-05, "loss": 0.0144, "step": 97770 }, { "epoch": 62.319949012109625, "grad_norm": 0.3716954290866852, "learning_rate": 8.058914232758078e-05, "loss": 0.0223, "step": 97780 }, { "epoch": 62.326322498406626, "grad_norm": 0.5336238741874695, "learning_rate": 8.05847823510306e-05, "loss": 0.0201, "step": 97790 }, { "epoch": 62.33269598470363, "grad_norm": 0.30397364497184753, "learning_rate": 8.058042200284645e-05, "loss": 0.0191, "step": 97800 }, { "epoch": 62.339069471000634, "grad_norm": 0.27326685190200806, "learning_rate": 8.057606128308132e-05, "loss": 0.0177, "step": 97810 }, { "epoch": 62.34544295729764, "grad_norm": 0.40142396092414856, "learning_rate": 8.057170019178817e-05, "loss": 0.0244, "step": 97820 }, { "epoch": 62.35181644359464, "grad_norm": 0.1699044406414032, "learning_rate": 8.056733872901997e-05, "loss": 0.0187, "step": 97830 }, { "epoch": 62.35818992989165, "grad_norm": 0.3968382477760315, "learning_rate": 8.056297689482978e-05, "loss": 0.0204, "step": 97840 }, { "epoch": 62.36456341618865, "grad_norm": 0.25534868240356445, "learning_rate": 8.055861468927055e-05, "loss": 0.025, "step": 97850 }, { "epoch": 62.37093690248566, "grad_norm": 0.5727357268333435, "learning_rate": 8.055425211239529e-05, "loss": 0.011, "step": 97860 }, { "epoch": 62.37731038878266, "grad_norm": 0.568152129650116, "learning_rate": 8.054988916425702e-05, "loss": 0.0245, "step": 97870 }, { "epoch": 62.38368387507967, "grad_norm": 1.2434951066970825, "learning_rate": 8.054552584490878e-05, "loss": 0.0205, "step": 97880 }, { "epoch": 62.39005736137667, "grad_norm": 0.45836806297302246, "learning_rate": 8.054116215440354e-05, "loss": 0.03, "step": 97890 }, { "epoch": 62.39643084767368, "grad_norm": 0.1967131495475769, "learning_rate": 8.053679809279434e-05, "loss": 0.0258, "step": 97900 }, { "epoch": 62.40280433397068, "grad_norm": 0.4425431191921234, "learning_rate": 8.053243366013422e-05, "loss": 0.0155, "step": 97910 }, { "epoch": 62.409177820267686, "grad_norm": 0.5517053604125977, "learning_rate": 8.052806885647621e-05, "loss": 0.0232, "step": 97920 }, { "epoch": 62.415551306564694, "grad_norm": 0.40846145153045654, "learning_rate": 8.052370368187332e-05, "loss": 0.023, "step": 97930 }, { "epoch": 62.421924792861695, "grad_norm": 0.46480873227119446, "learning_rate": 8.051933813637863e-05, "loss": 0.0199, "step": 97940 }, { "epoch": 62.4282982791587, "grad_norm": 0.18525362014770508, "learning_rate": 8.051497222004516e-05, "loss": 0.0187, "step": 97950 }, { "epoch": 62.434671765455704, "grad_norm": 0.3228902518749237, "learning_rate": 8.051060593292596e-05, "loss": 0.0221, "step": 97960 }, { "epoch": 62.44104525175271, "grad_norm": 0.316935658454895, "learning_rate": 8.05062392750741e-05, "loss": 0.0191, "step": 97970 }, { "epoch": 62.44741873804971, "grad_norm": 0.35347357392311096, "learning_rate": 8.050187224654263e-05, "loss": 0.0159, "step": 97980 }, { "epoch": 62.45379222434672, "grad_norm": 0.36566323041915894, "learning_rate": 8.049750484738461e-05, "loss": 0.0169, "step": 97990 }, { "epoch": 62.46016571064372, "grad_norm": 0.3364529311656952, "learning_rate": 8.049313707765312e-05, "loss": 0.0128, "step": 98000 }, { "epoch": 62.46653919694073, "grad_norm": 0.21278861165046692, "learning_rate": 8.048876893740121e-05, "loss": 0.0203, "step": 98010 }, { "epoch": 62.47291268323773, "grad_norm": 0.3901107907295227, "learning_rate": 8.048440042668198e-05, "loss": 0.0301, "step": 98020 }, { "epoch": 62.47928616953474, "grad_norm": 0.24456846714019775, "learning_rate": 8.048003154554851e-05, "loss": 0.0124, "step": 98030 }, { "epoch": 62.48565965583174, "grad_norm": 0.26253634691238403, "learning_rate": 8.047566229405386e-05, "loss": 0.0171, "step": 98040 }, { "epoch": 62.49203314212875, "grad_norm": 0.3108423054218292, "learning_rate": 8.047129267225115e-05, "loss": 0.0153, "step": 98050 }, { "epoch": 62.49840662842575, "grad_norm": 0.2625300884246826, "learning_rate": 8.046692268019345e-05, "loss": 0.0175, "step": 98060 }, { "epoch": 62.504780114722756, "grad_norm": 0.42729923129081726, "learning_rate": 8.046255231793389e-05, "loss": 0.0232, "step": 98070 }, { "epoch": 62.51115360101976, "grad_norm": 0.7011959552764893, "learning_rate": 8.045818158552555e-05, "loss": 0.0181, "step": 98080 }, { "epoch": 62.517527087316765, "grad_norm": 0.49554136395454407, "learning_rate": 8.045381048302157e-05, "loss": 0.0175, "step": 98090 }, { "epoch": 62.523900573613766, "grad_norm": 0.4114202857017517, "learning_rate": 8.044943901047502e-05, "loss": 0.0204, "step": 98100 }, { "epoch": 62.530274059910774, "grad_norm": 0.4409479796886444, "learning_rate": 8.044506716793904e-05, "loss": 0.0262, "step": 98110 }, { "epoch": 62.536647546207774, "grad_norm": 0.46096691489219666, "learning_rate": 8.044069495546674e-05, "loss": 0.0161, "step": 98120 }, { "epoch": 62.54302103250478, "grad_norm": 0.45059147477149963, "learning_rate": 8.043632237311127e-05, "loss": 0.0159, "step": 98130 }, { "epoch": 62.54939451880178, "grad_norm": 0.9806071519851685, "learning_rate": 8.043194942092574e-05, "loss": 0.0302, "step": 98140 }, { "epoch": 62.55576800509879, "grad_norm": 0.2080533653497696, "learning_rate": 8.042757609896327e-05, "loss": 0.0225, "step": 98150 }, { "epoch": 62.56214149139579, "grad_norm": 0.42297086119651794, "learning_rate": 8.042320240727705e-05, "loss": 0.0186, "step": 98160 }, { "epoch": 62.5685149776928, "grad_norm": 0.3181159496307373, "learning_rate": 8.04188283459202e-05, "loss": 0.0186, "step": 98170 }, { "epoch": 62.5748884639898, "grad_norm": 0.34635987877845764, "learning_rate": 8.041445391494584e-05, "loss": 0.0227, "step": 98180 }, { "epoch": 62.58126195028681, "grad_norm": 0.491744726896286, "learning_rate": 8.041007911440718e-05, "loss": 0.0199, "step": 98190 }, { "epoch": 62.58763543658381, "grad_norm": 0.4063069820404053, "learning_rate": 8.040570394435733e-05, "loss": 0.0224, "step": 98200 }, { "epoch": 62.59400892288082, "grad_norm": 0.39454934000968933, "learning_rate": 8.040132840484946e-05, "loss": 0.0172, "step": 98210 }, { "epoch": 62.60038240917782, "grad_norm": 0.2943691611289978, "learning_rate": 8.039695249593676e-05, "loss": 0.0172, "step": 98220 }, { "epoch": 62.606755895474826, "grad_norm": 0.478152334690094, "learning_rate": 8.039257621767238e-05, "loss": 0.0197, "step": 98230 }, { "epoch": 62.61312938177183, "grad_norm": 0.3306417763233185, "learning_rate": 8.038819957010952e-05, "loss": 0.0195, "step": 98240 }, { "epoch": 62.619502868068835, "grad_norm": 0.28508543968200684, "learning_rate": 8.038382255330133e-05, "loss": 0.0172, "step": 98250 }, { "epoch": 62.625876354365836, "grad_norm": 0.2696250379085541, "learning_rate": 8.037944516730101e-05, "loss": 0.0185, "step": 98260 }, { "epoch": 62.632249840662844, "grad_norm": 0.4812609851360321, "learning_rate": 8.037506741216174e-05, "loss": 0.0178, "step": 98270 }, { "epoch": 62.638623326959845, "grad_norm": 0.5007364749908447, "learning_rate": 8.037068928793674e-05, "loss": 0.0144, "step": 98280 }, { "epoch": 62.64499681325685, "grad_norm": 0.2383960783481598, "learning_rate": 8.036631079467917e-05, "loss": 0.0145, "step": 98290 }, { "epoch": 62.651370299553854, "grad_norm": 0.3832099437713623, "learning_rate": 8.036193193244228e-05, "loss": 0.0165, "step": 98300 }, { "epoch": 62.65774378585086, "grad_norm": 0.6132698655128479, "learning_rate": 8.035755270127922e-05, "loss": 0.0168, "step": 98310 }, { "epoch": 62.66411727214786, "grad_norm": 0.3570716679096222, "learning_rate": 8.035317310124327e-05, "loss": 0.0169, "step": 98320 }, { "epoch": 62.67049075844487, "grad_norm": 0.2383190095424652, "learning_rate": 8.034879313238758e-05, "loss": 0.0182, "step": 98330 }, { "epoch": 62.67686424474187, "grad_norm": 0.30801406502723694, "learning_rate": 8.03444127947654e-05, "loss": 0.0124, "step": 98340 }, { "epoch": 62.68323773103888, "grad_norm": 0.3085496723651886, "learning_rate": 8.034003208842997e-05, "loss": 0.0237, "step": 98350 }, { "epoch": 62.68961121733588, "grad_norm": 0.47724223136901855, "learning_rate": 8.033565101343448e-05, "loss": 0.0273, "step": 98360 }, { "epoch": 62.69598470363289, "grad_norm": 0.2284173220396042, "learning_rate": 8.033126956983221e-05, "loss": 0.0235, "step": 98370 }, { "epoch": 62.70235818992989, "grad_norm": 0.6044796109199524, "learning_rate": 8.032688775767637e-05, "loss": 0.0304, "step": 98380 }, { "epoch": 62.7087316762269, "grad_norm": 0.3556976020336151, "learning_rate": 8.032250557702021e-05, "loss": 0.017, "step": 98390 }, { "epoch": 62.7151051625239, "grad_norm": 0.36444106698036194, "learning_rate": 8.0318123027917e-05, "loss": 0.0263, "step": 98400 }, { "epoch": 62.721478648820906, "grad_norm": 0.33993998169898987, "learning_rate": 8.031374011041993e-05, "loss": 0.0179, "step": 98410 }, { "epoch": 62.72785213511791, "grad_norm": 1.283555269241333, "learning_rate": 8.030935682458231e-05, "loss": 0.0344, "step": 98420 }, { "epoch": 62.734225621414915, "grad_norm": 0.2985171675682068, "learning_rate": 8.03049731704574e-05, "loss": 0.0188, "step": 98430 }, { "epoch": 62.740599107711915, "grad_norm": 0.7376689314842224, "learning_rate": 8.030058914809844e-05, "loss": 0.018, "step": 98440 }, { "epoch": 62.74697259400892, "grad_norm": 0.5051977038383484, "learning_rate": 8.029620475755872e-05, "loss": 0.035, "step": 98450 }, { "epoch": 62.753346080305924, "grad_norm": 0.4614267945289612, "learning_rate": 8.029181999889152e-05, "loss": 0.0126, "step": 98460 }, { "epoch": 62.75971956660293, "grad_norm": 0.37160399556159973, "learning_rate": 8.028743487215009e-05, "loss": 0.0176, "step": 98470 }, { "epoch": 62.76609305289993, "grad_norm": 0.37515145540237427, "learning_rate": 8.028304937738773e-05, "loss": 0.02, "step": 98480 }, { "epoch": 62.77246653919694, "grad_norm": 0.33021432161331177, "learning_rate": 8.027866351465775e-05, "loss": 0.0314, "step": 98490 }, { "epoch": 62.77884002549394, "grad_norm": 0.47927749156951904, "learning_rate": 8.02742772840134e-05, "loss": 0.0155, "step": 98500 }, { "epoch": 62.78521351179095, "grad_norm": 0.6512821316719055, "learning_rate": 8.026989068550801e-05, "loss": 0.0283, "step": 98510 }, { "epoch": 62.79158699808795, "grad_norm": 0.29518041014671326, "learning_rate": 8.026550371919485e-05, "loss": 0.0199, "step": 98520 }, { "epoch": 62.79796048438496, "grad_norm": 0.48598840832710266, "learning_rate": 8.026111638512725e-05, "loss": 0.0169, "step": 98530 }, { "epoch": 62.80433397068197, "grad_norm": 0.48206278681755066, "learning_rate": 8.025672868335851e-05, "loss": 0.0309, "step": 98540 }, { "epoch": 62.81070745697897, "grad_norm": 0.26152753829956055, "learning_rate": 8.025234061394198e-05, "loss": 0.0129, "step": 98550 }, { "epoch": 62.817080943275975, "grad_norm": 0.5359047651290894, "learning_rate": 8.024795217693094e-05, "loss": 0.0144, "step": 98560 }, { "epoch": 62.823454429572976, "grad_norm": 0.4500282406806946, "learning_rate": 8.024356337237868e-05, "loss": 0.0197, "step": 98570 }, { "epoch": 62.829827915869984, "grad_norm": 0.3998371660709381, "learning_rate": 8.023917420033862e-05, "loss": 0.0245, "step": 98580 }, { "epoch": 62.836201402166985, "grad_norm": 0.4074002504348755, "learning_rate": 8.0234784660864e-05, "loss": 0.0194, "step": 98590 }, { "epoch": 62.84257488846399, "grad_norm": 0.33811888098716736, "learning_rate": 8.023039475400825e-05, "loss": 0.016, "step": 98600 }, { "epoch": 62.848948374760994, "grad_norm": 0.5253702402114868, "learning_rate": 8.022600447982462e-05, "loss": 0.0161, "step": 98610 }, { "epoch": 62.855321861058, "grad_norm": 0.668700098991394, "learning_rate": 8.022161383836652e-05, "loss": 0.0212, "step": 98620 }, { "epoch": 62.861695347355, "grad_norm": 0.7551515698432922, "learning_rate": 8.021722282968727e-05, "loss": 0.0218, "step": 98630 }, { "epoch": 62.86806883365201, "grad_norm": 0.4215162992477417, "learning_rate": 8.021283145384022e-05, "loss": 0.0188, "step": 98640 }, { "epoch": 62.87444231994901, "grad_norm": 0.5689658522605896, "learning_rate": 8.020843971087877e-05, "loss": 0.0194, "step": 98650 }, { "epoch": 62.88081580624602, "grad_norm": 0.5398138165473938, "learning_rate": 8.020404760085622e-05, "loss": 0.0191, "step": 98660 }, { "epoch": 62.88718929254302, "grad_norm": 0.4774570167064667, "learning_rate": 8.0199655123826e-05, "loss": 0.0193, "step": 98670 }, { "epoch": 62.89356277884003, "grad_norm": 0.3100110590457916, "learning_rate": 8.019526227984147e-05, "loss": 0.0209, "step": 98680 }, { "epoch": 62.89993626513703, "grad_norm": 0.5663490891456604, "learning_rate": 8.019086906895596e-05, "loss": 0.0256, "step": 98690 }, { "epoch": 62.90630975143404, "grad_norm": 0.5103445053100586, "learning_rate": 8.018647549122291e-05, "loss": 0.0267, "step": 98700 }, { "epoch": 62.91268323773104, "grad_norm": 0.4636227488517761, "learning_rate": 8.018208154669568e-05, "loss": 0.0198, "step": 98710 }, { "epoch": 62.919056724028046, "grad_norm": 0.283550500869751, "learning_rate": 8.017768723542765e-05, "loss": 0.0146, "step": 98720 }, { "epoch": 62.92543021032505, "grad_norm": 0.3818202614784241, "learning_rate": 8.017329255747223e-05, "loss": 0.0163, "step": 98730 }, { "epoch": 62.931803696622055, "grad_norm": 0.20966285467147827, "learning_rate": 8.016889751288283e-05, "loss": 0.02, "step": 98740 }, { "epoch": 62.938177182919055, "grad_norm": 0.35521215200424194, "learning_rate": 8.016450210171283e-05, "loss": 0.0242, "step": 98750 }, { "epoch": 62.94455066921606, "grad_norm": 0.6651269197463989, "learning_rate": 8.016010632401565e-05, "loss": 0.0247, "step": 98760 }, { "epoch": 62.950924155513064, "grad_norm": 0.539497971534729, "learning_rate": 8.015571017984471e-05, "loss": 0.0228, "step": 98770 }, { "epoch": 62.95729764181007, "grad_norm": 0.33614158630371094, "learning_rate": 8.015131366925341e-05, "loss": 0.019, "step": 98780 }, { "epoch": 62.96367112810707, "grad_norm": 0.4557303488254547, "learning_rate": 8.014691679229518e-05, "loss": 0.0319, "step": 98790 }, { "epoch": 62.97004461440408, "grad_norm": 0.2390744686126709, "learning_rate": 8.014251954902347e-05, "loss": 0.018, "step": 98800 }, { "epoch": 62.97641810070108, "grad_norm": 0.4323665201663971, "learning_rate": 8.013812193949166e-05, "loss": 0.0132, "step": 98810 }, { "epoch": 62.98279158699809, "grad_norm": 0.7057623863220215, "learning_rate": 8.013372396375323e-05, "loss": 0.0216, "step": 98820 }, { "epoch": 62.98916507329509, "grad_norm": 0.21151228249073029, "learning_rate": 8.012932562186159e-05, "loss": 0.02, "step": 98830 }, { "epoch": 62.9955385595921, "grad_norm": 0.37299472093582153, "learning_rate": 8.012492691387021e-05, "loss": 0.0147, "step": 98840 }, { "epoch": 63.0019120458891, "grad_norm": 0.5209152102470398, "learning_rate": 8.012052783983252e-05, "loss": 0.017, "step": 98850 }, { "epoch": 63.00828553218611, "grad_norm": 0.3561956286430359, "learning_rate": 8.011612839980198e-05, "loss": 0.0138, "step": 98860 }, { "epoch": 63.01465901848311, "grad_norm": 0.5387434363365173, "learning_rate": 8.011172859383204e-05, "loss": 0.0298, "step": 98870 }, { "epoch": 63.021032504780116, "grad_norm": 0.5193247199058533, "learning_rate": 8.010732842197616e-05, "loss": 0.0273, "step": 98880 }, { "epoch": 63.02740599107712, "grad_norm": 0.35978639125823975, "learning_rate": 8.010292788428784e-05, "loss": 0.0182, "step": 98890 }, { "epoch": 63.033779477374125, "grad_norm": 0.2811342775821686, "learning_rate": 8.00985269808205e-05, "loss": 0.0119, "step": 98900 }, { "epoch": 63.040152963671126, "grad_norm": 0.6008613705635071, "learning_rate": 8.009412571162765e-05, "loss": 0.029, "step": 98910 }, { "epoch": 63.046526449968134, "grad_norm": 0.4641428291797638, "learning_rate": 8.008972407676277e-05, "loss": 0.0209, "step": 98920 }, { "epoch": 63.052899936265135, "grad_norm": 0.3382489085197449, "learning_rate": 8.008532207627932e-05, "loss": 0.0183, "step": 98930 }, { "epoch": 63.05927342256214, "grad_norm": 0.5956593155860901, "learning_rate": 8.008091971023079e-05, "loss": 0.0147, "step": 98940 }, { "epoch": 63.065646908859144, "grad_norm": 0.32215505838394165, "learning_rate": 8.007651697867069e-05, "loss": 0.0107, "step": 98950 }, { "epoch": 63.07202039515615, "grad_norm": 0.2750360667705536, "learning_rate": 8.007211388165252e-05, "loss": 0.013, "step": 98960 }, { "epoch": 63.07839388145315, "grad_norm": 0.2838267982006073, "learning_rate": 8.006771041922978e-05, "loss": 0.0184, "step": 98970 }, { "epoch": 63.08476736775016, "grad_norm": 0.6140388250350952, "learning_rate": 8.006330659145595e-05, "loss": 0.0156, "step": 98980 }, { "epoch": 63.09114085404716, "grad_norm": 0.5899274349212646, "learning_rate": 8.005890239838458e-05, "loss": 0.0213, "step": 98990 }, { "epoch": 63.09751434034417, "grad_norm": 0.4996708035469055, "learning_rate": 8.005449784006917e-05, "loss": 0.0225, "step": 99000 }, { "epoch": 63.10388782664117, "grad_norm": 0.23874390125274658, "learning_rate": 8.005009291656322e-05, "loss": 0.0137, "step": 99010 }, { "epoch": 63.11026131293818, "grad_norm": 0.33274585008621216, "learning_rate": 8.004568762792026e-05, "loss": 0.0258, "step": 99020 }, { "epoch": 63.11663479923518, "grad_norm": 0.4330150783061981, "learning_rate": 8.004128197419385e-05, "loss": 0.0185, "step": 99030 }, { "epoch": 63.12300828553219, "grad_norm": 0.41974076628685, "learning_rate": 8.003687595543749e-05, "loss": 0.0105, "step": 99040 }, { "epoch": 63.12938177182919, "grad_norm": 0.40363675355911255, "learning_rate": 8.003246957170473e-05, "loss": 0.0189, "step": 99050 }, { "epoch": 63.135755258126196, "grad_norm": 0.3172776997089386, "learning_rate": 8.00280628230491e-05, "loss": 0.0163, "step": 99060 }, { "epoch": 63.1421287444232, "grad_norm": 0.4893060624599457, "learning_rate": 8.002365570952417e-05, "loss": 0.0122, "step": 99070 }, { "epoch": 63.148502230720204, "grad_norm": 0.3942698836326599, "learning_rate": 8.001924823118346e-05, "loss": 0.0153, "step": 99080 }, { "epoch": 63.154875717017205, "grad_norm": 0.3238273561000824, "learning_rate": 8.001484038808056e-05, "loss": 0.0171, "step": 99090 }, { "epoch": 63.16124920331421, "grad_norm": 0.36308565735816956, "learning_rate": 8.0010432180269e-05, "loss": 0.0141, "step": 99100 }, { "epoch": 63.167622689611214, "grad_norm": 0.5347263813018799, "learning_rate": 8.000602360780234e-05, "loss": 0.0193, "step": 99110 }, { "epoch": 63.17399617590822, "grad_norm": 0.5116556882858276, "learning_rate": 8.000161467073419e-05, "loss": 0.019, "step": 99120 }, { "epoch": 63.18036966220522, "grad_norm": 0.6092588305473328, "learning_rate": 7.999720536911807e-05, "loss": 0.0237, "step": 99130 }, { "epoch": 63.18674314850223, "grad_norm": 0.4888933300971985, "learning_rate": 7.999279570300761e-05, "loss": 0.0199, "step": 99140 }, { "epoch": 63.19311663479923, "grad_norm": 0.4203897714614868, "learning_rate": 7.998838567245634e-05, "loss": 0.0189, "step": 99150 }, { "epoch": 63.19949012109624, "grad_norm": 0.621531069278717, "learning_rate": 7.998397527751788e-05, "loss": 0.0195, "step": 99160 }, { "epoch": 63.20586360739325, "grad_norm": 0.4950043261051178, "learning_rate": 7.997956451824582e-05, "loss": 0.0244, "step": 99170 }, { "epoch": 63.21223709369025, "grad_norm": 0.4597020745277405, "learning_rate": 7.997515339469372e-05, "loss": 0.0235, "step": 99180 }, { "epoch": 63.218610579987256, "grad_norm": 0.24398472905158997, "learning_rate": 7.997074190691522e-05, "loss": 0.0187, "step": 99190 }, { "epoch": 63.22498406628426, "grad_norm": 0.3873862326145172, "learning_rate": 7.996633005496392e-05, "loss": 0.0253, "step": 99200 }, { "epoch": 63.231357552581265, "grad_norm": 0.6634920835494995, "learning_rate": 7.99619178388934e-05, "loss": 0.0181, "step": 99210 }, { "epoch": 63.237731038878266, "grad_norm": 0.3712419271469116, "learning_rate": 7.995750525875731e-05, "loss": 0.0163, "step": 99220 }, { "epoch": 63.244104525175274, "grad_norm": 0.5725627541542053, "learning_rate": 7.995309231460923e-05, "loss": 0.0172, "step": 99230 }, { "epoch": 63.250478011472275, "grad_norm": 0.45920196175575256, "learning_rate": 7.99486790065028e-05, "loss": 0.0146, "step": 99240 }, { "epoch": 63.25685149776928, "grad_norm": 0.24594923853874207, "learning_rate": 7.994426533449165e-05, "loss": 0.0244, "step": 99250 }, { "epoch": 63.263224984066284, "grad_norm": 0.2665552794933319, "learning_rate": 7.99398512986294e-05, "loss": 0.0129, "step": 99260 }, { "epoch": 63.26959847036329, "grad_norm": 0.5905922651290894, "learning_rate": 7.993543689896968e-05, "loss": 0.017, "step": 99270 }, { "epoch": 63.27597195666029, "grad_norm": 0.36916956305503845, "learning_rate": 7.993102213556616e-05, "loss": 0.0166, "step": 99280 }, { "epoch": 63.2823454429573, "grad_norm": 0.3216252326965332, "learning_rate": 7.992660700847244e-05, "loss": 0.0172, "step": 99290 }, { "epoch": 63.2887189292543, "grad_norm": 0.4990364909172058, "learning_rate": 7.99221915177422e-05, "loss": 0.0181, "step": 99300 }, { "epoch": 63.29509241555131, "grad_norm": 0.3280979096889496, "learning_rate": 7.991777566342909e-05, "loss": 0.0189, "step": 99310 }, { "epoch": 63.30146590184831, "grad_norm": 0.43353983759880066, "learning_rate": 7.991335944558673e-05, "loss": 0.0117, "step": 99320 }, { "epoch": 63.30783938814532, "grad_norm": 0.26578187942504883, "learning_rate": 7.990894286426883e-05, "loss": 0.0234, "step": 99330 }, { "epoch": 63.31421287444232, "grad_norm": 0.25533702969551086, "learning_rate": 7.990452591952902e-05, "loss": 0.0236, "step": 99340 }, { "epoch": 63.32058636073933, "grad_norm": 0.36629024147987366, "learning_rate": 7.990010861142101e-05, "loss": 0.0175, "step": 99350 }, { "epoch": 63.32695984703633, "grad_norm": 0.28898096084594727, "learning_rate": 7.989569093999843e-05, "loss": 0.0261, "step": 99360 }, { "epoch": 63.333333333333336, "grad_norm": 0.28239476680755615, "learning_rate": 7.989127290531499e-05, "loss": 0.0125, "step": 99370 }, { "epoch": 63.33970681963034, "grad_norm": 0.43497779965400696, "learning_rate": 7.988685450742437e-05, "loss": 0.0204, "step": 99380 }, { "epoch": 63.346080305927345, "grad_norm": 0.29697126150131226, "learning_rate": 7.988243574638024e-05, "loss": 0.0133, "step": 99390 }, { "epoch": 63.352453792224345, "grad_norm": 0.5886693000793457, "learning_rate": 7.987801662223631e-05, "loss": 0.0305, "step": 99400 }, { "epoch": 63.35882727852135, "grad_norm": 0.4977903664112091, "learning_rate": 7.987359713504626e-05, "loss": 0.0218, "step": 99410 }, { "epoch": 63.365200764818354, "grad_norm": 0.751430869102478, "learning_rate": 7.986917728486378e-05, "loss": 0.0186, "step": 99420 }, { "epoch": 63.37157425111536, "grad_norm": 0.8391770720481873, "learning_rate": 7.986475707174261e-05, "loss": 0.0256, "step": 99430 }, { "epoch": 63.37794773741236, "grad_norm": 0.6434748768806458, "learning_rate": 7.986033649573645e-05, "loss": 0.0219, "step": 99440 }, { "epoch": 63.38432122370937, "grad_norm": 0.4687001407146454, "learning_rate": 7.985591555689901e-05, "loss": 0.0301, "step": 99450 }, { "epoch": 63.39069471000637, "grad_norm": 0.539650559425354, "learning_rate": 7.9851494255284e-05, "loss": 0.0215, "step": 99460 }, { "epoch": 63.39706819630338, "grad_norm": 0.37941938638687134, "learning_rate": 7.984707259094517e-05, "loss": 0.0122, "step": 99470 }, { "epoch": 63.40344168260038, "grad_norm": 0.39016401767730713, "learning_rate": 7.98426505639362e-05, "loss": 0.0163, "step": 99480 }, { "epoch": 63.40981516889739, "grad_norm": 0.5059176683425903, "learning_rate": 7.983822817431088e-05, "loss": 0.0141, "step": 99490 }, { "epoch": 63.41618865519439, "grad_norm": 0.7039914131164551, "learning_rate": 7.983380542212288e-05, "loss": 0.0174, "step": 99500 }, { "epoch": 63.4225621414914, "grad_norm": 0.30721616744995117, "learning_rate": 7.9829382307426e-05, "loss": 0.013, "step": 99510 }, { "epoch": 63.4289356277884, "grad_norm": 0.7046541571617126, "learning_rate": 7.982495883027395e-05, "loss": 0.0256, "step": 99520 }, { "epoch": 63.435309114085406, "grad_norm": 0.2535940408706665, "learning_rate": 7.982053499072051e-05, "loss": 0.0213, "step": 99530 }, { "epoch": 63.44168260038241, "grad_norm": 0.27109014987945557, "learning_rate": 7.98161107888194e-05, "loss": 0.0245, "step": 99540 }, { "epoch": 63.448056086679415, "grad_norm": 0.4865108132362366, "learning_rate": 7.981168622462439e-05, "loss": 0.0164, "step": 99550 }, { "epoch": 63.454429572976416, "grad_norm": 0.35810279846191406, "learning_rate": 7.980726129818925e-05, "loss": 0.0121, "step": 99560 }, { "epoch": 63.460803059273424, "grad_norm": 0.5062312483787537, "learning_rate": 7.980283600956774e-05, "loss": 0.0192, "step": 99570 }, { "epoch": 63.467176545570425, "grad_norm": 0.6737575531005859, "learning_rate": 7.979841035881363e-05, "loss": 0.0176, "step": 99580 }, { "epoch": 63.47355003186743, "grad_norm": 0.5870066285133362, "learning_rate": 7.979398434598072e-05, "loss": 0.0218, "step": 99590 }, { "epoch": 63.47992351816443, "grad_norm": 0.37848570942878723, "learning_rate": 7.978955797112275e-05, "loss": 0.0142, "step": 99600 }, { "epoch": 63.48629700446144, "grad_norm": 0.36420461535453796, "learning_rate": 7.978513123429353e-05, "loss": 0.0171, "step": 99610 }, { "epoch": 63.49267049075844, "grad_norm": 0.46157515048980713, "learning_rate": 7.978070413554684e-05, "loss": 0.0135, "step": 99620 }, { "epoch": 63.49904397705545, "grad_norm": 0.536571741104126, "learning_rate": 7.977627667493648e-05, "loss": 0.019, "step": 99630 }, { "epoch": 63.50541746335245, "grad_norm": 0.25162723660469055, "learning_rate": 7.977184885251625e-05, "loss": 0.0154, "step": 99640 }, { "epoch": 63.51179094964946, "grad_norm": 0.2925228476524353, "learning_rate": 7.976742066833995e-05, "loss": 0.0156, "step": 99650 }, { "epoch": 63.51816443594646, "grad_norm": 0.35020291805267334, "learning_rate": 7.976299212246136e-05, "loss": 0.0202, "step": 99660 }, { "epoch": 63.52453792224347, "grad_norm": 0.683990478515625, "learning_rate": 7.975856321493434e-05, "loss": 0.015, "step": 99670 }, { "epoch": 63.53091140854047, "grad_norm": 0.3023894429206848, "learning_rate": 7.975413394581268e-05, "loss": 0.0173, "step": 99680 }, { "epoch": 63.53728489483748, "grad_norm": 0.5436844825744629, "learning_rate": 7.974970431515019e-05, "loss": 0.0155, "step": 99690 }, { "epoch": 63.54365838113448, "grad_norm": 0.260058730840683, "learning_rate": 7.97452743230007e-05, "loss": 0.0136, "step": 99700 }, { "epoch": 63.550031867431485, "grad_norm": 0.6079813838005066, "learning_rate": 7.974084396941805e-05, "loss": 0.0234, "step": 99710 }, { "epoch": 63.556405353728486, "grad_norm": 0.41311898827552795, "learning_rate": 7.973641325445607e-05, "loss": 0.0182, "step": 99720 }, { "epoch": 63.562778840025494, "grad_norm": 0.5203515291213989, "learning_rate": 7.973198217816858e-05, "loss": 0.0272, "step": 99730 }, { "epoch": 63.569152326322495, "grad_norm": 0.33635422587394714, "learning_rate": 7.972755074060945e-05, "loss": 0.0154, "step": 99740 }, { "epoch": 63.5755258126195, "grad_norm": 0.41816583275794983, "learning_rate": 7.97231189418325e-05, "loss": 0.032, "step": 99750 }, { "epoch": 63.581899298916504, "grad_norm": 0.40470996499061584, "learning_rate": 7.971868678189161e-05, "loss": 0.0172, "step": 99760 }, { "epoch": 63.58827278521351, "grad_norm": 0.25811755657196045, "learning_rate": 7.971425426084059e-05, "loss": 0.0237, "step": 99770 }, { "epoch": 63.59464627151051, "grad_norm": 0.3381985127925873, "learning_rate": 7.970982137873333e-05, "loss": 0.0323, "step": 99780 }, { "epoch": 63.60101975780752, "grad_norm": 0.42816609144210815, "learning_rate": 7.970538813562371e-05, "loss": 0.0125, "step": 99790 }, { "epoch": 63.60739324410453, "grad_norm": 0.3816288411617279, "learning_rate": 7.970095453156556e-05, "loss": 0.0145, "step": 99800 }, { "epoch": 63.61376673040153, "grad_norm": 0.6040284633636475, "learning_rate": 7.969652056661277e-05, "loss": 0.0282, "step": 99810 }, { "epoch": 63.62014021669854, "grad_norm": 0.20247100293636322, "learning_rate": 7.969208624081924e-05, "loss": 0.0132, "step": 99820 }, { "epoch": 63.62651370299554, "grad_norm": 0.4193212687969208, "learning_rate": 7.96876515542388e-05, "loss": 0.0199, "step": 99830 }, { "epoch": 63.632887189292546, "grad_norm": 0.5176677703857422, "learning_rate": 7.968321650692536e-05, "loss": 0.0242, "step": 99840 }, { "epoch": 63.63926067558955, "grad_norm": 0.18144778907299042, "learning_rate": 7.967878109893283e-05, "loss": 0.0171, "step": 99850 }, { "epoch": 63.645634161886555, "grad_norm": 0.5400227308273315, "learning_rate": 7.96743453303151e-05, "loss": 0.0151, "step": 99860 }, { "epoch": 63.652007648183556, "grad_norm": 0.7711611390113831, "learning_rate": 7.966990920112603e-05, "loss": 0.0211, "step": 99870 }, { "epoch": 63.658381134480564, "grad_norm": 0.36018404364585876, "learning_rate": 7.966547271141957e-05, "loss": 0.0309, "step": 99880 }, { "epoch": 63.664754620777565, "grad_norm": 0.475949227809906, "learning_rate": 7.966103586124959e-05, "loss": 0.0221, "step": 99890 }, { "epoch": 63.67112810707457, "grad_norm": 0.5908346176147461, "learning_rate": 7.965659865067004e-05, "loss": 0.0445, "step": 99900 }, { "epoch": 63.677501593371574, "grad_norm": 0.43027475476264954, "learning_rate": 7.965216107973481e-05, "loss": 0.0177, "step": 99910 }, { "epoch": 63.68387507966858, "grad_norm": 0.442116379737854, "learning_rate": 7.964772314849782e-05, "loss": 0.0193, "step": 99920 }, { "epoch": 63.69024856596558, "grad_norm": 0.4515080749988556, "learning_rate": 7.9643284857013e-05, "loss": 0.0275, "step": 99930 }, { "epoch": 63.69662205226259, "grad_norm": 0.39885076880455017, "learning_rate": 7.963884620533428e-05, "loss": 0.0178, "step": 99940 }, { "epoch": 63.70299553855959, "grad_norm": 0.4135817885398865, "learning_rate": 7.96344071935156e-05, "loss": 0.019, "step": 99950 }, { "epoch": 63.7093690248566, "grad_norm": 0.19746170938014984, "learning_rate": 7.962996782161089e-05, "loss": 0.0175, "step": 99960 }, { "epoch": 63.7157425111536, "grad_norm": 0.43266063928604126, "learning_rate": 7.962552808967412e-05, "loss": 0.0239, "step": 99970 }, { "epoch": 63.72211599745061, "grad_norm": 0.2739451229572296, "learning_rate": 7.962108799775917e-05, "loss": 0.0205, "step": 99980 }, { "epoch": 63.72848948374761, "grad_norm": 0.5102591514587402, "learning_rate": 7.961664754592006e-05, "loss": 0.0211, "step": 99990 }, { "epoch": 63.73486297004462, "grad_norm": 0.29910773038864136, "learning_rate": 7.961220673421072e-05, "loss": 0.0199, "step": 100000 }, { "epoch": 63.74123645634162, "grad_norm": 0.504512369632721, "learning_rate": 7.960776556268512e-05, "loss": 0.0189, "step": 100010 }, { "epoch": 63.747609942638626, "grad_norm": 0.5625543594360352, "learning_rate": 7.960332403139719e-05, "loss": 0.0142, "step": 100020 }, { "epoch": 63.753983428935626, "grad_norm": 0.7551673650741577, "learning_rate": 7.959888214040093e-05, "loss": 0.0304, "step": 100030 }, { "epoch": 63.760356915232634, "grad_norm": 0.6405827403068542, "learning_rate": 7.95944398897503e-05, "loss": 0.0251, "step": 100040 }, { "epoch": 63.766730401529635, "grad_norm": 0.23830997943878174, "learning_rate": 7.95899972794993e-05, "loss": 0.0221, "step": 100050 }, { "epoch": 63.77310388782664, "grad_norm": 0.25130775570869446, "learning_rate": 7.958555430970189e-05, "loss": 0.0162, "step": 100060 }, { "epoch": 63.779477374123644, "grad_norm": 0.6759980320930481, "learning_rate": 7.958111098041206e-05, "loss": 0.0246, "step": 100070 }, { "epoch": 63.78585086042065, "grad_norm": 0.41152045130729675, "learning_rate": 7.957666729168381e-05, "loss": 0.0166, "step": 100080 }, { "epoch": 63.79222434671765, "grad_norm": 0.48469844460487366, "learning_rate": 7.957222324357111e-05, "loss": 0.0169, "step": 100090 }, { "epoch": 63.79859783301466, "grad_norm": 0.4016260504722595, "learning_rate": 7.956777883612799e-05, "loss": 0.0263, "step": 100100 }, { "epoch": 63.80497131931166, "grad_norm": 0.38035067915916443, "learning_rate": 7.956333406940843e-05, "loss": 0.012, "step": 100110 }, { "epoch": 63.81134480560867, "grad_norm": 0.5274645090103149, "learning_rate": 7.955888894346646e-05, "loss": 0.0156, "step": 100120 }, { "epoch": 63.81771829190567, "grad_norm": 0.3349745571613312, "learning_rate": 7.955444345835607e-05, "loss": 0.0127, "step": 100130 }, { "epoch": 63.82409177820268, "grad_norm": 0.6604814529418945, "learning_rate": 7.954999761413128e-05, "loss": 0.0181, "step": 100140 }, { "epoch": 63.83046526449968, "grad_norm": 0.7396057844161987, "learning_rate": 7.954555141084614e-05, "loss": 0.0221, "step": 100150 }, { "epoch": 63.83683875079669, "grad_norm": 0.4282257556915283, "learning_rate": 7.954110484855465e-05, "loss": 0.0195, "step": 100160 }, { "epoch": 63.84321223709369, "grad_norm": 0.35531067848205566, "learning_rate": 7.953665792731084e-05, "loss": 0.0155, "step": 100170 }, { "epoch": 63.849585723390696, "grad_norm": 0.23789873719215393, "learning_rate": 7.953221064716876e-05, "loss": 0.0163, "step": 100180 }, { "epoch": 63.8559592096877, "grad_norm": 0.48693209886550903, "learning_rate": 7.952776300818243e-05, "loss": 0.0138, "step": 100190 }, { "epoch": 63.862332695984705, "grad_norm": 0.7536044716835022, "learning_rate": 7.95233150104059e-05, "loss": 0.0282, "step": 100200 }, { "epoch": 63.868706182281706, "grad_norm": 0.4548039436340332, "learning_rate": 7.951886665389322e-05, "loss": 0.0139, "step": 100210 }, { "epoch": 63.875079668578714, "grad_norm": 0.3638702929019928, "learning_rate": 7.951441793869843e-05, "loss": 0.0278, "step": 100220 }, { "epoch": 63.881453154875715, "grad_norm": 0.5413004159927368, "learning_rate": 7.95099688648756e-05, "loss": 0.0191, "step": 100230 }, { "epoch": 63.88782664117272, "grad_norm": 0.345094233751297, "learning_rate": 7.950551943247879e-05, "loss": 0.0136, "step": 100240 }, { "epoch": 63.89420012746972, "grad_norm": 0.42283883690834045, "learning_rate": 7.950106964156206e-05, "loss": 0.0205, "step": 100250 }, { "epoch": 63.90057361376673, "grad_norm": 0.3873358964920044, "learning_rate": 7.949661949217948e-05, "loss": 0.0157, "step": 100260 }, { "epoch": 63.90694710006373, "grad_norm": 0.8664463758468628, "learning_rate": 7.949216898438512e-05, "loss": 0.0285, "step": 100270 }, { "epoch": 63.91332058636074, "grad_norm": 0.3149782717227936, "learning_rate": 7.948771811823307e-05, "loss": 0.018, "step": 100280 }, { "epoch": 63.91969407265774, "grad_norm": 0.33077171444892883, "learning_rate": 7.94832668937774e-05, "loss": 0.0177, "step": 100290 }, { "epoch": 63.92606755895475, "grad_norm": 0.6482410430908203, "learning_rate": 7.94788153110722e-05, "loss": 0.0197, "step": 100300 }, { "epoch": 63.93244104525175, "grad_norm": 0.5424817800521851, "learning_rate": 7.947436337017155e-05, "loss": 0.0169, "step": 100310 }, { "epoch": 63.93881453154876, "grad_norm": 0.38601818680763245, "learning_rate": 7.946991107112957e-05, "loss": 0.0182, "step": 100320 }, { "epoch": 63.94518801784576, "grad_norm": 0.2997458875179291, "learning_rate": 7.946545841400035e-05, "loss": 0.0162, "step": 100330 }, { "epoch": 63.95156150414277, "grad_norm": 0.3400866985321045, "learning_rate": 7.946100539883798e-05, "loss": 0.0113, "step": 100340 }, { "epoch": 63.95793499043977, "grad_norm": 0.18522697687149048, "learning_rate": 7.945655202569659e-05, "loss": 0.0207, "step": 100350 }, { "epoch": 63.964308476736775, "grad_norm": 0.394366592168808, "learning_rate": 7.945209829463027e-05, "loss": 0.0119, "step": 100360 }, { "epoch": 63.970681963033776, "grad_norm": 0.4166446328163147, "learning_rate": 7.944764420569314e-05, "loss": 0.0182, "step": 100370 }, { "epoch": 63.977055449330784, "grad_norm": 0.1872115433216095, "learning_rate": 7.944318975893936e-05, "loss": 0.0209, "step": 100380 }, { "epoch": 63.983428935627785, "grad_norm": 0.3367049992084503, "learning_rate": 7.9438734954423e-05, "loss": 0.0131, "step": 100390 }, { "epoch": 63.98980242192479, "grad_norm": 0.5766119360923767, "learning_rate": 7.943427979219822e-05, "loss": 0.0222, "step": 100400 }, { "epoch": 63.9961759082218, "grad_norm": 0.3367582857608795, "learning_rate": 7.942982427231916e-05, "loss": 0.0244, "step": 100410 }, { "epoch": 64.00254939451881, "grad_norm": 0.3722878098487854, "learning_rate": 7.942536839483995e-05, "loss": 0.0149, "step": 100420 }, { "epoch": 64.0089228808158, "grad_norm": 0.3097875714302063, "learning_rate": 7.942091215981473e-05, "loss": 0.0168, "step": 100430 }, { "epoch": 64.01529636711281, "grad_norm": 0.6235575675964355, "learning_rate": 7.941645556729763e-05, "loss": 0.033, "step": 100440 }, { "epoch": 64.02166985340982, "grad_norm": 0.49927809834480286, "learning_rate": 7.941199861734285e-05, "loss": 0.0226, "step": 100450 }, { "epoch": 64.02804333970683, "grad_norm": 0.4408954381942749, "learning_rate": 7.940754131000449e-05, "loss": 0.0275, "step": 100460 }, { "epoch": 64.03441682600382, "grad_norm": 0.5775007009506226, "learning_rate": 7.940308364533675e-05, "loss": 0.0435, "step": 100470 }, { "epoch": 64.04079031230083, "grad_norm": 0.3866294026374817, "learning_rate": 7.939862562339378e-05, "loss": 0.0167, "step": 100480 }, { "epoch": 64.04716379859784, "grad_norm": 1.362955093383789, "learning_rate": 7.939416724422976e-05, "loss": 0.0205, "step": 100490 }, { "epoch": 64.05353728489484, "grad_norm": 0.4709736704826355, "learning_rate": 7.938970850789886e-05, "loss": 0.0217, "step": 100500 }, { "epoch": 64.05991077119184, "grad_norm": 0.27373307943344116, "learning_rate": 7.938524941445522e-05, "loss": 0.0219, "step": 100510 }, { "epoch": 64.06628425748885, "grad_norm": 0.7043091058731079, "learning_rate": 7.938078996395309e-05, "loss": 0.0257, "step": 100520 }, { "epoch": 64.07265774378585, "grad_norm": 0.6914076805114746, "learning_rate": 7.937633015644661e-05, "loss": 0.0213, "step": 100530 }, { "epoch": 64.07903123008286, "grad_norm": 0.7368538975715637, "learning_rate": 7.937186999198997e-05, "loss": 0.0145, "step": 100540 }, { "epoch": 64.08540471637986, "grad_norm": 1.0276824235916138, "learning_rate": 7.936740947063738e-05, "loss": 0.0231, "step": 100550 }, { "epoch": 64.09177820267686, "grad_norm": 0.5121086835861206, "learning_rate": 7.936294859244307e-05, "loss": 0.0189, "step": 100560 }, { "epoch": 64.09815168897387, "grad_norm": 0.6267324090003967, "learning_rate": 7.935848735746119e-05, "loss": 0.0241, "step": 100570 }, { "epoch": 64.10452517527088, "grad_norm": 0.4705946743488312, "learning_rate": 7.935402576574597e-05, "loss": 0.0189, "step": 100580 }, { "epoch": 64.11089866156787, "grad_norm": 0.4142254889011383, "learning_rate": 7.934956381735161e-05, "loss": 0.0307, "step": 100590 }, { "epoch": 64.11727214786488, "grad_norm": 0.5279407501220703, "learning_rate": 7.934510151233236e-05, "loss": 0.0301, "step": 100600 }, { "epoch": 64.12364563416189, "grad_norm": 0.8567071557044983, "learning_rate": 7.93406388507424e-05, "loss": 0.0205, "step": 100610 }, { "epoch": 64.1300191204589, "grad_norm": 0.5787873864173889, "learning_rate": 7.9336175832636e-05, "loss": 0.0329, "step": 100620 }, { "epoch": 64.13639260675589, "grad_norm": 0.9041554927825928, "learning_rate": 7.933171245806734e-05, "loss": 0.0231, "step": 100630 }, { "epoch": 64.1427660930529, "grad_norm": 0.4500020444393158, "learning_rate": 7.93272487270907e-05, "loss": 0.0143, "step": 100640 }, { "epoch": 64.1491395793499, "grad_norm": 0.5958545804023743, "learning_rate": 7.932278463976028e-05, "loss": 0.0299, "step": 100650 }, { "epoch": 64.15551306564691, "grad_norm": 0.48936063051223755, "learning_rate": 7.931832019613035e-05, "loss": 0.0169, "step": 100660 }, { "epoch": 64.16188655194391, "grad_norm": 0.4239228367805481, "learning_rate": 7.931385539625515e-05, "loss": 0.0167, "step": 100670 }, { "epoch": 64.16826003824092, "grad_norm": 0.3314990997314453, "learning_rate": 7.930939024018893e-05, "loss": 0.0156, "step": 100680 }, { "epoch": 64.17463352453792, "grad_norm": 0.37809112668037415, "learning_rate": 7.930492472798594e-05, "loss": 0.0251, "step": 100690 }, { "epoch": 64.18100701083493, "grad_norm": 0.3460645377635956, "learning_rate": 7.930045885970047e-05, "loss": 0.0175, "step": 100700 }, { "epoch": 64.18738049713193, "grad_norm": 0.3932236135005951, "learning_rate": 7.929599263538673e-05, "loss": 0.0145, "step": 100710 }, { "epoch": 64.19375398342893, "grad_norm": 0.30228695273399353, "learning_rate": 7.929152605509904e-05, "loss": 0.0166, "step": 100720 }, { "epoch": 64.20012746972594, "grad_norm": 0.538913905620575, "learning_rate": 7.928705911889164e-05, "loss": 0.024, "step": 100730 }, { "epoch": 64.20650095602295, "grad_norm": 0.4455079734325409, "learning_rate": 7.928259182681883e-05, "loss": 0.0201, "step": 100740 }, { "epoch": 64.21287444231994, "grad_norm": 0.38340333104133606, "learning_rate": 7.927812417893487e-05, "loss": 0.0096, "step": 100750 }, { "epoch": 64.21924792861695, "grad_norm": 0.37559008598327637, "learning_rate": 7.927365617529407e-05, "loss": 0.0203, "step": 100760 }, { "epoch": 64.22562141491396, "grad_norm": 0.1946892887353897, "learning_rate": 7.92691878159507e-05, "loss": 0.0134, "step": 100770 }, { "epoch": 64.23199490121097, "grad_norm": 0.5607951879501343, "learning_rate": 7.926471910095907e-05, "loss": 0.0196, "step": 100780 }, { "epoch": 64.23836838750796, "grad_norm": 0.8088657855987549, "learning_rate": 7.926025003037348e-05, "loss": 0.0192, "step": 100790 }, { "epoch": 64.24474187380497, "grad_norm": 0.2770121693611145, "learning_rate": 7.925578060424819e-05, "loss": 0.0158, "step": 100800 }, { "epoch": 64.25111536010198, "grad_norm": 0.2666962146759033, "learning_rate": 7.925131082263758e-05, "loss": 0.0184, "step": 100810 }, { "epoch": 64.25748884639899, "grad_norm": 0.2115778625011444, "learning_rate": 7.924684068559592e-05, "loss": 0.0238, "step": 100820 }, { "epoch": 64.26386233269598, "grad_norm": 0.19178056716918945, "learning_rate": 7.924237019317753e-05, "loss": 0.0189, "step": 100830 }, { "epoch": 64.27023581899299, "grad_norm": 0.2301441878080368, "learning_rate": 7.923789934543673e-05, "loss": 0.0144, "step": 100840 }, { "epoch": 64.27660930529, "grad_norm": 0.2912479043006897, "learning_rate": 7.923342814242785e-05, "loss": 0.0086, "step": 100850 }, { "epoch": 64.282982791587, "grad_norm": 0.1716517210006714, "learning_rate": 7.92289565842052e-05, "loss": 0.0168, "step": 100860 }, { "epoch": 64.289356277884, "grad_norm": 0.4291152358055115, "learning_rate": 7.922448467082315e-05, "loss": 0.0168, "step": 100870 }, { "epoch": 64.295729764181, "grad_norm": 0.455714613199234, "learning_rate": 7.922001240233601e-05, "loss": 0.018, "step": 100880 }, { "epoch": 64.30210325047801, "grad_norm": 0.46090081334114075, "learning_rate": 7.921553977879815e-05, "loss": 0.0178, "step": 100890 }, { "epoch": 64.30847673677502, "grad_norm": 0.3887077271938324, "learning_rate": 7.921106680026387e-05, "loss": 0.0182, "step": 100900 }, { "epoch": 64.31485022307201, "grad_norm": 0.5262818336486816, "learning_rate": 7.920659346678755e-05, "loss": 0.0223, "step": 100910 }, { "epoch": 64.32122370936902, "grad_norm": 0.21705761551856995, "learning_rate": 7.920211977842357e-05, "loss": 0.0183, "step": 100920 }, { "epoch": 64.32759719566603, "grad_norm": 0.37105482816696167, "learning_rate": 7.919764573522624e-05, "loss": 0.0198, "step": 100930 }, { "epoch": 64.33397068196304, "grad_norm": 0.4694282114505768, "learning_rate": 7.919317133724996e-05, "loss": 0.016, "step": 100940 }, { "epoch": 64.34034416826003, "grad_norm": 0.3237643539905548, "learning_rate": 7.918869658454908e-05, "loss": 0.0146, "step": 100950 }, { "epoch": 64.34671765455704, "grad_norm": 0.2847658693790436, "learning_rate": 7.918422147717798e-05, "loss": 0.0168, "step": 100960 }, { "epoch": 64.35309114085405, "grad_norm": 0.28392845392227173, "learning_rate": 7.917974601519102e-05, "loss": 0.0178, "step": 100970 }, { "epoch": 64.35946462715106, "grad_norm": 0.35983288288116455, "learning_rate": 7.917527019864262e-05, "loss": 0.0131, "step": 100980 }, { "epoch": 64.36583811344805, "grad_norm": 0.5620695352554321, "learning_rate": 7.917079402758713e-05, "loss": 0.0226, "step": 100990 }, { "epoch": 64.37221159974506, "grad_norm": 0.3042221665382385, "learning_rate": 7.916631750207894e-05, "loss": 0.0253, "step": 101000 }, { "epoch": 64.37858508604207, "grad_norm": 0.3776514232158661, "learning_rate": 7.916184062217247e-05, "loss": 0.0188, "step": 101010 }, { "epoch": 64.38495857233907, "grad_norm": 0.37415656447410583, "learning_rate": 7.91573633879221e-05, "loss": 0.0183, "step": 101020 }, { "epoch": 64.39133205863607, "grad_norm": 0.3529919981956482, "learning_rate": 7.915288579938224e-05, "loss": 0.0204, "step": 101030 }, { "epoch": 64.39770554493307, "grad_norm": 0.4255618453025818, "learning_rate": 7.914840785660729e-05, "loss": 0.021, "step": 101040 }, { "epoch": 64.40407903123008, "grad_norm": 0.2587633430957794, "learning_rate": 7.914392955965165e-05, "loss": 0.0178, "step": 101050 }, { "epoch": 64.41045251752709, "grad_norm": 0.46349823474884033, "learning_rate": 7.913945090856976e-05, "loss": 0.0191, "step": 101060 }, { "epoch": 64.4168260038241, "grad_norm": 0.3097410500049591, "learning_rate": 7.913497190341603e-05, "loss": 0.0239, "step": 101070 }, { "epoch": 64.42319949012109, "grad_norm": 0.29241394996643066, "learning_rate": 7.913049254424489e-05, "loss": 0.017, "step": 101080 }, { "epoch": 64.4295729764181, "grad_norm": 0.23196350038051605, "learning_rate": 7.912601283111077e-05, "loss": 0.0288, "step": 101090 }, { "epoch": 64.43594646271511, "grad_norm": 0.32502835988998413, "learning_rate": 7.912153276406807e-05, "loss": 0.0215, "step": 101100 }, { "epoch": 64.44231994901212, "grad_norm": 0.5920019745826721, "learning_rate": 7.911705234317126e-05, "loss": 0.0226, "step": 101110 }, { "epoch": 64.44869343530911, "grad_norm": 0.395711213350296, "learning_rate": 7.911257156847476e-05, "loss": 0.018, "step": 101120 }, { "epoch": 64.45506692160612, "grad_norm": 0.5862853527069092, "learning_rate": 7.910809044003304e-05, "loss": 0.0205, "step": 101130 }, { "epoch": 64.46144040790313, "grad_norm": 0.4178614914417267, "learning_rate": 7.910360895790055e-05, "loss": 0.0191, "step": 101140 }, { "epoch": 64.46781389420013, "grad_norm": 0.3227098286151886, "learning_rate": 7.909912712213172e-05, "loss": 0.0193, "step": 101150 }, { "epoch": 64.47418738049713, "grad_norm": 0.25548678636550903, "learning_rate": 7.909464493278103e-05, "loss": 0.0131, "step": 101160 }, { "epoch": 64.48056086679414, "grad_norm": 0.46040797233581543, "learning_rate": 7.909016238990292e-05, "loss": 0.0175, "step": 101170 }, { "epoch": 64.48693435309114, "grad_norm": 0.32197993993759155, "learning_rate": 7.90856794935519e-05, "loss": 0.0169, "step": 101180 }, { "epoch": 64.49330783938815, "grad_norm": 0.213952898979187, "learning_rate": 7.908119624378238e-05, "loss": 0.0172, "step": 101190 }, { "epoch": 64.49968132568515, "grad_norm": 0.4883028566837311, "learning_rate": 7.907671264064888e-05, "loss": 0.0224, "step": 101200 }, { "epoch": 64.50605481198215, "grad_norm": 0.7371419072151184, "learning_rate": 7.907222868420587e-05, "loss": 0.0208, "step": 101210 }, { "epoch": 64.51242829827916, "grad_norm": 0.18632501363754272, "learning_rate": 7.906774437450784e-05, "loss": 0.0103, "step": 101220 }, { "epoch": 64.51880178457617, "grad_norm": 0.5247831344604492, "learning_rate": 7.906325971160927e-05, "loss": 0.0281, "step": 101230 }, { "epoch": 64.52517527087316, "grad_norm": 0.5225400924682617, "learning_rate": 7.905877469556464e-05, "loss": 0.0236, "step": 101240 }, { "epoch": 64.53154875717017, "grad_norm": 0.2231888324022293, "learning_rate": 7.905428932642848e-05, "loss": 0.0123, "step": 101250 }, { "epoch": 64.53792224346718, "grad_norm": 0.2814962863922119, "learning_rate": 7.904980360425525e-05, "loss": 0.0167, "step": 101260 }, { "epoch": 64.54429572976419, "grad_norm": 0.3209356367588043, "learning_rate": 7.904531752909949e-05, "loss": 0.02, "step": 101270 }, { "epoch": 64.55066921606118, "grad_norm": 0.2787807881832123, "learning_rate": 7.90408311010157e-05, "loss": 0.0235, "step": 101280 }, { "epoch": 64.55704270235819, "grad_norm": 0.340372234582901, "learning_rate": 7.90363443200584e-05, "loss": 0.0218, "step": 101290 }, { "epoch": 64.5634161886552, "grad_norm": 0.4209139049053192, "learning_rate": 7.90318571862821e-05, "loss": 0.0182, "step": 101300 }, { "epoch": 64.5697896749522, "grad_norm": 0.32832300662994385, "learning_rate": 7.902736969974132e-05, "loss": 0.0211, "step": 101310 }, { "epoch": 64.5761631612492, "grad_norm": 0.49134567379951477, "learning_rate": 7.902288186049059e-05, "loss": 0.0311, "step": 101320 }, { "epoch": 64.5825366475462, "grad_norm": 0.2537376284599304, "learning_rate": 7.901839366858443e-05, "loss": 0.0281, "step": 101330 }, { "epoch": 64.58891013384321, "grad_norm": 0.30038487911224365, "learning_rate": 7.901390512407741e-05, "loss": 0.0153, "step": 101340 }, { "epoch": 64.59528362014022, "grad_norm": 0.2778017520904541, "learning_rate": 7.900941622702404e-05, "loss": 0.0164, "step": 101350 }, { "epoch": 64.60165710643722, "grad_norm": 0.3513677716255188, "learning_rate": 7.900492697747887e-05, "loss": 0.0171, "step": 101360 }, { "epoch": 64.60803059273422, "grad_norm": 0.4721972644329071, "learning_rate": 7.900043737549647e-05, "loss": 0.0239, "step": 101370 }, { "epoch": 64.61440407903123, "grad_norm": 0.2930135726928711, "learning_rate": 7.899594742113136e-05, "loss": 0.0127, "step": 101380 }, { "epoch": 64.62077756532824, "grad_norm": 0.3552321493625641, "learning_rate": 7.899145711443812e-05, "loss": 0.025, "step": 101390 }, { "epoch": 64.62715105162523, "grad_norm": 0.5242018699645996, "learning_rate": 7.89869664554713e-05, "loss": 0.0211, "step": 101400 }, { "epoch": 64.63352453792224, "grad_norm": 0.4109097719192505, "learning_rate": 7.898247544428547e-05, "loss": 0.0217, "step": 101410 }, { "epoch": 64.63989802421925, "grad_norm": 0.22159899771213531, "learning_rate": 7.89779840809352e-05, "loss": 0.0193, "step": 101420 }, { "epoch": 64.64627151051626, "grad_norm": 0.5730328559875488, "learning_rate": 7.897349236547506e-05, "loss": 0.0151, "step": 101430 }, { "epoch": 64.65264499681325, "grad_norm": 0.49782100319862366, "learning_rate": 7.896900029795961e-05, "loss": 0.0186, "step": 101440 }, { "epoch": 64.65901848311026, "grad_norm": 0.2766443192958832, "learning_rate": 7.89645078784435e-05, "loss": 0.0112, "step": 101450 }, { "epoch": 64.66539196940727, "grad_norm": 0.693105161190033, "learning_rate": 7.896001510698125e-05, "loss": 0.0254, "step": 101460 }, { "epoch": 64.67176545570427, "grad_norm": 0.5014377236366272, "learning_rate": 7.895552198362748e-05, "loss": 0.0131, "step": 101470 }, { "epoch": 64.67813894200127, "grad_norm": 0.3709430694580078, "learning_rate": 7.895102850843676e-05, "loss": 0.0215, "step": 101480 }, { "epoch": 64.68451242829828, "grad_norm": 0.39387303590774536, "learning_rate": 7.894653468146374e-05, "loss": 0.024, "step": 101490 }, { "epoch": 64.69088591459528, "grad_norm": 0.4278917908668518, "learning_rate": 7.894204050276297e-05, "loss": 0.0332, "step": 101500 }, { "epoch": 64.69725940089229, "grad_norm": 0.46081745624542236, "learning_rate": 7.893754597238908e-05, "loss": 0.0138, "step": 101510 }, { "epoch": 64.70363288718929, "grad_norm": 0.5743962526321411, "learning_rate": 7.893305109039671e-05, "loss": 0.0201, "step": 101520 }, { "epoch": 64.7100063734863, "grad_norm": 0.3479151427745819, "learning_rate": 7.892855585684043e-05, "loss": 0.0132, "step": 101530 }, { "epoch": 64.7163798597833, "grad_norm": 0.5232152342796326, "learning_rate": 7.892406027177488e-05, "loss": 0.024, "step": 101540 }, { "epoch": 64.72275334608031, "grad_norm": 0.3444971442222595, "learning_rate": 7.891956433525471e-05, "loss": 0.0358, "step": 101550 }, { "epoch": 64.7291268323773, "grad_norm": 0.41029149293899536, "learning_rate": 7.89150680473345e-05, "loss": 0.0149, "step": 101560 }, { "epoch": 64.73550031867431, "grad_norm": 0.23681877553462982, "learning_rate": 7.891057140806895e-05, "loss": 0.0191, "step": 101570 }, { "epoch": 64.74187380497132, "grad_norm": 0.41550537943840027, "learning_rate": 7.890607441751263e-05, "loss": 0.0186, "step": 101580 }, { "epoch": 64.74824729126833, "grad_norm": 0.35216400027275085, "learning_rate": 7.890157707572023e-05, "loss": 0.0112, "step": 101590 }, { "epoch": 64.75462077756532, "grad_norm": 0.49337026476860046, "learning_rate": 7.889707938274636e-05, "loss": 0.0135, "step": 101600 }, { "epoch": 64.76099426386233, "grad_norm": 0.3101850152015686, "learning_rate": 7.889258133864572e-05, "loss": 0.0167, "step": 101610 }, { "epoch": 64.76736775015934, "grad_norm": 0.3876095414161682, "learning_rate": 7.888808294347291e-05, "loss": 0.0175, "step": 101620 }, { "epoch": 64.77374123645635, "grad_norm": 0.563385546207428, "learning_rate": 7.888358419728263e-05, "loss": 0.0159, "step": 101630 }, { "epoch": 64.78011472275334, "grad_norm": 0.3106054365634918, "learning_rate": 7.887908510012954e-05, "loss": 0.0104, "step": 101640 }, { "epoch": 64.78648820905035, "grad_norm": 0.5170860886573792, "learning_rate": 7.88745856520683e-05, "loss": 0.0154, "step": 101650 }, { "epoch": 64.79286169534736, "grad_norm": 0.5176358222961426, "learning_rate": 7.887008585315357e-05, "loss": 0.0191, "step": 101660 }, { "epoch": 64.79923518164436, "grad_norm": 0.2890208959579468, "learning_rate": 7.886558570344005e-05, "loss": 0.0148, "step": 101670 }, { "epoch": 64.80560866794136, "grad_norm": 0.7593268752098083, "learning_rate": 7.886108520298241e-05, "loss": 0.0264, "step": 101680 }, { "epoch": 64.81198215423836, "grad_norm": 0.29270845651626587, "learning_rate": 7.885658435183532e-05, "loss": 0.0211, "step": 101690 }, { "epoch": 64.81835564053537, "grad_norm": 0.29428961873054504, "learning_rate": 7.88520831500535e-05, "loss": 0.0235, "step": 101700 }, { "epoch": 64.82472912683238, "grad_norm": 0.21894976496696472, "learning_rate": 7.884758159769164e-05, "loss": 0.0135, "step": 101710 }, { "epoch": 64.83110261312939, "grad_norm": 0.24917179346084595, "learning_rate": 7.884307969480439e-05, "loss": 0.0254, "step": 101720 }, { "epoch": 64.83747609942638, "grad_norm": 0.4067525863647461, "learning_rate": 7.883857744144651e-05, "loss": 0.0149, "step": 101730 }, { "epoch": 64.84384958572339, "grad_norm": 0.36483076214790344, "learning_rate": 7.883407483767269e-05, "loss": 0.0187, "step": 101740 }, { "epoch": 64.8502230720204, "grad_norm": 0.45491382479667664, "learning_rate": 7.882957188353764e-05, "loss": 0.0183, "step": 101750 }, { "epoch": 64.8565965583174, "grad_norm": 0.6092327833175659, "learning_rate": 7.882506857909606e-05, "loss": 0.0176, "step": 101760 }, { "epoch": 64.8629700446144, "grad_norm": 0.49361664056777954, "learning_rate": 7.882056492440269e-05, "loss": 0.0476, "step": 101770 }, { "epoch": 64.86934353091141, "grad_norm": 0.33111509680747986, "learning_rate": 7.881606091951223e-05, "loss": 0.0133, "step": 101780 }, { "epoch": 64.87571701720842, "grad_norm": 0.3197232484817505, "learning_rate": 7.881155656447944e-05, "loss": 0.0164, "step": 101790 }, { "epoch": 64.88209050350542, "grad_norm": 0.41550832986831665, "learning_rate": 7.880705185935904e-05, "loss": 0.0274, "step": 101800 }, { "epoch": 64.88846398980242, "grad_norm": 0.3176764249801636, "learning_rate": 7.880254680420574e-05, "loss": 0.0189, "step": 101810 }, { "epoch": 64.89483747609943, "grad_norm": 0.31465575098991394, "learning_rate": 7.879804139907431e-05, "loss": 0.021, "step": 101820 }, { "epoch": 64.90121096239643, "grad_norm": 0.39449888467788696, "learning_rate": 7.87935356440195e-05, "loss": 0.0213, "step": 101830 }, { "epoch": 64.90758444869344, "grad_norm": 0.4956305921077728, "learning_rate": 7.878902953909604e-05, "loss": 0.0375, "step": 101840 }, { "epoch": 64.91395793499044, "grad_norm": 0.3838893473148346, "learning_rate": 7.878452308435867e-05, "loss": 0.0263, "step": 101850 }, { "epoch": 64.92033142128744, "grad_norm": 0.39091402292251587, "learning_rate": 7.87800162798622e-05, "loss": 0.0175, "step": 101860 }, { "epoch": 64.92670490758445, "grad_norm": 0.3478544056415558, "learning_rate": 7.877550912566134e-05, "loss": 0.0086, "step": 101870 }, { "epoch": 64.93307839388146, "grad_norm": 0.4270361065864563, "learning_rate": 7.877100162181088e-05, "loss": 0.0164, "step": 101880 }, { "epoch": 64.93945188017845, "grad_norm": 0.3549550175666809, "learning_rate": 7.876649376836558e-05, "loss": 0.0211, "step": 101890 }, { "epoch": 64.94582536647546, "grad_norm": 0.3081165850162506, "learning_rate": 7.876198556538023e-05, "loss": 0.0119, "step": 101900 }, { "epoch": 64.95219885277247, "grad_norm": 0.8338729739189148, "learning_rate": 7.875747701290959e-05, "loss": 0.0133, "step": 101910 }, { "epoch": 64.95857233906948, "grad_norm": 0.3619285821914673, "learning_rate": 7.875296811100845e-05, "loss": 0.0106, "step": 101920 }, { "epoch": 64.96494582536647, "grad_norm": 0.3653522729873657, "learning_rate": 7.874845885973162e-05, "loss": 0.0226, "step": 101930 }, { "epoch": 64.97131931166348, "grad_norm": 0.3748699426651001, "learning_rate": 7.874394925913385e-05, "loss": 0.0183, "step": 101940 }, { "epoch": 64.97769279796049, "grad_norm": 0.6844147443771362, "learning_rate": 7.873943930926997e-05, "loss": 0.0212, "step": 101950 }, { "epoch": 64.9840662842575, "grad_norm": 0.2633829712867737, "learning_rate": 7.873492901019476e-05, "loss": 0.0184, "step": 101960 }, { "epoch": 64.99043977055449, "grad_norm": 0.42903703451156616, "learning_rate": 7.873041836196305e-05, "loss": 0.0195, "step": 101970 }, { "epoch": 64.9968132568515, "grad_norm": 0.4704042375087738, "learning_rate": 7.872590736462962e-05, "loss": 0.0136, "step": 101980 }, { "epoch": 65.0031867431485, "grad_norm": 0.20374086499214172, "learning_rate": 7.872139601824929e-05, "loss": 0.0311, "step": 101990 }, { "epoch": 65.00956022944551, "grad_norm": 0.5119425058364868, "learning_rate": 7.87168843228769e-05, "loss": 0.0437, "step": 102000 }, { "epoch": 65.0159337157425, "grad_norm": 0.31094542145729065, "learning_rate": 7.871237227856721e-05, "loss": 0.0227, "step": 102010 }, { "epoch": 65.02230720203951, "grad_norm": 0.26588964462280273, "learning_rate": 7.870785988537512e-05, "loss": 0.0239, "step": 102020 }, { "epoch": 65.02868068833652, "grad_norm": 0.3512263894081116, "learning_rate": 7.870334714335544e-05, "loss": 0.0283, "step": 102030 }, { "epoch": 65.03505417463353, "grad_norm": 0.3284313678741455, "learning_rate": 7.869883405256295e-05, "loss": 0.0172, "step": 102040 }, { "epoch": 65.04142766093052, "grad_norm": 0.42981940507888794, "learning_rate": 7.869432061305255e-05, "loss": 0.026, "step": 102050 }, { "epoch": 65.04780114722753, "grad_norm": 0.2298647165298462, "learning_rate": 7.868980682487907e-05, "loss": 0.0201, "step": 102060 }, { "epoch": 65.05417463352454, "grad_norm": 0.6069732308387756, "learning_rate": 7.868529268809735e-05, "loss": 0.0252, "step": 102070 }, { "epoch": 65.06054811982155, "grad_norm": 0.4655470848083496, "learning_rate": 7.868077820276223e-05, "loss": 0.0286, "step": 102080 }, { "epoch": 65.06692160611854, "grad_norm": 0.34539562463760376, "learning_rate": 7.867626336892858e-05, "loss": 0.0219, "step": 102090 }, { "epoch": 65.07329509241555, "grad_norm": 0.6529168486595154, "learning_rate": 7.867174818665126e-05, "loss": 0.0153, "step": 102100 }, { "epoch": 65.07966857871256, "grad_norm": 0.509239137172699, "learning_rate": 7.866723265598512e-05, "loss": 0.015, "step": 102110 }, { "epoch": 65.08604206500956, "grad_norm": 0.36583852767944336, "learning_rate": 7.866271677698504e-05, "loss": 0.0258, "step": 102120 }, { "epoch": 65.09241555130656, "grad_norm": 0.48289310932159424, "learning_rate": 7.86582005497059e-05, "loss": 0.0182, "step": 102130 }, { "epoch": 65.09878903760357, "grad_norm": 1.14234459400177, "learning_rate": 7.865368397420256e-05, "loss": 0.0324, "step": 102140 }, { "epoch": 65.10516252390057, "grad_norm": 0.47853198647499084, "learning_rate": 7.864916705052989e-05, "loss": 0.0196, "step": 102150 }, { "epoch": 65.11153601019758, "grad_norm": 0.3444884419441223, "learning_rate": 7.864464977874282e-05, "loss": 0.0194, "step": 102160 }, { "epoch": 65.11790949649458, "grad_norm": 0.3368479907512665, "learning_rate": 7.864013215889618e-05, "loss": 0.019, "step": 102170 }, { "epoch": 65.12428298279158, "grad_norm": 0.4835352897644043, "learning_rate": 7.863561419104491e-05, "loss": 0.0161, "step": 102180 }, { "epoch": 65.13065646908859, "grad_norm": 0.38151365518569946, "learning_rate": 7.863109587524389e-05, "loss": 0.0221, "step": 102190 }, { "epoch": 65.1370299553856, "grad_norm": 0.28853705525398254, "learning_rate": 7.862657721154802e-05, "loss": 0.0138, "step": 102200 }, { "epoch": 65.1434034416826, "grad_norm": 0.531181812286377, "learning_rate": 7.862205820001221e-05, "loss": 0.0143, "step": 102210 }, { "epoch": 65.1497769279796, "grad_norm": 0.5171051025390625, "learning_rate": 7.861753884069137e-05, "loss": 0.0161, "step": 102220 }, { "epoch": 65.15615041427661, "grad_norm": 0.47504138946533203, "learning_rate": 7.861301913364042e-05, "loss": 0.0233, "step": 102230 }, { "epoch": 65.16252390057362, "grad_norm": 0.4146222770214081, "learning_rate": 7.860849907891425e-05, "loss": 0.019, "step": 102240 }, { "epoch": 65.16889738687061, "grad_norm": 0.29952529072761536, "learning_rate": 7.860397867656784e-05, "loss": 0.0158, "step": 102250 }, { "epoch": 65.17527087316762, "grad_norm": 0.5573179125785828, "learning_rate": 7.859945792665605e-05, "loss": 0.0251, "step": 102260 }, { "epoch": 65.18164435946463, "grad_norm": 0.6797312498092651, "learning_rate": 7.859493682923387e-05, "loss": 0.0255, "step": 102270 }, { "epoch": 65.18801784576164, "grad_norm": 0.759575366973877, "learning_rate": 7.859041538435619e-05, "loss": 0.0262, "step": 102280 }, { "epoch": 65.19439133205863, "grad_norm": 0.6905320286750793, "learning_rate": 7.858589359207797e-05, "loss": 0.0268, "step": 102290 }, { "epoch": 65.20076481835564, "grad_norm": 0.23406296968460083, "learning_rate": 7.858137145245416e-05, "loss": 0.0173, "step": 102300 }, { "epoch": 65.20713830465264, "grad_norm": 0.30583304166793823, "learning_rate": 7.85768489655397e-05, "loss": 0.0194, "step": 102310 }, { "epoch": 65.21351179094965, "grad_norm": 0.44565093517303467, "learning_rate": 7.857232613138955e-05, "loss": 0.0297, "step": 102320 }, { "epoch": 65.21988527724666, "grad_norm": 0.4663126766681671, "learning_rate": 7.856780295005865e-05, "loss": 0.0253, "step": 102330 }, { "epoch": 65.22625876354365, "grad_norm": 0.43653979897499084, "learning_rate": 7.856327942160197e-05, "loss": 0.0131, "step": 102340 }, { "epoch": 65.23263224984066, "grad_norm": 0.5796127915382385, "learning_rate": 7.85587555460745e-05, "loss": 0.0215, "step": 102350 }, { "epoch": 65.23900573613767, "grad_norm": 0.5583221912384033, "learning_rate": 7.855423132353117e-05, "loss": 0.0191, "step": 102360 }, { "epoch": 65.24537922243468, "grad_norm": 0.5176437497138977, "learning_rate": 7.854970675402696e-05, "loss": 0.0215, "step": 102370 }, { "epoch": 65.25175270873167, "grad_norm": 0.49771931767463684, "learning_rate": 7.854518183761688e-05, "loss": 0.0131, "step": 102380 }, { "epoch": 65.25812619502868, "grad_norm": 0.4438225328922272, "learning_rate": 7.854065657435588e-05, "loss": 0.0395, "step": 102390 }, { "epoch": 65.26449968132569, "grad_norm": 0.7916486859321594, "learning_rate": 7.853613096429894e-05, "loss": 0.0188, "step": 102400 }, { "epoch": 65.2708731676227, "grad_norm": 0.37696781754493713, "learning_rate": 7.853160500750107e-05, "loss": 0.0146, "step": 102410 }, { "epoch": 65.27724665391969, "grad_norm": 0.2348046600818634, "learning_rate": 7.852707870401728e-05, "loss": 0.0212, "step": 102420 }, { "epoch": 65.2836201402167, "grad_norm": 0.25182101130485535, "learning_rate": 7.852255205390253e-05, "loss": 0.0163, "step": 102430 }, { "epoch": 65.2899936265137, "grad_norm": 0.16943129897117615, "learning_rate": 7.851802505721186e-05, "loss": 0.0223, "step": 102440 }, { "epoch": 65.29636711281071, "grad_norm": 0.3626856207847595, "learning_rate": 7.851349771400025e-05, "loss": 0.0194, "step": 102450 }, { "epoch": 65.30274059910771, "grad_norm": 0.3006613850593567, "learning_rate": 7.850897002432274e-05, "loss": 0.016, "step": 102460 }, { "epoch": 65.30911408540472, "grad_norm": 0.3690333664417267, "learning_rate": 7.850444198823431e-05, "loss": 0.0146, "step": 102470 }, { "epoch": 65.31548757170172, "grad_norm": 0.2569896876811981, "learning_rate": 7.849991360578999e-05, "loss": 0.0272, "step": 102480 }, { "epoch": 65.32186105799873, "grad_norm": 0.6101712584495544, "learning_rate": 7.849538487704483e-05, "loss": 0.0196, "step": 102490 }, { "epoch": 65.32823454429573, "grad_norm": 0.2981135845184326, "learning_rate": 7.849085580205382e-05, "loss": 0.0195, "step": 102500 }, { "epoch": 65.33460803059273, "grad_norm": 0.4539845287799835, "learning_rate": 7.848632638087202e-05, "loss": 0.021, "step": 102510 }, { "epoch": 65.34098151688974, "grad_norm": 0.25157296657562256, "learning_rate": 7.848179661355446e-05, "loss": 0.0261, "step": 102520 }, { "epoch": 65.34735500318675, "grad_norm": 0.3460191786289215, "learning_rate": 7.847726650015619e-05, "loss": 0.0248, "step": 102530 }, { "epoch": 65.35372848948374, "grad_norm": 0.37448257207870483, "learning_rate": 7.847273604073222e-05, "loss": 0.0184, "step": 102540 }, { "epoch": 65.36010197578075, "grad_norm": 0.48398613929748535, "learning_rate": 7.846820523533764e-05, "loss": 0.0268, "step": 102550 }, { "epoch": 65.36647546207776, "grad_norm": 0.41866376996040344, "learning_rate": 7.846367408402749e-05, "loss": 0.0241, "step": 102560 }, { "epoch": 65.37284894837477, "grad_norm": 0.43720099329948425, "learning_rate": 7.845914258685682e-05, "loss": 0.0171, "step": 102570 }, { "epoch": 65.37922243467176, "grad_norm": 0.1811823844909668, "learning_rate": 7.84546107438807e-05, "loss": 0.0181, "step": 102580 }, { "epoch": 65.38559592096877, "grad_norm": 0.35956311225891113, "learning_rate": 7.84500785551542e-05, "loss": 0.0163, "step": 102590 }, { "epoch": 65.39196940726578, "grad_norm": 0.3005768954753876, "learning_rate": 7.844554602073238e-05, "loss": 0.012, "step": 102600 }, { "epoch": 65.39834289356278, "grad_norm": 0.529806911945343, "learning_rate": 7.844101314067032e-05, "loss": 0.0276, "step": 102610 }, { "epoch": 65.40471637985978, "grad_norm": 0.7234546542167664, "learning_rate": 7.843647991502309e-05, "loss": 0.0185, "step": 102620 }, { "epoch": 65.41108986615679, "grad_norm": 0.676247239112854, "learning_rate": 7.843194634384578e-05, "loss": 0.0211, "step": 102630 }, { "epoch": 65.4174633524538, "grad_norm": 0.5337989330291748, "learning_rate": 7.84274124271935e-05, "loss": 0.0226, "step": 102640 }, { "epoch": 65.4238368387508, "grad_norm": 0.3495458960533142, "learning_rate": 7.84228781651213e-05, "loss": 0.0216, "step": 102650 }, { "epoch": 65.4302103250478, "grad_norm": 0.2972962558269501, "learning_rate": 7.841834355768429e-05, "loss": 0.0201, "step": 102660 }, { "epoch": 65.4365838113448, "grad_norm": 0.73860102891922, "learning_rate": 7.841380860493759e-05, "loss": 0.0193, "step": 102670 }, { "epoch": 65.44295729764181, "grad_norm": 0.5754537582397461, "learning_rate": 7.840927330693629e-05, "loss": 0.0127, "step": 102680 }, { "epoch": 65.44933078393882, "grad_norm": 1.4802402257919312, "learning_rate": 7.840473766373548e-05, "loss": 0.0342, "step": 102690 }, { "epoch": 65.45570427023581, "grad_norm": 0.335588663816452, "learning_rate": 7.840020167539028e-05, "loss": 0.0184, "step": 102700 }, { "epoch": 65.46207775653282, "grad_norm": 0.6408677697181702, "learning_rate": 7.839566534195584e-05, "loss": 0.015, "step": 102710 }, { "epoch": 65.46845124282983, "grad_norm": 0.4173913896083832, "learning_rate": 7.839112866348724e-05, "loss": 0.0187, "step": 102720 }, { "epoch": 65.47482472912684, "grad_norm": 0.3144106864929199, "learning_rate": 7.838659164003963e-05, "loss": 0.0181, "step": 102730 }, { "epoch": 65.48119821542383, "grad_norm": 0.2749316096305847, "learning_rate": 7.838205427166814e-05, "loss": 0.031, "step": 102740 }, { "epoch": 65.48757170172084, "grad_norm": 0.2788013517856598, "learning_rate": 7.837751655842788e-05, "loss": 0.0169, "step": 102750 }, { "epoch": 65.49394518801785, "grad_norm": 0.4586140513420105, "learning_rate": 7.837297850037399e-05, "loss": 0.0181, "step": 102760 }, { "epoch": 65.50031867431485, "grad_norm": 0.4848775863647461, "learning_rate": 7.836844009756164e-05, "loss": 0.0214, "step": 102770 }, { "epoch": 65.50669216061185, "grad_norm": 0.42470505833625793, "learning_rate": 7.836390135004596e-05, "loss": 0.0204, "step": 102780 }, { "epoch": 65.51306564690886, "grad_norm": 0.3330576717853546, "learning_rate": 7.83593622578821e-05, "loss": 0.0182, "step": 102790 }, { "epoch": 65.51943913320586, "grad_norm": 0.3691929876804352, "learning_rate": 7.83548228211252e-05, "loss": 0.02, "step": 102800 }, { "epoch": 65.52581261950287, "grad_norm": 0.6243945956230164, "learning_rate": 7.835028303983043e-05, "loss": 0.0318, "step": 102810 }, { "epoch": 65.53218610579987, "grad_norm": 0.9150430560112, "learning_rate": 7.834574291405297e-05, "loss": 0.0233, "step": 102820 }, { "epoch": 65.53855959209687, "grad_norm": 0.38269171118736267, "learning_rate": 7.834120244384795e-05, "loss": 0.0232, "step": 102830 }, { "epoch": 65.54493307839388, "grad_norm": 0.4879479706287384, "learning_rate": 7.833666162927058e-05, "loss": 0.0186, "step": 102840 }, { "epoch": 65.55130656469089, "grad_norm": 0.4995993673801422, "learning_rate": 7.833212047037602e-05, "loss": 0.0203, "step": 102850 }, { "epoch": 65.55768005098788, "grad_norm": 0.6419657468795776, "learning_rate": 7.832757896721943e-05, "loss": 0.0196, "step": 102860 }, { "epoch": 65.56405353728489, "grad_norm": 0.48423707485198975, "learning_rate": 7.832303711985602e-05, "loss": 0.0204, "step": 102870 }, { "epoch": 65.5704270235819, "grad_norm": 0.3181213140487671, "learning_rate": 7.831849492834096e-05, "loss": 0.023, "step": 102880 }, { "epoch": 65.57680050987891, "grad_norm": 0.4334322214126587, "learning_rate": 7.831395239272945e-05, "loss": 0.0199, "step": 102890 }, { "epoch": 65.5831739961759, "grad_norm": 0.4734525680541992, "learning_rate": 7.830940951307669e-05, "loss": 0.0179, "step": 102900 }, { "epoch": 65.58954748247291, "grad_norm": 0.3188893795013428, "learning_rate": 7.830486628943788e-05, "loss": 0.0252, "step": 102910 }, { "epoch": 65.59592096876992, "grad_norm": 0.36049970984458923, "learning_rate": 7.830032272186822e-05, "loss": 0.0275, "step": 102920 }, { "epoch": 65.60229445506693, "grad_norm": 0.4017672836780548, "learning_rate": 7.829577881042291e-05, "loss": 0.0203, "step": 102930 }, { "epoch": 65.60866794136393, "grad_norm": 0.7531338334083557, "learning_rate": 7.829123455515718e-05, "loss": 0.0281, "step": 102940 }, { "epoch": 65.61504142766093, "grad_norm": 0.28667938709259033, "learning_rate": 7.828668995612622e-05, "loss": 0.0245, "step": 102950 }, { "epoch": 65.62141491395793, "grad_norm": 0.4678466320037842, "learning_rate": 7.82821450133853e-05, "loss": 0.0271, "step": 102960 }, { "epoch": 65.62778840025494, "grad_norm": 0.5310764908790588, "learning_rate": 7.82775997269896e-05, "loss": 0.0188, "step": 102970 }, { "epoch": 65.63416188655195, "grad_norm": 0.5667780637741089, "learning_rate": 7.827305409699437e-05, "loss": 0.0163, "step": 102980 }, { "epoch": 65.64053537284894, "grad_norm": 0.2960823178291321, "learning_rate": 7.826850812345483e-05, "loss": 0.0131, "step": 102990 }, { "epoch": 65.64690885914595, "grad_norm": 0.28298553824424744, "learning_rate": 7.826396180642623e-05, "loss": 0.0151, "step": 103000 }, { "epoch": 65.65328234544296, "grad_norm": 0.3857429325580597, "learning_rate": 7.825941514596381e-05, "loss": 0.0146, "step": 103010 }, { "epoch": 65.65965583173997, "grad_norm": 0.24088533222675323, "learning_rate": 7.825486814212282e-05, "loss": 0.0162, "step": 103020 }, { "epoch": 65.66602931803696, "grad_norm": 0.3241817057132721, "learning_rate": 7.825032079495849e-05, "loss": 0.0283, "step": 103030 }, { "epoch": 65.67240280433397, "grad_norm": 0.371898353099823, "learning_rate": 7.82457731045261e-05, "loss": 0.018, "step": 103040 }, { "epoch": 65.67877629063098, "grad_norm": 0.6257537603378296, "learning_rate": 7.82412250708809e-05, "loss": 0.0195, "step": 103050 }, { "epoch": 65.68514977692799, "grad_norm": 0.533287763595581, "learning_rate": 7.823667669407815e-05, "loss": 0.0169, "step": 103060 }, { "epoch": 65.69152326322498, "grad_norm": 0.34694167971611023, "learning_rate": 7.823212797417311e-05, "loss": 0.0193, "step": 103070 }, { "epoch": 65.69789674952199, "grad_norm": 0.30658912658691406, "learning_rate": 7.822757891122107e-05, "loss": 0.0129, "step": 103080 }, { "epoch": 65.704270235819, "grad_norm": 0.3185724914073944, "learning_rate": 7.822302950527729e-05, "loss": 0.0192, "step": 103090 }, { "epoch": 65.710643722116, "grad_norm": 0.7462679743766785, "learning_rate": 7.821847975639705e-05, "loss": 0.018, "step": 103100 }, { "epoch": 65.717017208413, "grad_norm": 0.29510313272476196, "learning_rate": 7.821392966463564e-05, "loss": 0.0135, "step": 103110 }, { "epoch": 65.72339069471, "grad_norm": 0.32646939158439636, "learning_rate": 7.820937923004835e-05, "loss": 0.0176, "step": 103120 }, { "epoch": 65.72976418100701, "grad_norm": 0.5003933906555176, "learning_rate": 7.820482845269047e-05, "loss": 0.0204, "step": 103130 }, { "epoch": 65.73613766730402, "grad_norm": 0.5240833163261414, "learning_rate": 7.82002773326173e-05, "loss": 0.0407, "step": 103140 }, { "epoch": 65.74251115360101, "grad_norm": 0.349303275346756, "learning_rate": 7.819572586988413e-05, "loss": 0.0262, "step": 103150 }, { "epoch": 65.74888463989802, "grad_norm": 0.27445346117019653, "learning_rate": 7.819117406454626e-05, "loss": 0.0268, "step": 103160 }, { "epoch": 65.75525812619503, "grad_norm": 0.5163291692733765, "learning_rate": 7.818662191665901e-05, "loss": 0.0201, "step": 103170 }, { "epoch": 65.76163161249204, "grad_norm": 0.5932561159133911, "learning_rate": 7.81820694262777e-05, "loss": 0.0396, "step": 103180 }, { "epoch": 65.76800509878903, "grad_norm": 0.3425849974155426, "learning_rate": 7.817751659345765e-05, "loss": 0.0215, "step": 103190 }, { "epoch": 65.77437858508604, "grad_norm": 0.29944124817848206, "learning_rate": 7.817296341825413e-05, "loss": 0.0173, "step": 103200 }, { "epoch": 65.78075207138305, "grad_norm": 0.3647506535053253, "learning_rate": 7.816840990072254e-05, "loss": 0.0172, "step": 103210 }, { "epoch": 65.78712555768006, "grad_norm": 0.34036916494369507, "learning_rate": 7.816385604091815e-05, "loss": 0.0197, "step": 103220 }, { "epoch": 65.79349904397705, "grad_norm": 0.5521475672721863, "learning_rate": 7.815930183889635e-05, "loss": 0.0252, "step": 103230 }, { "epoch": 65.79987253027406, "grad_norm": 0.6016308665275574, "learning_rate": 7.815474729471241e-05, "loss": 0.0193, "step": 103240 }, { "epoch": 65.80624601657107, "grad_norm": 0.5104973316192627, "learning_rate": 7.815019240842174e-05, "loss": 0.0156, "step": 103250 }, { "epoch": 65.81261950286807, "grad_norm": 0.34524697065353394, "learning_rate": 7.814563718007963e-05, "loss": 0.0161, "step": 103260 }, { "epoch": 65.81899298916507, "grad_norm": 0.43747833371162415, "learning_rate": 7.814108160974148e-05, "loss": 0.0166, "step": 103270 }, { "epoch": 65.82536647546208, "grad_norm": 0.30830204486846924, "learning_rate": 7.81365256974626e-05, "loss": 0.017, "step": 103280 }, { "epoch": 65.83173996175908, "grad_norm": 0.4012058973312378, "learning_rate": 7.813196944329836e-05, "loss": 0.0144, "step": 103290 }, { "epoch": 65.83811344805609, "grad_norm": 0.3441618084907532, "learning_rate": 7.812741284730413e-05, "loss": 0.0174, "step": 103300 }, { "epoch": 65.84448693435309, "grad_norm": 0.4442175626754761, "learning_rate": 7.81228559095353e-05, "loss": 0.0178, "step": 103310 }, { "epoch": 65.8508604206501, "grad_norm": 0.35823187232017517, "learning_rate": 7.81182986300472e-05, "loss": 0.0166, "step": 103320 }, { "epoch": 65.8572339069471, "grad_norm": 0.522636353969574, "learning_rate": 7.811374100889524e-05, "loss": 0.0134, "step": 103330 }, { "epoch": 65.86360739324411, "grad_norm": 0.355129212141037, "learning_rate": 7.810918304613478e-05, "loss": 0.0223, "step": 103340 }, { "epoch": 65.8699808795411, "grad_norm": 0.3701978921890259, "learning_rate": 7.810462474182119e-05, "loss": 0.0246, "step": 103350 }, { "epoch": 65.87635436583811, "grad_norm": 0.6545206308364868, "learning_rate": 7.810006609600988e-05, "loss": 0.0212, "step": 103360 }, { "epoch": 65.88272785213512, "grad_norm": 0.4972892701625824, "learning_rate": 7.809550710875624e-05, "loss": 0.0277, "step": 103370 }, { "epoch": 65.88910133843213, "grad_norm": 0.5639822483062744, "learning_rate": 7.809094778011565e-05, "loss": 0.0231, "step": 103380 }, { "epoch": 65.89547482472912, "grad_norm": 0.44901537895202637, "learning_rate": 7.808638811014354e-05, "loss": 0.0316, "step": 103390 }, { "epoch": 65.90184831102613, "grad_norm": 0.46642786264419556, "learning_rate": 7.80818280988953e-05, "loss": 0.0197, "step": 103400 }, { "epoch": 65.90822179732314, "grad_norm": 0.30104172229766846, "learning_rate": 7.807726774642631e-05, "loss": 0.0192, "step": 103410 }, { "epoch": 65.91459528362014, "grad_norm": 0.6449616551399231, "learning_rate": 7.807270705279204e-05, "loss": 0.0157, "step": 103420 }, { "epoch": 65.92096876991714, "grad_norm": 0.2343538999557495, "learning_rate": 7.806814601804784e-05, "loss": 0.0211, "step": 103430 }, { "epoch": 65.92734225621415, "grad_norm": 0.6807414293289185, "learning_rate": 7.806358464224919e-05, "loss": 0.034, "step": 103440 }, { "epoch": 65.93371574251115, "grad_norm": 0.24237528443336487, "learning_rate": 7.805902292545147e-05, "loss": 0.0125, "step": 103450 }, { "epoch": 65.94008922880816, "grad_norm": 0.6697959899902344, "learning_rate": 7.805446086771016e-05, "loss": 0.0189, "step": 103460 }, { "epoch": 65.94646271510516, "grad_norm": 0.46799135208129883, "learning_rate": 7.804989846908063e-05, "loss": 0.017, "step": 103470 }, { "epoch": 65.95283620140216, "grad_norm": 0.268690288066864, "learning_rate": 7.804533572961836e-05, "loss": 0.0162, "step": 103480 }, { "epoch": 65.95920968769917, "grad_norm": 0.32006779313087463, "learning_rate": 7.80407726493788e-05, "loss": 0.0176, "step": 103490 }, { "epoch": 65.96558317399618, "grad_norm": 0.5648157596588135, "learning_rate": 7.803620922841736e-05, "loss": 0.0271, "step": 103500 }, { "epoch": 65.97195666029317, "grad_norm": 0.3454248309135437, "learning_rate": 7.80316454667895e-05, "loss": 0.0123, "step": 103510 }, { "epoch": 65.97833014659018, "grad_norm": 0.7566028833389282, "learning_rate": 7.802708136455071e-05, "loss": 0.0166, "step": 103520 }, { "epoch": 65.98470363288719, "grad_norm": 0.24555304646492004, "learning_rate": 7.802251692175638e-05, "loss": 0.0254, "step": 103530 }, { "epoch": 65.9910771191842, "grad_norm": 0.7986894845962524, "learning_rate": 7.801795213846204e-05, "loss": 0.0191, "step": 103540 }, { "epoch": 65.99745060548119, "grad_norm": 0.38679730892181396, "learning_rate": 7.801338701472311e-05, "loss": 0.0176, "step": 103550 }, { "epoch": 66.0038240917782, "grad_norm": 0.9482989311218262, "learning_rate": 7.800882155059511e-05, "loss": 0.0281, "step": 103560 }, { "epoch": 66.0101975780752, "grad_norm": 0.27796244621276855, "learning_rate": 7.800425574613345e-05, "loss": 0.015, "step": 103570 }, { "epoch": 66.01657106437222, "grad_norm": 0.2370956540107727, "learning_rate": 7.799968960139366e-05, "loss": 0.0151, "step": 103580 }, { "epoch": 66.02294455066922, "grad_norm": 0.3358617126941681, "learning_rate": 7.79951231164312e-05, "loss": 0.0269, "step": 103590 }, { "epoch": 66.02931803696622, "grad_norm": 0.5765673518180847, "learning_rate": 7.799055629130156e-05, "loss": 0.0155, "step": 103600 }, { "epoch": 66.03569152326322, "grad_norm": 0.4407922923564911, "learning_rate": 7.798598912606024e-05, "loss": 0.0221, "step": 103610 }, { "epoch": 66.04206500956023, "grad_norm": 0.325094074010849, "learning_rate": 7.798142162076273e-05, "loss": 0.0192, "step": 103620 }, { "epoch": 66.04843849585724, "grad_norm": 0.45933645963668823, "learning_rate": 7.797685377546451e-05, "loss": 0.0167, "step": 103630 }, { "epoch": 66.05481198215423, "grad_norm": 0.25464367866516113, "learning_rate": 7.797228559022113e-05, "loss": 0.0132, "step": 103640 }, { "epoch": 66.06118546845124, "grad_norm": 0.5071162581443787, "learning_rate": 7.796771706508805e-05, "loss": 0.0378, "step": 103650 }, { "epoch": 66.06755895474825, "grad_norm": 0.3177618384361267, "learning_rate": 7.796314820012081e-05, "loss": 0.0204, "step": 103660 }, { "epoch": 66.07393244104526, "grad_norm": 0.2850005328655243, "learning_rate": 7.795857899537492e-05, "loss": 0.0181, "step": 103670 }, { "epoch": 66.08030592734225, "grad_norm": 0.5369991660118103, "learning_rate": 7.79540094509059e-05, "loss": 0.0162, "step": 103680 }, { "epoch": 66.08667941363926, "grad_norm": 0.3731865882873535, "learning_rate": 7.794943956676925e-05, "loss": 0.0142, "step": 103690 }, { "epoch": 66.09305289993627, "grad_norm": 0.4750994145870209, "learning_rate": 7.794486934302054e-05, "loss": 0.0215, "step": 103700 }, { "epoch": 66.09942638623328, "grad_norm": 0.26797664165496826, "learning_rate": 7.794029877971529e-05, "loss": 0.0121, "step": 103710 }, { "epoch": 66.10579987253027, "grad_norm": 0.4954093098640442, "learning_rate": 7.793572787690902e-05, "loss": 0.0186, "step": 103720 }, { "epoch": 66.11217335882728, "grad_norm": 0.6190300583839417, "learning_rate": 7.793115663465727e-05, "loss": 0.0271, "step": 103730 }, { "epoch": 66.11854684512429, "grad_norm": 0.7718958258628845, "learning_rate": 7.792658505301561e-05, "loss": 0.0153, "step": 103740 }, { "epoch": 66.1249203314213, "grad_norm": 0.637861430644989, "learning_rate": 7.792201313203957e-05, "loss": 0.0153, "step": 103750 }, { "epoch": 66.13129381771829, "grad_norm": 0.39405620098114014, "learning_rate": 7.791744087178471e-05, "loss": 0.0174, "step": 103760 }, { "epoch": 66.1376673040153, "grad_norm": 0.2939327657222748, "learning_rate": 7.791286827230657e-05, "loss": 0.0184, "step": 103770 }, { "epoch": 66.1440407903123, "grad_norm": 0.3212132751941681, "learning_rate": 7.790829533366075e-05, "loss": 0.018, "step": 103780 }, { "epoch": 66.15041427660931, "grad_norm": 0.30288732051849365, "learning_rate": 7.790372205590277e-05, "loss": 0.0174, "step": 103790 }, { "epoch": 66.1567877629063, "grad_norm": 0.30808648467063904, "learning_rate": 7.789914843908824e-05, "loss": 0.02, "step": 103800 }, { "epoch": 66.16316124920331, "grad_norm": 0.21060657501220703, "learning_rate": 7.78945744832727e-05, "loss": 0.0325, "step": 103810 }, { "epoch": 66.16953473550032, "grad_norm": 0.48340728878974915, "learning_rate": 7.789000018851176e-05, "loss": 0.0148, "step": 103820 }, { "epoch": 66.17590822179733, "grad_norm": 0.30996447801589966, "learning_rate": 7.788542555486098e-05, "loss": 0.0184, "step": 103830 }, { "epoch": 66.18228170809432, "grad_norm": 0.4478999078273773, "learning_rate": 7.788085058237594e-05, "loss": 0.0191, "step": 103840 }, { "epoch": 66.18865519439133, "grad_norm": 0.6804331541061401, "learning_rate": 7.787627527111226e-05, "loss": 0.0189, "step": 103850 }, { "epoch": 66.19502868068834, "grad_norm": 0.8778570890426636, "learning_rate": 7.78716996211255e-05, "loss": 0.0174, "step": 103860 }, { "epoch": 66.20140216698535, "grad_norm": 0.680188775062561, "learning_rate": 7.786712363247129e-05, "loss": 0.0235, "step": 103870 }, { "epoch": 66.20777565328234, "grad_norm": 0.24531130492687225, "learning_rate": 7.786254730520519e-05, "loss": 0.0239, "step": 103880 }, { "epoch": 66.21414913957935, "grad_norm": 0.3847886919975281, "learning_rate": 7.785797063938284e-05, "loss": 0.0148, "step": 103890 }, { "epoch": 66.22052262587636, "grad_norm": 0.6601198315620422, "learning_rate": 7.785339363505987e-05, "loss": 0.0264, "step": 103900 }, { "epoch": 66.22689611217336, "grad_norm": 0.5291872620582581, "learning_rate": 7.784881629229185e-05, "loss": 0.0168, "step": 103910 }, { "epoch": 66.23326959847036, "grad_norm": 0.5817005634307861, "learning_rate": 7.784423861113443e-05, "loss": 0.0205, "step": 103920 }, { "epoch": 66.23964308476737, "grad_norm": 0.468478798866272, "learning_rate": 7.78396605916432e-05, "loss": 0.017, "step": 103930 }, { "epoch": 66.24601657106437, "grad_norm": 0.3667430579662323, "learning_rate": 7.783508223387383e-05, "loss": 0.0132, "step": 103940 }, { "epoch": 66.25239005736138, "grad_norm": 0.40649276971817017, "learning_rate": 7.783050353788193e-05, "loss": 0.0198, "step": 103950 }, { "epoch": 66.25876354365838, "grad_norm": 0.42306339740753174, "learning_rate": 7.782592450372312e-05, "loss": 0.0313, "step": 103960 }, { "epoch": 66.26513702995538, "grad_norm": 0.5502456426620483, "learning_rate": 7.782134513145307e-05, "loss": 0.0148, "step": 103970 }, { "epoch": 66.27151051625239, "grad_norm": 0.4431155323982239, "learning_rate": 7.781676542112739e-05, "loss": 0.0303, "step": 103980 }, { "epoch": 66.2778840025494, "grad_norm": 0.3459029197692871, "learning_rate": 7.781218537280178e-05, "loss": 0.0165, "step": 103990 }, { "epoch": 66.2842574888464, "grad_norm": 0.38782626390457153, "learning_rate": 7.780760498653183e-05, "loss": 0.0175, "step": 104000 }, { "epoch": 66.2906309751434, "grad_norm": 0.2680876851081848, "learning_rate": 7.780302426237323e-05, "loss": 0.0226, "step": 104010 }, { "epoch": 66.29700446144041, "grad_norm": 0.326782763004303, "learning_rate": 7.779844320038163e-05, "loss": 0.0203, "step": 104020 }, { "epoch": 66.30337794773742, "grad_norm": 0.7530038356781006, "learning_rate": 7.779386180061272e-05, "loss": 0.0271, "step": 104030 }, { "epoch": 66.30975143403441, "grad_norm": 0.2513943314552307, "learning_rate": 7.778928006312212e-05, "loss": 0.0174, "step": 104040 }, { "epoch": 66.31612492033142, "grad_norm": 0.35788676142692566, "learning_rate": 7.778469798796555e-05, "loss": 0.0274, "step": 104050 }, { "epoch": 66.32249840662843, "grad_norm": 0.290964275598526, "learning_rate": 7.778011557519867e-05, "loss": 0.0112, "step": 104060 }, { "epoch": 66.32887189292543, "grad_norm": 0.389392614364624, "learning_rate": 7.777553282487714e-05, "loss": 0.0259, "step": 104070 }, { "epoch": 66.33524537922243, "grad_norm": 0.32394707202911377, "learning_rate": 7.777094973705668e-05, "loss": 0.0219, "step": 104080 }, { "epoch": 66.34161886551944, "grad_norm": 0.5177465677261353, "learning_rate": 7.776636631179292e-05, "loss": 0.0188, "step": 104090 }, { "epoch": 66.34799235181644, "grad_norm": 0.6246376633644104, "learning_rate": 7.776178254914163e-05, "loss": 0.0192, "step": 104100 }, { "epoch": 66.35436583811345, "grad_norm": 0.2061404436826706, "learning_rate": 7.775719844915848e-05, "loss": 0.0175, "step": 104110 }, { "epoch": 66.36073932441045, "grad_norm": 0.5537052154541016, "learning_rate": 7.775261401189914e-05, "loss": 0.0189, "step": 104120 }, { "epoch": 66.36711281070745, "grad_norm": 0.38510414958000183, "learning_rate": 7.774802923741934e-05, "loss": 0.0372, "step": 104130 }, { "epoch": 66.37348629700446, "grad_norm": 0.30153512954711914, "learning_rate": 7.774344412577479e-05, "loss": 0.0149, "step": 104140 }, { "epoch": 66.37985978330147, "grad_norm": 0.5645362138748169, "learning_rate": 7.77388586770212e-05, "loss": 0.0263, "step": 104150 }, { "epoch": 66.38623326959846, "grad_norm": 0.2406090497970581, "learning_rate": 7.773427289121429e-05, "loss": 0.0126, "step": 104160 }, { "epoch": 66.39260675589547, "grad_norm": 0.29537901282310486, "learning_rate": 7.772968676840977e-05, "loss": 0.0162, "step": 104170 }, { "epoch": 66.39898024219248, "grad_norm": 0.33754897117614746, "learning_rate": 7.772510030866339e-05, "loss": 0.0227, "step": 104180 }, { "epoch": 66.40535372848949, "grad_norm": 0.6791728734970093, "learning_rate": 7.772051351203085e-05, "loss": 0.0172, "step": 104190 }, { "epoch": 66.4117272147865, "grad_norm": 0.4084722101688385, "learning_rate": 7.771592637856789e-05, "loss": 0.0123, "step": 104200 }, { "epoch": 66.41810070108349, "grad_norm": 0.31572774052619934, "learning_rate": 7.771133890833026e-05, "loss": 0.017, "step": 104210 }, { "epoch": 66.4244741873805, "grad_norm": 0.3071862459182739, "learning_rate": 7.770675110137371e-05, "loss": 0.0148, "step": 104220 }, { "epoch": 66.4308476736775, "grad_norm": 0.2591749131679535, "learning_rate": 7.770216295775398e-05, "loss": 0.0138, "step": 104230 }, { "epoch": 66.43722115997451, "grad_norm": 0.2720262408256531, "learning_rate": 7.76975744775268e-05, "loss": 0.0157, "step": 104240 }, { "epoch": 66.4435946462715, "grad_norm": 0.2967155873775482, "learning_rate": 7.769298566074794e-05, "loss": 0.0108, "step": 104250 }, { "epoch": 66.44996813256851, "grad_norm": 0.4368475377559662, "learning_rate": 7.768839650747317e-05, "loss": 0.0129, "step": 104260 }, { "epoch": 66.45634161886552, "grad_norm": 0.3891008496284485, "learning_rate": 7.768380701775822e-05, "loss": 0.0301, "step": 104270 }, { "epoch": 66.46271510516253, "grad_norm": 0.5583856701850891, "learning_rate": 7.767921719165889e-05, "loss": 0.0162, "step": 104280 }, { "epoch": 66.46908859145952, "grad_norm": 0.37496843934059143, "learning_rate": 7.767462702923095e-05, "loss": 0.0138, "step": 104290 }, { "epoch": 66.47546207775653, "grad_norm": 0.5572828650474548, "learning_rate": 7.767003653053015e-05, "loss": 0.0148, "step": 104300 }, { "epoch": 66.48183556405354, "grad_norm": 0.43109455704689026, "learning_rate": 7.766544569561228e-05, "loss": 0.0178, "step": 104310 }, { "epoch": 66.48820905035055, "grad_norm": 0.42166683077812195, "learning_rate": 7.766085452453312e-05, "loss": 0.0221, "step": 104320 }, { "epoch": 66.49458253664754, "grad_norm": 0.4453985095024109, "learning_rate": 7.765626301734848e-05, "loss": 0.0188, "step": 104330 }, { "epoch": 66.50095602294455, "grad_norm": 0.28248581290245056, "learning_rate": 7.765167117411412e-05, "loss": 0.0296, "step": 104340 }, { "epoch": 66.50732950924156, "grad_norm": 0.29273292422294617, "learning_rate": 7.764707899488584e-05, "loss": 0.0175, "step": 104350 }, { "epoch": 66.51370299553857, "grad_norm": 0.5694992542266846, "learning_rate": 7.764248647971945e-05, "loss": 0.0215, "step": 104360 }, { "epoch": 66.52007648183556, "grad_norm": 0.4754321873188019, "learning_rate": 7.763789362867075e-05, "loss": 0.0118, "step": 104370 }, { "epoch": 66.52644996813257, "grad_norm": 0.35538235306739807, "learning_rate": 7.763330044179557e-05, "loss": 0.0204, "step": 104380 }, { "epoch": 66.53282345442958, "grad_norm": 0.44605621695518494, "learning_rate": 7.762870691914968e-05, "loss": 0.0183, "step": 104390 }, { "epoch": 66.53919694072658, "grad_norm": 0.3377116322517395, "learning_rate": 7.762411306078892e-05, "loss": 0.0176, "step": 104400 }, { "epoch": 66.54557042702358, "grad_norm": 0.3596092462539673, "learning_rate": 7.761951886676908e-05, "loss": 0.0264, "step": 104410 }, { "epoch": 66.55194391332059, "grad_norm": 0.5701419115066528, "learning_rate": 7.761492433714604e-05, "loss": 0.0244, "step": 104420 }, { "epoch": 66.5583173996176, "grad_norm": 0.45920702815055847, "learning_rate": 7.761032947197558e-05, "loss": 0.0265, "step": 104430 }, { "epoch": 66.5646908859146, "grad_norm": 0.34056341648101807, "learning_rate": 7.760573427131356e-05, "loss": 0.0145, "step": 104440 }, { "epoch": 66.5710643722116, "grad_norm": 0.48435258865356445, "learning_rate": 7.760113873521581e-05, "loss": 0.0253, "step": 104450 }, { "epoch": 66.5774378585086, "grad_norm": 0.32668203115463257, "learning_rate": 7.759654286373814e-05, "loss": 0.0199, "step": 104460 }, { "epoch": 66.58381134480561, "grad_norm": 0.2782791256904602, "learning_rate": 7.759194665693646e-05, "loss": 0.0306, "step": 104470 }, { "epoch": 66.59018483110262, "grad_norm": 0.31007471680641174, "learning_rate": 7.758735011486653e-05, "loss": 0.0252, "step": 104480 }, { "epoch": 66.59655831739961, "grad_norm": 0.5312166810035706, "learning_rate": 7.758275323758429e-05, "loss": 0.011, "step": 104490 }, { "epoch": 66.60293180369662, "grad_norm": 0.33997178077697754, "learning_rate": 7.757815602514552e-05, "loss": 0.0185, "step": 104500 }, { "epoch": 66.60930528999363, "grad_norm": 0.30017730593681335, "learning_rate": 7.757355847760613e-05, "loss": 0.025, "step": 104510 }, { "epoch": 66.61567877629064, "grad_norm": 0.24849814176559448, "learning_rate": 7.756896059502197e-05, "loss": 0.0157, "step": 104520 }, { "epoch": 66.62205226258763, "grad_norm": 0.4428296387195587, "learning_rate": 7.756436237744891e-05, "loss": 0.0236, "step": 104530 }, { "epoch": 66.62842574888464, "grad_norm": 0.3234385848045349, "learning_rate": 7.755976382494284e-05, "loss": 0.0163, "step": 104540 }, { "epoch": 66.63479923518165, "grad_norm": 0.5087635517120361, "learning_rate": 7.75551649375596e-05, "loss": 0.0135, "step": 104550 }, { "epoch": 66.64117272147865, "grad_norm": 0.3576057553291321, "learning_rate": 7.755056571535508e-05, "loss": 0.0141, "step": 104560 }, { "epoch": 66.64754620777565, "grad_norm": 0.2613471448421478, "learning_rate": 7.754596615838518e-05, "loss": 0.0157, "step": 104570 }, { "epoch": 66.65391969407266, "grad_norm": 0.5001068711280823, "learning_rate": 7.754136626670579e-05, "loss": 0.0193, "step": 104580 }, { "epoch": 66.66029318036966, "grad_norm": 0.3789062798023224, "learning_rate": 7.753676604037282e-05, "loss": 0.0181, "step": 104590 }, { "epoch": 66.66666666666667, "grad_norm": 0.20813186466693878, "learning_rate": 7.753216547944211e-05, "loss": 0.031, "step": 104600 }, { "epoch": 66.67304015296367, "grad_norm": 0.35968807339668274, "learning_rate": 7.75275645839696e-05, "loss": 0.0227, "step": 104610 }, { "epoch": 66.67941363926067, "grad_norm": 0.5767723321914673, "learning_rate": 7.75229633540112e-05, "loss": 0.0244, "step": 104620 }, { "epoch": 66.68578712555768, "grad_norm": 0.6653258204460144, "learning_rate": 7.75183617896228e-05, "loss": 0.0301, "step": 104630 }, { "epoch": 66.69216061185469, "grad_norm": 0.28615257143974304, "learning_rate": 7.751375989086032e-05, "loss": 0.0228, "step": 104640 }, { "epoch": 66.69853409815168, "grad_norm": 0.4136192500591278, "learning_rate": 7.75091576577797e-05, "loss": 0.0121, "step": 104650 }, { "epoch": 66.70490758444869, "grad_norm": 0.567590594291687, "learning_rate": 7.750455509043681e-05, "loss": 0.029, "step": 104660 }, { "epoch": 66.7112810707457, "grad_norm": 0.3617996573448181, "learning_rate": 7.749995218888763e-05, "loss": 0.0262, "step": 104670 }, { "epoch": 66.7176545570427, "grad_norm": 0.5951144099235535, "learning_rate": 7.749534895318806e-05, "loss": 0.0099, "step": 104680 }, { "epoch": 66.7240280433397, "grad_norm": 0.501395583152771, "learning_rate": 7.749074538339405e-05, "loss": 0.0183, "step": 104690 }, { "epoch": 66.73040152963671, "grad_norm": 1.0098766088485718, "learning_rate": 7.748614147956153e-05, "loss": 0.0276, "step": 104700 }, { "epoch": 66.73677501593372, "grad_norm": 0.49463239312171936, "learning_rate": 7.748153724174642e-05, "loss": 0.0254, "step": 104710 }, { "epoch": 66.74314850223072, "grad_norm": 0.8672518134117126, "learning_rate": 7.747693267000469e-05, "loss": 0.0266, "step": 104720 }, { "epoch": 66.74952198852772, "grad_norm": 0.3697878420352936, "learning_rate": 7.74723277643923e-05, "loss": 0.0198, "step": 104730 }, { "epoch": 66.75589547482473, "grad_norm": 0.32793936133384705, "learning_rate": 7.746772252496519e-05, "loss": 0.0231, "step": 104740 }, { "epoch": 66.76226896112173, "grad_norm": 0.37291669845581055, "learning_rate": 7.74631169517793e-05, "loss": 0.0228, "step": 104750 }, { "epoch": 66.76864244741874, "grad_norm": 0.3412865996360779, "learning_rate": 7.745851104489061e-05, "loss": 0.0189, "step": 104760 }, { "epoch": 66.77501593371574, "grad_norm": 0.5117918252944946, "learning_rate": 7.74539048043551e-05, "loss": 0.0203, "step": 104770 }, { "epoch": 66.78138942001274, "grad_norm": 0.3399236798286438, "learning_rate": 7.744929823022872e-05, "loss": 0.0282, "step": 104780 }, { "epoch": 66.78776290630975, "grad_norm": 3.5026469230651855, "learning_rate": 7.744469132256746e-05, "loss": 0.0174, "step": 104790 }, { "epoch": 66.79413639260676, "grad_norm": 0.5342695713043213, "learning_rate": 7.744008408142726e-05, "loss": 0.0174, "step": 104800 }, { "epoch": 66.80050987890377, "grad_norm": 0.6316962242126465, "learning_rate": 7.743547650686415e-05, "loss": 0.0184, "step": 104810 }, { "epoch": 66.80688336520076, "grad_norm": 30.871097564697266, "learning_rate": 7.74308685989341e-05, "loss": 0.0254, "step": 104820 }, { "epoch": 66.81325685149777, "grad_norm": 0.3209657669067383, "learning_rate": 7.742626035769309e-05, "loss": 0.0261, "step": 104830 }, { "epoch": 66.81963033779478, "grad_norm": 0.3342027962207794, "learning_rate": 7.742165178319712e-05, "loss": 0.0222, "step": 104840 }, { "epoch": 66.82600382409179, "grad_norm": 0.35139378905296326, "learning_rate": 7.74170428755022e-05, "loss": 0.0142, "step": 104850 }, { "epoch": 66.83237731038878, "grad_norm": 0.923721969127655, "learning_rate": 7.741243363466433e-05, "loss": 0.0298, "step": 104860 }, { "epoch": 66.83875079668579, "grad_norm": 0.3711704909801483, "learning_rate": 7.74078240607395e-05, "loss": 0.0191, "step": 104870 }, { "epoch": 66.8451242829828, "grad_norm": 0.5540165305137634, "learning_rate": 7.740321415378373e-05, "loss": 0.0203, "step": 104880 }, { "epoch": 66.8514977692798, "grad_norm": 0.3774334192276001, "learning_rate": 7.739860391385303e-05, "loss": 0.0194, "step": 104890 }, { "epoch": 66.8578712555768, "grad_norm": 0.7168303728103638, "learning_rate": 7.739399334100342e-05, "loss": 0.0217, "step": 104900 }, { "epoch": 66.8642447418738, "grad_norm": 0.46840304136276245, "learning_rate": 7.738938243529094e-05, "loss": 0.0158, "step": 104910 }, { "epoch": 66.87061822817081, "grad_norm": 0.2594188451766968, "learning_rate": 7.73847711967716e-05, "loss": 0.0231, "step": 104920 }, { "epoch": 66.87699171446782, "grad_norm": 0.3350170850753784, "learning_rate": 7.738015962550142e-05, "loss": 0.0178, "step": 104930 }, { "epoch": 66.88336520076481, "grad_norm": 0.3165241777896881, "learning_rate": 7.737554772153647e-05, "loss": 0.0276, "step": 104940 }, { "epoch": 66.88973868706182, "grad_norm": 0.30383965373039246, "learning_rate": 7.737093548493276e-05, "loss": 0.0156, "step": 104950 }, { "epoch": 66.89611217335883, "grad_norm": 0.3588404059410095, "learning_rate": 7.736632291574634e-05, "loss": 0.0155, "step": 104960 }, { "epoch": 66.90248565965584, "grad_norm": 0.49292972683906555, "learning_rate": 7.736171001403326e-05, "loss": 0.0133, "step": 104970 }, { "epoch": 66.90885914595283, "grad_norm": 0.45548707246780396, "learning_rate": 7.735709677984957e-05, "loss": 0.0228, "step": 104980 }, { "epoch": 66.91523263224984, "grad_norm": 0.6077223420143127, "learning_rate": 7.735248321325133e-05, "loss": 0.0173, "step": 104990 }, { "epoch": 66.92160611854685, "grad_norm": 0.5410031676292419, "learning_rate": 7.734786931429457e-05, "loss": 0.0192, "step": 105000 }, { "epoch": 66.92797960484386, "grad_norm": 0.4797418713569641, "learning_rate": 7.73432550830354e-05, "loss": 0.0168, "step": 105010 }, { "epoch": 66.93435309114085, "grad_norm": 0.29916077852249146, "learning_rate": 7.733864051952987e-05, "loss": 0.0263, "step": 105020 }, { "epoch": 66.94072657743786, "grad_norm": 0.30439293384552, "learning_rate": 7.733402562383403e-05, "loss": 0.0115, "step": 105030 }, { "epoch": 66.94710006373487, "grad_norm": 0.3199256658554077, "learning_rate": 7.732941039600398e-05, "loss": 0.0187, "step": 105040 }, { "epoch": 66.95347355003187, "grad_norm": 0.8677883744239807, "learning_rate": 7.732479483609578e-05, "loss": 0.0318, "step": 105050 }, { "epoch": 66.95984703632887, "grad_norm": 0.7674688696861267, "learning_rate": 7.732017894416554e-05, "loss": 0.017, "step": 105060 }, { "epoch": 66.96622052262587, "grad_norm": 0.3315852880477905, "learning_rate": 7.731556272026932e-05, "loss": 0.0234, "step": 105070 }, { "epoch": 66.97259400892288, "grad_norm": 0.2697911858558655, "learning_rate": 7.731094616446323e-05, "loss": 0.0116, "step": 105080 }, { "epoch": 66.97896749521989, "grad_norm": 0.46022412180900574, "learning_rate": 7.730632927680334e-05, "loss": 0.0226, "step": 105090 }, { "epoch": 66.98534098151688, "grad_norm": 0.2678773105144501, "learning_rate": 7.730171205734578e-05, "loss": 0.0132, "step": 105100 }, { "epoch": 66.99171446781389, "grad_norm": 0.3796941936016083, "learning_rate": 7.729709450614666e-05, "loss": 0.0236, "step": 105110 }, { "epoch": 66.9980879541109, "grad_norm": 0.34052908420562744, "learning_rate": 7.729247662326205e-05, "loss": 0.0213, "step": 105120 }, { "epoch": 67.00446144040791, "grad_norm": 0.5408557057380676, "learning_rate": 7.728785840874808e-05, "loss": 0.0156, "step": 105130 }, { "epoch": 67.0108349267049, "grad_norm": 0.2174934446811676, "learning_rate": 7.728323986266086e-05, "loss": 0.018, "step": 105140 }, { "epoch": 67.01720841300191, "grad_norm": 0.40602123737335205, "learning_rate": 7.727862098505654e-05, "loss": 0.0144, "step": 105150 }, { "epoch": 67.02358189929892, "grad_norm": 0.37545037269592285, "learning_rate": 7.727400177599118e-05, "loss": 0.0215, "step": 105160 }, { "epoch": 67.02995538559593, "grad_norm": 0.533374547958374, "learning_rate": 7.726938223552098e-05, "loss": 0.033, "step": 105170 }, { "epoch": 67.03632887189292, "grad_norm": 0.4033690094947815, "learning_rate": 7.726476236370203e-05, "loss": 0.0146, "step": 105180 }, { "epoch": 67.04270235818993, "grad_norm": 0.3791067898273468, "learning_rate": 7.726014216059048e-05, "loss": 0.0163, "step": 105190 }, { "epoch": 67.04907584448694, "grad_norm": 0.5514476895332336, "learning_rate": 7.725552162624248e-05, "loss": 0.0161, "step": 105200 }, { "epoch": 67.05544933078394, "grad_norm": 0.43224847316741943, "learning_rate": 7.725090076071413e-05, "loss": 0.0245, "step": 105210 }, { "epoch": 67.06182281708094, "grad_norm": 0.2212410718202591, "learning_rate": 7.724627956406162e-05, "loss": 0.0129, "step": 105220 }, { "epoch": 67.06819630337795, "grad_norm": 0.19712330400943756, "learning_rate": 7.724165803634108e-05, "loss": 0.0141, "step": 105230 }, { "epoch": 67.07456978967495, "grad_norm": 0.39485466480255127, "learning_rate": 7.723703617760869e-05, "loss": 0.0274, "step": 105240 }, { "epoch": 67.08094327597196, "grad_norm": 0.21389883756637573, "learning_rate": 7.723241398792057e-05, "loss": 0.0133, "step": 105250 }, { "epoch": 67.08731676226896, "grad_norm": 0.4741755425930023, "learning_rate": 7.722779146733293e-05, "loss": 0.0207, "step": 105260 }, { "epoch": 67.09369024856596, "grad_norm": 0.41173914074897766, "learning_rate": 7.722316861590191e-05, "loss": 0.0188, "step": 105270 }, { "epoch": 67.10006373486297, "grad_norm": 0.2696644365787506, "learning_rate": 7.721854543368367e-05, "loss": 0.0113, "step": 105280 }, { "epoch": 67.10643722115998, "grad_norm": 0.30941760540008545, "learning_rate": 7.721392192073444e-05, "loss": 0.0109, "step": 105290 }, { "epoch": 67.11281070745697, "grad_norm": 0.21298426389694214, "learning_rate": 7.720929807711034e-05, "loss": 0.0185, "step": 105300 }, { "epoch": 67.11918419375398, "grad_norm": 0.7119069695472717, "learning_rate": 7.720467390286758e-05, "loss": 0.0263, "step": 105310 }, { "epoch": 67.12555768005099, "grad_norm": 0.6277884840965271, "learning_rate": 7.720004939806236e-05, "loss": 0.0153, "step": 105320 }, { "epoch": 67.131931166348, "grad_norm": 0.6795727610588074, "learning_rate": 7.719542456275084e-05, "loss": 0.0194, "step": 105330 }, { "epoch": 67.13830465264499, "grad_norm": 0.2808575928211212, "learning_rate": 7.719079939698925e-05, "loss": 0.017, "step": 105340 }, { "epoch": 67.144678138942, "grad_norm": 0.7953891158103943, "learning_rate": 7.718617390083376e-05, "loss": 0.0157, "step": 105350 }, { "epoch": 67.151051625239, "grad_norm": 0.2603432834148407, "learning_rate": 7.718154807434061e-05, "loss": 0.0196, "step": 105360 }, { "epoch": 67.15742511153601, "grad_norm": 0.5800109505653381, "learning_rate": 7.717692191756596e-05, "loss": 0.0231, "step": 105370 }, { "epoch": 67.16379859783301, "grad_norm": 0.36354097723960876, "learning_rate": 7.717229543056608e-05, "loss": 0.0169, "step": 105380 }, { "epoch": 67.17017208413002, "grad_norm": 0.3106566071510315, "learning_rate": 7.716766861339714e-05, "loss": 0.0164, "step": 105390 }, { "epoch": 67.17654557042702, "grad_norm": 0.5032699704170227, "learning_rate": 7.716304146611536e-05, "loss": 0.0243, "step": 105400 }, { "epoch": 67.18291905672403, "grad_norm": 0.4553208649158478, "learning_rate": 7.715841398877701e-05, "loss": 0.022, "step": 105410 }, { "epoch": 67.18929254302103, "grad_norm": 0.4614121615886688, "learning_rate": 7.715378618143827e-05, "loss": 0.0189, "step": 105420 }, { "epoch": 67.19566602931803, "grad_norm": 0.4059658944606781, "learning_rate": 7.714915804415542e-05, "loss": 0.0234, "step": 105430 }, { "epoch": 67.20203951561504, "grad_norm": 0.33004340529441833, "learning_rate": 7.714452957698462e-05, "loss": 0.0176, "step": 105440 }, { "epoch": 67.20841300191205, "grad_norm": 0.6053465604782104, "learning_rate": 7.713990077998219e-05, "loss": 0.0266, "step": 105450 }, { "epoch": 67.21478648820906, "grad_norm": 0.3490292429924011, "learning_rate": 7.713527165320432e-05, "loss": 0.0148, "step": 105460 }, { "epoch": 67.22115997450605, "grad_norm": 0.3693540394306183, "learning_rate": 7.713064219670729e-05, "loss": 0.0188, "step": 105470 }, { "epoch": 67.22753346080306, "grad_norm": 0.27278390526771545, "learning_rate": 7.712601241054734e-05, "loss": 0.0218, "step": 105480 }, { "epoch": 67.23390694710007, "grad_norm": 0.4609919786453247, "learning_rate": 7.712138229478075e-05, "loss": 0.0199, "step": 105490 }, { "epoch": 67.24028043339707, "grad_norm": 0.4214707911014557, "learning_rate": 7.711675184946373e-05, "loss": 0.0182, "step": 105500 }, { "epoch": 67.24665391969407, "grad_norm": 0.44127461314201355, "learning_rate": 7.711212107465257e-05, "loss": 0.0315, "step": 105510 }, { "epoch": 67.25302740599108, "grad_norm": 0.6578241586685181, "learning_rate": 7.710748997040356e-05, "loss": 0.0129, "step": 105520 }, { "epoch": 67.25940089228808, "grad_norm": 0.2958971858024597, "learning_rate": 7.710285853677295e-05, "loss": 0.0183, "step": 105530 }, { "epoch": 67.26577437858509, "grad_norm": 0.4070180356502533, "learning_rate": 7.709822677381702e-05, "loss": 0.0234, "step": 105540 }, { "epoch": 67.27214786488209, "grad_norm": 0.3588690161705017, "learning_rate": 7.709359468159203e-05, "loss": 0.0222, "step": 105550 }, { "epoch": 67.2785213511791, "grad_norm": 0.32245951890945435, "learning_rate": 7.70889622601543e-05, "loss": 0.011, "step": 105560 }, { "epoch": 67.2848948374761, "grad_norm": 0.2726690471172333, "learning_rate": 7.708432950956011e-05, "loss": 0.0109, "step": 105570 }, { "epoch": 67.29126832377311, "grad_norm": 0.25237929821014404, "learning_rate": 7.707969642986572e-05, "loss": 0.015, "step": 105580 }, { "epoch": 67.2976418100701, "grad_norm": 0.23396292328834534, "learning_rate": 7.707506302112746e-05, "loss": 0.0145, "step": 105590 }, { "epoch": 67.30401529636711, "grad_norm": 0.49927106499671936, "learning_rate": 7.707042928340164e-05, "loss": 0.0201, "step": 105600 }, { "epoch": 67.31038878266412, "grad_norm": 0.3887627124786377, "learning_rate": 7.706579521674453e-05, "loss": 0.0164, "step": 105610 }, { "epoch": 67.31676226896113, "grad_norm": 0.3990075886249542, "learning_rate": 7.706116082121245e-05, "loss": 0.0168, "step": 105620 }, { "epoch": 67.32313575525812, "grad_norm": 0.5167376399040222, "learning_rate": 7.705652609686173e-05, "loss": 0.0245, "step": 105630 }, { "epoch": 67.32950924155513, "grad_norm": 0.6399632096290588, "learning_rate": 7.705189104374866e-05, "loss": 0.0161, "step": 105640 }, { "epoch": 67.33588272785214, "grad_norm": 0.5228751301765442, "learning_rate": 7.704725566192958e-05, "loss": 0.0181, "step": 105650 }, { "epoch": 67.34225621414915, "grad_norm": 0.40531012415885925, "learning_rate": 7.70426199514608e-05, "loss": 0.0188, "step": 105660 }, { "epoch": 67.34862970044614, "grad_norm": 0.2802581191062927, "learning_rate": 7.703798391239866e-05, "loss": 0.0185, "step": 105670 }, { "epoch": 67.35500318674315, "grad_norm": 0.2423659712076187, "learning_rate": 7.703334754479949e-05, "loss": 0.0206, "step": 105680 }, { "epoch": 67.36137667304016, "grad_norm": 0.28774094581604004, "learning_rate": 7.70287108487196e-05, "loss": 0.0143, "step": 105690 }, { "epoch": 67.36775015933716, "grad_norm": 0.3289378881454468, "learning_rate": 7.702407382421539e-05, "loss": 0.0139, "step": 105700 }, { "epoch": 67.37412364563416, "grad_norm": 0.48083844780921936, "learning_rate": 7.701943647134315e-05, "loss": 0.0175, "step": 105710 }, { "epoch": 67.38049713193116, "grad_norm": 0.3376317620277405, "learning_rate": 7.701479879015925e-05, "loss": 0.0186, "step": 105720 }, { "epoch": 67.38687061822817, "grad_norm": 0.32823094725608826, "learning_rate": 7.701016078072005e-05, "loss": 0.0235, "step": 105730 }, { "epoch": 67.39324410452518, "grad_norm": 0.27656641602516174, "learning_rate": 7.700552244308188e-05, "loss": 0.015, "step": 105740 }, { "epoch": 67.39961759082217, "grad_norm": 0.23919177055358887, "learning_rate": 7.700088377730111e-05, "loss": 0.0099, "step": 105750 }, { "epoch": 67.40599107711918, "grad_norm": 0.29493051767349243, "learning_rate": 7.699624478343413e-05, "loss": 0.0167, "step": 105760 }, { "epoch": 67.41236456341619, "grad_norm": 0.32968592643737793, "learning_rate": 7.699160546153727e-05, "loss": 0.0254, "step": 105770 }, { "epoch": 67.4187380497132, "grad_norm": 0.5708123445510864, "learning_rate": 7.698696581166691e-05, "loss": 0.0184, "step": 105780 }, { "epoch": 67.42511153601019, "grad_norm": 0.7756767868995667, "learning_rate": 7.698232583387945e-05, "loss": 0.0206, "step": 105790 }, { "epoch": 67.4314850223072, "grad_norm": 0.44351449608802795, "learning_rate": 7.697768552823126e-05, "loss": 0.0275, "step": 105800 }, { "epoch": 67.43785850860421, "grad_norm": 0.2911220192909241, "learning_rate": 7.69730448947787e-05, "loss": 0.0174, "step": 105810 }, { "epoch": 67.44423199490122, "grad_norm": 0.38815680146217346, "learning_rate": 7.69684039335782e-05, "loss": 0.0162, "step": 105820 }, { "epoch": 67.45060548119821, "grad_norm": 0.33517536520957947, "learning_rate": 7.696376264468612e-05, "loss": 0.0242, "step": 105830 }, { "epoch": 67.45697896749522, "grad_norm": 0.5730772018432617, "learning_rate": 7.695912102815886e-05, "loss": 0.0216, "step": 105840 }, { "epoch": 67.46335245379223, "grad_norm": 0.3504476249217987, "learning_rate": 7.695447908405283e-05, "loss": 0.0141, "step": 105850 }, { "epoch": 67.46972594008923, "grad_norm": 0.23968160152435303, "learning_rate": 7.694983681242442e-05, "loss": 0.0269, "step": 105860 }, { "epoch": 67.47609942638623, "grad_norm": 0.4239415228366852, "learning_rate": 7.694519421333007e-05, "loss": 0.0118, "step": 105870 }, { "epoch": 67.48247291268324, "grad_norm": 0.24590325355529785, "learning_rate": 7.694055128682616e-05, "loss": 0.01, "step": 105880 }, { "epoch": 67.48884639898024, "grad_norm": 0.525265097618103, "learning_rate": 7.693590803296912e-05, "loss": 0.0266, "step": 105890 }, { "epoch": 67.49521988527725, "grad_norm": 0.2504498064517975, "learning_rate": 7.693126445181536e-05, "loss": 0.0164, "step": 105900 }, { "epoch": 67.50159337157424, "grad_norm": 0.20599915087223053, "learning_rate": 7.692662054342131e-05, "loss": 0.0138, "step": 105910 }, { "epoch": 67.50796685787125, "grad_norm": 0.8464505672454834, "learning_rate": 7.692197630784339e-05, "loss": 0.0178, "step": 105920 }, { "epoch": 67.51434034416826, "grad_norm": 0.29683056473731995, "learning_rate": 7.691733174513806e-05, "loss": 0.0156, "step": 105930 }, { "epoch": 67.52071383046527, "grad_norm": 0.45821741223335266, "learning_rate": 7.691268685536172e-05, "loss": 0.0204, "step": 105940 }, { "epoch": 67.52708731676226, "grad_norm": 0.3899914622306824, "learning_rate": 7.690804163857082e-05, "loss": 0.0177, "step": 105950 }, { "epoch": 67.53346080305927, "grad_norm": 0.39054110646247864, "learning_rate": 7.690339609482182e-05, "loss": 0.0133, "step": 105960 }, { "epoch": 67.53983428935628, "grad_norm": 0.32774627208709717, "learning_rate": 7.689875022417115e-05, "loss": 0.0182, "step": 105970 }, { "epoch": 67.54620777565329, "grad_norm": 0.4880833029747009, "learning_rate": 7.689410402667527e-05, "loss": 0.0126, "step": 105980 }, { "epoch": 67.55258126195028, "grad_norm": 0.3543986976146698, "learning_rate": 7.688945750239064e-05, "loss": 0.0185, "step": 105990 }, { "epoch": 67.55895474824729, "grad_norm": 0.6037893891334534, "learning_rate": 7.688481065137371e-05, "loss": 0.0233, "step": 106000 }, { "epoch": 67.5653282345443, "grad_norm": 0.34684044122695923, "learning_rate": 7.688016347368094e-05, "loss": 0.0172, "step": 106010 }, { "epoch": 67.5717017208413, "grad_norm": 0.41661015152931213, "learning_rate": 7.687551596936882e-05, "loss": 0.0197, "step": 106020 }, { "epoch": 67.5780752071383, "grad_norm": 0.2711317241191864, "learning_rate": 7.687086813849378e-05, "loss": 0.0128, "step": 106030 }, { "epoch": 67.5844486934353, "grad_norm": 0.602989137172699, "learning_rate": 7.686621998111234e-05, "loss": 0.0178, "step": 106040 }, { "epoch": 67.59082217973231, "grad_norm": 0.3907238245010376, "learning_rate": 7.686157149728096e-05, "loss": 0.0195, "step": 106050 }, { "epoch": 67.59719566602932, "grad_norm": 0.35256439447402954, "learning_rate": 7.685692268705614e-05, "loss": 0.0229, "step": 106060 }, { "epoch": 67.60356915232632, "grad_norm": 0.23378022015094757, "learning_rate": 7.685227355049434e-05, "loss": 0.0168, "step": 106070 }, { "epoch": 67.60994263862332, "grad_norm": 0.3681199848651886, "learning_rate": 7.684762408765205e-05, "loss": 0.0179, "step": 106080 }, { "epoch": 67.61631612492033, "grad_norm": 0.5050526261329651, "learning_rate": 7.684297429858579e-05, "loss": 0.0298, "step": 106090 }, { "epoch": 67.62268961121734, "grad_norm": 0.5170756578445435, "learning_rate": 7.683832418335204e-05, "loss": 0.0311, "step": 106100 }, { "epoch": 67.62906309751435, "grad_norm": 0.18287505209445953, "learning_rate": 7.683367374200732e-05, "loss": 0.0133, "step": 106110 }, { "epoch": 67.63543658381134, "grad_norm": 0.44249165058135986, "learning_rate": 7.682902297460812e-05, "loss": 0.0188, "step": 106120 }, { "epoch": 67.64181007010835, "grad_norm": 0.35246211290359497, "learning_rate": 7.682437188121097e-05, "loss": 0.0201, "step": 106130 }, { "epoch": 67.64818355640536, "grad_norm": 0.2629941999912262, "learning_rate": 7.681972046187238e-05, "loss": 0.0165, "step": 106140 }, { "epoch": 67.65455704270236, "grad_norm": 0.40199899673461914, "learning_rate": 7.681506871664886e-05, "loss": 0.0171, "step": 106150 }, { "epoch": 67.66093052899936, "grad_norm": 0.6481414437294006, "learning_rate": 7.681041664559693e-05, "loss": 0.033, "step": 106160 }, { "epoch": 67.66730401529637, "grad_norm": 0.23511670529842377, "learning_rate": 7.680576424877312e-05, "loss": 0.0162, "step": 106170 }, { "epoch": 67.67367750159337, "grad_norm": 0.34975284337997437, "learning_rate": 7.680111152623399e-05, "loss": 0.0142, "step": 106180 }, { "epoch": 67.68005098789038, "grad_norm": 0.55955970287323, "learning_rate": 7.679645847803603e-05, "loss": 0.0174, "step": 106190 }, { "epoch": 67.68642447418738, "grad_norm": 0.2405116707086563, "learning_rate": 7.67918051042358e-05, "loss": 0.0155, "step": 106200 }, { "epoch": 67.69279796048438, "grad_norm": 0.3025439381599426, "learning_rate": 7.678715140488985e-05, "loss": 0.0182, "step": 106210 }, { "epoch": 67.69917144678139, "grad_norm": 0.44550079107284546, "learning_rate": 7.678249738005472e-05, "loss": 0.0305, "step": 106220 }, { "epoch": 67.7055449330784, "grad_norm": 0.24607625603675842, "learning_rate": 7.677784302978697e-05, "loss": 0.0215, "step": 106230 }, { "epoch": 67.7119184193754, "grad_norm": 0.3581153452396393, "learning_rate": 7.67731883541431e-05, "loss": 0.0169, "step": 106240 }, { "epoch": 67.7182919056724, "grad_norm": 0.3534605801105499, "learning_rate": 7.676853335317977e-05, "loss": 0.0178, "step": 106250 }, { "epoch": 67.72466539196941, "grad_norm": 0.3339270055294037, "learning_rate": 7.676387802695345e-05, "loss": 0.0226, "step": 106260 }, { "epoch": 67.73103887826642, "grad_norm": 0.2533717453479767, "learning_rate": 7.675922237552075e-05, "loss": 0.0122, "step": 106270 }, { "epoch": 67.73741236456341, "grad_norm": 0.39450234174728394, "learning_rate": 7.675456639893823e-05, "loss": 0.0192, "step": 106280 }, { "epoch": 67.74378585086042, "grad_norm": 0.45683565735816956, "learning_rate": 7.674991009726249e-05, "loss": 0.0238, "step": 106290 }, { "epoch": 67.75015933715743, "grad_norm": 0.31965792179107666, "learning_rate": 7.674525347055006e-05, "loss": 0.0215, "step": 106300 }, { "epoch": 67.75653282345444, "grad_norm": 0.35750317573547363, "learning_rate": 7.674059651885757e-05, "loss": 0.0178, "step": 106310 }, { "epoch": 67.76290630975143, "grad_norm": 0.8767184615135193, "learning_rate": 7.673593924224158e-05, "loss": 0.0216, "step": 106320 }, { "epoch": 67.76927979604844, "grad_norm": 0.4145925045013428, "learning_rate": 7.673128164075867e-05, "loss": 0.039, "step": 106330 }, { "epoch": 67.77565328234544, "grad_norm": 0.19461078941822052, "learning_rate": 7.672662371446546e-05, "loss": 0.0192, "step": 106340 }, { "epoch": 67.78202676864245, "grad_norm": 0.35224634408950806, "learning_rate": 7.672196546341853e-05, "loss": 0.0137, "step": 106350 }, { "epoch": 67.78840025493945, "grad_norm": 0.38458341360092163, "learning_rate": 7.671730688767451e-05, "loss": 0.0165, "step": 106360 }, { "epoch": 67.79477374123645, "grad_norm": 0.5220276117324829, "learning_rate": 7.671264798728997e-05, "loss": 0.0221, "step": 106370 }, { "epoch": 67.80114722753346, "grad_norm": 0.5423200726509094, "learning_rate": 7.670798876232154e-05, "loss": 0.0224, "step": 106380 }, { "epoch": 67.80752071383047, "grad_norm": 0.3010280728340149, "learning_rate": 7.670332921282582e-05, "loss": 0.0173, "step": 106390 }, { "epoch": 67.81389420012746, "grad_norm": 0.46610260009765625, "learning_rate": 7.669866933885945e-05, "loss": 0.0142, "step": 106400 }, { "epoch": 67.82026768642447, "grad_norm": 0.4742877185344696, "learning_rate": 7.669400914047904e-05, "loss": 0.0138, "step": 106410 }, { "epoch": 67.82664117272148, "grad_norm": 0.28387489914894104, "learning_rate": 7.668934861774121e-05, "loss": 0.0342, "step": 106420 }, { "epoch": 67.83301465901849, "grad_norm": 0.4412648379802704, "learning_rate": 7.668468777070259e-05, "loss": 0.0219, "step": 106430 }, { "epoch": 67.83938814531548, "grad_norm": 0.7620427012443542, "learning_rate": 7.668002659941983e-05, "loss": 0.0224, "step": 106440 }, { "epoch": 67.84576163161249, "grad_norm": 0.43268439173698425, "learning_rate": 7.667536510394956e-05, "loss": 0.0157, "step": 106450 }, { "epoch": 67.8521351179095, "grad_norm": 0.4897032678127289, "learning_rate": 7.66707032843484e-05, "loss": 0.0234, "step": 106460 }, { "epoch": 67.8585086042065, "grad_norm": 0.7679625749588013, "learning_rate": 7.666604114067304e-05, "loss": 0.0152, "step": 106470 }, { "epoch": 67.8648820905035, "grad_norm": 0.35588929057121277, "learning_rate": 7.666137867298009e-05, "loss": 0.0182, "step": 106480 }, { "epoch": 67.87125557680051, "grad_norm": 0.7372584939002991, "learning_rate": 7.66567158813262e-05, "loss": 0.0282, "step": 106490 }, { "epoch": 67.87762906309752, "grad_norm": 0.4810059070587158, "learning_rate": 7.665205276576807e-05, "loss": 0.0164, "step": 106500 }, { "epoch": 67.88400254939452, "grad_norm": 0.5692694187164307, "learning_rate": 7.664738932636231e-05, "loss": 0.0143, "step": 106510 }, { "epoch": 67.89037603569152, "grad_norm": 0.278939813375473, "learning_rate": 7.664272556316564e-05, "loss": 0.0153, "step": 106520 }, { "epoch": 67.89674952198853, "grad_norm": 0.3903173804283142, "learning_rate": 7.663806147623467e-05, "loss": 0.0225, "step": 106530 }, { "epoch": 67.90312300828553, "grad_norm": 0.35728293657302856, "learning_rate": 7.663339706562612e-05, "loss": 0.0186, "step": 106540 }, { "epoch": 67.90949649458254, "grad_norm": 0.3167421817779541, "learning_rate": 7.662873233139664e-05, "loss": 0.0302, "step": 106550 }, { "epoch": 67.91586998087953, "grad_norm": 0.554246187210083, "learning_rate": 7.662406727360294e-05, "loss": 0.0266, "step": 106560 }, { "epoch": 67.92224346717654, "grad_norm": 0.5290642976760864, "learning_rate": 7.661940189230166e-05, "loss": 0.0132, "step": 106570 }, { "epoch": 67.92861695347355, "grad_norm": 0.4577910006046295, "learning_rate": 7.661473618754953e-05, "loss": 0.0331, "step": 106580 }, { "epoch": 67.93499043977056, "grad_norm": 0.518340528011322, "learning_rate": 7.661007015940323e-05, "loss": 0.0378, "step": 106590 }, { "epoch": 67.94136392606755, "grad_norm": 0.5243750810623169, "learning_rate": 7.660540380791942e-05, "loss": 0.016, "step": 106600 }, { "epoch": 67.94773741236456, "grad_norm": 0.5740789771080017, "learning_rate": 7.660073713315488e-05, "loss": 0.027, "step": 106610 }, { "epoch": 67.95411089866157, "grad_norm": 0.6874484419822693, "learning_rate": 7.659607013516624e-05, "loss": 0.023, "step": 106620 }, { "epoch": 67.96048438495858, "grad_norm": 0.6075157523155212, "learning_rate": 7.659140281401026e-05, "loss": 0.0208, "step": 106630 }, { "epoch": 67.96685787125557, "grad_norm": 0.44594675302505493, "learning_rate": 7.658673516974362e-05, "loss": 0.0186, "step": 106640 }, { "epoch": 67.97323135755258, "grad_norm": 0.39324626326560974, "learning_rate": 7.658206720242302e-05, "loss": 0.0151, "step": 106650 }, { "epoch": 67.97960484384959, "grad_norm": 0.6100729703903198, "learning_rate": 7.657739891210523e-05, "loss": 0.0249, "step": 106660 }, { "epoch": 67.9859783301466, "grad_norm": 0.3282598853111267, "learning_rate": 7.657273029884695e-05, "loss": 0.0146, "step": 106670 }, { "epoch": 67.99235181644359, "grad_norm": 0.337340772151947, "learning_rate": 7.656806136270492e-05, "loss": 0.0188, "step": 106680 }, { "epoch": 67.9987253027406, "grad_norm": 0.4167628884315491, "learning_rate": 7.656339210373584e-05, "loss": 0.0231, "step": 106690 }, { "epoch": 68.0050987890376, "grad_norm": 0.29945069551467896, "learning_rate": 7.655872252199646e-05, "loss": 0.0131, "step": 106700 }, { "epoch": 68.01147227533461, "grad_norm": 0.44897034764289856, "learning_rate": 7.655405261754354e-05, "loss": 0.0164, "step": 106710 }, { "epoch": 68.01784576163162, "grad_norm": 0.5033071637153625, "learning_rate": 7.654938239043381e-05, "loss": 0.0119, "step": 106720 }, { "epoch": 68.02421924792861, "grad_norm": 0.3997918367385864, "learning_rate": 7.654471184072401e-05, "loss": 0.0153, "step": 106730 }, { "epoch": 68.03059273422562, "grad_norm": 0.38981032371520996, "learning_rate": 7.65400409684709e-05, "loss": 0.0168, "step": 106740 }, { "epoch": 68.03696622052263, "grad_norm": 0.3457668125629425, "learning_rate": 7.653536977373123e-05, "loss": 0.047, "step": 106750 }, { "epoch": 68.04333970681964, "grad_norm": 0.4665244221687317, "learning_rate": 7.653069825656178e-05, "loss": 0.0136, "step": 106760 }, { "epoch": 68.04971319311663, "grad_norm": 0.2908143997192383, "learning_rate": 7.65260264170193e-05, "loss": 0.0145, "step": 106770 }, { "epoch": 68.05608667941364, "grad_norm": 0.48534539341926575, "learning_rate": 7.652135425516053e-05, "loss": 0.0164, "step": 106780 }, { "epoch": 68.06246016571065, "grad_norm": 0.5216912031173706, "learning_rate": 7.651668177104227e-05, "loss": 0.0166, "step": 106790 }, { "epoch": 68.06883365200765, "grad_norm": 0.44820600748062134, "learning_rate": 7.65120089647213e-05, "loss": 0.0192, "step": 106800 }, { "epoch": 68.07520713830465, "grad_norm": 0.4308840334415436, "learning_rate": 7.650733583625439e-05, "loss": 0.0154, "step": 106810 }, { "epoch": 68.08158062460166, "grad_norm": 0.6498591303825378, "learning_rate": 7.650266238569831e-05, "loss": 0.0294, "step": 106820 }, { "epoch": 68.08795411089866, "grad_norm": 0.7375837564468384, "learning_rate": 7.649798861310988e-05, "loss": 0.0161, "step": 106830 }, { "epoch": 68.09432759719567, "grad_norm": 0.36340686678886414, "learning_rate": 7.649331451854587e-05, "loss": 0.0187, "step": 106840 }, { "epoch": 68.10070108349267, "grad_norm": 0.39983057975769043, "learning_rate": 7.648864010206305e-05, "loss": 0.0205, "step": 106850 }, { "epoch": 68.10707456978967, "grad_norm": 0.33313754200935364, "learning_rate": 7.648396536371826e-05, "loss": 0.032, "step": 106860 }, { "epoch": 68.11344805608668, "grad_norm": 0.4528173804283142, "learning_rate": 7.647929030356829e-05, "loss": 0.0178, "step": 106870 }, { "epoch": 68.11982154238369, "grad_norm": 0.5261549949645996, "learning_rate": 7.647461492166993e-05, "loss": 0.0302, "step": 106880 }, { "epoch": 68.12619502868068, "grad_norm": 0.8389148712158203, "learning_rate": 7.646993921808001e-05, "loss": 0.0223, "step": 106890 }, { "epoch": 68.13256851497769, "grad_norm": 0.30468299984931946, "learning_rate": 7.646526319285533e-05, "loss": 0.0258, "step": 106900 }, { "epoch": 68.1389420012747, "grad_norm": 0.3896545469760895, "learning_rate": 7.646058684605273e-05, "loss": 0.0136, "step": 106910 }, { "epoch": 68.14531548757171, "grad_norm": 0.47277501225471497, "learning_rate": 7.6455910177729e-05, "loss": 0.023, "step": 106920 }, { "epoch": 68.1516889738687, "grad_norm": 0.30315983295440674, "learning_rate": 7.645123318794098e-05, "loss": 0.0126, "step": 106930 }, { "epoch": 68.15806246016571, "grad_norm": 0.2271626889705658, "learning_rate": 7.644655587674552e-05, "loss": 0.0191, "step": 106940 }, { "epoch": 68.16443594646272, "grad_norm": 0.21386992931365967, "learning_rate": 7.644187824419943e-05, "loss": 0.0265, "step": 106950 }, { "epoch": 68.17080943275973, "grad_norm": 0.7005563974380493, "learning_rate": 7.643720029035956e-05, "loss": 0.0203, "step": 106960 }, { "epoch": 68.17718291905672, "grad_norm": 0.3439531922340393, "learning_rate": 7.643252201528273e-05, "loss": 0.0155, "step": 106970 }, { "epoch": 68.18355640535373, "grad_norm": 0.5040580630302429, "learning_rate": 7.64278434190258e-05, "loss": 0.0208, "step": 106980 }, { "epoch": 68.18992989165073, "grad_norm": 0.6920354962348938, "learning_rate": 7.642316450164563e-05, "loss": 0.0168, "step": 106990 }, { "epoch": 68.19630337794774, "grad_norm": 0.30823206901550293, "learning_rate": 7.641848526319908e-05, "loss": 0.019, "step": 107000 }, { "epoch": 68.20267686424474, "grad_norm": 0.43885335326194763, "learning_rate": 7.641380570374296e-05, "loss": 0.0206, "step": 107010 }, { "epoch": 68.20905035054174, "grad_norm": 0.20622175931930542, "learning_rate": 7.640912582333418e-05, "loss": 0.0213, "step": 107020 }, { "epoch": 68.21542383683875, "grad_norm": 0.4821596145629883, "learning_rate": 7.640444562202959e-05, "loss": 0.0187, "step": 107030 }, { "epoch": 68.22179732313576, "grad_norm": 0.17340785264968872, "learning_rate": 7.639976509988605e-05, "loss": 0.0259, "step": 107040 }, { "epoch": 68.22817080943275, "grad_norm": 0.3724033236503601, "learning_rate": 7.639508425696042e-05, "loss": 0.0159, "step": 107050 }, { "epoch": 68.23454429572976, "grad_norm": 0.5715773701667786, "learning_rate": 7.639040309330963e-05, "loss": 0.0191, "step": 107060 }, { "epoch": 68.24091778202677, "grad_norm": 0.4976310133934021, "learning_rate": 7.63857216089905e-05, "loss": 0.0236, "step": 107070 }, { "epoch": 68.24729126832378, "grad_norm": 0.4499194920063019, "learning_rate": 7.638103980405995e-05, "loss": 0.0133, "step": 107080 }, { "epoch": 68.25366475462077, "grad_norm": 0.23533770442008972, "learning_rate": 7.637635767857486e-05, "loss": 0.0126, "step": 107090 }, { "epoch": 68.26003824091778, "grad_norm": 0.2602512538433075, "learning_rate": 7.637167523259211e-05, "loss": 0.0213, "step": 107100 }, { "epoch": 68.26641172721479, "grad_norm": 0.3403356075286865, "learning_rate": 7.636699246616862e-05, "loss": 0.0149, "step": 107110 }, { "epoch": 68.2727852135118, "grad_norm": 0.36458438634872437, "learning_rate": 7.636230937936128e-05, "loss": 0.0138, "step": 107120 }, { "epoch": 68.27915869980879, "grad_norm": 0.34950578212738037, "learning_rate": 7.635762597222698e-05, "loss": 0.0162, "step": 107130 }, { "epoch": 68.2855321861058, "grad_norm": 0.35765716433525085, "learning_rate": 7.635294224482263e-05, "loss": 0.0221, "step": 107140 }, { "epoch": 68.2919056724028, "grad_norm": 0.34105348587036133, "learning_rate": 7.634825819720517e-05, "loss": 0.0137, "step": 107150 }, { "epoch": 68.29827915869981, "grad_norm": 0.40952396392822266, "learning_rate": 7.634357382943149e-05, "loss": 0.0267, "step": 107160 }, { "epoch": 68.3046526449968, "grad_norm": 0.38946276903152466, "learning_rate": 7.63388891415585e-05, "loss": 0.0179, "step": 107170 }, { "epoch": 68.31102613129381, "grad_norm": 0.3432058095932007, "learning_rate": 7.633420413364317e-05, "loss": 0.0105, "step": 107180 }, { "epoch": 68.31739961759082, "grad_norm": 0.3013158142566681, "learning_rate": 7.632951880574237e-05, "loss": 0.0142, "step": 107190 }, { "epoch": 68.32377310388783, "grad_norm": 0.783240020275116, "learning_rate": 7.632483315791306e-05, "loss": 0.0302, "step": 107200 }, { "epoch": 68.33014659018482, "grad_norm": 0.3759140074253082, "learning_rate": 7.632014719021217e-05, "loss": 0.0152, "step": 107210 }, { "epoch": 68.33652007648183, "grad_norm": 0.5865796208381653, "learning_rate": 7.631546090269665e-05, "loss": 0.0271, "step": 107220 }, { "epoch": 68.34289356277884, "grad_norm": 0.3275274634361267, "learning_rate": 7.631077429542344e-05, "loss": 0.0258, "step": 107230 }, { "epoch": 68.34926704907585, "grad_norm": 0.2704906761646271, "learning_rate": 7.630608736844947e-05, "loss": 0.0224, "step": 107240 }, { "epoch": 68.35564053537284, "grad_norm": 0.3460225760936737, "learning_rate": 7.63014001218317e-05, "loss": 0.0171, "step": 107250 }, { "epoch": 68.36201402166985, "grad_norm": 0.39124923944473267, "learning_rate": 7.62967125556271e-05, "loss": 0.0124, "step": 107260 }, { "epoch": 68.36838750796686, "grad_norm": 0.3099464774131775, "learning_rate": 7.62920246698926e-05, "loss": 0.019, "step": 107270 }, { "epoch": 68.37476099426387, "grad_norm": 0.317867249250412, "learning_rate": 7.628733646468517e-05, "loss": 0.0173, "step": 107280 }, { "epoch": 68.38113448056086, "grad_norm": 0.2427472174167633, "learning_rate": 7.628264794006182e-05, "loss": 0.0151, "step": 107290 }, { "epoch": 68.38750796685787, "grad_norm": 0.4088838994503021, "learning_rate": 7.627795909607944e-05, "loss": 0.0209, "step": 107300 }, { "epoch": 68.39388145315488, "grad_norm": 0.5233646631240845, "learning_rate": 7.627326993279506e-05, "loss": 0.0253, "step": 107310 }, { "epoch": 68.40025493945188, "grad_norm": 0.19374558329582214, "learning_rate": 7.626858045026566e-05, "loss": 0.02, "step": 107320 }, { "epoch": 68.40662842574889, "grad_norm": 0.2931341230869293, "learning_rate": 7.626389064854818e-05, "loss": 0.0232, "step": 107330 }, { "epoch": 68.41300191204589, "grad_norm": 0.7147870659828186, "learning_rate": 7.625920052769966e-05, "loss": 0.0178, "step": 107340 }, { "epoch": 68.4193753983429, "grad_norm": 0.3726840913295746, "learning_rate": 7.625451008777705e-05, "loss": 0.0206, "step": 107350 }, { "epoch": 68.4257488846399, "grad_norm": 0.7866557240486145, "learning_rate": 7.624981932883734e-05, "loss": 0.0354, "step": 107360 }, { "epoch": 68.43212237093691, "grad_norm": 0.5845111608505249, "learning_rate": 7.624512825093755e-05, "loss": 0.0217, "step": 107370 }, { "epoch": 68.4384958572339, "grad_norm": 0.31185394525527954, "learning_rate": 7.624043685413467e-05, "loss": 0.0202, "step": 107380 }, { "epoch": 68.44486934353091, "grad_norm": 0.6676550507545471, "learning_rate": 7.623574513848572e-05, "loss": 0.0176, "step": 107390 }, { "epoch": 68.45124282982792, "grad_norm": 0.333631694316864, "learning_rate": 7.623105310404768e-05, "loss": 0.0189, "step": 107400 }, { "epoch": 68.45761631612493, "grad_norm": 0.5050011277198792, "learning_rate": 7.622636075087759e-05, "loss": 0.022, "step": 107410 }, { "epoch": 68.46398980242192, "grad_norm": 0.6961150169372559, "learning_rate": 7.622166807903243e-05, "loss": 0.0209, "step": 107420 }, { "epoch": 68.47036328871893, "grad_norm": 1.3149696588516235, "learning_rate": 7.621697508856929e-05, "loss": 0.024, "step": 107430 }, { "epoch": 68.47673677501594, "grad_norm": 0.33559754490852356, "learning_rate": 7.621228177954513e-05, "loss": 0.0138, "step": 107440 }, { "epoch": 68.48311026131294, "grad_norm": 0.4553198218345642, "learning_rate": 7.620758815201698e-05, "loss": 0.0292, "step": 107450 }, { "epoch": 68.48948374760994, "grad_norm": 0.40704768896102905, "learning_rate": 7.620289420604192e-05, "loss": 0.0331, "step": 107460 }, { "epoch": 68.49585723390695, "grad_norm": 0.6535449028015137, "learning_rate": 7.619819994167692e-05, "loss": 0.019, "step": 107470 }, { "epoch": 68.50223072020395, "grad_norm": 0.5713701844215393, "learning_rate": 7.619350535897909e-05, "loss": 0.0222, "step": 107480 }, { "epoch": 68.50860420650096, "grad_norm": 0.6810839772224426, "learning_rate": 7.61888104580054e-05, "loss": 0.023, "step": 107490 }, { "epoch": 68.51497769279796, "grad_norm": 0.41756436228752136, "learning_rate": 7.618411523881297e-05, "loss": 0.0209, "step": 107500 }, { "epoch": 68.52135117909496, "grad_norm": 0.5397517681121826, "learning_rate": 7.617941970145881e-05, "loss": 0.0206, "step": 107510 }, { "epoch": 68.52772466539197, "grad_norm": 0.3397981524467468, "learning_rate": 7.617472384599999e-05, "loss": 0.0148, "step": 107520 }, { "epoch": 68.53409815168898, "grad_norm": 0.5578193068504333, "learning_rate": 7.617002767249355e-05, "loss": 0.0302, "step": 107530 }, { "epoch": 68.54047163798597, "grad_norm": 0.4449341595172882, "learning_rate": 7.616533118099657e-05, "loss": 0.0246, "step": 107540 }, { "epoch": 68.54684512428298, "grad_norm": 0.4021769165992737, "learning_rate": 7.61606343715661e-05, "loss": 0.0124, "step": 107550 }, { "epoch": 68.55321861057999, "grad_norm": 0.3518792688846588, "learning_rate": 7.615593724425925e-05, "loss": 0.0145, "step": 107560 }, { "epoch": 68.559592096877, "grad_norm": 0.2632792592048645, "learning_rate": 7.615123979913305e-05, "loss": 0.0151, "step": 107570 }, { "epoch": 68.56596558317399, "grad_norm": 0.3523036241531372, "learning_rate": 7.614654203624458e-05, "loss": 0.0128, "step": 107580 }, { "epoch": 68.572339069471, "grad_norm": 0.30193230509757996, "learning_rate": 7.614184395565096e-05, "loss": 0.01, "step": 107590 }, { "epoch": 68.578712555768, "grad_norm": 0.6037343144416809, "learning_rate": 7.613714555740924e-05, "loss": 0.0228, "step": 107600 }, { "epoch": 68.58508604206502, "grad_norm": 0.37733855843544006, "learning_rate": 7.613244684157654e-05, "loss": 0.021, "step": 107610 }, { "epoch": 68.59145952836201, "grad_norm": 0.6305876970291138, "learning_rate": 7.612774780820994e-05, "loss": 0.0224, "step": 107620 }, { "epoch": 68.59783301465902, "grad_norm": 0.47082987427711487, "learning_rate": 7.612304845736651e-05, "loss": 0.0177, "step": 107630 }, { "epoch": 68.60420650095602, "grad_norm": 0.25820809602737427, "learning_rate": 7.61183487891034e-05, "loss": 0.0226, "step": 107640 }, { "epoch": 68.61057998725303, "grad_norm": 0.21953780949115753, "learning_rate": 7.61136488034777e-05, "loss": 0.0208, "step": 107650 }, { "epoch": 68.61695347355003, "grad_norm": 0.4183875024318695, "learning_rate": 7.610894850054648e-05, "loss": 0.0114, "step": 107660 }, { "epoch": 68.62332695984703, "grad_norm": 0.5300595164299011, "learning_rate": 7.61042478803669e-05, "loss": 0.0242, "step": 107670 }, { "epoch": 68.62970044614404, "grad_norm": 0.409984827041626, "learning_rate": 7.609954694299607e-05, "loss": 0.0119, "step": 107680 }, { "epoch": 68.63607393244105, "grad_norm": 0.3815591633319855, "learning_rate": 7.60948456884911e-05, "loss": 0.0137, "step": 107690 }, { "epoch": 68.64244741873804, "grad_norm": 0.35697662830352783, "learning_rate": 7.60901441169091e-05, "loss": 0.0241, "step": 107700 }, { "epoch": 68.64882090503505, "grad_norm": 0.39731234312057495, "learning_rate": 7.608544222830724e-05, "loss": 0.0103, "step": 107710 }, { "epoch": 68.65519439133206, "grad_norm": 0.3648849427700043, "learning_rate": 7.608074002274262e-05, "loss": 0.0175, "step": 107720 }, { "epoch": 68.66156787762907, "grad_norm": 0.42588314414024353, "learning_rate": 7.607603750027237e-05, "loss": 0.0156, "step": 107730 }, { "epoch": 68.66794136392606, "grad_norm": 0.3402612507343292, "learning_rate": 7.607133466095365e-05, "loss": 0.0112, "step": 107740 }, { "epoch": 68.67431485022307, "grad_norm": 0.47347623109817505, "learning_rate": 7.606663150484362e-05, "loss": 0.0276, "step": 107750 }, { "epoch": 68.68068833652008, "grad_norm": 0.35613930225372314, "learning_rate": 7.606192803199938e-05, "loss": 0.0126, "step": 107760 }, { "epoch": 68.68706182281709, "grad_norm": 0.4113629162311554, "learning_rate": 7.605722424247812e-05, "loss": 0.0334, "step": 107770 }, { "epoch": 68.69343530911408, "grad_norm": 0.2720666825771332, "learning_rate": 7.605252013633699e-05, "loss": 0.0282, "step": 107780 }, { "epoch": 68.69980879541109, "grad_norm": 0.38510674238204956, "learning_rate": 7.604781571363313e-05, "loss": 0.0233, "step": 107790 }, { "epoch": 68.7061822817081, "grad_norm": 0.5955936908721924, "learning_rate": 7.604311097442372e-05, "loss": 0.0402, "step": 107800 }, { "epoch": 68.7125557680051, "grad_norm": 0.5176739692687988, "learning_rate": 7.603840591876592e-05, "loss": 0.0202, "step": 107810 }, { "epoch": 68.7189292543021, "grad_norm": 0.20203040540218353, "learning_rate": 7.60337005467169e-05, "loss": 0.0114, "step": 107820 }, { "epoch": 68.7253027405991, "grad_norm": 0.34093043208122253, "learning_rate": 7.602899485833384e-05, "loss": 0.0186, "step": 107830 }, { "epoch": 68.73167622689611, "grad_norm": 0.3843778371810913, "learning_rate": 7.602428885367395e-05, "loss": 0.022, "step": 107840 }, { "epoch": 68.73804971319312, "grad_norm": 0.4147146940231323, "learning_rate": 7.601958253279435e-05, "loss": 0.0179, "step": 107850 }, { "epoch": 68.74442319949011, "grad_norm": 0.44987133145332336, "learning_rate": 7.601487589575224e-05, "loss": 0.0188, "step": 107860 }, { "epoch": 68.75079668578712, "grad_norm": 0.1732083559036255, "learning_rate": 7.601016894260485e-05, "loss": 0.0137, "step": 107870 }, { "epoch": 68.75717017208413, "grad_norm": 0.5704591274261475, "learning_rate": 7.600546167340934e-05, "loss": 0.015, "step": 107880 }, { "epoch": 68.76354365838114, "grad_norm": 0.4458199143409729, "learning_rate": 7.600075408822291e-05, "loss": 0.0103, "step": 107890 }, { "epoch": 68.76991714467813, "grad_norm": 0.1648261696100235, "learning_rate": 7.599604618710279e-05, "loss": 0.0181, "step": 107900 }, { "epoch": 68.77629063097514, "grad_norm": 0.2667195498943329, "learning_rate": 7.599133797010616e-05, "loss": 0.0199, "step": 107910 }, { "epoch": 68.78266411727215, "grad_norm": 0.23429815471172333, "learning_rate": 7.598662943729021e-05, "loss": 0.014, "step": 107920 }, { "epoch": 68.78903760356916, "grad_norm": 0.262127548456192, "learning_rate": 7.598192058871221e-05, "loss": 0.0189, "step": 107930 }, { "epoch": 68.79541108986615, "grad_norm": 0.22058309614658356, "learning_rate": 7.597721142442932e-05, "loss": 0.0218, "step": 107940 }, { "epoch": 68.80178457616316, "grad_norm": 0.36248114705085754, "learning_rate": 7.597250194449878e-05, "loss": 0.0168, "step": 107950 }, { "epoch": 68.80815806246017, "grad_norm": 0.2068217694759369, "learning_rate": 7.596779214897784e-05, "loss": 0.0184, "step": 107960 }, { "epoch": 68.81453154875717, "grad_norm": 0.6522926688194275, "learning_rate": 7.59630820379237e-05, "loss": 0.0313, "step": 107970 }, { "epoch": 68.82090503505418, "grad_norm": 0.6307727098464966, "learning_rate": 7.59583716113936e-05, "loss": 0.0157, "step": 107980 }, { "epoch": 68.82727852135118, "grad_norm": 0.3160719573497772, "learning_rate": 7.595366086944478e-05, "loss": 0.0201, "step": 107990 }, { "epoch": 68.83365200764818, "grad_norm": 0.5050654411315918, "learning_rate": 7.594894981213447e-05, "loss": 0.0183, "step": 108000 }, { "epoch": 68.84002549394519, "grad_norm": 0.4164694845676422, "learning_rate": 7.594423843951992e-05, "loss": 0.0133, "step": 108010 }, { "epoch": 68.8463989802422, "grad_norm": 0.32257139682769775, "learning_rate": 7.593952675165837e-05, "loss": 0.0145, "step": 108020 }, { "epoch": 68.8527724665392, "grad_norm": 0.4874633550643921, "learning_rate": 7.593481474860709e-05, "loss": 0.0304, "step": 108030 }, { "epoch": 68.8591459528362, "grad_norm": 0.7851148843765259, "learning_rate": 7.593010243042333e-05, "loss": 0.0179, "step": 108040 }, { "epoch": 68.86551943913321, "grad_norm": 0.29480835795402527, "learning_rate": 7.592538979716434e-05, "loss": 0.0212, "step": 108050 }, { "epoch": 68.87189292543022, "grad_norm": 0.6148595809936523, "learning_rate": 7.592067684888737e-05, "loss": 0.0208, "step": 108060 }, { "epoch": 68.87826641172721, "grad_norm": 0.4410087764263153, "learning_rate": 7.591596358564971e-05, "loss": 0.0172, "step": 108070 }, { "epoch": 68.88463989802422, "grad_norm": 0.5094189047813416, "learning_rate": 7.591125000750864e-05, "loss": 0.014, "step": 108080 }, { "epoch": 68.89101338432123, "grad_norm": 0.4877133369445801, "learning_rate": 7.59065361145214e-05, "loss": 0.0246, "step": 108090 }, { "epoch": 68.89738687061823, "grad_norm": 0.5226429104804993, "learning_rate": 7.59018219067453e-05, "loss": 0.0185, "step": 108100 }, { "epoch": 68.90376035691523, "grad_norm": 0.3255905210971832, "learning_rate": 7.58971073842376e-05, "loss": 0.0124, "step": 108110 }, { "epoch": 68.91013384321224, "grad_norm": 0.3183947801589966, "learning_rate": 7.58923925470556e-05, "loss": 0.0176, "step": 108120 }, { "epoch": 68.91650732950924, "grad_norm": 0.4008593261241913, "learning_rate": 7.588767739525659e-05, "loss": 0.0165, "step": 108130 }, { "epoch": 68.92288081580625, "grad_norm": 0.9146546721458435, "learning_rate": 7.588296192889784e-05, "loss": 0.0239, "step": 108140 }, { "epoch": 68.92925430210325, "grad_norm": 0.4135487675666809, "learning_rate": 7.587824614803667e-05, "loss": 0.0156, "step": 108150 }, { "epoch": 68.93562778840025, "grad_norm": 0.3348453938961029, "learning_rate": 7.58735300527304e-05, "loss": 0.0105, "step": 108160 }, { "epoch": 68.94200127469726, "grad_norm": 0.2534012496471405, "learning_rate": 7.586881364303629e-05, "loss": 0.0132, "step": 108170 }, { "epoch": 68.94837476099427, "grad_norm": 0.28328579664230347, "learning_rate": 7.586409691901167e-05, "loss": 0.0158, "step": 108180 }, { "epoch": 68.95474824729126, "grad_norm": 0.45651426911354065, "learning_rate": 7.585937988071386e-05, "loss": 0.0212, "step": 108190 }, { "epoch": 68.96112173358827, "grad_norm": 0.43294382095336914, "learning_rate": 7.585466252820014e-05, "loss": 0.0117, "step": 108200 }, { "epoch": 68.96749521988528, "grad_norm": 0.31954821944236755, "learning_rate": 7.584994486152788e-05, "loss": 0.0146, "step": 108210 }, { "epoch": 68.97386870618229, "grad_norm": 0.7792256474494934, "learning_rate": 7.58452268807544e-05, "loss": 0.0169, "step": 108220 }, { "epoch": 68.98024219247928, "grad_norm": 0.3245939612388611, "learning_rate": 7.584050858593699e-05, "loss": 0.0245, "step": 108230 }, { "epoch": 68.98661567877629, "grad_norm": 0.3179319500923157, "learning_rate": 7.5835789977133e-05, "loss": 0.0183, "step": 108240 }, { "epoch": 68.9929891650733, "grad_norm": 0.5070083141326904, "learning_rate": 7.583107105439979e-05, "loss": 0.0175, "step": 108250 }, { "epoch": 68.9993626513703, "grad_norm": 0.47172659635543823, "learning_rate": 7.582635181779467e-05, "loss": 0.0156, "step": 108260 }, { "epoch": 69.0057361376673, "grad_norm": 0.4728893041610718, "learning_rate": 7.582163226737499e-05, "loss": 0.0183, "step": 108270 }, { "epoch": 69.0121096239643, "grad_norm": 0.7240013480186462, "learning_rate": 7.58169124031981e-05, "loss": 0.0297, "step": 108280 }, { "epoch": 69.01848311026131, "grad_norm": 0.5990631580352783, "learning_rate": 7.581219222532134e-05, "loss": 0.0227, "step": 108290 }, { "epoch": 69.02485659655832, "grad_norm": 0.49733075499534607, "learning_rate": 7.580747173380209e-05, "loss": 0.0138, "step": 108300 }, { "epoch": 69.03123008285532, "grad_norm": 0.5253685712814331, "learning_rate": 7.580275092869766e-05, "loss": 0.0278, "step": 108310 }, { "epoch": 69.03760356915232, "grad_norm": 0.26072439551353455, "learning_rate": 7.579802981006548e-05, "loss": 0.0121, "step": 108320 }, { "epoch": 69.04397705544933, "grad_norm": 0.38928183913230896, "learning_rate": 7.579330837796287e-05, "loss": 0.0133, "step": 108330 }, { "epoch": 69.05035054174634, "grad_norm": 0.3050217032432556, "learning_rate": 7.57885866324472e-05, "loss": 0.0218, "step": 108340 }, { "epoch": 69.05672402804333, "grad_norm": 0.4450750946998596, "learning_rate": 7.578386457357587e-05, "loss": 0.0128, "step": 108350 }, { "epoch": 69.06309751434034, "grad_norm": 0.6043259501457214, "learning_rate": 7.577914220140623e-05, "loss": 0.0171, "step": 108360 }, { "epoch": 69.06947100063735, "grad_norm": 0.17940351366996765, "learning_rate": 7.577441951599567e-05, "loss": 0.0161, "step": 108370 }, { "epoch": 69.07584448693436, "grad_norm": 0.4216276705265045, "learning_rate": 7.576969651740157e-05, "loss": 0.0303, "step": 108380 }, { "epoch": 69.08221797323135, "grad_norm": 0.5170851349830627, "learning_rate": 7.576497320568135e-05, "loss": 0.0216, "step": 108390 }, { "epoch": 69.08859145952836, "grad_norm": 0.6065660715103149, "learning_rate": 7.576024958089236e-05, "loss": 0.0216, "step": 108400 }, { "epoch": 69.09496494582537, "grad_norm": 0.36588621139526367, "learning_rate": 7.575552564309202e-05, "loss": 0.0265, "step": 108410 }, { "epoch": 69.10133843212238, "grad_norm": 0.800003170967102, "learning_rate": 7.575080139233774e-05, "loss": 0.0162, "step": 108420 }, { "epoch": 69.10771191841937, "grad_norm": 0.4325380027294159, "learning_rate": 7.57460768286869e-05, "loss": 0.0171, "step": 108430 }, { "epoch": 69.11408540471638, "grad_norm": 0.40484583377838135, "learning_rate": 7.574135195219694e-05, "loss": 0.0234, "step": 108440 }, { "epoch": 69.12045889101339, "grad_norm": 0.407765656709671, "learning_rate": 7.57366267629252e-05, "loss": 0.0215, "step": 108450 }, { "epoch": 69.1268323773104, "grad_norm": 0.2727501094341278, "learning_rate": 7.573190126092919e-05, "loss": 0.0227, "step": 108460 }, { "epoch": 69.13320586360739, "grad_norm": 0.8473920822143555, "learning_rate": 7.572717544626627e-05, "loss": 0.0173, "step": 108470 }, { "epoch": 69.1395793499044, "grad_norm": 0.2807987332344055, "learning_rate": 7.572244931899388e-05, "loss": 0.0114, "step": 108480 }, { "epoch": 69.1459528362014, "grad_norm": 0.38958507776260376, "learning_rate": 7.571772287916945e-05, "loss": 0.0258, "step": 108490 }, { "epoch": 69.15232632249841, "grad_norm": 0.3884463906288147, "learning_rate": 7.57129961268504e-05, "loss": 0.0124, "step": 108500 }, { "epoch": 69.1586998087954, "grad_norm": 0.5014525055885315, "learning_rate": 7.570826906209417e-05, "loss": 0.0175, "step": 108510 }, { "epoch": 69.16507329509241, "grad_norm": 0.2555745840072632, "learning_rate": 7.570354168495818e-05, "loss": 0.0176, "step": 108520 }, { "epoch": 69.17144678138942, "grad_norm": 0.4079521596431732, "learning_rate": 7.569881399549992e-05, "loss": 0.0223, "step": 108530 }, { "epoch": 69.17782026768643, "grad_norm": 0.48922199010849, "learning_rate": 7.569408599377678e-05, "loss": 0.0259, "step": 108540 }, { "epoch": 69.18419375398342, "grad_norm": 0.36028140783309937, "learning_rate": 7.568935767984624e-05, "loss": 0.0127, "step": 108550 }, { "epoch": 69.19056724028043, "grad_norm": 0.3056938946247101, "learning_rate": 7.568462905376577e-05, "loss": 0.0171, "step": 108560 }, { "epoch": 69.19694072657744, "grad_norm": 0.48632609844207764, "learning_rate": 7.567990011559278e-05, "loss": 0.0143, "step": 108570 }, { "epoch": 69.20331421287445, "grad_norm": 0.2394338846206665, "learning_rate": 7.567517086538478e-05, "loss": 0.0207, "step": 108580 }, { "epoch": 69.20968769917145, "grad_norm": 0.325132817029953, "learning_rate": 7.56704413031992e-05, "loss": 0.0191, "step": 108590 }, { "epoch": 69.21606118546845, "grad_norm": 0.5196437239646912, "learning_rate": 7.566571142909353e-05, "loss": 0.0325, "step": 108600 }, { "epoch": 69.22243467176546, "grad_norm": 0.5635237693786621, "learning_rate": 7.566098124312524e-05, "loss": 0.0212, "step": 108610 }, { "epoch": 69.22880815806246, "grad_norm": 0.2639165222644806, "learning_rate": 7.565625074535179e-05, "loss": 0.0235, "step": 108620 }, { "epoch": 69.23518164435947, "grad_norm": 0.3383626341819763, "learning_rate": 7.565151993583066e-05, "loss": 0.0266, "step": 108630 }, { "epoch": 69.24155513065647, "grad_norm": 0.4056667685508728, "learning_rate": 7.564678881461935e-05, "loss": 0.0169, "step": 108640 }, { "epoch": 69.24792861695347, "grad_norm": 0.2835894823074341, "learning_rate": 7.564205738177535e-05, "loss": 0.0233, "step": 108650 }, { "epoch": 69.25430210325048, "grad_norm": 0.2543014585971832, "learning_rate": 7.563732563735613e-05, "loss": 0.0147, "step": 108660 }, { "epoch": 69.26067558954749, "grad_norm": 0.2068164050579071, "learning_rate": 7.56325935814192e-05, "loss": 0.016, "step": 108670 }, { "epoch": 69.26704907584448, "grad_norm": 0.19365648925304413, "learning_rate": 7.562786121402208e-05, "loss": 0.0157, "step": 108680 }, { "epoch": 69.27342256214149, "grad_norm": 0.27341821789741516, "learning_rate": 7.562312853522222e-05, "loss": 0.0164, "step": 108690 }, { "epoch": 69.2797960484385, "grad_norm": 0.3066481351852417, "learning_rate": 7.561839554507717e-05, "loss": 0.0234, "step": 108700 }, { "epoch": 69.2861695347355, "grad_norm": 0.363720178604126, "learning_rate": 7.561366224364443e-05, "loss": 0.021, "step": 108710 }, { "epoch": 69.2925430210325, "grad_norm": 0.5038548111915588, "learning_rate": 7.56089286309815e-05, "loss": 0.0335, "step": 108720 }, { "epoch": 69.29891650732951, "grad_norm": 0.5308862328529358, "learning_rate": 7.56041947071459e-05, "loss": 0.0248, "step": 108730 }, { "epoch": 69.30528999362652, "grad_norm": 0.759466826915741, "learning_rate": 7.559946047219518e-05, "loss": 0.0298, "step": 108740 }, { "epoch": 69.31166347992352, "grad_norm": 0.7935208082199097, "learning_rate": 7.559472592618683e-05, "loss": 0.0258, "step": 108750 }, { "epoch": 69.31803696622052, "grad_norm": 0.26888805627822876, "learning_rate": 7.558999106917841e-05, "loss": 0.0127, "step": 108760 }, { "epoch": 69.32441045251753, "grad_norm": 0.4722280204296112, "learning_rate": 7.558525590122743e-05, "loss": 0.0262, "step": 108770 }, { "epoch": 69.33078393881453, "grad_norm": 0.7265624403953552, "learning_rate": 7.558052042239144e-05, "loss": 0.0133, "step": 108780 }, { "epoch": 69.33715742511154, "grad_norm": 0.26435384154319763, "learning_rate": 7.557578463272797e-05, "loss": 0.0194, "step": 108790 }, { "epoch": 69.34353091140854, "grad_norm": 0.5723888278007507, "learning_rate": 7.557104853229457e-05, "loss": 0.0175, "step": 108800 }, { "epoch": 69.34990439770554, "grad_norm": 0.3668348789215088, "learning_rate": 7.556631212114878e-05, "loss": 0.0215, "step": 108810 }, { "epoch": 69.35627788400255, "grad_norm": 0.2692742943763733, "learning_rate": 7.556157539934817e-05, "loss": 0.0209, "step": 108820 }, { "epoch": 69.36265137029956, "grad_norm": 0.3940919041633606, "learning_rate": 7.555683836695028e-05, "loss": 0.0154, "step": 108830 }, { "epoch": 69.36902485659655, "grad_norm": 0.24596382677555084, "learning_rate": 7.555210102401267e-05, "loss": 0.0142, "step": 108840 }, { "epoch": 69.37539834289356, "grad_norm": 0.289396196603775, "learning_rate": 7.554736337059292e-05, "loss": 0.0173, "step": 108850 }, { "epoch": 69.38177182919057, "grad_norm": 0.40313586592674255, "learning_rate": 7.554262540674857e-05, "loss": 0.0171, "step": 108860 }, { "epoch": 69.38814531548758, "grad_norm": 0.6086888909339905, "learning_rate": 7.55378871325372e-05, "loss": 0.0266, "step": 108870 }, { "epoch": 69.39451880178457, "grad_norm": 0.2553968131542206, "learning_rate": 7.553314854801641e-05, "loss": 0.0145, "step": 108880 }, { "epoch": 69.40089228808158, "grad_norm": 0.46494343876838684, "learning_rate": 7.552840965324375e-05, "loss": 0.0171, "step": 108890 }, { "epoch": 69.40726577437859, "grad_norm": 0.45463627576828003, "learning_rate": 7.552367044827681e-05, "loss": 0.0158, "step": 108900 }, { "epoch": 69.4136392606756, "grad_norm": 0.26690274477005005, "learning_rate": 7.551893093317318e-05, "loss": 0.0175, "step": 108910 }, { "epoch": 69.42001274697259, "grad_norm": 0.6234889626502991, "learning_rate": 7.551419110799043e-05, "loss": 0.0161, "step": 108920 }, { "epoch": 69.4263862332696, "grad_norm": 0.40611496567726135, "learning_rate": 7.550945097278617e-05, "loss": 0.015, "step": 108930 }, { "epoch": 69.4327597195666, "grad_norm": 0.39796438813209534, "learning_rate": 7.550471052761801e-05, "loss": 0.0288, "step": 108940 }, { "epoch": 69.43913320586361, "grad_norm": 0.7154423594474792, "learning_rate": 7.549996977254352e-05, "loss": 0.018, "step": 108950 }, { "epoch": 69.4455066921606, "grad_norm": 0.39691969752311707, "learning_rate": 7.549522870762034e-05, "loss": 0.0169, "step": 108960 }, { "epoch": 69.45188017845761, "grad_norm": 0.2973479628562927, "learning_rate": 7.549048733290604e-05, "loss": 0.0137, "step": 108970 }, { "epoch": 69.45825366475462, "grad_norm": 0.23621079325675964, "learning_rate": 7.548574564845827e-05, "loss": 0.0218, "step": 108980 }, { "epoch": 69.46462715105163, "grad_norm": 0.5880181789398193, "learning_rate": 7.548100365433462e-05, "loss": 0.0235, "step": 108990 }, { "epoch": 69.47100063734862, "grad_norm": 0.28229820728302, "learning_rate": 7.547626135059272e-05, "loss": 0.0146, "step": 109000 }, { "epoch": 69.47737412364563, "grad_norm": 0.3866879343986511, "learning_rate": 7.547151873729019e-05, "loss": 0.0173, "step": 109010 }, { "epoch": 69.48374760994264, "grad_norm": 0.6951055526733398, "learning_rate": 7.546677581448466e-05, "loss": 0.0226, "step": 109020 }, { "epoch": 69.49012109623965, "grad_norm": 0.5561010837554932, "learning_rate": 7.546203258223375e-05, "loss": 0.0148, "step": 109030 }, { "epoch": 69.49649458253664, "grad_norm": 0.31488171219825745, "learning_rate": 7.545728904059511e-05, "loss": 0.0196, "step": 109040 }, { "epoch": 69.50286806883365, "grad_norm": 0.4457908868789673, "learning_rate": 7.545254518962638e-05, "loss": 0.0219, "step": 109050 }, { "epoch": 69.50924155513066, "grad_norm": 0.5393133163452148, "learning_rate": 7.54478010293852e-05, "loss": 0.0241, "step": 109060 }, { "epoch": 69.51561504142767, "grad_norm": 0.5015327334403992, "learning_rate": 7.544305655992919e-05, "loss": 0.0129, "step": 109070 }, { "epoch": 69.52198852772466, "grad_norm": 0.35094714164733887, "learning_rate": 7.543831178131603e-05, "loss": 0.0088, "step": 109080 }, { "epoch": 69.52836201402167, "grad_norm": 0.3180698752403259, "learning_rate": 7.543356669360335e-05, "loss": 0.0208, "step": 109090 }, { "epoch": 69.53473550031867, "grad_norm": 0.6049396395683289, "learning_rate": 7.542882129684885e-05, "loss": 0.0232, "step": 109100 }, { "epoch": 69.54110898661568, "grad_norm": 0.4450114965438843, "learning_rate": 7.542407559111013e-05, "loss": 0.0227, "step": 109110 }, { "epoch": 69.54748247291268, "grad_norm": 0.5628753304481506, "learning_rate": 7.541932957644492e-05, "loss": 0.0188, "step": 109120 }, { "epoch": 69.55385595920968, "grad_norm": 0.27134501934051514, "learning_rate": 7.541458325291085e-05, "loss": 0.0147, "step": 109130 }, { "epoch": 69.56022944550669, "grad_norm": 0.6742274761199951, "learning_rate": 7.540983662056557e-05, "loss": 0.0184, "step": 109140 }, { "epoch": 69.5666029318037, "grad_norm": 0.6480090022087097, "learning_rate": 7.540508967946682e-05, "loss": 0.0258, "step": 109150 }, { "epoch": 69.5729764181007, "grad_norm": 0.3208577036857605, "learning_rate": 7.540034242967223e-05, "loss": 0.0202, "step": 109160 }, { "epoch": 69.5793499043977, "grad_norm": 0.2661128640174866, "learning_rate": 7.539559487123949e-05, "loss": 0.0148, "step": 109170 }, { "epoch": 69.58572339069471, "grad_norm": 0.31732413172721863, "learning_rate": 7.53908470042263e-05, "loss": 0.0133, "step": 109180 }, { "epoch": 69.59209687699172, "grad_norm": 0.5046302676200867, "learning_rate": 7.538609882869035e-05, "loss": 0.0183, "step": 109190 }, { "epoch": 69.59847036328873, "grad_norm": 0.25919511914253235, "learning_rate": 7.538135034468931e-05, "loss": 0.0132, "step": 109200 }, { "epoch": 69.60484384958572, "grad_norm": 0.4193447530269623, "learning_rate": 7.537660155228092e-05, "loss": 0.0229, "step": 109210 }, { "epoch": 69.61121733588273, "grad_norm": 0.30501729249954224, "learning_rate": 7.537185245152286e-05, "loss": 0.018, "step": 109220 }, { "epoch": 69.61759082217974, "grad_norm": 0.40610194206237793, "learning_rate": 7.536710304247283e-05, "loss": 0.023, "step": 109230 }, { "epoch": 69.62396430847674, "grad_norm": 0.24277259409427643, "learning_rate": 7.536235332518855e-05, "loss": 0.0158, "step": 109240 }, { "epoch": 69.63033779477374, "grad_norm": 0.532320499420166, "learning_rate": 7.535760329972772e-05, "loss": 0.0176, "step": 109250 }, { "epoch": 69.63671128107075, "grad_norm": 0.2230072319507599, "learning_rate": 7.535285296614809e-05, "loss": 0.017, "step": 109260 }, { "epoch": 69.64308476736775, "grad_norm": 0.16715764999389648, "learning_rate": 7.534810232450736e-05, "loss": 0.0126, "step": 109270 }, { "epoch": 69.64945825366476, "grad_norm": 0.2856670022010803, "learning_rate": 7.534335137486323e-05, "loss": 0.0146, "step": 109280 }, { "epoch": 69.65583173996176, "grad_norm": 0.3535715341567993, "learning_rate": 7.533860011727348e-05, "loss": 0.0298, "step": 109290 }, { "epoch": 69.66220522625876, "grad_norm": 0.5474178791046143, "learning_rate": 7.533384855179581e-05, "loss": 0.0207, "step": 109300 }, { "epoch": 69.66857871255577, "grad_norm": 0.3114345967769623, "learning_rate": 7.532909667848795e-05, "loss": 0.0204, "step": 109310 }, { "epoch": 69.67495219885278, "grad_norm": 0.2399299591779709, "learning_rate": 7.532434449740765e-05, "loss": 0.0181, "step": 109320 }, { "epoch": 69.68132568514977, "grad_norm": 0.4175562858581543, "learning_rate": 7.531959200861267e-05, "loss": 0.0199, "step": 109330 }, { "epoch": 69.68769917144678, "grad_norm": 0.3727715015411377, "learning_rate": 7.531483921216072e-05, "loss": 0.0183, "step": 109340 }, { "epoch": 69.69407265774379, "grad_norm": 0.6036545634269714, "learning_rate": 7.531008610810958e-05, "loss": 0.017, "step": 109350 }, { "epoch": 69.7004461440408, "grad_norm": 0.314609169960022, "learning_rate": 7.530533269651701e-05, "loss": 0.0206, "step": 109360 }, { "epoch": 69.70681963033779, "grad_norm": 0.4172629714012146, "learning_rate": 7.530057897744074e-05, "loss": 0.014, "step": 109370 }, { "epoch": 69.7131931166348, "grad_norm": 0.2995571792125702, "learning_rate": 7.529582495093854e-05, "loss": 0.0187, "step": 109380 }, { "epoch": 69.7195666029318, "grad_norm": 0.4165508449077606, "learning_rate": 7.529107061706822e-05, "loss": 0.0277, "step": 109390 }, { "epoch": 69.72594008922881, "grad_norm": 0.21330003440380096, "learning_rate": 7.528631597588746e-05, "loss": 0.022, "step": 109400 }, { "epoch": 69.73231357552581, "grad_norm": 0.5039772391319275, "learning_rate": 7.528156102745413e-05, "loss": 0.0279, "step": 109410 }, { "epoch": 69.73868706182282, "grad_norm": 0.6918461322784424, "learning_rate": 7.527680577182595e-05, "loss": 0.0268, "step": 109420 }, { "epoch": 69.74506054811982, "grad_norm": 0.4251154959201813, "learning_rate": 7.527205020906071e-05, "loss": 0.0176, "step": 109430 }, { "epoch": 69.75143403441683, "grad_norm": 0.6789569854736328, "learning_rate": 7.52672943392162e-05, "loss": 0.0278, "step": 109440 }, { "epoch": 69.75780752071383, "grad_norm": 0.3526040315628052, "learning_rate": 7.526253816235023e-05, "loss": 0.0194, "step": 109450 }, { "epoch": 69.76418100701083, "grad_norm": 0.26495787501335144, "learning_rate": 7.525778167852054e-05, "loss": 0.0209, "step": 109460 }, { "epoch": 69.77055449330784, "grad_norm": 0.21037988364696503, "learning_rate": 7.525302488778497e-05, "loss": 0.0273, "step": 109470 }, { "epoch": 69.77692797960485, "grad_norm": 0.6247171759605408, "learning_rate": 7.52482677902013e-05, "loss": 0.0299, "step": 109480 }, { "epoch": 69.78330146590184, "grad_norm": 0.46315404772758484, "learning_rate": 7.524351038582735e-05, "loss": 0.0164, "step": 109490 }, { "epoch": 69.78967495219885, "grad_norm": 0.635569155216217, "learning_rate": 7.523875267472089e-05, "loss": 0.0328, "step": 109500 }, { "epoch": 69.79604843849586, "grad_norm": 0.41595327854156494, "learning_rate": 7.523399465693977e-05, "loss": 0.0159, "step": 109510 }, { "epoch": 69.80242192479287, "grad_norm": 0.28905194997787476, "learning_rate": 7.52292363325418e-05, "loss": 0.0188, "step": 109520 }, { "epoch": 69.80879541108986, "grad_norm": 0.2691091299057007, "learning_rate": 7.522447770158477e-05, "loss": 0.0149, "step": 109530 }, { "epoch": 69.81516889738687, "grad_norm": 0.4444141685962677, "learning_rate": 7.521971876412653e-05, "loss": 0.0189, "step": 109540 }, { "epoch": 69.82154238368388, "grad_norm": 0.44206503033638, "learning_rate": 7.521495952022489e-05, "loss": 0.0224, "step": 109550 }, { "epoch": 69.82791586998088, "grad_norm": 0.5703757405281067, "learning_rate": 7.52101999699377e-05, "loss": 0.0225, "step": 109560 }, { "epoch": 69.83428935627788, "grad_norm": 0.3079952299594879, "learning_rate": 7.520544011332276e-05, "loss": 0.0195, "step": 109570 }, { "epoch": 69.84066284257489, "grad_norm": 0.6065131425857544, "learning_rate": 7.520067995043794e-05, "loss": 0.0222, "step": 109580 }, { "epoch": 69.8470363288719, "grad_norm": 0.4130432903766632, "learning_rate": 7.519591948134104e-05, "loss": 0.0132, "step": 109590 }, { "epoch": 69.8534098151689, "grad_norm": 0.609714150428772, "learning_rate": 7.519115870608996e-05, "loss": 0.0415, "step": 109600 }, { "epoch": 69.8597833014659, "grad_norm": 0.5145106911659241, "learning_rate": 7.51863976247425e-05, "loss": 0.0207, "step": 109610 }, { "epoch": 69.8661567877629, "grad_norm": 0.2117522656917572, "learning_rate": 7.518163623735654e-05, "loss": 0.0203, "step": 109620 }, { "epoch": 69.87253027405991, "grad_norm": 0.25884926319122314, "learning_rate": 7.517687454398992e-05, "loss": 0.0157, "step": 109630 }, { "epoch": 69.87890376035692, "grad_norm": 0.31358516216278076, "learning_rate": 7.517211254470049e-05, "loss": 0.0198, "step": 109640 }, { "epoch": 69.88527724665391, "grad_norm": 0.6769780516624451, "learning_rate": 7.516735023954614e-05, "loss": 0.0202, "step": 109650 }, { "epoch": 69.89165073295092, "grad_norm": 0.28477713465690613, "learning_rate": 7.516258762858472e-05, "loss": 0.016, "step": 109660 }, { "epoch": 69.89802421924793, "grad_norm": 0.432976633310318, "learning_rate": 7.51578247118741e-05, "loss": 0.0181, "step": 109670 }, { "epoch": 69.90439770554494, "grad_norm": 0.6001273989677429, "learning_rate": 7.515306148947215e-05, "loss": 0.0184, "step": 109680 }, { "epoch": 69.91077119184193, "grad_norm": 0.4394828975200653, "learning_rate": 7.514829796143677e-05, "loss": 0.0155, "step": 109690 }, { "epoch": 69.91714467813894, "grad_norm": 0.6605792045593262, "learning_rate": 7.514353412782583e-05, "loss": 0.0301, "step": 109700 }, { "epoch": 69.92351816443595, "grad_norm": 0.20071323215961456, "learning_rate": 7.513876998869719e-05, "loss": 0.0128, "step": 109710 }, { "epoch": 69.92989165073296, "grad_norm": 0.456286758184433, "learning_rate": 7.513400554410877e-05, "loss": 0.0173, "step": 109720 }, { "epoch": 69.93626513702995, "grad_norm": 0.2944415807723999, "learning_rate": 7.512924079411846e-05, "loss": 0.0204, "step": 109730 }, { "epoch": 69.94263862332696, "grad_norm": 0.5973456501960754, "learning_rate": 7.512447573878414e-05, "loss": 0.0167, "step": 109740 }, { "epoch": 69.94901210962396, "grad_norm": 0.4693790078163147, "learning_rate": 7.51197103781637e-05, "loss": 0.0211, "step": 109750 }, { "epoch": 69.95538559592097, "grad_norm": 0.2627420723438263, "learning_rate": 7.511494471231509e-05, "loss": 0.0192, "step": 109760 }, { "epoch": 69.96175908221797, "grad_norm": 0.32957616448402405, "learning_rate": 7.511017874129618e-05, "loss": 0.0138, "step": 109770 }, { "epoch": 69.96813256851497, "grad_norm": 0.23858226835727692, "learning_rate": 7.51054124651649e-05, "loss": 0.0147, "step": 109780 }, { "epoch": 69.97450605481198, "grad_norm": 0.516528308391571, "learning_rate": 7.510064588397914e-05, "loss": 0.0176, "step": 109790 }, { "epoch": 69.98087954110899, "grad_norm": 0.45180732011795044, "learning_rate": 7.509587899779684e-05, "loss": 0.0185, "step": 109800 }, { "epoch": 69.98725302740598, "grad_norm": 0.11954141408205032, "learning_rate": 7.509111180667591e-05, "loss": 0.0189, "step": 109810 }, { "epoch": 69.99362651370299, "grad_norm": 0.4547651708126068, "learning_rate": 7.50863443106743e-05, "loss": 0.0124, "step": 109820 }, { "epoch": 70.0, "grad_norm": 0.34960684180259705, "learning_rate": 7.50815765098499e-05, "loss": 0.0145, "step": 109830 }, { "epoch": 70.00637348629701, "grad_norm": 0.2908296585083008, "learning_rate": 7.507680840426067e-05, "loss": 0.021, "step": 109840 }, { "epoch": 70.01274697259402, "grad_norm": 0.6088809370994568, "learning_rate": 7.507203999396455e-05, "loss": 0.0323, "step": 109850 }, { "epoch": 70.01912045889101, "grad_norm": 0.6494719982147217, "learning_rate": 7.506727127901946e-05, "loss": 0.0407, "step": 109860 }, { "epoch": 70.02549394518802, "grad_norm": 0.5126445293426514, "learning_rate": 7.506250225948336e-05, "loss": 0.0187, "step": 109870 }, { "epoch": 70.03186743148503, "grad_norm": 0.5078320503234863, "learning_rate": 7.505773293541421e-05, "loss": 0.0208, "step": 109880 }, { "epoch": 70.03824091778203, "grad_norm": 0.3333047926425934, "learning_rate": 7.505296330686993e-05, "loss": 0.0144, "step": 109890 }, { "epoch": 70.04461440407903, "grad_norm": 0.6727513074874878, "learning_rate": 7.50481933739085e-05, "loss": 0.0267, "step": 109900 }, { "epoch": 70.05098789037604, "grad_norm": 0.36027681827545166, "learning_rate": 7.504342313658785e-05, "loss": 0.0218, "step": 109910 }, { "epoch": 70.05736137667304, "grad_norm": 0.2214171439409256, "learning_rate": 7.503865259496599e-05, "loss": 0.013, "step": 109920 }, { "epoch": 70.06373486297005, "grad_norm": 0.40567103028297424, "learning_rate": 7.503388174910084e-05, "loss": 0.0172, "step": 109930 }, { "epoch": 70.07010834926704, "grad_norm": 0.785518229007721, "learning_rate": 7.50291105990504e-05, "loss": 0.0171, "step": 109940 }, { "epoch": 70.07648183556405, "grad_norm": 0.39964479207992554, "learning_rate": 7.502433914487265e-05, "loss": 0.02, "step": 109950 }, { "epoch": 70.08285532186106, "grad_norm": 0.5685421228408813, "learning_rate": 7.501956738662554e-05, "loss": 0.0305, "step": 109960 }, { "epoch": 70.08922880815807, "grad_norm": 0.9127949476242065, "learning_rate": 7.501479532436706e-05, "loss": 0.0193, "step": 109970 }, { "epoch": 70.09560229445506, "grad_norm": 0.5881022810935974, "learning_rate": 7.501002295815521e-05, "loss": 0.0175, "step": 109980 }, { "epoch": 70.10197578075207, "grad_norm": 0.5793927907943726, "learning_rate": 7.500525028804794e-05, "loss": 0.0283, "step": 109990 }, { "epoch": 70.10834926704908, "grad_norm": 0.33323097229003906, "learning_rate": 7.500047731410329e-05, "loss": 0.0244, "step": 110000 }, { "epoch": 70.11472275334609, "grad_norm": 0.3606049716472626, "learning_rate": 7.499570403637924e-05, "loss": 0.0187, "step": 110010 }, { "epoch": 70.12109623964308, "grad_norm": 0.5225717425346375, "learning_rate": 7.499093045493379e-05, "loss": 0.0197, "step": 110020 }, { "epoch": 70.12746972594009, "grad_norm": 0.35697054862976074, "learning_rate": 7.498615656982493e-05, "loss": 0.0227, "step": 110030 }, { "epoch": 70.1338432122371, "grad_norm": 0.4346321225166321, "learning_rate": 7.498138238111068e-05, "loss": 0.0158, "step": 110040 }, { "epoch": 70.1402166985341, "grad_norm": 0.5753293633460999, "learning_rate": 7.497660788884903e-05, "loss": 0.0237, "step": 110050 }, { "epoch": 70.1465901848311, "grad_norm": 0.507948637008667, "learning_rate": 7.497183309309804e-05, "loss": 0.0175, "step": 110060 }, { "epoch": 70.1529636711281, "grad_norm": 0.279093861579895, "learning_rate": 7.496705799391569e-05, "loss": 0.0144, "step": 110070 }, { "epoch": 70.15933715742511, "grad_norm": 0.26793280243873596, "learning_rate": 7.496228259136001e-05, "loss": 0.0209, "step": 110080 }, { "epoch": 70.16571064372212, "grad_norm": 0.666017472743988, "learning_rate": 7.495750688548902e-05, "loss": 0.0298, "step": 110090 }, { "epoch": 70.17208413001912, "grad_norm": 0.45712095499038696, "learning_rate": 7.495273087636077e-05, "loss": 0.0151, "step": 110100 }, { "epoch": 70.17845761631612, "grad_norm": 0.4813659191131592, "learning_rate": 7.494795456403329e-05, "loss": 0.0254, "step": 110110 }, { "epoch": 70.18483110261313, "grad_norm": 0.29053524136543274, "learning_rate": 7.494317794856459e-05, "loss": 0.0155, "step": 110120 }, { "epoch": 70.19120458891014, "grad_norm": 0.3293651044368744, "learning_rate": 7.493840103001273e-05, "loss": 0.0147, "step": 110130 }, { "epoch": 70.19757807520713, "grad_norm": 0.38272738456726074, "learning_rate": 7.493362380843576e-05, "loss": 0.0179, "step": 110140 }, { "epoch": 70.20395156150414, "grad_norm": 0.3908238410949707, "learning_rate": 7.49288462838917e-05, "loss": 0.0198, "step": 110150 }, { "epoch": 70.21032504780115, "grad_norm": 0.3286839425563812, "learning_rate": 7.492406845643864e-05, "loss": 0.0201, "step": 110160 }, { "epoch": 70.21669853409816, "grad_norm": 0.5370028614997864, "learning_rate": 7.491929032613461e-05, "loss": 0.0188, "step": 110170 }, { "epoch": 70.22307202039515, "grad_norm": 0.32326292991638184, "learning_rate": 7.491451189303769e-05, "loss": 0.0152, "step": 110180 }, { "epoch": 70.22944550669216, "grad_norm": 0.4833873510360718, "learning_rate": 7.49097331572059e-05, "loss": 0.0156, "step": 110190 }, { "epoch": 70.23581899298917, "grad_norm": 0.27931270003318787, "learning_rate": 7.490495411869735e-05, "loss": 0.0113, "step": 110200 }, { "epoch": 70.24219247928617, "grad_norm": 0.3658488094806671, "learning_rate": 7.490017477757009e-05, "loss": 0.0123, "step": 110210 }, { "epoch": 70.24856596558317, "grad_norm": 0.21749505400657654, "learning_rate": 7.489539513388219e-05, "loss": 0.0202, "step": 110220 }, { "epoch": 70.25493945188018, "grad_norm": 0.4358943700790405, "learning_rate": 7.489061518769173e-05, "loss": 0.0254, "step": 110230 }, { "epoch": 70.26131293817718, "grad_norm": 0.6098313927650452, "learning_rate": 7.488583493905682e-05, "loss": 0.0276, "step": 110240 }, { "epoch": 70.26768642447419, "grad_norm": 0.4728843867778778, "learning_rate": 7.48810543880355e-05, "loss": 0.0167, "step": 110250 }, { "epoch": 70.27405991077119, "grad_norm": 0.5907619595527649, "learning_rate": 7.487627353468589e-05, "loss": 0.0203, "step": 110260 }, { "epoch": 70.2804333970682, "grad_norm": 0.33721596002578735, "learning_rate": 7.487149237906606e-05, "loss": 0.01, "step": 110270 }, { "epoch": 70.2868068833652, "grad_norm": 0.6818324327468872, "learning_rate": 7.486671092123412e-05, "loss": 0.0251, "step": 110280 }, { "epoch": 70.29318036966221, "grad_norm": 0.20555658638477325, "learning_rate": 7.486192916124817e-05, "loss": 0.0165, "step": 110290 }, { "epoch": 70.2995538559592, "grad_norm": 0.28444740176200867, "learning_rate": 7.48571470991663e-05, "loss": 0.0104, "step": 110300 }, { "epoch": 70.30592734225621, "grad_norm": 0.6349973678588867, "learning_rate": 7.485236473504662e-05, "loss": 0.0223, "step": 110310 }, { "epoch": 70.31230082855322, "grad_norm": 0.49866020679473877, "learning_rate": 7.484758206894724e-05, "loss": 0.0116, "step": 110320 }, { "epoch": 70.31867431485023, "grad_norm": 0.8094007968902588, "learning_rate": 7.484279910092628e-05, "loss": 0.0165, "step": 110330 }, { "epoch": 70.32504780114722, "grad_norm": 0.5938400626182556, "learning_rate": 7.483801583104186e-05, "loss": 0.0144, "step": 110340 }, { "epoch": 70.33142128744423, "grad_norm": 0.24431605637073517, "learning_rate": 7.48332322593521e-05, "loss": 0.0213, "step": 110350 }, { "epoch": 70.33779477374124, "grad_norm": 0.4685380756855011, "learning_rate": 7.482844838591512e-05, "loss": 0.0224, "step": 110360 }, { "epoch": 70.34416826003824, "grad_norm": 0.25866037607192993, "learning_rate": 7.482366421078903e-05, "loss": 0.0151, "step": 110370 }, { "epoch": 70.35054174633524, "grad_norm": 0.5931476950645447, "learning_rate": 7.481887973403202e-05, "loss": 0.0152, "step": 110380 }, { "epoch": 70.35691523263225, "grad_norm": 0.24416254460811615, "learning_rate": 7.481409495570215e-05, "loss": 0.0204, "step": 110390 }, { "epoch": 70.36328871892925, "grad_norm": 0.3839692175388336, "learning_rate": 7.480930987585762e-05, "loss": 0.0256, "step": 110400 }, { "epoch": 70.36966220522626, "grad_norm": 0.6211572289466858, "learning_rate": 7.480452449455655e-05, "loss": 0.0167, "step": 110410 }, { "epoch": 70.37603569152326, "grad_norm": 0.42414870858192444, "learning_rate": 7.479973881185708e-05, "loss": 0.0171, "step": 110420 }, { "epoch": 70.38240917782026, "grad_norm": 0.40094777941703796, "learning_rate": 7.47949528278174e-05, "loss": 0.0142, "step": 110430 }, { "epoch": 70.38878266411727, "grad_norm": 0.1959487795829773, "learning_rate": 7.479016654249558e-05, "loss": 0.0159, "step": 110440 }, { "epoch": 70.39515615041428, "grad_norm": 0.2864038944244385, "learning_rate": 7.478537995594988e-05, "loss": 0.0219, "step": 110450 }, { "epoch": 70.40152963671127, "grad_norm": 0.5128370523452759, "learning_rate": 7.478059306823838e-05, "loss": 0.0203, "step": 110460 }, { "epoch": 70.40790312300828, "grad_norm": 0.4851933717727661, "learning_rate": 7.47758058794193e-05, "loss": 0.019, "step": 110470 }, { "epoch": 70.41427660930529, "grad_norm": 0.5393998026847839, "learning_rate": 7.477101838955077e-05, "loss": 0.022, "step": 110480 }, { "epoch": 70.4206500956023, "grad_norm": 0.5386297702789307, "learning_rate": 7.476623059869098e-05, "loss": 0.0224, "step": 110490 }, { "epoch": 70.4270235818993, "grad_norm": 0.34283939003944397, "learning_rate": 7.47614425068981e-05, "loss": 0.0121, "step": 110500 }, { "epoch": 70.4333970681963, "grad_norm": 0.3141005337238312, "learning_rate": 7.475665411423032e-05, "loss": 0.0131, "step": 110510 }, { "epoch": 70.43977055449331, "grad_norm": 0.37007036805152893, "learning_rate": 7.475186542074584e-05, "loss": 0.0216, "step": 110520 }, { "epoch": 70.44614404079032, "grad_norm": 0.2906089127063751, "learning_rate": 7.474707642650279e-05, "loss": 0.0144, "step": 110530 }, { "epoch": 70.45251752708732, "grad_norm": 0.7727821469306946, "learning_rate": 7.474228713155942e-05, "loss": 0.0292, "step": 110540 }, { "epoch": 70.45889101338432, "grad_norm": 0.6066847443580627, "learning_rate": 7.473749753597389e-05, "loss": 0.0239, "step": 110550 }, { "epoch": 70.46526449968133, "grad_norm": 0.6106574535369873, "learning_rate": 7.473270763980441e-05, "loss": 0.0205, "step": 110560 }, { "epoch": 70.47163798597833, "grad_norm": 0.7122835516929626, "learning_rate": 7.472791744310919e-05, "loss": 0.0174, "step": 110570 }, { "epoch": 70.47801147227534, "grad_norm": 0.32336845993995667, "learning_rate": 7.472312694594641e-05, "loss": 0.017, "step": 110580 }, { "epoch": 70.48438495857233, "grad_norm": 0.22056859731674194, "learning_rate": 7.471833614837431e-05, "loss": 0.0201, "step": 110590 }, { "epoch": 70.49075844486934, "grad_norm": 0.47530868649482727, "learning_rate": 7.471354505045108e-05, "loss": 0.0194, "step": 110600 }, { "epoch": 70.49713193116635, "grad_norm": 0.3424436151981354, "learning_rate": 7.470875365223496e-05, "loss": 0.0251, "step": 110610 }, { "epoch": 70.50350541746336, "grad_norm": 0.29361146688461304, "learning_rate": 7.470396195378413e-05, "loss": 0.0253, "step": 110620 }, { "epoch": 70.50987890376035, "grad_norm": 0.33092179894447327, "learning_rate": 7.469916995515684e-05, "loss": 0.0257, "step": 110630 }, { "epoch": 70.51625239005736, "grad_norm": 0.4215983748435974, "learning_rate": 7.469437765641132e-05, "loss": 0.0188, "step": 110640 }, { "epoch": 70.52262587635437, "grad_norm": 0.3229232430458069, "learning_rate": 7.468958505760581e-05, "loss": 0.0198, "step": 110650 }, { "epoch": 70.52899936265138, "grad_norm": 0.37529292702674866, "learning_rate": 7.468479215879852e-05, "loss": 0.0212, "step": 110660 }, { "epoch": 70.53537284894837, "grad_norm": 0.35298141837120056, "learning_rate": 7.467999896004769e-05, "loss": 0.0209, "step": 110670 }, { "epoch": 70.54174633524538, "grad_norm": 0.3835197985172272, "learning_rate": 7.46752054614116e-05, "loss": 0.0244, "step": 110680 }, { "epoch": 70.54811982154239, "grad_norm": 0.4667966663837433, "learning_rate": 7.467041166294844e-05, "loss": 0.0165, "step": 110690 }, { "epoch": 70.5544933078394, "grad_norm": 0.3145020008087158, "learning_rate": 7.466561756471648e-05, "loss": 0.0251, "step": 110700 }, { "epoch": 70.56086679413639, "grad_norm": 0.4987790286540985, "learning_rate": 7.4660823166774e-05, "loss": 0.0207, "step": 110710 }, { "epoch": 70.5672402804334, "grad_norm": 0.5521416664123535, "learning_rate": 7.465602846917923e-05, "loss": 0.0234, "step": 110720 }, { "epoch": 70.5736137667304, "grad_norm": 0.630126953125, "learning_rate": 7.465123347199041e-05, "loss": 0.0212, "step": 110730 }, { "epoch": 70.57998725302741, "grad_norm": 0.6182052493095398, "learning_rate": 7.464643817526586e-05, "loss": 0.014, "step": 110740 }, { "epoch": 70.5863607393244, "grad_norm": 0.7427298426628113, "learning_rate": 7.46416425790638e-05, "loss": 0.0189, "step": 110750 }, { "epoch": 70.59273422562141, "grad_norm": 0.3367733061313629, "learning_rate": 7.463684668344254e-05, "loss": 0.0135, "step": 110760 }, { "epoch": 70.59910771191842, "grad_norm": 0.23937629163265228, "learning_rate": 7.463205048846031e-05, "loss": 0.0125, "step": 110770 }, { "epoch": 70.60548119821543, "grad_norm": 0.4011872410774231, "learning_rate": 7.462725399417541e-05, "loss": 0.0182, "step": 110780 }, { "epoch": 70.61185468451242, "grad_norm": 0.3864860236644745, "learning_rate": 7.462245720064612e-05, "loss": 0.0178, "step": 110790 }, { "epoch": 70.61822817080943, "grad_norm": 0.3954959809780121, "learning_rate": 7.461766010793073e-05, "loss": 0.0253, "step": 110800 }, { "epoch": 70.62460165710644, "grad_norm": 0.3908807635307312, "learning_rate": 7.461286271608754e-05, "loss": 0.0139, "step": 110810 }, { "epoch": 70.63097514340345, "grad_norm": 0.5494047403335571, "learning_rate": 7.460806502517481e-05, "loss": 0.017, "step": 110820 }, { "epoch": 70.63734862970044, "grad_norm": 0.24690695106983185, "learning_rate": 7.460326703525086e-05, "loss": 0.0141, "step": 110830 }, { "epoch": 70.64372211599745, "grad_norm": 0.21273037791252136, "learning_rate": 7.4598468746374e-05, "loss": 0.0109, "step": 110840 }, { "epoch": 70.65009560229446, "grad_norm": 0.28744566440582275, "learning_rate": 7.459367015860252e-05, "loss": 0.018, "step": 110850 }, { "epoch": 70.65646908859146, "grad_norm": 1.0926573276519775, "learning_rate": 7.458887127199471e-05, "loss": 0.0216, "step": 110860 }, { "epoch": 70.66284257488846, "grad_norm": 0.7990713715553284, "learning_rate": 7.458407208660892e-05, "loss": 0.0178, "step": 110870 }, { "epoch": 70.66921606118547, "grad_norm": 0.38773682713508606, "learning_rate": 7.457927260250342e-05, "loss": 0.0142, "step": 110880 }, { "epoch": 70.67558954748247, "grad_norm": 0.36491456627845764, "learning_rate": 7.457447281973656e-05, "loss": 0.0179, "step": 110890 }, { "epoch": 70.68196303377948, "grad_norm": 0.2779029607772827, "learning_rate": 7.456967273836665e-05, "loss": 0.0162, "step": 110900 }, { "epoch": 70.68833652007648, "grad_norm": 0.22199025750160217, "learning_rate": 7.456487235845204e-05, "loss": 0.011, "step": 110910 }, { "epoch": 70.69471000637348, "grad_norm": 0.44108274579048157, "learning_rate": 7.4560071680051e-05, "loss": 0.0145, "step": 110920 }, { "epoch": 70.70108349267049, "grad_norm": 0.28117650747299194, "learning_rate": 7.455527070322193e-05, "loss": 0.0179, "step": 110930 }, { "epoch": 70.7074569789675, "grad_norm": 0.32963407039642334, "learning_rate": 7.455046942802313e-05, "loss": 0.0251, "step": 110940 }, { "epoch": 70.7138304652645, "grad_norm": 0.6334715485572815, "learning_rate": 7.454566785451294e-05, "loss": 0.0206, "step": 110950 }, { "epoch": 70.7202039515615, "grad_norm": 0.387898713350296, "learning_rate": 7.454086598274972e-05, "loss": 0.0151, "step": 110960 }, { "epoch": 70.72657743785851, "grad_norm": 0.5127922296524048, "learning_rate": 7.45360638127918e-05, "loss": 0.0228, "step": 110970 }, { "epoch": 70.73295092415552, "grad_norm": 0.32181069254875183, "learning_rate": 7.453126134469755e-05, "loss": 0.0193, "step": 110980 }, { "epoch": 70.73932441045251, "grad_norm": 0.31377163529396057, "learning_rate": 7.45264585785253e-05, "loss": 0.0154, "step": 110990 }, { "epoch": 70.74569789674952, "grad_norm": 0.4714060127735138, "learning_rate": 7.452165551433343e-05, "loss": 0.02, "step": 111000 }, { "epoch": 70.75207138304653, "grad_norm": 0.39887291193008423, "learning_rate": 7.45168521521803e-05, "loss": 0.0209, "step": 111010 }, { "epoch": 70.75844486934353, "grad_norm": 0.1710035353899002, "learning_rate": 7.451204849212425e-05, "loss": 0.0147, "step": 111020 }, { "epoch": 70.76481835564053, "grad_norm": 0.36765360832214355, "learning_rate": 7.450724453422368e-05, "loss": 0.0233, "step": 111030 }, { "epoch": 70.77119184193754, "grad_norm": 0.28101223707199097, "learning_rate": 7.450244027853695e-05, "loss": 0.015, "step": 111040 }, { "epoch": 70.77756532823454, "grad_norm": 0.46950262784957886, "learning_rate": 7.449763572512242e-05, "loss": 0.0141, "step": 111050 }, { "epoch": 70.78393881453155, "grad_norm": 0.24851472675800323, "learning_rate": 7.449283087403851e-05, "loss": 0.0187, "step": 111060 }, { "epoch": 70.79031230082855, "grad_norm": 0.5380865931510925, "learning_rate": 7.448802572534356e-05, "loss": 0.0202, "step": 111070 }, { "epoch": 70.79668578712555, "grad_norm": 0.2650468945503235, "learning_rate": 7.4483220279096e-05, "loss": 0.02, "step": 111080 }, { "epoch": 70.80305927342256, "grad_norm": 0.4117903411388397, "learning_rate": 7.447841453535418e-05, "loss": 0.016, "step": 111090 }, { "epoch": 70.80943275971957, "grad_norm": 0.704867959022522, "learning_rate": 7.447360849417652e-05, "loss": 0.0208, "step": 111100 }, { "epoch": 70.81580624601658, "grad_norm": 0.25766023993492126, "learning_rate": 7.44688021556214e-05, "loss": 0.0195, "step": 111110 }, { "epoch": 70.82217973231357, "grad_norm": 0.3519233465194702, "learning_rate": 7.446399551974723e-05, "loss": 0.0178, "step": 111120 }, { "epoch": 70.82855321861058, "grad_norm": 0.6770761013031006, "learning_rate": 7.445918858661244e-05, "loss": 0.0306, "step": 111130 }, { "epoch": 70.83492670490759, "grad_norm": 0.35930171608924866, "learning_rate": 7.445438135627539e-05, "loss": 0.0149, "step": 111140 }, { "epoch": 70.8413001912046, "grad_norm": 0.48433917760849, "learning_rate": 7.444957382879454e-05, "loss": 0.0162, "step": 111150 }, { "epoch": 70.84767367750159, "grad_norm": 0.4752691090106964, "learning_rate": 7.444476600422828e-05, "loss": 0.0197, "step": 111160 }, { "epoch": 70.8540471637986, "grad_norm": 0.6118513345718384, "learning_rate": 7.4439957882635e-05, "loss": 0.0125, "step": 111170 }, { "epoch": 70.8604206500956, "grad_norm": 0.2735186815261841, "learning_rate": 7.44351494640732e-05, "loss": 0.0182, "step": 111180 }, { "epoch": 70.86679413639261, "grad_norm": 0.6872962713241577, "learning_rate": 7.443034074860124e-05, "loss": 0.0247, "step": 111190 }, { "epoch": 70.8731676226896, "grad_norm": 0.37924760580062866, "learning_rate": 7.442553173627757e-05, "loss": 0.0126, "step": 111200 }, { "epoch": 70.87954110898661, "grad_norm": 0.425611287355423, "learning_rate": 7.442072242716064e-05, "loss": 0.0141, "step": 111210 }, { "epoch": 70.88591459528362, "grad_norm": 0.3564702570438385, "learning_rate": 7.441591282130888e-05, "loss": 0.0121, "step": 111220 }, { "epoch": 70.89228808158063, "grad_norm": 0.286515474319458, "learning_rate": 7.441110291878072e-05, "loss": 0.0138, "step": 111230 }, { "epoch": 70.89866156787762, "grad_norm": 0.43566083908081055, "learning_rate": 7.44062927196346e-05, "loss": 0.0251, "step": 111240 }, { "epoch": 70.90503505417463, "grad_norm": 0.5162158608436584, "learning_rate": 7.4401482223929e-05, "loss": 0.0336, "step": 111250 }, { "epoch": 70.91140854047164, "grad_norm": 0.5080259442329407, "learning_rate": 7.439667143172232e-05, "loss": 0.0196, "step": 111260 }, { "epoch": 70.91778202676865, "grad_norm": 0.29871058464050293, "learning_rate": 7.439186034307308e-05, "loss": 0.0148, "step": 111270 }, { "epoch": 70.92415551306564, "grad_norm": 0.4005752205848694, "learning_rate": 7.43870489580397e-05, "loss": 0.0179, "step": 111280 }, { "epoch": 70.93052899936265, "grad_norm": 0.2640437185764313, "learning_rate": 7.438223727668065e-05, "loss": 0.0268, "step": 111290 }, { "epoch": 70.93690248565966, "grad_norm": 0.5144799947738647, "learning_rate": 7.437742529905438e-05, "loss": 0.0174, "step": 111300 }, { "epoch": 70.94327597195667, "grad_norm": 0.4153822958469391, "learning_rate": 7.437261302521941e-05, "loss": 0.0184, "step": 111310 }, { "epoch": 70.94964945825366, "grad_norm": 0.30273035168647766, "learning_rate": 7.436780045523416e-05, "loss": 0.0225, "step": 111320 }, { "epoch": 70.95602294455067, "grad_norm": 0.347936749458313, "learning_rate": 7.436298758915712e-05, "loss": 0.0202, "step": 111330 }, { "epoch": 70.96239643084768, "grad_norm": 0.2888943552970886, "learning_rate": 7.435817442704679e-05, "loss": 0.0142, "step": 111340 }, { "epoch": 70.96876991714468, "grad_norm": 0.45582571625709534, "learning_rate": 7.435336096896164e-05, "loss": 0.0134, "step": 111350 }, { "epoch": 70.97514340344168, "grad_norm": 0.17989078164100647, "learning_rate": 7.434854721496017e-05, "loss": 0.0197, "step": 111360 }, { "epoch": 70.98151688973869, "grad_norm": 0.37364208698272705, "learning_rate": 7.434373316510083e-05, "loss": 0.0234, "step": 111370 }, { "epoch": 70.9878903760357, "grad_norm": 0.4869678318500519, "learning_rate": 7.433891881944218e-05, "loss": 0.0251, "step": 111380 }, { "epoch": 70.9942638623327, "grad_norm": 0.4083375036716461, "learning_rate": 7.433410417804268e-05, "loss": 0.0129, "step": 111390 }, { "epoch": 71.0006373486297, "grad_norm": 0.2743567228317261, "learning_rate": 7.432928924096083e-05, "loss": 0.0174, "step": 111400 }, { "epoch": 71.0070108349267, "grad_norm": 0.23553141951560974, "learning_rate": 7.432447400825515e-05, "loss": 0.0187, "step": 111410 }, { "epoch": 71.01338432122371, "grad_norm": 0.2498624324798584, "learning_rate": 7.431965847998414e-05, "loss": 0.0186, "step": 111420 }, { "epoch": 71.01975780752072, "grad_norm": 0.2529947757720947, "learning_rate": 7.431484265620633e-05, "loss": 0.0166, "step": 111430 }, { "epoch": 71.02613129381771, "grad_norm": 0.49505311250686646, "learning_rate": 7.431002653698021e-05, "loss": 0.0107, "step": 111440 }, { "epoch": 71.03250478011472, "grad_norm": 0.5335766673088074, "learning_rate": 7.430521012236433e-05, "loss": 0.0133, "step": 111450 }, { "epoch": 71.03887826641173, "grad_norm": 0.3321877717971802, "learning_rate": 7.430039341241719e-05, "loss": 0.0098, "step": 111460 }, { "epoch": 71.04525175270874, "grad_norm": 0.6093344688415527, "learning_rate": 7.429557640719733e-05, "loss": 0.0173, "step": 111470 }, { "epoch": 71.05162523900573, "grad_norm": 0.20159131288528442, "learning_rate": 7.429075910676328e-05, "loss": 0.0136, "step": 111480 }, { "epoch": 71.05799872530274, "grad_norm": 0.2729185223579407, "learning_rate": 7.428594151117357e-05, "loss": 0.0156, "step": 111490 }, { "epoch": 71.06437221159975, "grad_norm": 0.6778607964515686, "learning_rate": 7.428112362048675e-05, "loss": 0.0227, "step": 111500 }, { "epoch": 71.07074569789675, "grad_norm": 0.3649970591068268, "learning_rate": 7.427630543476134e-05, "loss": 0.0135, "step": 111510 }, { "epoch": 71.07711918419375, "grad_norm": 0.2877497971057892, "learning_rate": 7.42714869540559e-05, "loss": 0.0143, "step": 111520 }, { "epoch": 71.08349267049076, "grad_norm": 0.28144481778144836, "learning_rate": 7.426666817842898e-05, "loss": 0.0102, "step": 111530 }, { "epoch": 71.08986615678776, "grad_norm": 0.5119896531105042, "learning_rate": 7.426184910793912e-05, "loss": 0.017, "step": 111540 }, { "epoch": 71.09623964308477, "grad_norm": 0.24396082758903503, "learning_rate": 7.42570297426449e-05, "loss": 0.0243, "step": 111550 }, { "epoch": 71.10261312938177, "grad_norm": 0.25271913409233093, "learning_rate": 7.425221008260487e-05, "loss": 0.0187, "step": 111560 }, { "epoch": 71.10898661567877, "grad_norm": 0.574061393737793, "learning_rate": 7.424739012787758e-05, "loss": 0.0241, "step": 111570 }, { "epoch": 71.11536010197578, "grad_norm": 0.42652013897895813, "learning_rate": 7.424256987852161e-05, "loss": 0.0201, "step": 111580 }, { "epoch": 71.12173358827279, "grad_norm": 0.2635291516780853, "learning_rate": 7.423774933459551e-05, "loss": 0.0142, "step": 111590 }, { "epoch": 71.12810707456978, "grad_norm": 0.5189371109008789, "learning_rate": 7.423292849615789e-05, "loss": 0.0197, "step": 111600 }, { "epoch": 71.13448056086679, "grad_norm": 0.36636531352996826, "learning_rate": 7.42281073632673e-05, "loss": 0.0179, "step": 111610 }, { "epoch": 71.1408540471638, "grad_norm": 0.2982560694217682, "learning_rate": 7.422328593598234e-05, "loss": 0.0129, "step": 111620 }, { "epoch": 71.1472275334608, "grad_norm": 0.1597938984632492, "learning_rate": 7.421846421436158e-05, "loss": 0.0117, "step": 111630 }, { "epoch": 71.1536010197578, "grad_norm": 0.6215270757675171, "learning_rate": 7.421364219846361e-05, "loss": 0.0206, "step": 111640 }, { "epoch": 71.15997450605481, "grad_norm": 0.6858156323432922, "learning_rate": 7.420881988834704e-05, "loss": 0.0189, "step": 111650 }, { "epoch": 71.16634799235182, "grad_norm": 0.5712109208106995, "learning_rate": 7.420399728407044e-05, "loss": 0.0288, "step": 111660 }, { "epoch": 71.17272147864882, "grad_norm": 0.6001536250114441, "learning_rate": 7.419917438569242e-05, "loss": 0.0292, "step": 111670 }, { "epoch": 71.17909496494582, "grad_norm": 0.27020469307899475, "learning_rate": 7.419435119327158e-05, "loss": 0.0193, "step": 111680 }, { "epoch": 71.18546845124283, "grad_norm": 0.5612889528274536, "learning_rate": 7.418952770686653e-05, "loss": 0.0212, "step": 111690 }, { "epoch": 71.19184193753983, "grad_norm": 0.5953080058097839, "learning_rate": 7.41847039265359e-05, "loss": 0.0181, "step": 111700 }, { "epoch": 71.19821542383684, "grad_norm": 0.5372666120529175, "learning_rate": 7.417987985233828e-05, "loss": 0.0191, "step": 111710 }, { "epoch": 71.20458891013385, "grad_norm": 0.3364112377166748, "learning_rate": 7.417505548433227e-05, "loss": 0.0337, "step": 111720 }, { "epoch": 71.21096239643084, "grad_norm": 0.26506349444389343, "learning_rate": 7.417023082257652e-05, "loss": 0.0254, "step": 111730 }, { "epoch": 71.21733588272785, "grad_norm": 0.48038312792778015, "learning_rate": 7.416540586712964e-05, "loss": 0.015, "step": 111740 }, { "epoch": 71.22370936902486, "grad_norm": 0.4602726995944977, "learning_rate": 7.416058061805027e-05, "loss": 0.0244, "step": 111750 }, { "epoch": 71.23008285532187, "grad_norm": 0.322104811668396, "learning_rate": 7.415575507539702e-05, "loss": 0.0159, "step": 111760 }, { "epoch": 71.23645634161886, "grad_norm": 0.3177740275859833, "learning_rate": 7.415092923922856e-05, "loss": 0.0204, "step": 111770 }, { "epoch": 71.24282982791587, "grad_norm": 0.38239848613739014, "learning_rate": 7.414610310960349e-05, "loss": 0.0285, "step": 111780 }, { "epoch": 71.24920331421288, "grad_norm": 0.4733118414878845, "learning_rate": 7.414127668658049e-05, "loss": 0.0139, "step": 111790 }, { "epoch": 71.25557680050989, "grad_norm": 0.4304279685020447, "learning_rate": 7.413644997021819e-05, "loss": 0.0256, "step": 111800 }, { "epoch": 71.26195028680688, "grad_norm": 0.3381074368953705, "learning_rate": 7.41316229605752e-05, "loss": 0.0216, "step": 111810 }, { "epoch": 71.26832377310389, "grad_norm": 0.32345151901245117, "learning_rate": 7.412679565771022e-05, "loss": 0.0156, "step": 111820 }, { "epoch": 71.2746972594009, "grad_norm": 0.2584652304649353, "learning_rate": 7.41219680616819e-05, "loss": 0.0138, "step": 111830 }, { "epoch": 71.2810707456979, "grad_norm": 0.3220323324203491, "learning_rate": 7.41171401725489e-05, "loss": 0.0105, "step": 111840 }, { "epoch": 71.2874442319949, "grad_norm": 0.33952227234840393, "learning_rate": 7.411231199036986e-05, "loss": 0.0253, "step": 111850 }, { "epoch": 71.2938177182919, "grad_norm": 0.48257675766944885, "learning_rate": 7.410748351520346e-05, "loss": 0.0195, "step": 111860 }, { "epoch": 71.30019120458891, "grad_norm": 0.4043757915496826, "learning_rate": 7.41026547471084e-05, "loss": 0.0175, "step": 111870 }, { "epoch": 71.30656469088592, "grad_norm": 0.5093423128128052, "learning_rate": 7.409782568614331e-05, "loss": 0.0159, "step": 111880 }, { "epoch": 71.31293817718291, "grad_norm": 0.49829819798469543, "learning_rate": 7.409299633236689e-05, "loss": 0.0156, "step": 111890 }, { "epoch": 71.31931166347992, "grad_norm": 0.3054064214229584, "learning_rate": 7.408816668583781e-05, "loss": 0.0147, "step": 111900 }, { "epoch": 71.32568514977693, "grad_norm": 0.2895759642124176, "learning_rate": 7.408333674661477e-05, "loss": 0.0212, "step": 111910 }, { "epoch": 71.33205863607394, "grad_norm": 0.6152570247650146, "learning_rate": 7.407850651475645e-05, "loss": 0.0175, "step": 111920 }, { "epoch": 71.33843212237093, "grad_norm": 0.4498957097530365, "learning_rate": 7.407367599032153e-05, "loss": 0.0145, "step": 111930 }, { "epoch": 71.34480560866794, "grad_norm": 0.5411485433578491, "learning_rate": 7.406884517336873e-05, "loss": 0.0206, "step": 111940 }, { "epoch": 71.35117909496495, "grad_norm": 0.2691161334514618, "learning_rate": 7.406401406395673e-05, "loss": 0.0259, "step": 111950 }, { "epoch": 71.35755258126196, "grad_norm": 0.2796146273612976, "learning_rate": 7.405918266214426e-05, "loss": 0.0125, "step": 111960 }, { "epoch": 71.36392606755895, "grad_norm": 0.58985835313797, "learning_rate": 7.405435096798999e-05, "loss": 0.0196, "step": 111970 }, { "epoch": 71.37029955385596, "grad_norm": 0.46045950055122375, "learning_rate": 7.404951898155263e-05, "loss": 0.0093, "step": 111980 }, { "epoch": 71.37667304015297, "grad_norm": 0.30071139335632324, "learning_rate": 7.404468670289092e-05, "loss": 0.0086, "step": 111990 }, { "epoch": 71.38304652644997, "grad_norm": 0.4210141599178314, "learning_rate": 7.403985413206357e-05, "loss": 0.019, "step": 112000 }, { "epoch": 71.38942001274697, "grad_norm": 0.43543553352355957, "learning_rate": 7.403502126912929e-05, "loss": 0.0186, "step": 112010 }, { "epoch": 71.39579349904398, "grad_norm": 0.476346492767334, "learning_rate": 7.40301881141468e-05, "loss": 0.0164, "step": 112020 }, { "epoch": 71.40216698534098, "grad_norm": 0.1843661665916443, "learning_rate": 7.402535466717484e-05, "loss": 0.0161, "step": 112030 }, { "epoch": 71.40854047163799, "grad_norm": 0.5556214451789856, "learning_rate": 7.402052092827214e-05, "loss": 0.0203, "step": 112040 }, { "epoch": 71.41491395793498, "grad_norm": 0.3086717426776886, "learning_rate": 7.401568689749745e-05, "loss": 0.0147, "step": 112050 }, { "epoch": 71.421287444232, "grad_norm": 0.26795196533203125, "learning_rate": 7.401085257490947e-05, "loss": 0.0142, "step": 112060 }, { "epoch": 71.427660930529, "grad_norm": 0.22148117423057556, "learning_rate": 7.400601796056696e-05, "loss": 0.0161, "step": 112070 }, { "epoch": 71.43403441682601, "grad_norm": 0.649986743927002, "learning_rate": 7.400118305452867e-05, "loss": 0.0333, "step": 112080 }, { "epoch": 71.440407903123, "grad_norm": 0.40546172857284546, "learning_rate": 7.399634785685335e-05, "loss": 0.0192, "step": 112090 }, { "epoch": 71.44678138942001, "grad_norm": 0.4664226174354553, "learning_rate": 7.399151236759974e-05, "loss": 0.0154, "step": 112100 }, { "epoch": 71.45315487571702, "grad_norm": 0.2864699363708496, "learning_rate": 7.39866765868266e-05, "loss": 0.029, "step": 112110 }, { "epoch": 71.45952836201403, "grad_norm": 0.3281095325946808, "learning_rate": 7.398184051459271e-05, "loss": 0.0112, "step": 112120 }, { "epoch": 71.46590184831102, "grad_norm": 0.3200412690639496, "learning_rate": 7.397700415095679e-05, "loss": 0.0314, "step": 112130 }, { "epoch": 71.47227533460803, "grad_norm": 0.5504675507545471, "learning_rate": 7.397216749597765e-05, "loss": 0.0178, "step": 112140 }, { "epoch": 71.47864882090504, "grad_norm": 0.38527223467826843, "learning_rate": 7.396733054971403e-05, "loss": 0.0168, "step": 112150 }, { "epoch": 71.48502230720204, "grad_norm": 0.43689268827438354, "learning_rate": 7.396249331222472e-05, "loss": 0.0135, "step": 112160 }, { "epoch": 71.49139579349904, "grad_norm": 0.529055655002594, "learning_rate": 7.395765578356849e-05, "loss": 0.021, "step": 112170 }, { "epoch": 71.49776927979605, "grad_norm": 0.3452892005443573, "learning_rate": 7.395281796380414e-05, "loss": 0.0256, "step": 112180 }, { "epoch": 71.50414276609305, "grad_norm": 0.3483910858631134, "learning_rate": 7.39479798529904e-05, "loss": 0.0117, "step": 112190 }, { "epoch": 71.51051625239006, "grad_norm": 0.36640042066574097, "learning_rate": 7.394314145118613e-05, "loss": 0.0142, "step": 112200 }, { "epoch": 71.51688973868706, "grad_norm": 0.5262042284011841, "learning_rate": 7.393830275845007e-05, "loss": 0.0204, "step": 112210 }, { "epoch": 71.52326322498406, "grad_norm": 0.16639140248298645, "learning_rate": 7.393346377484103e-05, "loss": 0.017, "step": 112220 }, { "epoch": 71.52963671128107, "grad_norm": 0.3318454921245575, "learning_rate": 7.392862450041781e-05, "loss": 0.0164, "step": 112230 }, { "epoch": 71.53601019757808, "grad_norm": 0.28175681829452515, "learning_rate": 7.39237849352392e-05, "loss": 0.0151, "step": 112240 }, { "epoch": 71.54238368387507, "grad_norm": 0.5442802309989929, "learning_rate": 7.391894507936403e-05, "loss": 0.0161, "step": 112250 }, { "epoch": 71.54875717017208, "grad_norm": 0.3890407085418701, "learning_rate": 7.391410493285109e-05, "loss": 0.0215, "step": 112260 }, { "epoch": 71.55513065646909, "grad_norm": 0.25443580746650696, "learning_rate": 7.39092644957592e-05, "loss": 0.007, "step": 112270 }, { "epoch": 71.5615041427661, "grad_norm": 0.35709139704704285, "learning_rate": 7.390442376814716e-05, "loss": 0.017, "step": 112280 }, { "epoch": 71.56787762906309, "grad_norm": 0.31479161977767944, "learning_rate": 7.389958275007381e-05, "loss": 0.0156, "step": 112290 }, { "epoch": 71.5742511153601, "grad_norm": 0.35649216175079346, "learning_rate": 7.389474144159796e-05, "loss": 0.0175, "step": 112300 }, { "epoch": 71.5806246016571, "grad_norm": 0.43991512060165405, "learning_rate": 7.388989984277845e-05, "loss": 0.0118, "step": 112310 }, { "epoch": 71.58699808795411, "grad_norm": 0.28414320945739746, "learning_rate": 7.388505795367409e-05, "loss": 0.0221, "step": 112320 }, { "epoch": 71.59337157425111, "grad_norm": 0.37139230966567993, "learning_rate": 7.388021577434374e-05, "loss": 0.0212, "step": 112330 }, { "epoch": 71.59974506054812, "grad_norm": 0.48078304529190063, "learning_rate": 7.38753733048462e-05, "loss": 0.0123, "step": 112340 }, { "epoch": 71.60611854684512, "grad_norm": 0.4004465341567993, "learning_rate": 7.387053054524035e-05, "loss": 0.0256, "step": 112350 }, { "epoch": 71.61249203314213, "grad_norm": 0.47982630133628845, "learning_rate": 7.3865687495585e-05, "loss": 0.017, "step": 112360 }, { "epoch": 71.61886551943914, "grad_norm": 0.3756134808063507, "learning_rate": 7.386084415593904e-05, "loss": 0.0177, "step": 112370 }, { "epoch": 71.62523900573613, "grad_norm": 0.2892858386039734, "learning_rate": 7.385600052636127e-05, "loss": 0.0203, "step": 112380 }, { "epoch": 71.63161249203314, "grad_norm": 0.28233975172042847, "learning_rate": 7.385115660691058e-05, "loss": 0.0196, "step": 112390 }, { "epoch": 71.63798597833015, "grad_norm": 0.41853615641593933, "learning_rate": 7.384631239764581e-05, "loss": 0.0161, "step": 112400 }, { "epoch": 71.64435946462716, "grad_norm": 0.5323163866996765, "learning_rate": 7.384146789862585e-05, "loss": 0.0231, "step": 112410 }, { "epoch": 71.65073295092415, "grad_norm": 0.7301079034805298, "learning_rate": 7.383662310990952e-05, "loss": 0.0243, "step": 112420 }, { "epoch": 71.65710643722116, "grad_norm": 0.26666516065597534, "learning_rate": 7.383177803155573e-05, "loss": 0.0198, "step": 112430 }, { "epoch": 71.66347992351817, "grad_norm": 0.5283469557762146, "learning_rate": 7.382693266362334e-05, "loss": 0.0161, "step": 112440 }, { "epoch": 71.66985340981518, "grad_norm": 0.32770204544067383, "learning_rate": 7.382208700617121e-05, "loss": 0.0154, "step": 112450 }, { "epoch": 71.67622689611217, "grad_norm": 0.4272671937942505, "learning_rate": 7.381724105925824e-05, "loss": 0.0189, "step": 112460 }, { "epoch": 71.68260038240918, "grad_norm": 0.3193916380405426, "learning_rate": 7.381239482294329e-05, "loss": 0.0122, "step": 112470 }, { "epoch": 71.68897386870619, "grad_norm": 0.3643152117729187, "learning_rate": 7.380754829728528e-05, "loss": 0.0233, "step": 112480 }, { "epoch": 71.6953473550032, "grad_norm": 0.46829432249069214, "learning_rate": 7.380270148234306e-05, "loss": 0.0193, "step": 112490 }, { "epoch": 71.70172084130019, "grad_norm": 0.5239201784133911, "learning_rate": 7.379785437817556e-05, "loss": 0.0222, "step": 112500 }, { "epoch": 71.7080943275972, "grad_norm": 0.24278642237186432, "learning_rate": 7.379300698484164e-05, "loss": 0.0121, "step": 112510 }, { "epoch": 71.7144678138942, "grad_norm": 0.28867438435554504, "learning_rate": 7.378815930240024e-05, "loss": 0.0146, "step": 112520 }, { "epoch": 71.72084130019121, "grad_norm": 0.40474647283554077, "learning_rate": 7.378331133091024e-05, "loss": 0.0161, "step": 112530 }, { "epoch": 71.7272147864882, "grad_norm": 0.442464679479599, "learning_rate": 7.377846307043055e-05, "loss": 0.0132, "step": 112540 }, { "epoch": 71.73358827278521, "grad_norm": 0.19204314053058624, "learning_rate": 7.377361452102009e-05, "loss": 0.0105, "step": 112550 }, { "epoch": 71.73996175908222, "grad_norm": 0.3332919180393219, "learning_rate": 7.376876568273776e-05, "loss": 0.0104, "step": 112560 }, { "epoch": 71.74633524537923, "grad_norm": 0.44868332147598267, "learning_rate": 7.376391655564249e-05, "loss": 0.018, "step": 112570 }, { "epoch": 71.75270873167622, "grad_norm": 0.3221135437488556, "learning_rate": 7.375906713979318e-05, "loss": 0.0108, "step": 112580 }, { "epoch": 71.75908221797323, "grad_norm": 0.7078045606613159, "learning_rate": 7.37542174352488e-05, "loss": 0.0109, "step": 112590 }, { "epoch": 71.76545570427024, "grad_norm": 0.43876132369041443, "learning_rate": 7.374936744206824e-05, "loss": 0.0194, "step": 112600 }, { "epoch": 71.77182919056725, "grad_norm": 0.566847026348114, "learning_rate": 7.374451716031043e-05, "loss": 0.0154, "step": 112610 }, { "epoch": 71.77820267686424, "grad_norm": 0.4566158950328827, "learning_rate": 7.373966659003433e-05, "loss": 0.0121, "step": 112620 }, { "epoch": 71.78457616316125, "grad_norm": 0.2289484441280365, "learning_rate": 7.373481573129886e-05, "loss": 0.0106, "step": 112630 }, { "epoch": 71.79094964945826, "grad_norm": 0.14812730252742767, "learning_rate": 7.372996458416298e-05, "loss": 0.0112, "step": 112640 }, { "epoch": 71.79732313575526, "grad_norm": 0.2204786241054535, "learning_rate": 7.372511314868561e-05, "loss": 0.0196, "step": 112650 }, { "epoch": 71.80369662205226, "grad_norm": 0.4022834300994873, "learning_rate": 7.372026142492573e-05, "loss": 0.013, "step": 112660 }, { "epoch": 71.81007010834927, "grad_norm": 0.328564316034317, "learning_rate": 7.371540941294226e-05, "loss": 0.0137, "step": 112670 }, { "epoch": 71.81644359464627, "grad_norm": 0.9040935635566711, "learning_rate": 7.37105571127942e-05, "loss": 0.0193, "step": 112680 }, { "epoch": 71.82281708094328, "grad_norm": 0.6369282007217407, "learning_rate": 7.370570452454047e-05, "loss": 0.0218, "step": 112690 }, { "epoch": 71.82919056724027, "grad_norm": 0.35568514466285706, "learning_rate": 7.370085164824004e-05, "loss": 0.0149, "step": 112700 }, { "epoch": 71.83556405353728, "grad_norm": 0.2949157953262329, "learning_rate": 7.36959984839519e-05, "loss": 0.0124, "step": 112710 }, { "epoch": 71.84193753983429, "grad_norm": 0.46655571460723877, "learning_rate": 7.3691145031735e-05, "loss": 0.0124, "step": 112720 }, { "epoch": 71.8483110261313, "grad_norm": 0.44947993755340576, "learning_rate": 7.368629129164833e-05, "loss": 0.0141, "step": 112730 }, { "epoch": 71.85468451242829, "grad_norm": 0.38359084725379944, "learning_rate": 7.368143726375083e-05, "loss": 0.0198, "step": 112740 }, { "epoch": 71.8610579987253, "grad_norm": 0.3058016896247864, "learning_rate": 7.367658294810154e-05, "loss": 0.0137, "step": 112750 }, { "epoch": 71.86743148502231, "grad_norm": 0.29920321702957153, "learning_rate": 7.367172834475939e-05, "loss": 0.0132, "step": 112760 }, { "epoch": 71.87380497131932, "grad_norm": 0.4284398555755615, "learning_rate": 7.36668734537834e-05, "loss": 0.0132, "step": 112770 }, { "epoch": 71.88017845761631, "grad_norm": 0.6181581616401672, "learning_rate": 7.366201827523256e-05, "loss": 0.0176, "step": 112780 }, { "epoch": 71.88655194391332, "grad_norm": 0.265926718711853, "learning_rate": 7.365716280916584e-05, "loss": 0.0105, "step": 112790 }, { "epoch": 71.89292543021033, "grad_norm": 0.472455769777298, "learning_rate": 7.365230705564227e-05, "loss": 0.0217, "step": 112800 }, { "epoch": 71.89929891650733, "grad_norm": 0.6640032529830933, "learning_rate": 7.364745101472083e-05, "loss": 0.0191, "step": 112810 }, { "epoch": 71.90567240280433, "grad_norm": 0.36768969893455505, "learning_rate": 7.364259468646053e-05, "loss": 0.0156, "step": 112820 }, { "epoch": 71.91204588910134, "grad_norm": 0.4640871286392212, "learning_rate": 7.363773807092038e-05, "loss": 0.0232, "step": 112830 }, { "epoch": 71.91841937539834, "grad_norm": 0.41178372502326965, "learning_rate": 7.363288116815941e-05, "loss": 0.0185, "step": 112840 }, { "epoch": 71.92479286169535, "grad_norm": 0.36267292499542236, "learning_rate": 7.362802397823663e-05, "loss": 0.0121, "step": 112850 }, { "epoch": 71.93116634799235, "grad_norm": 0.2497832328081131, "learning_rate": 7.362316650121102e-05, "loss": 0.0166, "step": 112860 }, { "epoch": 71.93753983428935, "grad_norm": 0.5851818323135376, "learning_rate": 7.361830873714165e-05, "loss": 0.0234, "step": 112870 }, { "epoch": 71.94391332058636, "grad_norm": 0.20047226548194885, "learning_rate": 7.361345068608752e-05, "loss": 0.0122, "step": 112880 }, { "epoch": 71.95028680688337, "grad_norm": 0.5246102213859558, "learning_rate": 7.360859234810767e-05, "loss": 0.0175, "step": 112890 }, { "epoch": 71.95666029318036, "grad_norm": 0.43953388929367065, "learning_rate": 7.360373372326114e-05, "loss": 0.0235, "step": 112900 }, { "epoch": 71.96303377947737, "grad_norm": 0.305100679397583, "learning_rate": 7.359887481160697e-05, "loss": 0.0265, "step": 112910 }, { "epoch": 71.96940726577438, "grad_norm": 0.6466211080551147, "learning_rate": 7.359401561320417e-05, "loss": 0.0159, "step": 112920 }, { "epoch": 71.97578075207139, "grad_norm": 0.7118947505950928, "learning_rate": 7.358915612811181e-05, "loss": 0.0292, "step": 112930 }, { "epoch": 71.98215423836838, "grad_norm": 0.23722565174102783, "learning_rate": 7.358429635638894e-05, "loss": 0.0227, "step": 112940 }, { "epoch": 71.98852772466539, "grad_norm": 0.3455184996128082, "learning_rate": 7.357943629809458e-05, "loss": 0.0166, "step": 112950 }, { "epoch": 71.9949012109624, "grad_norm": 0.40463805198669434, "learning_rate": 7.357457595328783e-05, "loss": 0.0155, "step": 112960 }, { "epoch": 72.0012746972594, "grad_norm": 0.5718831419944763, "learning_rate": 7.356971532202771e-05, "loss": 0.0149, "step": 112970 }, { "epoch": 72.00764818355641, "grad_norm": 0.3989647626876831, "learning_rate": 7.35648544043733e-05, "loss": 0.0307, "step": 112980 }, { "epoch": 72.0140216698534, "grad_norm": 0.702725350856781, "learning_rate": 7.355999320038366e-05, "loss": 0.0203, "step": 112990 }, { "epoch": 72.02039515615041, "grad_norm": 0.39064455032348633, "learning_rate": 7.355513171011787e-05, "loss": 0.0207, "step": 113000 }, { "epoch": 72.02676864244742, "grad_norm": 0.3058459758758545, "learning_rate": 7.355026993363496e-05, "loss": 0.0252, "step": 113010 }, { "epoch": 72.03314212874443, "grad_norm": 0.3144052028656006, "learning_rate": 7.354540787099406e-05, "loss": 0.0142, "step": 113020 }, { "epoch": 72.03951561504142, "grad_norm": 0.6099256277084351, "learning_rate": 7.354054552225422e-05, "loss": 0.02, "step": 113030 }, { "epoch": 72.04588910133843, "grad_norm": 0.46430057287216187, "learning_rate": 7.353568288747452e-05, "loss": 0.0183, "step": 113040 }, { "epoch": 72.05226258763544, "grad_norm": 0.2648930251598358, "learning_rate": 7.353081996671404e-05, "loss": 0.0229, "step": 113050 }, { "epoch": 72.05863607393245, "grad_norm": 0.23276522755622864, "learning_rate": 7.352595676003192e-05, "loss": 0.0132, "step": 113060 }, { "epoch": 72.06500956022944, "grad_norm": 0.4818512797355652, "learning_rate": 7.352109326748717e-05, "loss": 0.0209, "step": 113070 }, { "epoch": 72.07138304652645, "grad_norm": 0.575624942779541, "learning_rate": 7.351622948913896e-05, "loss": 0.0164, "step": 113080 }, { "epoch": 72.07775653282346, "grad_norm": 0.5992916822433472, "learning_rate": 7.351136542504634e-05, "loss": 0.0194, "step": 113090 }, { "epoch": 72.08413001912047, "grad_norm": 0.7430914640426636, "learning_rate": 7.350650107526845e-05, "loss": 0.0183, "step": 113100 }, { "epoch": 72.09050350541746, "grad_norm": 0.36094510555267334, "learning_rate": 7.350163643986435e-05, "loss": 0.0139, "step": 113110 }, { "epoch": 72.09687699171447, "grad_norm": 0.27551546692848206, "learning_rate": 7.349677151889321e-05, "loss": 0.0236, "step": 113120 }, { "epoch": 72.10325047801147, "grad_norm": 0.37647610902786255, "learning_rate": 7.349190631241408e-05, "loss": 0.0115, "step": 113130 }, { "epoch": 72.10962396430848, "grad_norm": 0.3141384422779083, "learning_rate": 7.348704082048613e-05, "loss": 0.0127, "step": 113140 }, { "epoch": 72.11599745060548, "grad_norm": 0.3731916844844818, "learning_rate": 7.348217504316845e-05, "loss": 0.021, "step": 113150 }, { "epoch": 72.12237093690248, "grad_norm": 0.5014616250991821, "learning_rate": 7.347730898052018e-05, "loss": 0.0146, "step": 113160 }, { "epoch": 72.12874442319949, "grad_norm": 0.7035441994667053, "learning_rate": 7.347244263260043e-05, "loss": 0.0181, "step": 113170 }, { "epoch": 72.1351179094965, "grad_norm": 0.22204901278018951, "learning_rate": 7.346757599946834e-05, "loss": 0.014, "step": 113180 }, { "epoch": 72.1414913957935, "grad_norm": 0.3503003716468811, "learning_rate": 7.346270908118306e-05, "loss": 0.0164, "step": 113190 }, { "epoch": 72.1478648820905, "grad_norm": 0.5039528012275696, "learning_rate": 7.34578418778037e-05, "loss": 0.0154, "step": 113200 }, { "epoch": 72.15423836838751, "grad_norm": 0.3733975887298584, "learning_rate": 7.345297438938942e-05, "loss": 0.0085, "step": 113210 }, { "epoch": 72.16061185468452, "grad_norm": 0.34849977493286133, "learning_rate": 7.344810661599935e-05, "loss": 0.0216, "step": 113220 }, { "epoch": 72.16698534098151, "grad_norm": 0.5941341519355774, "learning_rate": 7.344323855769265e-05, "loss": 0.0192, "step": 113230 }, { "epoch": 72.17335882727852, "grad_norm": 0.44508278369903564, "learning_rate": 7.343837021452846e-05, "loss": 0.0155, "step": 113240 }, { "epoch": 72.17973231357553, "grad_norm": 0.43203622102737427, "learning_rate": 7.343350158656596e-05, "loss": 0.0196, "step": 113250 }, { "epoch": 72.18610579987254, "grad_norm": 0.21516522765159607, "learning_rate": 7.342863267386428e-05, "loss": 0.0116, "step": 113260 }, { "epoch": 72.19247928616953, "grad_norm": 0.42442163825035095, "learning_rate": 7.342376347648259e-05, "loss": 0.0211, "step": 113270 }, { "epoch": 72.19885277246654, "grad_norm": 0.5309235453605652, "learning_rate": 7.341889399448006e-05, "loss": 0.025, "step": 113280 }, { "epoch": 72.20522625876355, "grad_norm": 0.19177095592021942, "learning_rate": 7.341402422791586e-05, "loss": 0.0135, "step": 113290 }, { "epoch": 72.21159974506055, "grad_norm": 0.35628488659858704, "learning_rate": 7.340915417684915e-05, "loss": 0.0123, "step": 113300 }, { "epoch": 72.21797323135755, "grad_norm": 0.3526313602924347, "learning_rate": 7.340428384133912e-05, "loss": 0.0162, "step": 113310 }, { "epoch": 72.22434671765456, "grad_norm": 0.2816762924194336, "learning_rate": 7.339941322144494e-05, "loss": 0.0237, "step": 113320 }, { "epoch": 72.23072020395156, "grad_norm": 0.32740554213523865, "learning_rate": 7.339454231722582e-05, "loss": 0.0412, "step": 113330 }, { "epoch": 72.23709369024857, "grad_norm": 0.24438093602657318, "learning_rate": 7.338967112874089e-05, "loss": 0.0158, "step": 113340 }, { "epoch": 72.24346717654556, "grad_norm": 0.5879038572311401, "learning_rate": 7.33847996560494e-05, "loss": 0.0132, "step": 113350 }, { "epoch": 72.24984066284257, "grad_norm": 0.21791058778762817, "learning_rate": 7.337992789921049e-05, "loss": 0.0125, "step": 113360 }, { "epoch": 72.25621414913958, "grad_norm": 0.42603424191474915, "learning_rate": 7.33750558582834e-05, "loss": 0.0183, "step": 113370 }, { "epoch": 72.26258763543659, "grad_norm": 0.34699559211730957, "learning_rate": 7.33701835333273e-05, "loss": 0.016, "step": 113380 }, { "epoch": 72.26896112173358, "grad_norm": 0.41082480549812317, "learning_rate": 7.336531092440142e-05, "loss": 0.0139, "step": 113390 }, { "epoch": 72.27533460803059, "grad_norm": 0.22218084335327148, "learning_rate": 7.336043803156493e-05, "loss": 0.0199, "step": 113400 }, { "epoch": 72.2817080943276, "grad_norm": 0.31930333375930786, "learning_rate": 7.335556485487708e-05, "loss": 0.0166, "step": 113410 }, { "epoch": 72.2880815806246, "grad_norm": 0.36845988035202026, "learning_rate": 7.335069139439706e-05, "loss": 0.0234, "step": 113420 }, { "epoch": 72.2944550669216, "grad_norm": 0.39159321784973145, "learning_rate": 7.334581765018409e-05, "loss": 0.0167, "step": 113430 }, { "epoch": 72.30082855321861, "grad_norm": 0.45930665731430054, "learning_rate": 7.33409436222974e-05, "loss": 0.0226, "step": 113440 }, { "epoch": 72.30720203951562, "grad_norm": 0.34750086069107056, "learning_rate": 7.33360693107962e-05, "loss": 0.02, "step": 113450 }, { "epoch": 72.31357552581262, "grad_norm": 0.4266999661922455, "learning_rate": 7.333119471573974e-05, "loss": 0.0107, "step": 113460 }, { "epoch": 72.31994901210962, "grad_norm": 0.34591278433799744, "learning_rate": 7.332631983718722e-05, "loss": 0.0224, "step": 113470 }, { "epoch": 72.32632249840663, "grad_norm": 0.2823264002799988, "learning_rate": 7.33214446751979e-05, "loss": 0.023, "step": 113480 }, { "epoch": 72.33269598470363, "grad_norm": 0.40549349784851074, "learning_rate": 7.3316569229831e-05, "loss": 0.0139, "step": 113490 }, { "epoch": 72.33906947100064, "grad_norm": 0.5716128349304199, "learning_rate": 7.331169350114577e-05, "loss": 0.016, "step": 113500 }, { "epoch": 72.34544295729764, "grad_norm": 0.22423169016838074, "learning_rate": 7.330681748920146e-05, "loss": 0.0146, "step": 113510 }, { "epoch": 72.35181644359464, "grad_norm": 0.2838788628578186, "learning_rate": 7.330194119405731e-05, "loss": 0.0265, "step": 113520 }, { "epoch": 72.35818992989165, "grad_norm": 0.25387415289878845, "learning_rate": 7.329706461577257e-05, "loss": 0.0195, "step": 113530 }, { "epoch": 72.36456341618866, "grad_norm": 0.5474233627319336, "learning_rate": 7.32921877544065e-05, "loss": 0.0144, "step": 113540 }, { "epoch": 72.37093690248565, "grad_norm": 0.22595922648906708, "learning_rate": 7.328731061001837e-05, "loss": 0.0119, "step": 113550 }, { "epoch": 72.37731038878266, "grad_norm": 0.3871650695800781, "learning_rate": 7.328243318266741e-05, "loss": 0.0221, "step": 113560 }, { "epoch": 72.38368387507967, "grad_norm": 0.2002009004354477, "learning_rate": 7.327755547241291e-05, "loss": 0.0097, "step": 113570 }, { "epoch": 72.39005736137668, "grad_norm": 0.5796409845352173, "learning_rate": 7.327267747931414e-05, "loss": 0.0169, "step": 113580 }, { "epoch": 72.39643084767367, "grad_norm": 0.3281729519367218, "learning_rate": 7.326779920343035e-05, "loss": 0.0096, "step": 113590 }, { "epoch": 72.40280433397068, "grad_norm": 0.2723643481731415, "learning_rate": 7.326292064482085e-05, "loss": 0.0252, "step": 113600 }, { "epoch": 72.40917782026769, "grad_norm": 0.33537307381629944, "learning_rate": 7.32580418035449e-05, "loss": 0.0184, "step": 113610 }, { "epoch": 72.4155513065647, "grad_norm": 0.43124207854270935, "learning_rate": 7.325316267966177e-05, "loss": 0.0096, "step": 113620 }, { "epoch": 72.4219247928617, "grad_norm": 0.3130299150943756, "learning_rate": 7.324828327323076e-05, "loss": 0.0198, "step": 113630 }, { "epoch": 72.4282982791587, "grad_norm": 0.4134792983531952, "learning_rate": 7.324340358431117e-05, "loss": 0.017, "step": 113640 }, { "epoch": 72.4346717654557, "grad_norm": 0.5460556149482727, "learning_rate": 7.323852361296227e-05, "loss": 0.0208, "step": 113650 }, { "epoch": 72.44104525175271, "grad_norm": 0.5248019695281982, "learning_rate": 7.323364335924337e-05, "loss": 0.0241, "step": 113660 }, { "epoch": 72.44741873804972, "grad_norm": 0.33820849657058716, "learning_rate": 7.322876282321379e-05, "loss": 0.0213, "step": 113670 }, { "epoch": 72.45379222434671, "grad_norm": 0.319362074136734, "learning_rate": 7.322388200493277e-05, "loss": 0.0163, "step": 113680 }, { "epoch": 72.46016571064372, "grad_norm": 0.3042007088661194, "learning_rate": 7.321900090445967e-05, "loss": 0.0276, "step": 113690 }, { "epoch": 72.46653919694073, "grad_norm": 0.5222786068916321, "learning_rate": 7.321411952185379e-05, "loss": 0.0106, "step": 113700 }, { "epoch": 72.47291268323774, "grad_norm": 0.44042277336120605, "learning_rate": 7.320923785717445e-05, "loss": 0.0202, "step": 113710 }, { "epoch": 72.47928616953473, "grad_norm": 0.2913137972354889, "learning_rate": 7.320435591048094e-05, "loss": 0.018, "step": 113720 }, { "epoch": 72.48565965583174, "grad_norm": 0.31539174914360046, "learning_rate": 7.319947368183262e-05, "loss": 0.0168, "step": 113730 }, { "epoch": 72.49203314212875, "grad_norm": 0.23334136605262756, "learning_rate": 7.319459117128877e-05, "loss": 0.0202, "step": 113740 }, { "epoch": 72.49840662842576, "grad_norm": 0.39079874753952026, "learning_rate": 7.318970837890875e-05, "loss": 0.0145, "step": 113750 }, { "epoch": 72.50478011472275, "grad_norm": 0.46653157472610474, "learning_rate": 7.318482530475186e-05, "loss": 0.0172, "step": 113760 }, { "epoch": 72.51115360101976, "grad_norm": 0.2498743236064911, "learning_rate": 7.317994194887747e-05, "loss": 0.0149, "step": 113770 }, { "epoch": 72.51752708731676, "grad_norm": 0.5402017831802368, "learning_rate": 7.317505831134489e-05, "loss": 0.0386, "step": 113780 }, { "epoch": 72.52390057361377, "grad_norm": 0.37883448600769043, "learning_rate": 7.317017439221347e-05, "loss": 0.0115, "step": 113790 }, { "epoch": 72.53027405991077, "grad_norm": 0.453969806432724, "learning_rate": 7.316529019154257e-05, "loss": 0.0206, "step": 113800 }, { "epoch": 72.53664754620777, "grad_norm": 0.34084638953208923, "learning_rate": 7.316040570939151e-05, "loss": 0.0115, "step": 113810 }, { "epoch": 72.54302103250478, "grad_norm": 0.4473154544830322, "learning_rate": 7.315552094581966e-05, "loss": 0.0113, "step": 113820 }, { "epoch": 72.54939451880179, "grad_norm": 0.2327394187450409, "learning_rate": 7.315063590088637e-05, "loss": 0.02, "step": 113830 }, { "epoch": 72.55576800509878, "grad_norm": 0.256333589553833, "learning_rate": 7.314575057465098e-05, "loss": 0.0147, "step": 113840 }, { "epoch": 72.56214149139579, "grad_norm": 0.4698444604873657, "learning_rate": 7.314086496717289e-05, "loss": 0.0259, "step": 113850 }, { "epoch": 72.5685149776928, "grad_norm": 0.32647955417633057, "learning_rate": 7.313597907851142e-05, "loss": 0.0122, "step": 113860 }, { "epoch": 72.57488846398981, "grad_norm": 0.48328351974487305, "learning_rate": 7.313109290872598e-05, "loss": 0.0309, "step": 113870 }, { "epoch": 72.5812619502868, "grad_norm": 0.2738121449947357, "learning_rate": 7.31262064578759e-05, "loss": 0.0177, "step": 113880 }, { "epoch": 72.58763543658381, "grad_norm": 0.3332400619983673, "learning_rate": 7.312131972602058e-05, "loss": 0.0139, "step": 113890 }, { "epoch": 72.59400892288082, "grad_norm": 0.5476122498512268, "learning_rate": 7.311643271321942e-05, "loss": 0.0143, "step": 113900 }, { "epoch": 72.60038240917783, "grad_norm": 0.4550575911998749, "learning_rate": 7.311154541953176e-05, "loss": 0.0199, "step": 113910 }, { "epoch": 72.60675589547482, "grad_norm": 0.25108546018600464, "learning_rate": 7.3106657845017e-05, "loss": 0.015, "step": 113920 }, { "epoch": 72.61312938177183, "grad_norm": 0.7585155963897705, "learning_rate": 7.310176998973453e-05, "loss": 0.0281, "step": 113930 }, { "epoch": 72.61950286806884, "grad_norm": 0.30048468708992004, "learning_rate": 7.309688185374376e-05, "loss": 0.0115, "step": 113940 }, { "epoch": 72.62587635436584, "grad_norm": 0.6492212414741516, "learning_rate": 7.309199343710405e-05, "loss": 0.0243, "step": 113950 }, { "epoch": 72.63224984066284, "grad_norm": 0.5122234225273132, "learning_rate": 7.308710473987483e-05, "loss": 0.0188, "step": 113960 }, { "epoch": 72.63862332695984, "grad_norm": 0.609464168548584, "learning_rate": 7.308221576211548e-05, "loss": 0.0226, "step": 113970 }, { "epoch": 72.64499681325685, "grad_norm": 0.30804866552352905, "learning_rate": 7.307732650388542e-05, "loss": 0.0148, "step": 113980 }, { "epoch": 72.65137029955386, "grad_norm": 0.4142485558986664, "learning_rate": 7.307243696524405e-05, "loss": 0.0182, "step": 113990 }, { "epoch": 72.65774378585085, "grad_norm": 0.30982667207717896, "learning_rate": 7.306754714625078e-05, "loss": 0.0175, "step": 114000 }, { "epoch": 72.66411727214786, "grad_norm": 0.3742220401763916, "learning_rate": 7.306265704696504e-05, "loss": 0.0194, "step": 114010 }, { "epoch": 72.67049075844487, "grad_norm": 0.6224552392959595, "learning_rate": 7.305776666744625e-05, "loss": 0.0201, "step": 114020 }, { "epoch": 72.67686424474188, "grad_norm": 0.35640913248062134, "learning_rate": 7.305287600775382e-05, "loss": 0.0174, "step": 114030 }, { "epoch": 72.68323773103887, "grad_norm": 0.2707843482494354, "learning_rate": 7.304798506794718e-05, "loss": 0.0103, "step": 114040 }, { "epoch": 72.68961121733588, "grad_norm": 0.4198243021965027, "learning_rate": 7.304309384808577e-05, "loss": 0.0125, "step": 114050 }, { "epoch": 72.69598470363289, "grad_norm": 0.3541308045387268, "learning_rate": 7.303820234822901e-05, "loss": 0.0195, "step": 114060 }, { "epoch": 72.7023581899299, "grad_norm": 0.38320985436439514, "learning_rate": 7.303331056843631e-05, "loss": 0.0182, "step": 114070 }, { "epoch": 72.70873167622689, "grad_norm": 0.5441397428512573, "learning_rate": 7.302841850876718e-05, "loss": 0.0191, "step": 114080 }, { "epoch": 72.7151051625239, "grad_norm": 0.40555664896965027, "learning_rate": 7.302352616928099e-05, "loss": 0.0204, "step": 114090 }, { "epoch": 72.7214786488209, "grad_norm": 0.270123690366745, "learning_rate": 7.301863355003725e-05, "loss": 0.017, "step": 114100 }, { "epoch": 72.72785213511791, "grad_norm": 0.401199609041214, "learning_rate": 7.301374065109535e-05, "loss": 0.0198, "step": 114110 }, { "epoch": 72.73422562141491, "grad_norm": 0.30923980474472046, "learning_rate": 7.30088474725148e-05, "loss": 0.0189, "step": 114120 }, { "epoch": 72.74059910771192, "grad_norm": 0.317912220954895, "learning_rate": 7.300395401435501e-05, "loss": 0.0146, "step": 114130 }, { "epoch": 72.74697259400892, "grad_norm": 0.3823492228984833, "learning_rate": 7.299906027667547e-05, "loss": 0.0138, "step": 114140 }, { "epoch": 72.75334608030593, "grad_norm": 0.6895157694816589, "learning_rate": 7.299416625953564e-05, "loss": 0.0189, "step": 114150 }, { "epoch": 72.75971956660293, "grad_norm": 0.6077188849449158, "learning_rate": 7.298927196299496e-05, "loss": 0.0215, "step": 114160 }, { "epoch": 72.76609305289993, "grad_norm": 0.447308212518692, "learning_rate": 7.298437738711292e-05, "loss": 0.0267, "step": 114170 }, { "epoch": 72.77246653919694, "grad_norm": 0.3975064754486084, "learning_rate": 7.297948253194899e-05, "loss": 0.0146, "step": 114180 }, { "epoch": 72.77884002549395, "grad_norm": 0.24420438706874847, "learning_rate": 7.297458739756267e-05, "loss": 0.0268, "step": 114190 }, { "epoch": 72.78521351179094, "grad_norm": 0.3586350083351135, "learning_rate": 7.296969198401342e-05, "loss": 0.0208, "step": 114200 }, { "epoch": 72.79158699808795, "grad_norm": 0.805204451084137, "learning_rate": 7.296479629136073e-05, "loss": 0.0238, "step": 114210 }, { "epoch": 72.79796048438496, "grad_norm": 0.48016560077667236, "learning_rate": 7.295990031966407e-05, "loss": 0.0191, "step": 114220 }, { "epoch": 72.80433397068197, "grad_norm": 0.4322308897972107, "learning_rate": 7.295500406898295e-05, "loss": 0.021, "step": 114230 }, { "epoch": 72.81070745697897, "grad_norm": 0.28259584307670593, "learning_rate": 7.295010753937686e-05, "loss": 0.0124, "step": 114240 }, { "epoch": 72.81708094327597, "grad_norm": 0.2607811391353607, "learning_rate": 7.294521073090527e-05, "loss": 0.0156, "step": 114250 }, { "epoch": 72.82345442957298, "grad_norm": 0.43748167157173157, "learning_rate": 7.294031364362773e-05, "loss": 0.0181, "step": 114260 }, { "epoch": 72.82982791586998, "grad_norm": 0.21345020830631256, "learning_rate": 7.29354162776037e-05, "loss": 0.017, "step": 114270 }, { "epoch": 72.83620140216699, "grad_norm": 0.5328529477119446, "learning_rate": 7.293051863289274e-05, "loss": 0.0157, "step": 114280 }, { "epoch": 72.84257488846399, "grad_norm": 0.9353928565979004, "learning_rate": 7.29256207095543e-05, "loss": 0.0181, "step": 114290 }, { "epoch": 72.848948374761, "grad_norm": 0.3485811650753021, "learning_rate": 7.292072250764795e-05, "loss": 0.0127, "step": 114300 }, { "epoch": 72.855321861058, "grad_norm": 0.39055851101875305, "learning_rate": 7.291582402723316e-05, "loss": 0.0103, "step": 114310 }, { "epoch": 72.86169534735501, "grad_norm": 0.4384983777999878, "learning_rate": 7.291092526836947e-05, "loss": 0.0194, "step": 114320 }, { "epoch": 72.868068833652, "grad_norm": 0.6139965057373047, "learning_rate": 7.290602623111642e-05, "loss": 0.0297, "step": 114330 }, { "epoch": 72.87444231994901, "grad_norm": 0.5330699682235718, "learning_rate": 7.290112691553351e-05, "loss": 0.0184, "step": 114340 }, { "epoch": 72.88081580624602, "grad_norm": 0.5116077065467834, "learning_rate": 7.28962273216803e-05, "loss": 0.0124, "step": 114350 }, { "epoch": 72.88718929254303, "grad_norm": 0.5688324570655823, "learning_rate": 7.28913274496163e-05, "loss": 0.0175, "step": 114360 }, { "epoch": 72.89356277884002, "grad_norm": 0.3196817636489868, "learning_rate": 7.288642729940106e-05, "loss": 0.0201, "step": 114370 }, { "epoch": 72.89993626513703, "grad_norm": 0.5364440679550171, "learning_rate": 7.288152687109413e-05, "loss": 0.0169, "step": 114380 }, { "epoch": 72.90630975143404, "grad_norm": 0.43039670586586, "learning_rate": 7.287662616475504e-05, "loss": 0.0181, "step": 114390 }, { "epoch": 72.91268323773104, "grad_norm": 0.220923513174057, "learning_rate": 7.287172518044334e-05, "loss": 0.0312, "step": 114400 }, { "epoch": 72.91905672402804, "grad_norm": 0.3401090204715729, "learning_rate": 7.286682391821858e-05, "loss": 0.0148, "step": 114410 }, { "epoch": 72.92543021032505, "grad_norm": 0.7129428386688232, "learning_rate": 7.286192237814033e-05, "loss": 0.0256, "step": 114420 }, { "epoch": 72.93180369662205, "grad_norm": 1.0186797380447388, "learning_rate": 7.285702056026813e-05, "loss": 0.0231, "step": 114430 }, { "epoch": 72.93817718291906, "grad_norm": 0.5374343991279602, "learning_rate": 7.285211846466154e-05, "loss": 0.0209, "step": 114440 }, { "epoch": 72.94455066921606, "grad_norm": 0.5231232047080994, "learning_rate": 7.284721609138014e-05, "loss": 0.02, "step": 114450 }, { "epoch": 72.95092415551306, "grad_norm": 0.3980078399181366, "learning_rate": 7.284231344048351e-05, "loss": 0.0133, "step": 114460 }, { "epoch": 72.95729764181007, "grad_norm": 0.6895831227302551, "learning_rate": 7.28374105120312e-05, "loss": 0.0209, "step": 114470 }, { "epoch": 72.96367112810708, "grad_norm": 0.47740039229393005, "learning_rate": 7.283250730608277e-05, "loss": 0.0272, "step": 114480 }, { "epoch": 72.97004461440407, "grad_norm": 0.41901305317878723, "learning_rate": 7.282760382269783e-05, "loss": 0.0249, "step": 114490 }, { "epoch": 72.97641810070108, "grad_norm": 0.3793869614601135, "learning_rate": 7.282270006193595e-05, "loss": 0.0242, "step": 114500 }, { "epoch": 72.98279158699809, "grad_norm": 0.27045634388923645, "learning_rate": 7.281779602385673e-05, "loss": 0.0139, "step": 114510 }, { "epoch": 72.9891650732951, "grad_norm": 0.29091525077819824, "learning_rate": 7.281289170851972e-05, "loss": 0.0179, "step": 114520 }, { "epoch": 72.99553855959209, "grad_norm": 0.27438822388648987, "learning_rate": 7.280798711598456e-05, "loss": 0.0174, "step": 114530 }, { "epoch": 73.0019120458891, "grad_norm": 0.4468340277671814, "learning_rate": 7.280308224631081e-05, "loss": 0.0235, "step": 114540 }, { "epoch": 73.00828553218611, "grad_norm": 0.4306821823120117, "learning_rate": 7.279817709955809e-05, "loss": 0.0182, "step": 114550 }, { "epoch": 73.01465901848312, "grad_norm": 0.3407873809337616, "learning_rate": 7.279327167578599e-05, "loss": 0.0157, "step": 114560 }, { "epoch": 73.02103250478011, "grad_norm": 0.4185697138309479, "learning_rate": 7.27883659750541e-05, "loss": 0.0151, "step": 114570 }, { "epoch": 73.02740599107712, "grad_norm": 0.37426844239234924, "learning_rate": 7.278345999742208e-05, "loss": 0.0252, "step": 114580 }, { "epoch": 73.03377947737413, "grad_norm": 0.8414224982261658, "learning_rate": 7.277855374294948e-05, "loss": 0.0253, "step": 114590 }, { "epoch": 73.04015296367113, "grad_norm": 0.4614699184894562, "learning_rate": 7.277364721169597e-05, "loss": 0.0172, "step": 114600 }, { "epoch": 73.04652644996813, "grad_norm": 0.697331428527832, "learning_rate": 7.276874040372113e-05, "loss": 0.0165, "step": 114610 }, { "epoch": 73.05289993626513, "grad_norm": 0.2998666763305664, "learning_rate": 7.27638333190846e-05, "loss": 0.0178, "step": 114620 }, { "epoch": 73.05927342256214, "grad_norm": 0.24222257733345032, "learning_rate": 7.275892595784601e-05, "loss": 0.0128, "step": 114630 }, { "epoch": 73.06564690885915, "grad_norm": 0.38814419507980347, "learning_rate": 7.275401832006496e-05, "loss": 0.0122, "step": 114640 }, { "epoch": 73.07202039515614, "grad_norm": 0.2249758094549179, "learning_rate": 7.274911040580114e-05, "loss": 0.0099, "step": 114650 }, { "epoch": 73.07839388145315, "grad_norm": 0.465024471282959, "learning_rate": 7.274420221511411e-05, "loss": 0.0221, "step": 114660 }, { "epoch": 73.08476736775016, "grad_norm": 0.5032816529273987, "learning_rate": 7.273929374806358e-05, "loss": 0.0132, "step": 114670 }, { "epoch": 73.09114085404717, "grad_norm": 0.3518093526363373, "learning_rate": 7.273438500470914e-05, "loss": 0.0231, "step": 114680 }, { "epoch": 73.09751434034416, "grad_norm": 0.18249323964118958, "learning_rate": 7.272947598511047e-05, "loss": 0.0197, "step": 114690 }, { "epoch": 73.10388782664117, "grad_norm": 0.2743316888809204, "learning_rate": 7.272456668932721e-05, "loss": 0.0178, "step": 114700 }, { "epoch": 73.11026131293818, "grad_norm": 0.48399490118026733, "learning_rate": 7.2719657117419e-05, "loss": 0.0178, "step": 114710 }, { "epoch": 73.11663479923519, "grad_norm": 0.3588631749153137, "learning_rate": 7.271474726944551e-05, "loss": 0.0216, "step": 114720 }, { "epoch": 73.12300828553218, "grad_norm": 0.3495731055736542, "learning_rate": 7.270983714546639e-05, "loss": 0.0144, "step": 114730 }, { "epoch": 73.12938177182919, "grad_norm": 0.36568087339401245, "learning_rate": 7.270492674554131e-05, "loss": 0.0241, "step": 114740 }, { "epoch": 73.1357552581262, "grad_norm": 0.26062512397766113, "learning_rate": 7.270001606972993e-05, "loss": 0.0106, "step": 114750 }, { "epoch": 73.1421287444232, "grad_norm": 0.5435271859169006, "learning_rate": 7.269510511809192e-05, "loss": 0.0251, "step": 114760 }, { "epoch": 73.1485022307202, "grad_norm": 0.3497425615787506, "learning_rate": 7.269019389068696e-05, "loss": 0.0185, "step": 114770 }, { "epoch": 73.1548757170172, "grad_norm": 0.6228663325309753, "learning_rate": 7.268528238757472e-05, "loss": 0.0105, "step": 114780 }, { "epoch": 73.16124920331421, "grad_norm": 0.6168334484100342, "learning_rate": 7.268037060881489e-05, "loss": 0.0161, "step": 114790 }, { "epoch": 73.16762268961122, "grad_norm": 0.37382084131240845, "learning_rate": 7.267545855446713e-05, "loss": 0.0148, "step": 114800 }, { "epoch": 73.17399617590821, "grad_norm": 0.4541049599647522, "learning_rate": 7.267054622459115e-05, "loss": 0.0159, "step": 114810 }, { "epoch": 73.18036966220522, "grad_norm": 0.2264547049999237, "learning_rate": 7.266563361924663e-05, "loss": 0.0181, "step": 114820 }, { "epoch": 73.18674314850223, "grad_norm": 0.45743653178215027, "learning_rate": 7.266072073849325e-05, "loss": 0.0158, "step": 114830 }, { "epoch": 73.19311663479924, "grad_norm": 0.5299502611160278, "learning_rate": 7.26558075823907e-05, "loss": 0.0127, "step": 114840 }, { "epoch": 73.19949012109623, "grad_norm": 0.24945341050624847, "learning_rate": 7.265089415099874e-05, "loss": 0.0214, "step": 114850 }, { "epoch": 73.20586360739324, "grad_norm": 0.6723679900169373, "learning_rate": 7.2645980444377e-05, "loss": 0.0231, "step": 114860 }, { "epoch": 73.21223709369025, "grad_norm": 0.3296510875225067, "learning_rate": 7.264106646258523e-05, "loss": 0.0125, "step": 114870 }, { "epoch": 73.21861057998726, "grad_norm": 0.47656840085983276, "learning_rate": 7.263615220568313e-05, "loss": 0.0164, "step": 114880 }, { "epoch": 73.22498406628426, "grad_norm": 0.421760231256485, "learning_rate": 7.263123767373039e-05, "loss": 0.0123, "step": 114890 }, { "epoch": 73.23135755258126, "grad_norm": 0.47046104073524475, "learning_rate": 7.262632286678675e-05, "loss": 0.028, "step": 114900 }, { "epoch": 73.23773103887827, "grad_norm": 0.2811891436576843, "learning_rate": 7.262140778491194e-05, "loss": 0.0108, "step": 114910 }, { "epoch": 73.24410452517527, "grad_norm": 0.33337539434432983, "learning_rate": 7.261649242816566e-05, "loss": 0.0176, "step": 114920 }, { "epoch": 73.25047801147228, "grad_norm": 0.575538158416748, "learning_rate": 7.261157679660763e-05, "loss": 0.0197, "step": 114930 }, { "epoch": 73.25685149776928, "grad_norm": 0.2701853811740875, "learning_rate": 7.260666089029761e-05, "loss": 0.0183, "step": 114940 }, { "epoch": 73.26322498406628, "grad_norm": 0.5189904570579529, "learning_rate": 7.260174470929531e-05, "loss": 0.015, "step": 114950 }, { "epoch": 73.26959847036329, "grad_norm": 0.5346341133117676, "learning_rate": 7.259682825366046e-05, "loss": 0.0124, "step": 114960 }, { "epoch": 73.2759719566603, "grad_norm": 0.6046198606491089, "learning_rate": 7.259191152345284e-05, "loss": 0.0193, "step": 114970 }, { "epoch": 73.2823454429573, "grad_norm": 0.23794376850128174, "learning_rate": 7.258699451873214e-05, "loss": 0.0185, "step": 114980 }, { "epoch": 73.2887189292543, "grad_norm": 0.5603199005126953, "learning_rate": 7.258207723955814e-05, "loss": 0.0142, "step": 114990 }, { "epoch": 73.29509241555131, "grad_norm": 0.23493841290473938, "learning_rate": 7.257715968599057e-05, "loss": 0.013, "step": 115000 }, { "epoch": 73.30146590184832, "grad_norm": 0.2693469822406769, "learning_rate": 7.25722418580892e-05, "loss": 0.0164, "step": 115010 }, { "epoch": 73.30783938814531, "grad_norm": 0.26224181056022644, "learning_rate": 7.25673237559138e-05, "loss": 0.0206, "step": 115020 }, { "epoch": 73.31421287444232, "grad_norm": 0.27086660265922546, "learning_rate": 7.256240537952408e-05, "loss": 0.0129, "step": 115030 }, { "epoch": 73.32058636073933, "grad_norm": 0.2586776912212372, "learning_rate": 7.255748672897985e-05, "loss": 0.0158, "step": 115040 }, { "epoch": 73.32695984703633, "grad_norm": 0.848334550857544, "learning_rate": 7.255256780434085e-05, "loss": 0.025, "step": 115050 }, { "epoch": 73.33333333333333, "grad_norm": 0.39143237471580505, "learning_rate": 7.254764860566686e-05, "loss": 0.0227, "step": 115060 }, { "epoch": 73.33970681963034, "grad_norm": 0.37229445576667786, "learning_rate": 7.254272913301765e-05, "loss": 0.0216, "step": 115070 }, { "epoch": 73.34608030592734, "grad_norm": 0.30652183294296265, "learning_rate": 7.2537809386453e-05, "loss": 0.0269, "step": 115080 }, { "epoch": 73.35245379222435, "grad_norm": 0.31502023339271545, "learning_rate": 7.253288936603269e-05, "loss": 0.0169, "step": 115090 }, { "epoch": 73.35882727852135, "grad_norm": 0.3458635210990906, "learning_rate": 7.252796907181649e-05, "loss": 0.013, "step": 115100 }, { "epoch": 73.36520076481835, "grad_norm": 0.3791997730731964, "learning_rate": 7.25230485038642e-05, "loss": 0.0174, "step": 115110 }, { "epoch": 73.37157425111536, "grad_norm": 0.5891581177711487, "learning_rate": 7.25181276622356e-05, "loss": 0.0249, "step": 115120 }, { "epoch": 73.37794773741237, "grad_norm": 0.35299086570739746, "learning_rate": 7.25132065469905e-05, "loss": 0.0144, "step": 115130 }, { "epoch": 73.38432122370936, "grad_norm": 0.31718409061431885, "learning_rate": 7.250828515818868e-05, "loss": 0.02, "step": 115140 }, { "epoch": 73.39069471000637, "grad_norm": 0.35293900966644287, "learning_rate": 7.250336349588994e-05, "loss": 0.0244, "step": 115150 }, { "epoch": 73.39706819630338, "grad_norm": 0.3104652166366577, "learning_rate": 7.249844156015408e-05, "loss": 0.0124, "step": 115160 }, { "epoch": 73.40344168260039, "grad_norm": 0.3197007477283478, "learning_rate": 7.249351935104094e-05, "loss": 0.017, "step": 115170 }, { "epoch": 73.40981516889738, "grad_norm": 0.36745020747184753, "learning_rate": 7.248859686861027e-05, "loss": 0.0181, "step": 115180 }, { "epoch": 73.41618865519439, "grad_norm": 0.6316176056861877, "learning_rate": 7.248367411292194e-05, "loss": 0.0164, "step": 115190 }, { "epoch": 73.4225621414914, "grad_norm": 0.4720275402069092, "learning_rate": 7.247875108403572e-05, "loss": 0.0168, "step": 115200 }, { "epoch": 73.4289356277884, "grad_norm": 0.3026297092437744, "learning_rate": 7.247382778201147e-05, "loss": 0.0144, "step": 115210 }, { "epoch": 73.4353091140854, "grad_norm": 0.5124156475067139, "learning_rate": 7.246890420690898e-05, "loss": 0.0206, "step": 115220 }, { "epoch": 73.4416826003824, "grad_norm": 0.8104950785636902, "learning_rate": 7.246398035878811e-05, "loss": 0.0173, "step": 115230 }, { "epoch": 73.44805608667941, "grad_norm": 0.45218324661254883, "learning_rate": 7.245905623770864e-05, "loss": 0.0229, "step": 115240 }, { "epoch": 73.45442957297642, "grad_norm": 0.2616206407546997, "learning_rate": 7.245413184373045e-05, "loss": 0.0151, "step": 115250 }, { "epoch": 73.46080305927342, "grad_norm": 0.31167906522750854, "learning_rate": 7.244920717691336e-05, "loss": 0.0139, "step": 115260 }, { "epoch": 73.46717654557042, "grad_norm": 0.6149519681930542, "learning_rate": 7.24442822373172e-05, "loss": 0.0174, "step": 115270 }, { "epoch": 73.47355003186743, "grad_norm": 0.7584963440895081, "learning_rate": 7.243935702500182e-05, "loss": 0.0192, "step": 115280 }, { "epoch": 73.47992351816444, "grad_norm": 0.35027632117271423, "learning_rate": 7.243443154002707e-05, "loss": 0.0107, "step": 115290 }, { "epoch": 73.48629700446143, "grad_norm": 0.25112730264663696, "learning_rate": 7.242950578245279e-05, "loss": 0.0163, "step": 115300 }, { "epoch": 73.49267049075844, "grad_norm": 0.3585738241672516, "learning_rate": 7.242457975233885e-05, "loss": 0.0135, "step": 115310 }, { "epoch": 73.49904397705545, "grad_norm": 0.5342162847518921, "learning_rate": 7.241965344974508e-05, "loss": 0.0134, "step": 115320 }, { "epoch": 73.50541746335246, "grad_norm": 0.3074877858161926, "learning_rate": 7.241472687473136e-05, "loss": 0.0123, "step": 115330 }, { "epoch": 73.51179094964945, "grad_norm": 0.3166631758213043, "learning_rate": 7.240980002735753e-05, "loss": 0.0124, "step": 115340 }, { "epoch": 73.51816443594646, "grad_norm": 0.3163415789604187, "learning_rate": 7.24048729076835e-05, "loss": 0.0104, "step": 115350 }, { "epoch": 73.52453792224347, "grad_norm": 0.3791068196296692, "learning_rate": 7.23999455157691e-05, "loss": 0.0164, "step": 115360 }, { "epoch": 73.53091140854048, "grad_norm": 0.4049588739871979, "learning_rate": 7.23950178516742e-05, "loss": 0.0181, "step": 115370 }, { "epoch": 73.53728489483747, "grad_norm": 0.27487578988075256, "learning_rate": 7.23900899154587e-05, "loss": 0.0145, "step": 115380 }, { "epoch": 73.54365838113448, "grad_norm": 0.5929043889045715, "learning_rate": 7.238516170718245e-05, "loss": 0.0164, "step": 115390 }, { "epoch": 73.55003186743149, "grad_norm": 0.5540155172348022, "learning_rate": 7.238023322690537e-05, "loss": 0.0168, "step": 115400 }, { "epoch": 73.5564053537285, "grad_norm": 0.4740724563598633, "learning_rate": 7.237530447468732e-05, "loss": 0.0153, "step": 115410 }, { "epoch": 73.56277884002549, "grad_norm": 0.6215957999229431, "learning_rate": 7.237037545058821e-05, "loss": 0.016, "step": 115420 }, { "epoch": 73.5691523263225, "grad_norm": 0.24161942303180695, "learning_rate": 7.236544615466789e-05, "loss": 0.0167, "step": 115430 }, { "epoch": 73.5755258126195, "grad_norm": 0.4381413459777832, "learning_rate": 7.236051658698629e-05, "loss": 0.0224, "step": 115440 }, { "epoch": 73.58189929891651, "grad_norm": 0.24443212151527405, "learning_rate": 7.235558674760331e-05, "loss": 0.0162, "step": 115450 }, { "epoch": 73.5882727852135, "grad_norm": 0.24450792372226715, "learning_rate": 7.235065663657883e-05, "loss": 0.0152, "step": 115460 }, { "epoch": 73.59464627151051, "grad_norm": 0.5791457295417786, "learning_rate": 7.234572625397279e-05, "loss": 0.0242, "step": 115470 }, { "epoch": 73.60101975780752, "grad_norm": 0.33753135800361633, "learning_rate": 7.234079559984505e-05, "loss": 0.0188, "step": 115480 }, { "epoch": 73.60739324410453, "grad_norm": 0.4831898510456085, "learning_rate": 7.233586467425558e-05, "loss": 0.018, "step": 115490 }, { "epoch": 73.61376673040154, "grad_norm": 0.536468505859375, "learning_rate": 7.233093347726426e-05, "loss": 0.0228, "step": 115500 }, { "epoch": 73.62014021669853, "grad_norm": 0.32197025418281555, "learning_rate": 7.232600200893101e-05, "loss": 0.0225, "step": 115510 }, { "epoch": 73.62651370299554, "grad_norm": 0.2870180308818817, "learning_rate": 7.232107026931576e-05, "loss": 0.018, "step": 115520 }, { "epoch": 73.63288718929255, "grad_norm": 0.4758233428001404, "learning_rate": 7.231613825847841e-05, "loss": 0.0176, "step": 115530 }, { "epoch": 73.63926067558955, "grad_norm": 0.5439596176147461, "learning_rate": 7.231120597647893e-05, "loss": 0.0135, "step": 115540 }, { "epoch": 73.64563416188655, "grad_norm": 0.22396385669708252, "learning_rate": 7.230627342337724e-05, "loss": 0.0119, "step": 115550 }, { "epoch": 73.65200764818356, "grad_norm": 0.30205848813056946, "learning_rate": 7.230134059923325e-05, "loss": 0.017, "step": 115560 }, { "epoch": 73.65838113448056, "grad_norm": 0.4344211220741272, "learning_rate": 7.229640750410692e-05, "loss": 0.0196, "step": 115570 }, { "epoch": 73.66475462077757, "grad_norm": 0.5814890265464783, "learning_rate": 7.229147413805819e-05, "loss": 0.016, "step": 115580 }, { "epoch": 73.67112810707457, "grad_norm": 0.5645961165428162, "learning_rate": 7.2286540501147e-05, "loss": 0.0196, "step": 115590 }, { "epoch": 73.67750159337157, "grad_norm": 0.5897126197814941, "learning_rate": 7.22816065934333e-05, "loss": 0.0223, "step": 115600 }, { "epoch": 73.68387507966858, "grad_norm": 0.2617107331752777, "learning_rate": 7.227667241497705e-05, "loss": 0.0166, "step": 115610 }, { "epoch": 73.69024856596559, "grad_norm": 0.3378779888153076, "learning_rate": 7.227173796583819e-05, "loss": 0.0201, "step": 115620 }, { "epoch": 73.69662205226258, "grad_norm": 0.36935514211654663, "learning_rate": 7.226680324607668e-05, "loss": 0.0169, "step": 115630 }, { "epoch": 73.70299553855959, "grad_norm": 0.33140432834625244, "learning_rate": 7.22618682557525e-05, "loss": 0.0155, "step": 115640 }, { "epoch": 73.7093690248566, "grad_norm": 0.3313341438770294, "learning_rate": 7.22569329949256e-05, "loss": 0.016, "step": 115650 }, { "epoch": 73.7157425111536, "grad_norm": 0.5798285603523254, "learning_rate": 7.225199746365594e-05, "loss": 0.0224, "step": 115660 }, { "epoch": 73.7221159974506, "grad_norm": 0.35749414563179016, "learning_rate": 7.224706166200352e-05, "loss": 0.0125, "step": 115670 }, { "epoch": 73.72848948374761, "grad_norm": 0.3144128918647766, "learning_rate": 7.224212559002829e-05, "loss": 0.0172, "step": 115680 }, { "epoch": 73.73486297004462, "grad_norm": 0.34504345059394836, "learning_rate": 7.223718924779023e-05, "loss": 0.0148, "step": 115690 }, { "epoch": 73.74123645634162, "grad_norm": 0.299193412065506, "learning_rate": 7.223225263534932e-05, "loss": 0.0262, "step": 115700 }, { "epoch": 73.74760994263862, "grad_norm": 0.605874240398407, "learning_rate": 7.222731575276555e-05, "loss": 0.0128, "step": 115710 }, { "epoch": 73.75398342893563, "grad_norm": 0.48805171251296997, "learning_rate": 7.222237860009892e-05, "loss": 0.017, "step": 115720 }, { "epoch": 73.76035691523263, "grad_norm": 0.25587064027786255, "learning_rate": 7.22174411774094e-05, "loss": 0.034, "step": 115730 }, { "epoch": 73.76673040152964, "grad_norm": 0.9557430744171143, "learning_rate": 7.221250348475698e-05, "loss": 0.024, "step": 115740 }, { "epoch": 73.77310388782664, "grad_norm": 0.41542255878448486, "learning_rate": 7.22075655222017e-05, "loss": 0.0174, "step": 115750 }, { "epoch": 73.77947737412364, "grad_norm": 0.6677271723747253, "learning_rate": 7.22026272898035e-05, "loss": 0.0169, "step": 115760 }, { "epoch": 73.78585086042065, "grad_norm": 0.5032698512077332, "learning_rate": 7.219768878762244e-05, "loss": 0.0106, "step": 115770 }, { "epoch": 73.79222434671766, "grad_norm": 0.7453247308731079, "learning_rate": 7.219275001571849e-05, "loss": 0.0133, "step": 115780 }, { "epoch": 73.79859783301465, "grad_norm": 0.4247680604457855, "learning_rate": 7.218781097415168e-05, "loss": 0.0138, "step": 115790 }, { "epoch": 73.80497131931166, "grad_norm": 0.5640400648117065, "learning_rate": 7.218287166298201e-05, "loss": 0.0201, "step": 115800 }, { "epoch": 73.81134480560867, "grad_norm": 0.33291658759117126, "learning_rate": 7.217793208226952e-05, "loss": 0.0182, "step": 115810 }, { "epoch": 73.81771829190568, "grad_norm": 0.4215115010738373, "learning_rate": 7.21729922320742e-05, "loss": 0.0164, "step": 115820 }, { "epoch": 73.82409177820267, "grad_norm": 0.3280388116836548, "learning_rate": 7.216805211245609e-05, "loss": 0.0225, "step": 115830 }, { "epoch": 73.83046526449968, "grad_norm": 0.3067978024482727, "learning_rate": 7.216311172347525e-05, "loss": 0.0121, "step": 115840 }, { "epoch": 73.83683875079669, "grad_norm": 0.7962836623191833, "learning_rate": 7.215817106519164e-05, "loss": 0.0155, "step": 115850 }, { "epoch": 73.8432122370937, "grad_norm": 0.23182956874370575, "learning_rate": 7.215323013766533e-05, "loss": 0.0232, "step": 115860 }, { "epoch": 73.84958572339069, "grad_norm": 0.48458367586135864, "learning_rate": 7.214828894095638e-05, "loss": 0.0208, "step": 115870 }, { "epoch": 73.8559592096877, "grad_norm": 0.19405071437358856, "learning_rate": 7.21433474751248e-05, "loss": 0.0152, "step": 115880 }, { "epoch": 73.8623326959847, "grad_norm": 0.5217065811157227, "learning_rate": 7.213840574023063e-05, "loss": 0.0177, "step": 115890 }, { "epoch": 73.86870618228171, "grad_norm": 0.26839619874954224, "learning_rate": 7.213346373633393e-05, "loss": 0.0197, "step": 115900 }, { "epoch": 73.8750796685787, "grad_norm": 0.2389526069164276, "learning_rate": 7.212852146349475e-05, "loss": 0.0102, "step": 115910 }, { "epoch": 73.88145315487571, "grad_norm": 0.26983219385147095, "learning_rate": 7.212357892177315e-05, "loss": 0.0202, "step": 115920 }, { "epoch": 73.88782664117272, "grad_norm": 0.2893909513950348, "learning_rate": 7.211863611122918e-05, "loss": 0.0191, "step": 115930 }, { "epoch": 73.89420012746973, "grad_norm": 0.2626626193523407, "learning_rate": 7.211369303192288e-05, "loss": 0.0214, "step": 115940 }, { "epoch": 73.90057361376672, "grad_norm": 0.4110710620880127, "learning_rate": 7.210874968391435e-05, "loss": 0.0191, "step": 115950 }, { "epoch": 73.90694710006373, "grad_norm": 0.3613704442977905, "learning_rate": 7.210380606726362e-05, "loss": 0.0196, "step": 115960 }, { "epoch": 73.91332058636074, "grad_norm": 0.5233836770057678, "learning_rate": 7.20988621820308e-05, "loss": 0.0261, "step": 115970 }, { "epoch": 73.91969407265775, "grad_norm": 0.28794923424720764, "learning_rate": 7.209391802827592e-05, "loss": 0.0221, "step": 115980 }, { "epoch": 73.92606755895474, "grad_norm": 0.6523963809013367, "learning_rate": 7.208897360605909e-05, "loss": 0.0244, "step": 115990 }, { "epoch": 73.93244104525175, "grad_norm": 0.2600138187408447, "learning_rate": 7.208402891544038e-05, "loss": 0.0128, "step": 116000 }, { "epoch": 73.93881453154876, "grad_norm": 0.6235347986221313, "learning_rate": 7.207908395647983e-05, "loss": 0.0169, "step": 116010 }, { "epoch": 73.94518801784577, "grad_norm": 0.23931175470352173, "learning_rate": 7.20741387292376e-05, "loss": 0.0139, "step": 116020 }, { "epoch": 73.95156150414276, "grad_norm": 0.5392659306526184, "learning_rate": 7.206919323377373e-05, "loss": 0.0161, "step": 116030 }, { "epoch": 73.95793499043977, "grad_norm": 0.5339849591255188, "learning_rate": 7.206424747014833e-05, "loss": 0.0282, "step": 116040 }, { "epoch": 73.96430847673678, "grad_norm": 0.4119715988636017, "learning_rate": 7.205930143842147e-05, "loss": 0.0154, "step": 116050 }, { "epoch": 73.97068196303378, "grad_norm": 0.3263351321220398, "learning_rate": 7.20543551386533e-05, "loss": 0.0177, "step": 116060 }, { "epoch": 73.97705544933078, "grad_norm": 0.621549129486084, "learning_rate": 7.204940857090389e-05, "loss": 0.0139, "step": 116070 }, { "epoch": 73.98342893562778, "grad_norm": 0.8502930998802185, "learning_rate": 7.204446173523333e-05, "loss": 0.0242, "step": 116080 }, { "epoch": 73.9898024219248, "grad_norm": 0.33467820286750793, "learning_rate": 7.203951463170176e-05, "loss": 0.0129, "step": 116090 }, { "epoch": 73.9961759082218, "grad_norm": 0.6854860186576843, "learning_rate": 7.203456726036927e-05, "loss": 0.0131, "step": 116100 }, { "epoch": 74.00254939451881, "grad_norm": 0.3075568377971649, "learning_rate": 7.2029619621296e-05, "loss": 0.0117, "step": 116110 }, { "epoch": 74.0089228808158, "grad_norm": 0.40048810839653015, "learning_rate": 7.202467171454202e-05, "loss": 0.0149, "step": 116120 }, { "epoch": 74.01529636711281, "grad_norm": 0.3538587987422943, "learning_rate": 7.201972354016752e-05, "loss": 0.0093, "step": 116130 }, { "epoch": 74.02166985340982, "grad_norm": 0.41513967514038086, "learning_rate": 7.201477509823255e-05, "loss": 0.0109, "step": 116140 }, { "epoch": 74.02804333970683, "grad_norm": 0.4846752882003784, "learning_rate": 7.200982638879733e-05, "loss": 0.0222, "step": 116150 }, { "epoch": 74.03441682600382, "grad_norm": 0.4269276261329651, "learning_rate": 7.20048774119219e-05, "loss": 0.0152, "step": 116160 }, { "epoch": 74.04079031230083, "grad_norm": 0.30178314447402954, "learning_rate": 7.199992816766644e-05, "loss": 0.0161, "step": 116170 }, { "epoch": 74.04716379859784, "grad_norm": 0.521265983581543, "learning_rate": 7.199497865609108e-05, "loss": 0.0207, "step": 116180 }, { "epoch": 74.05353728489484, "grad_norm": 0.4082285761833191, "learning_rate": 7.199002887725597e-05, "loss": 0.0222, "step": 116190 }, { "epoch": 74.05991077119184, "grad_norm": 0.3966520428657532, "learning_rate": 7.198507883122124e-05, "loss": 0.0155, "step": 116200 }, { "epoch": 74.06628425748885, "grad_norm": 0.31399857997894287, "learning_rate": 7.198012851804705e-05, "loss": 0.0192, "step": 116210 }, { "epoch": 74.07265774378585, "grad_norm": 0.5151280164718628, "learning_rate": 7.197517793779354e-05, "loss": 0.0126, "step": 116220 }, { "epoch": 74.07903123008286, "grad_norm": 0.8411592245101929, "learning_rate": 7.197022709052088e-05, "loss": 0.0207, "step": 116230 }, { "epoch": 74.08540471637986, "grad_norm": 0.29851576685905457, "learning_rate": 7.19652759762892e-05, "loss": 0.02, "step": 116240 }, { "epoch": 74.09177820267686, "grad_norm": 0.5149940848350525, "learning_rate": 7.19603245951587e-05, "loss": 0.0172, "step": 116250 }, { "epoch": 74.09815168897387, "grad_norm": 0.4416622221469879, "learning_rate": 7.19553729471895e-05, "loss": 0.0168, "step": 116260 }, { "epoch": 74.10452517527088, "grad_norm": 0.4956913888454437, "learning_rate": 7.195042103244181e-05, "loss": 0.0134, "step": 116270 }, { "epoch": 74.11089866156787, "grad_norm": 0.37001433968544006, "learning_rate": 7.194546885097576e-05, "loss": 0.0166, "step": 116280 }, { "epoch": 74.11727214786488, "grad_norm": 0.2170042246580124, "learning_rate": 7.194051640285157e-05, "loss": 0.0147, "step": 116290 }, { "epoch": 74.12364563416189, "grad_norm": 0.42293381690979004, "learning_rate": 7.193556368812936e-05, "loss": 0.0136, "step": 116300 }, { "epoch": 74.1300191204589, "grad_norm": 0.3197706937789917, "learning_rate": 7.193061070686936e-05, "loss": 0.0189, "step": 116310 }, { "epoch": 74.13639260675589, "grad_norm": 0.2805660367012024, "learning_rate": 7.192565745913173e-05, "loss": 0.0109, "step": 116320 }, { "epoch": 74.1427660930529, "grad_norm": 0.3782398998737335, "learning_rate": 7.192070394497667e-05, "loss": 0.0106, "step": 116330 }, { "epoch": 74.1491395793499, "grad_norm": 0.4187425374984741, "learning_rate": 7.191575016446435e-05, "loss": 0.0167, "step": 116340 }, { "epoch": 74.15551306564691, "grad_norm": 0.44776809215545654, "learning_rate": 7.191079611765496e-05, "loss": 0.0178, "step": 116350 }, { "epoch": 74.16188655194391, "grad_norm": 0.7005776762962341, "learning_rate": 7.190584180460874e-05, "loss": 0.0322, "step": 116360 }, { "epoch": 74.16826003824092, "grad_norm": 0.23176611959934235, "learning_rate": 7.190088722538582e-05, "loss": 0.022, "step": 116370 }, { "epoch": 74.17463352453792, "grad_norm": 0.4133422076702118, "learning_rate": 7.189593238004648e-05, "loss": 0.0209, "step": 116380 }, { "epoch": 74.18100701083493, "grad_norm": 0.49787554144859314, "learning_rate": 7.189097726865088e-05, "loss": 0.0175, "step": 116390 }, { "epoch": 74.18738049713193, "grad_norm": 0.40401026606559753, "learning_rate": 7.188602189125923e-05, "loss": 0.0137, "step": 116400 }, { "epoch": 74.19375398342893, "grad_norm": 0.579494297504425, "learning_rate": 7.188106624793175e-05, "loss": 0.0203, "step": 116410 }, { "epoch": 74.20012746972594, "grad_norm": 0.3028266131877899, "learning_rate": 7.187611033872865e-05, "loss": 0.0137, "step": 116420 }, { "epoch": 74.20650095602295, "grad_norm": 0.2473432868719101, "learning_rate": 7.187115416371018e-05, "loss": 0.0145, "step": 116430 }, { "epoch": 74.21287444231994, "grad_norm": 0.6884672045707703, "learning_rate": 7.18661977229365e-05, "loss": 0.0375, "step": 116440 }, { "epoch": 74.21924792861695, "grad_norm": 0.14972451329231262, "learning_rate": 7.18612410164679e-05, "loss": 0.0281, "step": 116450 }, { "epoch": 74.22562141491396, "grad_norm": 0.22430667281150818, "learning_rate": 7.185628404436458e-05, "loss": 0.0105, "step": 116460 }, { "epoch": 74.23199490121097, "grad_norm": 0.4549750089645386, "learning_rate": 7.185132680668676e-05, "loss": 0.0173, "step": 116470 }, { "epoch": 74.23836838750796, "grad_norm": 0.4119848906993866, "learning_rate": 7.184636930349472e-05, "loss": 0.0101, "step": 116480 }, { "epoch": 74.24474187380497, "grad_norm": 0.550010621547699, "learning_rate": 7.184141153484862e-05, "loss": 0.0242, "step": 116490 }, { "epoch": 74.25111536010198, "grad_norm": 0.5055806040763855, "learning_rate": 7.183645350080877e-05, "loss": 0.0238, "step": 116500 }, { "epoch": 74.25748884639899, "grad_norm": 0.2379290759563446, "learning_rate": 7.18314952014354e-05, "loss": 0.0128, "step": 116510 }, { "epoch": 74.26386233269598, "grad_norm": 0.5193620920181274, "learning_rate": 7.182653663678873e-05, "loss": 0.0191, "step": 116520 }, { "epoch": 74.27023581899299, "grad_norm": 0.3642948567867279, "learning_rate": 7.182157780692905e-05, "loss": 0.0157, "step": 116530 }, { "epoch": 74.27660930529, "grad_norm": 0.3807300627231598, "learning_rate": 7.18166187119166e-05, "loss": 0.0105, "step": 116540 }, { "epoch": 74.282982791587, "grad_norm": 0.4102737009525299, "learning_rate": 7.181165935181163e-05, "loss": 0.015, "step": 116550 }, { "epoch": 74.289356277884, "grad_norm": 0.35219132900238037, "learning_rate": 7.18066997266744e-05, "loss": 0.0165, "step": 116560 }, { "epoch": 74.295729764181, "grad_norm": 0.316741943359375, "learning_rate": 7.180173983656518e-05, "loss": 0.0245, "step": 116570 }, { "epoch": 74.30210325047801, "grad_norm": 0.6073954701423645, "learning_rate": 7.179677968154424e-05, "loss": 0.0146, "step": 116580 }, { "epoch": 74.30847673677502, "grad_norm": 0.45314398407936096, "learning_rate": 7.179181926167184e-05, "loss": 0.0142, "step": 116590 }, { "epoch": 74.31485022307201, "grad_norm": 0.4378422796726227, "learning_rate": 7.178685857700827e-05, "loss": 0.0128, "step": 116600 }, { "epoch": 74.32122370936902, "grad_norm": 0.8679520487785339, "learning_rate": 7.178189762761378e-05, "loss": 0.0211, "step": 116610 }, { "epoch": 74.32759719566603, "grad_norm": 0.2875358760356903, "learning_rate": 7.177693641354868e-05, "loss": 0.012, "step": 116620 }, { "epoch": 74.33397068196304, "grad_norm": 0.33438658714294434, "learning_rate": 7.177197493487326e-05, "loss": 0.0125, "step": 116630 }, { "epoch": 74.34034416826003, "grad_norm": 0.5345471501350403, "learning_rate": 7.176701319164777e-05, "loss": 0.0194, "step": 116640 }, { "epoch": 74.34671765455704, "grad_norm": 0.4707488715648651, "learning_rate": 7.176205118393251e-05, "loss": 0.0164, "step": 116650 }, { "epoch": 74.35309114085405, "grad_norm": 0.2933320999145508, "learning_rate": 7.17570889117878e-05, "loss": 0.0177, "step": 116660 }, { "epoch": 74.35946462715106, "grad_norm": 0.2661857306957245, "learning_rate": 7.17521263752739e-05, "loss": 0.0105, "step": 116670 }, { "epoch": 74.36583811344805, "grad_norm": 0.20921756327152252, "learning_rate": 7.174716357445115e-05, "loss": 0.0163, "step": 116680 }, { "epoch": 74.37221159974506, "grad_norm": 0.4425075948238373, "learning_rate": 7.17422005093798e-05, "loss": 0.0255, "step": 116690 }, { "epoch": 74.37858508604207, "grad_norm": 0.590911328792572, "learning_rate": 7.173723718012022e-05, "loss": 0.0164, "step": 116700 }, { "epoch": 74.38495857233907, "grad_norm": 0.37575894594192505, "learning_rate": 7.173227358673267e-05, "loss": 0.0265, "step": 116710 }, { "epoch": 74.39133205863607, "grad_norm": 0.35875996947288513, "learning_rate": 7.172730972927748e-05, "loss": 0.0162, "step": 116720 }, { "epoch": 74.39770554493307, "grad_norm": 0.32842519879341125, "learning_rate": 7.172234560781495e-05, "loss": 0.0209, "step": 116730 }, { "epoch": 74.40407903123008, "grad_norm": 0.3168003261089325, "learning_rate": 7.171738122240543e-05, "loss": 0.0157, "step": 116740 }, { "epoch": 74.41045251752709, "grad_norm": 0.29454731941223145, "learning_rate": 7.171241657310921e-05, "loss": 0.0168, "step": 116750 }, { "epoch": 74.4168260038241, "grad_norm": 0.4793085753917694, "learning_rate": 7.170745165998663e-05, "loss": 0.0105, "step": 116760 }, { "epoch": 74.42319949012109, "grad_norm": 0.5265362858772278, "learning_rate": 7.170248648309801e-05, "loss": 0.0166, "step": 116770 }, { "epoch": 74.4295729764181, "grad_norm": 0.15961089730262756, "learning_rate": 7.16975210425037e-05, "loss": 0.0177, "step": 116780 }, { "epoch": 74.43594646271511, "grad_norm": 0.3948343098163605, "learning_rate": 7.169255533826403e-05, "loss": 0.0178, "step": 116790 }, { "epoch": 74.44231994901212, "grad_norm": 0.3741171956062317, "learning_rate": 7.168758937043933e-05, "loss": 0.0213, "step": 116800 }, { "epoch": 74.44869343530911, "grad_norm": 0.2917334735393524, "learning_rate": 7.168262313908993e-05, "loss": 0.0234, "step": 116810 }, { "epoch": 74.45506692160612, "grad_norm": 0.3895876705646515, "learning_rate": 7.16776566442762e-05, "loss": 0.0199, "step": 116820 }, { "epoch": 74.46144040790313, "grad_norm": 0.22522903978824615, "learning_rate": 7.167268988605845e-05, "loss": 0.0154, "step": 116830 }, { "epoch": 74.46781389420013, "grad_norm": 0.4412127435207367, "learning_rate": 7.166772286449708e-05, "loss": 0.0232, "step": 116840 }, { "epoch": 74.47418738049713, "grad_norm": 0.4267738461494446, "learning_rate": 7.166275557965241e-05, "loss": 0.0192, "step": 116850 }, { "epoch": 74.48056086679414, "grad_norm": 0.41632208228111267, "learning_rate": 7.16577880315848e-05, "loss": 0.0135, "step": 116860 }, { "epoch": 74.48693435309114, "grad_norm": 0.4077979028224945, "learning_rate": 7.165282022035462e-05, "loss": 0.0245, "step": 116870 }, { "epoch": 74.49330783938815, "grad_norm": 0.3422371447086334, "learning_rate": 7.164785214602226e-05, "loss": 0.0314, "step": 116880 }, { "epoch": 74.49968132568515, "grad_norm": 0.45189401507377625, "learning_rate": 7.164288380864803e-05, "loss": 0.0215, "step": 116890 }, { "epoch": 74.50605481198215, "grad_norm": 0.8089586496353149, "learning_rate": 7.163791520829232e-05, "loss": 0.028, "step": 116900 }, { "epoch": 74.51242829827916, "grad_norm": 0.42305925488471985, "learning_rate": 7.163294634501553e-05, "loss": 0.0176, "step": 116910 }, { "epoch": 74.51880178457617, "grad_norm": 0.6178303360939026, "learning_rate": 7.162797721887801e-05, "loss": 0.0173, "step": 116920 }, { "epoch": 74.52517527087316, "grad_norm": 0.3254499137401581, "learning_rate": 7.162300782994015e-05, "loss": 0.0188, "step": 116930 }, { "epoch": 74.53154875717017, "grad_norm": 0.4442274868488312, "learning_rate": 7.161803817826231e-05, "loss": 0.0118, "step": 116940 }, { "epoch": 74.53792224346718, "grad_norm": 0.19798272848129272, "learning_rate": 7.161306826390491e-05, "loss": 0.014, "step": 116950 }, { "epoch": 74.54429572976419, "grad_norm": 0.49758380651474, "learning_rate": 7.160809808692835e-05, "loss": 0.0182, "step": 116960 }, { "epoch": 74.55066921606118, "grad_norm": 0.23390142619609833, "learning_rate": 7.160312764739295e-05, "loss": 0.0175, "step": 116970 }, { "epoch": 74.55704270235819, "grad_norm": 0.2280474305152893, "learning_rate": 7.159815694535919e-05, "loss": 0.0154, "step": 116980 }, { "epoch": 74.5634161886552, "grad_norm": 0.39231476187705994, "learning_rate": 7.15931859808874e-05, "loss": 0.0125, "step": 116990 }, { "epoch": 74.5697896749522, "grad_norm": 0.4148441255092621, "learning_rate": 7.158821475403803e-05, "loss": 0.013, "step": 117000 }, { "epoch": 74.5761631612492, "grad_norm": 0.37935805320739746, "learning_rate": 7.158324326487147e-05, "loss": 0.0162, "step": 117010 }, { "epoch": 74.5825366475462, "grad_norm": 0.2716043293476105, "learning_rate": 7.157827151344813e-05, "loss": 0.0116, "step": 117020 }, { "epoch": 74.58891013384321, "grad_norm": 0.2592540979385376, "learning_rate": 7.15732994998284e-05, "loss": 0.0139, "step": 117030 }, { "epoch": 74.59528362014022, "grad_norm": 0.552582323551178, "learning_rate": 7.156832722407271e-05, "loss": 0.0172, "step": 117040 }, { "epoch": 74.60165710643722, "grad_norm": 0.5816164016723633, "learning_rate": 7.156335468624151e-05, "loss": 0.0307, "step": 117050 }, { "epoch": 74.60803059273422, "grad_norm": 0.44948694109916687, "learning_rate": 7.155838188639517e-05, "loss": 0.0154, "step": 117060 }, { "epoch": 74.61440407903123, "grad_norm": 0.34117797017097473, "learning_rate": 7.155340882459414e-05, "loss": 0.019, "step": 117070 }, { "epoch": 74.62077756532824, "grad_norm": 0.32258760929107666, "learning_rate": 7.154843550089885e-05, "loss": 0.0211, "step": 117080 }, { "epoch": 74.62715105162523, "grad_norm": 0.2637089788913727, "learning_rate": 7.154346191536971e-05, "loss": 0.0176, "step": 117090 }, { "epoch": 74.63352453792224, "grad_norm": 0.4709986746311188, "learning_rate": 7.153848806806718e-05, "loss": 0.0176, "step": 117100 }, { "epoch": 74.63989802421925, "grad_norm": 0.41009974479675293, "learning_rate": 7.153351395905168e-05, "loss": 0.0179, "step": 117110 }, { "epoch": 74.64627151051626, "grad_norm": 0.6144423484802246, "learning_rate": 7.152853958838366e-05, "loss": 0.0136, "step": 117120 }, { "epoch": 74.65264499681325, "grad_norm": 0.3654935657978058, "learning_rate": 7.152356495612354e-05, "loss": 0.0193, "step": 117130 }, { "epoch": 74.65901848311026, "grad_norm": 0.2829592227935791, "learning_rate": 7.151859006233179e-05, "loss": 0.0112, "step": 117140 }, { "epoch": 74.66539196940727, "grad_norm": 0.6396158933639526, "learning_rate": 7.151361490706886e-05, "loss": 0.0197, "step": 117150 }, { "epoch": 74.67176545570427, "grad_norm": 0.2542078197002411, "learning_rate": 7.15086394903952e-05, "loss": 0.0139, "step": 117160 }, { "epoch": 74.67813894200127, "grad_norm": 0.29929178953170776, "learning_rate": 7.150366381237125e-05, "loss": 0.0255, "step": 117170 }, { "epoch": 74.68451242829828, "grad_norm": 0.545782208442688, "learning_rate": 7.149868787305749e-05, "loss": 0.0162, "step": 117180 }, { "epoch": 74.69088591459528, "grad_norm": 0.34052392840385437, "learning_rate": 7.149371167251438e-05, "loss": 0.0206, "step": 117190 }, { "epoch": 74.69725940089229, "grad_norm": 0.27173107862472534, "learning_rate": 7.148873521080237e-05, "loss": 0.0177, "step": 117200 }, { "epoch": 74.70363288718929, "grad_norm": 0.3928282856941223, "learning_rate": 7.148375848798194e-05, "loss": 0.0142, "step": 117210 }, { "epoch": 74.7100063734863, "grad_norm": 0.21264420449733734, "learning_rate": 7.147878150411356e-05, "loss": 0.0128, "step": 117220 }, { "epoch": 74.7163798597833, "grad_norm": 0.8242433667182922, "learning_rate": 7.147380425925771e-05, "loss": 0.019, "step": 117230 }, { "epoch": 74.72275334608031, "grad_norm": 0.9869304895401001, "learning_rate": 7.146882675347486e-05, "loss": 0.0136, "step": 117240 }, { "epoch": 74.7291268323773, "grad_norm": 0.42476144433021545, "learning_rate": 7.14638489868255e-05, "loss": 0.0282, "step": 117250 }, { "epoch": 74.73550031867431, "grad_norm": 0.22038203477859497, "learning_rate": 7.14588709593701e-05, "loss": 0.0161, "step": 117260 }, { "epoch": 74.74187380497132, "grad_norm": 0.38567858934402466, "learning_rate": 7.145389267116917e-05, "loss": 0.0101, "step": 117270 }, { "epoch": 74.74824729126833, "grad_norm": 0.23413538932800293, "learning_rate": 7.144891412228318e-05, "loss": 0.0193, "step": 117280 }, { "epoch": 74.75462077756532, "grad_norm": 0.24906465411186218, "learning_rate": 7.144393531277264e-05, "loss": 0.0149, "step": 117290 }, { "epoch": 74.76099426386233, "grad_norm": 0.4420420229434967, "learning_rate": 7.143895624269802e-05, "loss": 0.0144, "step": 117300 }, { "epoch": 74.76736775015934, "grad_norm": 0.5890982151031494, "learning_rate": 7.143397691211986e-05, "loss": 0.0173, "step": 117310 }, { "epoch": 74.77374123645635, "grad_norm": 0.5032193064689636, "learning_rate": 7.142899732109864e-05, "loss": 0.0272, "step": 117320 }, { "epoch": 74.78011472275334, "grad_norm": 0.31868594884872437, "learning_rate": 7.142401746969487e-05, "loss": 0.0194, "step": 117330 }, { "epoch": 74.78648820905035, "grad_norm": 0.4130663275718689, "learning_rate": 7.141903735796904e-05, "loss": 0.018, "step": 117340 }, { "epoch": 74.79286169534736, "grad_norm": 0.2160739302635193, "learning_rate": 7.141405698598172e-05, "loss": 0.018, "step": 117350 }, { "epoch": 74.79923518164436, "grad_norm": 0.2437848150730133, "learning_rate": 7.140907635379336e-05, "loss": 0.0194, "step": 117360 }, { "epoch": 74.80560866794136, "grad_norm": 0.3845278322696686, "learning_rate": 7.140409546146452e-05, "loss": 0.0192, "step": 117370 }, { "epoch": 74.81198215423836, "grad_norm": 0.254384845495224, "learning_rate": 7.139911430905571e-05, "loss": 0.0123, "step": 117380 }, { "epoch": 74.81835564053537, "grad_norm": 0.2830893099308014, "learning_rate": 7.139413289662745e-05, "loss": 0.0231, "step": 117390 }, { "epoch": 74.82472912683238, "grad_norm": 0.5557066798210144, "learning_rate": 7.138915122424029e-05, "loss": 0.0213, "step": 117400 }, { "epoch": 74.83110261312939, "grad_norm": 0.7901528477668762, "learning_rate": 7.138416929195474e-05, "loss": 0.0251, "step": 117410 }, { "epoch": 74.83747609942638, "grad_norm": 0.4900665283203125, "learning_rate": 7.137918709983134e-05, "loss": 0.0188, "step": 117420 }, { "epoch": 74.84384958572339, "grad_norm": 0.40884748101234436, "learning_rate": 7.137420464793063e-05, "loss": 0.0234, "step": 117430 }, { "epoch": 74.8502230720204, "grad_norm": 0.5936700105667114, "learning_rate": 7.136922193631317e-05, "loss": 0.0181, "step": 117440 }, { "epoch": 74.8565965583174, "grad_norm": 0.4353640675544739, "learning_rate": 7.136423896503948e-05, "loss": 0.0083, "step": 117450 }, { "epoch": 74.8629700446144, "grad_norm": 0.21858566999435425, "learning_rate": 7.13592557341701e-05, "loss": 0.0138, "step": 117460 }, { "epoch": 74.86934353091141, "grad_norm": 0.2760002613067627, "learning_rate": 7.135427224376561e-05, "loss": 0.0139, "step": 117470 }, { "epoch": 74.87571701720842, "grad_norm": 0.6029525399208069, "learning_rate": 7.134928849388655e-05, "loss": 0.0124, "step": 117480 }, { "epoch": 74.88209050350542, "grad_norm": 0.6593682169914246, "learning_rate": 7.134430448459348e-05, "loss": 0.0142, "step": 117490 }, { "epoch": 74.88846398980242, "grad_norm": 0.3489951193332672, "learning_rate": 7.133932021594694e-05, "loss": 0.0115, "step": 117500 }, { "epoch": 74.89483747609943, "grad_norm": 0.5161749720573425, "learning_rate": 7.133433568800752e-05, "loss": 0.0182, "step": 117510 }, { "epoch": 74.90121096239643, "grad_norm": 0.3992754817008972, "learning_rate": 7.13293509008358e-05, "loss": 0.015, "step": 117520 }, { "epoch": 74.90758444869344, "grad_norm": 0.5144767165184021, "learning_rate": 7.132436585449231e-05, "loss": 0.0113, "step": 117530 }, { "epoch": 74.91395793499044, "grad_norm": 0.34124207496643066, "learning_rate": 7.131938054903763e-05, "loss": 0.0172, "step": 117540 }, { "epoch": 74.92033142128744, "grad_norm": 0.46153032779693604, "learning_rate": 7.131439498453237e-05, "loss": 0.0285, "step": 117550 }, { "epoch": 74.92670490758445, "grad_norm": 0.5381916761398315, "learning_rate": 7.130940916103707e-05, "loss": 0.0126, "step": 117560 }, { "epoch": 74.93307839388146, "grad_norm": 0.6206822991371155, "learning_rate": 7.130442307861235e-05, "loss": 0.0243, "step": 117570 }, { "epoch": 74.93945188017845, "grad_norm": 0.282241553068161, "learning_rate": 7.129943673731876e-05, "loss": 0.0157, "step": 117580 }, { "epoch": 74.94582536647546, "grad_norm": 0.5282867550849915, "learning_rate": 7.12944501372169e-05, "loss": 0.0235, "step": 117590 }, { "epoch": 74.95219885277247, "grad_norm": 0.326990008354187, "learning_rate": 7.128946327836738e-05, "loss": 0.0219, "step": 117600 }, { "epoch": 74.95857233906948, "grad_norm": 0.27237987518310547, "learning_rate": 7.128447616083077e-05, "loss": 0.0229, "step": 117610 }, { "epoch": 74.96494582536647, "grad_norm": 0.30156657099723816, "learning_rate": 7.127948878466768e-05, "loss": 0.0232, "step": 117620 }, { "epoch": 74.97131931166348, "grad_norm": 0.563775360584259, "learning_rate": 7.127450114993871e-05, "loss": 0.0221, "step": 117630 }, { "epoch": 74.97769279796049, "grad_norm": 0.37508416175842285, "learning_rate": 7.126951325670447e-05, "loss": 0.0183, "step": 117640 }, { "epoch": 74.9840662842575, "grad_norm": 0.7179257869720459, "learning_rate": 7.126452510502554e-05, "loss": 0.0189, "step": 117650 }, { "epoch": 74.99043977055449, "grad_norm": 0.3970891237258911, "learning_rate": 7.125953669496259e-05, "loss": 0.0179, "step": 117660 }, { "epoch": 74.9968132568515, "grad_norm": 0.15423694252967834, "learning_rate": 7.125454802657617e-05, "loss": 0.0158, "step": 117670 }, { "epoch": 75.0031867431485, "grad_norm": 0.47962063550949097, "learning_rate": 7.124955909992693e-05, "loss": 0.0153, "step": 117680 }, { "epoch": 75.00956022944551, "grad_norm": 0.31093651056289673, "learning_rate": 7.124456991507548e-05, "loss": 0.0144, "step": 117690 }, { "epoch": 75.0159337157425, "grad_norm": 0.34098923206329346, "learning_rate": 7.123958047208246e-05, "loss": 0.0199, "step": 117700 }, { "epoch": 75.02230720203951, "grad_norm": 0.4469543397426605, "learning_rate": 7.123459077100848e-05, "loss": 0.0141, "step": 117710 }, { "epoch": 75.02868068833652, "grad_norm": 0.4541991651058197, "learning_rate": 7.122960081191417e-05, "loss": 0.0212, "step": 117720 }, { "epoch": 75.03505417463353, "grad_norm": 0.39003786444664, "learning_rate": 7.122461059486016e-05, "loss": 0.0158, "step": 117730 }, { "epoch": 75.04142766093052, "grad_norm": 0.5544167757034302, "learning_rate": 7.12196201199071e-05, "loss": 0.0208, "step": 117740 }, { "epoch": 75.04780114722753, "grad_norm": 0.27241212129592896, "learning_rate": 7.121462938711561e-05, "loss": 0.022, "step": 117750 }, { "epoch": 75.05417463352454, "grad_norm": 0.6401001811027527, "learning_rate": 7.120963839654635e-05, "loss": 0.0164, "step": 117760 }, { "epoch": 75.06054811982155, "grad_norm": 0.23535755276679993, "learning_rate": 7.120464714825998e-05, "loss": 0.014, "step": 117770 }, { "epoch": 75.06692160611854, "grad_norm": 0.30301934480667114, "learning_rate": 7.11996556423171e-05, "loss": 0.0134, "step": 117780 }, { "epoch": 75.07329509241555, "grad_norm": 0.38700070977211, "learning_rate": 7.119466387877839e-05, "loss": 0.0126, "step": 117790 }, { "epoch": 75.07966857871256, "grad_norm": 0.5298253297805786, "learning_rate": 7.11896718577045e-05, "loss": 0.0103, "step": 117800 }, { "epoch": 75.08604206500956, "grad_norm": 0.24552221596240997, "learning_rate": 7.11846795791561e-05, "loss": 0.0162, "step": 117810 }, { "epoch": 75.09241555130656, "grad_norm": 0.3840981721878052, "learning_rate": 7.117968704319384e-05, "loss": 0.0218, "step": 117820 }, { "epoch": 75.09878903760357, "grad_norm": 0.3721402883529663, "learning_rate": 7.117469424987839e-05, "loss": 0.016, "step": 117830 }, { "epoch": 75.10516252390057, "grad_norm": 0.42303621768951416, "learning_rate": 7.116970119927041e-05, "loss": 0.0199, "step": 117840 }, { "epoch": 75.11153601019758, "grad_norm": 0.5831379294395447, "learning_rate": 7.116470789143057e-05, "loss": 0.0251, "step": 117850 }, { "epoch": 75.11790949649458, "grad_norm": 0.40739545226097107, "learning_rate": 7.115971432641955e-05, "loss": 0.0149, "step": 117860 }, { "epoch": 75.12428298279158, "grad_norm": 0.8486936092376709, "learning_rate": 7.115472050429802e-05, "loss": 0.0301, "step": 117870 }, { "epoch": 75.13065646908859, "grad_norm": 0.5703004002571106, "learning_rate": 7.114972642512665e-05, "loss": 0.0212, "step": 117880 }, { "epoch": 75.1370299553856, "grad_norm": 0.350324422121048, "learning_rate": 7.114473208896616e-05, "loss": 0.0118, "step": 117890 }, { "epoch": 75.1434034416826, "grad_norm": 0.4853611886501312, "learning_rate": 7.113973749587719e-05, "loss": 0.0155, "step": 117900 }, { "epoch": 75.1497769279796, "grad_norm": 0.4184267222881317, "learning_rate": 7.113474264592045e-05, "loss": 0.0103, "step": 117910 }, { "epoch": 75.15615041427661, "grad_norm": 0.7300262451171875, "learning_rate": 7.112974753915665e-05, "loss": 0.0172, "step": 117920 }, { "epoch": 75.16252390057362, "grad_norm": 0.34682875871658325, "learning_rate": 7.112475217564647e-05, "loss": 0.0129, "step": 117930 }, { "epoch": 75.16889738687061, "grad_norm": 0.3113180696964264, "learning_rate": 7.111975655545059e-05, "loss": 0.0181, "step": 117940 }, { "epoch": 75.17527087316762, "grad_norm": 0.39826396107673645, "learning_rate": 7.111476067862973e-05, "loss": 0.0159, "step": 117950 }, { "epoch": 75.18164435946463, "grad_norm": 0.48751622438430786, "learning_rate": 7.110976454524459e-05, "loss": 0.019, "step": 117960 }, { "epoch": 75.18801784576164, "grad_norm": 0.3592997193336487, "learning_rate": 7.110476815535588e-05, "loss": 0.0152, "step": 117970 }, { "epoch": 75.19439133205863, "grad_norm": 0.1926124542951584, "learning_rate": 7.109977150902433e-05, "loss": 0.0108, "step": 117980 }, { "epoch": 75.20076481835564, "grad_norm": 0.5644398331642151, "learning_rate": 7.10947746063106e-05, "loss": 0.0158, "step": 117990 }, { "epoch": 75.20713830465264, "grad_norm": 0.45703577995300293, "learning_rate": 7.108977744727547e-05, "loss": 0.0171, "step": 118000 }, { "epoch": 75.21351179094965, "grad_norm": 0.4369147717952728, "learning_rate": 7.108478003197963e-05, "loss": 0.0096, "step": 118010 }, { "epoch": 75.21988527724666, "grad_norm": 0.45491623878479004, "learning_rate": 7.10797823604838e-05, "loss": 0.0104, "step": 118020 }, { "epoch": 75.22625876354365, "grad_norm": 0.453640341758728, "learning_rate": 7.107478443284869e-05, "loss": 0.0185, "step": 118030 }, { "epoch": 75.23263224984066, "grad_norm": 0.44394561648368835, "learning_rate": 7.106978624913509e-05, "loss": 0.0203, "step": 118040 }, { "epoch": 75.23900573613767, "grad_norm": 0.29633545875549316, "learning_rate": 7.106478780940366e-05, "loss": 0.0202, "step": 118050 }, { "epoch": 75.24537922243468, "grad_norm": 0.3647107183933258, "learning_rate": 7.10597891137152e-05, "loss": 0.016, "step": 118060 }, { "epoch": 75.25175270873167, "grad_norm": 0.24277363717556, "learning_rate": 7.105479016213039e-05, "loss": 0.0249, "step": 118070 }, { "epoch": 75.25812619502868, "grad_norm": 0.3760528564453125, "learning_rate": 7.104979095471003e-05, "loss": 0.0171, "step": 118080 }, { "epoch": 75.26449968132569, "grad_norm": 0.2430296689271927, "learning_rate": 7.10447914915148e-05, "loss": 0.0166, "step": 118090 }, { "epoch": 75.2708731676227, "grad_norm": 0.5031399726867676, "learning_rate": 7.103979177260551e-05, "loss": 0.0197, "step": 118100 }, { "epoch": 75.27724665391969, "grad_norm": 0.5476059317588806, "learning_rate": 7.103479179804287e-05, "loss": 0.0258, "step": 118110 }, { "epoch": 75.2836201402167, "grad_norm": 0.20177757740020752, "learning_rate": 7.102979156788767e-05, "loss": 0.0189, "step": 118120 }, { "epoch": 75.2899936265137, "grad_norm": 0.3523573875427246, "learning_rate": 7.102479108220063e-05, "loss": 0.0222, "step": 118130 }, { "epoch": 75.29636711281071, "grad_norm": 0.5069718360900879, "learning_rate": 7.101979034104254e-05, "loss": 0.0115, "step": 118140 }, { "epoch": 75.30274059910771, "grad_norm": 0.4522167444229126, "learning_rate": 7.101478934447414e-05, "loss": 0.0118, "step": 118150 }, { "epoch": 75.30911408540472, "grad_norm": 0.3507945239543915, "learning_rate": 7.100978809255622e-05, "loss": 0.0143, "step": 118160 }, { "epoch": 75.31548757170172, "grad_norm": 0.6283950805664062, "learning_rate": 7.100478658534952e-05, "loss": 0.0188, "step": 118170 }, { "epoch": 75.32186105799873, "grad_norm": 0.5681741833686829, "learning_rate": 7.099978482291484e-05, "loss": 0.0171, "step": 118180 }, { "epoch": 75.32823454429573, "grad_norm": 0.3189215064048767, "learning_rate": 7.099478280531295e-05, "loss": 0.0171, "step": 118190 }, { "epoch": 75.33460803059273, "grad_norm": 1.1620513200759888, "learning_rate": 7.098978053260464e-05, "loss": 0.0212, "step": 118200 }, { "epoch": 75.34098151688974, "grad_norm": 0.421426922082901, "learning_rate": 7.098477800485066e-05, "loss": 0.0173, "step": 118210 }, { "epoch": 75.34735500318675, "grad_norm": 0.2165140062570572, "learning_rate": 7.09797752221118e-05, "loss": 0.0119, "step": 118220 }, { "epoch": 75.35372848948374, "grad_norm": 0.7376498579978943, "learning_rate": 7.09747721844489e-05, "loss": 0.0173, "step": 118230 }, { "epoch": 75.36010197578075, "grad_norm": 0.22096237540245056, "learning_rate": 7.096976889192268e-05, "loss": 0.0256, "step": 118240 }, { "epoch": 75.36647546207776, "grad_norm": 0.3184158504009247, "learning_rate": 7.0964765344594e-05, "loss": 0.0179, "step": 118250 }, { "epoch": 75.37284894837477, "grad_norm": 0.28278058767318726, "learning_rate": 7.095976154252362e-05, "loss": 0.0167, "step": 118260 }, { "epoch": 75.37922243467176, "grad_norm": 0.3692629635334015, "learning_rate": 7.095475748577233e-05, "loss": 0.0213, "step": 118270 }, { "epoch": 75.38559592096877, "grad_norm": 0.4559711217880249, "learning_rate": 7.094975317440097e-05, "loss": 0.0175, "step": 118280 }, { "epoch": 75.39196940726578, "grad_norm": 0.3263126611709595, "learning_rate": 7.094474860847032e-05, "loss": 0.0154, "step": 118290 }, { "epoch": 75.39834289356278, "grad_norm": 0.2649509906768799, "learning_rate": 7.09397437880412e-05, "loss": 0.0132, "step": 118300 }, { "epoch": 75.40471637985978, "grad_norm": 0.36323386430740356, "learning_rate": 7.093473871317442e-05, "loss": 0.0103, "step": 118310 }, { "epoch": 75.41108986615679, "grad_norm": 0.20992234349250793, "learning_rate": 7.092973338393081e-05, "loss": 0.0182, "step": 118320 }, { "epoch": 75.4174633524538, "grad_norm": 0.47000283002853394, "learning_rate": 7.092472780037117e-05, "loss": 0.0189, "step": 118330 }, { "epoch": 75.4238368387508, "grad_norm": 0.3510236442089081, "learning_rate": 7.091972196255632e-05, "loss": 0.0159, "step": 118340 }, { "epoch": 75.4302103250478, "grad_norm": 0.325776606798172, "learning_rate": 7.091471587054712e-05, "loss": 0.0195, "step": 118350 }, { "epoch": 75.4365838113448, "grad_norm": 0.7340406775474548, "learning_rate": 7.090970952440435e-05, "loss": 0.0264, "step": 118360 }, { "epoch": 75.44295729764181, "grad_norm": 0.476479709148407, "learning_rate": 7.090470292418889e-05, "loss": 0.0155, "step": 118370 }, { "epoch": 75.44933078393882, "grad_norm": 0.3958926796913147, "learning_rate": 7.089969606996154e-05, "loss": 0.0141, "step": 118380 }, { "epoch": 75.45570427023581, "grad_norm": 0.3670749366283417, "learning_rate": 7.089468896178316e-05, "loss": 0.014, "step": 118390 }, { "epoch": 75.46207775653282, "grad_norm": 0.4032856523990631, "learning_rate": 7.088968159971457e-05, "loss": 0.0251, "step": 118400 }, { "epoch": 75.46845124282983, "grad_norm": 0.2790076732635498, "learning_rate": 7.088467398381663e-05, "loss": 0.0163, "step": 118410 }, { "epoch": 75.47482472912684, "grad_norm": 0.4327918291091919, "learning_rate": 7.087966611415019e-05, "loss": 0.0213, "step": 118420 }, { "epoch": 75.48119821542383, "grad_norm": 0.5310767889022827, "learning_rate": 7.087465799077608e-05, "loss": 0.0134, "step": 118430 }, { "epoch": 75.48757170172084, "grad_norm": 0.45143043994903564, "learning_rate": 7.086964961375518e-05, "loss": 0.0198, "step": 118440 }, { "epoch": 75.49394518801785, "grad_norm": 0.3090333640575409, "learning_rate": 7.086464098314832e-05, "loss": 0.0145, "step": 118450 }, { "epoch": 75.50031867431485, "grad_norm": 0.2475583404302597, "learning_rate": 7.085963209901638e-05, "loss": 0.0201, "step": 118460 }, { "epoch": 75.50669216061185, "grad_norm": 0.41350194811820984, "learning_rate": 7.08546229614202e-05, "loss": 0.0127, "step": 118470 }, { "epoch": 75.51306564690886, "grad_norm": 0.2784697115421295, "learning_rate": 7.084961357042068e-05, "loss": 0.0139, "step": 118480 }, { "epoch": 75.51943913320586, "grad_norm": 0.3643914759159088, "learning_rate": 7.084460392607868e-05, "loss": 0.0159, "step": 118490 }, { "epoch": 75.52581261950287, "grad_norm": 0.6785404086112976, "learning_rate": 7.083959402845504e-05, "loss": 0.0156, "step": 118500 }, { "epoch": 75.53218610579987, "grad_norm": 0.4057476818561554, "learning_rate": 7.083458387761065e-05, "loss": 0.0139, "step": 118510 }, { "epoch": 75.53855959209687, "grad_norm": 0.5370882153511047, "learning_rate": 7.082957347360642e-05, "loss": 0.023, "step": 118520 }, { "epoch": 75.54493307839388, "grad_norm": 0.5640504956245422, "learning_rate": 7.08245628165032e-05, "loss": 0.0163, "step": 118530 }, { "epoch": 75.55130656469089, "grad_norm": 0.3632296919822693, "learning_rate": 7.081955190636185e-05, "loss": 0.0108, "step": 118540 }, { "epoch": 75.55768005098788, "grad_norm": 0.5030195713043213, "learning_rate": 7.081454074324332e-05, "loss": 0.0118, "step": 118550 }, { "epoch": 75.56405353728489, "grad_norm": 0.6293888092041016, "learning_rate": 7.080952932720845e-05, "loss": 0.0154, "step": 118560 }, { "epoch": 75.5704270235819, "grad_norm": 0.4529464840888977, "learning_rate": 7.080451765831817e-05, "loss": 0.0201, "step": 118570 }, { "epoch": 75.57680050987891, "grad_norm": 0.31795647740364075, "learning_rate": 7.079950573663336e-05, "loss": 0.0163, "step": 118580 }, { "epoch": 75.5831739961759, "grad_norm": 0.3560052216053009, "learning_rate": 7.079449356221489e-05, "loss": 0.0167, "step": 118590 }, { "epoch": 75.58954748247291, "grad_norm": 0.20952339470386505, "learning_rate": 7.078948113512372e-05, "loss": 0.0188, "step": 118600 }, { "epoch": 75.59592096876992, "grad_norm": 0.4707447588443756, "learning_rate": 7.07844684554207e-05, "loss": 0.0256, "step": 118610 }, { "epoch": 75.60229445506693, "grad_norm": 0.7181344628334045, "learning_rate": 7.077945552316678e-05, "loss": 0.024, "step": 118620 }, { "epoch": 75.60866794136393, "grad_norm": 0.5379713177680969, "learning_rate": 7.077444233842285e-05, "loss": 0.017, "step": 118630 }, { "epoch": 75.61504142766093, "grad_norm": 0.4485417306423187, "learning_rate": 7.076942890124983e-05, "loss": 0.0243, "step": 118640 }, { "epoch": 75.62141491395793, "grad_norm": 0.3172924220561981, "learning_rate": 7.076441521170865e-05, "loss": 0.0176, "step": 118650 }, { "epoch": 75.62778840025494, "grad_norm": 0.49425095319747925, "learning_rate": 7.07594012698602e-05, "loss": 0.0162, "step": 118660 }, { "epoch": 75.63416188655195, "grad_norm": 0.26472529768943787, "learning_rate": 7.075438707576544e-05, "loss": 0.0183, "step": 118670 }, { "epoch": 75.64053537284894, "grad_norm": 0.3454269468784332, "learning_rate": 7.074937262948527e-05, "loss": 0.0124, "step": 118680 }, { "epoch": 75.64690885914595, "grad_norm": 0.6954070925712585, "learning_rate": 7.074435793108065e-05, "loss": 0.021, "step": 118690 }, { "epoch": 75.65328234544296, "grad_norm": 0.2862941026687622, "learning_rate": 7.073934298061247e-05, "loss": 0.0186, "step": 118700 }, { "epoch": 75.65965583173997, "grad_norm": 0.4144957363605499, "learning_rate": 7.073432777814172e-05, "loss": 0.0115, "step": 118710 }, { "epoch": 75.66602931803696, "grad_norm": 0.1956450641155243, "learning_rate": 7.072931232372929e-05, "loss": 0.0213, "step": 118720 }, { "epoch": 75.67240280433397, "grad_norm": 0.42568740248680115, "learning_rate": 7.072429661743616e-05, "loss": 0.0172, "step": 118730 }, { "epoch": 75.67877629063098, "grad_norm": 0.2669760584831238, "learning_rate": 7.071928065932324e-05, "loss": 0.0211, "step": 118740 }, { "epoch": 75.68514977692799, "grad_norm": 0.33795788884162903, "learning_rate": 7.071426444945151e-05, "loss": 0.0183, "step": 118750 }, { "epoch": 75.69152326322498, "grad_norm": 0.46806928515434265, "learning_rate": 7.070924798788191e-05, "loss": 0.0206, "step": 118760 }, { "epoch": 75.69789674952199, "grad_norm": 0.48012083768844604, "learning_rate": 7.070423127467539e-05, "loss": 0.0123, "step": 118770 }, { "epoch": 75.704270235819, "grad_norm": 0.2523433268070221, "learning_rate": 7.069921430989292e-05, "loss": 0.0118, "step": 118780 }, { "epoch": 75.710643722116, "grad_norm": 0.3420585095882416, "learning_rate": 7.069419709359544e-05, "loss": 0.0164, "step": 118790 }, { "epoch": 75.717017208413, "grad_norm": 0.29377248883247375, "learning_rate": 7.068917962584394e-05, "loss": 0.0119, "step": 118800 }, { "epoch": 75.72339069471, "grad_norm": 0.5734994411468506, "learning_rate": 7.068416190669937e-05, "loss": 0.0152, "step": 118810 }, { "epoch": 75.72976418100701, "grad_norm": 0.3722682595252991, "learning_rate": 7.067914393622269e-05, "loss": 0.0165, "step": 118820 }, { "epoch": 75.73613766730402, "grad_norm": 0.42218488454818726, "learning_rate": 7.06741257144749e-05, "loss": 0.0111, "step": 118830 }, { "epoch": 75.74251115360101, "grad_norm": 0.5269044637680054, "learning_rate": 7.066910724151696e-05, "loss": 0.0204, "step": 118840 }, { "epoch": 75.74888463989802, "grad_norm": 0.3886585235595703, "learning_rate": 7.066408851740986e-05, "loss": 0.0235, "step": 118850 }, { "epoch": 75.75525812619503, "grad_norm": 0.25223931670188904, "learning_rate": 7.065906954221455e-05, "loss": 0.0163, "step": 118860 }, { "epoch": 75.76163161249204, "grad_norm": 1.0044302940368652, "learning_rate": 7.065405031599207e-05, "loss": 0.0149, "step": 118870 }, { "epoch": 75.76800509878903, "grad_norm": 0.3749907314777374, "learning_rate": 7.064903083880335e-05, "loss": 0.0181, "step": 118880 }, { "epoch": 75.77437858508604, "grad_norm": 0.26379361748695374, "learning_rate": 7.064401111070944e-05, "loss": 0.0131, "step": 118890 }, { "epoch": 75.78075207138305, "grad_norm": 0.7485328912734985, "learning_rate": 7.063899113177128e-05, "loss": 0.025, "step": 118900 }, { "epoch": 75.78712555768006, "grad_norm": 0.3414473235607147, "learning_rate": 7.06339709020499e-05, "loss": 0.0158, "step": 118910 }, { "epoch": 75.79349904397705, "grad_norm": 0.32937902212142944, "learning_rate": 7.06289504216063e-05, "loss": 0.0221, "step": 118920 }, { "epoch": 75.79987253027406, "grad_norm": 0.34725791215896606, "learning_rate": 7.062392969050145e-05, "loss": 0.0146, "step": 118930 }, { "epoch": 75.80624601657107, "grad_norm": 0.7936139702796936, "learning_rate": 7.061890870879642e-05, "loss": 0.0205, "step": 118940 }, { "epoch": 75.81261950286807, "grad_norm": 0.26154300570487976, "learning_rate": 7.061388747655214e-05, "loss": 0.0149, "step": 118950 }, { "epoch": 75.81899298916507, "grad_norm": 0.4940328598022461, "learning_rate": 7.060886599382969e-05, "loss": 0.0205, "step": 118960 }, { "epoch": 75.82536647546208, "grad_norm": 0.27195629477500916, "learning_rate": 7.060384426069006e-05, "loss": 0.0206, "step": 118970 }, { "epoch": 75.83173996175908, "grad_norm": 0.38097116351127625, "learning_rate": 7.059882227719427e-05, "loss": 0.0167, "step": 118980 }, { "epoch": 75.83811344805609, "grad_norm": 0.23440241813659668, "learning_rate": 7.059380004340333e-05, "loss": 0.0158, "step": 118990 }, { "epoch": 75.84448693435309, "grad_norm": 0.606343150138855, "learning_rate": 7.058877755937828e-05, "loss": 0.0243, "step": 119000 }, { "epoch": 75.8508604206501, "grad_norm": 0.3098571002483368, "learning_rate": 7.058375482518014e-05, "loss": 0.0232, "step": 119010 }, { "epoch": 75.8572339069471, "grad_norm": 0.34063032269477844, "learning_rate": 7.057873184086994e-05, "loss": 0.012, "step": 119020 }, { "epoch": 75.86360739324411, "grad_norm": 0.4766545295715332, "learning_rate": 7.057370860650873e-05, "loss": 0.0121, "step": 119030 }, { "epoch": 75.8699808795411, "grad_norm": 0.21780772507190704, "learning_rate": 7.056868512215754e-05, "loss": 0.0241, "step": 119040 }, { "epoch": 75.87635436583811, "grad_norm": 0.426311731338501, "learning_rate": 7.05636613878774e-05, "loss": 0.0182, "step": 119050 }, { "epoch": 75.88272785213512, "grad_norm": 0.3660537302494049, "learning_rate": 7.055863740372936e-05, "loss": 0.013, "step": 119060 }, { "epoch": 75.88910133843213, "grad_norm": 0.2848835587501526, "learning_rate": 7.055361316977446e-05, "loss": 0.012, "step": 119070 }, { "epoch": 75.89547482472912, "grad_norm": 0.27099817991256714, "learning_rate": 7.054858868607377e-05, "loss": 0.0162, "step": 119080 }, { "epoch": 75.90184831102613, "grad_norm": 0.31384721398353577, "learning_rate": 7.05435639526883e-05, "loss": 0.0175, "step": 119090 }, { "epoch": 75.90822179732314, "grad_norm": 0.2895788252353668, "learning_rate": 7.053853896967914e-05, "loss": 0.0185, "step": 119100 }, { "epoch": 75.91459528362014, "grad_norm": 0.38014233112335205, "learning_rate": 7.053351373710735e-05, "loss": 0.0115, "step": 119110 }, { "epoch": 75.92096876991714, "grad_norm": 0.36204490065574646, "learning_rate": 7.052848825503398e-05, "loss": 0.0129, "step": 119120 }, { "epoch": 75.92734225621415, "grad_norm": 0.2689323127269745, "learning_rate": 7.052346252352008e-05, "loss": 0.0125, "step": 119130 }, { "epoch": 75.93371574251115, "grad_norm": 0.39021265506744385, "learning_rate": 7.051843654262676e-05, "loss": 0.0173, "step": 119140 }, { "epoch": 75.94008922880816, "grad_norm": 0.3947398066520691, "learning_rate": 7.051341031241505e-05, "loss": 0.0185, "step": 119150 }, { "epoch": 75.94646271510516, "grad_norm": 0.3476082384586334, "learning_rate": 7.050838383294603e-05, "loss": 0.0281, "step": 119160 }, { "epoch": 75.95283620140216, "grad_norm": 0.2957853078842163, "learning_rate": 7.050335710428078e-05, "loss": 0.0234, "step": 119170 }, { "epoch": 75.95920968769917, "grad_norm": 0.28981685638427734, "learning_rate": 7.04983301264804e-05, "loss": 0.0138, "step": 119180 }, { "epoch": 75.96558317399618, "grad_norm": 0.37735751271247864, "learning_rate": 7.049330289960594e-05, "loss": 0.0166, "step": 119190 }, { "epoch": 75.97195666029317, "grad_norm": 0.38681334257125854, "learning_rate": 7.048827542371851e-05, "loss": 0.0405, "step": 119200 }, { "epoch": 75.97833014659018, "grad_norm": 0.32490020990371704, "learning_rate": 7.048324769887919e-05, "loss": 0.0129, "step": 119210 }, { "epoch": 75.98470363288719, "grad_norm": 0.5360503196716309, "learning_rate": 7.047821972514907e-05, "loss": 0.015, "step": 119220 }, { "epoch": 75.9910771191842, "grad_norm": 0.7611535787582397, "learning_rate": 7.047319150258925e-05, "loss": 0.0205, "step": 119230 }, { "epoch": 75.99745060548119, "grad_norm": 0.2923852205276489, "learning_rate": 7.04681630312608e-05, "loss": 0.0196, "step": 119240 }, { "epoch": 76.0038240917782, "grad_norm": 0.5442422032356262, "learning_rate": 7.046313431122487e-05, "loss": 0.0153, "step": 119250 }, { "epoch": 76.0101975780752, "grad_norm": 0.47718143463134766, "learning_rate": 7.045810534254252e-05, "loss": 0.0266, "step": 119260 }, { "epoch": 76.01657106437222, "grad_norm": 0.5469136238098145, "learning_rate": 7.045307612527489e-05, "loss": 0.0178, "step": 119270 }, { "epoch": 76.02294455066922, "grad_norm": 0.6368625164031982, "learning_rate": 7.044804665948305e-05, "loss": 0.0258, "step": 119280 }, { "epoch": 76.02931803696622, "grad_norm": 0.48599711060523987, "learning_rate": 7.044301694522817e-05, "loss": 0.0169, "step": 119290 }, { "epoch": 76.03569152326322, "grad_norm": 0.33294788002967834, "learning_rate": 7.043798698257132e-05, "loss": 0.0151, "step": 119300 }, { "epoch": 76.04206500956023, "grad_norm": 0.3763900399208069, "learning_rate": 7.043295677157362e-05, "loss": 0.0147, "step": 119310 }, { "epoch": 76.04843849585724, "grad_norm": 0.3118327856063843, "learning_rate": 7.04279263122962e-05, "loss": 0.0105, "step": 119320 }, { "epoch": 76.05481198215423, "grad_norm": 0.25498318672180176, "learning_rate": 7.042289560480023e-05, "loss": 0.0137, "step": 119330 }, { "epoch": 76.06118546845124, "grad_norm": 0.4376508593559265, "learning_rate": 7.041786464914674e-05, "loss": 0.0182, "step": 119340 }, { "epoch": 76.06755895474825, "grad_norm": 0.23315739631652832, "learning_rate": 7.041283344539696e-05, "loss": 0.0135, "step": 119350 }, { "epoch": 76.07393244104526, "grad_norm": 0.4076998829841614, "learning_rate": 7.040780199361195e-05, "loss": 0.0225, "step": 119360 }, { "epoch": 76.08030592734225, "grad_norm": 0.24937200546264648, "learning_rate": 7.04027702938529e-05, "loss": 0.0123, "step": 119370 }, { "epoch": 76.08667941363926, "grad_norm": 0.5979923009872437, "learning_rate": 7.039773834618092e-05, "loss": 0.0256, "step": 119380 }, { "epoch": 76.09305289993627, "grad_norm": 0.34047478437423706, "learning_rate": 7.039270615065715e-05, "loss": 0.0141, "step": 119390 }, { "epoch": 76.09942638623328, "grad_norm": 0.41216030716896057, "learning_rate": 7.038767370734276e-05, "loss": 0.0159, "step": 119400 }, { "epoch": 76.10579987253027, "grad_norm": 0.48933830857276917, "learning_rate": 7.038264101629888e-05, "loss": 0.014, "step": 119410 }, { "epoch": 76.11217335882728, "grad_norm": 0.44872793555259705, "learning_rate": 7.037760807758666e-05, "loss": 0.0229, "step": 119420 }, { "epoch": 76.11854684512429, "grad_norm": 0.7171058654785156, "learning_rate": 7.037257489126728e-05, "loss": 0.0225, "step": 119430 }, { "epoch": 76.1249203314213, "grad_norm": 0.46266770362854004, "learning_rate": 7.036754145740186e-05, "loss": 0.0198, "step": 119440 }, { "epoch": 76.13129381771829, "grad_norm": 0.42266812920570374, "learning_rate": 7.03625077760516e-05, "loss": 0.0183, "step": 119450 }, { "epoch": 76.1376673040153, "grad_norm": 0.31465527415275574, "learning_rate": 7.035747384727764e-05, "loss": 0.017, "step": 119460 }, { "epoch": 76.1440407903123, "grad_norm": 0.3554028570652008, "learning_rate": 7.035243967114114e-05, "loss": 0.0307, "step": 119470 }, { "epoch": 76.15041427660931, "grad_norm": 0.3877589702606201, "learning_rate": 7.03474052477033e-05, "loss": 0.0348, "step": 119480 }, { "epoch": 76.1567877629063, "grad_norm": 0.3062985837459564, "learning_rate": 7.034237057702525e-05, "loss": 0.025, "step": 119490 }, { "epoch": 76.16316124920331, "grad_norm": 0.2593017816543579, "learning_rate": 7.033733565916821e-05, "loss": 0.0178, "step": 119500 }, { "epoch": 76.16953473550032, "grad_norm": 0.5586726665496826, "learning_rate": 7.033230049419333e-05, "loss": 0.0281, "step": 119510 }, { "epoch": 76.17590822179733, "grad_norm": 0.29772013425827026, "learning_rate": 7.032726508216182e-05, "loss": 0.0137, "step": 119520 }, { "epoch": 76.18228170809432, "grad_norm": 0.25305941700935364, "learning_rate": 7.032222942313482e-05, "loss": 0.016, "step": 119530 }, { "epoch": 76.18865519439133, "grad_norm": 0.3606245517730713, "learning_rate": 7.031719351717357e-05, "loss": 0.0132, "step": 119540 }, { "epoch": 76.19502868068834, "grad_norm": 0.26501190662384033, "learning_rate": 7.031215736433922e-05, "loss": 0.0102, "step": 119550 }, { "epoch": 76.20140216698535, "grad_norm": 0.2294761836528778, "learning_rate": 7.030712096469299e-05, "loss": 0.0139, "step": 119560 }, { "epoch": 76.20777565328234, "grad_norm": 0.3854679465293884, "learning_rate": 7.030208431829607e-05, "loss": 0.0108, "step": 119570 }, { "epoch": 76.21414913957935, "grad_norm": 0.3587517738342285, "learning_rate": 7.029704742520966e-05, "loss": 0.0148, "step": 119580 }, { "epoch": 76.22052262587636, "grad_norm": 0.29181337356567383, "learning_rate": 7.029201028549493e-05, "loss": 0.0117, "step": 119590 }, { "epoch": 76.22689611217336, "grad_norm": 0.26849082112312317, "learning_rate": 7.028697289921316e-05, "loss": 0.0198, "step": 119600 }, { "epoch": 76.23326959847036, "grad_norm": 0.4411601424217224, "learning_rate": 7.028193526642548e-05, "loss": 0.0167, "step": 119610 }, { "epoch": 76.23964308476737, "grad_norm": 0.7212768793106079, "learning_rate": 7.027689738719315e-05, "loss": 0.0248, "step": 119620 }, { "epoch": 76.24601657106437, "grad_norm": 0.3983801305294037, "learning_rate": 7.027185926157738e-05, "loss": 0.0128, "step": 119630 }, { "epoch": 76.25239005736138, "grad_norm": 0.2677469551563263, "learning_rate": 7.026682088963938e-05, "loss": 0.0118, "step": 119640 }, { "epoch": 76.25876354365838, "grad_norm": 0.4805612564086914, "learning_rate": 7.026178227144036e-05, "loss": 0.0099, "step": 119650 }, { "epoch": 76.26513702995538, "grad_norm": 0.6268524527549744, "learning_rate": 7.025674340704156e-05, "loss": 0.0196, "step": 119660 }, { "epoch": 76.27151051625239, "grad_norm": 0.22509434819221497, "learning_rate": 7.025170429650419e-05, "loss": 0.0243, "step": 119670 }, { "epoch": 76.2778840025494, "grad_norm": 0.2901289165019989, "learning_rate": 7.02466649398895e-05, "loss": 0.0131, "step": 119680 }, { "epoch": 76.2842574888464, "grad_norm": 0.31110355257987976, "learning_rate": 7.024162533725873e-05, "loss": 0.0077, "step": 119690 }, { "epoch": 76.2906309751434, "grad_norm": 0.3637370467185974, "learning_rate": 7.023658548867308e-05, "loss": 0.0169, "step": 119700 }, { "epoch": 76.29700446144041, "grad_norm": 0.3088567852973938, "learning_rate": 7.023154539419384e-05, "loss": 0.0151, "step": 119710 }, { "epoch": 76.30337794773742, "grad_norm": 0.22625067830085754, "learning_rate": 7.02265050538822e-05, "loss": 0.0118, "step": 119720 }, { "epoch": 76.30975143403441, "grad_norm": 0.3720156252384186, "learning_rate": 7.022146446779942e-05, "loss": 0.0162, "step": 119730 }, { "epoch": 76.31612492033142, "grad_norm": 0.4158764183521271, "learning_rate": 7.021642363600676e-05, "loss": 0.0221, "step": 119740 }, { "epoch": 76.32249840662843, "grad_norm": 0.3193633556365967, "learning_rate": 7.021138255856547e-05, "loss": 0.0147, "step": 119750 }, { "epoch": 76.32887189292543, "grad_norm": 0.7858723998069763, "learning_rate": 7.02063412355368e-05, "loss": 0.0164, "step": 119760 }, { "epoch": 76.33524537922243, "grad_norm": 0.5369969606399536, "learning_rate": 7.020129966698198e-05, "loss": 0.0253, "step": 119770 }, { "epoch": 76.34161886551944, "grad_norm": 0.5215790867805481, "learning_rate": 7.019625785296233e-05, "loss": 0.0154, "step": 119780 }, { "epoch": 76.34799235181644, "grad_norm": 0.30377906560897827, "learning_rate": 7.019121579353906e-05, "loss": 0.0127, "step": 119790 }, { "epoch": 76.35436583811345, "grad_norm": 0.363247275352478, "learning_rate": 7.018617348877346e-05, "loss": 0.0168, "step": 119800 }, { "epoch": 76.36073932441045, "grad_norm": 0.4339054822921753, "learning_rate": 7.01811309387268e-05, "loss": 0.0186, "step": 119810 }, { "epoch": 76.36711281070745, "grad_norm": 0.33712658286094666, "learning_rate": 7.017608814346035e-05, "loss": 0.0147, "step": 119820 }, { "epoch": 76.37348629700446, "grad_norm": 0.4331188499927521, "learning_rate": 7.017104510303536e-05, "loss": 0.0155, "step": 119830 }, { "epoch": 76.37985978330147, "grad_norm": 0.46566882729530334, "learning_rate": 7.016600181751313e-05, "loss": 0.0218, "step": 119840 }, { "epoch": 76.38623326959846, "grad_norm": 0.5132261514663696, "learning_rate": 7.016095828695496e-05, "loss": 0.0352, "step": 119850 }, { "epoch": 76.39260675589547, "grad_norm": 0.32452791929244995, "learning_rate": 7.01559145114221e-05, "loss": 0.0147, "step": 119860 }, { "epoch": 76.39898024219248, "grad_norm": 0.387556791305542, "learning_rate": 7.015087049097584e-05, "loss": 0.0115, "step": 119870 }, { "epoch": 76.40535372848949, "grad_norm": 0.6907685995101929, "learning_rate": 7.014582622567747e-05, "loss": 0.0205, "step": 119880 }, { "epoch": 76.4117272147865, "grad_norm": 0.5552417039871216, "learning_rate": 7.014078171558831e-05, "loss": 0.0104, "step": 119890 }, { "epoch": 76.41810070108349, "grad_norm": 0.23790223896503448, "learning_rate": 7.013573696076964e-05, "loss": 0.0222, "step": 119900 }, { "epoch": 76.4244741873805, "grad_norm": 0.41896069049835205, "learning_rate": 7.013069196128274e-05, "loss": 0.0186, "step": 119910 }, { "epoch": 76.4308476736775, "grad_norm": 0.5840412378311157, "learning_rate": 7.012564671718893e-05, "loss": 0.0172, "step": 119920 }, { "epoch": 76.43722115997451, "grad_norm": 0.43002501130104065, "learning_rate": 7.01206012285495e-05, "loss": 0.0159, "step": 119930 }, { "epoch": 76.4435946462715, "grad_norm": 0.2577081024646759, "learning_rate": 7.011555549542578e-05, "loss": 0.0177, "step": 119940 }, { "epoch": 76.44996813256851, "grad_norm": 0.4418856203556061, "learning_rate": 7.011050951787908e-05, "loss": 0.02, "step": 119950 }, { "epoch": 76.45634161886552, "grad_norm": 0.5544539093971252, "learning_rate": 7.010546329597069e-05, "loss": 0.0164, "step": 119960 }, { "epoch": 76.46271510516253, "grad_norm": 0.4831506907939911, "learning_rate": 7.010041682976193e-05, "loss": 0.0182, "step": 119970 }, { "epoch": 76.46908859145952, "grad_norm": 0.25018179416656494, "learning_rate": 7.009537011931413e-05, "loss": 0.0151, "step": 119980 }, { "epoch": 76.47546207775653, "grad_norm": 0.5172871947288513, "learning_rate": 7.009032316468861e-05, "loss": 0.0204, "step": 119990 }, { "epoch": 76.48183556405354, "grad_norm": 0.5293229818344116, "learning_rate": 7.00852759659467e-05, "loss": 0.0251, "step": 120000 }, { "epoch": 76.48820905035055, "grad_norm": 0.4393138289451599, "learning_rate": 7.008022852314974e-05, "loss": 0.0107, "step": 120010 }, { "epoch": 76.49458253664754, "grad_norm": 0.34186798334121704, "learning_rate": 7.007518083635903e-05, "loss": 0.0201, "step": 120020 }, { "epoch": 76.50095602294455, "grad_norm": 0.3845565617084503, "learning_rate": 7.00701329056359e-05, "loss": 0.0134, "step": 120030 }, { "epoch": 76.50732950924156, "grad_norm": 0.2562335133552551, "learning_rate": 7.006508473104174e-05, "loss": 0.0264, "step": 120040 }, { "epoch": 76.51370299553857, "grad_norm": 0.4415591061115265, "learning_rate": 7.006003631263784e-05, "loss": 0.038, "step": 120050 }, { "epoch": 76.52007648183556, "grad_norm": 0.3340368866920471, "learning_rate": 7.005498765048554e-05, "loss": 0.0135, "step": 120060 }, { "epoch": 76.52644996813257, "grad_norm": 0.5183016061782837, "learning_rate": 7.004993874464623e-05, "loss": 0.0199, "step": 120070 }, { "epoch": 76.53282345442958, "grad_norm": 0.6553449034690857, "learning_rate": 7.004488959518123e-05, "loss": 0.0144, "step": 120080 }, { "epoch": 76.53919694072658, "grad_norm": 0.2988850176334381, "learning_rate": 7.003984020215188e-05, "loss": 0.0146, "step": 120090 }, { "epoch": 76.54557042702358, "grad_norm": 0.2388104796409607, "learning_rate": 7.003479056561957e-05, "loss": 0.0096, "step": 120100 }, { "epoch": 76.55194391332059, "grad_norm": 0.19145923852920532, "learning_rate": 7.002974068564563e-05, "loss": 0.0208, "step": 120110 }, { "epoch": 76.5583173996176, "grad_norm": 0.6082251667976379, "learning_rate": 7.002469056229144e-05, "loss": 0.0157, "step": 120120 }, { "epoch": 76.5646908859146, "grad_norm": 0.6472329497337341, "learning_rate": 7.001964019561836e-05, "loss": 0.0251, "step": 120130 }, { "epoch": 76.5710643722116, "grad_norm": 0.3972470760345459, "learning_rate": 7.001458958568772e-05, "loss": 0.0127, "step": 120140 }, { "epoch": 76.5774378585086, "grad_norm": 0.6075165867805481, "learning_rate": 7.000953873256094e-05, "loss": 0.0222, "step": 120150 }, { "epoch": 76.58381134480561, "grad_norm": 0.18439415097236633, "learning_rate": 7.000448763629936e-05, "loss": 0.0107, "step": 120160 }, { "epoch": 76.59018483110262, "grad_norm": 0.3768170177936554, "learning_rate": 6.999943629696437e-05, "loss": 0.0156, "step": 120170 }, { "epoch": 76.59655831739961, "grad_norm": 0.4187471568584442, "learning_rate": 6.999438471461735e-05, "loss": 0.0261, "step": 120180 }, { "epoch": 76.60293180369662, "grad_norm": 0.4538813829421997, "learning_rate": 6.998933288931968e-05, "loss": 0.0151, "step": 120190 }, { "epoch": 76.60930528999363, "grad_norm": 0.2580612003803253, "learning_rate": 6.998428082113273e-05, "loss": 0.0107, "step": 120200 }, { "epoch": 76.61567877629064, "grad_norm": 0.7332178354263306, "learning_rate": 6.997922851011792e-05, "loss": 0.018, "step": 120210 }, { "epoch": 76.62205226258763, "grad_norm": 0.25342512130737305, "learning_rate": 6.997417595633661e-05, "loss": 0.0139, "step": 120220 }, { "epoch": 76.62842574888464, "grad_norm": 0.19051136076450348, "learning_rate": 6.99691231598502e-05, "loss": 0.0125, "step": 120230 }, { "epoch": 76.63479923518165, "grad_norm": 0.374942809343338, "learning_rate": 6.99640701207201e-05, "loss": 0.0131, "step": 120240 }, { "epoch": 76.64117272147865, "grad_norm": 0.32361602783203125, "learning_rate": 6.99590168390077e-05, "loss": 0.0138, "step": 120250 }, { "epoch": 76.64754620777565, "grad_norm": 0.32478588819503784, "learning_rate": 6.995396331477437e-05, "loss": 0.0157, "step": 120260 }, { "epoch": 76.65391969407266, "grad_norm": 0.7987448573112488, "learning_rate": 6.994890954808157e-05, "loss": 0.0168, "step": 120270 }, { "epoch": 76.66029318036966, "grad_norm": 0.38661864399909973, "learning_rate": 6.994385553899069e-05, "loss": 0.0118, "step": 120280 }, { "epoch": 76.66666666666667, "grad_norm": 0.4990396797657013, "learning_rate": 6.993880128756314e-05, "loss": 0.0117, "step": 120290 }, { "epoch": 76.67304015296367, "grad_norm": 0.2263706475496292, "learning_rate": 6.993374679386031e-05, "loss": 0.0157, "step": 120300 }, { "epoch": 76.67941363926067, "grad_norm": 0.25576871633529663, "learning_rate": 6.992869205794365e-05, "loss": 0.0192, "step": 120310 }, { "epoch": 76.68578712555768, "grad_norm": 0.4239945411682129, "learning_rate": 6.992363707987455e-05, "loss": 0.0106, "step": 120320 }, { "epoch": 76.69216061185469, "grad_norm": 0.5215150713920593, "learning_rate": 6.991858185971447e-05, "loss": 0.0165, "step": 120330 }, { "epoch": 76.69853409815168, "grad_norm": 0.39891546964645386, "learning_rate": 6.991352639752479e-05, "loss": 0.0132, "step": 120340 }, { "epoch": 76.70490758444869, "grad_norm": 0.5601545572280884, "learning_rate": 6.990847069336698e-05, "loss": 0.013, "step": 120350 }, { "epoch": 76.7112810707457, "grad_norm": 0.4320802390575409, "learning_rate": 6.990341474730244e-05, "loss": 0.0173, "step": 120360 }, { "epoch": 76.7176545570427, "grad_norm": 0.32223349809646606, "learning_rate": 6.989835855939263e-05, "loss": 0.031, "step": 120370 }, { "epoch": 76.7240280433397, "grad_norm": 0.5285472273826599, "learning_rate": 6.989330212969898e-05, "loss": 0.0155, "step": 120380 }, { "epoch": 76.73040152963671, "grad_norm": 0.5540227890014648, "learning_rate": 6.988824545828292e-05, "loss": 0.0143, "step": 120390 }, { "epoch": 76.73677501593372, "grad_norm": 0.317134290933609, "learning_rate": 6.98831885452059e-05, "loss": 0.0153, "step": 120400 }, { "epoch": 76.74314850223072, "grad_norm": 0.5720508694648743, "learning_rate": 6.987813139052937e-05, "loss": 0.0155, "step": 120410 }, { "epoch": 76.74952198852772, "grad_norm": 0.3130979537963867, "learning_rate": 6.987307399431477e-05, "loss": 0.0219, "step": 120420 }, { "epoch": 76.75589547482473, "grad_norm": 0.3876934349536896, "learning_rate": 6.986801635662356e-05, "loss": 0.0224, "step": 120430 }, { "epoch": 76.76226896112173, "grad_norm": 0.37582266330718994, "learning_rate": 6.986295847751719e-05, "loss": 0.0191, "step": 120440 }, { "epoch": 76.76864244741874, "grad_norm": 0.3939870595932007, "learning_rate": 6.985790035705713e-05, "loss": 0.0131, "step": 120450 }, { "epoch": 76.77501593371574, "grad_norm": 0.3004836440086365, "learning_rate": 6.985284199530481e-05, "loss": 0.0102, "step": 120460 }, { "epoch": 76.78138942001274, "grad_norm": 0.39778104424476624, "learning_rate": 6.984778339232174e-05, "loss": 0.0171, "step": 120470 }, { "epoch": 76.78776290630975, "grad_norm": 0.3575454652309418, "learning_rate": 6.984272454816936e-05, "loss": 0.0205, "step": 120480 }, { "epoch": 76.79413639260676, "grad_norm": 0.35549396276474, "learning_rate": 6.983766546290913e-05, "loss": 0.0219, "step": 120490 }, { "epoch": 76.80050987890377, "grad_norm": 0.4208260178565979, "learning_rate": 6.983260613660254e-05, "loss": 0.0136, "step": 120500 }, { "epoch": 76.80688336520076, "grad_norm": 0.37947744131088257, "learning_rate": 6.982754656931106e-05, "loss": 0.0131, "step": 120510 }, { "epoch": 76.81325685149777, "grad_norm": 0.34810665249824524, "learning_rate": 6.982248676109619e-05, "loss": 0.0204, "step": 120520 }, { "epoch": 76.81963033779478, "grad_norm": 0.5307003259658813, "learning_rate": 6.981742671201936e-05, "loss": 0.0241, "step": 120530 }, { "epoch": 76.82600382409179, "grad_norm": 0.4547986388206482, "learning_rate": 6.98123664221421e-05, "loss": 0.0156, "step": 120540 }, { "epoch": 76.83237731038878, "grad_norm": 0.3267855942249298, "learning_rate": 6.980730589152586e-05, "loss": 0.0179, "step": 120550 }, { "epoch": 76.83875079668579, "grad_norm": 0.40358152985572815, "learning_rate": 6.980224512023218e-05, "loss": 0.0138, "step": 120560 }, { "epoch": 76.8451242829828, "grad_norm": 0.464021772146225, "learning_rate": 6.979718410832251e-05, "loss": 0.0206, "step": 120570 }, { "epoch": 76.8514977692798, "grad_norm": 0.31981149315834045, "learning_rate": 6.979212285585837e-05, "loss": 0.0289, "step": 120580 }, { "epoch": 76.8578712555768, "grad_norm": 0.6751660108566284, "learning_rate": 6.978706136290125e-05, "loss": 0.0203, "step": 120590 }, { "epoch": 76.8642447418738, "grad_norm": 0.6354386806488037, "learning_rate": 6.978199962951264e-05, "loss": 0.0233, "step": 120600 }, { "epoch": 76.87061822817081, "grad_norm": 0.5301569104194641, "learning_rate": 6.977693765575407e-05, "loss": 0.0316, "step": 120610 }, { "epoch": 76.87699171446782, "grad_norm": 0.7087173461914062, "learning_rate": 6.977187544168703e-05, "loss": 0.0229, "step": 120620 }, { "epoch": 76.88336520076481, "grad_norm": 0.4721750319004059, "learning_rate": 6.976681298737304e-05, "loss": 0.0256, "step": 120630 }, { "epoch": 76.88973868706182, "grad_norm": 0.5929004549980164, "learning_rate": 6.976175029287359e-05, "loss": 0.0251, "step": 120640 }, { "epoch": 76.89611217335883, "grad_norm": 0.40192368626594543, "learning_rate": 6.975668735825025e-05, "loss": 0.0206, "step": 120650 }, { "epoch": 76.90248565965584, "grad_norm": 0.778732419013977, "learning_rate": 6.975162418356448e-05, "loss": 0.0328, "step": 120660 }, { "epoch": 76.90885914595283, "grad_norm": 0.3958010971546173, "learning_rate": 6.974656076887783e-05, "loss": 0.0163, "step": 120670 }, { "epoch": 76.91523263224984, "grad_norm": 0.40840908885002136, "learning_rate": 6.974149711425184e-05, "loss": 0.017, "step": 120680 }, { "epoch": 76.92160611854685, "grad_norm": 0.4346381723880768, "learning_rate": 6.9736433219748e-05, "loss": 0.0248, "step": 120690 }, { "epoch": 76.92797960484386, "grad_norm": 0.25700050592422485, "learning_rate": 6.973136908542788e-05, "loss": 0.0242, "step": 120700 }, { "epoch": 76.93435309114085, "grad_norm": 0.5753760933876038, "learning_rate": 6.972630471135297e-05, "loss": 0.0177, "step": 120710 }, { "epoch": 76.94072657743786, "grad_norm": 0.3882943391799927, "learning_rate": 6.972124009758486e-05, "loss": 0.0249, "step": 120720 }, { "epoch": 76.94710006373487, "grad_norm": 0.5235735177993774, "learning_rate": 6.971617524418504e-05, "loss": 0.026, "step": 120730 }, { "epoch": 76.95347355003187, "grad_norm": 0.28797486424446106, "learning_rate": 6.971111015121509e-05, "loss": 0.0185, "step": 120740 }, { "epoch": 76.95984703632887, "grad_norm": 0.49461284279823303, "learning_rate": 6.970604481873654e-05, "loss": 0.0211, "step": 120750 }, { "epoch": 76.96622052262587, "grad_norm": 0.45036908984184265, "learning_rate": 6.970097924681095e-05, "loss": 0.0265, "step": 120760 }, { "epoch": 76.97259400892288, "grad_norm": 0.6251679062843323, "learning_rate": 6.969591343549985e-05, "loss": 0.0161, "step": 120770 }, { "epoch": 76.97896749521989, "grad_norm": 0.3549080789089203, "learning_rate": 6.96908473848648e-05, "loss": 0.0198, "step": 120780 }, { "epoch": 76.98534098151688, "grad_norm": 0.6185206174850464, "learning_rate": 6.968578109496737e-05, "loss": 0.0226, "step": 120790 }, { "epoch": 76.99171446781389, "grad_norm": 0.3077760338783264, "learning_rate": 6.96807145658691e-05, "loss": 0.0228, "step": 120800 }, { "epoch": 76.9980879541109, "grad_norm": 0.218795046210289, "learning_rate": 6.967564779763159e-05, "loss": 0.0135, "step": 120810 }, { "epoch": 77.00446144040791, "grad_norm": 0.296843945980072, "learning_rate": 6.967058079031637e-05, "loss": 0.0256, "step": 120820 }, { "epoch": 77.0108349267049, "grad_norm": 0.32512179017066956, "learning_rate": 6.966551354398502e-05, "loss": 0.0131, "step": 120830 }, { "epoch": 77.01720841300191, "grad_norm": 0.3759230375289917, "learning_rate": 6.966044605869913e-05, "loss": 0.0247, "step": 120840 }, { "epoch": 77.02358189929892, "grad_norm": 0.35798123478889465, "learning_rate": 6.965537833452024e-05, "loss": 0.009, "step": 120850 }, { "epoch": 77.02995538559593, "grad_norm": 0.32153579592704773, "learning_rate": 6.965031037150996e-05, "loss": 0.0254, "step": 120860 }, { "epoch": 77.03632887189292, "grad_norm": 0.32079213857650757, "learning_rate": 6.964524216972983e-05, "loss": 0.0178, "step": 120870 }, { "epoch": 77.04270235818993, "grad_norm": 0.5773524641990662, "learning_rate": 6.964017372924147e-05, "loss": 0.0148, "step": 120880 }, { "epoch": 77.04907584448694, "grad_norm": 0.24179412424564362, "learning_rate": 6.963510505010646e-05, "loss": 0.0124, "step": 120890 }, { "epoch": 77.05544933078394, "grad_norm": 0.6741291880607605, "learning_rate": 6.963003613238639e-05, "loss": 0.0298, "step": 120900 }, { "epoch": 77.06182281708094, "grad_norm": 0.21513473987579346, "learning_rate": 6.962496697614283e-05, "loss": 0.0251, "step": 120910 }, { "epoch": 77.06819630337795, "grad_norm": 0.6130028367042542, "learning_rate": 6.961989758143741e-05, "loss": 0.0165, "step": 120920 }, { "epoch": 77.07456978967495, "grad_norm": 0.1981063336133957, "learning_rate": 6.961482794833171e-05, "loss": 0.0228, "step": 120930 }, { "epoch": 77.08094327597196, "grad_norm": 0.40909233689308167, "learning_rate": 6.960975807688732e-05, "loss": 0.0179, "step": 120940 }, { "epoch": 77.08731676226896, "grad_norm": 0.20108391344547272, "learning_rate": 6.960468796716585e-05, "loss": 0.0117, "step": 120950 }, { "epoch": 77.09369024856596, "grad_norm": 0.4932329058647156, "learning_rate": 6.959961761922894e-05, "loss": 0.016, "step": 120960 }, { "epoch": 77.10006373486297, "grad_norm": 0.460509717464447, "learning_rate": 6.959454703313815e-05, "loss": 0.0191, "step": 120970 }, { "epoch": 77.10643722115998, "grad_norm": 0.3285568356513977, "learning_rate": 6.958947620895511e-05, "loss": 0.0142, "step": 120980 }, { "epoch": 77.11281070745697, "grad_norm": 0.3266829550266266, "learning_rate": 6.958440514674144e-05, "loss": 0.0143, "step": 120990 }, { "epoch": 77.11918419375398, "grad_norm": 0.2708703279495239, "learning_rate": 6.957933384655877e-05, "loss": 0.0104, "step": 121000 }, { "epoch": 77.12555768005099, "grad_norm": 0.3283633887767792, "learning_rate": 6.957426230846868e-05, "loss": 0.0113, "step": 121010 }, { "epoch": 77.131931166348, "grad_norm": 0.13865064084529877, "learning_rate": 6.956919053253286e-05, "loss": 0.0156, "step": 121020 }, { "epoch": 77.13830465264499, "grad_norm": 0.45916593074798584, "learning_rate": 6.956411851881287e-05, "loss": 0.0132, "step": 121030 }, { "epoch": 77.144678138942, "grad_norm": 0.257213294506073, "learning_rate": 6.955904626737038e-05, "loss": 0.013, "step": 121040 }, { "epoch": 77.151051625239, "grad_norm": 0.3984711468219757, "learning_rate": 6.9553973778267e-05, "loss": 0.0233, "step": 121050 }, { "epoch": 77.15742511153601, "grad_norm": 0.49317264556884766, "learning_rate": 6.95489010515644e-05, "loss": 0.0149, "step": 121060 }, { "epoch": 77.16379859783301, "grad_norm": 0.3807309567928314, "learning_rate": 6.954382808732418e-05, "loss": 0.009, "step": 121070 }, { "epoch": 77.17017208413002, "grad_norm": 0.3420100510120392, "learning_rate": 6.953875488560799e-05, "loss": 0.0183, "step": 121080 }, { "epoch": 77.17654557042702, "grad_norm": 0.5449411273002625, "learning_rate": 6.953368144647749e-05, "loss": 0.0222, "step": 121090 }, { "epoch": 77.18291905672403, "grad_norm": 0.25654295086860657, "learning_rate": 6.95286077699943e-05, "loss": 0.0099, "step": 121100 }, { "epoch": 77.18929254302103, "grad_norm": 0.2492886781692505, "learning_rate": 6.952353385622011e-05, "loss": 0.0135, "step": 121110 }, { "epoch": 77.19566602931803, "grad_norm": 0.5579444766044617, "learning_rate": 6.951845970521654e-05, "loss": 0.0172, "step": 121120 }, { "epoch": 77.20203951561504, "grad_norm": 0.3125736713409424, "learning_rate": 6.951338531704526e-05, "loss": 0.0199, "step": 121130 }, { "epoch": 77.20841300191205, "grad_norm": 0.23040452599525452, "learning_rate": 6.950831069176793e-05, "loss": 0.0165, "step": 121140 }, { "epoch": 77.21478648820906, "grad_norm": 0.6188251376152039, "learning_rate": 6.95032358294462e-05, "loss": 0.0207, "step": 121150 }, { "epoch": 77.22115997450605, "grad_norm": 0.2550593614578247, "learning_rate": 6.949816073014174e-05, "loss": 0.015, "step": 121160 }, { "epoch": 77.22753346080306, "grad_norm": 0.33948245644569397, "learning_rate": 6.949308539391622e-05, "loss": 0.0097, "step": 121170 }, { "epoch": 77.23390694710007, "grad_norm": 0.23025597631931305, "learning_rate": 6.94880098208313e-05, "loss": 0.0125, "step": 121180 }, { "epoch": 77.24028043339707, "grad_norm": 0.7384238839149475, "learning_rate": 6.948293401094866e-05, "loss": 0.0222, "step": 121190 }, { "epoch": 77.24665391969407, "grad_norm": 0.265072226524353, "learning_rate": 6.947785796432999e-05, "loss": 0.0141, "step": 121200 }, { "epoch": 77.25302740599108, "grad_norm": 0.3657500147819519, "learning_rate": 6.947278168103694e-05, "loss": 0.0122, "step": 121210 }, { "epoch": 77.25940089228808, "grad_norm": 0.4296131134033203, "learning_rate": 6.946770516113122e-05, "loss": 0.017, "step": 121220 }, { "epoch": 77.26577437858509, "grad_norm": 0.23262923955917358, "learning_rate": 6.94626284046745e-05, "loss": 0.0108, "step": 121230 }, { "epoch": 77.27214786488209, "grad_norm": 0.3266640305519104, "learning_rate": 6.945755141172848e-05, "loss": 0.0155, "step": 121240 }, { "epoch": 77.2785213511791, "grad_norm": 0.35070186853408813, "learning_rate": 6.945247418235482e-05, "loss": 0.0185, "step": 121250 }, { "epoch": 77.2848948374761, "grad_norm": 0.2177131474018097, "learning_rate": 6.944739671661523e-05, "loss": 0.0116, "step": 121260 }, { "epoch": 77.29126832377311, "grad_norm": 0.3152422606945038, "learning_rate": 6.944231901457144e-05, "loss": 0.0116, "step": 121270 }, { "epoch": 77.2976418100701, "grad_norm": 0.18078164756298065, "learning_rate": 6.943724107628508e-05, "loss": 0.0174, "step": 121280 }, { "epoch": 77.30401529636711, "grad_norm": 0.21065227687358856, "learning_rate": 6.943216290181791e-05, "loss": 0.0175, "step": 121290 }, { "epoch": 77.31038878266412, "grad_norm": 0.17832772433757782, "learning_rate": 6.94270844912316e-05, "loss": 0.0089, "step": 121300 }, { "epoch": 77.31676226896113, "grad_norm": 0.45380157232284546, "learning_rate": 6.942200584458788e-05, "loss": 0.0183, "step": 121310 }, { "epoch": 77.32313575525812, "grad_norm": 0.48307523131370544, "learning_rate": 6.941692696194846e-05, "loss": 0.0152, "step": 121320 }, { "epoch": 77.32950924155513, "grad_norm": 0.32400044798851013, "learning_rate": 6.941184784337502e-05, "loss": 0.0168, "step": 121330 }, { "epoch": 77.33588272785214, "grad_norm": 0.3296479880809784, "learning_rate": 6.940676848892933e-05, "loss": 0.0212, "step": 121340 }, { "epoch": 77.34225621414915, "grad_norm": 0.224664106965065, "learning_rate": 6.940168889867306e-05, "loss": 0.0219, "step": 121350 }, { "epoch": 77.34862970044614, "grad_norm": 0.5276059508323669, "learning_rate": 6.939660907266796e-05, "loss": 0.0117, "step": 121360 }, { "epoch": 77.35500318674315, "grad_norm": 0.20155705511569977, "learning_rate": 6.939152901097575e-05, "loss": 0.0187, "step": 121370 }, { "epoch": 77.36137667304016, "grad_norm": 0.24087291955947876, "learning_rate": 6.938644871365814e-05, "loss": 0.0127, "step": 121380 }, { "epoch": 77.36775015933716, "grad_norm": 0.48428475856781006, "learning_rate": 6.938136818077687e-05, "loss": 0.0151, "step": 121390 }, { "epoch": 77.37412364563416, "grad_norm": 0.3187786340713501, "learning_rate": 6.93762874123937e-05, "loss": 0.0161, "step": 121400 }, { "epoch": 77.38049713193116, "grad_norm": 0.511940062046051, "learning_rate": 6.937120640857034e-05, "loss": 0.0155, "step": 121410 }, { "epoch": 77.38687061822817, "grad_norm": 0.3615458309650421, "learning_rate": 6.936612516936852e-05, "loss": 0.0161, "step": 121420 }, { "epoch": 77.39324410452518, "grad_norm": 0.34313055872917175, "learning_rate": 6.936104369485e-05, "loss": 0.0086, "step": 121430 }, { "epoch": 77.39961759082217, "grad_norm": 0.4214589595794678, "learning_rate": 6.935596198507652e-05, "loss": 0.0114, "step": 121440 }, { "epoch": 77.40599107711918, "grad_norm": 0.754612386226654, "learning_rate": 6.935088004010983e-05, "loss": 0.02, "step": 121450 }, { "epoch": 77.41236456341619, "grad_norm": 0.413110613822937, "learning_rate": 6.934579786001167e-05, "loss": 0.0144, "step": 121460 }, { "epoch": 77.4187380497132, "grad_norm": 0.365928590297699, "learning_rate": 6.93407154448438e-05, "loss": 0.0214, "step": 121470 }, { "epoch": 77.42511153601019, "grad_norm": 0.4973861873149872, "learning_rate": 6.9335632794668e-05, "loss": 0.0188, "step": 121480 }, { "epoch": 77.4314850223072, "grad_norm": 0.5191419124603271, "learning_rate": 6.933054990954597e-05, "loss": 0.0238, "step": 121490 }, { "epoch": 77.43785850860421, "grad_norm": 0.4629170000553131, "learning_rate": 6.932546678953953e-05, "loss": 0.0164, "step": 121500 }, { "epoch": 77.44423199490122, "grad_norm": 0.38150128722190857, "learning_rate": 6.932038343471041e-05, "loss": 0.018, "step": 121510 }, { "epoch": 77.45060548119821, "grad_norm": 0.3358640968799591, "learning_rate": 6.931529984512041e-05, "loss": 0.0195, "step": 121520 }, { "epoch": 77.45697896749522, "grad_norm": 0.331556111574173, "learning_rate": 6.931021602083125e-05, "loss": 0.017, "step": 121530 }, { "epoch": 77.46335245379223, "grad_norm": 0.4522685110569, "learning_rate": 6.930513196190477e-05, "loss": 0.0315, "step": 121540 }, { "epoch": 77.46972594008923, "grad_norm": 0.6638670563697815, "learning_rate": 6.930004766840268e-05, "loss": 0.0275, "step": 121550 }, { "epoch": 77.47609942638623, "grad_norm": 0.3923209309577942, "learning_rate": 6.92949631403868e-05, "loss": 0.0131, "step": 121560 }, { "epoch": 77.48247291268324, "grad_norm": 0.4300227761268616, "learning_rate": 6.928987837791891e-05, "loss": 0.02, "step": 121570 }, { "epoch": 77.48884639898024, "grad_norm": 0.4284272789955139, "learning_rate": 6.928479338106077e-05, "loss": 0.0189, "step": 121580 }, { "epoch": 77.49521988527725, "grad_norm": 0.351423442363739, "learning_rate": 6.927970814987419e-05, "loss": 0.0164, "step": 121590 }, { "epoch": 77.50159337157424, "grad_norm": 0.43340548872947693, "learning_rate": 6.927462268442093e-05, "loss": 0.0183, "step": 121600 }, { "epoch": 77.50796685787125, "grad_norm": 0.3433035910129547, "learning_rate": 6.926953698476284e-05, "loss": 0.0174, "step": 121610 }, { "epoch": 77.51434034416826, "grad_norm": 0.46647587418556213, "learning_rate": 6.926445105096167e-05, "loss": 0.0243, "step": 121620 }, { "epoch": 77.52071383046527, "grad_norm": 0.28962430357933044, "learning_rate": 6.925936488307923e-05, "loss": 0.0143, "step": 121630 }, { "epoch": 77.52708731676226, "grad_norm": 0.2893587350845337, "learning_rate": 6.925427848117731e-05, "loss": 0.0114, "step": 121640 }, { "epoch": 77.53346080305927, "grad_norm": 0.559873640537262, "learning_rate": 6.924919184531774e-05, "loss": 0.0219, "step": 121650 }, { "epoch": 77.53983428935628, "grad_norm": 0.34016141295433044, "learning_rate": 6.92441049755623e-05, "loss": 0.0136, "step": 121660 }, { "epoch": 77.54620777565329, "grad_norm": 0.23608146607875824, "learning_rate": 6.923901787197284e-05, "loss": 0.0135, "step": 121670 }, { "epoch": 77.55258126195028, "grad_norm": 0.24244184792041779, "learning_rate": 6.923393053461114e-05, "loss": 0.0191, "step": 121680 }, { "epoch": 77.55895474824729, "grad_norm": 0.5104697942733765, "learning_rate": 6.922884296353901e-05, "loss": 0.0136, "step": 121690 }, { "epoch": 77.5653282345443, "grad_norm": 0.33493471145629883, "learning_rate": 6.922375515881829e-05, "loss": 0.0177, "step": 121700 }, { "epoch": 77.5717017208413, "grad_norm": 0.3044620156288147, "learning_rate": 6.921866712051081e-05, "loss": 0.0199, "step": 121710 }, { "epoch": 77.5780752071383, "grad_norm": 0.372275710105896, "learning_rate": 6.921357884867836e-05, "loss": 0.0178, "step": 121720 }, { "epoch": 77.5844486934353, "grad_norm": 0.4293034076690674, "learning_rate": 6.92084903433828e-05, "loss": 0.0167, "step": 121730 }, { "epoch": 77.59082217973231, "grad_norm": 0.3320325016975403, "learning_rate": 6.920340160468593e-05, "loss": 0.0145, "step": 121740 }, { "epoch": 77.59719566602932, "grad_norm": 0.45226767659187317, "learning_rate": 6.919831263264962e-05, "loss": 0.0179, "step": 121750 }, { "epoch": 77.60356915232632, "grad_norm": 0.37698057293891907, "learning_rate": 6.919322342733567e-05, "loss": 0.0135, "step": 121760 }, { "epoch": 77.60994263862332, "grad_norm": 0.4043039083480835, "learning_rate": 6.918813398880595e-05, "loss": 0.0156, "step": 121770 }, { "epoch": 77.61631612492033, "grad_norm": 0.5637972354888916, "learning_rate": 6.918304431712227e-05, "loss": 0.0195, "step": 121780 }, { "epoch": 77.62268961121734, "grad_norm": 0.30011045932769775, "learning_rate": 6.91779544123465e-05, "loss": 0.013, "step": 121790 }, { "epoch": 77.62906309751435, "grad_norm": 0.3574414849281311, "learning_rate": 6.917286427454047e-05, "loss": 0.0155, "step": 121800 }, { "epoch": 77.63543658381134, "grad_norm": 0.43574419617652893, "learning_rate": 6.916777390376603e-05, "loss": 0.0124, "step": 121810 }, { "epoch": 77.64181007010835, "grad_norm": 0.4471307098865509, "learning_rate": 6.916268330008507e-05, "loss": 0.0185, "step": 121820 }, { "epoch": 77.64818355640536, "grad_norm": 0.3900284469127655, "learning_rate": 6.915759246355937e-05, "loss": 0.0146, "step": 121830 }, { "epoch": 77.65455704270236, "grad_norm": 0.28427740931510925, "learning_rate": 6.915250139425087e-05, "loss": 0.0176, "step": 121840 }, { "epoch": 77.66093052899936, "grad_norm": 0.4913356602191925, "learning_rate": 6.914741009222138e-05, "loss": 0.021, "step": 121850 }, { "epoch": 77.66730401529637, "grad_norm": 0.5813433527946472, "learning_rate": 6.914231855753278e-05, "loss": 0.0187, "step": 121860 }, { "epoch": 77.67367750159337, "grad_norm": 0.2852868139743805, "learning_rate": 6.913722679024693e-05, "loss": 0.0212, "step": 121870 }, { "epoch": 77.68005098789038, "grad_norm": 0.3290872871875763, "learning_rate": 6.913213479042572e-05, "loss": 0.0204, "step": 121880 }, { "epoch": 77.68642447418738, "grad_norm": 0.24395370483398438, "learning_rate": 6.9127042558131e-05, "loss": 0.0116, "step": 121890 }, { "epoch": 77.69279796048438, "grad_norm": 0.312213659286499, "learning_rate": 6.912195009342466e-05, "loss": 0.0213, "step": 121900 }, { "epoch": 77.69917144678139, "grad_norm": 0.35331323742866516, "learning_rate": 6.911685739636858e-05, "loss": 0.0103, "step": 121910 }, { "epoch": 77.7055449330784, "grad_norm": 0.3034515380859375, "learning_rate": 6.91117644670246e-05, "loss": 0.0151, "step": 121920 }, { "epoch": 77.7119184193754, "grad_norm": 0.35450872778892517, "learning_rate": 6.910667130545468e-05, "loss": 0.0138, "step": 121930 }, { "epoch": 77.7182919056724, "grad_norm": 0.26445645093917847, "learning_rate": 6.910157791172062e-05, "loss": 0.0182, "step": 121940 }, { "epoch": 77.72466539196941, "grad_norm": 0.40601012110710144, "learning_rate": 6.909648428588439e-05, "loss": 0.0201, "step": 121950 }, { "epoch": 77.73103887826642, "grad_norm": 0.37410447001457214, "learning_rate": 6.909139042800783e-05, "loss": 0.0244, "step": 121960 }, { "epoch": 77.73741236456341, "grad_norm": 0.2934815287590027, "learning_rate": 6.908629633815285e-05, "loss": 0.0075, "step": 121970 }, { "epoch": 77.74378585086042, "grad_norm": 0.45327824354171753, "learning_rate": 6.908120201638136e-05, "loss": 0.0238, "step": 121980 }, { "epoch": 77.75015933715743, "grad_norm": 0.4717637896537781, "learning_rate": 6.907610746275523e-05, "loss": 0.0233, "step": 121990 }, { "epoch": 77.75653282345444, "grad_norm": 0.3918783366680145, "learning_rate": 6.907101267733641e-05, "loss": 0.0134, "step": 122000 }, { "epoch": 77.76290630975143, "grad_norm": 0.3900119662284851, "learning_rate": 6.906591766018675e-05, "loss": 0.0148, "step": 122010 }, { "epoch": 77.76927979604844, "grad_norm": 0.5096040964126587, "learning_rate": 6.906082241136822e-05, "loss": 0.0176, "step": 122020 }, { "epoch": 77.77565328234544, "grad_norm": 0.6446693539619446, "learning_rate": 6.905572693094268e-05, "loss": 0.031, "step": 122030 }, { "epoch": 77.78202676864245, "grad_norm": 0.3987652659416199, "learning_rate": 6.905063121897207e-05, "loss": 0.0147, "step": 122040 }, { "epoch": 77.78840025493945, "grad_norm": 0.5443147420883179, "learning_rate": 6.904553527551831e-05, "loss": 0.0135, "step": 122050 }, { "epoch": 77.79477374123645, "grad_norm": 0.5124820470809937, "learning_rate": 6.90404391006433e-05, "loss": 0.0159, "step": 122060 }, { "epoch": 77.80114722753346, "grad_norm": 0.3089767098426819, "learning_rate": 6.903534269440902e-05, "loss": 0.0154, "step": 122070 }, { "epoch": 77.80752071383047, "grad_norm": 0.543022096157074, "learning_rate": 6.90302460568773e-05, "loss": 0.0165, "step": 122080 }, { "epoch": 77.81389420012746, "grad_norm": 0.37379905581474304, "learning_rate": 6.902514918811015e-05, "loss": 0.0221, "step": 122090 }, { "epoch": 77.82026768642447, "grad_norm": 0.3803763687610626, "learning_rate": 6.902005208816946e-05, "loss": 0.0123, "step": 122100 }, { "epoch": 77.82664117272148, "grad_norm": 0.21133831143379211, "learning_rate": 6.90149547571172e-05, "loss": 0.0185, "step": 122110 }, { "epoch": 77.83301465901849, "grad_norm": 0.5882111191749573, "learning_rate": 6.900985719501528e-05, "loss": 0.029, "step": 122120 }, { "epoch": 77.83938814531548, "grad_norm": 0.4107116460800171, "learning_rate": 6.900475940192564e-05, "loss": 0.0197, "step": 122130 }, { "epoch": 77.84576163161249, "grad_norm": 0.28202855587005615, "learning_rate": 6.899966137791022e-05, "loss": 0.0173, "step": 122140 }, { "epoch": 77.8521351179095, "grad_norm": 0.2856064736843109, "learning_rate": 6.8994563123031e-05, "loss": 0.0235, "step": 122150 }, { "epoch": 77.8585086042065, "grad_norm": 0.31416404247283936, "learning_rate": 6.89894646373499e-05, "loss": 0.0149, "step": 122160 }, { "epoch": 77.8648820905035, "grad_norm": 0.4551234245300293, "learning_rate": 6.898436592092885e-05, "loss": 0.0192, "step": 122170 }, { "epoch": 77.87125557680051, "grad_norm": 0.42161476612091064, "learning_rate": 6.897926697382986e-05, "loss": 0.0182, "step": 122180 }, { "epoch": 77.87762906309752, "grad_norm": 0.30027756094932556, "learning_rate": 6.897416779611485e-05, "loss": 0.0137, "step": 122190 }, { "epoch": 77.88400254939452, "grad_norm": 0.23313355445861816, "learning_rate": 6.896906838784577e-05, "loss": 0.0136, "step": 122200 }, { "epoch": 77.89037603569152, "grad_norm": 0.33784210681915283, "learning_rate": 6.896396874908462e-05, "loss": 0.018, "step": 122210 }, { "epoch": 77.89674952198853, "grad_norm": 0.48688432574272156, "learning_rate": 6.895886887989334e-05, "loss": 0.0208, "step": 122220 }, { "epoch": 77.90312300828553, "grad_norm": 0.4813896119594574, "learning_rate": 6.89537687803339e-05, "loss": 0.0227, "step": 122230 }, { "epoch": 77.90949649458254, "grad_norm": 0.41594409942626953, "learning_rate": 6.894866845046825e-05, "loss": 0.0151, "step": 122240 }, { "epoch": 77.91586998087953, "grad_norm": 0.5649278163909912, "learning_rate": 6.894356789035843e-05, "loss": 0.015, "step": 122250 }, { "epoch": 77.92224346717654, "grad_norm": 0.47628235816955566, "learning_rate": 6.893846710006635e-05, "loss": 0.0199, "step": 122260 }, { "epoch": 77.92861695347355, "grad_norm": 0.6539028286933899, "learning_rate": 6.893336607965402e-05, "loss": 0.0173, "step": 122270 }, { "epoch": 77.93499043977056, "grad_norm": 0.41415587067604065, "learning_rate": 6.892826482918342e-05, "loss": 0.0102, "step": 122280 }, { "epoch": 77.94136392606755, "grad_norm": 0.34293362498283386, "learning_rate": 6.892316334871652e-05, "loss": 0.0167, "step": 122290 }, { "epoch": 77.94773741236456, "grad_norm": 0.4101725220680237, "learning_rate": 6.891806163831532e-05, "loss": 0.0256, "step": 122300 }, { "epoch": 77.95411089866157, "grad_norm": 0.5718145370483398, "learning_rate": 6.891295969804181e-05, "loss": 0.0225, "step": 122310 }, { "epoch": 77.96048438495858, "grad_norm": 0.6408422589302063, "learning_rate": 6.8907857527958e-05, "loss": 0.0221, "step": 122320 }, { "epoch": 77.96685787125557, "grad_norm": 0.32891184091567993, "learning_rate": 6.890275512812584e-05, "loss": 0.0267, "step": 122330 }, { "epoch": 77.97323135755258, "grad_norm": 0.5236144065856934, "learning_rate": 6.889765249860738e-05, "loss": 0.0148, "step": 122340 }, { "epoch": 77.97960484384959, "grad_norm": 0.11620648205280304, "learning_rate": 6.889254963946459e-05, "loss": 0.016, "step": 122350 }, { "epoch": 77.9859783301466, "grad_norm": 0.6122729182243347, "learning_rate": 6.88874465507595e-05, "loss": 0.0153, "step": 122360 }, { "epoch": 77.99235181644359, "grad_norm": 0.14491863548755646, "learning_rate": 6.888234323255408e-05, "loss": 0.026, "step": 122370 }, { "epoch": 77.9987253027406, "grad_norm": 1.0197968482971191, "learning_rate": 6.887723968491037e-05, "loss": 0.0158, "step": 122380 }, { "epoch": 78.0050987890376, "grad_norm": 0.36731505393981934, "learning_rate": 6.887213590789038e-05, "loss": 0.011, "step": 122390 }, { "epoch": 78.01147227533461, "grad_norm": 0.8004700541496277, "learning_rate": 6.886703190155612e-05, "loss": 0.0169, "step": 122400 }, { "epoch": 78.01784576163162, "grad_norm": 0.5506035089492798, "learning_rate": 6.88619276659696e-05, "loss": 0.0264, "step": 122410 }, { "epoch": 78.02421924792861, "grad_norm": 0.3063824772834778, "learning_rate": 6.885682320119286e-05, "loss": 0.0196, "step": 122420 }, { "epoch": 78.03059273422562, "grad_norm": 0.28915634751319885, "learning_rate": 6.88517185072879e-05, "loss": 0.027, "step": 122430 }, { "epoch": 78.03696622052263, "grad_norm": 0.28449276089668274, "learning_rate": 6.884661358431678e-05, "loss": 0.0205, "step": 122440 }, { "epoch": 78.04333970681964, "grad_norm": 0.2267657369375229, "learning_rate": 6.884150843234149e-05, "loss": 0.0218, "step": 122450 }, { "epoch": 78.04971319311663, "grad_norm": 0.30294060707092285, "learning_rate": 6.883640305142411e-05, "loss": 0.0163, "step": 122460 }, { "epoch": 78.05608667941364, "grad_norm": 0.20734252035617828, "learning_rate": 6.883129744162662e-05, "loss": 0.0113, "step": 122470 }, { "epoch": 78.06246016571065, "grad_norm": 0.7200838923454285, "learning_rate": 6.882619160301112e-05, "loss": 0.0189, "step": 122480 }, { "epoch": 78.06883365200765, "grad_norm": 0.34621474146842957, "learning_rate": 6.882108553563959e-05, "loss": 0.01, "step": 122490 }, { "epoch": 78.07520713830465, "grad_norm": 0.3410166800022125, "learning_rate": 6.88159792395741e-05, "loss": 0.0087, "step": 122500 }, { "epoch": 78.08158062460166, "grad_norm": 0.4359540641307831, "learning_rate": 6.881087271487671e-05, "loss": 0.0119, "step": 122510 }, { "epoch": 78.08795411089866, "grad_norm": 0.19677318632602692, "learning_rate": 6.880576596160945e-05, "loss": 0.0132, "step": 122520 }, { "epoch": 78.09432759719567, "grad_norm": 0.3519301414489746, "learning_rate": 6.880065897983438e-05, "loss": 0.016, "step": 122530 }, { "epoch": 78.10070108349267, "grad_norm": 0.33513250946998596, "learning_rate": 6.879555176961355e-05, "loss": 0.0103, "step": 122540 }, { "epoch": 78.10707456978967, "grad_norm": 0.4781673848628998, "learning_rate": 6.879044433100904e-05, "loss": 0.0156, "step": 122550 }, { "epoch": 78.11344805608668, "grad_norm": 0.7219738364219666, "learning_rate": 6.878533666408286e-05, "loss": 0.0103, "step": 122560 }, { "epoch": 78.11982154238369, "grad_norm": 0.24531111121177673, "learning_rate": 6.878022876889711e-05, "loss": 0.0092, "step": 122570 }, { "epoch": 78.12619502868068, "grad_norm": 0.6606714725494385, "learning_rate": 6.877512064551386e-05, "loss": 0.0185, "step": 122580 }, { "epoch": 78.13256851497769, "grad_norm": 0.36494237184524536, "learning_rate": 6.877001229399516e-05, "loss": 0.0098, "step": 122590 }, { "epoch": 78.1389420012747, "grad_norm": 0.28938886523246765, "learning_rate": 6.876490371440308e-05, "loss": 0.0122, "step": 122600 }, { "epoch": 78.14531548757171, "grad_norm": 0.14077922701835632, "learning_rate": 6.875979490679972e-05, "loss": 0.0133, "step": 122610 }, { "epoch": 78.1516889738687, "grad_norm": 0.5218281149864197, "learning_rate": 6.875468587124714e-05, "loss": 0.0149, "step": 122620 }, { "epoch": 78.15806246016571, "grad_norm": 0.4053407311439514, "learning_rate": 6.874957660780741e-05, "loss": 0.0132, "step": 122630 }, { "epoch": 78.16443594646272, "grad_norm": 0.37988582253456116, "learning_rate": 6.874446711654263e-05, "loss": 0.0146, "step": 122640 }, { "epoch": 78.17080943275973, "grad_norm": 0.32150617241859436, "learning_rate": 6.873935739751486e-05, "loss": 0.0156, "step": 122650 }, { "epoch": 78.17718291905672, "grad_norm": 0.3478430509567261, "learning_rate": 6.873424745078623e-05, "loss": 0.0314, "step": 122660 }, { "epoch": 78.18355640535373, "grad_norm": 0.3678928017616272, "learning_rate": 6.872913727641877e-05, "loss": 0.01, "step": 122670 }, { "epoch": 78.18992989165073, "grad_norm": 0.3248390257358551, "learning_rate": 6.872402687447464e-05, "loss": 0.0143, "step": 122680 }, { "epoch": 78.19630337794774, "grad_norm": 0.27577346563339233, "learning_rate": 6.87189162450159e-05, "loss": 0.0118, "step": 122690 }, { "epoch": 78.20267686424474, "grad_norm": 0.42649486660957336, "learning_rate": 6.871380538810464e-05, "loss": 0.0144, "step": 122700 }, { "epoch": 78.20905035054174, "grad_norm": 0.6105812788009644, "learning_rate": 6.8708694303803e-05, "loss": 0.0178, "step": 122710 }, { "epoch": 78.21542383683875, "grad_norm": 0.4149724245071411, "learning_rate": 6.870358299217303e-05, "loss": 0.0116, "step": 122720 }, { "epoch": 78.22179732313576, "grad_norm": 0.3241637051105499, "learning_rate": 6.869847145327689e-05, "loss": 0.0103, "step": 122730 }, { "epoch": 78.22817080943275, "grad_norm": 0.353291392326355, "learning_rate": 6.869335968717666e-05, "loss": 0.021, "step": 122740 }, { "epoch": 78.23454429572976, "grad_norm": 0.3571188449859619, "learning_rate": 6.868824769393446e-05, "loss": 0.0193, "step": 122750 }, { "epoch": 78.24091778202677, "grad_norm": 0.4770832359790802, "learning_rate": 6.868313547361241e-05, "loss": 0.01, "step": 122760 }, { "epoch": 78.24729126832378, "grad_norm": 0.2476295828819275, "learning_rate": 6.867802302627262e-05, "loss": 0.0118, "step": 122770 }, { "epoch": 78.25366475462077, "grad_norm": 0.191560760140419, "learning_rate": 6.867291035197723e-05, "loss": 0.0215, "step": 122780 }, { "epoch": 78.26003824091778, "grad_norm": 0.36439791321754456, "learning_rate": 6.866779745078832e-05, "loss": 0.0168, "step": 122790 }, { "epoch": 78.26641172721479, "grad_norm": 0.38987231254577637, "learning_rate": 6.866268432276806e-05, "loss": 0.0233, "step": 122800 }, { "epoch": 78.2727852135118, "grad_norm": 0.42037877440452576, "learning_rate": 6.865757096797856e-05, "loss": 0.0165, "step": 122810 }, { "epoch": 78.27915869980879, "grad_norm": 0.4059255123138428, "learning_rate": 6.865245738648197e-05, "loss": 0.0117, "step": 122820 }, { "epoch": 78.2855321861058, "grad_norm": 0.4492069482803345, "learning_rate": 6.86473435783404e-05, "loss": 0.0124, "step": 122830 }, { "epoch": 78.2919056724028, "grad_norm": 0.4361444413661957, "learning_rate": 6.8642229543616e-05, "loss": 0.024, "step": 122840 }, { "epoch": 78.29827915869981, "grad_norm": 0.35346418619155884, "learning_rate": 6.863711528237093e-05, "loss": 0.0167, "step": 122850 }, { "epoch": 78.3046526449968, "grad_norm": 0.3377143442630768, "learning_rate": 6.86320007946673e-05, "loss": 0.0265, "step": 122860 }, { "epoch": 78.31102613129381, "grad_norm": 0.32617509365081787, "learning_rate": 6.862688608056727e-05, "loss": 0.0119, "step": 122870 }, { "epoch": 78.31739961759082, "grad_norm": 0.45718562602996826, "learning_rate": 6.862177114013297e-05, "loss": 0.0324, "step": 122880 }, { "epoch": 78.32377310388783, "grad_norm": 0.2335909754037857, "learning_rate": 6.861665597342659e-05, "loss": 0.0142, "step": 122890 }, { "epoch": 78.33014659018482, "grad_norm": 0.3062015175819397, "learning_rate": 6.861154058051024e-05, "loss": 0.0144, "step": 122900 }, { "epoch": 78.33652007648183, "grad_norm": 0.3479377329349518, "learning_rate": 6.860642496144612e-05, "loss": 0.0166, "step": 122910 }, { "epoch": 78.34289356277884, "grad_norm": 0.6701332926750183, "learning_rate": 6.860130911629636e-05, "loss": 0.0266, "step": 122920 }, { "epoch": 78.34926704907585, "grad_norm": 0.3406379818916321, "learning_rate": 6.859619304512315e-05, "loss": 0.0262, "step": 122930 }, { "epoch": 78.35564053537284, "grad_norm": 0.3330257534980774, "learning_rate": 6.859107674798863e-05, "loss": 0.0162, "step": 122940 }, { "epoch": 78.36201402166985, "grad_norm": 0.44840726256370544, "learning_rate": 6.858596022495496e-05, "loss": 0.0165, "step": 122950 }, { "epoch": 78.36838750796686, "grad_norm": 0.27725595235824585, "learning_rate": 6.858084347608433e-05, "loss": 0.0128, "step": 122960 }, { "epoch": 78.37476099426387, "grad_norm": 0.286251962184906, "learning_rate": 6.857572650143891e-05, "loss": 0.0278, "step": 122970 }, { "epoch": 78.38113448056086, "grad_norm": 0.39910373091697693, "learning_rate": 6.857060930108088e-05, "loss": 0.0181, "step": 122980 }, { "epoch": 78.38750796685787, "grad_norm": 0.1990775465965271, "learning_rate": 6.85654918750724e-05, "loss": 0.0198, "step": 122990 }, { "epoch": 78.39388145315488, "grad_norm": 0.19122517108917236, "learning_rate": 6.856037422347567e-05, "loss": 0.0159, "step": 123000 }, { "epoch": 78.40025493945188, "grad_norm": 0.2408302277326584, "learning_rate": 6.855525634635289e-05, "loss": 0.0242, "step": 123010 }, { "epoch": 78.40662842574889, "grad_norm": 0.26401007175445557, "learning_rate": 6.85501382437662e-05, "loss": 0.0121, "step": 123020 }, { "epoch": 78.41300191204589, "grad_norm": 0.24770453572273254, "learning_rate": 6.854501991577784e-05, "loss": 0.0183, "step": 123030 }, { "epoch": 78.4193753983429, "grad_norm": 0.30025139451026917, "learning_rate": 6.853990136244996e-05, "loss": 0.014, "step": 123040 }, { "epoch": 78.4257488846399, "grad_norm": 0.5051948428153992, "learning_rate": 6.853478258384476e-05, "loss": 0.0142, "step": 123050 }, { "epoch": 78.43212237093691, "grad_norm": 0.4362858831882477, "learning_rate": 6.852966358002449e-05, "loss": 0.0196, "step": 123060 }, { "epoch": 78.4384958572339, "grad_norm": 0.31164684891700745, "learning_rate": 6.852454435105129e-05, "loss": 0.0107, "step": 123070 }, { "epoch": 78.44486934353091, "grad_norm": 0.42629143595695496, "learning_rate": 6.851942489698737e-05, "loss": 0.0175, "step": 123080 }, { "epoch": 78.45124282982792, "grad_norm": 0.25977206230163574, "learning_rate": 6.851430521789499e-05, "loss": 0.0137, "step": 123090 }, { "epoch": 78.45761631612493, "grad_norm": 0.5357429385185242, "learning_rate": 6.850918531383629e-05, "loss": 0.03, "step": 123100 }, { "epoch": 78.46398980242192, "grad_norm": 0.49596142768859863, "learning_rate": 6.850406518487354e-05, "loss": 0.0306, "step": 123110 }, { "epoch": 78.47036328871893, "grad_norm": 0.21735702455043793, "learning_rate": 6.84989448310689e-05, "loss": 0.0205, "step": 123120 }, { "epoch": 78.47673677501594, "grad_norm": 0.4362592399120331, "learning_rate": 6.849382425248464e-05, "loss": 0.0122, "step": 123130 }, { "epoch": 78.48311026131294, "grad_norm": 0.45920586585998535, "learning_rate": 6.848870344918292e-05, "loss": 0.0172, "step": 123140 }, { "epoch": 78.48948374760994, "grad_norm": 0.19502292573451996, "learning_rate": 6.848358242122604e-05, "loss": 0.02, "step": 123150 }, { "epoch": 78.49585723390695, "grad_norm": 0.415803462266922, "learning_rate": 6.847846116867614e-05, "loss": 0.0189, "step": 123160 }, { "epoch": 78.50223072020395, "grad_norm": 0.4810436964035034, "learning_rate": 6.847333969159551e-05, "loss": 0.0158, "step": 123170 }, { "epoch": 78.50860420650096, "grad_norm": 0.39929306507110596, "learning_rate": 6.846821799004637e-05, "loss": 0.0367, "step": 123180 }, { "epoch": 78.51497769279796, "grad_norm": 0.2582939863204956, "learning_rate": 6.846309606409094e-05, "loss": 0.0182, "step": 123190 }, { "epoch": 78.52135117909496, "grad_norm": 0.31870555877685547, "learning_rate": 6.845797391379144e-05, "loss": 0.0172, "step": 123200 }, { "epoch": 78.52772466539197, "grad_norm": 0.2942270338535309, "learning_rate": 6.845285153921014e-05, "loss": 0.0125, "step": 123210 }, { "epoch": 78.53409815168898, "grad_norm": 0.336407333612442, "learning_rate": 6.844772894040929e-05, "loss": 0.0188, "step": 123220 }, { "epoch": 78.54047163798597, "grad_norm": 0.22782959043979645, "learning_rate": 6.84426061174511e-05, "loss": 0.012, "step": 123230 }, { "epoch": 78.54684512428298, "grad_norm": 0.2748051881790161, "learning_rate": 6.843748307039783e-05, "loss": 0.0231, "step": 123240 }, { "epoch": 78.55321861057999, "grad_norm": 0.2406461238861084, "learning_rate": 6.843235979931175e-05, "loss": 0.0135, "step": 123250 }, { "epoch": 78.559592096877, "grad_norm": 0.3856358230113983, "learning_rate": 6.84272363042551e-05, "loss": 0.0177, "step": 123260 }, { "epoch": 78.56596558317399, "grad_norm": 0.26624780893325806, "learning_rate": 6.842211258529009e-05, "loss": 0.0101, "step": 123270 }, { "epoch": 78.572339069471, "grad_norm": 0.2951872944831848, "learning_rate": 6.841698864247906e-05, "loss": 0.0165, "step": 123280 }, { "epoch": 78.578712555768, "grad_norm": 0.3321335017681122, "learning_rate": 6.841186447588421e-05, "loss": 0.0154, "step": 123290 }, { "epoch": 78.58508604206502, "grad_norm": 0.38842177391052246, "learning_rate": 6.840674008556782e-05, "loss": 0.0105, "step": 123300 }, { "epoch": 78.59145952836201, "grad_norm": 0.6308919191360474, "learning_rate": 6.840161547159216e-05, "loss": 0.0208, "step": 123310 }, { "epoch": 78.59783301465902, "grad_norm": 0.3894574046134949, "learning_rate": 6.839649063401951e-05, "loss": 0.0312, "step": 123320 }, { "epoch": 78.60420650095602, "grad_norm": 0.21011745929718018, "learning_rate": 6.839136557291214e-05, "loss": 0.0212, "step": 123330 }, { "epoch": 78.61057998725303, "grad_norm": 0.36269330978393555, "learning_rate": 6.838624028833229e-05, "loss": 0.0151, "step": 123340 }, { "epoch": 78.61695347355003, "grad_norm": 0.24412178993225098, "learning_rate": 6.838111478034228e-05, "loss": 0.0178, "step": 123350 }, { "epoch": 78.62332695984703, "grad_norm": 0.3549131453037262, "learning_rate": 6.837598904900434e-05, "loss": 0.032, "step": 123360 }, { "epoch": 78.62970044614404, "grad_norm": 0.41879183053970337, "learning_rate": 6.837086309438081e-05, "loss": 0.0185, "step": 123370 }, { "epoch": 78.63607393244105, "grad_norm": 0.3784198462963104, "learning_rate": 6.836573691653394e-05, "loss": 0.0131, "step": 123380 }, { "epoch": 78.64244741873804, "grad_norm": 0.5755842328071594, "learning_rate": 6.836061051552602e-05, "loss": 0.0184, "step": 123390 }, { "epoch": 78.64882090503505, "grad_norm": 0.4400772452354431, "learning_rate": 6.835548389141936e-05, "loss": 0.0223, "step": 123400 }, { "epoch": 78.65519439133206, "grad_norm": 0.3973684012889862, "learning_rate": 6.835035704427623e-05, "loss": 0.0185, "step": 123410 }, { "epoch": 78.66156787762907, "grad_norm": 0.4347219467163086, "learning_rate": 6.834522997415895e-05, "loss": 0.019, "step": 123420 }, { "epoch": 78.66794136392606, "grad_norm": 0.49584612250328064, "learning_rate": 6.834010268112979e-05, "loss": 0.0153, "step": 123430 }, { "epoch": 78.67431485022307, "grad_norm": 0.5026621222496033, "learning_rate": 6.833497516525108e-05, "loss": 0.0196, "step": 123440 }, { "epoch": 78.68068833652008, "grad_norm": 0.44700005650520325, "learning_rate": 6.83298474265851e-05, "loss": 0.0155, "step": 123450 }, { "epoch": 78.68706182281709, "grad_norm": 0.5249901413917542, "learning_rate": 6.832471946519416e-05, "loss": 0.014, "step": 123460 }, { "epoch": 78.69343530911408, "grad_norm": 0.5060945153236389, "learning_rate": 6.83195912811406e-05, "loss": 0.0165, "step": 123470 }, { "epoch": 78.69980879541109, "grad_norm": 0.3246866464614868, "learning_rate": 6.83144628744867e-05, "loss": 0.0145, "step": 123480 }, { "epoch": 78.7061822817081, "grad_norm": 0.3250579237937927, "learning_rate": 6.830933424529477e-05, "loss": 0.023, "step": 123490 }, { "epoch": 78.7125557680051, "grad_norm": 0.47065550088882446, "learning_rate": 6.830420539362716e-05, "loss": 0.0287, "step": 123500 }, { "epoch": 78.7189292543021, "grad_norm": 0.5402032732963562, "learning_rate": 6.829907631954617e-05, "loss": 0.0148, "step": 123510 }, { "epoch": 78.7253027405991, "grad_norm": 0.6417707204818726, "learning_rate": 6.829394702311411e-05, "loss": 0.0163, "step": 123520 }, { "epoch": 78.73167622689611, "grad_norm": 0.5371558666229248, "learning_rate": 6.828881750439334e-05, "loss": 0.0214, "step": 123530 }, { "epoch": 78.73804971319312, "grad_norm": 0.5166661739349365, "learning_rate": 6.828368776344615e-05, "loss": 0.0144, "step": 123540 }, { "epoch": 78.74442319949011, "grad_norm": 0.8493273854255676, "learning_rate": 6.827855780033492e-05, "loss": 0.0167, "step": 123550 }, { "epoch": 78.75079668578712, "grad_norm": 0.7652248740196228, "learning_rate": 6.827342761512193e-05, "loss": 0.0189, "step": 123560 }, { "epoch": 78.75717017208413, "grad_norm": 0.23265600204467773, "learning_rate": 6.826829720786955e-05, "loss": 0.0105, "step": 123570 }, { "epoch": 78.76354365838114, "grad_norm": 0.38748666644096375, "learning_rate": 6.826316657864012e-05, "loss": 0.0135, "step": 123580 }, { "epoch": 78.76991714467813, "grad_norm": 0.4733855128288269, "learning_rate": 6.825803572749596e-05, "loss": 0.0147, "step": 123590 }, { "epoch": 78.77629063097514, "grad_norm": 0.2693250775337219, "learning_rate": 6.825290465449943e-05, "loss": 0.0154, "step": 123600 }, { "epoch": 78.78266411727215, "grad_norm": 0.3528660535812378, "learning_rate": 6.824777335971287e-05, "loss": 0.01, "step": 123610 }, { "epoch": 78.78903760356916, "grad_norm": 0.2684353291988373, "learning_rate": 6.824264184319865e-05, "loss": 0.0133, "step": 123620 }, { "epoch": 78.79541108986615, "grad_norm": 0.5610007047653198, "learning_rate": 6.823751010501909e-05, "loss": 0.0151, "step": 123630 }, { "epoch": 78.80178457616316, "grad_norm": 0.29709136486053467, "learning_rate": 6.823237814523658e-05, "loss": 0.0148, "step": 123640 }, { "epoch": 78.80815806246017, "grad_norm": 0.24438971281051636, "learning_rate": 6.822724596391344e-05, "loss": 0.0114, "step": 123650 }, { "epoch": 78.81453154875717, "grad_norm": 0.29563039541244507, "learning_rate": 6.822211356111208e-05, "loss": 0.0207, "step": 123660 }, { "epoch": 78.82090503505418, "grad_norm": 0.9763038158416748, "learning_rate": 6.821698093689481e-05, "loss": 0.0135, "step": 123670 }, { "epoch": 78.82727852135118, "grad_norm": 0.4119942784309387, "learning_rate": 6.821184809132404e-05, "loss": 0.0144, "step": 123680 }, { "epoch": 78.83365200764818, "grad_norm": 0.3480439782142639, "learning_rate": 6.820671502446211e-05, "loss": 0.0183, "step": 123690 }, { "epoch": 78.84002549394519, "grad_norm": 0.45418667793273926, "learning_rate": 6.820158173637142e-05, "loss": 0.0146, "step": 123700 }, { "epoch": 78.8463989802422, "grad_norm": 0.6552848815917969, "learning_rate": 6.81964482271143e-05, "loss": 0.0159, "step": 123710 }, { "epoch": 78.8527724665392, "grad_norm": 0.42525896430015564, "learning_rate": 6.819131449675317e-05, "loss": 0.0161, "step": 123720 }, { "epoch": 78.8591459528362, "grad_norm": 0.30015668272972107, "learning_rate": 6.818618054535039e-05, "loss": 0.011, "step": 123730 }, { "epoch": 78.86551943913321, "grad_norm": 0.6558993458747864, "learning_rate": 6.818104637296835e-05, "loss": 0.0134, "step": 123740 }, { "epoch": 78.87189292543022, "grad_norm": 0.27336692810058594, "learning_rate": 6.817591197966941e-05, "loss": 0.0156, "step": 123750 }, { "epoch": 78.87826641172721, "grad_norm": 0.46464288234710693, "learning_rate": 6.8170777365516e-05, "loss": 0.0107, "step": 123760 }, { "epoch": 78.88463989802422, "grad_norm": 0.5010867118835449, "learning_rate": 6.816564253057048e-05, "loss": 0.0189, "step": 123770 }, { "epoch": 78.89101338432123, "grad_norm": 0.4154602885246277, "learning_rate": 6.816050747489526e-05, "loss": 0.0147, "step": 123780 }, { "epoch": 78.89738687061823, "grad_norm": 0.4458504617214203, "learning_rate": 6.815537219855273e-05, "loss": 0.013, "step": 123790 }, { "epoch": 78.90376035691523, "grad_norm": 0.2146216779947281, "learning_rate": 6.815023670160529e-05, "loss": 0.0156, "step": 123800 }, { "epoch": 78.91013384321224, "grad_norm": 0.19892099499702454, "learning_rate": 6.814510098411532e-05, "loss": 0.0317, "step": 123810 }, { "epoch": 78.91650732950924, "grad_norm": 0.3107859492301941, "learning_rate": 6.813996504614524e-05, "loss": 0.0134, "step": 123820 }, { "epoch": 78.92288081580625, "grad_norm": 0.6733356714248657, "learning_rate": 6.813482888775746e-05, "loss": 0.0233, "step": 123830 }, { "epoch": 78.92925430210325, "grad_norm": 0.31189224123954773, "learning_rate": 6.81296925090144e-05, "loss": 0.0169, "step": 123840 }, { "epoch": 78.93562778840025, "grad_norm": 0.18726645410060883, "learning_rate": 6.812455590997845e-05, "loss": 0.0188, "step": 123850 }, { "epoch": 78.94200127469726, "grad_norm": 0.4826422929763794, "learning_rate": 6.811941909071205e-05, "loss": 0.0112, "step": 123860 }, { "epoch": 78.94837476099427, "grad_norm": 0.5559147000312805, "learning_rate": 6.811428205127756e-05, "loss": 0.0214, "step": 123870 }, { "epoch": 78.95474824729126, "grad_norm": 0.32084429264068604, "learning_rate": 6.810914479173749e-05, "loss": 0.0105, "step": 123880 }, { "epoch": 78.96112173358827, "grad_norm": 0.6392157077789307, "learning_rate": 6.810400731215419e-05, "loss": 0.0274, "step": 123890 }, { "epoch": 78.96749521988528, "grad_norm": 0.32043755054473877, "learning_rate": 6.80988696125901e-05, "loss": 0.0175, "step": 123900 }, { "epoch": 78.97386870618229, "grad_norm": 0.45797303318977356, "learning_rate": 6.809373169310766e-05, "loss": 0.0159, "step": 123910 }, { "epoch": 78.98024219247928, "grad_norm": 0.24299174547195435, "learning_rate": 6.80885935537693e-05, "loss": 0.0205, "step": 123920 }, { "epoch": 78.98661567877629, "grad_norm": 0.28977566957473755, "learning_rate": 6.808345519463744e-05, "loss": 0.0241, "step": 123930 }, { "epoch": 78.9929891650733, "grad_norm": 0.7594200372695923, "learning_rate": 6.807831661577455e-05, "loss": 0.0165, "step": 123940 }, { "epoch": 78.9993626513703, "grad_norm": 0.3583877384662628, "learning_rate": 6.807317781724302e-05, "loss": 0.0126, "step": 123950 }, { "epoch": 79.0057361376673, "grad_norm": 0.2698872983455658, "learning_rate": 6.806803879910533e-05, "loss": 0.0195, "step": 123960 }, { "epoch": 79.0121096239643, "grad_norm": 0.2844659686088562, "learning_rate": 6.80628995614239e-05, "loss": 0.0142, "step": 123970 }, { "epoch": 79.01848311026131, "grad_norm": 0.39553436636924744, "learning_rate": 6.805776010426119e-05, "loss": 0.0224, "step": 123980 }, { "epoch": 79.02485659655832, "grad_norm": 0.34019938111305237, "learning_rate": 6.805262042767966e-05, "loss": 0.0124, "step": 123990 }, { "epoch": 79.03123008285532, "grad_norm": 0.6314989924430847, "learning_rate": 6.804748053174172e-05, "loss": 0.023, "step": 124000 }, { "epoch": 79.03760356915232, "grad_norm": 0.31770408153533936, "learning_rate": 6.804234041650988e-05, "loss": 0.0142, "step": 124010 }, { "epoch": 79.04397705544933, "grad_norm": 0.581812858581543, "learning_rate": 6.803720008204655e-05, "loss": 0.0282, "step": 124020 }, { "epoch": 79.05035054174634, "grad_norm": 0.3969649076461792, "learning_rate": 6.803205952841421e-05, "loss": 0.0169, "step": 124030 }, { "epoch": 79.05672402804333, "grad_norm": 0.5441523790359497, "learning_rate": 6.802691875567533e-05, "loss": 0.0156, "step": 124040 }, { "epoch": 79.06309751434034, "grad_norm": 0.3006282448768616, "learning_rate": 6.802177776389236e-05, "loss": 0.0132, "step": 124050 }, { "epoch": 79.06947100063735, "grad_norm": 0.5456255674362183, "learning_rate": 6.801663655312778e-05, "loss": 0.0197, "step": 124060 }, { "epoch": 79.07584448693436, "grad_norm": 0.3816102147102356, "learning_rate": 6.801149512344404e-05, "loss": 0.0115, "step": 124070 }, { "epoch": 79.08221797323135, "grad_norm": 0.2637821137905121, "learning_rate": 6.800635347490366e-05, "loss": 0.0139, "step": 124080 }, { "epoch": 79.08859145952836, "grad_norm": 0.3666629493236542, "learning_rate": 6.800121160756906e-05, "loss": 0.0164, "step": 124090 }, { "epoch": 79.09496494582537, "grad_norm": 0.5454000234603882, "learning_rate": 6.799606952150275e-05, "loss": 0.023, "step": 124100 }, { "epoch": 79.10133843212238, "grad_norm": 0.27976348996162415, "learning_rate": 6.79909272167672e-05, "loss": 0.0203, "step": 124110 }, { "epoch": 79.10771191841937, "grad_norm": 0.9234357476234436, "learning_rate": 6.798578469342491e-05, "loss": 0.0222, "step": 124120 }, { "epoch": 79.11408540471638, "grad_norm": 0.3904895484447479, "learning_rate": 6.798064195153834e-05, "loss": 0.0183, "step": 124130 }, { "epoch": 79.12045889101339, "grad_norm": 0.3978463113307953, "learning_rate": 6.797549899117e-05, "loss": 0.0159, "step": 124140 }, { "epoch": 79.1268323773104, "grad_norm": 0.48421919345855713, "learning_rate": 6.797035581238238e-05, "loss": 0.0219, "step": 124150 }, { "epoch": 79.13320586360739, "grad_norm": 0.4921092092990875, "learning_rate": 6.796521241523795e-05, "loss": 0.0177, "step": 124160 }, { "epoch": 79.1395793499044, "grad_norm": 0.5840530395507812, "learning_rate": 6.796006879979927e-05, "loss": 0.0162, "step": 124170 }, { "epoch": 79.1459528362014, "grad_norm": 0.6610860228538513, "learning_rate": 6.795492496612874e-05, "loss": 0.0264, "step": 124180 }, { "epoch": 79.15232632249841, "grad_norm": 0.3641248643398285, "learning_rate": 6.794978091428896e-05, "loss": 0.0161, "step": 124190 }, { "epoch": 79.1586998087954, "grad_norm": 0.3426484763622284, "learning_rate": 6.794463664434237e-05, "loss": 0.0142, "step": 124200 }, { "epoch": 79.16507329509241, "grad_norm": 0.5096228122711182, "learning_rate": 6.793949215635152e-05, "loss": 0.0119, "step": 124210 }, { "epoch": 79.17144678138942, "grad_norm": 0.47674989700317383, "learning_rate": 6.793434745037891e-05, "loss": 0.0197, "step": 124220 }, { "epoch": 79.17782026768643, "grad_norm": 0.3248569071292877, "learning_rate": 6.792920252648703e-05, "loss": 0.0151, "step": 124230 }, { "epoch": 79.18419375398342, "grad_norm": 0.430462509393692, "learning_rate": 6.792405738473841e-05, "loss": 0.0149, "step": 124240 }, { "epoch": 79.19056724028043, "grad_norm": 0.30890515446662903, "learning_rate": 6.791891202519558e-05, "loss": 0.0126, "step": 124250 }, { "epoch": 79.19694072657744, "grad_norm": 0.40231853723526, "learning_rate": 6.791376644792105e-05, "loss": 0.0128, "step": 124260 }, { "epoch": 79.20331421287445, "grad_norm": 0.4438477158546448, "learning_rate": 6.790862065297732e-05, "loss": 0.0351, "step": 124270 }, { "epoch": 79.20968769917145, "grad_norm": 0.5814865231513977, "learning_rate": 6.790347464042698e-05, "loss": 0.0178, "step": 124280 }, { "epoch": 79.21606118546845, "grad_norm": 0.3947550356388092, "learning_rate": 6.789832841033249e-05, "loss": 0.0132, "step": 124290 }, { "epoch": 79.22243467176546, "grad_norm": 0.5559412837028503, "learning_rate": 6.789318196275642e-05, "loss": 0.0201, "step": 124300 }, { "epoch": 79.22880815806246, "grad_norm": 0.37741851806640625, "learning_rate": 6.788803529776131e-05, "loss": 0.0196, "step": 124310 }, { "epoch": 79.23518164435947, "grad_norm": 0.31301259994506836, "learning_rate": 6.788288841540966e-05, "loss": 0.0138, "step": 124320 }, { "epoch": 79.24155513065647, "grad_norm": 0.6295177936553955, "learning_rate": 6.787774131576404e-05, "loss": 0.021, "step": 124330 }, { "epoch": 79.24792861695347, "grad_norm": 0.27624720335006714, "learning_rate": 6.787259399888697e-05, "loss": 0.0127, "step": 124340 }, { "epoch": 79.25430210325048, "grad_norm": 0.5697582364082336, "learning_rate": 6.786744646484103e-05, "loss": 0.0186, "step": 124350 }, { "epoch": 79.26067558954749, "grad_norm": 0.2906741499900818, "learning_rate": 6.786229871368873e-05, "loss": 0.0225, "step": 124360 }, { "epoch": 79.26704907584448, "grad_norm": 0.4944457709789276, "learning_rate": 6.785715074549265e-05, "loss": 0.0228, "step": 124370 }, { "epoch": 79.27342256214149, "grad_norm": 0.26455989480018616, "learning_rate": 6.785200256031534e-05, "loss": 0.0219, "step": 124380 }, { "epoch": 79.2797960484385, "grad_norm": 0.30772116780281067, "learning_rate": 6.784685415821931e-05, "loss": 0.0132, "step": 124390 }, { "epoch": 79.2861695347355, "grad_norm": 0.4650837182998657, "learning_rate": 6.784170553926716e-05, "loss": 0.0268, "step": 124400 }, { "epoch": 79.2925430210325, "grad_norm": 0.23812608420848846, "learning_rate": 6.783655670352145e-05, "loss": 0.0127, "step": 124410 }, { "epoch": 79.29891650732951, "grad_norm": 0.6308455467224121, "learning_rate": 6.783140765104474e-05, "loss": 0.0139, "step": 124420 }, { "epoch": 79.30528999362652, "grad_norm": 0.2753642797470093, "learning_rate": 6.782625838189957e-05, "loss": 0.0137, "step": 124430 }, { "epoch": 79.31166347992352, "grad_norm": 0.3419796824455261, "learning_rate": 6.782110889614857e-05, "loss": 0.0243, "step": 124440 }, { "epoch": 79.31803696622052, "grad_norm": 0.51396644115448, "learning_rate": 6.781595919385424e-05, "loss": 0.0224, "step": 124450 }, { "epoch": 79.32441045251753, "grad_norm": 0.40464600920677185, "learning_rate": 6.781080927507919e-05, "loss": 0.0206, "step": 124460 }, { "epoch": 79.33078393881453, "grad_norm": 0.5071026086807251, "learning_rate": 6.780565913988599e-05, "loss": 0.0153, "step": 124470 }, { "epoch": 79.33715742511154, "grad_norm": 0.5155244469642639, "learning_rate": 6.780050878833722e-05, "loss": 0.0202, "step": 124480 }, { "epoch": 79.34353091140854, "grad_norm": 0.5394009351730347, "learning_rate": 6.779535822049547e-05, "loss": 0.0151, "step": 124490 }, { "epoch": 79.34990439770554, "grad_norm": 0.6136733889579773, "learning_rate": 6.77902074364233e-05, "loss": 0.0191, "step": 124500 }, { "epoch": 79.35627788400255, "grad_norm": 0.3775118589401245, "learning_rate": 6.778505643618334e-05, "loss": 0.0099, "step": 124510 }, { "epoch": 79.36265137029956, "grad_norm": 0.45429179072380066, "learning_rate": 6.777990521983812e-05, "loss": 0.0142, "step": 124520 }, { "epoch": 79.36902485659655, "grad_norm": 0.4027152955532074, "learning_rate": 6.77747537874503e-05, "loss": 0.0178, "step": 124530 }, { "epoch": 79.37539834289356, "grad_norm": 0.27589869499206543, "learning_rate": 6.77696021390824e-05, "loss": 0.0203, "step": 124540 }, { "epoch": 79.38177182919057, "grad_norm": 0.5553091764450073, "learning_rate": 6.776445027479707e-05, "loss": 0.0153, "step": 124550 }, { "epoch": 79.38814531548758, "grad_norm": 0.2851516008377075, "learning_rate": 6.775929819465691e-05, "loss": 0.0139, "step": 124560 }, { "epoch": 79.39451880178457, "grad_norm": 0.33986395597457886, "learning_rate": 6.775414589872451e-05, "loss": 0.0239, "step": 124570 }, { "epoch": 79.40089228808158, "grad_norm": 0.43518608808517456, "learning_rate": 6.774899338706247e-05, "loss": 0.0185, "step": 124580 }, { "epoch": 79.40726577437859, "grad_norm": 0.4383088946342468, "learning_rate": 6.77438406597334e-05, "loss": 0.016, "step": 124590 }, { "epoch": 79.4136392606756, "grad_norm": 0.35903435945510864, "learning_rate": 6.773868771679991e-05, "loss": 0.0174, "step": 124600 }, { "epoch": 79.42001274697259, "grad_norm": 0.5579381585121155, "learning_rate": 6.77335345583246e-05, "loss": 0.0281, "step": 124610 }, { "epoch": 79.4263862332696, "grad_norm": 0.4031221568584442, "learning_rate": 6.772838118437013e-05, "loss": 0.0256, "step": 124620 }, { "epoch": 79.4327597195666, "grad_norm": 0.4691297113895416, "learning_rate": 6.77232275949991e-05, "loss": 0.0131, "step": 124630 }, { "epoch": 79.43913320586361, "grad_norm": 0.5654774308204651, "learning_rate": 6.771807379027408e-05, "loss": 0.0241, "step": 124640 }, { "epoch": 79.4455066921606, "grad_norm": 0.5557680726051331, "learning_rate": 6.771291977025778e-05, "loss": 0.016, "step": 124650 }, { "epoch": 79.45188017845761, "grad_norm": 0.22522112727165222, "learning_rate": 6.770776553501275e-05, "loss": 0.0174, "step": 124660 }, { "epoch": 79.45825366475462, "grad_norm": 0.23699238896369934, "learning_rate": 6.770261108460166e-05, "loss": 0.0118, "step": 124670 }, { "epoch": 79.46462715105163, "grad_norm": 0.25070518255233765, "learning_rate": 6.769745641908712e-05, "loss": 0.013, "step": 124680 }, { "epoch": 79.47100063734862, "grad_norm": 0.3604833781719208, "learning_rate": 6.769230153853179e-05, "loss": 0.0094, "step": 124690 }, { "epoch": 79.47737412364563, "grad_norm": 0.47783318161964417, "learning_rate": 6.768714644299829e-05, "loss": 0.0186, "step": 124700 }, { "epoch": 79.48374760994264, "grad_norm": 0.30699992179870605, "learning_rate": 6.768199113254925e-05, "loss": 0.0152, "step": 124710 }, { "epoch": 79.49012109623965, "grad_norm": 0.32178446650505066, "learning_rate": 6.767683560724733e-05, "loss": 0.011, "step": 124720 }, { "epoch": 79.49649458253664, "grad_norm": 0.4799908399581909, "learning_rate": 6.767167986715516e-05, "loss": 0.0167, "step": 124730 }, { "epoch": 79.50286806883365, "grad_norm": 0.16650621592998505, "learning_rate": 6.766652391233542e-05, "loss": 0.014, "step": 124740 }, { "epoch": 79.50924155513066, "grad_norm": 0.4994553029537201, "learning_rate": 6.766136774285071e-05, "loss": 0.0168, "step": 124750 }, { "epoch": 79.51561504142767, "grad_norm": 0.4704348146915436, "learning_rate": 6.765621135876372e-05, "loss": 0.0075, "step": 124760 }, { "epoch": 79.52198852772466, "grad_norm": 0.4244696795940399, "learning_rate": 6.765105476013709e-05, "loss": 0.0192, "step": 124770 }, { "epoch": 79.52836201402167, "grad_norm": 0.5259035229682922, "learning_rate": 6.764589794703348e-05, "loss": 0.0134, "step": 124780 }, { "epoch": 79.53473550031867, "grad_norm": 0.25295814871788025, "learning_rate": 6.764074091951554e-05, "loss": 0.0142, "step": 124790 }, { "epoch": 79.54110898661568, "grad_norm": 0.5579010844230652, "learning_rate": 6.763558367764595e-05, "loss": 0.0226, "step": 124800 }, { "epoch": 79.54748247291268, "grad_norm": 0.4375065267086029, "learning_rate": 6.763042622148736e-05, "loss": 0.0121, "step": 124810 }, { "epoch": 79.55385595920968, "grad_norm": 0.2199287712574005, "learning_rate": 6.762526855110245e-05, "loss": 0.0158, "step": 124820 }, { "epoch": 79.56022944550669, "grad_norm": 0.3574201464653015, "learning_rate": 6.762011066655389e-05, "loss": 0.0119, "step": 124830 }, { "epoch": 79.5666029318037, "grad_norm": 0.32563516497612, "learning_rate": 6.761495256790435e-05, "loss": 0.0131, "step": 124840 }, { "epoch": 79.5729764181007, "grad_norm": 0.3432236313819885, "learning_rate": 6.76097942552165e-05, "loss": 0.0118, "step": 124850 }, { "epoch": 79.5793499043977, "grad_norm": 0.17910881340503693, "learning_rate": 6.760463572855302e-05, "loss": 0.0115, "step": 124860 }, { "epoch": 79.58572339069471, "grad_norm": 0.47910794615745544, "learning_rate": 6.759947698797661e-05, "loss": 0.0154, "step": 124870 }, { "epoch": 79.59209687699172, "grad_norm": 0.6364296078681946, "learning_rate": 6.759431803354992e-05, "loss": 0.0167, "step": 124880 }, { "epoch": 79.59847036328873, "grad_norm": 0.3372046649456024, "learning_rate": 6.758915886533566e-05, "loss": 0.018, "step": 124890 }, { "epoch": 79.60484384958572, "grad_norm": 0.5765656232833862, "learning_rate": 6.758399948339653e-05, "loss": 0.0203, "step": 124900 }, { "epoch": 79.61121733588273, "grad_norm": 0.5448253154754639, "learning_rate": 6.757883988779517e-05, "loss": 0.0173, "step": 124910 }, { "epoch": 79.61759082217974, "grad_norm": 0.470099538564682, "learning_rate": 6.757368007859434e-05, "loss": 0.0139, "step": 124920 }, { "epoch": 79.62396430847674, "grad_norm": 0.5501277446746826, "learning_rate": 6.756852005585668e-05, "loss": 0.0129, "step": 124930 }, { "epoch": 79.63033779477374, "grad_norm": 0.1706847995519638, "learning_rate": 6.756335981964492e-05, "loss": 0.0101, "step": 124940 }, { "epoch": 79.63671128107075, "grad_norm": 0.2411523461341858, "learning_rate": 6.755819937002177e-05, "loss": 0.0117, "step": 124950 }, { "epoch": 79.64308476736775, "grad_norm": 0.4624946415424347, "learning_rate": 6.75530387070499e-05, "loss": 0.0284, "step": 124960 }, { "epoch": 79.64945825366476, "grad_norm": 0.20828039944171906, "learning_rate": 6.754787783079206e-05, "loss": 0.0135, "step": 124970 }, { "epoch": 79.65583173996176, "grad_norm": 0.45076823234558105, "learning_rate": 6.754271674131091e-05, "loss": 0.016, "step": 124980 }, { "epoch": 79.66220522625876, "grad_norm": 0.19420182704925537, "learning_rate": 6.753755543866922e-05, "loss": 0.0203, "step": 124990 }, { "epoch": 79.66857871255577, "grad_norm": 0.323278546333313, "learning_rate": 6.753239392292965e-05, "loss": 0.0142, "step": 125000 }, { "epoch": 79.67495219885278, "grad_norm": 0.19359712302684784, "learning_rate": 6.752723219415494e-05, "loss": 0.0169, "step": 125010 }, { "epoch": 79.68132568514977, "grad_norm": 0.3531534671783447, "learning_rate": 6.752207025240782e-05, "loss": 0.016, "step": 125020 }, { "epoch": 79.68769917144678, "grad_norm": 0.44073188304901123, "learning_rate": 6.7516908097751e-05, "loss": 0.0158, "step": 125030 }, { "epoch": 79.69407265774379, "grad_norm": 0.42768749594688416, "learning_rate": 6.75117457302472e-05, "loss": 0.0145, "step": 125040 }, { "epoch": 79.7004461440408, "grad_norm": 0.3125505745410919, "learning_rate": 6.750658314995916e-05, "loss": 0.0104, "step": 125050 }, { "epoch": 79.70681963033779, "grad_norm": 0.27193912863731384, "learning_rate": 6.750142035694962e-05, "loss": 0.0149, "step": 125060 }, { "epoch": 79.7131931166348, "grad_norm": 0.16986165940761566, "learning_rate": 6.74962573512813e-05, "loss": 0.0141, "step": 125070 }, { "epoch": 79.7195666029318, "grad_norm": 0.49531176686286926, "learning_rate": 6.749109413301691e-05, "loss": 0.0151, "step": 125080 }, { "epoch": 79.72594008922881, "grad_norm": 0.4428061544895172, "learning_rate": 6.748593070221923e-05, "loss": 0.0185, "step": 125090 }, { "epoch": 79.73231357552581, "grad_norm": 0.38770952820777893, "learning_rate": 6.748076705895097e-05, "loss": 0.0186, "step": 125100 }, { "epoch": 79.73868706182282, "grad_norm": 0.5201800465583801, "learning_rate": 6.74756032032749e-05, "loss": 0.0155, "step": 125110 }, { "epoch": 79.74506054811982, "grad_norm": 0.6536782383918762, "learning_rate": 6.747043913525373e-05, "loss": 0.013, "step": 125120 }, { "epoch": 79.75143403441683, "grad_norm": 0.32139503955841064, "learning_rate": 6.746527485495025e-05, "loss": 0.0166, "step": 125130 }, { "epoch": 79.75780752071383, "grad_norm": 0.2603851556777954, "learning_rate": 6.746011036242717e-05, "loss": 0.0155, "step": 125140 }, { "epoch": 79.76418100701083, "grad_norm": 0.3858010470867157, "learning_rate": 6.745494565774728e-05, "loss": 0.0188, "step": 125150 }, { "epoch": 79.77055449330784, "grad_norm": 0.33930355310440063, "learning_rate": 6.744978074097332e-05, "loss": 0.0153, "step": 125160 }, { "epoch": 79.77692797960485, "grad_norm": 0.5987381935119629, "learning_rate": 6.744461561216805e-05, "loss": 0.0178, "step": 125170 }, { "epoch": 79.78330146590184, "grad_norm": 0.3014277517795563, "learning_rate": 6.743945027139423e-05, "loss": 0.0196, "step": 125180 }, { "epoch": 79.78967495219885, "grad_norm": 0.2712571322917938, "learning_rate": 6.74342847187146e-05, "loss": 0.0219, "step": 125190 }, { "epoch": 79.79604843849586, "grad_norm": 0.5042886734008789, "learning_rate": 6.742911895419198e-05, "loss": 0.0242, "step": 125200 }, { "epoch": 79.80242192479287, "grad_norm": 0.3758966624736786, "learning_rate": 6.74239529778891e-05, "loss": 0.0096, "step": 125210 }, { "epoch": 79.80879541108986, "grad_norm": 0.4786214530467987, "learning_rate": 6.741878678986873e-05, "loss": 0.0123, "step": 125220 }, { "epoch": 79.81516889738687, "grad_norm": 0.35215839743614197, "learning_rate": 6.741362039019367e-05, "loss": 0.0164, "step": 125230 }, { "epoch": 79.82154238368388, "grad_norm": 0.3008069694042206, "learning_rate": 6.740845377892667e-05, "loss": 0.0175, "step": 125240 }, { "epoch": 79.82791586998088, "grad_norm": 0.2933357357978821, "learning_rate": 6.740328695613053e-05, "loss": 0.0114, "step": 125250 }, { "epoch": 79.83428935627788, "grad_norm": 0.3213497996330261, "learning_rate": 6.7398119921868e-05, "loss": 0.0153, "step": 125260 }, { "epoch": 79.84066284257489, "grad_norm": 0.3545714318752289, "learning_rate": 6.739295267620192e-05, "loss": 0.0066, "step": 125270 }, { "epoch": 79.8470363288719, "grad_norm": 0.2940559983253479, "learning_rate": 6.738778521919501e-05, "loss": 0.0224, "step": 125280 }, { "epoch": 79.8534098151689, "grad_norm": 0.6588410139083862, "learning_rate": 6.738261755091012e-05, "loss": 0.0135, "step": 125290 }, { "epoch": 79.8597833014659, "grad_norm": 0.38771483302116394, "learning_rate": 6.737744967141e-05, "loss": 0.0264, "step": 125300 }, { "epoch": 79.8661567877629, "grad_norm": 0.39279651641845703, "learning_rate": 6.737228158075747e-05, "loss": 0.0251, "step": 125310 }, { "epoch": 79.87253027405991, "grad_norm": 0.5684505105018616, "learning_rate": 6.73671132790153e-05, "loss": 0.0175, "step": 125320 }, { "epoch": 79.87890376035692, "grad_norm": 0.2768491804599762, "learning_rate": 6.736194476624632e-05, "loss": 0.012, "step": 125330 }, { "epoch": 79.88527724665391, "grad_norm": 0.44127583503723145, "learning_rate": 6.735677604251332e-05, "loss": 0.0116, "step": 125340 }, { "epoch": 79.89165073295092, "grad_norm": 0.581598162651062, "learning_rate": 6.735160710787909e-05, "loss": 0.0183, "step": 125350 }, { "epoch": 79.89802421924793, "grad_norm": 0.25878554582595825, "learning_rate": 6.734643796240647e-05, "loss": 0.0142, "step": 125360 }, { "epoch": 79.90439770554494, "grad_norm": 0.5962002873420715, "learning_rate": 6.734126860615824e-05, "loss": 0.0161, "step": 125370 }, { "epoch": 79.91077119184193, "grad_norm": 0.23590587079524994, "learning_rate": 6.733609903919722e-05, "loss": 0.0146, "step": 125380 }, { "epoch": 79.91714467813894, "grad_norm": 0.3890500068664551, "learning_rate": 6.733092926158623e-05, "loss": 0.0146, "step": 125390 }, { "epoch": 79.92351816443595, "grad_norm": 0.3760499358177185, "learning_rate": 6.732575927338808e-05, "loss": 0.0113, "step": 125400 }, { "epoch": 79.92989165073296, "grad_norm": 0.32836225628852844, "learning_rate": 6.73205890746656e-05, "loss": 0.0134, "step": 125410 }, { "epoch": 79.93626513702995, "grad_norm": 0.37851718068122864, "learning_rate": 6.731541866548164e-05, "loss": 0.0137, "step": 125420 }, { "epoch": 79.94263862332696, "grad_norm": 0.4003143906593323, "learning_rate": 6.731024804589896e-05, "loss": 0.0166, "step": 125430 }, { "epoch": 79.94901210962396, "grad_norm": 0.3304022252559662, "learning_rate": 6.730507721598043e-05, "loss": 0.0153, "step": 125440 }, { "epoch": 79.95538559592097, "grad_norm": 0.3601759374141693, "learning_rate": 6.729990617578889e-05, "loss": 0.0212, "step": 125450 }, { "epoch": 79.96175908221797, "grad_norm": 0.5092607140541077, "learning_rate": 6.729473492538713e-05, "loss": 0.0302, "step": 125460 }, { "epoch": 79.96813256851497, "grad_norm": 0.346640944480896, "learning_rate": 6.728956346483802e-05, "loss": 0.0115, "step": 125470 }, { "epoch": 79.97450605481198, "grad_norm": 0.5884768962860107, "learning_rate": 6.72843917942044e-05, "loss": 0.0203, "step": 125480 }, { "epoch": 79.98087954110899, "grad_norm": 0.39155152440071106, "learning_rate": 6.727921991354909e-05, "loss": 0.0097, "step": 125490 }, { "epoch": 79.98725302740598, "grad_norm": 0.48818764090538025, "learning_rate": 6.727404782293494e-05, "loss": 0.0171, "step": 125500 }, { "epoch": 79.99362651370299, "grad_norm": 0.4754868447780609, "learning_rate": 6.726887552242482e-05, "loss": 0.0133, "step": 125510 }, { "epoch": 80.0, "grad_norm": 0.4132716953754425, "learning_rate": 6.726370301208154e-05, "loss": 0.0129, "step": 125520 }, { "epoch": 80.00637348629701, "grad_norm": 0.26044467091560364, "learning_rate": 6.725853029196797e-05, "loss": 0.0196, "step": 125530 }, { "epoch": 80.01274697259402, "grad_norm": 0.34793469309806824, "learning_rate": 6.725335736214697e-05, "loss": 0.0114, "step": 125540 }, { "epoch": 80.01912045889101, "grad_norm": 0.26328396797180176, "learning_rate": 6.724818422268137e-05, "loss": 0.0157, "step": 125550 }, { "epoch": 80.02549394518802, "grad_norm": 0.5081735253334045, "learning_rate": 6.724301087363407e-05, "loss": 0.0197, "step": 125560 }, { "epoch": 80.03186743148503, "grad_norm": 0.39210349321365356, "learning_rate": 6.72378373150679e-05, "loss": 0.0155, "step": 125570 }, { "epoch": 80.03824091778203, "grad_norm": 0.3234591484069824, "learning_rate": 6.723266354704574e-05, "loss": 0.0152, "step": 125580 }, { "epoch": 80.04461440407903, "grad_norm": 0.46330076456069946, "learning_rate": 6.722748956963043e-05, "loss": 0.0154, "step": 125590 }, { "epoch": 80.05098789037604, "grad_norm": 0.5016193389892578, "learning_rate": 6.722231538288486e-05, "loss": 0.0155, "step": 125600 }, { "epoch": 80.05736137667304, "grad_norm": 0.3275073766708374, "learning_rate": 6.721714098687191e-05, "loss": 0.0133, "step": 125610 }, { "epoch": 80.06373486297005, "grad_norm": 0.7065107226371765, "learning_rate": 6.721196638165443e-05, "loss": 0.0225, "step": 125620 }, { "epoch": 80.07010834926704, "grad_norm": 0.3828933835029602, "learning_rate": 6.720679156729532e-05, "loss": 0.0146, "step": 125630 }, { "epoch": 80.07648183556405, "grad_norm": 0.4238526225090027, "learning_rate": 6.720161654385743e-05, "loss": 0.0268, "step": 125640 }, { "epoch": 80.08285532186106, "grad_norm": 0.2987675964832306, "learning_rate": 6.719644131140367e-05, "loss": 0.0224, "step": 125650 }, { "epoch": 80.08922880815807, "grad_norm": 0.42434296011924744, "learning_rate": 6.71912658699969e-05, "loss": 0.02, "step": 125660 }, { "epoch": 80.09560229445506, "grad_norm": 0.24993960559368134, "learning_rate": 6.718609021970003e-05, "loss": 0.0143, "step": 125670 }, { "epoch": 80.10197578075207, "grad_norm": 0.4516327977180481, "learning_rate": 6.718091436057593e-05, "loss": 0.0098, "step": 125680 }, { "epoch": 80.10834926704908, "grad_norm": 0.3569120764732361, "learning_rate": 6.717573829268749e-05, "loss": 0.0135, "step": 125690 }, { "epoch": 80.11472275334609, "grad_norm": 0.4617933928966522, "learning_rate": 6.717056201609764e-05, "loss": 0.0165, "step": 125700 }, { "epoch": 80.12109623964308, "grad_norm": 0.6558597683906555, "learning_rate": 6.716538553086923e-05, "loss": 0.0143, "step": 125710 }, { "epoch": 80.12746972594009, "grad_norm": 0.4239666759967804, "learning_rate": 6.716020883706518e-05, "loss": 0.0209, "step": 125720 }, { "epoch": 80.1338432122371, "grad_norm": 0.2596076428890228, "learning_rate": 6.715503193474839e-05, "loss": 0.0194, "step": 125730 }, { "epoch": 80.1402166985341, "grad_norm": 0.26778683066368103, "learning_rate": 6.714985482398179e-05, "loss": 0.0135, "step": 125740 }, { "epoch": 80.1465901848311, "grad_norm": 0.2040417641401291, "learning_rate": 6.714467750482823e-05, "loss": 0.0097, "step": 125750 }, { "epoch": 80.1529636711281, "grad_norm": 0.45506352186203003, "learning_rate": 6.713949997735067e-05, "loss": 0.0215, "step": 125760 }, { "epoch": 80.15933715742511, "grad_norm": 0.38872867822647095, "learning_rate": 6.713432224161199e-05, "loss": 0.0177, "step": 125770 }, { "epoch": 80.16571064372212, "grad_norm": 0.18489481508731842, "learning_rate": 6.712914429767511e-05, "loss": 0.0125, "step": 125780 }, { "epoch": 80.17208413001912, "grad_norm": 0.3248577415943146, "learning_rate": 6.712396614560297e-05, "loss": 0.0131, "step": 125790 }, { "epoch": 80.17845761631612, "grad_norm": 0.43897250294685364, "learning_rate": 6.711878778545848e-05, "loss": 0.0237, "step": 125800 }, { "epoch": 80.18483110261313, "grad_norm": 0.453967422246933, "learning_rate": 6.711360921730454e-05, "loss": 0.0141, "step": 125810 }, { "epoch": 80.19120458891014, "grad_norm": 0.2577480375766754, "learning_rate": 6.710843044120409e-05, "loss": 0.0145, "step": 125820 }, { "epoch": 80.19757807520713, "grad_norm": 0.32862624526023865, "learning_rate": 6.710325145722007e-05, "loss": 0.014, "step": 125830 }, { "epoch": 80.20395156150414, "grad_norm": 0.508571982383728, "learning_rate": 6.70980722654154e-05, "loss": 0.0178, "step": 125840 }, { "epoch": 80.21032504780115, "grad_norm": 0.37741729617118835, "learning_rate": 6.709289286585299e-05, "loss": 0.0113, "step": 125850 }, { "epoch": 80.21669853409816, "grad_norm": 0.688094973564148, "learning_rate": 6.708771325859581e-05, "loss": 0.0172, "step": 125860 }, { "epoch": 80.22307202039515, "grad_norm": 0.48461732268333435, "learning_rate": 6.708253344370676e-05, "loss": 0.0126, "step": 125870 }, { "epoch": 80.22944550669216, "grad_norm": 0.316806823015213, "learning_rate": 6.707735342124882e-05, "loss": 0.0101, "step": 125880 }, { "epoch": 80.23581899298917, "grad_norm": 0.4391910433769226, "learning_rate": 6.707217319128489e-05, "loss": 0.0203, "step": 125890 }, { "epoch": 80.24219247928617, "grad_norm": 0.3578508496284485, "learning_rate": 6.706699275387796e-05, "loss": 0.0309, "step": 125900 }, { "epoch": 80.24856596558317, "grad_norm": 0.35560983419418335, "learning_rate": 6.706181210909095e-05, "loss": 0.0139, "step": 125910 }, { "epoch": 80.25493945188018, "grad_norm": 0.3086462914943695, "learning_rate": 6.705663125698682e-05, "loss": 0.0159, "step": 125920 }, { "epoch": 80.26131293817718, "grad_norm": 0.5664320588111877, "learning_rate": 6.705145019762849e-05, "loss": 0.019, "step": 125930 }, { "epoch": 80.26768642447419, "grad_norm": 0.3132663667201996, "learning_rate": 6.704626893107897e-05, "loss": 0.0095, "step": 125940 }, { "epoch": 80.27405991077119, "grad_norm": 0.48230257630348206, "learning_rate": 6.704108745740117e-05, "loss": 0.0218, "step": 125950 }, { "epoch": 80.2804333970682, "grad_norm": 0.6967734694480896, "learning_rate": 6.703590577665808e-05, "loss": 0.02, "step": 125960 }, { "epoch": 80.2868068833652, "grad_norm": 0.391135573387146, "learning_rate": 6.703072388891263e-05, "loss": 0.0169, "step": 125970 }, { "epoch": 80.29318036966221, "grad_norm": 0.41381028294563293, "learning_rate": 6.702554179422783e-05, "loss": 0.0249, "step": 125980 }, { "epoch": 80.2995538559592, "grad_norm": 1.3740564584732056, "learning_rate": 6.702035949266658e-05, "loss": 0.027, "step": 125990 }, { "epoch": 80.30592734225621, "grad_norm": 0.3906605541706085, "learning_rate": 6.701517698429194e-05, "loss": 0.0208, "step": 126000 }, { "epoch": 80.31230082855322, "grad_norm": 0.3945595622062683, "learning_rate": 6.70099942691668e-05, "loss": 0.0189, "step": 126010 }, { "epoch": 80.31867431485023, "grad_norm": 0.6461365222930908, "learning_rate": 6.700481134735421e-05, "loss": 0.0234, "step": 126020 }, { "epoch": 80.32504780114722, "grad_norm": 0.6638172268867493, "learning_rate": 6.699962821891706e-05, "loss": 0.0148, "step": 126030 }, { "epoch": 80.33142128744423, "grad_norm": 0.5848672389984131, "learning_rate": 6.699444488391842e-05, "loss": 0.0255, "step": 126040 }, { "epoch": 80.33779477374124, "grad_norm": 0.3781450092792511, "learning_rate": 6.698926134242119e-05, "loss": 0.0155, "step": 126050 }, { "epoch": 80.34416826003824, "grad_norm": 0.6222744584083557, "learning_rate": 6.698407759448843e-05, "loss": 0.0187, "step": 126060 }, { "epoch": 80.35054174633524, "grad_norm": 0.767084538936615, "learning_rate": 6.697889364018306e-05, "loss": 0.019, "step": 126070 }, { "epoch": 80.35691523263225, "grad_norm": 0.43843960762023926, "learning_rate": 6.697370947956812e-05, "loss": 0.0177, "step": 126080 }, { "epoch": 80.36328871892925, "grad_norm": 0.4191601872444153, "learning_rate": 6.696852511270659e-05, "loss": 0.0128, "step": 126090 }, { "epoch": 80.36966220522626, "grad_norm": 0.3066982328891754, "learning_rate": 6.696334053966144e-05, "loss": 0.01, "step": 126100 }, { "epoch": 80.37603569152326, "grad_norm": 0.2755712568759918, "learning_rate": 6.69581557604957e-05, "loss": 0.0239, "step": 126110 }, { "epoch": 80.38240917782026, "grad_norm": 0.34613803029060364, "learning_rate": 6.695297077527236e-05, "loss": 0.0128, "step": 126120 }, { "epoch": 80.38878266411727, "grad_norm": 0.5091246366500854, "learning_rate": 6.694778558405441e-05, "loss": 0.0203, "step": 126130 }, { "epoch": 80.39515615041428, "grad_norm": 0.2727690637111664, "learning_rate": 6.694260018690485e-05, "loss": 0.0175, "step": 126140 }, { "epoch": 80.40152963671127, "grad_norm": 0.5497962832450867, "learning_rate": 6.693741458388674e-05, "loss": 0.0232, "step": 126150 }, { "epoch": 80.40790312300828, "grad_norm": 0.35026729106903076, "learning_rate": 6.693222877506302e-05, "loss": 0.0179, "step": 126160 }, { "epoch": 80.41427660930529, "grad_norm": 0.23851560056209564, "learning_rate": 6.692704276049674e-05, "loss": 0.0227, "step": 126170 }, { "epoch": 80.4206500956023, "grad_norm": 0.5691242218017578, "learning_rate": 6.692185654025092e-05, "loss": 0.0217, "step": 126180 }, { "epoch": 80.4270235818993, "grad_norm": 0.501129686832428, "learning_rate": 6.691667011438855e-05, "loss": 0.0092, "step": 126190 }, { "epoch": 80.4333970681963, "grad_norm": 0.38817015290260315, "learning_rate": 6.691148348297268e-05, "loss": 0.0152, "step": 126200 }, { "epoch": 80.43977055449331, "grad_norm": 0.6953297853469849, "learning_rate": 6.69062966460663e-05, "loss": 0.0101, "step": 126210 }, { "epoch": 80.44614404079032, "grad_norm": 0.4346459209918976, "learning_rate": 6.690110960373246e-05, "loss": 0.0115, "step": 126220 }, { "epoch": 80.45251752708732, "grad_norm": 0.4636843204498291, "learning_rate": 6.689592235603419e-05, "loss": 0.0182, "step": 126230 }, { "epoch": 80.45889101338432, "grad_norm": 0.3680875301361084, "learning_rate": 6.68907349030345e-05, "loss": 0.0221, "step": 126240 }, { "epoch": 80.46526449968133, "grad_norm": 0.4416764974594116, "learning_rate": 6.688554724479646e-05, "loss": 0.0242, "step": 126250 }, { "epoch": 80.47163798597833, "grad_norm": 0.37517237663269043, "learning_rate": 6.688035938138303e-05, "loss": 0.0116, "step": 126260 }, { "epoch": 80.47801147227534, "grad_norm": 0.35759076476097107, "learning_rate": 6.687517131285735e-05, "loss": 0.017, "step": 126270 }, { "epoch": 80.48438495857233, "grad_norm": 0.3682042062282562, "learning_rate": 6.686998303928237e-05, "loss": 0.0204, "step": 126280 }, { "epoch": 80.49075844486934, "grad_norm": 0.6756048798561096, "learning_rate": 6.686479456072118e-05, "loss": 0.0194, "step": 126290 }, { "epoch": 80.49713193116635, "grad_norm": 0.6227577328681946, "learning_rate": 6.685960587723681e-05, "loss": 0.0241, "step": 126300 }, { "epoch": 80.50350541746336, "grad_norm": 0.4742136299610138, "learning_rate": 6.685441698889232e-05, "loss": 0.0211, "step": 126310 }, { "epoch": 80.50987890376035, "grad_norm": 0.2923162579536438, "learning_rate": 6.684922789575074e-05, "loss": 0.0222, "step": 126320 }, { "epoch": 80.51625239005736, "grad_norm": 0.5508079528808594, "learning_rate": 6.684403859787514e-05, "loss": 0.0159, "step": 126330 }, { "epoch": 80.52262587635437, "grad_norm": 0.3177070617675781, "learning_rate": 6.683884909532857e-05, "loss": 0.0129, "step": 126340 }, { "epoch": 80.52899936265138, "grad_norm": 0.5068110227584839, "learning_rate": 6.683365938817406e-05, "loss": 0.0128, "step": 126350 }, { "epoch": 80.53537284894837, "grad_norm": 0.33872923254966736, "learning_rate": 6.682846947647472e-05, "loss": 0.0167, "step": 126360 }, { "epoch": 80.54174633524538, "grad_norm": 0.23440749943256378, "learning_rate": 6.682327936029357e-05, "loss": 0.0156, "step": 126370 }, { "epoch": 80.54811982154239, "grad_norm": 0.16772273182868958, "learning_rate": 6.68180890396937e-05, "loss": 0.0114, "step": 126380 }, { "epoch": 80.5544933078394, "grad_norm": 0.7510534524917603, "learning_rate": 6.681289851473818e-05, "loss": 0.0197, "step": 126390 }, { "epoch": 80.56086679413639, "grad_norm": 0.3557160496711731, "learning_rate": 6.680770778549006e-05, "loss": 0.0185, "step": 126400 }, { "epoch": 80.5672402804334, "grad_norm": 0.3356439173221588, "learning_rate": 6.680251685201242e-05, "loss": 0.0247, "step": 126410 }, { "epoch": 80.5736137667304, "grad_norm": 0.6048323512077332, "learning_rate": 6.679732571436834e-05, "loss": 0.0279, "step": 126420 }, { "epoch": 80.57998725302741, "grad_norm": 0.3814190924167633, "learning_rate": 6.679213437262088e-05, "loss": 0.0211, "step": 126430 }, { "epoch": 80.5863607393244, "grad_norm": 0.35746559500694275, "learning_rate": 6.678694282683315e-05, "loss": 0.0139, "step": 126440 }, { "epoch": 80.59273422562141, "grad_norm": 0.38779416680336, "learning_rate": 6.678175107706821e-05, "loss": 0.0123, "step": 126450 }, { "epoch": 80.59910771191842, "grad_norm": 0.4195821285247803, "learning_rate": 6.677655912338913e-05, "loss": 0.0175, "step": 126460 }, { "epoch": 80.60548119821543, "grad_norm": 0.2711597979068756, "learning_rate": 6.677136696585906e-05, "loss": 0.0228, "step": 126470 }, { "epoch": 80.61185468451242, "grad_norm": 0.5140336155891418, "learning_rate": 6.676617460454102e-05, "loss": 0.0135, "step": 126480 }, { "epoch": 80.61822817080943, "grad_norm": 0.3927972912788391, "learning_rate": 6.676098203949812e-05, "loss": 0.0099, "step": 126490 }, { "epoch": 80.62460165710644, "grad_norm": 0.12783552706241608, "learning_rate": 6.675578927079348e-05, "loss": 0.0162, "step": 126500 }, { "epoch": 80.63097514340345, "grad_norm": 0.2168223112821579, "learning_rate": 6.675059629849017e-05, "loss": 0.015, "step": 126510 }, { "epoch": 80.63734862970044, "grad_norm": 0.31361573934555054, "learning_rate": 6.674540312265131e-05, "loss": 0.0141, "step": 126520 }, { "epoch": 80.64372211599745, "grad_norm": 0.6191925406455994, "learning_rate": 6.674020974333998e-05, "loss": 0.0162, "step": 126530 }, { "epoch": 80.65009560229446, "grad_norm": 0.3114016652107239, "learning_rate": 6.67350161606193e-05, "loss": 0.0139, "step": 126540 }, { "epoch": 80.65646908859146, "grad_norm": 0.34173473715782166, "learning_rate": 6.672982237455237e-05, "loss": 0.0117, "step": 126550 }, { "epoch": 80.66284257488846, "grad_norm": 0.488564133644104, "learning_rate": 6.672462838520232e-05, "loss": 0.0208, "step": 126560 }, { "epoch": 80.66921606118547, "grad_norm": 0.2971140444278717, "learning_rate": 6.671943419263223e-05, "loss": 0.0212, "step": 126570 }, { "epoch": 80.67558954748247, "grad_norm": 0.40928199887275696, "learning_rate": 6.671423979690523e-05, "loss": 0.0099, "step": 126580 }, { "epoch": 80.68196303377948, "grad_norm": 0.40064939856529236, "learning_rate": 6.670904519808443e-05, "loss": 0.0216, "step": 126590 }, { "epoch": 80.68833652007648, "grad_norm": 0.5325694680213928, "learning_rate": 6.670385039623295e-05, "loss": 0.0105, "step": 126600 }, { "epoch": 80.69471000637348, "grad_norm": 0.15836618840694427, "learning_rate": 6.669865539141394e-05, "loss": 0.0198, "step": 126610 }, { "epoch": 80.70108349267049, "grad_norm": 0.5138508677482605, "learning_rate": 6.669346018369049e-05, "loss": 0.0148, "step": 126620 }, { "epoch": 80.7074569789675, "grad_norm": 0.3755810856819153, "learning_rate": 6.668826477312574e-05, "loss": 0.0178, "step": 126630 }, { "epoch": 80.7138304652645, "grad_norm": 0.22888292372226715, "learning_rate": 6.668306915978281e-05, "loss": 0.0164, "step": 126640 }, { "epoch": 80.7202039515615, "grad_norm": 0.2579191327095032, "learning_rate": 6.667787334372484e-05, "loss": 0.0155, "step": 126650 }, { "epoch": 80.72657743785851, "grad_norm": 0.265672892332077, "learning_rate": 6.667267732501496e-05, "loss": 0.0087, "step": 126660 }, { "epoch": 80.73295092415552, "grad_norm": 0.8445600867271423, "learning_rate": 6.666748110371632e-05, "loss": 0.0274, "step": 126670 }, { "epoch": 80.73932441045251, "grad_norm": 0.6260095238685608, "learning_rate": 6.666228467989203e-05, "loss": 0.0168, "step": 126680 }, { "epoch": 80.74569789674952, "grad_norm": 0.25650957226753235, "learning_rate": 6.665708805360527e-05, "loss": 0.0265, "step": 126690 }, { "epoch": 80.75207138304653, "grad_norm": 0.44495564699172974, "learning_rate": 6.665189122491915e-05, "loss": 0.0146, "step": 126700 }, { "epoch": 80.75844486934353, "grad_norm": 0.3959968388080597, "learning_rate": 6.664669419389684e-05, "loss": 0.0196, "step": 126710 }, { "epoch": 80.76481835564053, "grad_norm": 0.7253341674804688, "learning_rate": 6.664149696060146e-05, "loss": 0.0148, "step": 126720 }, { "epoch": 80.77119184193754, "grad_norm": 0.3937951326370239, "learning_rate": 6.663629952509619e-05, "loss": 0.0136, "step": 126730 }, { "epoch": 80.77756532823454, "grad_norm": 0.3259277939796448, "learning_rate": 6.663110188744417e-05, "loss": 0.0086, "step": 126740 }, { "epoch": 80.78393881453155, "grad_norm": 0.3008105754852295, "learning_rate": 6.662590404770857e-05, "loss": 0.0144, "step": 126750 }, { "epoch": 80.79031230082855, "grad_norm": 0.29666775465011597, "learning_rate": 6.662070600595252e-05, "loss": 0.0217, "step": 126760 }, { "epoch": 80.79668578712555, "grad_norm": 0.6915923953056335, "learning_rate": 6.661550776223922e-05, "loss": 0.0254, "step": 126770 }, { "epoch": 80.80305927342256, "grad_norm": 0.3394468426704407, "learning_rate": 6.66103093166318e-05, "loss": 0.0125, "step": 126780 }, { "epoch": 80.80943275971957, "grad_norm": 0.3648572862148285, "learning_rate": 6.660511066919344e-05, "loss": 0.0179, "step": 126790 }, { "epoch": 80.81580624601658, "grad_norm": 0.31661874055862427, "learning_rate": 6.659991181998733e-05, "loss": 0.0172, "step": 126800 }, { "epoch": 80.82217973231357, "grad_norm": 0.18575285375118256, "learning_rate": 6.659471276907658e-05, "loss": 0.0169, "step": 126810 }, { "epoch": 80.82855321861058, "grad_norm": 0.19795109331607819, "learning_rate": 6.658951351652444e-05, "loss": 0.0208, "step": 126820 }, { "epoch": 80.83492670490759, "grad_norm": 0.41461706161499023, "learning_rate": 6.658431406239402e-05, "loss": 0.0121, "step": 126830 }, { "epoch": 80.8413001912046, "grad_norm": 0.1891605705022812, "learning_rate": 6.657911440674855e-05, "loss": 0.0129, "step": 126840 }, { "epoch": 80.84767367750159, "grad_norm": 0.2017437368631363, "learning_rate": 6.657391454965117e-05, "loss": 0.0155, "step": 126850 }, { "epoch": 80.8540471637986, "grad_norm": 0.2239898294210434, "learning_rate": 6.656871449116508e-05, "loss": 0.0291, "step": 126860 }, { "epoch": 80.8604206500956, "grad_norm": 0.48643994331359863, "learning_rate": 6.656351423135346e-05, "loss": 0.0204, "step": 126870 }, { "epoch": 80.86679413639261, "grad_norm": 0.5520294308662415, "learning_rate": 6.655831377027952e-05, "loss": 0.0216, "step": 126880 }, { "epoch": 80.8731676226896, "grad_norm": 0.6381688117980957, "learning_rate": 6.655311310800643e-05, "loss": 0.0215, "step": 126890 }, { "epoch": 80.87954110898661, "grad_norm": 0.29683762788772583, "learning_rate": 6.654791224459737e-05, "loss": 0.0204, "step": 126900 }, { "epoch": 80.88591459528362, "grad_norm": 0.5710203051567078, "learning_rate": 6.654271118011557e-05, "loss": 0.016, "step": 126910 }, { "epoch": 80.89228808158063, "grad_norm": 0.4914308786392212, "learning_rate": 6.65375099146242e-05, "loss": 0.0198, "step": 126920 }, { "epoch": 80.89866156787762, "grad_norm": 0.32553476095199585, "learning_rate": 6.653230844818647e-05, "loss": 0.0124, "step": 126930 }, { "epoch": 80.90503505417463, "grad_norm": 0.39785251021385193, "learning_rate": 6.652710678086558e-05, "loss": 0.0128, "step": 126940 }, { "epoch": 80.91140854047164, "grad_norm": 0.40677180886268616, "learning_rate": 6.652190491272476e-05, "loss": 0.0212, "step": 126950 }, { "epoch": 80.91778202676865, "grad_norm": 0.4438180923461914, "learning_rate": 6.651670284382718e-05, "loss": 0.0229, "step": 126960 }, { "epoch": 80.92415551306564, "grad_norm": 0.911753237247467, "learning_rate": 6.651150057423607e-05, "loss": 0.0095, "step": 126970 }, { "epoch": 80.93052899936265, "grad_norm": 0.6160798072814941, "learning_rate": 6.650629810401465e-05, "loss": 0.0226, "step": 126980 }, { "epoch": 80.93690248565966, "grad_norm": 0.4749008119106293, "learning_rate": 6.65010954332261e-05, "loss": 0.0172, "step": 126990 }, { "epoch": 80.94327597195667, "grad_norm": 0.29766616225242615, "learning_rate": 6.649589256193367e-05, "loss": 0.0189, "step": 127000 }, { "epoch": 80.94964945825366, "grad_norm": 0.47174158692359924, "learning_rate": 6.649068949020058e-05, "loss": 0.0127, "step": 127010 }, { "epoch": 80.95602294455067, "grad_norm": 0.32872432470321655, "learning_rate": 6.648548621809003e-05, "loss": 0.0171, "step": 127020 }, { "epoch": 80.96239643084768, "grad_norm": 0.28067752718925476, "learning_rate": 6.648028274566526e-05, "loss": 0.0123, "step": 127030 }, { "epoch": 80.96876991714468, "grad_norm": 0.23956497013568878, "learning_rate": 6.647507907298949e-05, "loss": 0.0196, "step": 127040 }, { "epoch": 80.97514340344168, "grad_norm": 0.29510581493377686, "learning_rate": 6.646987520012597e-05, "loss": 0.0127, "step": 127050 }, { "epoch": 80.98151688973869, "grad_norm": 0.23039419949054718, "learning_rate": 6.64646711271379e-05, "loss": 0.0203, "step": 127060 }, { "epoch": 80.9878903760357, "grad_norm": 0.253165602684021, "learning_rate": 6.645946685408856e-05, "loss": 0.0147, "step": 127070 }, { "epoch": 80.9942638623327, "grad_norm": 0.4904017746448517, "learning_rate": 6.645426238104113e-05, "loss": 0.0208, "step": 127080 }, { "epoch": 81.0006373486297, "grad_norm": 0.7988378405570984, "learning_rate": 6.644905770805889e-05, "loss": 0.0122, "step": 127090 }, { "epoch": 81.0070108349267, "grad_norm": 0.43837496638298035, "learning_rate": 6.644385283520504e-05, "loss": 0.0202, "step": 127100 }, { "epoch": 81.01338432122371, "grad_norm": 0.2410060167312622, "learning_rate": 6.64386477625429e-05, "loss": 0.0118, "step": 127110 }, { "epoch": 81.01975780752072, "grad_norm": 0.38906365633010864, "learning_rate": 6.643344249013562e-05, "loss": 0.0139, "step": 127120 }, { "epoch": 81.02613129381771, "grad_norm": 0.3276345729827881, "learning_rate": 6.642823701804653e-05, "loss": 0.0181, "step": 127130 }, { "epoch": 81.03250478011472, "grad_norm": 0.6254652142524719, "learning_rate": 6.642303134633885e-05, "loss": 0.0253, "step": 127140 }, { "epoch": 81.03887826641173, "grad_norm": 0.8940832614898682, "learning_rate": 6.641782547507581e-05, "loss": 0.0192, "step": 127150 }, { "epoch": 81.04525175270874, "grad_norm": 0.6184812784194946, "learning_rate": 6.641261940432071e-05, "loss": 0.016, "step": 127160 }, { "epoch": 81.05162523900573, "grad_norm": 0.47905582189559937, "learning_rate": 6.640741313413678e-05, "loss": 0.0138, "step": 127170 }, { "epoch": 81.05799872530274, "grad_norm": 0.4322684705257416, "learning_rate": 6.640220666458729e-05, "loss": 0.0142, "step": 127180 }, { "epoch": 81.06437221159975, "grad_norm": 0.3111843168735504, "learning_rate": 6.63969999957355e-05, "loss": 0.013, "step": 127190 }, { "epoch": 81.07074569789675, "grad_norm": 0.3068602383136749, "learning_rate": 6.639179312764469e-05, "loss": 0.0176, "step": 127200 }, { "epoch": 81.07711918419375, "grad_norm": 0.3148795962333679, "learning_rate": 6.638658606037811e-05, "loss": 0.0217, "step": 127210 }, { "epoch": 81.08349267049076, "grad_norm": 0.17243891954421997, "learning_rate": 6.638137879399903e-05, "loss": 0.0108, "step": 127220 }, { "epoch": 81.08986615678776, "grad_norm": 0.2453475147485733, "learning_rate": 6.637617132857075e-05, "loss": 0.021, "step": 127230 }, { "epoch": 81.09623964308477, "grad_norm": 0.15608802437782288, "learning_rate": 6.63709636641565e-05, "loss": 0.0111, "step": 127240 }, { "epoch": 81.10261312938177, "grad_norm": 0.42830178141593933, "learning_rate": 6.636575580081961e-05, "loss": 0.0148, "step": 127250 }, { "epoch": 81.10898661567877, "grad_norm": 0.21630507707595825, "learning_rate": 6.636054773862332e-05, "loss": 0.0137, "step": 127260 }, { "epoch": 81.11536010197578, "grad_norm": 0.21450023353099823, "learning_rate": 6.635533947763093e-05, "loss": 0.0106, "step": 127270 }, { "epoch": 81.12173358827279, "grad_norm": 0.2385348379611969, "learning_rate": 6.635013101790573e-05, "loss": 0.0143, "step": 127280 }, { "epoch": 81.12810707456978, "grad_norm": 0.3169389069080353, "learning_rate": 6.6344922359511e-05, "loss": 0.0208, "step": 127290 }, { "epoch": 81.13448056086679, "grad_norm": 0.48389434814453125, "learning_rate": 6.633971350251002e-05, "loss": 0.0204, "step": 127300 }, { "epoch": 81.1408540471638, "grad_norm": 0.34694239497184753, "learning_rate": 6.63345044469661e-05, "loss": 0.0188, "step": 127310 }, { "epoch": 81.1472275334608, "grad_norm": 0.36945629119873047, "learning_rate": 6.632929519294253e-05, "loss": 0.0264, "step": 127320 }, { "epoch": 81.1536010197578, "grad_norm": 0.4333002269268036, "learning_rate": 6.632408574050261e-05, "loss": 0.012, "step": 127330 }, { "epoch": 81.15997450605481, "grad_norm": 0.4793090522289276, "learning_rate": 6.631887608970963e-05, "loss": 0.0173, "step": 127340 }, { "epoch": 81.16634799235182, "grad_norm": 0.5290791988372803, "learning_rate": 6.631366624062689e-05, "loss": 0.0147, "step": 127350 }, { "epoch": 81.17272147864882, "grad_norm": 0.3516162633895874, "learning_rate": 6.630845619331773e-05, "loss": 0.0196, "step": 127360 }, { "epoch": 81.17909496494582, "grad_norm": 0.21909646689891815, "learning_rate": 6.63032459478454e-05, "loss": 0.0153, "step": 127370 }, { "epoch": 81.18546845124283, "grad_norm": 0.5707698464393616, "learning_rate": 6.629803550427326e-05, "loss": 0.0167, "step": 127380 }, { "epoch": 81.19184193753983, "grad_norm": 0.29888975620269775, "learning_rate": 6.629282486266458e-05, "loss": 0.018, "step": 127390 }, { "epoch": 81.19821542383684, "grad_norm": 0.3173043131828308, "learning_rate": 6.628761402308273e-05, "loss": 0.0123, "step": 127400 }, { "epoch": 81.20458891013385, "grad_norm": 0.29929792881011963, "learning_rate": 6.628240298559096e-05, "loss": 0.026, "step": 127410 }, { "epoch": 81.21096239643084, "grad_norm": 0.2227499634027481, "learning_rate": 6.627719175025261e-05, "loss": 0.0081, "step": 127420 }, { "epoch": 81.21733588272785, "grad_norm": 0.23302891850471497, "learning_rate": 6.627198031713105e-05, "loss": 0.0174, "step": 127430 }, { "epoch": 81.22370936902486, "grad_norm": 0.3801746666431427, "learning_rate": 6.626676868628954e-05, "loss": 0.0091, "step": 127440 }, { "epoch": 81.23008285532187, "grad_norm": 0.3989483118057251, "learning_rate": 6.626155685779143e-05, "loss": 0.0186, "step": 127450 }, { "epoch": 81.23645634161886, "grad_norm": 0.5053424835205078, "learning_rate": 6.625634483170007e-05, "loss": 0.0231, "step": 127460 }, { "epoch": 81.24282982791587, "grad_norm": 0.7118179798126221, "learning_rate": 6.625113260807875e-05, "loss": 0.0198, "step": 127470 }, { "epoch": 81.24920331421288, "grad_norm": 0.38172170519828796, "learning_rate": 6.624592018699084e-05, "loss": 0.01, "step": 127480 }, { "epoch": 81.25557680050989, "grad_norm": 0.16911453008651733, "learning_rate": 6.624070756849964e-05, "loss": 0.0238, "step": 127490 }, { "epoch": 81.26195028680688, "grad_norm": 0.3202678859233856, "learning_rate": 6.623549475266855e-05, "loss": 0.017, "step": 127500 }, { "epoch": 81.26832377310389, "grad_norm": 0.42701807618141174, "learning_rate": 6.623028173956083e-05, "loss": 0.0155, "step": 127510 }, { "epoch": 81.2746972594009, "grad_norm": 0.26585322618484497, "learning_rate": 6.622506852923988e-05, "loss": 0.0106, "step": 127520 }, { "epoch": 81.2810707456979, "grad_norm": 0.23967081308364868, "learning_rate": 6.621985512176905e-05, "loss": 0.0093, "step": 127530 }, { "epoch": 81.2874442319949, "grad_norm": 0.2693195641040802, "learning_rate": 6.621464151721164e-05, "loss": 0.0148, "step": 127540 }, { "epoch": 81.2938177182919, "grad_norm": 0.2564235031604767, "learning_rate": 6.620942771563104e-05, "loss": 0.0125, "step": 127550 }, { "epoch": 81.30019120458891, "grad_norm": 0.34550246596336365, "learning_rate": 6.620421371709055e-05, "loss": 0.0197, "step": 127560 }, { "epoch": 81.30656469088592, "grad_norm": 0.26747581362724304, "learning_rate": 6.61989995216536e-05, "loss": 0.0128, "step": 127570 }, { "epoch": 81.31293817718291, "grad_norm": 0.1312360018491745, "learning_rate": 6.619378512938351e-05, "loss": 0.0139, "step": 127580 }, { "epoch": 81.31931166347992, "grad_norm": 0.4966040849685669, "learning_rate": 6.618857054034365e-05, "loss": 0.0158, "step": 127590 }, { "epoch": 81.32568514977693, "grad_norm": 0.4939136803150177, "learning_rate": 6.618335575459734e-05, "loss": 0.0266, "step": 127600 }, { "epoch": 81.33205863607394, "grad_norm": 0.2748204171657562, "learning_rate": 6.617814077220801e-05, "loss": 0.0297, "step": 127610 }, { "epoch": 81.33843212237093, "grad_norm": 0.3028007745742798, "learning_rate": 6.617292559323899e-05, "loss": 0.0151, "step": 127620 }, { "epoch": 81.34480560866794, "grad_norm": 0.4295285642147064, "learning_rate": 6.616771021775363e-05, "loss": 0.0195, "step": 127630 }, { "epoch": 81.35117909496495, "grad_norm": 0.3242585062980652, "learning_rate": 6.616249464581535e-05, "loss": 0.013, "step": 127640 }, { "epoch": 81.35755258126196, "grad_norm": 0.28624168038368225, "learning_rate": 6.615727887748749e-05, "loss": 0.0116, "step": 127650 }, { "epoch": 81.36392606755895, "grad_norm": 0.18329140543937683, "learning_rate": 6.615206291283344e-05, "loss": 0.0141, "step": 127660 }, { "epoch": 81.37029955385596, "grad_norm": 0.3322015106678009, "learning_rate": 6.614684675191658e-05, "loss": 0.0201, "step": 127670 }, { "epoch": 81.37667304015297, "grad_norm": 0.38822224736213684, "learning_rate": 6.614163039480027e-05, "loss": 0.0117, "step": 127680 }, { "epoch": 81.38304652644997, "grad_norm": 0.5598871111869812, "learning_rate": 6.613641384154794e-05, "loss": 0.0185, "step": 127690 }, { "epoch": 81.38942001274697, "grad_norm": 0.30626851320266724, "learning_rate": 6.613119709222292e-05, "loss": 0.0141, "step": 127700 }, { "epoch": 81.39579349904398, "grad_norm": 0.24263986945152283, "learning_rate": 6.612598014688864e-05, "loss": 0.0157, "step": 127710 }, { "epoch": 81.40216698534098, "grad_norm": 0.19710896909236908, "learning_rate": 6.612076300560847e-05, "loss": 0.0126, "step": 127720 }, { "epoch": 81.40854047163799, "grad_norm": 0.43520018458366394, "learning_rate": 6.611554566844581e-05, "loss": 0.014, "step": 127730 }, { "epoch": 81.41491395793498, "grad_norm": 0.23662208020687103, "learning_rate": 6.611032813546405e-05, "loss": 0.0148, "step": 127740 }, { "epoch": 81.421287444232, "grad_norm": 0.26984718441963196, "learning_rate": 6.61051104067266e-05, "loss": 0.0157, "step": 127750 }, { "epoch": 81.427660930529, "grad_norm": 0.4992947280406952, "learning_rate": 6.609989248229684e-05, "loss": 0.0195, "step": 127760 }, { "epoch": 81.43403441682601, "grad_norm": 0.40786072611808777, "learning_rate": 6.609467436223819e-05, "loss": 0.0171, "step": 127770 }, { "epoch": 81.440407903123, "grad_norm": 0.3281767964363098, "learning_rate": 6.608945604661407e-05, "loss": 0.0133, "step": 127780 }, { "epoch": 81.44678138942001, "grad_norm": 0.3373706042766571, "learning_rate": 6.608423753548785e-05, "loss": 0.0191, "step": 127790 }, { "epoch": 81.45315487571702, "grad_norm": 0.36204326152801514, "learning_rate": 6.607901882892296e-05, "loss": 0.0144, "step": 127800 }, { "epoch": 81.45952836201403, "grad_norm": 0.2809852957725525, "learning_rate": 6.607379992698281e-05, "loss": 0.0138, "step": 127810 }, { "epoch": 81.46590184831102, "grad_norm": 0.24130919575691223, "learning_rate": 6.606858082973082e-05, "loss": 0.0138, "step": 127820 }, { "epoch": 81.47227533460803, "grad_norm": 0.40450048446655273, "learning_rate": 6.606336153723038e-05, "loss": 0.0104, "step": 127830 }, { "epoch": 81.47864882090504, "grad_norm": 0.2732711732387543, "learning_rate": 6.605814204954496e-05, "loss": 0.0206, "step": 127840 }, { "epoch": 81.48502230720204, "grad_norm": 0.5569290518760681, "learning_rate": 6.605292236673794e-05, "loss": 0.0175, "step": 127850 }, { "epoch": 81.49139579349904, "grad_norm": 0.26931798458099365, "learning_rate": 6.604770248887275e-05, "loss": 0.0182, "step": 127860 }, { "epoch": 81.49776927979605, "grad_norm": 0.4693372845649719, "learning_rate": 6.604248241601285e-05, "loss": 0.0254, "step": 127870 }, { "epoch": 81.50414276609305, "grad_norm": 0.42395004630088806, "learning_rate": 6.60372621482216e-05, "loss": 0.0139, "step": 127880 }, { "epoch": 81.51051625239006, "grad_norm": 0.4101807475090027, "learning_rate": 6.603204168556251e-05, "loss": 0.0143, "step": 127890 }, { "epoch": 81.51688973868706, "grad_norm": 0.20198851823806763, "learning_rate": 6.602682102809896e-05, "loss": 0.0115, "step": 127900 }, { "epoch": 81.52326322498406, "grad_norm": 0.2849816679954529, "learning_rate": 6.60216001758944e-05, "loss": 0.012, "step": 127910 }, { "epoch": 81.52963671128107, "grad_norm": 0.43186095356941223, "learning_rate": 6.601637912901227e-05, "loss": 0.0168, "step": 127920 }, { "epoch": 81.53601019757808, "grad_norm": 0.37529632449150085, "learning_rate": 6.601115788751602e-05, "loss": 0.0225, "step": 127930 }, { "epoch": 81.54238368387507, "grad_norm": 0.16300275921821594, "learning_rate": 6.600593645146909e-05, "loss": 0.0245, "step": 127940 }, { "epoch": 81.54875717017208, "grad_norm": 0.4236455261707306, "learning_rate": 6.600071482093491e-05, "loss": 0.0125, "step": 127950 }, { "epoch": 81.55513065646909, "grad_norm": 0.6016177535057068, "learning_rate": 6.599549299597695e-05, "loss": 0.0162, "step": 127960 }, { "epoch": 81.5615041427661, "grad_norm": 0.42569202184677124, "learning_rate": 6.599027097665863e-05, "loss": 0.0184, "step": 127970 }, { "epoch": 81.56787762906309, "grad_norm": 0.3088478147983551, "learning_rate": 6.598504876304343e-05, "loss": 0.0117, "step": 127980 }, { "epoch": 81.5742511153601, "grad_norm": 0.34524720907211304, "learning_rate": 6.59798263551948e-05, "loss": 0.0128, "step": 127990 }, { "epoch": 81.5806246016571, "grad_norm": 0.3614957630634308, "learning_rate": 6.597460375317619e-05, "loss": 0.0134, "step": 128000 }, { "epoch": 81.58699808795411, "grad_norm": 0.18238256871700287, "learning_rate": 6.596938095705105e-05, "loss": 0.0092, "step": 128010 }, { "epoch": 81.59337157425111, "grad_norm": 0.6764192581176758, "learning_rate": 6.596415796688287e-05, "loss": 0.0183, "step": 128020 }, { "epoch": 81.59974506054812, "grad_norm": 0.7099656462669373, "learning_rate": 6.59589347827351e-05, "loss": 0.029, "step": 128030 }, { "epoch": 81.60611854684512, "grad_norm": 0.663221001625061, "learning_rate": 6.595371140467117e-05, "loss": 0.016, "step": 128040 }, { "epoch": 81.61249203314213, "grad_norm": 0.4533836245536804, "learning_rate": 6.594848783275462e-05, "loss": 0.0155, "step": 128050 }, { "epoch": 81.61886551943914, "grad_norm": 0.3718253970146179, "learning_rate": 6.594326406704887e-05, "loss": 0.0131, "step": 128060 }, { "epoch": 81.62523900573613, "grad_norm": 0.32075798511505127, "learning_rate": 6.593804010761741e-05, "loss": 0.0162, "step": 128070 }, { "epoch": 81.63161249203314, "grad_norm": 0.41828203201293945, "learning_rate": 6.593281595452372e-05, "loss": 0.0141, "step": 128080 }, { "epoch": 81.63798597833015, "grad_norm": 0.13964274525642395, "learning_rate": 6.592759160783127e-05, "loss": 0.0092, "step": 128090 }, { "epoch": 81.64435946462716, "grad_norm": 0.29005011916160583, "learning_rate": 6.592236706760354e-05, "loss": 0.0123, "step": 128100 }, { "epoch": 81.65073295092415, "grad_norm": 0.5831001996994019, "learning_rate": 6.591714233390402e-05, "loss": 0.0235, "step": 128110 }, { "epoch": 81.65710643722116, "grad_norm": 0.20778940618038177, "learning_rate": 6.591191740679619e-05, "loss": 0.0214, "step": 128120 }, { "epoch": 81.66347992351817, "grad_norm": 0.7422183156013489, "learning_rate": 6.590669228634354e-05, "loss": 0.0223, "step": 128130 }, { "epoch": 81.66985340981518, "grad_norm": 0.483907014131546, "learning_rate": 6.590146697260957e-05, "loss": 0.0145, "step": 128140 }, { "epoch": 81.67622689611217, "grad_norm": 0.3647027313709259, "learning_rate": 6.589624146565775e-05, "loss": 0.0125, "step": 128150 }, { "epoch": 81.68260038240918, "grad_norm": 0.2211553305387497, "learning_rate": 6.58910157655516e-05, "loss": 0.0157, "step": 128160 }, { "epoch": 81.68897386870619, "grad_norm": 0.8959919810295105, "learning_rate": 6.588578987235459e-05, "loss": 0.0163, "step": 128170 }, { "epoch": 81.6953473550032, "grad_norm": 0.289741575717926, "learning_rate": 6.588056378613023e-05, "loss": 0.0125, "step": 128180 }, { "epoch": 81.70172084130019, "grad_norm": 0.1956545114517212, "learning_rate": 6.587533750694206e-05, "loss": 0.0245, "step": 128190 }, { "epoch": 81.7080943275972, "grad_norm": 0.751069188117981, "learning_rate": 6.587011103485352e-05, "loss": 0.0165, "step": 128200 }, { "epoch": 81.7144678138942, "grad_norm": 0.44637590646743774, "learning_rate": 6.586488436992816e-05, "loss": 0.0131, "step": 128210 }, { "epoch": 81.72084130019121, "grad_norm": 0.35311785340309143, "learning_rate": 6.585965751222947e-05, "loss": 0.0121, "step": 128220 }, { "epoch": 81.7272147864882, "grad_norm": 0.2387092262506485, "learning_rate": 6.585443046182097e-05, "loss": 0.0125, "step": 128230 }, { "epoch": 81.73358827278521, "grad_norm": 0.3070967495441437, "learning_rate": 6.584920321876617e-05, "loss": 0.0202, "step": 128240 }, { "epoch": 81.73996175908222, "grad_norm": 0.23998437821865082, "learning_rate": 6.584397578312859e-05, "loss": 0.0129, "step": 128250 }, { "epoch": 81.74633524537923, "grad_norm": 0.1991690993309021, "learning_rate": 6.583874815497175e-05, "loss": 0.0162, "step": 128260 }, { "epoch": 81.75270873167622, "grad_norm": 0.2373322695493698, "learning_rate": 6.583352033435914e-05, "loss": 0.0146, "step": 128270 }, { "epoch": 81.75908221797323, "grad_norm": 0.48872095346450806, "learning_rate": 6.582829232135433e-05, "loss": 0.0134, "step": 128280 }, { "epoch": 81.76545570427024, "grad_norm": 0.20950189232826233, "learning_rate": 6.582306411602081e-05, "loss": 0.0113, "step": 128290 }, { "epoch": 81.77182919056725, "grad_norm": 0.2452247440814972, "learning_rate": 6.581783571842216e-05, "loss": 0.0116, "step": 128300 }, { "epoch": 81.77820267686424, "grad_norm": 0.48979443311691284, "learning_rate": 6.581260712862183e-05, "loss": 0.0127, "step": 128310 }, { "epoch": 81.78457616316125, "grad_norm": 0.2909833788871765, "learning_rate": 6.580737834668341e-05, "loss": 0.0258, "step": 128320 }, { "epoch": 81.79094964945826, "grad_norm": 0.5084080100059509, "learning_rate": 6.58021493726704e-05, "loss": 0.019, "step": 128330 }, { "epoch": 81.79732313575526, "grad_norm": 0.5048345923423767, "learning_rate": 6.579692020664639e-05, "loss": 0.0138, "step": 128340 }, { "epoch": 81.80369662205226, "grad_norm": 0.23233503103256226, "learning_rate": 6.579169084867486e-05, "loss": 0.0147, "step": 128350 }, { "epoch": 81.81007010834927, "grad_norm": 0.44075509905815125, "learning_rate": 6.578646129881938e-05, "loss": 0.0103, "step": 128360 }, { "epoch": 81.81644359464627, "grad_norm": 0.3790730834007263, "learning_rate": 6.57812315571435e-05, "loss": 0.0271, "step": 128370 }, { "epoch": 81.82281708094328, "grad_norm": 0.2620769441127777, "learning_rate": 6.577600162371074e-05, "loss": 0.0231, "step": 128380 }, { "epoch": 81.82919056724027, "grad_norm": 0.37984275817871094, "learning_rate": 6.577077149858468e-05, "loss": 0.0177, "step": 128390 }, { "epoch": 81.83556405353728, "grad_norm": 0.2711968421936035, "learning_rate": 6.576554118182885e-05, "loss": 0.0141, "step": 128400 }, { "epoch": 81.84193753983429, "grad_norm": 0.312850683927536, "learning_rate": 6.576031067350681e-05, "loss": 0.0155, "step": 128410 }, { "epoch": 81.8483110261313, "grad_norm": 0.6217361092567444, "learning_rate": 6.575507997368212e-05, "loss": 0.0166, "step": 128420 }, { "epoch": 81.85468451242829, "grad_norm": 0.2619372606277466, "learning_rate": 6.574984908241833e-05, "loss": 0.0177, "step": 128430 }, { "epoch": 81.8610579987253, "grad_norm": 0.1730402708053589, "learning_rate": 6.5744617999779e-05, "loss": 0.0204, "step": 128440 }, { "epoch": 81.86743148502231, "grad_norm": 0.5761787295341492, "learning_rate": 6.573938672582769e-05, "loss": 0.0183, "step": 128450 }, { "epoch": 81.87380497131932, "grad_norm": 0.22725649178028107, "learning_rate": 6.573415526062798e-05, "loss": 0.0187, "step": 128460 }, { "epoch": 81.88017845761631, "grad_norm": 0.29454728960990906, "learning_rate": 6.572892360424342e-05, "loss": 0.0176, "step": 128470 }, { "epoch": 81.88655194391332, "grad_norm": 0.22280116379261017, "learning_rate": 6.57236917567376e-05, "loss": 0.0117, "step": 128480 }, { "epoch": 81.89292543021033, "grad_norm": 0.4208506643772125, "learning_rate": 6.571845971817407e-05, "loss": 0.0288, "step": 128490 }, { "epoch": 81.89929891650733, "grad_norm": 0.532431960105896, "learning_rate": 6.571322748861643e-05, "loss": 0.0351, "step": 128500 }, { "epoch": 81.90567240280433, "grad_norm": 0.3497880697250366, "learning_rate": 6.570799506812824e-05, "loss": 0.0239, "step": 128510 }, { "epoch": 81.91204588910134, "grad_norm": 0.6106723546981812, "learning_rate": 6.570276245677305e-05, "loss": 0.0134, "step": 128520 }, { "epoch": 81.91841937539834, "grad_norm": 0.42493873834609985, "learning_rate": 6.569752965461448e-05, "loss": 0.0131, "step": 128530 }, { "epoch": 81.92479286169535, "grad_norm": 0.3729230761528015, "learning_rate": 6.569229666171613e-05, "loss": 0.0159, "step": 128540 }, { "epoch": 81.93116634799235, "grad_norm": 0.3722967803478241, "learning_rate": 6.568706347814154e-05, "loss": 0.0124, "step": 128550 }, { "epoch": 81.93753983428935, "grad_norm": 0.28316137194633484, "learning_rate": 6.56818301039543e-05, "loss": 0.0234, "step": 128560 }, { "epoch": 81.94391332058636, "grad_norm": 0.6038443446159363, "learning_rate": 6.567659653921805e-05, "loss": 0.0158, "step": 128570 }, { "epoch": 81.95028680688337, "grad_norm": 0.5959979295730591, "learning_rate": 6.567136278399634e-05, "loss": 0.0154, "step": 128580 }, { "epoch": 81.95666029318036, "grad_norm": 0.31616199016571045, "learning_rate": 6.566612883835276e-05, "loss": 0.0157, "step": 128590 }, { "epoch": 81.96303377947737, "grad_norm": 0.26260602474212646, "learning_rate": 6.566089470235095e-05, "loss": 0.0113, "step": 128600 }, { "epoch": 81.96940726577438, "grad_norm": 0.28833502531051636, "learning_rate": 6.565566037605447e-05, "loss": 0.0234, "step": 128610 }, { "epoch": 81.97578075207139, "grad_norm": 0.24887816607952118, "learning_rate": 6.565042585952693e-05, "loss": 0.011, "step": 128620 }, { "epoch": 81.98215423836838, "grad_norm": 0.21246270835399628, "learning_rate": 6.564519115283194e-05, "loss": 0.0142, "step": 128630 }, { "epoch": 81.98852772466539, "grad_norm": 0.8406859040260315, "learning_rate": 6.563995625603312e-05, "loss": 0.0311, "step": 128640 }, { "epoch": 81.9949012109624, "grad_norm": 0.36603865027427673, "learning_rate": 6.563472116919405e-05, "loss": 0.0143, "step": 128650 }, { "epoch": 82.0012746972594, "grad_norm": 0.2759566307067871, "learning_rate": 6.562948589237837e-05, "loss": 0.0217, "step": 128660 }, { "epoch": 82.00764818355641, "grad_norm": 0.33693820238113403, "learning_rate": 6.562425042564967e-05, "loss": 0.0131, "step": 128670 }, { "epoch": 82.0140216698534, "grad_norm": 0.6064636707305908, "learning_rate": 6.561901476907157e-05, "loss": 0.0315, "step": 128680 }, { "epoch": 82.02039515615041, "grad_norm": 0.3349491059780121, "learning_rate": 6.56137789227077e-05, "loss": 0.012, "step": 128690 }, { "epoch": 82.02676864244742, "grad_norm": 0.432344913482666, "learning_rate": 6.560854288662167e-05, "loss": 0.0134, "step": 128700 }, { "epoch": 82.03314212874443, "grad_norm": 0.563667893409729, "learning_rate": 6.560330666087711e-05, "loss": 0.0239, "step": 128710 }, { "epoch": 82.03951561504142, "grad_norm": 0.7653446197509766, "learning_rate": 6.559807024553765e-05, "loss": 0.0201, "step": 128720 }, { "epoch": 82.04588910133843, "grad_norm": 0.40037381649017334, "learning_rate": 6.55928336406669e-05, "loss": 0.0197, "step": 128730 }, { "epoch": 82.05226258763544, "grad_norm": 0.4187275469303131, "learning_rate": 6.558759684632851e-05, "loss": 0.0151, "step": 128740 }, { "epoch": 82.05863607393245, "grad_norm": 0.5199059247970581, "learning_rate": 6.558235986258608e-05, "loss": 0.0233, "step": 128750 }, { "epoch": 82.06500956022944, "grad_norm": 0.29151448607444763, "learning_rate": 6.557712268950329e-05, "loss": 0.0141, "step": 128760 }, { "epoch": 82.07138304652645, "grad_norm": 0.2693638503551483, "learning_rate": 6.557188532714374e-05, "loss": 0.01, "step": 128770 }, { "epoch": 82.07775653282346, "grad_norm": 0.3176136612892151, "learning_rate": 6.556664777557109e-05, "loss": 0.0122, "step": 128780 }, { "epoch": 82.08413001912047, "grad_norm": 0.43023332953453064, "learning_rate": 6.556141003484895e-05, "loss": 0.0161, "step": 128790 }, { "epoch": 82.09050350541746, "grad_norm": 0.5234009027481079, "learning_rate": 6.5556172105041e-05, "loss": 0.0136, "step": 128800 }, { "epoch": 82.09687699171447, "grad_norm": 0.37645745277404785, "learning_rate": 6.555093398621088e-05, "loss": 0.0117, "step": 128810 }, { "epoch": 82.10325047801147, "grad_norm": 0.27266931533813477, "learning_rate": 6.554569567842221e-05, "loss": 0.0149, "step": 128820 }, { "epoch": 82.10962396430848, "grad_norm": 0.34110593795776367, "learning_rate": 6.554045718173867e-05, "loss": 0.0118, "step": 128830 }, { "epoch": 82.11599745060548, "grad_norm": 0.46295127272605896, "learning_rate": 6.553521849622391e-05, "loss": 0.02, "step": 128840 }, { "epoch": 82.12237093690248, "grad_norm": 0.3209141194820404, "learning_rate": 6.552997962194157e-05, "loss": 0.0118, "step": 128850 }, { "epoch": 82.12874442319949, "grad_norm": 0.2872767746448517, "learning_rate": 6.552474055895531e-05, "loss": 0.0117, "step": 128860 }, { "epoch": 82.1351179094965, "grad_norm": 0.25526511669158936, "learning_rate": 6.55195013073288e-05, "loss": 0.0141, "step": 128870 }, { "epoch": 82.1414913957935, "grad_norm": 0.4619637727737427, "learning_rate": 6.55142618671257e-05, "loss": 0.0185, "step": 128880 }, { "epoch": 82.1478648820905, "grad_norm": 0.9284997582435608, "learning_rate": 6.550902223840967e-05, "loss": 0.0151, "step": 128890 }, { "epoch": 82.15423836838751, "grad_norm": 0.43440350890159607, "learning_rate": 6.550378242124437e-05, "loss": 0.0124, "step": 128900 }, { "epoch": 82.16061185468452, "grad_norm": 0.24251821637153625, "learning_rate": 6.549854241569347e-05, "loss": 0.0159, "step": 128910 }, { "epoch": 82.16698534098151, "grad_norm": 0.35185766220092773, "learning_rate": 6.549330222182067e-05, "loss": 0.0132, "step": 128920 }, { "epoch": 82.17335882727852, "grad_norm": 0.35577017068862915, "learning_rate": 6.54880618396896e-05, "loss": 0.0315, "step": 128930 }, { "epoch": 82.17973231357553, "grad_norm": 0.2870731055736542, "learning_rate": 6.548282126936397e-05, "loss": 0.0128, "step": 128940 }, { "epoch": 82.18610579987254, "grad_norm": 0.3392900824546814, "learning_rate": 6.547758051090742e-05, "loss": 0.0166, "step": 128950 }, { "epoch": 82.19247928616953, "grad_norm": 0.4017775058746338, "learning_rate": 6.547233956438368e-05, "loss": 0.0185, "step": 128960 }, { "epoch": 82.19885277246654, "grad_norm": 0.15848763287067413, "learning_rate": 6.546709842985639e-05, "loss": 0.0116, "step": 128970 }, { "epoch": 82.20522625876355, "grad_norm": 0.8483330011367798, "learning_rate": 6.546185710738926e-05, "loss": 0.0211, "step": 128980 }, { "epoch": 82.21159974506055, "grad_norm": 0.41901329159736633, "learning_rate": 6.545661559704596e-05, "loss": 0.0168, "step": 128990 }, { "epoch": 82.21797323135755, "grad_norm": 0.4196595847606659, "learning_rate": 6.545137389889018e-05, "loss": 0.0248, "step": 129000 }, { "epoch": 82.22434671765456, "grad_norm": 0.26421019434928894, "learning_rate": 6.544613201298564e-05, "loss": 0.0272, "step": 129010 }, { "epoch": 82.23072020395156, "grad_norm": 0.35071468353271484, "learning_rate": 6.544088993939599e-05, "loss": 0.0206, "step": 129020 }, { "epoch": 82.23709369024857, "grad_norm": 0.20422649383544922, "learning_rate": 6.543564767818497e-05, "loss": 0.01, "step": 129030 }, { "epoch": 82.24346717654556, "grad_norm": 0.3799484670162201, "learning_rate": 6.543040522941624e-05, "loss": 0.0117, "step": 129040 }, { "epoch": 82.24984066284257, "grad_norm": 0.34045475721359253, "learning_rate": 6.542516259315352e-05, "loss": 0.0096, "step": 129050 }, { "epoch": 82.25621414913958, "grad_norm": 0.49087706208229065, "learning_rate": 6.541991976946052e-05, "loss": 0.0189, "step": 129060 }, { "epoch": 82.26258763543659, "grad_norm": 0.5129827857017517, "learning_rate": 6.541467675840093e-05, "loss": 0.0174, "step": 129070 }, { "epoch": 82.26896112173358, "grad_norm": 0.3261740505695343, "learning_rate": 6.540943356003847e-05, "loss": 0.015, "step": 129080 }, { "epoch": 82.27533460803059, "grad_norm": 0.2976730763912201, "learning_rate": 6.540419017443683e-05, "loss": 0.0153, "step": 129090 }, { "epoch": 82.2817080943276, "grad_norm": 0.4898168444633484, "learning_rate": 6.539894660165975e-05, "loss": 0.0189, "step": 129100 }, { "epoch": 82.2880815806246, "grad_norm": 0.4398637115955353, "learning_rate": 6.539370284177092e-05, "loss": 0.0152, "step": 129110 }, { "epoch": 82.2944550669216, "grad_norm": 0.24904365837574005, "learning_rate": 6.538845889483407e-05, "loss": 0.0113, "step": 129120 }, { "epoch": 82.30082855321861, "grad_norm": 0.13801807165145874, "learning_rate": 6.538321476091292e-05, "loss": 0.0184, "step": 129130 }, { "epoch": 82.30720203951562, "grad_norm": 0.4048437178134918, "learning_rate": 6.537797044007118e-05, "loss": 0.0183, "step": 129140 }, { "epoch": 82.31357552581262, "grad_norm": 0.32958391308784485, "learning_rate": 6.537272593237257e-05, "loss": 0.0179, "step": 129150 }, { "epoch": 82.31994901210962, "grad_norm": 0.44573330879211426, "learning_rate": 6.536748123788084e-05, "loss": 0.0157, "step": 129160 }, { "epoch": 82.32632249840663, "grad_norm": 0.44613009691238403, "learning_rate": 6.536223635665971e-05, "loss": 0.0185, "step": 129170 }, { "epoch": 82.33269598470363, "grad_norm": 0.2173072099685669, "learning_rate": 6.535699128877288e-05, "loss": 0.0164, "step": 129180 }, { "epoch": 82.33906947100064, "grad_norm": 0.4045519232749939, "learning_rate": 6.535174603428413e-05, "loss": 0.0191, "step": 129190 }, { "epoch": 82.34544295729764, "grad_norm": 0.4143335223197937, "learning_rate": 6.534650059325715e-05, "loss": 0.011, "step": 129200 }, { "epoch": 82.35181644359464, "grad_norm": 0.24158607423305511, "learning_rate": 6.534125496575573e-05, "loss": 0.0102, "step": 129210 }, { "epoch": 82.35818992989165, "grad_norm": 0.5204176306724548, "learning_rate": 6.533600915184356e-05, "loss": 0.0148, "step": 129220 }, { "epoch": 82.36456341618866, "grad_norm": 0.23025505244731903, "learning_rate": 6.533076315158438e-05, "loss": 0.0153, "step": 129230 }, { "epoch": 82.37093690248565, "grad_norm": 0.16277673840522766, "learning_rate": 6.532551696504198e-05, "loss": 0.0153, "step": 129240 }, { "epoch": 82.37731038878266, "grad_norm": 0.3283427655696869, "learning_rate": 6.532027059228006e-05, "loss": 0.0166, "step": 129250 }, { "epoch": 82.38368387507967, "grad_norm": 0.2775934338569641, "learning_rate": 6.53150240333624e-05, "loss": 0.0129, "step": 129260 }, { "epoch": 82.39005736137668, "grad_norm": 0.3321723937988281, "learning_rate": 6.530977728835274e-05, "loss": 0.0105, "step": 129270 }, { "epoch": 82.39643084767367, "grad_norm": 0.2265108823776245, "learning_rate": 6.530453035731482e-05, "loss": 0.0127, "step": 129280 }, { "epoch": 82.40280433397068, "grad_norm": 0.5545596480369568, "learning_rate": 6.529928324031242e-05, "loss": 0.0168, "step": 129290 }, { "epoch": 82.40917782026769, "grad_norm": 0.36006003618240356, "learning_rate": 6.529403593740929e-05, "loss": 0.0106, "step": 129300 }, { "epoch": 82.4155513065647, "grad_norm": 0.26289042830467224, "learning_rate": 6.528878844866916e-05, "loss": 0.0122, "step": 129310 }, { "epoch": 82.4219247928617, "grad_norm": 0.188180074095726, "learning_rate": 6.528354077415582e-05, "loss": 0.0134, "step": 129320 }, { "epoch": 82.4282982791587, "grad_norm": 0.38672778010368347, "learning_rate": 6.527829291393304e-05, "loss": 0.0141, "step": 129330 }, { "epoch": 82.4346717654557, "grad_norm": 0.5343759059906006, "learning_rate": 6.527304486806456e-05, "loss": 0.0179, "step": 129340 }, { "epoch": 82.44104525175271, "grad_norm": 0.2163008153438568, "learning_rate": 6.526779663661419e-05, "loss": 0.0101, "step": 129350 }, { "epoch": 82.44741873804972, "grad_norm": 0.41879984736442566, "learning_rate": 6.526254821964565e-05, "loss": 0.0147, "step": 129360 }, { "epoch": 82.45379222434671, "grad_norm": 0.8188528418540955, "learning_rate": 6.525729961722276e-05, "loss": 0.0201, "step": 129370 }, { "epoch": 82.46016571064372, "grad_norm": 0.864723265171051, "learning_rate": 6.525205082940925e-05, "loss": 0.0158, "step": 129380 }, { "epoch": 82.46653919694073, "grad_norm": 0.47532087564468384, "learning_rate": 6.524680185626894e-05, "loss": 0.0176, "step": 129390 }, { "epoch": 82.47291268323774, "grad_norm": 0.44338297843933105, "learning_rate": 6.52415526978656e-05, "loss": 0.0165, "step": 129400 }, { "epoch": 82.47928616953473, "grad_norm": 0.42772194743156433, "learning_rate": 6.523630335426298e-05, "loss": 0.0162, "step": 129410 }, { "epoch": 82.48565965583174, "grad_norm": 0.5249453186988831, "learning_rate": 6.523105382552491e-05, "loss": 0.013, "step": 129420 }, { "epoch": 82.49203314212875, "grad_norm": 0.6009605526924133, "learning_rate": 6.522580411171515e-05, "loss": 0.018, "step": 129430 }, { "epoch": 82.49840662842576, "grad_norm": 0.46704569458961487, "learning_rate": 6.522055421289749e-05, "loss": 0.0135, "step": 129440 }, { "epoch": 82.50478011472275, "grad_norm": 0.3131047487258911, "learning_rate": 6.521530412913573e-05, "loss": 0.0136, "step": 129450 }, { "epoch": 82.51115360101976, "grad_norm": 0.6475018262863159, "learning_rate": 6.521005386049366e-05, "loss": 0.0243, "step": 129460 }, { "epoch": 82.51752708731676, "grad_norm": 0.21416524052619934, "learning_rate": 6.520480340703508e-05, "loss": 0.0173, "step": 129470 }, { "epoch": 82.52390057361377, "grad_norm": 0.5421136021614075, "learning_rate": 6.519955276882375e-05, "loss": 0.0211, "step": 129480 }, { "epoch": 82.53027405991077, "grad_norm": 0.40036237239837646, "learning_rate": 6.519430194592355e-05, "loss": 0.0148, "step": 129490 }, { "epoch": 82.53664754620777, "grad_norm": 0.4670766592025757, "learning_rate": 6.51890509383982e-05, "loss": 0.0117, "step": 129500 }, { "epoch": 82.54302103250478, "grad_norm": 0.3858421742916107, "learning_rate": 6.518379974631156e-05, "loss": 0.0172, "step": 129510 }, { "epoch": 82.54939451880179, "grad_norm": 0.3115980625152588, "learning_rate": 6.51785483697274e-05, "loss": 0.0198, "step": 129520 }, { "epoch": 82.55576800509878, "grad_norm": 0.15540215373039246, "learning_rate": 6.517329680870956e-05, "loss": 0.0117, "step": 129530 }, { "epoch": 82.56214149139579, "grad_norm": 0.3400847613811493, "learning_rate": 6.516804506332183e-05, "loss": 0.0129, "step": 129540 }, { "epoch": 82.5685149776928, "grad_norm": 0.6210516095161438, "learning_rate": 6.516279313362804e-05, "loss": 0.0239, "step": 129550 }, { "epoch": 82.57488846398981, "grad_norm": 0.6158271431922913, "learning_rate": 6.515754101969199e-05, "loss": 0.0227, "step": 129560 }, { "epoch": 82.5812619502868, "grad_norm": 0.3773817718029022, "learning_rate": 6.515228872157752e-05, "loss": 0.0202, "step": 129570 }, { "epoch": 82.58763543658381, "grad_norm": 0.275083065032959, "learning_rate": 6.514703623934841e-05, "loss": 0.0186, "step": 129580 }, { "epoch": 82.59400892288082, "grad_norm": 0.5150445699691772, "learning_rate": 6.514178357306852e-05, "loss": 0.0225, "step": 129590 }, { "epoch": 82.60038240917783, "grad_norm": 0.2508932948112488, "learning_rate": 6.513653072280166e-05, "loss": 0.022, "step": 129600 }, { "epoch": 82.60675589547482, "grad_norm": 0.25177550315856934, "learning_rate": 6.513127768861169e-05, "loss": 0.014, "step": 129610 }, { "epoch": 82.61312938177183, "grad_norm": 0.3836192488670349, "learning_rate": 6.512602447056238e-05, "loss": 0.0193, "step": 129620 }, { "epoch": 82.61950286806884, "grad_norm": 0.26287126541137695, "learning_rate": 6.51207710687176e-05, "loss": 0.0142, "step": 129630 }, { "epoch": 82.62587635436584, "grad_norm": 0.2941252291202545, "learning_rate": 6.511551748314117e-05, "loss": 0.0148, "step": 129640 }, { "epoch": 82.63224984066284, "grad_norm": 0.4325369894504547, "learning_rate": 6.511026371389693e-05, "loss": 0.0126, "step": 129650 }, { "epoch": 82.63862332695984, "grad_norm": 0.35974374413490295, "learning_rate": 6.510500976104872e-05, "loss": 0.0147, "step": 129660 }, { "epoch": 82.64499681325685, "grad_norm": 0.8904023766517639, "learning_rate": 6.50997556246604e-05, "loss": 0.0244, "step": 129670 }, { "epoch": 82.65137029955386, "grad_norm": 0.4447114169597626, "learning_rate": 6.509450130479577e-05, "loss": 0.0179, "step": 129680 }, { "epoch": 82.65774378585085, "grad_norm": 0.378740519285202, "learning_rate": 6.50892468015187e-05, "loss": 0.0162, "step": 129690 }, { "epoch": 82.66411727214786, "grad_norm": 0.6596896052360535, "learning_rate": 6.508399211489303e-05, "loss": 0.0219, "step": 129700 }, { "epoch": 82.67049075844487, "grad_norm": 0.4516632854938507, "learning_rate": 6.507873724498264e-05, "loss": 0.0134, "step": 129710 }, { "epoch": 82.67686424474188, "grad_norm": 0.45621487498283386, "learning_rate": 6.507348219185134e-05, "loss": 0.0122, "step": 129720 }, { "epoch": 82.68323773103887, "grad_norm": 0.3335203528404236, "learning_rate": 6.506822695556298e-05, "loss": 0.0242, "step": 129730 }, { "epoch": 82.68961121733588, "grad_norm": 0.46405524015426636, "learning_rate": 6.506297153618147e-05, "loss": 0.0154, "step": 129740 }, { "epoch": 82.69598470363289, "grad_norm": 0.30529388785362244, "learning_rate": 6.505771593377061e-05, "loss": 0.0166, "step": 129750 }, { "epoch": 82.7023581899299, "grad_norm": 0.6802051663398743, "learning_rate": 6.50524601483943e-05, "loss": 0.0158, "step": 129760 }, { "epoch": 82.70873167622689, "grad_norm": 0.4951118528842926, "learning_rate": 6.504720418011637e-05, "loss": 0.0152, "step": 129770 }, { "epoch": 82.7151051625239, "grad_norm": 0.4721625745296478, "learning_rate": 6.504194802900072e-05, "loss": 0.0129, "step": 129780 }, { "epoch": 82.7214786488209, "grad_norm": 0.4822039008140564, "learning_rate": 6.50366916951112e-05, "loss": 0.0277, "step": 129790 }, { "epoch": 82.72785213511791, "grad_norm": 0.3109472990036011, "learning_rate": 6.503143517851167e-05, "loss": 0.0108, "step": 129800 }, { "epoch": 82.73422562141491, "grad_norm": 0.48140203952789307, "learning_rate": 6.5026178479266e-05, "loss": 0.0155, "step": 129810 }, { "epoch": 82.74059910771192, "grad_norm": 0.39591076970100403, "learning_rate": 6.502092159743809e-05, "loss": 0.0209, "step": 129820 }, { "epoch": 82.74697259400892, "grad_norm": 0.23049214482307434, "learning_rate": 6.50156645330918e-05, "loss": 0.0128, "step": 129830 }, { "epoch": 82.75334608030593, "grad_norm": 0.2730417251586914, "learning_rate": 6.5010407286291e-05, "loss": 0.0293, "step": 129840 }, { "epoch": 82.75971956660293, "grad_norm": 0.24825452268123627, "learning_rate": 6.50051498570996e-05, "loss": 0.0175, "step": 129850 }, { "epoch": 82.76609305289993, "grad_norm": 0.5648978352546692, "learning_rate": 6.499989224558143e-05, "loss": 0.0121, "step": 129860 }, { "epoch": 82.77246653919694, "grad_norm": 0.3948519229888916, "learning_rate": 6.499463445180043e-05, "loss": 0.0106, "step": 129870 }, { "epoch": 82.77884002549395, "grad_norm": 0.6222568154335022, "learning_rate": 6.498937647582046e-05, "loss": 0.0169, "step": 129880 }, { "epoch": 82.78521351179094, "grad_norm": 0.43854808807373047, "learning_rate": 6.49841183177054e-05, "loss": 0.0239, "step": 129890 }, { "epoch": 82.79158699808795, "grad_norm": 0.37719249725341797, "learning_rate": 6.497885997751918e-05, "loss": 0.018, "step": 129900 }, { "epoch": 82.79796048438496, "grad_norm": 0.4126072824001312, "learning_rate": 6.497360145532565e-05, "loss": 0.0136, "step": 129910 }, { "epoch": 82.80433397068197, "grad_norm": 0.442438542842865, "learning_rate": 6.496834275118872e-05, "loss": 0.0142, "step": 129920 }, { "epoch": 82.81070745697897, "grad_norm": 0.44087305665016174, "learning_rate": 6.496308386517232e-05, "loss": 0.0101, "step": 129930 }, { "epoch": 82.81708094327597, "grad_norm": 0.46413296461105347, "learning_rate": 6.49578247973403e-05, "loss": 0.0097, "step": 129940 }, { "epoch": 82.82345442957298, "grad_norm": 0.2547758221626282, "learning_rate": 6.49525655477566e-05, "loss": 0.0105, "step": 129950 }, { "epoch": 82.82982791586998, "grad_norm": 0.5181885957717896, "learning_rate": 6.494730611648509e-05, "loss": 0.0163, "step": 129960 }, { "epoch": 82.83620140216699, "grad_norm": 0.2818664610385895, "learning_rate": 6.494204650358973e-05, "loss": 0.0128, "step": 129970 }, { "epoch": 82.84257488846399, "grad_norm": 0.6530259847640991, "learning_rate": 6.493678670913436e-05, "loss": 0.0127, "step": 129980 }, { "epoch": 82.848948374761, "grad_norm": 0.30274635553359985, "learning_rate": 6.493152673318296e-05, "loss": 0.029, "step": 129990 }, { "epoch": 82.855321861058, "grad_norm": 0.5398173928260803, "learning_rate": 6.49262665757994e-05, "loss": 0.0272, "step": 130000 }, { "epoch": 82.86169534735501, "grad_norm": 0.31484073400497437, "learning_rate": 6.49210062370476e-05, "loss": 0.0146, "step": 130010 }, { "epoch": 82.868068833652, "grad_norm": 0.338051438331604, "learning_rate": 6.491574571699149e-05, "loss": 0.0147, "step": 130020 }, { "epoch": 82.87444231994901, "grad_norm": 0.4722456634044647, "learning_rate": 6.4910485015695e-05, "loss": 0.0237, "step": 130030 }, { "epoch": 82.88081580624602, "grad_norm": 0.20754572749137878, "learning_rate": 6.490522413322205e-05, "loss": 0.0111, "step": 130040 }, { "epoch": 82.88718929254303, "grad_norm": 0.25654107332229614, "learning_rate": 6.489996306963652e-05, "loss": 0.0155, "step": 130050 }, { "epoch": 82.89356277884002, "grad_norm": 0.43719482421875, "learning_rate": 6.489470182500239e-05, "loss": 0.0136, "step": 130060 }, { "epoch": 82.89993626513703, "grad_norm": 0.4859764277935028, "learning_rate": 6.488944039938355e-05, "loss": 0.0172, "step": 130070 }, { "epoch": 82.90630975143404, "grad_norm": 0.35515403747558594, "learning_rate": 6.488417879284399e-05, "loss": 0.0269, "step": 130080 }, { "epoch": 82.91268323773104, "grad_norm": 0.4701721966266632, "learning_rate": 6.487891700544757e-05, "loss": 0.0114, "step": 130090 }, { "epoch": 82.91905672402804, "grad_norm": 0.2615286707878113, "learning_rate": 6.487365503725827e-05, "loss": 0.014, "step": 130100 }, { "epoch": 82.92543021032505, "grad_norm": 0.27387645840644836, "learning_rate": 6.486839288834003e-05, "loss": 0.0105, "step": 130110 }, { "epoch": 82.93180369662205, "grad_norm": 0.5262463092803955, "learning_rate": 6.486313055875676e-05, "loss": 0.0179, "step": 130120 }, { "epoch": 82.93817718291906, "grad_norm": 0.5102649331092834, "learning_rate": 6.485786804857244e-05, "loss": 0.0211, "step": 130130 }, { "epoch": 82.94455066921606, "grad_norm": 0.2652188539505005, "learning_rate": 6.485260535785099e-05, "loss": 0.0179, "step": 130140 }, { "epoch": 82.95092415551306, "grad_norm": 0.22846364974975586, "learning_rate": 6.484734248665634e-05, "loss": 0.0208, "step": 130150 }, { "epoch": 82.95729764181007, "grad_norm": 0.35175174474716187, "learning_rate": 6.484207943505249e-05, "loss": 0.0163, "step": 130160 }, { "epoch": 82.96367112810708, "grad_norm": 0.3670831620693207, "learning_rate": 6.483681620310334e-05, "loss": 0.0195, "step": 130170 }, { "epoch": 82.97004461440407, "grad_norm": 0.21414296329021454, "learning_rate": 6.483155279087287e-05, "loss": 0.0161, "step": 130180 }, { "epoch": 82.97641810070108, "grad_norm": 0.38655275106430054, "learning_rate": 6.482628919842503e-05, "loss": 0.0269, "step": 130190 }, { "epoch": 82.98279158699809, "grad_norm": 0.5764787793159485, "learning_rate": 6.48210254258238e-05, "loss": 0.0166, "step": 130200 }, { "epoch": 82.9891650732951, "grad_norm": 0.23272280395030975, "learning_rate": 6.481576147313308e-05, "loss": 0.0165, "step": 130210 }, { "epoch": 82.99553855959209, "grad_norm": 0.3530321419239044, "learning_rate": 6.48104973404169e-05, "loss": 0.0178, "step": 130220 }, { "epoch": 83.0019120458891, "grad_norm": 0.5218722820281982, "learning_rate": 6.480523302773918e-05, "loss": 0.0168, "step": 130230 }, { "epoch": 83.00828553218611, "grad_norm": 0.31366652250289917, "learning_rate": 6.47999685351639e-05, "loss": 0.0139, "step": 130240 }, { "epoch": 83.01465901848312, "grad_norm": 0.3952755928039551, "learning_rate": 6.479470386275503e-05, "loss": 0.0198, "step": 130250 }, { "epoch": 83.02103250478011, "grad_norm": 0.319271445274353, "learning_rate": 6.478943901057654e-05, "loss": 0.0164, "step": 130260 }, { "epoch": 83.02740599107712, "grad_norm": 0.2707465887069702, "learning_rate": 6.478417397869241e-05, "loss": 0.0093, "step": 130270 }, { "epoch": 83.03377947737413, "grad_norm": 0.43230336904525757, "learning_rate": 6.47789087671666e-05, "loss": 0.0164, "step": 130280 }, { "epoch": 83.04015296367113, "grad_norm": 0.2873488664627075, "learning_rate": 6.477364337606309e-05, "loss": 0.0109, "step": 130290 }, { "epoch": 83.04652644996813, "grad_norm": 0.4224412739276886, "learning_rate": 6.476837780544588e-05, "loss": 0.0305, "step": 130300 }, { "epoch": 83.05289993626513, "grad_norm": 0.24483788013458252, "learning_rate": 6.476311205537892e-05, "loss": 0.0136, "step": 130310 }, { "epoch": 83.05927342256214, "grad_norm": 0.32233986258506775, "learning_rate": 6.475784612592622e-05, "loss": 0.0242, "step": 130320 }, { "epoch": 83.06564690885915, "grad_norm": 0.5004836916923523, "learning_rate": 6.475258001715175e-05, "loss": 0.0269, "step": 130330 }, { "epoch": 83.07202039515614, "grad_norm": 0.46104440093040466, "learning_rate": 6.47473137291195e-05, "loss": 0.0272, "step": 130340 }, { "epoch": 83.07839388145315, "grad_norm": 0.47214803099632263, "learning_rate": 6.474204726189348e-05, "loss": 0.0177, "step": 130350 }, { "epoch": 83.08476736775016, "grad_norm": 0.7090554237365723, "learning_rate": 6.473678061553765e-05, "loss": 0.0194, "step": 130360 }, { "epoch": 83.09114085404717, "grad_norm": 0.5701329708099365, "learning_rate": 6.473151379011603e-05, "loss": 0.02, "step": 130370 }, { "epoch": 83.09751434034416, "grad_norm": 0.4428839087486267, "learning_rate": 6.47262467856926e-05, "loss": 0.0136, "step": 130380 }, { "epoch": 83.10388782664117, "grad_norm": 0.23075000941753387, "learning_rate": 6.472097960233139e-05, "loss": 0.0144, "step": 130390 }, { "epoch": 83.11026131293818, "grad_norm": 0.3052363693714142, "learning_rate": 6.471571224009636e-05, "loss": 0.0191, "step": 130400 }, { "epoch": 83.11663479923519, "grad_norm": 0.32743799686431885, "learning_rate": 6.471044469905153e-05, "loss": 0.0323, "step": 130410 }, { "epoch": 83.12300828553218, "grad_norm": 0.6052147746086121, "learning_rate": 6.470517697926093e-05, "loss": 0.0194, "step": 130420 }, { "epoch": 83.12938177182919, "grad_norm": 0.49153396487236023, "learning_rate": 6.469990908078854e-05, "loss": 0.0109, "step": 130430 }, { "epoch": 83.1357552581262, "grad_norm": 0.32386717200279236, "learning_rate": 6.469464100369835e-05, "loss": 0.0137, "step": 130440 }, { "epoch": 83.1421287444232, "grad_norm": 0.3301803767681122, "learning_rate": 6.468937274805442e-05, "loss": 0.0159, "step": 130450 }, { "epoch": 83.1485022307202, "grad_norm": 0.2878493666648865, "learning_rate": 6.468410431392072e-05, "loss": 0.012, "step": 130460 }, { "epoch": 83.1548757170172, "grad_norm": 0.23186154663562775, "learning_rate": 6.467883570136132e-05, "loss": 0.0112, "step": 130470 }, { "epoch": 83.16124920331421, "grad_norm": 0.35240933299064636, "learning_rate": 6.467356691044019e-05, "loss": 0.0126, "step": 130480 }, { "epoch": 83.16762268961122, "grad_norm": 0.7586543560028076, "learning_rate": 6.466829794122136e-05, "loss": 0.0225, "step": 130490 }, { "epoch": 83.17399617590821, "grad_norm": 0.25580722093582153, "learning_rate": 6.466302879376887e-05, "loss": 0.0153, "step": 130500 }, { "epoch": 83.18036966220522, "grad_norm": 0.38005533814430237, "learning_rate": 6.465775946814674e-05, "loss": 0.0144, "step": 130510 }, { "epoch": 83.18674314850223, "grad_norm": 0.33872169256210327, "learning_rate": 6.465248996441899e-05, "loss": 0.0138, "step": 130520 }, { "epoch": 83.19311663479924, "grad_norm": 0.2315920889377594, "learning_rate": 6.464722028264966e-05, "loss": 0.0117, "step": 130530 }, { "epoch": 83.19949012109623, "grad_norm": 0.26399996876716614, "learning_rate": 6.464195042290277e-05, "loss": 0.0091, "step": 130540 }, { "epoch": 83.20586360739324, "grad_norm": 0.3699013292789459, "learning_rate": 6.463668038524234e-05, "loss": 0.021, "step": 130550 }, { "epoch": 83.21223709369025, "grad_norm": 0.21113309264183044, "learning_rate": 6.463141016973245e-05, "loss": 0.0099, "step": 130560 }, { "epoch": 83.21861057998726, "grad_norm": 0.6450787782669067, "learning_rate": 6.46261397764371e-05, "loss": 0.0136, "step": 130570 }, { "epoch": 83.22498406628426, "grad_norm": 0.2625298798084259, "learning_rate": 6.462086920542034e-05, "loss": 0.0132, "step": 130580 }, { "epoch": 83.23135755258126, "grad_norm": 0.23705129325389862, "learning_rate": 6.461559845674621e-05, "loss": 0.0177, "step": 130590 }, { "epoch": 83.23773103887827, "grad_norm": 0.5440493226051331, "learning_rate": 6.461032753047878e-05, "loss": 0.0222, "step": 130600 }, { "epoch": 83.24410452517527, "grad_norm": 0.23699304461479187, "learning_rate": 6.460505642668206e-05, "loss": 0.0171, "step": 130610 }, { "epoch": 83.25047801147228, "grad_norm": 0.42001861333847046, "learning_rate": 6.459978514542011e-05, "loss": 0.0138, "step": 130620 }, { "epoch": 83.25685149776928, "grad_norm": 0.3001210689544678, "learning_rate": 6.4594513686757e-05, "loss": 0.024, "step": 130630 }, { "epoch": 83.26322498406628, "grad_norm": 0.2653592526912689, "learning_rate": 6.458924205075676e-05, "loss": 0.0296, "step": 130640 }, { "epoch": 83.26959847036329, "grad_norm": 0.4558902680873871, "learning_rate": 6.458397023748346e-05, "loss": 0.0128, "step": 130650 }, { "epoch": 83.2759719566603, "grad_norm": 0.4752671420574188, "learning_rate": 6.457869824700116e-05, "loss": 0.0117, "step": 130660 }, { "epoch": 83.2823454429573, "grad_norm": 0.20817027986049652, "learning_rate": 6.45734260793739e-05, "loss": 0.0099, "step": 130670 }, { "epoch": 83.2887189292543, "grad_norm": 0.3515315353870392, "learning_rate": 6.456815373466575e-05, "loss": 0.024, "step": 130680 }, { "epoch": 83.29509241555131, "grad_norm": 0.5317285060882568, "learning_rate": 6.456288121294079e-05, "loss": 0.0113, "step": 130690 }, { "epoch": 83.30146590184832, "grad_norm": 0.26010337471961975, "learning_rate": 6.455760851426307e-05, "loss": 0.0098, "step": 130700 }, { "epoch": 83.30783938814531, "grad_norm": 0.40324848890304565, "learning_rate": 6.455233563869665e-05, "loss": 0.0267, "step": 130710 }, { "epoch": 83.31421287444232, "grad_norm": 0.15192781388759613, "learning_rate": 6.454706258630564e-05, "loss": 0.019, "step": 130720 }, { "epoch": 83.32058636073933, "grad_norm": 0.37542569637298584, "learning_rate": 6.454178935715405e-05, "loss": 0.0175, "step": 130730 }, { "epoch": 83.32695984703633, "grad_norm": 0.1971392035484314, "learning_rate": 6.4536515951306e-05, "loss": 0.0183, "step": 130740 }, { "epoch": 83.33333333333333, "grad_norm": 0.3407244384288788, "learning_rate": 6.453124236882557e-05, "loss": 0.0112, "step": 130750 }, { "epoch": 83.33970681963034, "grad_norm": 0.2081877589225769, "learning_rate": 6.452596860977681e-05, "loss": 0.0119, "step": 130760 }, { "epoch": 83.34608030592734, "grad_norm": 0.2930329442024231, "learning_rate": 6.452069467422384e-05, "loss": 0.0093, "step": 130770 }, { "epoch": 83.35245379222435, "grad_norm": 0.41556593775749207, "learning_rate": 6.451542056223069e-05, "loss": 0.0113, "step": 130780 }, { "epoch": 83.35882727852135, "grad_norm": 0.31988170742988586, "learning_rate": 6.451014627386149e-05, "loss": 0.0161, "step": 130790 }, { "epoch": 83.36520076481835, "grad_norm": 0.17635323107242584, "learning_rate": 6.45048718091803e-05, "loss": 0.0124, "step": 130800 }, { "epoch": 83.37157425111536, "grad_norm": 0.2683553993701935, "learning_rate": 6.449959716825124e-05, "loss": 0.0154, "step": 130810 }, { "epoch": 83.37794773741237, "grad_norm": 0.6304535865783691, "learning_rate": 6.449432235113839e-05, "loss": 0.0132, "step": 130820 }, { "epoch": 83.38432122370936, "grad_norm": 0.42291203141212463, "learning_rate": 6.448904735790582e-05, "loss": 0.0131, "step": 130830 }, { "epoch": 83.39069471000637, "grad_norm": 0.3073234558105469, "learning_rate": 6.448377218861765e-05, "loss": 0.0258, "step": 130840 }, { "epoch": 83.39706819630338, "grad_norm": 0.4783661365509033, "learning_rate": 6.447849684333797e-05, "loss": 0.0119, "step": 130850 }, { "epoch": 83.40344168260039, "grad_norm": 0.24375177919864655, "learning_rate": 6.44732213221309e-05, "loss": 0.0188, "step": 130860 }, { "epoch": 83.40981516889738, "grad_norm": 0.2408261001110077, "learning_rate": 6.44679456250605e-05, "loss": 0.0226, "step": 130870 }, { "epoch": 83.41618865519439, "grad_norm": 0.29230937361717224, "learning_rate": 6.446266975219093e-05, "loss": 0.0101, "step": 130880 }, { "epoch": 83.4225621414914, "grad_norm": 0.2254548966884613, "learning_rate": 6.445739370358625e-05, "loss": 0.0139, "step": 130890 }, { "epoch": 83.4289356277884, "grad_norm": 0.6339810490608215, "learning_rate": 6.44521174793106e-05, "loss": 0.0145, "step": 130900 }, { "epoch": 83.4353091140854, "grad_norm": 0.4953218996524811, "learning_rate": 6.444684107942805e-05, "loss": 0.0219, "step": 130910 }, { "epoch": 83.4416826003824, "grad_norm": 0.34992629289627075, "learning_rate": 6.444156450400276e-05, "loss": 0.0142, "step": 130920 }, { "epoch": 83.44805608667941, "grad_norm": 0.6577014923095703, "learning_rate": 6.443628775309883e-05, "loss": 0.0168, "step": 130930 }, { "epoch": 83.45442957297642, "grad_norm": 0.32354632019996643, "learning_rate": 6.443101082678036e-05, "loss": 0.0115, "step": 130940 }, { "epoch": 83.46080305927342, "grad_norm": 0.29779189825057983, "learning_rate": 6.44257337251115e-05, "loss": 0.0114, "step": 130950 }, { "epoch": 83.46717654557042, "grad_norm": 0.6189366579055786, "learning_rate": 6.442045644815634e-05, "loss": 0.0181, "step": 130960 }, { "epoch": 83.47355003186743, "grad_norm": 0.36857569217681885, "learning_rate": 6.441517899597904e-05, "loss": 0.0144, "step": 130970 }, { "epoch": 83.47992351816444, "grad_norm": 0.38785719871520996, "learning_rate": 6.440990136864368e-05, "loss": 0.0207, "step": 130980 }, { "epoch": 83.48629700446143, "grad_norm": 0.21416784822940826, "learning_rate": 6.440462356621441e-05, "loss": 0.0171, "step": 130990 }, { "epoch": 83.49267049075844, "grad_norm": 0.7020421028137207, "learning_rate": 6.439934558875537e-05, "loss": 0.0147, "step": 131000 }, { "epoch": 83.49904397705545, "grad_norm": 0.2223960906267166, "learning_rate": 6.439406743633071e-05, "loss": 0.0131, "step": 131010 }, { "epoch": 83.50541746335246, "grad_norm": 0.8205466866493225, "learning_rate": 6.438878910900452e-05, "loss": 0.0187, "step": 131020 }, { "epoch": 83.51179094964945, "grad_norm": 0.3020743727684021, "learning_rate": 6.438351060684096e-05, "loss": 0.0186, "step": 131030 }, { "epoch": 83.51816443594646, "grad_norm": 0.43736720085144043, "learning_rate": 6.437823192990415e-05, "loss": 0.0142, "step": 131040 }, { "epoch": 83.52453792224347, "grad_norm": 0.5254865884780884, "learning_rate": 6.437295307825828e-05, "loss": 0.0093, "step": 131050 }, { "epoch": 83.53091140854048, "grad_norm": 0.2729909420013428, "learning_rate": 6.436767405196743e-05, "loss": 0.0167, "step": 131060 }, { "epoch": 83.53728489483747, "grad_norm": 0.3024376928806305, "learning_rate": 6.43623948510958e-05, "loss": 0.0115, "step": 131070 }, { "epoch": 83.54365838113448, "grad_norm": 0.3646664321422577, "learning_rate": 6.435711547570748e-05, "loss": 0.0109, "step": 131080 }, { "epoch": 83.55003186743149, "grad_norm": 0.6028757691383362, "learning_rate": 6.435183592586669e-05, "loss": 0.0237, "step": 131090 }, { "epoch": 83.5564053537285, "grad_norm": 0.16825492680072784, "learning_rate": 6.43465562016375e-05, "loss": 0.0145, "step": 131100 }, { "epoch": 83.56277884002549, "grad_norm": 0.6626102328300476, "learning_rate": 6.434127630308416e-05, "loss": 0.0326, "step": 131110 }, { "epoch": 83.5691523263225, "grad_norm": 0.3023071885108948, "learning_rate": 6.433599623027072e-05, "loss": 0.0124, "step": 131120 }, { "epoch": 83.5755258126195, "grad_norm": 0.3567672371864319, "learning_rate": 6.433071598326142e-05, "loss": 0.0128, "step": 131130 }, { "epoch": 83.58189929891651, "grad_norm": 0.42792874574661255, "learning_rate": 6.432543556212038e-05, "loss": 0.0192, "step": 131140 }, { "epoch": 83.5882727852135, "grad_norm": 0.1426030844449997, "learning_rate": 6.432015496691179e-05, "loss": 0.0232, "step": 131150 }, { "epoch": 83.59464627151051, "grad_norm": 0.16682283580303192, "learning_rate": 6.431487419769978e-05, "loss": 0.0131, "step": 131160 }, { "epoch": 83.60101975780752, "grad_norm": 0.30909544229507446, "learning_rate": 6.430959325454852e-05, "loss": 0.0187, "step": 131170 }, { "epoch": 83.60739324410453, "grad_norm": 0.27484557032585144, "learning_rate": 6.430431213752222e-05, "loss": 0.0138, "step": 131180 }, { "epoch": 83.61376673040154, "grad_norm": 0.5512232780456543, "learning_rate": 6.4299030846685e-05, "loss": 0.0177, "step": 131190 }, { "epoch": 83.62014021669853, "grad_norm": 0.3164660334587097, "learning_rate": 6.429374938210108e-05, "loss": 0.013, "step": 131200 }, { "epoch": 83.62651370299554, "grad_norm": 0.5080949664115906, "learning_rate": 6.42884677438346e-05, "loss": 0.0129, "step": 131210 }, { "epoch": 83.63288718929255, "grad_norm": 0.650016725063324, "learning_rate": 6.428318593194974e-05, "loss": 0.0202, "step": 131220 }, { "epoch": 83.63926067558955, "grad_norm": 0.8462024331092834, "learning_rate": 6.427790394651067e-05, "loss": 0.016, "step": 131230 }, { "epoch": 83.64563416188655, "grad_norm": 0.5229950547218323, "learning_rate": 6.42726217875816e-05, "loss": 0.0195, "step": 131240 }, { "epoch": 83.65200764818356, "grad_norm": 0.3396455645561218, "learning_rate": 6.426733945522672e-05, "loss": 0.0155, "step": 131250 }, { "epoch": 83.65838113448056, "grad_norm": 0.3319931924343109, "learning_rate": 6.426205694951016e-05, "loss": 0.0154, "step": 131260 }, { "epoch": 83.66475462077757, "grad_norm": 0.5990366339683533, "learning_rate": 6.425677427049616e-05, "loss": 0.0163, "step": 131270 }, { "epoch": 83.67112810707457, "grad_norm": 0.4167865812778473, "learning_rate": 6.425149141824889e-05, "loss": 0.0107, "step": 131280 }, { "epoch": 83.67750159337157, "grad_norm": 0.27583926916122437, "learning_rate": 6.424620839283255e-05, "loss": 0.0121, "step": 131290 }, { "epoch": 83.68387507966858, "grad_norm": 0.252920925617218, "learning_rate": 6.424092519431132e-05, "loss": 0.0138, "step": 131300 }, { "epoch": 83.69024856596559, "grad_norm": 0.6350367665290833, "learning_rate": 6.423564182274942e-05, "loss": 0.0156, "step": 131310 }, { "epoch": 83.69662205226258, "grad_norm": 0.7880165576934814, "learning_rate": 6.423035827821101e-05, "loss": 0.016, "step": 131320 }, { "epoch": 83.70299553855959, "grad_norm": 0.4576292634010315, "learning_rate": 6.42250745607603e-05, "loss": 0.0129, "step": 131330 }, { "epoch": 83.7093690248566, "grad_norm": 0.30360135436058044, "learning_rate": 6.421979067046153e-05, "loss": 0.0139, "step": 131340 }, { "epoch": 83.7157425111536, "grad_norm": 0.32643282413482666, "learning_rate": 6.421450660737887e-05, "loss": 0.0169, "step": 131350 }, { "epoch": 83.7221159974506, "grad_norm": 0.42639976739883423, "learning_rate": 6.420922237157654e-05, "loss": 0.0206, "step": 131360 }, { "epoch": 83.72848948374761, "grad_norm": 0.4696943759918213, "learning_rate": 6.420393796311874e-05, "loss": 0.0209, "step": 131370 }, { "epoch": 83.73486297004462, "grad_norm": 0.36730653047561646, "learning_rate": 6.419865338206968e-05, "loss": 0.0118, "step": 131380 }, { "epoch": 83.74123645634162, "grad_norm": 0.38335904479026794, "learning_rate": 6.419336862849356e-05, "loss": 0.0132, "step": 131390 }, { "epoch": 83.74760994263862, "grad_norm": 0.32962265610694885, "learning_rate": 6.418808370245462e-05, "loss": 0.0228, "step": 131400 }, { "epoch": 83.75398342893563, "grad_norm": 0.5191961526870728, "learning_rate": 6.418279860401708e-05, "loss": 0.0211, "step": 131410 }, { "epoch": 83.76035691523263, "grad_norm": 0.44134435057640076, "learning_rate": 6.417751333324513e-05, "loss": 0.0246, "step": 131420 }, { "epoch": 83.76673040152964, "grad_norm": 0.4587678015232086, "learning_rate": 6.417222789020303e-05, "loss": 0.0123, "step": 131430 }, { "epoch": 83.77310388782664, "grad_norm": 0.3685089349746704, "learning_rate": 6.416694227495496e-05, "loss": 0.014, "step": 131440 }, { "epoch": 83.77947737412364, "grad_norm": 0.23504896461963654, "learning_rate": 6.416165648756517e-05, "loss": 0.0142, "step": 131450 }, { "epoch": 83.78585086042065, "grad_norm": 0.4158063232898712, "learning_rate": 6.415637052809788e-05, "loss": 0.0143, "step": 131460 }, { "epoch": 83.79222434671766, "grad_norm": 0.2853351831436157, "learning_rate": 6.415108439661733e-05, "loss": 0.014, "step": 131470 }, { "epoch": 83.79859783301465, "grad_norm": 0.3286244869232178, "learning_rate": 6.414579809318775e-05, "loss": 0.0155, "step": 131480 }, { "epoch": 83.80497131931166, "grad_norm": 0.4894074499607086, "learning_rate": 6.414051161787334e-05, "loss": 0.0183, "step": 131490 }, { "epoch": 83.81134480560867, "grad_norm": 0.2464938759803772, "learning_rate": 6.413522497073839e-05, "loss": 0.0116, "step": 131500 }, { "epoch": 83.81771829190568, "grad_norm": 0.15380670130252838, "learning_rate": 6.41299381518471e-05, "loss": 0.0121, "step": 131510 }, { "epoch": 83.82409177820267, "grad_norm": 0.5270677208900452, "learning_rate": 6.412465116126371e-05, "loss": 0.0123, "step": 131520 }, { "epoch": 83.83046526449968, "grad_norm": 0.35303524136543274, "learning_rate": 6.41193639990525e-05, "loss": 0.0105, "step": 131530 }, { "epoch": 83.83683875079669, "grad_norm": 0.36112990975379944, "learning_rate": 6.411407666527767e-05, "loss": 0.0202, "step": 131540 }, { "epoch": 83.8432122370937, "grad_norm": 0.3976000249385834, "learning_rate": 6.410878916000348e-05, "loss": 0.0129, "step": 131550 }, { "epoch": 83.84958572339069, "grad_norm": 0.3795400857925415, "learning_rate": 6.410350148329419e-05, "loss": 0.0116, "step": 131560 }, { "epoch": 83.8559592096877, "grad_norm": 0.31290340423583984, "learning_rate": 6.409821363521403e-05, "loss": 0.012, "step": 131570 }, { "epoch": 83.8623326959847, "grad_norm": 0.4476834535598755, "learning_rate": 6.409292561582727e-05, "loss": 0.015, "step": 131580 }, { "epoch": 83.86870618228171, "grad_norm": 0.4359557330608368, "learning_rate": 6.408763742519815e-05, "loss": 0.0228, "step": 131590 }, { "epoch": 83.8750796685787, "grad_norm": 0.3220438063144684, "learning_rate": 6.408234906339094e-05, "loss": 0.0149, "step": 131600 }, { "epoch": 83.88145315487571, "grad_norm": 0.3882409632205963, "learning_rate": 6.40770605304699e-05, "loss": 0.0147, "step": 131610 }, { "epoch": 83.88782664117272, "grad_norm": 0.34638234972953796, "learning_rate": 6.407177182649927e-05, "loss": 0.0289, "step": 131620 }, { "epoch": 83.89420012746973, "grad_norm": 0.3443375527858734, "learning_rate": 6.406648295154333e-05, "loss": 0.0115, "step": 131630 }, { "epoch": 83.90057361376672, "grad_norm": 0.23756569623947144, "learning_rate": 6.406119390566637e-05, "loss": 0.0102, "step": 131640 }, { "epoch": 83.90694710006373, "grad_norm": 0.2672894299030304, "learning_rate": 6.405590468893258e-05, "loss": 0.0184, "step": 131650 }, { "epoch": 83.91332058636074, "grad_norm": 0.3958364427089691, "learning_rate": 6.40506153014063e-05, "loss": 0.0136, "step": 131660 }, { "epoch": 83.91969407265775, "grad_norm": 0.2411256730556488, "learning_rate": 6.404532574315179e-05, "loss": 0.0205, "step": 131670 }, { "epoch": 83.92606755895474, "grad_norm": 0.3888923227787018, "learning_rate": 6.404003601423329e-05, "loss": 0.01, "step": 131680 }, { "epoch": 83.93244104525175, "grad_norm": 0.27967751026153564, "learning_rate": 6.40347461147151e-05, "loss": 0.0156, "step": 131690 }, { "epoch": 83.93881453154876, "grad_norm": 0.34002095460891724, "learning_rate": 6.40294560446615e-05, "loss": 0.011, "step": 131700 }, { "epoch": 83.94518801784577, "grad_norm": 0.38990098237991333, "learning_rate": 6.402416580413676e-05, "loss": 0.014, "step": 131710 }, { "epoch": 83.95156150414276, "grad_norm": 0.2602407932281494, "learning_rate": 6.401887539320517e-05, "loss": 0.0115, "step": 131720 }, { "epoch": 83.95793499043977, "grad_norm": 0.5918154716491699, "learning_rate": 6.4013584811931e-05, "loss": 0.0139, "step": 131730 }, { "epoch": 83.96430847673678, "grad_norm": 0.260928213596344, "learning_rate": 6.400829406037856e-05, "loss": 0.0112, "step": 131740 }, { "epoch": 83.97068196303378, "grad_norm": 0.5408462285995483, "learning_rate": 6.40030031386121e-05, "loss": 0.0135, "step": 131750 }, { "epoch": 83.97705544933078, "grad_norm": 0.43934616446495056, "learning_rate": 6.399771204669594e-05, "loss": 0.0135, "step": 131760 }, { "epoch": 83.98342893562778, "grad_norm": 0.27909010648727417, "learning_rate": 6.399242078469437e-05, "loss": 0.0173, "step": 131770 }, { "epoch": 83.9898024219248, "grad_norm": 0.2956847548484802, "learning_rate": 6.398712935267166e-05, "loss": 0.0152, "step": 131780 }, { "epoch": 83.9961759082218, "grad_norm": 0.555381178855896, "learning_rate": 6.398183775069214e-05, "loss": 0.0122, "step": 131790 }, { "epoch": 84.00254939451881, "grad_norm": 0.1625937521457672, "learning_rate": 6.397654597882008e-05, "loss": 0.0151, "step": 131800 }, { "epoch": 84.0089228808158, "grad_norm": 0.17557282745838165, "learning_rate": 6.397125403711979e-05, "loss": 0.0173, "step": 131810 }, { "epoch": 84.01529636711281, "grad_norm": 0.5437086820602417, "learning_rate": 6.396596192565557e-05, "loss": 0.026, "step": 131820 }, { "epoch": 84.02166985340982, "grad_norm": 0.308824360370636, "learning_rate": 6.396066964449173e-05, "loss": 0.0181, "step": 131830 }, { "epoch": 84.02804333970683, "grad_norm": 0.41311559081077576, "learning_rate": 6.395537719369257e-05, "loss": 0.0142, "step": 131840 }, { "epoch": 84.03441682600382, "grad_norm": 0.42375484108924866, "learning_rate": 6.395008457332241e-05, "loss": 0.0179, "step": 131850 }, { "epoch": 84.04079031230083, "grad_norm": 0.5022709965705872, "learning_rate": 6.394479178344554e-05, "loss": 0.0189, "step": 131860 }, { "epoch": 84.04716379859784, "grad_norm": 0.18057653307914734, "learning_rate": 6.393949882412629e-05, "loss": 0.01, "step": 131870 }, { "epoch": 84.05353728489484, "grad_norm": 0.3134704530239105, "learning_rate": 6.393420569542897e-05, "loss": 0.0204, "step": 131880 }, { "epoch": 84.05991077119184, "grad_norm": 0.31656214594841003, "learning_rate": 6.39289123974179e-05, "loss": 0.019, "step": 131890 }, { "epoch": 84.06628425748885, "grad_norm": 0.20310260355472565, "learning_rate": 6.392361893015737e-05, "loss": 0.0133, "step": 131900 }, { "epoch": 84.07265774378585, "grad_norm": 0.6723842620849609, "learning_rate": 6.391832529371176e-05, "loss": 0.013, "step": 131910 }, { "epoch": 84.07903123008286, "grad_norm": 0.3332955539226532, "learning_rate": 6.391303148814534e-05, "loss": 0.0154, "step": 131920 }, { "epoch": 84.08540471637986, "grad_norm": 0.4290412664413452, "learning_rate": 6.390773751352243e-05, "loss": 0.0175, "step": 131930 }, { "epoch": 84.09177820267686, "grad_norm": 0.435187429189682, "learning_rate": 6.390244336990741e-05, "loss": 0.0204, "step": 131940 }, { "epoch": 84.09815168897387, "grad_norm": 0.5755150318145752, "learning_rate": 6.389714905736456e-05, "loss": 0.0243, "step": 131950 }, { "epoch": 84.10452517527088, "grad_norm": 0.36383485794067383, "learning_rate": 6.389185457595823e-05, "loss": 0.0139, "step": 131960 }, { "epoch": 84.11089866156787, "grad_norm": 0.2933783531188965, "learning_rate": 6.388655992575275e-05, "loss": 0.01, "step": 131970 }, { "epoch": 84.11727214786488, "grad_norm": 0.2072402834892273, "learning_rate": 6.388126510681247e-05, "loss": 0.0124, "step": 131980 }, { "epoch": 84.12364563416189, "grad_norm": 0.43248656392097473, "learning_rate": 6.38759701192017e-05, "loss": 0.0126, "step": 131990 }, { "epoch": 84.1300191204589, "grad_norm": 0.37408560514450073, "learning_rate": 6.387067496298479e-05, "loss": 0.0119, "step": 132000 }, { "epoch": 84.13639260675589, "grad_norm": 0.4798544645309448, "learning_rate": 6.386537963822607e-05, "loss": 0.0111, "step": 132010 }, { "epoch": 84.1427660930529, "grad_norm": 0.37076908349990845, "learning_rate": 6.386008414498991e-05, "loss": 0.0179, "step": 132020 }, { "epoch": 84.1491395793499, "grad_norm": 0.49032679200172424, "learning_rate": 6.385478848334066e-05, "loss": 0.0145, "step": 132030 }, { "epoch": 84.15551306564691, "grad_norm": 0.384633332490921, "learning_rate": 6.384949265334263e-05, "loss": 0.0121, "step": 132040 }, { "epoch": 84.16188655194391, "grad_norm": 0.26955392956733704, "learning_rate": 6.384419665506019e-05, "loss": 0.0223, "step": 132050 }, { "epoch": 84.16826003824092, "grad_norm": 0.41410180926322937, "learning_rate": 6.38389004885577e-05, "loss": 0.0107, "step": 132060 }, { "epoch": 84.17463352453792, "grad_norm": 0.3894641101360321, "learning_rate": 6.383360415389949e-05, "loss": 0.0169, "step": 132070 }, { "epoch": 84.18100701083493, "grad_norm": 0.33781740069389343, "learning_rate": 6.382830765114992e-05, "loss": 0.0116, "step": 132080 }, { "epoch": 84.18738049713193, "grad_norm": 0.2935221791267395, "learning_rate": 6.382301098037338e-05, "loss": 0.0116, "step": 132090 }, { "epoch": 84.19375398342893, "grad_norm": 0.4740540087223053, "learning_rate": 6.381771414163418e-05, "loss": 0.0152, "step": 132100 }, { "epoch": 84.20012746972594, "grad_norm": 0.3439091145992279, "learning_rate": 6.381241713499673e-05, "loss": 0.0127, "step": 132110 }, { "epoch": 84.20650095602295, "grad_norm": 0.435615599155426, "learning_rate": 6.380711996052535e-05, "loss": 0.0153, "step": 132120 }, { "epoch": 84.21287444231994, "grad_norm": 0.4336388409137726, "learning_rate": 6.380182261828445e-05, "loss": 0.019, "step": 132130 }, { "epoch": 84.21924792861695, "grad_norm": 0.43241459131240845, "learning_rate": 6.379652510833836e-05, "loss": 0.01, "step": 132140 }, { "epoch": 84.22562141491396, "grad_norm": 0.40375828742980957, "learning_rate": 6.379122743075145e-05, "loss": 0.0175, "step": 132150 }, { "epoch": 84.23199490121097, "grad_norm": 0.33931031823158264, "learning_rate": 6.378592958558814e-05, "loss": 0.0216, "step": 132160 }, { "epoch": 84.23836838750796, "grad_norm": 0.6289101839065552, "learning_rate": 6.378063157291275e-05, "loss": 0.0151, "step": 132170 }, { "epoch": 84.24474187380497, "grad_norm": 0.293766051530838, "learning_rate": 6.377533339278967e-05, "loss": 0.0097, "step": 132180 }, { "epoch": 84.25111536010198, "grad_norm": 0.4110451638698578, "learning_rate": 6.377003504528328e-05, "loss": 0.013, "step": 132190 }, { "epoch": 84.25748884639899, "grad_norm": 0.23686547577381134, "learning_rate": 6.376473653045798e-05, "loss": 0.0151, "step": 132200 }, { "epoch": 84.26386233269598, "grad_norm": 0.3428138494491577, "learning_rate": 6.375943784837813e-05, "loss": 0.0144, "step": 132210 }, { "epoch": 84.27023581899299, "grad_norm": 0.44701525568962097, "learning_rate": 6.375413899910812e-05, "loss": 0.0251, "step": 132220 }, { "epoch": 84.27660930529, "grad_norm": 0.3759009540081024, "learning_rate": 6.374883998271233e-05, "loss": 0.0158, "step": 132230 }, { "epoch": 84.282982791587, "grad_norm": 0.34278950095176697, "learning_rate": 6.374354079925515e-05, "loss": 0.0137, "step": 132240 }, { "epoch": 84.289356277884, "grad_norm": 0.6174064874649048, "learning_rate": 6.373824144880098e-05, "loss": 0.0125, "step": 132250 }, { "epoch": 84.295729764181, "grad_norm": 0.2763179838657379, "learning_rate": 6.373294193141419e-05, "loss": 0.0087, "step": 132260 }, { "epoch": 84.30210325047801, "grad_norm": 0.3914702832698822, "learning_rate": 6.372764224715922e-05, "loss": 0.018, "step": 132270 }, { "epoch": 84.30847673677502, "grad_norm": 0.7302253246307373, "learning_rate": 6.37223423961004e-05, "loss": 0.0143, "step": 132280 }, { "epoch": 84.31485022307201, "grad_norm": 0.27895912528038025, "learning_rate": 6.371704237830219e-05, "loss": 0.0091, "step": 132290 }, { "epoch": 84.32122370936902, "grad_norm": 0.6531727910041809, "learning_rate": 6.371174219382897e-05, "loss": 0.0179, "step": 132300 }, { "epoch": 84.32759719566603, "grad_norm": 0.586678147315979, "learning_rate": 6.37064418427451e-05, "loss": 0.0116, "step": 132310 }, { "epoch": 84.33397068196304, "grad_norm": 0.41012513637542725, "learning_rate": 6.370114132511505e-05, "loss": 0.0198, "step": 132320 }, { "epoch": 84.34034416826003, "grad_norm": 0.44647157192230225, "learning_rate": 6.369584064100318e-05, "loss": 0.0201, "step": 132330 }, { "epoch": 84.34671765455704, "grad_norm": 0.5076871514320374, "learning_rate": 6.369053979047395e-05, "loss": 0.0112, "step": 132340 }, { "epoch": 84.35309114085405, "grad_norm": 0.5035755038261414, "learning_rate": 6.368523877359169e-05, "loss": 0.0106, "step": 132350 }, { "epoch": 84.35946462715106, "grad_norm": 0.3012668490409851, "learning_rate": 6.36799375904209e-05, "loss": 0.0113, "step": 132360 }, { "epoch": 84.36583811344805, "grad_norm": 0.3172634541988373, "learning_rate": 6.367463624102592e-05, "loss": 0.0125, "step": 132370 }, { "epoch": 84.37221159974506, "grad_norm": 0.2987251877784729, "learning_rate": 6.366933472547121e-05, "loss": 0.0241, "step": 132380 }, { "epoch": 84.37858508604207, "grad_norm": 0.22357140481472015, "learning_rate": 6.366403304382118e-05, "loss": 0.0097, "step": 132390 }, { "epoch": 84.38495857233907, "grad_norm": 0.18303710222244263, "learning_rate": 6.365873119614024e-05, "loss": 0.0154, "step": 132400 }, { "epoch": 84.39133205863607, "grad_norm": 0.5183787941932678, "learning_rate": 6.365342918249283e-05, "loss": 0.0189, "step": 132410 }, { "epoch": 84.39770554493307, "grad_norm": 0.4597075879573822, "learning_rate": 6.364812700294334e-05, "loss": 0.016, "step": 132420 }, { "epoch": 84.40407903123008, "grad_norm": 0.317028671503067, "learning_rate": 6.364282465755626e-05, "loss": 0.0084, "step": 132430 }, { "epoch": 84.41045251752709, "grad_norm": 0.37524738907814026, "learning_rate": 6.363752214639595e-05, "loss": 0.015, "step": 132440 }, { "epoch": 84.4168260038241, "grad_norm": 0.4002605974674225, "learning_rate": 6.363221946952688e-05, "loss": 0.0156, "step": 132450 }, { "epoch": 84.42319949012109, "grad_norm": 0.4532420337200165, "learning_rate": 6.362691662701347e-05, "loss": 0.0172, "step": 132460 }, { "epoch": 84.4295729764181, "grad_norm": 0.41567230224609375, "learning_rate": 6.362161361892017e-05, "loss": 0.0164, "step": 132470 }, { "epoch": 84.43594646271511, "grad_norm": 0.6296383738517761, "learning_rate": 6.361631044531138e-05, "loss": 0.0205, "step": 132480 }, { "epoch": 84.44231994901212, "grad_norm": 0.6126928925514221, "learning_rate": 6.361100710625158e-05, "loss": 0.0112, "step": 132490 }, { "epoch": 84.44869343530911, "grad_norm": 0.3111700713634491, "learning_rate": 6.360570360180518e-05, "loss": 0.0164, "step": 132500 }, { "epoch": 84.45506692160612, "grad_norm": 0.1879359930753708, "learning_rate": 6.360039993203664e-05, "loss": 0.0088, "step": 132510 }, { "epoch": 84.46144040790313, "grad_norm": 0.1623217612504959, "learning_rate": 6.35950960970104e-05, "loss": 0.0127, "step": 132520 }, { "epoch": 84.46781389420013, "grad_norm": 0.38969889283180237, "learning_rate": 6.358979209679091e-05, "loss": 0.0165, "step": 132530 }, { "epoch": 84.47418738049713, "grad_norm": 0.44266587495803833, "learning_rate": 6.35844879314426e-05, "loss": 0.0168, "step": 132540 }, { "epoch": 84.48056086679414, "grad_norm": 0.8355053067207336, "learning_rate": 6.357918360102996e-05, "loss": 0.0153, "step": 132550 }, { "epoch": 84.48693435309114, "grad_norm": 0.2436361163854599, "learning_rate": 6.357387910561741e-05, "loss": 0.0141, "step": 132560 }, { "epoch": 84.49330783938815, "grad_norm": 0.3407157361507416, "learning_rate": 6.356857444526941e-05, "loss": 0.0165, "step": 132570 }, { "epoch": 84.49968132568515, "grad_norm": 0.21425671875476837, "learning_rate": 6.356326962005041e-05, "loss": 0.0146, "step": 132580 }, { "epoch": 84.50605481198215, "grad_norm": 0.4523334205150604, "learning_rate": 6.355796463002491e-05, "loss": 0.0157, "step": 132590 }, { "epoch": 84.51242829827916, "grad_norm": 0.24077092111110687, "learning_rate": 6.35526594752573e-05, "loss": 0.0219, "step": 132600 }, { "epoch": 84.51880178457617, "grad_norm": 0.3217729926109314, "learning_rate": 6.35473541558121e-05, "loss": 0.0245, "step": 132610 }, { "epoch": 84.52517527087316, "grad_norm": 0.2782188951969147, "learning_rate": 6.354204867175376e-05, "loss": 0.0179, "step": 132620 }, { "epoch": 84.53154875717017, "grad_norm": 0.5291478633880615, "learning_rate": 6.353674302314674e-05, "loss": 0.0127, "step": 132630 }, { "epoch": 84.53792224346718, "grad_norm": 0.35054436326026917, "learning_rate": 6.353143721005551e-05, "loss": 0.0115, "step": 132640 }, { "epoch": 84.54429572976419, "grad_norm": 0.30296698212623596, "learning_rate": 6.352613123254454e-05, "loss": 0.0147, "step": 132650 }, { "epoch": 84.55066921606118, "grad_norm": 0.21419648826122284, "learning_rate": 6.35208250906783e-05, "loss": 0.0097, "step": 132660 }, { "epoch": 84.55704270235819, "grad_norm": 0.5438587069511414, "learning_rate": 6.351551878452127e-05, "loss": 0.0161, "step": 132670 }, { "epoch": 84.5634161886552, "grad_norm": 0.3780128061771393, "learning_rate": 6.351021231413794e-05, "loss": 0.0219, "step": 132680 }, { "epoch": 84.5697896749522, "grad_norm": 0.3828471302986145, "learning_rate": 6.350490567959277e-05, "loss": 0.016, "step": 132690 }, { "epoch": 84.5761631612492, "grad_norm": 0.4249163568019867, "learning_rate": 6.349959888095024e-05, "loss": 0.0211, "step": 132700 }, { "epoch": 84.5825366475462, "grad_norm": 0.42051926255226135, "learning_rate": 6.349429191827486e-05, "loss": 0.0232, "step": 132710 }, { "epoch": 84.58891013384321, "grad_norm": 0.36169934272766113, "learning_rate": 6.348898479163105e-05, "loss": 0.0188, "step": 132720 }, { "epoch": 84.59528362014022, "grad_norm": 0.33523333072662354, "learning_rate": 6.348367750108337e-05, "loss": 0.0145, "step": 132730 }, { "epoch": 84.60165710643722, "grad_norm": 0.23613344132900238, "learning_rate": 6.347837004669625e-05, "loss": 0.0119, "step": 132740 }, { "epoch": 84.60803059273422, "grad_norm": 0.42854073643684387, "learning_rate": 6.347306242853425e-05, "loss": 0.0206, "step": 132750 }, { "epoch": 84.61440407903123, "grad_norm": 0.6108058094978333, "learning_rate": 6.346775464666179e-05, "loss": 0.0127, "step": 132760 }, { "epoch": 84.62077756532824, "grad_norm": 0.37020063400268555, "learning_rate": 6.346244670114341e-05, "loss": 0.011, "step": 132770 }, { "epoch": 84.62715105162523, "grad_norm": 0.3274647891521454, "learning_rate": 6.345713859204359e-05, "loss": 0.0172, "step": 132780 }, { "epoch": 84.63352453792224, "grad_norm": 0.30806925892829895, "learning_rate": 6.345183031942681e-05, "loss": 0.0232, "step": 132790 }, { "epoch": 84.63989802421925, "grad_norm": 0.3634806275367737, "learning_rate": 6.344652188335762e-05, "loss": 0.0381, "step": 132800 }, { "epoch": 84.64627151051626, "grad_norm": 0.6183315515518188, "learning_rate": 6.344121328390047e-05, "loss": 0.0162, "step": 132810 }, { "epoch": 84.65264499681325, "grad_norm": 0.3550736904144287, "learning_rate": 6.34359045211199e-05, "loss": 0.02, "step": 132820 }, { "epoch": 84.65901848311026, "grad_norm": 0.2917729616165161, "learning_rate": 6.34305955950804e-05, "loss": 0.0136, "step": 132830 }, { "epoch": 84.66539196940727, "grad_norm": 0.6440348625183105, "learning_rate": 6.34252865058465e-05, "loss": 0.0107, "step": 132840 }, { "epoch": 84.67176545570427, "grad_norm": 0.22162780165672302, "learning_rate": 6.341997725348268e-05, "loss": 0.0195, "step": 132850 }, { "epoch": 84.67813894200127, "grad_norm": 0.38527876138687134, "learning_rate": 6.341466783805345e-05, "loss": 0.0197, "step": 132860 }, { "epoch": 84.68451242829828, "grad_norm": 0.2322668582201004, "learning_rate": 6.340935825962336e-05, "loss": 0.0131, "step": 132870 }, { "epoch": 84.69088591459528, "grad_norm": 0.2872855067253113, "learning_rate": 6.34040485182569e-05, "loss": 0.015, "step": 132880 }, { "epoch": 84.69725940089229, "grad_norm": 0.30338212847709656, "learning_rate": 6.33987386140186e-05, "loss": 0.0199, "step": 132890 }, { "epoch": 84.70363288718929, "grad_norm": 0.2765369117259979, "learning_rate": 6.339342854697297e-05, "loss": 0.0147, "step": 132900 }, { "epoch": 84.7100063734863, "grad_norm": 0.23655110597610474, "learning_rate": 6.338811831718454e-05, "loss": 0.0179, "step": 132910 }, { "epoch": 84.7163798597833, "grad_norm": 0.34605106711387634, "learning_rate": 6.338280792471782e-05, "loss": 0.0109, "step": 132920 }, { "epoch": 84.72275334608031, "grad_norm": 0.24316366016864777, "learning_rate": 6.337749736963738e-05, "loss": 0.0197, "step": 132930 }, { "epoch": 84.7291268323773, "grad_norm": 0.2694697678089142, "learning_rate": 6.33721866520077e-05, "loss": 0.016, "step": 132940 }, { "epoch": 84.73550031867431, "grad_norm": 0.28915509581565857, "learning_rate": 6.336687577189331e-05, "loss": 0.0152, "step": 132950 }, { "epoch": 84.74187380497132, "grad_norm": 0.3840300440788269, "learning_rate": 6.336156472935877e-05, "loss": 0.0152, "step": 132960 }, { "epoch": 84.74824729126833, "grad_norm": 0.2756035029888153, "learning_rate": 6.33562535244686e-05, "loss": 0.0135, "step": 132970 }, { "epoch": 84.75462077756532, "grad_norm": 0.6210448145866394, "learning_rate": 6.335094215728735e-05, "loss": 0.0189, "step": 132980 }, { "epoch": 84.76099426386233, "grad_norm": 0.2618992328643799, "learning_rate": 6.334563062787953e-05, "loss": 0.0102, "step": 132990 }, { "epoch": 84.76736775015934, "grad_norm": 0.5321381092071533, "learning_rate": 6.334031893630972e-05, "loss": 0.0209, "step": 133000 }, { "epoch": 84.77374123645635, "grad_norm": 0.20945508778095245, "learning_rate": 6.333500708264243e-05, "loss": 0.0183, "step": 133010 }, { "epoch": 84.78011472275334, "grad_norm": 0.45514699816703796, "learning_rate": 6.332969506694222e-05, "loss": 0.0129, "step": 133020 }, { "epoch": 84.78648820905035, "grad_norm": 0.2396371066570282, "learning_rate": 6.332438288927362e-05, "loss": 0.0072, "step": 133030 }, { "epoch": 84.79286169534736, "grad_norm": 0.37780600786209106, "learning_rate": 6.331907054970118e-05, "loss": 0.0277, "step": 133040 }, { "epoch": 84.79923518164436, "grad_norm": 0.44274020195007324, "learning_rate": 6.331375804828947e-05, "loss": 0.023, "step": 133050 }, { "epoch": 84.80560866794136, "grad_norm": 0.32657960057258606, "learning_rate": 6.330844538510304e-05, "loss": 0.0092, "step": 133060 }, { "epoch": 84.81198215423836, "grad_norm": 0.3026450276374817, "learning_rate": 6.330313256020641e-05, "loss": 0.0134, "step": 133070 }, { "epoch": 84.81835564053537, "grad_norm": 0.30539625883102417, "learning_rate": 6.329781957366416e-05, "loss": 0.0115, "step": 133080 }, { "epoch": 84.82472912683238, "grad_norm": 0.5972595810890198, "learning_rate": 6.329250642554087e-05, "loss": 0.0095, "step": 133090 }, { "epoch": 84.83110261312939, "grad_norm": 0.2875621020793915, "learning_rate": 6.328719311590107e-05, "loss": 0.0153, "step": 133100 }, { "epoch": 84.83747609942638, "grad_norm": 0.5255768895149231, "learning_rate": 6.328187964480931e-05, "loss": 0.0307, "step": 133110 }, { "epoch": 84.84384958572339, "grad_norm": 0.5431597828865051, "learning_rate": 6.327656601233018e-05, "loss": 0.0173, "step": 133120 }, { "epoch": 84.8502230720204, "grad_norm": 0.31676965951919556, "learning_rate": 6.327125221852825e-05, "loss": 0.0235, "step": 133130 }, { "epoch": 84.8565965583174, "grad_norm": 0.5273497700691223, "learning_rate": 6.326593826346806e-05, "loss": 0.0149, "step": 133140 }, { "epoch": 84.8629700446144, "grad_norm": 0.4538824260234833, "learning_rate": 6.326062414721419e-05, "loss": 0.0147, "step": 133150 }, { "epoch": 84.86934353091141, "grad_norm": 0.43102169036865234, "learning_rate": 6.325530986983124e-05, "loss": 0.0132, "step": 133160 }, { "epoch": 84.87571701720842, "grad_norm": 0.4645758867263794, "learning_rate": 6.324999543138374e-05, "loss": 0.0129, "step": 133170 }, { "epoch": 84.88209050350542, "grad_norm": 0.5049845576286316, "learning_rate": 6.324468083193631e-05, "loss": 0.0206, "step": 133180 }, { "epoch": 84.88846398980242, "grad_norm": 0.20524127781391144, "learning_rate": 6.323936607155348e-05, "loss": 0.0093, "step": 133190 }, { "epoch": 84.89483747609943, "grad_norm": 0.3368780314922333, "learning_rate": 6.323405115029985e-05, "loss": 0.0271, "step": 133200 }, { "epoch": 84.90121096239643, "grad_norm": 0.21306592226028442, "learning_rate": 6.322873606824001e-05, "loss": 0.0108, "step": 133210 }, { "epoch": 84.90758444869344, "grad_norm": 0.21386587619781494, "learning_rate": 6.322342082543856e-05, "loss": 0.0116, "step": 133220 }, { "epoch": 84.91395793499044, "grad_norm": 0.4438561201095581, "learning_rate": 6.321810542196002e-05, "loss": 0.0246, "step": 133230 }, { "epoch": 84.92033142128744, "grad_norm": 0.3850349187850952, "learning_rate": 6.321278985786904e-05, "loss": 0.0161, "step": 133240 }, { "epoch": 84.92670490758445, "grad_norm": 0.37547019124031067, "learning_rate": 6.320747413323019e-05, "loss": 0.0167, "step": 133250 }, { "epoch": 84.93307839388146, "grad_norm": 0.5453292727470398, "learning_rate": 6.320215824810807e-05, "loss": 0.0161, "step": 133260 }, { "epoch": 84.93945188017845, "grad_norm": 0.5185712575912476, "learning_rate": 6.319684220256723e-05, "loss": 0.0111, "step": 133270 }, { "epoch": 84.94582536647546, "grad_norm": 0.354581356048584, "learning_rate": 6.319152599667232e-05, "loss": 0.0132, "step": 133280 }, { "epoch": 84.95219885277247, "grad_norm": 0.47317975759506226, "learning_rate": 6.318620963048791e-05, "loss": 0.0228, "step": 133290 }, { "epoch": 84.95857233906948, "grad_norm": 0.4889851212501526, "learning_rate": 6.318089310407862e-05, "loss": 0.0141, "step": 133300 }, { "epoch": 84.96494582536647, "grad_norm": 0.5003365278244019, "learning_rate": 6.317557641750901e-05, "loss": 0.0203, "step": 133310 }, { "epoch": 84.97131931166348, "grad_norm": 0.49199649691581726, "learning_rate": 6.317025957084372e-05, "loss": 0.0218, "step": 133320 }, { "epoch": 84.97769279796049, "grad_norm": 0.31489062309265137, "learning_rate": 6.316494256414733e-05, "loss": 0.0178, "step": 133330 }, { "epoch": 84.9840662842575, "grad_norm": 0.46738216280937195, "learning_rate": 6.315962539748447e-05, "loss": 0.0224, "step": 133340 }, { "epoch": 84.99043977055449, "grad_norm": 0.34236806631088257, "learning_rate": 6.315430807091973e-05, "loss": 0.0122, "step": 133350 }, { "epoch": 84.9968132568515, "grad_norm": 0.4099246859550476, "learning_rate": 6.314899058451773e-05, "loss": 0.0187, "step": 133360 }, { "epoch": 85.0031867431485, "grad_norm": 0.621935248374939, "learning_rate": 6.314367293834309e-05, "loss": 0.0224, "step": 133370 }, { "epoch": 85.00956022944551, "grad_norm": 0.3139297068119049, "learning_rate": 6.313835513246038e-05, "loss": 0.012, "step": 133380 }, { "epoch": 85.0159337157425, "grad_norm": 0.5313336849212646, "learning_rate": 6.313303716693428e-05, "loss": 0.0139, "step": 133390 }, { "epoch": 85.02230720203951, "grad_norm": 0.34900930523872375, "learning_rate": 6.312771904182937e-05, "loss": 0.0175, "step": 133400 }, { "epoch": 85.02868068833652, "grad_norm": 0.315639853477478, "learning_rate": 6.312240075721029e-05, "loss": 0.01, "step": 133410 }, { "epoch": 85.03505417463353, "grad_norm": 0.5061761140823364, "learning_rate": 6.311708231314164e-05, "loss": 0.012, "step": 133420 }, { "epoch": 85.04142766093052, "grad_norm": 0.21184015274047852, "learning_rate": 6.311176370968806e-05, "loss": 0.0153, "step": 133430 }, { "epoch": 85.04780114722753, "grad_norm": 0.49752309918403625, "learning_rate": 6.310644494691417e-05, "loss": 0.0128, "step": 133440 }, { "epoch": 85.05417463352454, "grad_norm": 0.5022903084754944, "learning_rate": 6.31011260248846e-05, "loss": 0.0146, "step": 133450 }, { "epoch": 85.06054811982155, "grad_norm": 0.46964195370674133, "learning_rate": 6.309580694366397e-05, "loss": 0.024, "step": 133460 }, { "epoch": 85.06692160611854, "grad_norm": 0.23028768599033356, "learning_rate": 6.309048770331692e-05, "loss": 0.01, "step": 133470 }, { "epoch": 85.07329509241555, "grad_norm": 0.3037722706794739, "learning_rate": 6.30851683039081e-05, "loss": 0.0106, "step": 133480 }, { "epoch": 85.07966857871256, "grad_norm": 0.16038651764392853, "learning_rate": 6.307984874550213e-05, "loss": 0.017, "step": 133490 }, { "epoch": 85.08604206500956, "grad_norm": 0.23609168827533722, "learning_rate": 6.307452902816364e-05, "loss": 0.0184, "step": 133500 }, { "epoch": 85.09241555130656, "grad_norm": 0.4985032081604004, "learning_rate": 6.306920915195728e-05, "loss": 0.0322, "step": 133510 }, { "epoch": 85.09878903760357, "grad_norm": 0.8257176280021667, "learning_rate": 6.30638891169477e-05, "loss": 0.0239, "step": 133520 }, { "epoch": 85.10516252390057, "grad_norm": 0.2936190366744995, "learning_rate": 6.305856892319952e-05, "loss": 0.012, "step": 133530 }, { "epoch": 85.11153601019758, "grad_norm": 0.34523454308509827, "learning_rate": 6.305324857077739e-05, "loss": 0.0166, "step": 133540 }, { "epoch": 85.11790949649458, "grad_norm": 0.28248846530914307, "learning_rate": 6.304792805974599e-05, "loss": 0.0238, "step": 133550 }, { "epoch": 85.12428298279158, "grad_norm": 0.4663640558719635, "learning_rate": 6.304260739016992e-05, "loss": 0.026, "step": 133560 }, { "epoch": 85.13065646908859, "grad_norm": 0.25383907556533813, "learning_rate": 6.303728656211387e-05, "loss": 0.0103, "step": 133570 }, { "epoch": 85.1370299553856, "grad_norm": 0.30031919479370117, "learning_rate": 6.303196557564249e-05, "loss": 0.0122, "step": 133580 }, { "epoch": 85.1434034416826, "grad_norm": 0.657975971698761, "learning_rate": 6.302664443082041e-05, "loss": 0.0196, "step": 133590 }, { "epoch": 85.1497769279796, "grad_norm": 0.39710748195648193, "learning_rate": 6.30213231277123e-05, "loss": 0.0183, "step": 133600 }, { "epoch": 85.15615041427661, "grad_norm": 0.3141266405582428, "learning_rate": 6.30160016663828e-05, "loss": 0.0105, "step": 133610 }, { "epoch": 85.16252390057362, "grad_norm": 0.306724488735199, "learning_rate": 6.301068004689662e-05, "loss": 0.0153, "step": 133620 }, { "epoch": 85.16889738687061, "grad_norm": 0.7298356294631958, "learning_rate": 6.300535826931838e-05, "loss": 0.0176, "step": 133630 }, { "epoch": 85.17527087316762, "grad_norm": 0.35468268394470215, "learning_rate": 6.300003633371277e-05, "loss": 0.014, "step": 133640 }, { "epoch": 85.18164435946463, "grad_norm": 0.30317091941833496, "learning_rate": 6.299471424014443e-05, "loss": 0.0138, "step": 133650 }, { "epoch": 85.18801784576164, "grad_norm": 0.47335687279701233, "learning_rate": 6.298939198867806e-05, "loss": 0.0144, "step": 133660 }, { "epoch": 85.19439133205863, "grad_norm": 0.616987407207489, "learning_rate": 6.298406957937832e-05, "loss": 0.0137, "step": 133670 }, { "epoch": 85.20076481835564, "grad_norm": 0.29102134704589844, "learning_rate": 6.297874701230984e-05, "loss": 0.0097, "step": 133680 }, { "epoch": 85.20713830465264, "grad_norm": 0.5683382153511047, "learning_rate": 6.297342428753736e-05, "loss": 0.0138, "step": 133690 }, { "epoch": 85.21351179094965, "grad_norm": 0.32119202613830566, "learning_rate": 6.29681014051255e-05, "loss": 0.0123, "step": 133700 }, { "epoch": 85.21988527724666, "grad_norm": 0.5160893201828003, "learning_rate": 6.2962778365139e-05, "loss": 0.0236, "step": 133710 }, { "epoch": 85.22625876354365, "grad_norm": 0.35252028703689575, "learning_rate": 6.295745516764247e-05, "loss": 0.0077, "step": 133720 }, { "epoch": 85.23263224984066, "grad_norm": 0.31651997566223145, "learning_rate": 6.295213181270064e-05, "loss": 0.0187, "step": 133730 }, { "epoch": 85.23900573613767, "grad_norm": 0.7795909643173218, "learning_rate": 6.294680830037819e-05, "loss": 0.0248, "step": 133740 }, { "epoch": 85.24537922243468, "grad_norm": 0.35686424374580383, "learning_rate": 6.294148463073976e-05, "loss": 0.0123, "step": 133750 }, { "epoch": 85.25175270873167, "grad_norm": 0.42998915910720825, "learning_rate": 6.293616080385012e-05, "loss": 0.0088, "step": 133760 }, { "epoch": 85.25812619502868, "grad_norm": 0.27756747603416443, "learning_rate": 6.293083681977387e-05, "loss": 0.0139, "step": 133770 }, { "epoch": 85.26449968132569, "grad_norm": 0.5360267162322998, "learning_rate": 6.292551267857578e-05, "loss": 0.0123, "step": 133780 }, { "epoch": 85.2708731676227, "grad_norm": 0.37585267424583435, "learning_rate": 6.292018838032047e-05, "loss": 0.0125, "step": 133790 }, { "epoch": 85.27724665391969, "grad_norm": 0.431610643863678, "learning_rate": 6.29148639250727e-05, "loss": 0.0086, "step": 133800 }, { "epoch": 85.2836201402167, "grad_norm": 0.386801153421402, "learning_rate": 6.290953931289713e-05, "loss": 0.0103, "step": 133810 }, { "epoch": 85.2899936265137, "grad_norm": 0.3985896408557892, "learning_rate": 6.290421454385849e-05, "loss": 0.0111, "step": 133820 }, { "epoch": 85.29636711281071, "grad_norm": 0.630875289440155, "learning_rate": 6.289888961802143e-05, "loss": 0.0164, "step": 133830 }, { "epoch": 85.30274059910771, "grad_norm": 0.28274887800216675, "learning_rate": 6.28935645354507e-05, "loss": 0.0116, "step": 133840 }, { "epoch": 85.30911408540472, "grad_norm": 0.308476984500885, "learning_rate": 6.288823929621098e-05, "loss": 0.0171, "step": 133850 }, { "epoch": 85.31548757170172, "grad_norm": 0.4567665755748749, "learning_rate": 6.288291390036698e-05, "loss": 0.0158, "step": 133860 }, { "epoch": 85.32186105799873, "grad_norm": 0.3329651653766632, "learning_rate": 6.287758834798341e-05, "loss": 0.0187, "step": 133870 }, { "epoch": 85.32823454429573, "grad_norm": 0.40971994400024414, "learning_rate": 6.2872262639125e-05, "loss": 0.0223, "step": 133880 }, { "epoch": 85.33460803059273, "grad_norm": 0.25484341382980347, "learning_rate": 6.286693677385643e-05, "loss": 0.0118, "step": 133890 }, { "epoch": 85.34098151688974, "grad_norm": 0.48147693276405334, "learning_rate": 6.286161075224243e-05, "loss": 0.0148, "step": 133900 }, { "epoch": 85.34735500318675, "grad_norm": 0.336171418428421, "learning_rate": 6.285628457434772e-05, "loss": 0.0124, "step": 133910 }, { "epoch": 85.35372848948374, "grad_norm": 0.5017493963241577, "learning_rate": 6.285095824023703e-05, "loss": 0.0124, "step": 133920 }, { "epoch": 85.36010197578075, "grad_norm": 0.2354864925146103, "learning_rate": 6.284563174997505e-05, "loss": 0.0164, "step": 133930 }, { "epoch": 85.36647546207776, "grad_norm": 0.3976515829563141, "learning_rate": 6.28403051036265e-05, "loss": 0.0108, "step": 133940 }, { "epoch": 85.37284894837477, "grad_norm": 0.1592789739370346, "learning_rate": 6.283497830125614e-05, "loss": 0.0155, "step": 133950 }, { "epoch": 85.37922243467176, "grad_norm": 0.3049215078353882, "learning_rate": 6.282965134292868e-05, "loss": 0.0162, "step": 133960 }, { "epoch": 85.38559592096877, "grad_norm": 0.3121669292449951, "learning_rate": 6.282432422870883e-05, "loss": 0.0162, "step": 133970 }, { "epoch": 85.39196940726578, "grad_norm": 0.43142643570899963, "learning_rate": 6.281899695866133e-05, "loss": 0.0169, "step": 133980 }, { "epoch": 85.39834289356278, "grad_norm": 0.371711403131485, "learning_rate": 6.281366953285094e-05, "loss": 0.0182, "step": 133990 }, { "epoch": 85.40471637985978, "grad_norm": 0.4096178710460663, "learning_rate": 6.280834195134234e-05, "loss": 0.0145, "step": 134000 }, { "epoch": 85.41108986615679, "grad_norm": 0.8119091987609863, "learning_rate": 6.28030142142003e-05, "loss": 0.0275, "step": 134010 }, { "epoch": 85.4174633524538, "grad_norm": 0.3857542872428894, "learning_rate": 6.279768632148955e-05, "loss": 0.0211, "step": 134020 }, { "epoch": 85.4238368387508, "grad_norm": 0.354531854391098, "learning_rate": 6.279235827327484e-05, "loss": 0.0131, "step": 134030 }, { "epoch": 85.4302103250478, "grad_norm": 0.2181270718574524, "learning_rate": 6.278703006962087e-05, "loss": 0.0179, "step": 134040 }, { "epoch": 85.4365838113448, "grad_norm": 0.423397958278656, "learning_rate": 6.278170171059243e-05, "loss": 0.024, "step": 134050 }, { "epoch": 85.44295729764181, "grad_norm": 0.46015915274620056, "learning_rate": 6.277637319625425e-05, "loss": 0.0126, "step": 134060 }, { "epoch": 85.44933078393882, "grad_norm": 0.39262688159942627, "learning_rate": 6.277104452667106e-05, "loss": 0.0217, "step": 134070 }, { "epoch": 85.45570427023581, "grad_norm": 0.1965094357728958, "learning_rate": 6.276571570190764e-05, "loss": 0.0105, "step": 134080 }, { "epoch": 85.46207775653282, "grad_norm": 0.35573405027389526, "learning_rate": 6.27603867220287e-05, "loss": 0.0139, "step": 134090 }, { "epoch": 85.46845124282983, "grad_norm": 0.2517143785953522, "learning_rate": 6.275505758709903e-05, "loss": 0.0117, "step": 134100 }, { "epoch": 85.47482472912684, "grad_norm": 0.3976518213748932, "learning_rate": 6.274972829718335e-05, "loss": 0.0134, "step": 134110 }, { "epoch": 85.48119821542383, "grad_norm": 0.3639957010746002, "learning_rate": 6.274439885234645e-05, "loss": 0.0192, "step": 134120 }, { "epoch": 85.48757170172084, "grad_norm": 0.5718289613723755, "learning_rate": 6.273906925265306e-05, "loss": 0.0177, "step": 134130 }, { "epoch": 85.49394518801785, "grad_norm": 0.2690408229827881, "learning_rate": 6.273373949816797e-05, "loss": 0.013, "step": 134140 }, { "epoch": 85.50031867431485, "grad_norm": 0.5210650563240051, "learning_rate": 6.27284095889559e-05, "loss": 0.0142, "step": 134150 }, { "epoch": 85.50669216061185, "grad_norm": 0.28243574500083923, "learning_rate": 6.272307952508164e-05, "loss": 0.0126, "step": 134160 }, { "epoch": 85.51306564690886, "grad_norm": 0.2538597881793976, "learning_rate": 6.271774930660997e-05, "loss": 0.0097, "step": 134170 }, { "epoch": 85.51943913320586, "grad_norm": 0.4388836622238159, "learning_rate": 6.271241893360561e-05, "loss": 0.0128, "step": 134180 }, { "epoch": 85.52581261950287, "grad_norm": 0.37384334206581116, "learning_rate": 6.270708840613337e-05, "loss": 0.0337, "step": 134190 }, { "epoch": 85.53218610579987, "grad_norm": 0.7318226099014282, "learning_rate": 6.270175772425802e-05, "loss": 0.0147, "step": 134200 }, { "epoch": 85.53855959209687, "grad_norm": 0.29239773750305176, "learning_rate": 6.26964268880443e-05, "loss": 0.0076, "step": 134210 }, { "epoch": 85.54493307839388, "grad_norm": 0.6223142147064209, "learning_rate": 6.269109589755701e-05, "loss": 0.0283, "step": 134220 }, { "epoch": 85.55130656469089, "grad_norm": 0.19736704230308533, "learning_rate": 6.268576475286095e-05, "loss": 0.0102, "step": 134230 }, { "epoch": 85.55768005098788, "grad_norm": 0.9032602310180664, "learning_rate": 6.268043345402085e-05, "loss": 0.0228, "step": 134240 }, { "epoch": 85.56405353728489, "grad_norm": 0.2433578222990036, "learning_rate": 6.26751020011015e-05, "loss": 0.0119, "step": 134250 }, { "epoch": 85.5704270235819, "grad_norm": 0.5338777899742126, "learning_rate": 6.266977039416771e-05, "loss": 0.0116, "step": 134260 }, { "epoch": 85.57680050987891, "grad_norm": 0.28763332962989807, "learning_rate": 6.266443863328424e-05, "loss": 0.0165, "step": 134270 }, { "epoch": 85.5831739961759, "grad_norm": 0.32613518834114075, "learning_rate": 6.26591067185159e-05, "loss": 0.0113, "step": 134280 }, { "epoch": 85.58954748247291, "grad_norm": 0.1906759887933731, "learning_rate": 6.265377464992743e-05, "loss": 0.0139, "step": 134290 }, { "epoch": 85.59592096876992, "grad_norm": 0.5102590322494507, "learning_rate": 6.264844242758368e-05, "loss": 0.0181, "step": 134300 }, { "epoch": 85.60229445506693, "grad_norm": 0.2676417827606201, "learning_rate": 6.26431100515494e-05, "loss": 0.0111, "step": 134310 }, { "epoch": 85.60866794136393, "grad_norm": 0.2396395355463028, "learning_rate": 6.263777752188939e-05, "loss": 0.024, "step": 134320 }, { "epoch": 85.61504142766093, "grad_norm": 0.5930048227310181, "learning_rate": 6.263244483866846e-05, "loss": 0.0165, "step": 134330 }, { "epoch": 85.62141491395793, "grad_norm": 0.19959649443626404, "learning_rate": 6.26271120019514e-05, "loss": 0.0229, "step": 134340 }, { "epoch": 85.62778840025494, "grad_norm": 0.24350512027740479, "learning_rate": 6.2621779011803e-05, "loss": 0.0148, "step": 134350 }, { "epoch": 85.63416188655195, "grad_norm": 0.23164118826389313, "learning_rate": 6.261644586828806e-05, "loss": 0.0175, "step": 134360 }, { "epoch": 85.64053537284894, "grad_norm": 0.7929355502128601, "learning_rate": 6.26111125714714e-05, "loss": 0.0161, "step": 134370 }, { "epoch": 85.64690885914595, "grad_norm": 0.26290762424468994, "learning_rate": 6.260577912141783e-05, "loss": 0.0182, "step": 134380 }, { "epoch": 85.65328234544296, "grad_norm": 0.45211490988731384, "learning_rate": 6.260044551819213e-05, "loss": 0.0161, "step": 134390 }, { "epoch": 85.65965583173997, "grad_norm": 0.602666974067688, "learning_rate": 6.259511176185913e-05, "loss": 0.0179, "step": 134400 }, { "epoch": 85.66602931803696, "grad_norm": 0.5500831604003906, "learning_rate": 6.258977785248362e-05, "loss": 0.0368, "step": 134410 }, { "epoch": 85.67240280433397, "grad_norm": 0.543546199798584, "learning_rate": 6.258444379013041e-05, "loss": 0.0131, "step": 134420 }, { "epoch": 85.67877629063098, "grad_norm": 0.45119503140449524, "learning_rate": 6.257910957486433e-05, "loss": 0.0156, "step": 134430 }, { "epoch": 85.68514977692799, "grad_norm": 0.3457984924316406, "learning_rate": 6.25737752067502e-05, "loss": 0.0124, "step": 134440 }, { "epoch": 85.69152326322498, "grad_norm": 0.4464523196220398, "learning_rate": 6.256844068585284e-05, "loss": 0.0127, "step": 134450 }, { "epoch": 85.69789674952199, "grad_norm": 0.4291567802429199, "learning_rate": 6.256310601223705e-05, "loss": 0.0188, "step": 134460 }, { "epoch": 85.704270235819, "grad_norm": 0.3365817368030548, "learning_rate": 6.255777118596766e-05, "loss": 0.0095, "step": 134470 }, { "epoch": 85.710643722116, "grad_norm": 0.2675381898880005, "learning_rate": 6.255243620710949e-05, "loss": 0.0084, "step": 134480 }, { "epoch": 85.717017208413, "grad_norm": 0.625413179397583, "learning_rate": 6.254710107572737e-05, "loss": 0.0202, "step": 134490 }, { "epoch": 85.72339069471, "grad_norm": 0.28728628158569336, "learning_rate": 6.254176579188612e-05, "loss": 0.0133, "step": 134500 }, { "epoch": 85.72976418100701, "grad_norm": 0.5359876155853271, "learning_rate": 6.253643035565058e-05, "loss": 0.0328, "step": 134510 }, { "epoch": 85.73613766730402, "grad_norm": 0.2798033654689789, "learning_rate": 6.253109476708558e-05, "loss": 0.0151, "step": 134520 }, { "epoch": 85.74251115360101, "grad_norm": 0.35901927947998047, "learning_rate": 6.252575902625595e-05, "loss": 0.0118, "step": 134530 }, { "epoch": 85.74888463989802, "grad_norm": 0.5089362859725952, "learning_rate": 6.25204231332265e-05, "loss": 0.0164, "step": 134540 }, { "epoch": 85.75525812619503, "grad_norm": 0.28204217553138733, "learning_rate": 6.251508708806211e-05, "loss": 0.015, "step": 134550 }, { "epoch": 85.76163161249204, "grad_norm": 0.32606256008148193, "learning_rate": 6.250975089082759e-05, "loss": 0.0172, "step": 134560 }, { "epoch": 85.76800509878903, "grad_norm": 0.40596604347229004, "learning_rate": 6.250441454158778e-05, "loss": 0.0148, "step": 134570 }, { "epoch": 85.77437858508604, "grad_norm": 0.22786960005760193, "learning_rate": 6.249907804040754e-05, "loss": 0.0094, "step": 134580 }, { "epoch": 85.78075207138305, "grad_norm": 0.2618916928768158, "learning_rate": 6.249374138735168e-05, "loss": 0.0276, "step": 134590 }, { "epoch": 85.78712555768006, "grad_norm": 0.4987781345844269, "learning_rate": 6.248840458248508e-05, "loss": 0.0134, "step": 134600 }, { "epoch": 85.79349904397705, "grad_norm": 0.4804847538471222, "learning_rate": 6.248306762587257e-05, "loss": 0.0213, "step": 134610 }, { "epoch": 85.79987253027406, "grad_norm": 0.21705777943134308, "learning_rate": 6.2477730517579e-05, "loss": 0.015, "step": 134620 }, { "epoch": 85.80624601657107, "grad_norm": 0.495553582906723, "learning_rate": 6.247239325766922e-05, "loss": 0.0298, "step": 134630 }, { "epoch": 85.81261950286807, "grad_norm": 0.23771072924137115, "learning_rate": 6.24670558462081e-05, "loss": 0.02, "step": 134640 }, { "epoch": 85.81899298916507, "grad_norm": 0.37462735176086426, "learning_rate": 6.246171828326048e-05, "loss": 0.0136, "step": 134650 }, { "epoch": 85.82536647546208, "grad_norm": 0.22058837115764618, "learning_rate": 6.24563805688912e-05, "loss": 0.0206, "step": 134660 }, { "epoch": 85.83173996175908, "grad_norm": 0.7076476216316223, "learning_rate": 6.245104270316516e-05, "loss": 0.0193, "step": 134670 }, { "epoch": 85.83811344805609, "grad_norm": 0.45809313654899597, "learning_rate": 6.244570468614717e-05, "loss": 0.0207, "step": 134680 }, { "epoch": 85.84448693435309, "grad_norm": 0.3392370939254761, "learning_rate": 6.244036651790212e-05, "loss": 0.0164, "step": 134690 }, { "epoch": 85.8508604206501, "grad_norm": 0.6087072491645813, "learning_rate": 6.243502819849488e-05, "loss": 0.0191, "step": 134700 }, { "epoch": 85.8572339069471, "grad_norm": 0.3491577208042145, "learning_rate": 6.242968972799032e-05, "loss": 0.0104, "step": 134710 }, { "epoch": 85.86360739324411, "grad_norm": 0.5296474695205688, "learning_rate": 6.242435110645329e-05, "loss": 0.0098, "step": 134720 }, { "epoch": 85.8699808795411, "grad_norm": 0.23673203587532043, "learning_rate": 6.241901233394864e-05, "loss": 0.0123, "step": 134730 }, { "epoch": 85.87635436583811, "grad_norm": 0.5240015983581543, "learning_rate": 6.241367341054128e-05, "loss": 0.0124, "step": 134740 }, { "epoch": 85.88272785213512, "grad_norm": 0.5271756649017334, "learning_rate": 6.240833433629606e-05, "loss": 0.0082, "step": 134750 }, { "epoch": 85.88910133843213, "grad_norm": 0.6531194448471069, "learning_rate": 6.240299511127787e-05, "loss": 0.0158, "step": 134760 }, { "epoch": 85.89547482472912, "grad_norm": 0.4254186749458313, "learning_rate": 6.239765573555157e-05, "loss": 0.012, "step": 134770 }, { "epoch": 85.90184831102613, "grad_norm": 0.20464026927947998, "learning_rate": 6.239231620918207e-05, "loss": 0.0114, "step": 134780 }, { "epoch": 85.90822179732314, "grad_norm": 0.3883301913738251, "learning_rate": 6.238697653223422e-05, "loss": 0.02, "step": 134790 }, { "epoch": 85.91459528362014, "grad_norm": 0.6050004363059998, "learning_rate": 6.23816367047729e-05, "loss": 0.0174, "step": 134800 }, { "epoch": 85.92096876991714, "grad_norm": 0.2954604923725128, "learning_rate": 6.2376296726863e-05, "loss": 0.0134, "step": 134810 }, { "epoch": 85.92734225621415, "grad_norm": 0.3161962926387787, "learning_rate": 6.237095659856941e-05, "loss": 0.0138, "step": 134820 }, { "epoch": 85.93371574251115, "grad_norm": 0.30038729310035706, "learning_rate": 6.236561631995703e-05, "loss": 0.0187, "step": 134830 }, { "epoch": 85.94008922880816, "grad_norm": 0.8933187127113342, "learning_rate": 6.236027589109073e-05, "loss": 0.0152, "step": 134840 }, { "epoch": 85.94646271510516, "grad_norm": 0.3523712158203125, "learning_rate": 6.235493531203542e-05, "loss": 0.0181, "step": 134850 }, { "epoch": 85.95283620140216, "grad_norm": 0.40961790084838867, "learning_rate": 6.234959458285595e-05, "loss": 0.0099, "step": 134860 }, { "epoch": 85.95920968769917, "grad_norm": 0.32076549530029297, "learning_rate": 6.234425370361728e-05, "loss": 0.0133, "step": 134870 }, { "epoch": 85.96558317399618, "grad_norm": 0.2806931138038635, "learning_rate": 6.233891267438425e-05, "loss": 0.018, "step": 134880 }, { "epoch": 85.97195666029317, "grad_norm": 0.4500206410884857, "learning_rate": 6.23335714952218e-05, "loss": 0.0125, "step": 134890 }, { "epoch": 85.97833014659018, "grad_norm": 0.4107638895511627, "learning_rate": 6.23282301661948e-05, "loss": 0.0151, "step": 134900 }, { "epoch": 85.98470363288719, "grad_norm": 1.0616971254348755, "learning_rate": 6.232288868736815e-05, "loss": 0.0248, "step": 134910 }, { "epoch": 85.9910771191842, "grad_norm": 0.7169125080108643, "learning_rate": 6.231754705880679e-05, "loss": 0.0164, "step": 134920 }, { "epoch": 85.99745060548119, "grad_norm": 0.2793859839439392, "learning_rate": 6.231220528057559e-05, "loss": 0.0089, "step": 134930 }, { "epoch": 86.0038240917782, "grad_norm": 0.6584393382072449, "learning_rate": 6.230686335273947e-05, "loss": 0.0136, "step": 134940 }, { "epoch": 86.0101975780752, "grad_norm": 0.4580683708190918, "learning_rate": 6.230152127536337e-05, "loss": 0.0088, "step": 134950 }, { "epoch": 86.01657106437222, "grad_norm": 0.6158672571182251, "learning_rate": 6.229617904851214e-05, "loss": 0.0159, "step": 134960 }, { "epoch": 86.02294455066922, "grad_norm": 0.5652865171432495, "learning_rate": 6.229083667225073e-05, "loss": 0.0231, "step": 134970 }, { "epoch": 86.02931803696622, "grad_norm": 0.5845011472702026, "learning_rate": 6.228549414664406e-05, "loss": 0.0119, "step": 134980 }, { "epoch": 86.03569152326322, "grad_norm": 0.3714328408241272, "learning_rate": 6.228015147175703e-05, "loss": 0.0199, "step": 134990 }, { "epoch": 86.04206500956023, "grad_norm": 0.5463960766792297, "learning_rate": 6.227480864765456e-05, "loss": 0.0169, "step": 135000 }, { "epoch": 86.04843849585724, "grad_norm": 0.21562433242797852, "learning_rate": 6.226946567440158e-05, "loss": 0.0133, "step": 135010 }, { "epoch": 86.05481198215423, "grad_norm": 0.3090166449546814, "learning_rate": 6.2264122552063e-05, "loss": 0.014, "step": 135020 }, { "epoch": 86.06118546845124, "grad_norm": 0.2305338829755783, "learning_rate": 6.225877928070377e-05, "loss": 0.0143, "step": 135030 }, { "epoch": 86.06755895474825, "grad_norm": 0.433553546667099, "learning_rate": 6.22534358603888e-05, "loss": 0.0114, "step": 135040 }, { "epoch": 86.07393244104526, "grad_norm": 0.3043432831764221, "learning_rate": 6.224809229118299e-05, "loss": 0.0145, "step": 135050 }, { "epoch": 86.08030592734225, "grad_norm": 0.2510030269622803, "learning_rate": 6.22427485731513e-05, "loss": 0.019, "step": 135060 }, { "epoch": 86.08667941363926, "grad_norm": 0.19729898869991302, "learning_rate": 6.223740470635867e-05, "loss": 0.0138, "step": 135070 }, { "epoch": 86.09305289993627, "grad_norm": 0.3960331678390503, "learning_rate": 6.223206069087002e-05, "loss": 0.0149, "step": 135080 }, { "epoch": 86.09942638623328, "grad_norm": 0.31779107451438904, "learning_rate": 6.222671652675027e-05, "loss": 0.0169, "step": 135090 }, { "epoch": 86.10579987253027, "grad_norm": 0.21725769340991974, "learning_rate": 6.222137221406439e-05, "loss": 0.0162, "step": 135100 }, { "epoch": 86.11217335882728, "grad_norm": 0.48256155848503113, "learning_rate": 6.221602775287728e-05, "loss": 0.0195, "step": 135110 }, { "epoch": 86.11854684512429, "grad_norm": 0.17646318674087524, "learning_rate": 6.22106831432539e-05, "loss": 0.0095, "step": 135120 }, { "epoch": 86.1249203314213, "grad_norm": 0.39146724343299866, "learning_rate": 6.22053383852592e-05, "loss": 0.0206, "step": 135130 }, { "epoch": 86.13129381771829, "grad_norm": 0.49523472785949707, "learning_rate": 6.219999347895811e-05, "loss": 0.0152, "step": 135140 }, { "epoch": 86.1376673040153, "grad_norm": 0.1939106434583664, "learning_rate": 6.219464842441558e-05, "loss": 0.0087, "step": 135150 }, { "epoch": 86.1440407903123, "grad_norm": 0.5579684376716614, "learning_rate": 6.218930322169655e-05, "loss": 0.013, "step": 135160 }, { "epoch": 86.15041427660931, "grad_norm": 0.42011529207229614, "learning_rate": 6.2183957870866e-05, "loss": 0.0133, "step": 135170 }, { "epoch": 86.1567877629063, "grad_norm": 0.33165866136550903, "learning_rate": 6.217861237198884e-05, "loss": 0.0105, "step": 135180 }, { "epoch": 86.16316124920331, "grad_norm": 0.16643035411834717, "learning_rate": 6.217326672513004e-05, "loss": 0.0125, "step": 135190 }, { "epoch": 86.16953473550032, "grad_norm": 0.18035632371902466, "learning_rate": 6.216792093035457e-05, "loss": 0.0115, "step": 135200 }, { "epoch": 86.17590822179733, "grad_norm": 0.6001988053321838, "learning_rate": 6.216257498772734e-05, "loss": 0.0313, "step": 135210 }, { "epoch": 86.18228170809432, "grad_norm": 0.22946247458457947, "learning_rate": 6.21572288973134e-05, "loss": 0.0222, "step": 135220 }, { "epoch": 86.18865519439133, "grad_norm": 0.5913975238800049, "learning_rate": 6.21518826591776e-05, "loss": 0.0292, "step": 135230 }, { "epoch": 86.19502868068834, "grad_norm": 0.3534841239452362, "learning_rate": 6.214653627338498e-05, "loss": 0.0102, "step": 135240 }, { "epoch": 86.20140216698535, "grad_norm": 0.20196394622325897, "learning_rate": 6.214118974000044e-05, "loss": 0.0146, "step": 135250 }, { "epoch": 86.20777565328234, "grad_norm": 0.26523587107658386, "learning_rate": 6.213584305908901e-05, "loss": 0.0125, "step": 135260 }, { "epoch": 86.21414913957935, "grad_norm": 0.4626877009868622, "learning_rate": 6.213049623071562e-05, "loss": 0.0129, "step": 135270 }, { "epoch": 86.22052262587636, "grad_norm": 0.5268779397010803, "learning_rate": 6.212514925494526e-05, "loss": 0.023, "step": 135280 }, { "epoch": 86.22689611217336, "grad_norm": 0.27981695532798767, "learning_rate": 6.211980213184287e-05, "loss": 0.0167, "step": 135290 }, { "epoch": 86.23326959847036, "grad_norm": 0.3353387415409088, "learning_rate": 6.211445486147346e-05, "loss": 0.0128, "step": 135300 }, { "epoch": 86.23964308476737, "grad_norm": 0.7728697061538696, "learning_rate": 6.210910744390197e-05, "loss": 0.0183, "step": 135310 }, { "epoch": 86.24601657106437, "grad_norm": 0.4499590992927551, "learning_rate": 6.210375987919341e-05, "loss": 0.0157, "step": 135320 }, { "epoch": 86.25239005736138, "grad_norm": 0.1418088972568512, "learning_rate": 6.209841216741271e-05, "loss": 0.0071, "step": 135330 }, { "epoch": 86.25876354365838, "grad_norm": 0.2138492912054062, "learning_rate": 6.20930643086249e-05, "loss": 0.0084, "step": 135340 }, { "epoch": 86.26513702995538, "grad_norm": 0.4573669731616974, "learning_rate": 6.208771630289493e-05, "loss": 0.0121, "step": 135350 }, { "epoch": 86.27151051625239, "grad_norm": 0.9010759592056274, "learning_rate": 6.208236815028781e-05, "loss": 0.0175, "step": 135360 }, { "epoch": 86.2778840025494, "grad_norm": 0.4801674485206604, "learning_rate": 6.207701985086849e-05, "loss": 0.0194, "step": 135370 }, { "epoch": 86.2842574888464, "grad_norm": 0.49767372012138367, "learning_rate": 6.2071671404702e-05, "loss": 0.0231, "step": 135380 }, { "epoch": 86.2906309751434, "grad_norm": 0.649364709854126, "learning_rate": 6.206632281185328e-05, "loss": 0.0193, "step": 135390 }, { "epoch": 86.29700446144041, "grad_norm": 0.34288161993026733, "learning_rate": 6.206097407238736e-05, "loss": 0.0168, "step": 135400 }, { "epoch": 86.30337794773742, "grad_norm": 0.25144845247268677, "learning_rate": 6.205562518636921e-05, "loss": 0.0138, "step": 135410 }, { "epoch": 86.30975143403441, "grad_norm": 0.43395283818244934, "learning_rate": 6.205027615386385e-05, "loss": 0.0172, "step": 135420 }, { "epoch": 86.31612492033142, "grad_norm": 0.30412763357162476, "learning_rate": 6.204492697493624e-05, "loss": 0.0126, "step": 135430 }, { "epoch": 86.32249840662843, "grad_norm": 0.2962929308414459, "learning_rate": 6.20395776496514e-05, "loss": 0.0123, "step": 135440 }, { "epoch": 86.32887189292543, "grad_norm": 0.2828255593776703, "learning_rate": 6.203422817807433e-05, "loss": 0.0148, "step": 135450 }, { "epoch": 86.33524537922243, "grad_norm": 0.665999174118042, "learning_rate": 6.202887856027001e-05, "loss": 0.0179, "step": 135460 }, { "epoch": 86.34161886551944, "grad_norm": 0.3620237410068512, "learning_rate": 6.202352879630347e-05, "loss": 0.0063, "step": 135470 }, { "epoch": 86.34799235181644, "grad_norm": 0.32914337515830994, "learning_rate": 6.20181788862397e-05, "loss": 0.014, "step": 135480 }, { "epoch": 86.35436583811345, "grad_norm": 0.3205731213092804, "learning_rate": 6.201282883014371e-05, "loss": 0.0165, "step": 135490 }, { "epoch": 86.36073932441045, "grad_norm": 0.341469943523407, "learning_rate": 6.20074786280805e-05, "loss": 0.0129, "step": 135500 }, { "epoch": 86.36711281070745, "grad_norm": 0.2432442158460617, "learning_rate": 6.200212828011511e-05, "loss": 0.0103, "step": 135510 }, { "epoch": 86.37348629700446, "grad_norm": 0.20887355506420135, "learning_rate": 6.199677778631251e-05, "loss": 0.0075, "step": 135520 }, { "epoch": 86.37985978330147, "grad_norm": 0.5486912727355957, "learning_rate": 6.199142714673774e-05, "loss": 0.0134, "step": 135530 }, { "epoch": 86.38623326959846, "grad_norm": 0.5357344150543213, "learning_rate": 6.19860763614558e-05, "loss": 0.0161, "step": 135540 }, { "epoch": 86.39260675589547, "grad_norm": 0.6181392073631287, "learning_rate": 6.198072543053173e-05, "loss": 0.0117, "step": 135550 }, { "epoch": 86.39898024219248, "grad_norm": 0.40764039754867554, "learning_rate": 6.197537435403052e-05, "loss": 0.0095, "step": 135560 }, { "epoch": 86.40535372848949, "grad_norm": 0.20449642837047577, "learning_rate": 6.197002313201721e-05, "loss": 0.0196, "step": 135570 }, { "epoch": 86.4117272147865, "grad_norm": 0.3925415575504303, "learning_rate": 6.196467176455683e-05, "loss": 0.0088, "step": 135580 }, { "epoch": 86.41810070108349, "grad_norm": 0.3782031536102295, "learning_rate": 6.195932025171436e-05, "loss": 0.0195, "step": 135590 }, { "epoch": 86.4244741873805, "grad_norm": 0.3787313401699066, "learning_rate": 6.195396859355488e-05, "loss": 0.011, "step": 135600 }, { "epoch": 86.4308476736775, "grad_norm": 0.2139919102191925, "learning_rate": 6.194861679014341e-05, "loss": 0.0141, "step": 135610 }, { "epoch": 86.43722115997451, "grad_norm": 0.26279762387275696, "learning_rate": 6.194326484154493e-05, "loss": 0.0182, "step": 135620 }, { "epoch": 86.4435946462715, "grad_norm": 0.451067179441452, "learning_rate": 6.193791274782453e-05, "loss": 0.0168, "step": 135630 }, { "epoch": 86.44996813256851, "grad_norm": 0.47460275888442993, "learning_rate": 6.19325605090472e-05, "loss": 0.0109, "step": 135640 }, { "epoch": 86.45634161886552, "grad_norm": 0.455019474029541, "learning_rate": 6.192720812527802e-05, "loss": 0.0108, "step": 135650 }, { "epoch": 86.46271510516253, "grad_norm": 0.4023546874523163, "learning_rate": 6.192185559658198e-05, "loss": 0.014, "step": 135660 }, { "epoch": 86.46908859145952, "grad_norm": 0.5042657256126404, "learning_rate": 6.191650292302414e-05, "loss": 0.027, "step": 135670 }, { "epoch": 86.47546207775653, "grad_norm": 0.2793891727924347, "learning_rate": 6.191115010466956e-05, "loss": 0.0098, "step": 135680 }, { "epoch": 86.48183556405354, "grad_norm": 0.24504855275154114, "learning_rate": 6.190579714158324e-05, "loss": 0.0158, "step": 135690 }, { "epoch": 86.48820905035055, "grad_norm": 0.5600149035453796, "learning_rate": 6.190044403383024e-05, "loss": 0.013, "step": 135700 }, { "epoch": 86.49458253664754, "grad_norm": 0.5385920405387878, "learning_rate": 6.189509078147561e-05, "loss": 0.0155, "step": 135710 }, { "epoch": 86.50095602294455, "grad_norm": 0.35997435450553894, "learning_rate": 6.188973738458441e-05, "loss": 0.016, "step": 135720 }, { "epoch": 86.50732950924156, "grad_norm": 0.6061364412307739, "learning_rate": 6.188438384322166e-05, "loss": 0.015, "step": 135730 }, { "epoch": 86.51370299553857, "grad_norm": 0.4451411962509155, "learning_rate": 6.187903015745243e-05, "loss": 0.0156, "step": 135740 }, { "epoch": 86.52007648183556, "grad_norm": 0.5176655650138855, "learning_rate": 6.187367632734177e-05, "loss": 0.0112, "step": 135750 }, { "epoch": 86.52644996813257, "grad_norm": 0.37311169505119324, "learning_rate": 6.186832235295474e-05, "loss": 0.0117, "step": 135760 }, { "epoch": 86.53282345442958, "grad_norm": 0.24009248614311218, "learning_rate": 6.186296823435639e-05, "loss": 0.0089, "step": 135770 }, { "epoch": 86.53919694072658, "grad_norm": 0.20149506628513336, "learning_rate": 6.185761397161175e-05, "loss": 0.0131, "step": 135780 }, { "epoch": 86.54557042702358, "grad_norm": 0.5700642466545105, "learning_rate": 6.185225956478593e-05, "loss": 0.0121, "step": 135790 }, { "epoch": 86.55194391332059, "grad_norm": 0.33434686064720154, "learning_rate": 6.184690501394394e-05, "loss": 0.0151, "step": 135800 }, { "epoch": 86.5583173996176, "grad_norm": 0.3418165445327759, "learning_rate": 6.184155031915089e-05, "loss": 0.0163, "step": 135810 }, { "epoch": 86.5646908859146, "grad_norm": 0.23490965366363525, "learning_rate": 6.183619548047183e-05, "loss": 0.0164, "step": 135820 }, { "epoch": 86.5710643722116, "grad_norm": 0.22476395964622498, "learning_rate": 6.183084049797181e-05, "loss": 0.0174, "step": 135830 }, { "epoch": 86.5774378585086, "grad_norm": 0.29840990900993347, "learning_rate": 6.182548537171591e-05, "loss": 0.0147, "step": 135840 }, { "epoch": 86.58381134480561, "grad_norm": 0.27412399649620056, "learning_rate": 6.182013010176918e-05, "loss": 0.0108, "step": 135850 }, { "epoch": 86.59018483110262, "grad_norm": 0.28438565135002136, "learning_rate": 6.181477468819673e-05, "loss": 0.016, "step": 135860 }, { "epoch": 86.59655831739961, "grad_norm": 0.379455029964447, "learning_rate": 6.18094191310636e-05, "loss": 0.0181, "step": 135870 }, { "epoch": 86.60293180369662, "grad_norm": 0.5012579560279846, "learning_rate": 6.180406343043489e-05, "loss": 0.022, "step": 135880 }, { "epoch": 86.60930528999363, "grad_norm": 0.4783099889755249, "learning_rate": 6.179870758637565e-05, "loss": 0.0279, "step": 135890 }, { "epoch": 86.61567877629064, "grad_norm": 0.2859387993812561, "learning_rate": 6.179335159895099e-05, "loss": 0.0107, "step": 135900 }, { "epoch": 86.62205226258763, "grad_norm": 0.5143576860427856, "learning_rate": 6.178799546822595e-05, "loss": 0.0185, "step": 135910 }, { "epoch": 86.62842574888464, "grad_norm": 0.43622055649757385, "learning_rate": 6.178263919426566e-05, "loss": 0.0205, "step": 135920 }, { "epoch": 86.63479923518165, "grad_norm": 0.5519411563873291, "learning_rate": 6.177728277713516e-05, "loss": 0.0139, "step": 135930 }, { "epoch": 86.64117272147865, "grad_norm": 0.6829575300216675, "learning_rate": 6.177192621689957e-05, "loss": 0.0186, "step": 135940 }, { "epoch": 86.64754620777565, "grad_norm": 0.38545626401901245, "learning_rate": 6.176656951362395e-05, "loss": 0.0257, "step": 135950 }, { "epoch": 86.65391969407266, "grad_norm": 0.27389034628868103, "learning_rate": 6.17612126673734e-05, "loss": 0.0326, "step": 135960 }, { "epoch": 86.66029318036966, "grad_norm": 0.21246659755706787, "learning_rate": 6.175585567821303e-05, "loss": 0.0106, "step": 135970 }, { "epoch": 86.66666666666667, "grad_norm": 0.2679625153541565, "learning_rate": 6.17504985462079e-05, "loss": 0.0129, "step": 135980 }, { "epoch": 86.67304015296367, "grad_norm": 0.26560646295547485, "learning_rate": 6.174514127142312e-05, "loss": 0.0203, "step": 135990 }, { "epoch": 86.67941363926067, "grad_norm": 0.432272732257843, "learning_rate": 6.173978385392377e-05, "loss": 0.0236, "step": 136000 }, { "epoch": 86.68578712555768, "grad_norm": 0.6627709269523621, "learning_rate": 6.173442629377498e-05, "loss": 0.0202, "step": 136010 }, { "epoch": 86.69216061185469, "grad_norm": 0.5887337923049927, "learning_rate": 6.172906859104183e-05, "loss": 0.0145, "step": 136020 }, { "epoch": 86.69853409815168, "grad_norm": 0.47394421696662903, "learning_rate": 6.172371074578942e-05, "loss": 0.0158, "step": 136030 }, { "epoch": 86.70490758444869, "grad_norm": 0.4661499857902527, "learning_rate": 6.171835275808286e-05, "loss": 0.0247, "step": 136040 }, { "epoch": 86.7112810707457, "grad_norm": 0.27699604630470276, "learning_rate": 6.171299462798724e-05, "loss": 0.0196, "step": 136050 }, { "epoch": 86.7176545570427, "grad_norm": 0.31430625915527344, "learning_rate": 6.170763635556767e-05, "loss": 0.0132, "step": 136060 }, { "epoch": 86.7240280433397, "grad_norm": 0.866858184337616, "learning_rate": 6.170227794088928e-05, "loss": 0.0147, "step": 136070 }, { "epoch": 86.73040152963671, "grad_norm": 0.3130281865596771, "learning_rate": 6.169691938401716e-05, "loss": 0.0136, "step": 136080 }, { "epoch": 86.73677501593372, "grad_norm": 0.566281259059906, "learning_rate": 6.169156068501643e-05, "loss": 0.0157, "step": 136090 }, { "epoch": 86.74314850223072, "grad_norm": 0.24743251502513885, "learning_rate": 6.16862018439522e-05, "loss": 0.0191, "step": 136100 }, { "epoch": 86.74952198852772, "grad_norm": 0.3813864588737488, "learning_rate": 6.168084286088958e-05, "loss": 0.0201, "step": 136110 }, { "epoch": 86.75589547482473, "grad_norm": 0.3566763699054718, "learning_rate": 6.167548373589368e-05, "loss": 0.0248, "step": 136120 }, { "epoch": 86.76226896112173, "grad_norm": 0.4163026213645935, "learning_rate": 6.167012446902964e-05, "loss": 0.0241, "step": 136130 }, { "epoch": 86.76864244741874, "grad_norm": 0.3649134039878845, "learning_rate": 6.166476506036257e-05, "loss": 0.0166, "step": 136140 }, { "epoch": 86.77501593371574, "grad_norm": 0.19617947936058044, "learning_rate": 6.165940550995759e-05, "loss": 0.0183, "step": 136150 }, { "epoch": 86.78138942001274, "grad_norm": 0.20727837085723877, "learning_rate": 6.165404581787983e-05, "loss": 0.0086, "step": 136160 }, { "epoch": 86.78776290630975, "grad_norm": 0.2628084719181061, "learning_rate": 6.164868598419439e-05, "loss": 0.0138, "step": 136170 }, { "epoch": 86.79413639260676, "grad_norm": 0.7333850860595703, "learning_rate": 6.164332600896645e-05, "loss": 0.0184, "step": 136180 }, { "epoch": 86.80050987890377, "grad_norm": 0.19258369505405426, "learning_rate": 6.163796589226108e-05, "loss": 0.0093, "step": 136190 }, { "epoch": 86.80688336520076, "grad_norm": 0.48198413848876953, "learning_rate": 6.163260563414344e-05, "loss": 0.0245, "step": 136200 }, { "epoch": 86.81325685149777, "grad_norm": 0.5114803910255432, "learning_rate": 6.162724523467867e-05, "loss": 0.0131, "step": 136210 }, { "epoch": 86.81963033779478, "grad_norm": 0.2419361174106598, "learning_rate": 6.162188469393188e-05, "loss": 0.0121, "step": 136220 }, { "epoch": 86.82600382409179, "grad_norm": 0.5240670442581177, "learning_rate": 6.161652401196822e-05, "loss": 0.0141, "step": 136230 }, { "epoch": 86.83237731038878, "grad_norm": 0.49525779485702515, "learning_rate": 6.161116318885283e-05, "loss": 0.0099, "step": 136240 }, { "epoch": 86.83875079668579, "grad_norm": 0.38083189725875854, "learning_rate": 6.160580222465085e-05, "loss": 0.0165, "step": 136250 }, { "epoch": 86.8451242829828, "grad_norm": 0.5293951630592346, "learning_rate": 6.16004411194274e-05, "loss": 0.0174, "step": 136260 }, { "epoch": 86.8514977692798, "grad_norm": 0.2664691209793091, "learning_rate": 6.159507987324765e-05, "loss": 0.0166, "step": 136270 }, { "epoch": 86.8578712555768, "grad_norm": 0.20196571946144104, "learning_rate": 6.158971848617673e-05, "loss": 0.0271, "step": 136280 }, { "epoch": 86.8642447418738, "grad_norm": 0.14810959994792938, "learning_rate": 6.158435695827979e-05, "loss": 0.018, "step": 136290 }, { "epoch": 86.87061822817081, "grad_norm": 0.2296142280101776, "learning_rate": 6.157899528962197e-05, "loss": 0.0138, "step": 136300 }, { "epoch": 86.87699171446782, "grad_norm": 0.15450845658779144, "learning_rate": 6.157363348026842e-05, "loss": 0.0151, "step": 136310 }, { "epoch": 86.88336520076481, "grad_norm": 0.48244673013687134, "learning_rate": 6.15682715302843e-05, "loss": 0.0218, "step": 136320 }, { "epoch": 86.88973868706182, "grad_norm": 0.3250318467617035, "learning_rate": 6.156290943973476e-05, "loss": 0.011, "step": 136330 }, { "epoch": 86.89611217335883, "grad_norm": 0.3073652386665344, "learning_rate": 6.155754720868496e-05, "loss": 0.013, "step": 136340 }, { "epoch": 86.90248565965584, "grad_norm": 0.3023199141025543, "learning_rate": 6.155218483720004e-05, "loss": 0.0113, "step": 136350 }, { "epoch": 86.90885914595283, "grad_norm": 0.46760857105255127, "learning_rate": 6.154682232534517e-05, "loss": 0.0118, "step": 136360 }, { "epoch": 86.91523263224984, "grad_norm": 0.3629573583602905, "learning_rate": 6.154145967318551e-05, "loss": 0.0163, "step": 136370 }, { "epoch": 86.92160611854685, "grad_norm": 0.4137875437736511, "learning_rate": 6.15360968807862e-05, "loss": 0.0185, "step": 136380 }, { "epoch": 86.92797960484386, "grad_norm": 0.2590354084968567, "learning_rate": 6.153073394821244e-05, "loss": 0.012, "step": 136390 }, { "epoch": 86.93435309114085, "grad_norm": 0.36579543352127075, "learning_rate": 6.152537087552938e-05, "loss": 0.0127, "step": 136400 }, { "epoch": 86.94072657743786, "grad_norm": 0.2452632337808609, "learning_rate": 6.152000766280218e-05, "loss": 0.0101, "step": 136410 }, { "epoch": 86.94710006373487, "grad_norm": 0.3453703224658966, "learning_rate": 6.151464431009599e-05, "loss": 0.0141, "step": 136420 }, { "epoch": 86.95347355003187, "grad_norm": 0.46439236402511597, "learning_rate": 6.150928081747603e-05, "loss": 0.0152, "step": 136430 }, { "epoch": 86.95984703632887, "grad_norm": 0.18831168115139008, "learning_rate": 6.150391718500741e-05, "loss": 0.0197, "step": 136440 }, { "epoch": 86.96622052262587, "grad_norm": 0.2990996539592743, "learning_rate": 6.149855341275536e-05, "loss": 0.0199, "step": 136450 }, { "epoch": 86.97259400892288, "grad_norm": 0.2614096701145172, "learning_rate": 6.149318950078501e-05, "loss": 0.0089, "step": 136460 }, { "epoch": 86.97896749521989, "grad_norm": 0.4590532183647156, "learning_rate": 6.148782544916157e-05, "loss": 0.0238, "step": 136470 }, { "epoch": 86.98534098151688, "grad_norm": 0.4219270348548889, "learning_rate": 6.14824612579502e-05, "loss": 0.0169, "step": 136480 }, { "epoch": 86.99171446781389, "grad_norm": 0.412659227848053, "learning_rate": 6.147709692721608e-05, "loss": 0.0299, "step": 136490 }, { "epoch": 86.9980879541109, "grad_norm": 0.28372716903686523, "learning_rate": 6.147173245702442e-05, "loss": 0.0168, "step": 136500 }, { "epoch": 87.00446144040791, "grad_norm": 0.36065414547920227, "learning_rate": 6.146636784744036e-05, "loss": 0.0107, "step": 136510 }, { "epoch": 87.0108349267049, "grad_norm": 0.21776160597801208, "learning_rate": 6.14610030985291e-05, "loss": 0.0103, "step": 136520 }, { "epoch": 87.01720841300191, "grad_norm": 0.377972811460495, "learning_rate": 6.145563821035582e-05, "loss": 0.0081, "step": 136530 }, { "epoch": 87.02358189929892, "grad_norm": 0.32573941349983215, "learning_rate": 6.145027318298576e-05, "loss": 0.0147, "step": 136540 }, { "epoch": 87.02995538559593, "grad_norm": 0.19913597404956818, "learning_rate": 6.144490801648404e-05, "loss": 0.0086, "step": 136550 }, { "epoch": 87.03632887189292, "grad_norm": 0.15542803704738617, "learning_rate": 6.143954271091589e-05, "loss": 0.0133, "step": 136560 }, { "epoch": 87.04270235818993, "grad_norm": 0.19216316938400269, "learning_rate": 6.14341772663465e-05, "loss": 0.0112, "step": 136570 }, { "epoch": 87.04907584448694, "grad_norm": 0.3057840168476105, "learning_rate": 6.142881168284104e-05, "loss": 0.0141, "step": 136580 }, { "epoch": 87.05544933078394, "grad_norm": 0.6105427145957947, "learning_rate": 6.142344596046475e-05, "loss": 0.0262, "step": 136590 }, { "epoch": 87.06182281708094, "grad_norm": 0.439746230840683, "learning_rate": 6.14180800992828e-05, "loss": 0.0213, "step": 136600 }, { "epoch": 87.06819630337795, "grad_norm": 0.24846015870571136, "learning_rate": 6.141271409936039e-05, "loss": 0.0165, "step": 136610 }, { "epoch": 87.07456978967495, "grad_norm": 0.2127065807580948, "learning_rate": 6.140734796076273e-05, "loss": 0.0252, "step": 136620 }, { "epoch": 87.08094327597196, "grad_norm": 0.22384266555309296, "learning_rate": 6.140198168355502e-05, "loss": 0.015, "step": 136630 }, { "epoch": 87.08731676226896, "grad_norm": 0.3309597671031952, "learning_rate": 6.139661526780247e-05, "loss": 0.0099, "step": 136640 }, { "epoch": 87.09369024856596, "grad_norm": 0.3901521563529968, "learning_rate": 6.13912487135703e-05, "loss": 0.0191, "step": 136650 }, { "epoch": 87.10006373486297, "grad_norm": 0.4719346761703491, "learning_rate": 6.138588202092367e-05, "loss": 0.0145, "step": 136660 }, { "epoch": 87.10643722115998, "grad_norm": 0.3398447334766388, "learning_rate": 6.138051518992784e-05, "loss": 0.0121, "step": 136670 }, { "epoch": 87.11281070745697, "grad_norm": 0.5052052736282349, "learning_rate": 6.137514822064802e-05, "loss": 0.0226, "step": 136680 }, { "epoch": 87.11918419375398, "grad_norm": 0.3320371210575104, "learning_rate": 6.13697811131494e-05, "loss": 0.0115, "step": 136690 }, { "epoch": 87.12555768005099, "grad_norm": 0.15817736089229584, "learning_rate": 6.136441386749719e-05, "loss": 0.0177, "step": 136700 }, { "epoch": 87.131931166348, "grad_norm": 0.25440025329589844, "learning_rate": 6.135904648375663e-05, "loss": 0.0132, "step": 136710 }, { "epoch": 87.13830465264499, "grad_norm": 0.9420782923698425, "learning_rate": 6.135367896199294e-05, "loss": 0.0233, "step": 136720 }, { "epoch": 87.144678138942, "grad_norm": 0.38487347960472107, "learning_rate": 6.134831130227132e-05, "loss": 0.0103, "step": 136730 }, { "epoch": 87.151051625239, "grad_norm": 0.25661471486091614, "learning_rate": 6.134294350465701e-05, "loss": 0.013, "step": 136740 }, { "epoch": 87.15742511153601, "grad_norm": 0.25249311327934265, "learning_rate": 6.133757556921523e-05, "loss": 0.0132, "step": 136750 }, { "epoch": 87.16379859783301, "grad_norm": 0.6065831780433655, "learning_rate": 6.133220749601119e-05, "loss": 0.0161, "step": 136760 }, { "epoch": 87.17017208413002, "grad_norm": 0.36369338631629944, "learning_rate": 6.132683928511014e-05, "loss": 0.0139, "step": 136770 }, { "epoch": 87.17654557042702, "grad_norm": 0.5983407497406006, "learning_rate": 6.13214709365773e-05, "loss": 0.0151, "step": 136780 }, { "epoch": 87.18291905672403, "grad_norm": 0.25521352887153625, "learning_rate": 6.131610245047789e-05, "loss": 0.0116, "step": 136790 }, { "epoch": 87.18929254302103, "grad_norm": 0.33713090419769287, "learning_rate": 6.131073382687715e-05, "loss": 0.0094, "step": 136800 }, { "epoch": 87.19566602931803, "grad_norm": 0.4561706781387329, "learning_rate": 6.130536506584032e-05, "loss": 0.0265, "step": 136810 }, { "epoch": 87.20203951561504, "grad_norm": 0.6540097594261169, "learning_rate": 6.129999616743264e-05, "loss": 0.0143, "step": 136820 }, { "epoch": 87.20841300191205, "grad_norm": 0.5582777857780457, "learning_rate": 6.129462713171932e-05, "loss": 0.0151, "step": 136830 }, { "epoch": 87.21478648820906, "grad_norm": 0.38014474511146545, "learning_rate": 6.128925795876562e-05, "loss": 0.022, "step": 136840 }, { "epoch": 87.22115997450605, "grad_norm": 0.6847732663154602, "learning_rate": 6.128388864863678e-05, "loss": 0.0227, "step": 136850 }, { "epoch": 87.22753346080306, "grad_norm": 0.41780000925064087, "learning_rate": 6.127851920139804e-05, "loss": 0.0117, "step": 136860 }, { "epoch": 87.23390694710007, "grad_norm": 0.35452067852020264, "learning_rate": 6.127314961711464e-05, "loss": 0.0215, "step": 136870 }, { "epoch": 87.24028043339707, "grad_norm": 0.17160610854625702, "learning_rate": 6.126777989585185e-05, "loss": 0.0214, "step": 136880 }, { "epoch": 87.24665391969407, "grad_norm": 0.47455185651779175, "learning_rate": 6.126241003767487e-05, "loss": 0.021, "step": 136890 }, { "epoch": 87.25302740599108, "grad_norm": 0.27651721239089966, "learning_rate": 6.125704004264898e-05, "loss": 0.0122, "step": 136900 }, { "epoch": 87.25940089228808, "grad_norm": 0.3424924910068512, "learning_rate": 6.125166991083942e-05, "loss": 0.0141, "step": 136910 }, { "epoch": 87.26577437858509, "grad_norm": 0.37724751234054565, "learning_rate": 6.124629964231145e-05, "loss": 0.0099, "step": 136920 }, { "epoch": 87.27214786488209, "grad_norm": 0.3622196316719055, "learning_rate": 6.124092923713033e-05, "loss": 0.0199, "step": 136930 }, { "epoch": 87.2785213511791, "grad_norm": 0.15154772996902466, "learning_rate": 6.12355586953613e-05, "loss": 0.0094, "step": 136940 }, { "epoch": 87.2848948374761, "grad_norm": 0.2815091609954834, "learning_rate": 6.123018801706963e-05, "loss": 0.0127, "step": 136950 }, { "epoch": 87.29126832377311, "grad_norm": 0.20359069108963013, "learning_rate": 6.122481720232058e-05, "loss": 0.0125, "step": 136960 }, { "epoch": 87.2976418100701, "grad_norm": 0.643278181552887, "learning_rate": 6.121944625117939e-05, "loss": 0.013, "step": 136970 }, { "epoch": 87.30401529636711, "grad_norm": 0.24554336071014404, "learning_rate": 6.121407516371134e-05, "loss": 0.0114, "step": 136980 }, { "epoch": 87.31038878266412, "grad_norm": 0.2733408212661743, "learning_rate": 6.120870393998168e-05, "loss": 0.0088, "step": 136990 }, { "epoch": 87.31676226896113, "grad_norm": 0.3001162111759186, "learning_rate": 6.120333258005569e-05, "loss": 0.0199, "step": 137000 }, { "epoch": 87.32313575525812, "grad_norm": 0.25674840807914734, "learning_rate": 6.119796108399862e-05, "loss": 0.017, "step": 137010 }, { "epoch": 87.32950924155513, "grad_norm": 0.3565239906311035, "learning_rate": 6.119258945187577e-05, "loss": 0.0196, "step": 137020 }, { "epoch": 87.33588272785214, "grad_norm": 0.26449480652809143, "learning_rate": 6.118721768375238e-05, "loss": 0.0089, "step": 137030 }, { "epoch": 87.34225621414915, "grad_norm": 0.2986873388290405, "learning_rate": 6.118184577969374e-05, "loss": 0.0098, "step": 137040 }, { "epoch": 87.34862970044614, "grad_norm": 0.5817490816116333, "learning_rate": 6.117647373976512e-05, "loss": 0.016, "step": 137050 }, { "epoch": 87.35500318674315, "grad_norm": 0.1572684645652771, "learning_rate": 6.117110156403178e-05, "loss": 0.0116, "step": 137060 }, { "epoch": 87.36137667304016, "grad_norm": 0.2864267826080322, "learning_rate": 6.116572925255902e-05, "loss": 0.0167, "step": 137070 }, { "epoch": 87.36775015933716, "grad_norm": 0.3274248540401459, "learning_rate": 6.116035680541208e-05, "loss": 0.0106, "step": 137080 }, { "epoch": 87.37412364563416, "grad_norm": 0.40401580929756165, "learning_rate": 6.11549842226563e-05, "loss": 0.0158, "step": 137090 }, { "epoch": 87.38049713193116, "grad_norm": 0.40803083777427673, "learning_rate": 6.114961150435691e-05, "loss": 0.0176, "step": 137100 }, { "epoch": 87.38687061822817, "grad_norm": 0.662192165851593, "learning_rate": 6.114423865057922e-05, "loss": 0.0199, "step": 137110 }, { "epoch": 87.39324410452518, "grad_norm": 0.4642355144023895, "learning_rate": 6.11388656613885e-05, "loss": 0.016, "step": 137120 }, { "epoch": 87.39961759082217, "grad_norm": 0.6393150091171265, "learning_rate": 6.113349253685006e-05, "loss": 0.0181, "step": 137130 }, { "epoch": 87.40599107711918, "grad_norm": 0.29093173146247864, "learning_rate": 6.112811927702917e-05, "loss": 0.0142, "step": 137140 }, { "epoch": 87.41236456341619, "grad_norm": 0.5215398073196411, "learning_rate": 6.112274588199111e-05, "loss": 0.0143, "step": 137150 }, { "epoch": 87.4187380497132, "grad_norm": 0.49087029695510864, "learning_rate": 6.111737235180121e-05, "loss": 0.0146, "step": 137160 }, { "epoch": 87.42511153601019, "grad_norm": 0.20683076977729797, "learning_rate": 6.111199868652472e-05, "loss": 0.0153, "step": 137170 }, { "epoch": 87.4314850223072, "grad_norm": 0.40812626481056213, "learning_rate": 6.110662488622696e-05, "loss": 0.0177, "step": 137180 }, { "epoch": 87.43785850860421, "grad_norm": 0.3843812048435211, "learning_rate": 6.110125095097321e-05, "loss": 0.0115, "step": 137190 }, { "epoch": 87.44423199490122, "grad_norm": 0.3612349033355713, "learning_rate": 6.109587688082881e-05, "loss": 0.0085, "step": 137200 }, { "epoch": 87.45060548119821, "grad_norm": 0.3207167387008667, "learning_rate": 6.109050267585901e-05, "loss": 0.011, "step": 137210 }, { "epoch": 87.45697896749522, "grad_norm": 0.41377007961273193, "learning_rate": 6.108512833612913e-05, "loss": 0.0125, "step": 137220 }, { "epoch": 87.46335245379223, "grad_norm": 0.43529173731803894, "learning_rate": 6.107975386170448e-05, "loss": 0.0201, "step": 137230 }, { "epoch": 87.46972594008923, "grad_norm": 0.5296605825424194, "learning_rate": 6.107437925265035e-05, "loss": 0.0115, "step": 137240 }, { "epoch": 87.47609942638623, "grad_norm": 0.4453665614128113, "learning_rate": 6.106900450903209e-05, "loss": 0.0155, "step": 137250 }, { "epoch": 87.48247291268324, "grad_norm": 0.6092602014541626, "learning_rate": 6.106362963091493e-05, "loss": 0.0236, "step": 137260 }, { "epoch": 87.48884639898024, "grad_norm": 0.38884127140045166, "learning_rate": 6.105825461836425e-05, "loss": 0.0151, "step": 137270 }, { "epoch": 87.49521988527725, "grad_norm": 0.2980034351348877, "learning_rate": 6.10528794714453e-05, "loss": 0.0107, "step": 137280 }, { "epoch": 87.50159337157424, "grad_norm": 0.49606195092201233, "learning_rate": 6.104750419022348e-05, "loss": 0.0177, "step": 137290 }, { "epoch": 87.50796685787125, "grad_norm": 0.26936498284339905, "learning_rate": 6.104212877476403e-05, "loss": 0.0169, "step": 137300 }, { "epoch": 87.51434034416826, "grad_norm": 0.14850039780139923, "learning_rate": 6.103675322513228e-05, "loss": 0.0104, "step": 137310 }, { "epoch": 87.52071383046527, "grad_norm": 0.3515681028366089, "learning_rate": 6.1031377541393573e-05, "loss": 0.0086, "step": 137320 }, { "epoch": 87.52708731676226, "grad_norm": 0.5784274339675903, "learning_rate": 6.102600172361321e-05, "loss": 0.0159, "step": 137330 }, { "epoch": 87.53346080305927, "grad_norm": 0.22007283568382263, "learning_rate": 6.102062577185652e-05, "loss": 0.0208, "step": 137340 }, { "epoch": 87.53983428935628, "grad_norm": 0.3635593056678772, "learning_rate": 6.10152496861888e-05, "loss": 0.0167, "step": 137350 }, { "epoch": 87.54620777565329, "grad_norm": 0.33827078342437744, "learning_rate": 6.1009873466675424e-05, "loss": 0.0149, "step": 137360 }, { "epoch": 87.55258126195028, "grad_norm": 0.23772995173931122, "learning_rate": 6.100449711338169e-05, "loss": 0.0121, "step": 137370 }, { "epoch": 87.55895474824729, "grad_norm": 0.30862006545066833, "learning_rate": 6.0999120626372894e-05, "loss": 0.0171, "step": 137380 }, { "epoch": 87.5653282345443, "grad_norm": 0.384147971868515, "learning_rate": 6.099374400571443e-05, "loss": 0.0227, "step": 137390 }, { "epoch": 87.5717017208413, "grad_norm": 0.3602431118488312, "learning_rate": 6.0988367251471577e-05, "loss": 0.0124, "step": 137400 }, { "epoch": 87.5780752071383, "grad_norm": 0.27384674549102783, "learning_rate": 6.09829903637097e-05, "loss": 0.0175, "step": 137410 }, { "epoch": 87.5844486934353, "grad_norm": 0.5771077871322632, "learning_rate": 6.097761334249411e-05, "loss": 0.0132, "step": 137420 }, { "epoch": 87.59082217973231, "grad_norm": 0.28370943665504456, "learning_rate": 6.097223618789017e-05, "loss": 0.0163, "step": 137430 }, { "epoch": 87.59719566602932, "grad_norm": 0.2920137345790863, "learning_rate": 6.0966858899963175e-05, "loss": 0.024, "step": 137440 }, { "epoch": 87.60356915232632, "grad_norm": 0.2804521918296814, "learning_rate": 6.096148147877851e-05, "loss": 0.0176, "step": 137450 }, { "epoch": 87.60994263862332, "grad_norm": 0.26651501655578613, "learning_rate": 6.0956103924401484e-05, "loss": 0.0172, "step": 137460 }, { "epoch": 87.61631612492033, "grad_norm": 0.30589309334754944, "learning_rate": 6.095072623689746e-05, "loss": 0.0123, "step": 137470 }, { "epoch": 87.62268961121734, "grad_norm": 0.5025419592857361, "learning_rate": 6.094534841633176e-05, "loss": 0.0172, "step": 137480 }, { "epoch": 87.62906309751435, "grad_norm": 0.5468724966049194, "learning_rate": 6.0939970462769756e-05, "loss": 0.016, "step": 137490 }, { "epoch": 87.63543658381134, "grad_norm": 0.2551041841506958, "learning_rate": 6.093459237627678e-05, "loss": 0.0124, "step": 137500 }, { "epoch": 87.64181007010835, "grad_norm": 0.23530267179012299, "learning_rate": 6.092921415691818e-05, "loss": 0.0162, "step": 137510 }, { "epoch": 87.64818355640536, "grad_norm": 0.5390749573707581, "learning_rate": 6.092383580475931e-05, "loss": 0.0114, "step": 137520 }, { "epoch": 87.65455704270236, "grad_norm": 0.3395254611968994, "learning_rate": 6.091845731986552e-05, "loss": 0.0133, "step": 137530 }, { "epoch": 87.66093052899936, "grad_norm": 0.3853604197502136, "learning_rate": 6.091307870230216e-05, "loss": 0.0168, "step": 137540 }, { "epoch": 87.66730401529637, "grad_norm": 0.49970465898513794, "learning_rate": 6.09076999521346e-05, "loss": 0.0125, "step": 137550 }, { "epoch": 87.67367750159337, "grad_norm": 0.19122712314128876, "learning_rate": 6.090232106942818e-05, "loss": 0.0141, "step": 137560 }, { "epoch": 87.68005098789038, "grad_norm": 0.19156301021575928, "learning_rate": 6.089694205424828e-05, "loss": 0.0131, "step": 137570 }, { "epoch": 87.68642447418738, "grad_norm": 0.29423239827156067, "learning_rate": 6.0891562906660226e-05, "loss": 0.0137, "step": 137580 }, { "epoch": 87.69279796048438, "grad_norm": 0.4403967559337616, "learning_rate": 6.088618362672941e-05, "loss": 0.013, "step": 137590 }, { "epoch": 87.69917144678139, "grad_norm": 0.5550444722175598, "learning_rate": 6.088080421452118e-05, "loss": 0.0121, "step": 137600 }, { "epoch": 87.7055449330784, "grad_norm": 0.30383792519569397, "learning_rate": 6.087542467010091e-05, "loss": 0.0211, "step": 137610 }, { "epoch": 87.7119184193754, "grad_norm": 0.44512856006622314, "learning_rate": 6.087004499353396e-05, "loss": 0.0185, "step": 137620 }, { "epoch": 87.7182919056724, "grad_norm": 0.3710156977176666, "learning_rate": 6.0864665184885694e-05, "loss": 0.0137, "step": 137630 }, { "epoch": 87.72466539196941, "grad_norm": 0.5007081031799316, "learning_rate": 6.085928524422151e-05, "loss": 0.0097, "step": 137640 }, { "epoch": 87.73103887826642, "grad_norm": 0.5960485339164734, "learning_rate": 6.085390517160673e-05, "loss": 0.0176, "step": 137650 }, { "epoch": 87.73741236456341, "grad_norm": 0.44741541147232056, "learning_rate": 6.084852496710677e-05, "loss": 0.0103, "step": 137660 }, { "epoch": 87.74378585086042, "grad_norm": 0.2819337844848633, "learning_rate": 6.0843144630786975e-05, "loss": 0.0113, "step": 137670 }, { "epoch": 87.75015933715743, "grad_norm": 0.37044236063957214, "learning_rate": 6.083776416271275e-05, "loss": 0.0218, "step": 137680 }, { "epoch": 87.75653282345444, "grad_norm": 0.48306572437286377, "learning_rate": 6.0832383562949444e-05, "loss": 0.0304, "step": 137690 }, { "epoch": 87.76290630975143, "grad_norm": 0.28179633617401123, "learning_rate": 6.082700283156247e-05, "loss": 0.0105, "step": 137700 }, { "epoch": 87.76927979604844, "grad_norm": 0.26671460270881653, "learning_rate": 6.082162196861718e-05, "loss": 0.0088, "step": 137710 }, { "epoch": 87.77565328234544, "grad_norm": 0.23846182227134705, "learning_rate": 6.081624097417896e-05, "loss": 0.0192, "step": 137720 }, { "epoch": 87.78202676864245, "grad_norm": 0.280593603849411, "learning_rate": 6.0810859848313206e-05, "loss": 0.0196, "step": 137730 }, { "epoch": 87.78840025493945, "grad_norm": 0.369967520236969, "learning_rate": 6.08054785910853e-05, "loss": 0.013, "step": 137740 }, { "epoch": 87.79477374123645, "grad_norm": 0.3321811258792877, "learning_rate": 6.0800097202560624e-05, "loss": 0.0121, "step": 137750 }, { "epoch": 87.80114722753346, "grad_norm": 0.2858947813510895, "learning_rate": 6.079471568280456e-05, "loss": 0.034, "step": 137760 }, { "epoch": 87.80752071383047, "grad_norm": 0.2513634264469147, "learning_rate": 6.078933403188253e-05, "loss": 0.0098, "step": 137770 }, { "epoch": 87.81389420012746, "grad_norm": 0.46375611424446106, "learning_rate": 6.078395224985989e-05, "loss": 0.0158, "step": 137780 }, { "epoch": 87.82026768642447, "grad_norm": 0.2529365122318268, "learning_rate": 6.0778570336802044e-05, "loss": 0.0226, "step": 137790 }, { "epoch": 87.82664117272148, "grad_norm": 0.41690701246261597, "learning_rate": 6.077318829277441e-05, "loss": 0.0116, "step": 137800 }, { "epoch": 87.83301465901849, "grad_norm": 0.9054744839668274, "learning_rate": 6.076780611784234e-05, "loss": 0.0135, "step": 137810 }, { "epoch": 87.83938814531548, "grad_norm": 0.16648191213607788, "learning_rate": 6.076242381207129e-05, "loss": 0.0109, "step": 137820 }, { "epoch": 87.84576163161249, "grad_norm": 0.2318059504032135, "learning_rate": 6.07570413755266e-05, "loss": 0.0249, "step": 137830 }, { "epoch": 87.8521351179095, "grad_norm": 0.355350524187088, "learning_rate": 6.0751658808273724e-05, "loss": 0.0158, "step": 137840 }, { "epoch": 87.8585086042065, "grad_norm": 0.46908000111579895, "learning_rate": 6.074627611037802e-05, "loss": 0.0228, "step": 137850 }, { "epoch": 87.8648820905035, "grad_norm": 0.4624278247356415, "learning_rate": 6.074089328190493e-05, "loss": 0.0135, "step": 137860 }, { "epoch": 87.87125557680051, "grad_norm": 0.29126614332199097, "learning_rate": 6.0735510322919855e-05, "loss": 0.013, "step": 137870 }, { "epoch": 87.87762906309752, "grad_norm": 0.3741316497325897, "learning_rate": 6.073012723348818e-05, "loss": 0.0186, "step": 137880 }, { "epoch": 87.88400254939452, "grad_norm": 0.46564021706581116, "learning_rate": 6.0724744013675325e-05, "loss": 0.0169, "step": 137890 }, { "epoch": 87.89037603569152, "grad_norm": 0.28127071261405945, "learning_rate": 6.071936066354671e-05, "loss": 0.0085, "step": 137900 }, { "epoch": 87.89674952198853, "grad_norm": 0.26654067635536194, "learning_rate": 6.071397718316775e-05, "loss": 0.0132, "step": 137910 }, { "epoch": 87.90312300828553, "grad_norm": 0.32834935188293457, "learning_rate": 6.070859357260383e-05, "loss": 0.0215, "step": 137920 }, { "epoch": 87.90949649458254, "grad_norm": 0.3755296468734741, "learning_rate": 6.070320983192042e-05, "loss": 0.0165, "step": 137930 }, { "epoch": 87.91586998087953, "grad_norm": 0.596771240234375, "learning_rate": 6.069782596118288e-05, "loss": 0.0119, "step": 137940 }, { "epoch": 87.92224346717654, "grad_norm": 0.3252980411052704, "learning_rate": 6.0692441960456657e-05, "loss": 0.0107, "step": 137950 }, { "epoch": 87.92861695347355, "grad_norm": 0.441940039396286, "learning_rate": 6.068705782980717e-05, "loss": 0.014, "step": 137960 }, { "epoch": 87.93499043977056, "grad_norm": 0.5034086108207703, "learning_rate": 6.0681673569299834e-05, "loss": 0.0175, "step": 137970 }, { "epoch": 87.94136392606755, "grad_norm": 0.7978722453117371, "learning_rate": 6.06762891790001e-05, "loss": 0.029, "step": 137980 }, { "epoch": 87.94773741236456, "grad_norm": 0.461183100938797, "learning_rate": 6.067090465897335e-05, "loss": 0.0211, "step": 137990 }, { "epoch": 87.95411089866157, "grad_norm": 0.3494621515274048, "learning_rate": 6.066552000928505e-05, "loss": 0.0209, "step": 138000 }, { "epoch": 87.96048438495858, "grad_norm": 0.3269343078136444, "learning_rate": 6.0660135230000604e-05, "loss": 0.0136, "step": 138010 }, { "epoch": 87.96685787125557, "grad_norm": 0.3699321746826172, "learning_rate": 6.065475032118545e-05, "loss": 0.011, "step": 138020 }, { "epoch": 87.97323135755258, "grad_norm": 0.2585626244544983, "learning_rate": 6.064936528290503e-05, "loss": 0.015, "step": 138030 }, { "epoch": 87.97960484384959, "grad_norm": 0.18687231838703156, "learning_rate": 6.064398011522474e-05, "loss": 0.0185, "step": 138040 }, { "epoch": 87.9859783301466, "grad_norm": 0.6049677133560181, "learning_rate": 6.063859481821007e-05, "loss": 0.0149, "step": 138050 }, { "epoch": 87.99235181644359, "grad_norm": 0.49252572655677795, "learning_rate": 6.063320939192641e-05, "loss": 0.0108, "step": 138060 }, { "epoch": 87.9987253027406, "grad_norm": 0.41785725951194763, "learning_rate": 6.0627823836439234e-05, "loss": 0.009, "step": 138070 }, { "epoch": 88.0050987890376, "grad_norm": 0.21481095254421234, "learning_rate": 6.062243815181394e-05, "loss": 0.0112, "step": 138080 }, { "epoch": 88.01147227533461, "grad_norm": 0.20791058242321014, "learning_rate": 6.061705233811602e-05, "loss": 0.0174, "step": 138090 }, { "epoch": 88.01784576163162, "grad_norm": 0.4167209565639496, "learning_rate": 6.0611666395410885e-05, "loss": 0.0188, "step": 138100 }, { "epoch": 88.02421924792861, "grad_norm": 0.46749648451805115, "learning_rate": 6.0606280323763966e-05, "loss": 0.0209, "step": 138110 }, { "epoch": 88.03059273422562, "grad_norm": 0.2618454396724701, "learning_rate": 6.060089412324075e-05, "loss": 0.017, "step": 138120 }, { "epoch": 88.03696622052263, "grad_norm": 0.5788019895553589, "learning_rate": 6.059550779390664e-05, "loss": 0.0153, "step": 138130 }, { "epoch": 88.04333970681964, "grad_norm": 0.30578893423080444, "learning_rate": 6.059012133582712e-05, "loss": 0.016, "step": 138140 }, { "epoch": 88.04971319311663, "grad_norm": 0.42068061232566833, "learning_rate": 6.058473474906762e-05, "loss": 0.0141, "step": 138150 }, { "epoch": 88.05608667941364, "grad_norm": 0.5631097555160522, "learning_rate": 6.0579348033693604e-05, "loss": 0.0189, "step": 138160 }, { "epoch": 88.06246016571065, "grad_norm": 0.21441565454006195, "learning_rate": 6.0573961189770515e-05, "loss": 0.0148, "step": 138170 }, { "epoch": 88.06883365200765, "grad_norm": 0.5174910426139832, "learning_rate": 6.056857421736383e-05, "loss": 0.0125, "step": 138180 }, { "epoch": 88.07520713830465, "grad_norm": 0.48113512992858887, "learning_rate": 6.056318711653897e-05, "loss": 0.0076, "step": 138190 }, { "epoch": 88.08158062460166, "grad_norm": 0.5321341753005981, "learning_rate": 6.055779988736142e-05, "loss": 0.0161, "step": 138200 }, { "epoch": 88.08795411089866, "grad_norm": 0.5354197025299072, "learning_rate": 6.055241252989663e-05, "loss": 0.0154, "step": 138210 }, { "epoch": 88.09432759719567, "grad_norm": 0.4129907488822937, "learning_rate": 6.054702504421007e-05, "loss": 0.0163, "step": 138220 }, { "epoch": 88.10070108349267, "grad_norm": 0.30221715569496155, "learning_rate": 6.054163743036719e-05, "loss": 0.0166, "step": 138230 }, { "epoch": 88.10707456978967, "grad_norm": 0.3528828024864197, "learning_rate": 6.053624968843348e-05, "loss": 0.0128, "step": 138240 }, { "epoch": 88.11344805608668, "grad_norm": 0.2371608316898346, "learning_rate": 6.053086181847437e-05, "loss": 0.0171, "step": 138250 }, { "epoch": 88.11982154238369, "grad_norm": 0.32056957483291626, "learning_rate": 6.0525473820555356e-05, "loss": 0.0169, "step": 138260 }, { "epoch": 88.12619502868068, "grad_norm": 0.36592888832092285, "learning_rate": 6.052008569474189e-05, "loss": 0.0169, "step": 138270 }, { "epoch": 88.13256851497769, "grad_norm": 0.27562564611434937, "learning_rate": 6.0514697441099465e-05, "loss": 0.0206, "step": 138280 }, { "epoch": 88.1389420012747, "grad_norm": 0.4157578945159912, "learning_rate": 6.050930905969352e-05, "loss": 0.0085, "step": 138290 }, { "epoch": 88.14531548757171, "grad_norm": 0.22583967447280884, "learning_rate": 6.050392055058955e-05, "loss": 0.0092, "step": 138300 }, { "epoch": 88.1516889738687, "grad_norm": 0.3714357912540436, "learning_rate": 6.049853191385304e-05, "loss": 0.0169, "step": 138310 }, { "epoch": 88.15806246016571, "grad_norm": 0.35761114954948425, "learning_rate": 6.049314314954946e-05, "loss": 0.0105, "step": 138320 }, { "epoch": 88.16443594646272, "grad_norm": 0.5029021501541138, "learning_rate": 6.048775425774425e-05, "loss": 0.0176, "step": 138330 }, { "epoch": 88.17080943275973, "grad_norm": 0.41381770372390747, "learning_rate": 6.0482365238502956e-05, "loss": 0.0091, "step": 138340 }, { "epoch": 88.17718291905672, "grad_norm": 0.4459671378135681, "learning_rate": 6.0476976091891025e-05, "loss": 0.0136, "step": 138350 }, { "epoch": 88.18355640535373, "grad_norm": 0.5066105723381042, "learning_rate": 6.047158681797394e-05, "loss": 0.0091, "step": 138360 }, { "epoch": 88.18992989165073, "grad_norm": 0.2561478316783905, "learning_rate": 6.046619741681718e-05, "loss": 0.0134, "step": 138370 }, { "epoch": 88.19630337794774, "grad_norm": 0.9254859685897827, "learning_rate": 6.046080788848624e-05, "loss": 0.0197, "step": 138380 }, { "epoch": 88.20267686424474, "grad_norm": 0.6464645266532898, "learning_rate": 6.0455418233046614e-05, "loss": 0.0145, "step": 138390 }, { "epoch": 88.20905035054174, "grad_norm": 0.5584646463394165, "learning_rate": 6.0450028450563776e-05, "loss": 0.0283, "step": 138400 }, { "epoch": 88.21542383683875, "grad_norm": 0.32420289516448975, "learning_rate": 6.044463854110324e-05, "loss": 0.0211, "step": 138410 }, { "epoch": 88.22179732313576, "grad_norm": 0.353691428899765, "learning_rate": 6.043924850473048e-05, "loss": 0.0077, "step": 138420 }, { "epoch": 88.22817080943275, "grad_norm": 0.32140982151031494, "learning_rate": 6.043385834151099e-05, "loss": 0.0163, "step": 138430 }, { "epoch": 88.23454429572976, "grad_norm": 0.41910308599472046, "learning_rate": 6.042846805151027e-05, "loss": 0.0129, "step": 138440 }, { "epoch": 88.24091778202677, "grad_norm": 0.27970653772354126, "learning_rate": 6.042307763479381e-05, "loss": 0.0154, "step": 138450 }, { "epoch": 88.24729126832378, "grad_norm": 0.17582742869853973, "learning_rate": 6.0417687091427124e-05, "loss": 0.0191, "step": 138460 }, { "epoch": 88.25366475462077, "grad_norm": 0.42742201685905457, "learning_rate": 6.04122964214757e-05, "loss": 0.0121, "step": 138470 }, { "epoch": 88.26003824091778, "grad_norm": 0.29840150475502014, "learning_rate": 6.040690562500505e-05, "loss": 0.0145, "step": 138480 }, { "epoch": 88.26641172721479, "grad_norm": 0.24473686516284943, "learning_rate": 6.0401514702080666e-05, "loss": 0.0181, "step": 138490 }, { "epoch": 88.2727852135118, "grad_norm": 0.45903193950653076, "learning_rate": 6.039612365276807e-05, "loss": 0.0258, "step": 138500 }, { "epoch": 88.27915869980879, "grad_norm": 0.3691752254962921, "learning_rate": 6.0390732477132736e-05, "loss": 0.0197, "step": 138510 }, { "epoch": 88.2855321861058, "grad_norm": 0.3191525638103485, "learning_rate": 6.0385341175240205e-05, "loss": 0.0252, "step": 138520 }, { "epoch": 88.2919056724028, "grad_norm": 0.45346665382385254, "learning_rate": 6.0379949747155974e-05, "loss": 0.0203, "step": 138530 }, { "epoch": 88.29827915869981, "grad_norm": 0.788385808467865, "learning_rate": 6.037455819294554e-05, "loss": 0.0188, "step": 138540 }, { "epoch": 88.3046526449968, "grad_norm": 0.4732201099395752, "learning_rate": 6.036916651267445e-05, "loss": 0.0157, "step": 138550 }, { "epoch": 88.31102613129381, "grad_norm": 0.6238333582878113, "learning_rate": 6.036377470640817e-05, "loss": 0.0175, "step": 138560 }, { "epoch": 88.31739961759082, "grad_norm": 0.3412695825099945, "learning_rate": 6.035838277421226e-05, "loss": 0.0112, "step": 138570 }, { "epoch": 88.32377310388783, "grad_norm": 0.4377163350582123, "learning_rate": 6.0352990716152224e-05, "loss": 0.0193, "step": 138580 }, { "epoch": 88.33014659018482, "grad_norm": 0.2816741466522217, "learning_rate": 6.034759853229356e-05, "loss": 0.0116, "step": 138590 }, { "epoch": 88.33652007648183, "grad_norm": 0.5502408146858215, "learning_rate": 6.034220622270181e-05, "loss": 0.0137, "step": 138600 }, { "epoch": 88.34289356277884, "grad_norm": 0.39333054423332214, "learning_rate": 6.033681378744249e-05, "loss": 0.0106, "step": 138610 }, { "epoch": 88.34926704907585, "grad_norm": 0.21571218967437744, "learning_rate": 6.033142122658113e-05, "loss": 0.0166, "step": 138620 }, { "epoch": 88.35564053537284, "grad_norm": 0.38808146119117737, "learning_rate": 6.032602854018324e-05, "loss": 0.0247, "step": 138630 }, { "epoch": 88.36201402166985, "grad_norm": 0.27111512422561646, "learning_rate": 6.032063572831437e-05, "loss": 0.0095, "step": 138640 }, { "epoch": 88.36838750796686, "grad_norm": 0.6893284320831299, "learning_rate": 6.031524279104001e-05, "loss": 0.0212, "step": 138650 }, { "epoch": 88.37476099426387, "grad_norm": 0.20324267446994781, "learning_rate": 6.030984972842571e-05, "loss": 0.0138, "step": 138660 }, { "epoch": 88.38113448056086, "grad_norm": 0.2440280318260193, "learning_rate": 6.030445654053702e-05, "loss": 0.0195, "step": 138670 }, { "epoch": 88.38750796685787, "grad_norm": 0.25892162322998047, "learning_rate": 6.0299063227439445e-05, "loss": 0.0139, "step": 138680 }, { "epoch": 88.39388145315488, "grad_norm": 0.25729477405548096, "learning_rate": 6.0293669789198524e-05, "loss": 0.0164, "step": 138690 }, { "epoch": 88.40025493945188, "grad_norm": 0.2350446879863739, "learning_rate": 6.0288276225879805e-05, "loss": 0.0085, "step": 138700 }, { "epoch": 88.40662842574889, "grad_norm": 0.2607162296772003, "learning_rate": 6.0282882537548815e-05, "loss": 0.0133, "step": 138710 }, { "epoch": 88.41300191204589, "grad_norm": 0.5814048051834106, "learning_rate": 6.027748872427108e-05, "loss": 0.0171, "step": 138720 }, { "epoch": 88.4193753983429, "grad_norm": 0.608008861541748, "learning_rate": 6.027209478611217e-05, "loss": 0.0114, "step": 138730 }, { "epoch": 88.4257488846399, "grad_norm": 0.2913368046283722, "learning_rate": 6.026670072313759e-05, "loss": 0.0093, "step": 138740 }, { "epoch": 88.43212237093691, "grad_norm": 0.45396164059638977, "learning_rate": 6.026130653541292e-05, "loss": 0.0191, "step": 138750 }, { "epoch": 88.4384958572339, "grad_norm": 0.3579753637313843, "learning_rate": 6.025591222300368e-05, "loss": 0.0133, "step": 138760 }, { "epoch": 88.44486934353091, "grad_norm": 0.3730071187019348, "learning_rate": 6.025051778597543e-05, "loss": 0.0199, "step": 138770 }, { "epoch": 88.45124282982792, "grad_norm": 0.24438969790935516, "learning_rate": 6.0245123224393705e-05, "loss": 0.0128, "step": 138780 }, { "epoch": 88.45761631612493, "grad_norm": 0.3155497610569, "learning_rate": 6.023972853832406e-05, "loss": 0.0158, "step": 138790 }, { "epoch": 88.46398980242192, "grad_norm": 0.33482393622398376, "learning_rate": 6.023433372783205e-05, "loss": 0.0143, "step": 138800 }, { "epoch": 88.47036328871893, "grad_norm": 0.3560594618320465, "learning_rate": 6.022893879298321e-05, "loss": 0.0128, "step": 138810 }, { "epoch": 88.47673677501594, "grad_norm": 0.2742050886154175, "learning_rate": 6.022354373384312e-05, "loss": 0.012, "step": 138820 }, { "epoch": 88.48311026131294, "grad_norm": 0.4026924967765808, "learning_rate": 6.0218148550477324e-05, "loss": 0.0165, "step": 138830 }, { "epoch": 88.48948374760994, "grad_norm": 0.12349523603916168, "learning_rate": 6.021275324295137e-05, "loss": 0.0185, "step": 138840 }, { "epoch": 88.49585723390695, "grad_norm": 0.2741256058216095, "learning_rate": 6.0207357811330825e-05, "loss": 0.0142, "step": 138850 }, { "epoch": 88.50223072020395, "grad_norm": 0.3824670612812042, "learning_rate": 6.020196225568123e-05, "loss": 0.0115, "step": 138860 }, { "epoch": 88.50860420650096, "grad_norm": 0.3469003140926361, "learning_rate": 6.0196566576068184e-05, "loss": 0.0145, "step": 138870 }, { "epoch": 88.51497769279796, "grad_norm": 0.33676785230636597, "learning_rate": 6.01911707725572e-05, "loss": 0.0114, "step": 138880 }, { "epoch": 88.52135117909496, "grad_norm": 0.3251839578151703, "learning_rate": 6.0185774845213884e-05, "loss": 0.0161, "step": 138890 }, { "epoch": 88.52772466539197, "grad_norm": 0.33980852365493774, "learning_rate": 6.018037879410379e-05, "loss": 0.0156, "step": 138900 }, { "epoch": 88.53409815168898, "grad_norm": 0.36193445324897766, "learning_rate": 6.0174982619292476e-05, "loss": 0.0259, "step": 138910 }, { "epoch": 88.54047163798597, "grad_norm": 0.4486226737499237, "learning_rate": 6.016958632084552e-05, "loss": 0.0167, "step": 138920 }, { "epoch": 88.54684512428298, "grad_norm": 0.39785337448120117, "learning_rate": 6.016418989882848e-05, "loss": 0.0139, "step": 138930 }, { "epoch": 88.55321861057999, "grad_norm": 0.33903101086616516, "learning_rate": 6.015879335330694e-05, "loss": 0.0316, "step": 138940 }, { "epoch": 88.559592096877, "grad_norm": 0.46979644894599915, "learning_rate": 6.0153396684346475e-05, "loss": 0.0126, "step": 138950 }, { "epoch": 88.56596558317399, "grad_norm": 0.420225590467453, "learning_rate": 6.014799989201265e-05, "loss": 0.0138, "step": 138960 }, { "epoch": 88.572339069471, "grad_norm": 0.2955191433429718, "learning_rate": 6.0142602976371044e-05, "loss": 0.0154, "step": 138970 }, { "epoch": 88.578712555768, "grad_norm": 0.34962522983551025, "learning_rate": 6.013720593748724e-05, "loss": 0.0142, "step": 138980 }, { "epoch": 88.58508604206502, "grad_norm": 0.5609009861946106, "learning_rate": 6.013180877542681e-05, "loss": 0.0183, "step": 138990 }, { "epoch": 88.59145952836201, "grad_norm": 0.5953995585441589, "learning_rate": 6.012641149025533e-05, "loss": 0.0257, "step": 139000 }, { "epoch": 88.59783301465902, "grad_norm": 0.44574853777885437, "learning_rate": 6.012101408203841e-05, "loss": 0.0167, "step": 139010 }, { "epoch": 88.60420650095602, "grad_norm": 0.4414599537849426, "learning_rate": 6.0115616550841593e-05, "loss": 0.0163, "step": 139020 }, { "epoch": 88.61057998725303, "grad_norm": 0.23228652775287628, "learning_rate": 6.01102188967305e-05, "loss": 0.0074, "step": 139030 }, { "epoch": 88.61695347355003, "grad_norm": 0.3525727093219757, "learning_rate": 6.010482111977068e-05, "loss": 0.0108, "step": 139040 }, { "epoch": 88.62332695984703, "grad_norm": 0.2600214183330536, "learning_rate": 6.0099423220027764e-05, "loss": 0.0119, "step": 139050 }, { "epoch": 88.62970044614404, "grad_norm": 0.48309341073036194, "learning_rate": 6.0094025197567304e-05, "loss": 0.018, "step": 139060 }, { "epoch": 88.63607393244105, "grad_norm": 0.38054925203323364, "learning_rate": 6.0088627052454914e-05, "loss": 0.0092, "step": 139070 }, { "epoch": 88.64244741873804, "grad_norm": 0.2867327928543091, "learning_rate": 6.0083228784756196e-05, "loss": 0.0122, "step": 139080 }, { "epoch": 88.64882090503505, "grad_norm": 0.2530948519706726, "learning_rate": 6.00778303945367e-05, "loss": 0.0125, "step": 139090 }, { "epoch": 88.65519439133206, "grad_norm": 0.4281449317932129, "learning_rate": 6.007243188186207e-05, "loss": 0.0131, "step": 139100 }, { "epoch": 88.66156787762907, "grad_norm": 0.16592271625995636, "learning_rate": 6.006703324679788e-05, "loss": 0.0132, "step": 139110 }, { "epoch": 88.66794136392606, "grad_norm": 0.5478826761245728, "learning_rate": 6.006163448940972e-05, "loss": 0.0119, "step": 139120 }, { "epoch": 88.67431485022307, "grad_norm": 0.2892785668373108, "learning_rate": 6.0056235609763215e-05, "loss": 0.021, "step": 139130 }, { "epoch": 88.68068833652008, "grad_norm": 0.47925692796707153, "learning_rate": 6.0050836607923945e-05, "loss": 0.0186, "step": 139140 }, { "epoch": 88.68706182281709, "grad_norm": 0.4407724440097809, "learning_rate": 6.004543748395752e-05, "loss": 0.0158, "step": 139150 }, { "epoch": 88.69343530911408, "grad_norm": 0.6699056625366211, "learning_rate": 6.0040038237929554e-05, "loss": 0.0185, "step": 139160 }, { "epoch": 88.69980879541109, "grad_norm": 0.2031799852848053, "learning_rate": 6.0034638869905636e-05, "loss": 0.0154, "step": 139170 }, { "epoch": 88.7061822817081, "grad_norm": 0.274306982755661, "learning_rate": 6.002923937995137e-05, "loss": 0.0208, "step": 139180 }, { "epoch": 88.7125557680051, "grad_norm": 0.35753464698791504, "learning_rate": 6.00238397681324e-05, "loss": 0.0101, "step": 139190 }, { "epoch": 88.7189292543021, "grad_norm": 0.469097375869751, "learning_rate": 6.00184400345143e-05, "loss": 0.0184, "step": 139200 }, { "epoch": 88.7253027405991, "grad_norm": 0.5874054431915283, "learning_rate": 6.0013040179162694e-05, "loss": 0.0226, "step": 139210 }, { "epoch": 88.73167622689611, "grad_norm": 0.1944512128829956, "learning_rate": 6.00076402021432e-05, "loss": 0.0192, "step": 139220 }, { "epoch": 88.73804971319312, "grad_norm": 0.18699809908866882, "learning_rate": 6.000224010352142e-05, "loss": 0.0161, "step": 139230 }, { "epoch": 88.74442319949011, "grad_norm": 0.1878189742565155, "learning_rate": 5.9996839883363e-05, "loss": 0.0233, "step": 139240 }, { "epoch": 88.75079668578712, "grad_norm": 0.2968621551990509, "learning_rate": 5.999143954173352e-05, "loss": 0.0185, "step": 139250 }, { "epoch": 88.75717017208413, "grad_norm": 0.7470718026161194, "learning_rate": 5.998603907869862e-05, "loss": 0.0166, "step": 139260 }, { "epoch": 88.76354365838114, "grad_norm": 0.44994062185287476, "learning_rate": 5.998063849432391e-05, "loss": 0.0136, "step": 139270 }, { "epoch": 88.76991714467813, "grad_norm": 0.9254646301269531, "learning_rate": 5.997523778867503e-05, "loss": 0.0164, "step": 139280 }, { "epoch": 88.77629063097514, "grad_norm": 0.30074551701545715, "learning_rate": 5.996983696181758e-05, "loss": 0.0137, "step": 139290 }, { "epoch": 88.78266411727215, "grad_norm": 0.529801070690155, "learning_rate": 5.996443601381722e-05, "loss": 0.0101, "step": 139300 }, { "epoch": 88.78903760356916, "grad_norm": 0.4674713611602783, "learning_rate": 5.995903494473953e-05, "loss": 0.0191, "step": 139310 }, { "epoch": 88.79541108986615, "grad_norm": 0.6590204834938049, "learning_rate": 5.995363375465017e-05, "loss": 0.0127, "step": 139320 }, { "epoch": 88.80178457616316, "grad_norm": 0.3950510621070862, "learning_rate": 5.9948232443614774e-05, "loss": 0.0124, "step": 139330 }, { "epoch": 88.80815806246017, "grad_norm": 0.4036603569984436, "learning_rate": 5.9942831011698944e-05, "loss": 0.0092, "step": 139340 }, { "epoch": 88.81453154875717, "grad_norm": 0.5487477779388428, "learning_rate": 5.993742945896834e-05, "loss": 0.0129, "step": 139350 }, { "epoch": 88.82090503505418, "grad_norm": 0.7304756045341492, "learning_rate": 5.993202778548858e-05, "loss": 0.0161, "step": 139360 }, { "epoch": 88.82727852135118, "grad_norm": 0.3287215828895569, "learning_rate": 5.992662599132531e-05, "loss": 0.015, "step": 139370 }, { "epoch": 88.83365200764818, "grad_norm": 0.13390077650547028, "learning_rate": 5.9921224076544155e-05, "loss": 0.0169, "step": 139380 }, { "epoch": 88.84002549394519, "grad_norm": 0.2724994421005249, "learning_rate": 5.9915822041210775e-05, "loss": 0.0163, "step": 139390 }, { "epoch": 88.8463989802422, "grad_norm": 0.465359628200531, "learning_rate": 5.991041988539079e-05, "loss": 0.0141, "step": 139400 }, { "epoch": 88.8527724665392, "grad_norm": 0.30984529852867126, "learning_rate": 5.990501760914984e-05, "loss": 0.0125, "step": 139410 }, { "epoch": 88.8591459528362, "grad_norm": 0.3201258182525635, "learning_rate": 5.989961521255357e-05, "loss": 0.0161, "step": 139420 }, { "epoch": 88.86551943913321, "grad_norm": 0.4278925359249115, "learning_rate": 5.989421269566763e-05, "loss": 0.0216, "step": 139430 }, { "epoch": 88.87189292543022, "grad_norm": 0.3441429138183594, "learning_rate": 5.9888810058557674e-05, "loss": 0.0119, "step": 139440 }, { "epoch": 88.87826641172721, "grad_norm": 0.28052857518196106, "learning_rate": 5.988340730128933e-05, "loss": 0.02, "step": 139450 }, { "epoch": 88.88463989802422, "grad_norm": 0.34180745482444763, "learning_rate": 5.9878004423928256e-05, "loss": 0.0096, "step": 139460 }, { "epoch": 88.89101338432123, "grad_norm": 0.4202592670917511, "learning_rate": 5.987260142654012e-05, "loss": 0.0118, "step": 139470 }, { "epoch": 88.89738687061823, "grad_norm": 0.7288104891777039, "learning_rate": 5.986719830919053e-05, "loss": 0.0122, "step": 139480 }, { "epoch": 88.90376035691523, "grad_norm": 0.5493438839912415, "learning_rate": 5.986179507194518e-05, "loss": 0.0156, "step": 139490 }, { "epoch": 88.91013384321224, "grad_norm": 0.17476196587085724, "learning_rate": 5.98563917148697e-05, "loss": 0.0114, "step": 139500 }, { "epoch": 88.91650732950924, "grad_norm": 0.3440963625907898, "learning_rate": 5.985098823802976e-05, "loss": 0.0179, "step": 139510 }, { "epoch": 88.92288081580625, "grad_norm": 0.3376140892505646, "learning_rate": 5.9845584641491e-05, "loss": 0.011, "step": 139520 }, { "epoch": 88.92925430210325, "grad_norm": 0.5809616446495056, "learning_rate": 5.984018092531911e-05, "loss": 0.0137, "step": 139530 }, { "epoch": 88.93562778840025, "grad_norm": 0.5299781560897827, "learning_rate": 5.983477708957972e-05, "loss": 0.0168, "step": 139540 }, { "epoch": 88.94200127469726, "grad_norm": 0.3779919743537903, "learning_rate": 5.9829373134338503e-05, "loss": 0.0104, "step": 139550 }, { "epoch": 88.94837476099427, "grad_norm": 0.27828145027160645, "learning_rate": 5.9823969059661135e-05, "loss": 0.0094, "step": 139560 }, { "epoch": 88.95474824729126, "grad_norm": 0.30722352862358093, "learning_rate": 5.981856486561325e-05, "loss": 0.0182, "step": 139570 }, { "epoch": 88.96112173358827, "grad_norm": 0.32641685009002686, "learning_rate": 5.981316055226054e-05, "loss": 0.0071, "step": 139580 }, { "epoch": 88.96749521988528, "grad_norm": 0.21725726127624512, "learning_rate": 5.980775611966866e-05, "loss": 0.0129, "step": 139590 }, { "epoch": 88.97386870618229, "grad_norm": 0.4590458273887634, "learning_rate": 5.980235156790329e-05, "loss": 0.0105, "step": 139600 }, { "epoch": 88.98024219247928, "grad_norm": 0.5501264929771423, "learning_rate": 5.979694689703008e-05, "loss": 0.0159, "step": 139610 }, { "epoch": 88.98661567877629, "grad_norm": 0.3600703179836273, "learning_rate": 5.979154210711473e-05, "loss": 0.008, "step": 139620 }, { "epoch": 88.9929891650733, "grad_norm": 0.23298178613185883, "learning_rate": 5.978613719822289e-05, "loss": 0.0133, "step": 139630 }, { "epoch": 88.9993626513703, "grad_norm": 0.16055364906787872, "learning_rate": 5.978073217042025e-05, "loss": 0.0063, "step": 139640 }, { "epoch": 89.0057361376673, "grad_norm": 0.34845900535583496, "learning_rate": 5.9775327023772465e-05, "loss": 0.0193, "step": 139650 }, { "epoch": 89.0121096239643, "grad_norm": 0.5981891751289368, "learning_rate": 5.9769921758345246e-05, "loss": 0.0142, "step": 139660 }, { "epoch": 89.01848311026131, "grad_norm": 0.22142700850963593, "learning_rate": 5.976451637420424e-05, "loss": 0.0131, "step": 139670 }, { "epoch": 89.02485659655832, "grad_norm": 0.12764482200145721, "learning_rate": 5.975911087141516e-05, "loss": 0.016, "step": 139680 }, { "epoch": 89.03123008285532, "grad_norm": 0.3739071190357208, "learning_rate": 5.975370525004365e-05, "loss": 0.0181, "step": 139690 }, { "epoch": 89.03760356915232, "grad_norm": 0.2990763783454895, "learning_rate": 5.974829951015542e-05, "loss": 0.0101, "step": 139700 }, { "epoch": 89.04397705544933, "grad_norm": 0.4330497682094574, "learning_rate": 5.974289365181613e-05, "loss": 0.0126, "step": 139710 }, { "epoch": 89.05035054174634, "grad_norm": 0.312810480594635, "learning_rate": 5.9737487675091506e-05, "loss": 0.0227, "step": 139720 }, { "epoch": 89.05672402804333, "grad_norm": 0.34370365738868713, "learning_rate": 5.973208158004719e-05, "loss": 0.0159, "step": 139730 }, { "epoch": 89.06309751434034, "grad_norm": 0.2227131724357605, "learning_rate": 5.972667536674892e-05, "loss": 0.0231, "step": 139740 }, { "epoch": 89.06947100063735, "grad_norm": 0.3727726638317108, "learning_rate": 5.972126903526235e-05, "loss": 0.0182, "step": 139750 }, { "epoch": 89.07584448693436, "grad_norm": 0.21760053932666779, "learning_rate": 5.971586258565318e-05, "loss": 0.0108, "step": 139760 }, { "epoch": 89.08221797323135, "grad_norm": 0.4540652632713318, "learning_rate": 5.971045601798711e-05, "loss": 0.0131, "step": 139770 }, { "epoch": 89.08859145952836, "grad_norm": 0.35886117815971375, "learning_rate": 5.970504933232983e-05, "loss": 0.0101, "step": 139780 }, { "epoch": 89.09496494582537, "grad_norm": 0.4414829909801483, "learning_rate": 5.9699642528747044e-05, "loss": 0.0179, "step": 139790 }, { "epoch": 89.10133843212238, "grad_norm": 0.18439535796642303, "learning_rate": 5.9694235607304426e-05, "loss": 0.0086, "step": 139800 }, { "epoch": 89.10771191841937, "grad_norm": 0.321682333946228, "learning_rate": 5.9688828568067715e-05, "loss": 0.0203, "step": 139810 }, { "epoch": 89.11408540471638, "grad_norm": 0.2979534864425659, "learning_rate": 5.968342141110257e-05, "loss": 0.0097, "step": 139820 }, { "epoch": 89.12045889101339, "grad_norm": 0.28446081280708313, "learning_rate": 5.9678014136474726e-05, "loss": 0.0115, "step": 139830 }, { "epoch": 89.1268323773104, "grad_norm": 0.4428972601890564, "learning_rate": 5.967260674424986e-05, "loss": 0.0149, "step": 139840 }, { "epoch": 89.13320586360739, "grad_norm": 0.23701967298984528, "learning_rate": 5.96671992344937e-05, "loss": 0.0154, "step": 139850 }, { "epoch": 89.1395793499044, "grad_norm": 0.17305836081504822, "learning_rate": 5.966179160727193e-05, "loss": 0.0151, "step": 139860 }, { "epoch": 89.1459528362014, "grad_norm": 0.24773304164409637, "learning_rate": 5.965638386265028e-05, "loss": 0.0127, "step": 139870 }, { "epoch": 89.15232632249841, "grad_norm": 0.6708719730377197, "learning_rate": 5.965097600069446e-05, "loss": 0.0196, "step": 139880 }, { "epoch": 89.1586998087954, "grad_norm": 0.47111719846725464, "learning_rate": 5.964556802147015e-05, "loss": 0.0174, "step": 139890 }, { "epoch": 89.16507329509241, "grad_norm": 0.1898910403251648, "learning_rate": 5.964015992504308e-05, "loss": 0.0177, "step": 139900 }, { "epoch": 89.17144678138942, "grad_norm": 0.21255114674568176, "learning_rate": 5.963475171147898e-05, "loss": 0.0084, "step": 139910 }, { "epoch": 89.17782026768643, "grad_norm": 0.2475043386220932, "learning_rate": 5.9629343380843536e-05, "loss": 0.0173, "step": 139920 }, { "epoch": 89.18419375398342, "grad_norm": 0.23626920580863953, "learning_rate": 5.9623934933202486e-05, "loss": 0.0157, "step": 139930 }, { "epoch": 89.19056724028043, "grad_norm": 0.24922537803649902, "learning_rate": 5.961852636862154e-05, "loss": 0.0226, "step": 139940 }, { "epoch": 89.19694072657744, "grad_norm": 0.46172213554382324, "learning_rate": 5.961311768716641e-05, "loss": 0.0094, "step": 139950 }, { "epoch": 89.20331421287445, "grad_norm": 0.45166119933128357, "learning_rate": 5.960770888890283e-05, "loss": 0.0185, "step": 139960 }, { "epoch": 89.20968769917145, "grad_norm": 0.663911759853363, "learning_rate": 5.960229997389652e-05, "loss": 0.0112, "step": 139970 }, { "epoch": 89.21606118546845, "grad_norm": 0.2624914348125458, "learning_rate": 5.9596890942213204e-05, "loss": 0.0102, "step": 139980 }, { "epoch": 89.22243467176546, "grad_norm": 0.8327314853668213, "learning_rate": 5.959148179391859e-05, "loss": 0.0126, "step": 139990 }, { "epoch": 89.22880815806246, "grad_norm": 0.3748314678668976, "learning_rate": 5.958607252907842e-05, "loss": 0.0115, "step": 140000 }, { "epoch": 89.23518164435947, "grad_norm": 0.5831470489501953, "learning_rate": 5.958066314775842e-05, "loss": 0.0219, "step": 140010 }, { "epoch": 89.24155513065647, "grad_norm": 0.24455757439136505, "learning_rate": 5.957525365002431e-05, "loss": 0.0165, "step": 140020 }, { "epoch": 89.24792861695347, "grad_norm": 0.35460546612739563, "learning_rate": 5.956984403594184e-05, "loss": 0.014, "step": 140030 }, { "epoch": 89.25430210325048, "grad_norm": 0.3716607689857483, "learning_rate": 5.956443430557672e-05, "loss": 0.0112, "step": 140040 }, { "epoch": 89.26067558954749, "grad_norm": 0.40056151151657104, "learning_rate": 5.9559024458994695e-05, "loss": 0.012, "step": 140050 }, { "epoch": 89.26704907584448, "grad_norm": 0.509259819984436, "learning_rate": 5.95536144962615e-05, "loss": 0.0153, "step": 140060 }, { "epoch": 89.27342256214149, "grad_norm": 0.31718045473098755, "learning_rate": 5.954820441744285e-05, "loss": 0.0145, "step": 140070 }, { "epoch": 89.2797960484385, "grad_norm": 0.3457518517971039, "learning_rate": 5.9542794222604526e-05, "loss": 0.0167, "step": 140080 }, { "epoch": 89.2861695347355, "grad_norm": 0.2683109939098358, "learning_rate": 5.953738391181223e-05, "loss": 0.011, "step": 140090 }, { "epoch": 89.2925430210325, "grad_norm": 0.224969282746315, "learning_rate": 5.953197348513172e-05, "loss": 0.0133, "step": 140100 }, { "epoch": 89.29891650732951, "grad_norm": 0.3990812599658966, "learning_rate": 5.952656294262873e-05, "loss": 0.0179, "step": 140110 }, { "epoch": 89.30528999362652, "grad_norm": 0.2900805175304413, "learning_rate": 5.952115228436901e-05, "loss": 0.0115, "step": 140120 }, { "epoch": 89.31166347992352, "grad_norm": 0.34667378664016724, "learning_rate": 5.95157415104183e-05, "loss": 0.0181, "step": 140130 }, { "epoch": 89.31803696622052, "grad_norm": 0.2615636885166168, "learning_rate": 5.9510330620842333e-05, "loss": 0.013, "step": 140140 }, { "epoch": 89.32441045251753, "grad_norm": 0.20190392434597015, "learning_rate": 5.9504919615706886e-05, "loss": 0.018, "step": 140150 }, { "epoch": 89.33078393881453, "grad_norm": 0.3678366243839264, "learning_rate": 5.949950849507768e-05, "loss": 0.0226, "step": 140160 }, { "epoch": 89.33715742511154, "grad_norm": 0.3506247103214264, "learning_rate": 5.949409725902049e-05, "loss": 0.0138, "step": 140170 }, { "epoch": 89.34353091140854, "grad_norm": 0.231230229139328, "learning_rate": 5.948868590760104e-05, "loss": 0.0101, "step": 140180 }, { "epoch": 89.34990439770554, "grad_norm": 0.4461985230445862, "learning_rate": 5.9483274440885103e-05, "loss": 0.0149, "step": 140190 }, { "epoch": 89.35627788400255, "grad_norm": 0.6397733092308044, "learning_rate": 5.947786285893844e-05, "loss": 0.0125, "step": 140200 }, { "epoch": 89.36265137029956, "grad_norm": 0.37002941966056824, "learning_rate": 5.947245116182678e-05, "loss": 0.0231, "step": 140210 }, { "epoch": 89.36902485659655, "grad_norm": 0.5077957510948181, "learning_rate": 5.9467039349615903e-05, "loss": 0.0202, "step": 140220 }, { "epoch": 89.37539834289356, "grad_norm": 0.5691283345222473, "learning_rate": 5.946162742237154e-05, "loss": 0.0241, "step": 140230 }, { "epoch": 89.38177182919057, "grad_norm": 0.3201838433742523, "learning_rate": 5.9456215380159496e-05, "loss": 0.0147, "step": 140240 }, { "epoch": 89.38814531548758, "grad_norm": 0.41356635093688965, "learning_rate": 5.945080322304549e-05, "loss": 0.0116, "step": 140250 }, { "epoch": 89.39451880178457, "grad_norm": 0.2714832127094269, "learning_rate": 5.944539095109531e-05, "loss": 0.0204, "step": 140260 }, { "epoch": 89.40089228808158, "grad_norm": 0.16562718152999878, "learning_rate": 5.943997856437471e-05, "loss": 0.0166, "step": 140270 }, { "epoch": 89.40726577437859, "grad_norm": 0.2938501536846161, "learning_rate": 5.943456606294947e-05, "loss": 0.0131, "step": 140280 }, { "epoch": 89.4136392606756, "grad_norm": 0.32314610481262207, "learning_rate": 5.9429153446885335e-05, "loss": 0.0118, "step": 140290 }, { "epoch": 89.42001274697259, "grad_norm": 0.198868066072464, "learning_rate": 5.942374071624808e-05, "loss": 0.0217, "step": 140300 }, { "epoch": 89.4263862332696, "grad_norm": 0.44732466340065, "learning_rate": 5.9418327871103476e-05, "loss": 0.0214, "step": 140310 }, { "epoch": 89.4327597195666, "grad_norm": 0.3951016366481781, "learning_rate": 5.9412914911517294e-05, "loss": 0.014, "step": 140320 }, { "epoch": 89.43913320586361, "grad_norm": 0.20579995214939117, "learning_rate": 5.940750183755531e-05, "loss": 0.0139, "step": 140330 }, { "epoch": 89.4455066921606, "grad_norm": 0.26660069823265076, "learning_rate": 5.9402088649283306e-05, "loss": 0.0167, "step": 140340 }, { "epoch": 89.45188017845761, "grad_norm": 0.28563156723976135, "learning_rate": 5.939667534676704e-05, "loss": 0.0179, "step": 140350 }, { "epoch": 89.45825366475462, "grad_norm": 0.2541399598121643, "learning_rate": 5.93912619300723e-05, "loss": 0.0168, "step": 140360 }, { "epoch": 89.46462715105163, "grad_norm": 0.22344167530536652, "learning_rate": 5.938584839926485e-05, "loss": 0.0109, "step": 140370 }, { "epoch": 89.47100063734862, "grad_norm": 0.3174265921115875, "learning_rate": 5.938043475441049e-05, "loss": 0.0154, "step": 140380 }, { "epoch": 89.47737412364563, "grad_norm": 0.45084941387176514, "learning_rate": 5.9375020995574996e-05, "loss": 0.0151, "step": 140390 }, { "epoch": 89.48374760994264, "grad_norm": 0.2869286835193634, "learning_rate": 5.9369607122824135e-05, "loss": 0.012, "step": 140400 }, { "epoch": 89.49012109623965, "grad_norm": 0.29156023263931274, "learning_rate": 5.936419313622371e-05, "loss": 0.0122, "step": 140410 }, { "epoch": 89.49649458253664, "grad_norm": 0.45514580607414246, "learning_rate": 5.935877903583948e-05, "loss": 0.0196, "step": 140420 }, { "epoch": 89.50286806883365, "grad_norm": 0.4401954710483551, "learning_rate": 5.9353364821737265e-05, "loss": 0.0195, "step": 140430 }, { "epoch": 89.50924155513066, "grad_norm": 0.3170454800128937, "learning_rate": 5.934795049398283e-05, "loss": 0.0157, "step": 140440 }, { "epoch": 89.51561504142767, "grad_norm": 0.26170599460601807, "learning_rate": 5.934253605264198e-05, "loss": 0.0148, "step": 140450 }, { "epoch": 89.52198852772466, "grad_norm": 0.27041780948638916, "learning_rate": 5.9337121497780476e-05, "loss": 0.0235, "step": 140460 }, { "epoch": 89.52836201402167, "grad_norm": 0.5133333206176758, "learning_rate": 5.933170682946416e-05, "loss": 0.0209, "step": 140470 }, { "epoch": 89.53473550031867, "grad_norm": 0.5285617709159851, "learning_rate": 5.9326292047758766e-05, "loss": 0.0123, "step": 140480 }, { "epoch": 89.54110898661568, "grad_norm": 0.3402102589607239, "learning_rate": 5.932087715273014e-05, "loss": 0.0113, "step": 140490 }, { "epoch": 89.54748247291268, "grad_norm": 0.44394609332084656, "learning_rate": 5.931546214444405e-05, "loss": 0.0173, "step": 140500 }, { "epoch": 89.55385595920968, "grad_norm": 0.22288595139980316, "learning_rate": 5.93100470229663e-05, "loss": 0.0172, "step": 140510 }, { "epoch": 89.56022944550669, "grad_norm": 0.48308584094047546, "learning_rate": 5.930463178836269e-05, "loss": 0.0159, "step": 140520 }, { "epoch": 89.5666029318037, "grad_norm": 0.24752068519592285, "learning_rate": 5.9299216440699005e-05, "loss": 0.0288, "step": 140530 }, { "epoch": 89.5729764181007, "grad_norm": 0.29382896423339844, "learning_rate": 5.929380098004107e-05, "loss": 0.0117, "step": 140540 }, { "epoch": 89.5793499043977, "grad_norm": 0.5225654244422913, "learning_rate": 5.9288385406454674e-05, "loss": 0.0122, "step": 140550 }, { "epoch": 89.58572339069471, "grad_norm": 0.30615851283073425, "learning_rate": 5.9282969720005635e-05, "loss": 0.0072, "step": 140560 }, { "epoch": 89.59209687699172, "grad_norm": 0.5407669544219971, "learning_rate": 5.927755392075973e-05, "loss": 0.0253, "step": 140570 }, { "epoch": 89.59847036328873, "grad_norm": 0.31899207830429077, "learning_rate": 5.9272138008782805e-05, "loss": 0.0171, "step": 140580 }, { "epoch": 89.60484384958572, "grad_norm": 0.22335997223854065, "learning_rate": 5.9266721984140636e-05, "loss": 0.0118, "step": 140590 }, { "epoch": 89.61121733588273, "grad_norm": 0.17497819662094116, "learning_rate": 5.9261305846899063e-05, "loss": 0.0185, "step": 140600 }, { "epoch": 89.61759082217974, "grad_norm": 0.19810721278190613, "learning_rate": 5.925588959712387e-05, "loss": 0.0147, "step": 140610 }, { "epoch": 89.62396430847674, "grad_norm": 0.5289632678031921, "learning_rate": 5.925047323488088e-05, "loss": 0.0228, "step": 140620 }, { "epoch": 89.63033779477374, "grad_norm": 0.3366800844669342, "learning_rate": 5.924505676023591e-05, "loss": 0.0121, "step": 140630 }, { "epoch": 89.63671128107075, "grad_norm": 0.3393937647342682, "learning_rate": 5.9239640173254766e-05, "loss": 0.0084, "step": 140640 }, { "epoch": 89.64308476736775, "grad_norm": 0.4404216706752777, "learning_rate": 5.923422347400327e-05, "loss": 0.014, "step": 140650 }, { "epoch": 89.64945825366476, "grad_norm": 0.4747549891471863, "learning_rate": 5.9228806662547234e-05, "loss": 0.0126, "step": 140660 }, { "epoch": 89.65583173996176, "grad_norm": 0.5650514960289001, "learning_rate": 5.9223389738952495e-05, "loss": 0.0171, "step": 140670 }, { "epoch": 89.66220522625876, "grad_norm": 0.47267434000968933, "learning_rate": 5.921797270328486e-05, "loss": 0.0123, "step": 140680 }, { "epoch": 89.66857871255577, "grad_norm": 0.4096846878528595, "learning_rate": 5.921255555561015e-05, "loss": 0.0228, "step": 140690 }, { "epoch": 89.67495219885278, "grad_norm": 0.6346778869628906, "learning_rate": 5.92071382959942e-05, "loss": 0.018, "step": 140700 }, { "epoch": 89.68132568514977, "grad_norm": 0.42007216811180115, "learning_rate": 5.920172092450281e-05, "loss": 0.0164, "step": 140710 }, { "epoch": 89.68769917144678, "grad_norm": 0.2826941907405853, "learning_rate": 5.919630344120184e-05, "loss": 0.0145, "step": 140720 }, { "epoch": 89.69407265774379, "grad_norm": 0.33132630586624146, "learning_rate": 5.919088584615708e-05, "loss": 0.0084, "step": 140730 }, { "epoch": 89.7004461440408, "grad_norm": 0.344742089509964, "learning_rate": 5.918546813943441e-05, "loss": 0.01, "step": 140740 }, { "epoch": 89.70681963033779, "grad_norm": 0.19326291978359222, "learning_rate": 5.91800503210996e-05, "loss": 0.0108, "step": 140750 }, { "epoch": 89.7131931166348, "grad_norm": 0.5279132723808289, "learning_rate": 5.917463239121853e-05, "loss": 0.0292, "step": 140760 }, { "epoch": 89.7195666029318, "grad_norm": 0.1725301593542099, "learning_rate": 5.916921434985702e-05, "loss": 0.012, "step": 140770 }, { "epoch": 89.72594008922881, "grad_norm": 0.2936982214450836, "learning_rate": 5.916379619708088e-05, "loss": 0.0092, "step": 140780 }, { "epoch": 89.73231357552581, "grad_norm": 0.2299092561006546, "learning_rate": 5.915837793295598e-05, "loss": 0.0318, "step": 140790 }, { "epoch": 89.73868706182282, "grad_norm": 0.3931572139263153, "learning_rate": 5.9152959557548117e-05, "loss": 0.0182, "step": 140800 }, { "epoch": 89.74506054811982, "grad_norm": 0.26847386360168457, "learning_rate": 5.9147541070923174e-05, "loss": 0.0095, "step": 140810 }, { "epoch": 89.75143403441683, "grad_norm": 0.5482473373413086, "learning_rate": 5.914212247314696e-05, "loss": 0.0208, "step": 140820 }, { "epoch": 89.75780752071383, "grad_norm": 0.39805102348327637, "learning_rate": 5.913670376428533e-05, "loss": 0.0146, "step": 140830 }, { "epoch": 89.76418100701083, "grad_norm": 0.570382297039032, "learning_rate": 5.913128494440413e-05, "loss": 0.0158, "step": 140840 }, { "epoch": 89.77055449330784, "grad_norm": 0.4761384427547455, "learning_rate": 5.9125866013569186e-05, "loss": 0.0184, "step": 140850 }, { "epoch": 89.77692797960485, "grad_norm": 0.5653547048568726, "learning_rate": 5.9120446971846366e-05, "loss": 0.0154, "step": 140860 }, { "epoch": 89.78330146590184, "grad_norm": 0.2734260857105255, "learning_rate": 5.911502781930148e-05, "loss": 0.0101, "step": 140870 }, { "epoch": 89.78967495219885, "grad_norm": 0.3417934775352478, "learning_rate": 5.910960855600043e-05, "loss": 0.0121, "step": 140880 }, { "epoch": 89.79604843849586, "grad_norm": 0.2784649729728699, "learning_rate": 5.9104189182009016e-05, "loss": 0.0117, "step": 140890 }, { "epoch": 89.80242192479287, "grad_norm": 0.2368330955505371, "learning_rate": 5.909876969739312e-05, "loss": 0.013, "step": 140900 }, { "epoch": 89.80879541108986, "grad_norm": 0.4651283025741577, "learning_rate": 5.909335010221857e-05, "loss": 0.0124, "step": 140910 }, { "epoch": 89.81516889738687, "grad_norm": 0.3086391091346741, "learning_rate": 5.9087930396551237e-05, "loss": 0.0108, "step": 140920 }, { "epoch": 89.82154238368388, "grad_norm": 0.329018771648407, "learning_rate": 5.908251058045697e-05, "loss": 0.0133, "step": 140930 }, { "epoch": 89.82791586998088, "grad_norm": 0.3858122229576111, "learning_rate": 5.907709065400162e-05, "loss": 0.0124, "step": 140940 }, { "epoch": 89.83428935627788, "grad_norm": 0.25778019428253174, "learning_rate": 5.907167061725105e-05, "loss": 0.0118, "step": 140950 }, { "epoch": 89.84066284257489, "grad_norm": 0.424681693315506, "learning_rate": 5.906625047027111e-05, "loss": 0.0137, "step": 140960 }, { "epoch": 89.8470363288719, "grad_norm": 0.15689992904663086, "learning_rate": 5.906083021312768e-05, "loss": 0.0093, "step": 140970 }, { "epoch": 89.8534098151689, "grad_norm": 0.5167582631111145, "learning_rate": 5.90554098458866e-05, "loss": 0.0138, "step": 140980 }, { "epoch": 89.8597833014659, "grad_norm": 0.3342542350292206, "learning_rate": 5.9049989368613745e-05, "loss": 0.0107, "step": 140990 }, { "epoch": 89.8661567877629, "grad_norm": 0.5277170538902283, "learning_rate": 5.904456878137498e-05, "loss": 0.0122, "step": 141000 }, { "epoch": 89.87253027405991, "grad_norm": 0.26743364334106445, "learning_rate": 5.903914808423614e-05, "loss": 0.0195, "step": 141010 }, { "epoch": 89.87890376035692, "grad_norm": 0.24508143961429596, "learning_rate": 5.903372727726314e-05, "loss": 0.0133, "step": 141020 }, { "epoch": 89.88527724665391, "grad_norm": 0.2685932219028473, "learning_rate": 5.9028306360521815e-05, "loss": 0.0211, "step": 141030 }, { "epoch": 89.89165073295092, "grad_norm": 0.161499485373497, "learning_rate": 5.902288533407805e-05, "loss": 0.0149, "step": 141040 }, { "epoch": 89.89802421924793, "grad_norm": 0.43055227398872375, "learning_rate": 5.9017464197997695e-05, "loss": 0.0129, "step": 141050 }, { "epoch": 89.90439770554494, "grad_norm": 0.705884575843811, "learning_rate": 5.901204295234666e-05, "loss": 0.0156, "step": 141060 }, { "epoch": 89.91077119184193, "grad_norm": 0.3250654637813568, "learning_rate": 5.900662159719077e-05, "loss": 0.0218, "step": 141070 }, { "epoch": 89.91714467813894, "grad_norm": 0.25999534130096436, "learning_rate": 5.9001200132595934e-05, "loss": 0.0188, "step": 141080 }, { "epoch": 89.92351816443595, "grad_norm": 0.4639572203159332, "learning_rate": 5.899577855862802e-05, "loss": 0.0185, "step": 141090 }, { "epoch": 89.92989165073296, "grad_norm": 0.27911245822906494, "learning_rate": 5.89903568753529e-05, "loss": 0.0145, "step": 141100 }, { "epoch": 89.93626513702995, "grad_norm": 0.46071237325668335, "learning_rate": 5.8984935082836455e-05, "loss": 0.013, "step": 141110 }, { "epoch": 89.94263862332696, "grad_norm": 0.2991214394569397, "learning_rate": 5.897951318114458e-05, "loss": 0.018, "step": 141120 }, { "epoch": 89.94901210962396, "grad_norm": 0.22761650383472443, "learning_rate": 5.8974091170343114e-05, "loss": 0.0096, "step": 141130 }, { "epoch": 89.95538559592097, "grad_norm": 0.24365517497062683, "learning_rate": 5.8968669050497995e-05, "loss": 0.0121, "step": 141140 }, { "epoch": 89.96175908221797, "grad_norm": 0.2850628197193146, "learning_rate": 5.8963246821675055e-05, "loss": 0.0223, "step": 141150 }, { "epoch": 89.96813256851497, "grad_norm": 0.5716148614883423, "learning_rate": 5.8957824483940225e-05, "loss": 0.0203, "step": 141160 }, { "epoch": 89.97450605481198, "grad_norm": 0.3009267747402191, "learning_rate": 5.895240203735936e-05, "loss": 0.0127, "step": 141170 }, { "epoch": 89.98087954110899, "grad_norm": 0.7312164306640625, "learning_rate": 5.894697948199835e-05, "loss": 0.0154, "step": 141180 }, { "epoch": 89.98725302740598, "grad_norm": 0.5479052662849426, "learning_rate": 5.8941556817923106e-05, "loss": 0.0203, "step": 141190 }, { "epoch": 89.99362651370299, "grad_norm": 0.28416159749031067, "learning_rate": 5.89361340451995e-05, "loss": 0.0177, "step": 141200 }, { "epoch": 90.0, "grad_norm": 0.8861802220344543, "learning_rate": 5.893071116389341e-05, "loss": 0.0233, "step": 141210 }, { "epoch": 90.00637348629701, "grad_norm": 0.4462067782878876, "learning_rate": 5.8925288174070783e-05, "loss": 0.0195, "step": 141220 }, { "epoch": 90.01274697259402, "grad_norm": 0.18536436557769775, "learning_rate": 5.891986507579745e-05, "loss": 0.0129, "step": 141230 }, { "epoch": 90.01912045889101, "grad_norm": 0.6052350997924805, "learning_rate": 5.891444186913935e-05, "loss": 0.0247, "step": 141240 }, { "epoch": 90.02549394518802, "grad_norm": 0.21407094597816467, "learning_rate": 5.890901855416235e-05, "loss": 0.015, "step": 141250 }, { "epoch": 90.03186743148503, "grad_norm": 0.3702368140220642, "learning_rate": 5.890359513093238e-05, "loss": 0.0085, "step": 141260 }, { "epoch": 90.03824091778203, "grad_norm": 0.3234679698944092, "learning_rate": 5.8898171599515316e-05, "loss": 0.0189, "step": 141270 }, { "epoch": 90.04461440407903, "grad_norm": 0.42784059047698975, "learning_rate": 5.889274795997706e-05, "loss": 0.0213, "step": 141280 }, { "epoch": 90.05098789037604, "grad_norm": 0.25479423999786377, "learning_rate": 5.8887324212383523e-05, "loss": 0.0136, "step": 141290 }, { "epoch": 90.05736137667304, "grad_norm": 0.37001538276672363, "learning_rate": 5.8881900356800604e-05, "loss": 0.0085, "step": 141300 }, { "epoch": 90.06373486297005, "grad_norm": 0.19190993905067444, "learning_rate": 5.887647639329421e-05, "loss": 0.0107, "step": 141310 }, { "epoch": 90.07010834926704, "grad_norm": 0.18902304768562317, "learning_rate": 5.8871052321930255e-05, "loss": 0.0192, "step": 141320 }, { "epoch": 90.07648183556405, "grad_norm": 0.17648647725582123, "learning_rate": 5.886562814277463e-05, "loss": 0.0146, "step": 141330 }, { "epoch": 90.08285532186106, "grad_norm": 0.3028877377510071, "learning_rate": 5.886020385589326e-05, "loss": 0.0105, "step": 141340 }, { "epoch": 90.08922880815807, "grad_norm": 0.18507160246372223, "learning_rate": 5.885477946135204e-05, "loss": 0.0111, "step": 141350 }, { "epoch": 90.09560229445506, "grad_norm": 0.34752485156059265, "learning_rate": 5.88493549592169e-05, "loss": 0.0107, "step": 141360 }, { "epoch": 90.10197578075207, "grad_norm": 0.2544264793395996, "learning_rate": 5.8843930349553736e-05, "loss": 0.0131, "step": 141370 }, { "epoch": 90.10834926704908, "grad_norm": 0.2853991389274597, "learning_rate": 5.8838505632428466e-05, "loss": 0.0117, "step": 141380 }, { "epoch": 90.11472275334609, "grad_norm": 0.3517368733882904, "learning_rate": 5.8833080807907e-05, "loss": 0.0124, "step": 141390 }, { "epoch": 90.12109623964308, "grad_norm": 0.25236478447914124, "learning_rate": 5.8827655876055274e-05, "loss": 0.0134, "step": 141400 }, { "epoch": 90.12746972594009, "grad_norm": 0.35362616181373596, "learning_rate": 5.88222308369392e-05, "loss": 0.0092, "step": 141410 }, { "epoch": 90.1338432122371, "grad_norm": 0.4418635666370392, "learning_rate": 5.881680569062468e-05, "loss": 0.0159, "step": 141420 }, { "epoch": 90.1402166985341, "grad_norm": 0.46173015236854553, "learning_rate": 5.8811380437177654e-05, "loss": 0.0097, "step": 141430 }, { "epoch": 90.1465901848311, "grad_norm": 0.2991622984409332, "learning_rate": 5.880595507666402e-05, "loss": 0.011, "step": 141440 }, { "epoch": 90.1529636711281, "grad_norm": 0.40630093216896057, "learning_rate": 5.8800529609149734e-05, "loss": 0.018, "step": 141450 }, { "epoch": 90.15933715742511, "grad_norm": 0.2692103981971741, "learning_rate": 5.87951040347007e-05, "loss": 0.0093, "step": 141460 }, { "epoch": 90.16571064372212, "grad_norm": 0.38832351565361023, "learning_rate": 5.878967835338285e-05, "loss": 0.0189, "step": 141470 }, { "epoch": 90.17208413001912, "grad_norm": 0.31903260946273804, "learning_rate": 5.87842525652621e-05, "loss": 0.0178, "step": 141480 }, { "epoch": 90.17845761631612, "grad_norm": 0.6249162554740906, "learning_rate": 5.87788266704044e-05, "loss": 0.0155, "step": 141490 }, { "epoch": 90.18483110261313, "grad_norm": 0.5182027816772461, "learning_rate": 5.877340066887567e-05, "loss": 0.0284, "step": 141500 }, { "epoch": 90.19120458891014, "grad_norm": 0.5540375113487244, "learning_rate": 5.876797456074182e-05, "loss": 0.0228, "step": 141510 }, { "epoch": 90.19757807520713, "grad_norm": 0.40690553188323975, "learning_rate": 5.8762548346068826e-05, "loss": 0.0124, "step": 141520 }, { "epoch": 90.20395156150414, "grad_norm": 0.6415233016014099, "learning_rate": 5.8757122024922576e-05, "loss": 0.0241, "step": 141530 }, { "epoch": 90.21032504780115, "grad_norm": 0.46244391798973083, "learning_rate": 5.8751695597369025e-05, "loss": 0.0122, "step": 141540 }, { "epoch": 90.21669853409816, "grad_norm": 0.2667366862297058, "learning_rate": 5.8746269063474114e-05, "loss": 0.0164, "step": 141550 }, { "epoch": 90.22307202039515, "grad_norm": 0.3400197923183441, "learning_rate": 5.8740842423303773e-05, "loss": 0.0093, "step": 141560 }, { "epoch": 90.22944550669216, "grad_norm": 0.32450181245803833, "learning_rate": 5.8735415676923965e-05, "loss": 0.0073, "step": 141570 }, { "epoch": 90.23581899298917, "grad_norm": 0.608775794506073, "learning_rate": 5.872998882440058e-05, "loss": 0.0257, "step": 141580 }, { "epoch": 90.24219247928617, "grad_norm": 0.6961585283279419, "learning_rate": 5.872456186579961e-05, "loss": 0.0108, "step": 141590 }, { "epoch": 90.24856596558317, "grad_norm": 0.39716508984565735, "learning_rate": 5.871913480118696e-05, "loss": 0.0148, "step": 141600 }, { "epoch": 90.25493945188018, "grad_norm": 0.4208233952522278, "learning_rate": 5.87137076306286e-05, "loss": 0.0099, "step": 141610 }, { "epoch": 90.26131293817718, "grad_norm": 0.2669081687927246, "learning_rate": 5.870828035419046e-05, "loss": 0.0096, "step": 141620 }, { "epoch": 90.26768642447419, "grad_norm": 0.34281426668167114, "learning_rate": 5.870285297193849e-05, "loss": 0.0095, "step": 141630 }, { "epoch": 90.27405991077119, "grad_norm": 0.6416864395141602, "learning_rate": 5.869742548393864e-05, "loss": 0.0177, "step": 141640 }, { "epoch": 90.2804333970682, "grad_norm": 0.3858131468296051, "learning_rate": 5.869199789025687e-05, "loss": 0.0127, "step": 141650 }, { "epoch": 90.2868068833652, "grad_norm": 0.2078344076871872, "learning_rate": 5.868657019095911e-05, "loss": 0.0087, "step": 141660 }, { "epoch": 90.29318036966221, "grad_norm": 0.2613237798213959, "learning_rate": 5.868114238611132e-05, "loss": 0.0089, "step": 141670 }, { "epoch": 90.2995538559592, "grad_norm": 0.31048911809921265, "learning_rate": 5.867571447577945e-05, "loss": 0.0115, "step": 141680 }, { "epoch": 90.30592734225621, "grad_norm": 0.5604158639907837, "learning_rate": 5.867028646002946e-05, "loss": 0.0135, "step": 141690 }, { "epoch": 90.31230082855322, "grad_norm": 0.3619856834411621, "learning_rate": 5.866485833892732e-05, "loss": 0.0168, "step": 141700 }, { "epoch": 90.31867431485023, "grad_norm": 0.26110339164733887, "learning_rate": 5.865943011253895e-05, "loss": 0.0099, "step": 141710 }, { "epoch": 90.32504780114722, "grad_norm": 0.601672887802124, "learning_rate": 5.8654001780930334e-05, "loss": 0.0083, "step": 141720 }, { "epoch": 90.33142128744423, "grad_norm": 0.4413936734199524, "learning_rate": 5.8648573344167436e-05, "loss": 0.0114, "step": 141730 }, { "epoch": 90.33779477374124, "grad_norm": 0.22006887197494507, "learning_rate": 5.864314480231619e-05, "loss": 0.0109, "step": 141740 }, { "epoch": 90.34416826003824, "grad_norm": 0.41744983196258545, "learning_rate": 5.863771615544258e-05, "loss": 0.0163, "step": 141750 }, { "epoch": 90.35054174633524, "grad_norm": 0.34062570333480835, "learning_rate": 5.8632287403612564e-05, "loss": 0.0155, "step": 141760 }, { "epoch": 90.35691523263225, "grad_norm": 0.5805906653404236, "learning_rate": 5.8626858546892106e-05, "loss": 0.013, "step": 141770 }, { "epoch": 90.36328871892925, "grad_norm": 0.5107944011688232, "learning_rate": 5.862142958534717e-05, "loss": 0.0176, "step": 141780 }, { "epoch": 90.36966220522626, "grad_norm": 0.5608083605766296, "learning_rate": 5.861600051904372e-05, "loss": 0.0163, "step": 141790 }, { "epoch": 90.37603569152326, "grad_norm": 0.2686828374862671, "learning_rate": 5.861057134804774e-05, "loss": 0.0096, "step": 141800 }, { "epoch": 90.38240917782026, "grad_norm": 0.3189029395580292, "learning_rate": 5.860514207242518e-05, "loss": 0.0163, "step": 141810 }, { "epoch": 90.38878266411727, "grad_norm": 0.28966790437698364, "learning_rate": 5.859971269224203e-05, "loss": 0.015, "step": 141820 }, { "epoch": 90.39515615041428, "grad_norm": 0.2575758397579193, "learning_rate": 5.859428320756424e-05, "loss": 0.0212, "step": 141830 }, { "epoch": 90.40152963671127, "grad_norm": 0.24366039037704468, "learning_rate": 5.8588853618457795e-05, "loss": 0.0108, "step": 141840 }, { "epoch": 90.40790312300828, "grad_norm": 0.4146634638309479, "learning_rate": 5.858342392498868e-05, "loss": 0.0174, "step": 141850 }, { "epoch": 90.41427660930529, "grad_norm": 0.33235785365104675, "learning_rate": 5.857799412722285e-05, "loss": 0.0099, "step": 141860 }, { "epoch": 90.4206500956023, "grad_norm": 0.5032908916473389, "learning_rate": 5.857256422522629e-05, "loss": 0.018, "step": 141870 }, { "epoch": 90.4270235818993, "grad_norm": 0.39025989174842834, "learning_rate": 5.856713421906498e-05, "loss": 0.0114, "step": 141880 }, { "epoch": 90.4333970681963, "grad_norm": 0.5562689900398254, "learning_rate": 5.8561704108804916e-05, "loss": 0.0109, "step": 141890 }, { "epoch": 90.43977055449331, "grad_norm": 0.9375810027122498, "learning_rate": 5.855627389451205e-05, "loss": 0.0192, "step": 141900 }, { "epoch": 90.44614404079032, "grad_norm": 0.335371732711792, "learning_rate": 5.855084357625238e-05, "loss": 0.014, "step": 141910 }, { "epoch": 90.45251752708732, "grad_norm": 0.39729705452919006, "learning_rate": 5.854541315409189e-05, "loss": 0.0101, "step": 141920 }, { "epoch": 90.45889101338432, "grad_norm": 0.3047296404838562, "learning_rate": 5.8539982628096556e-05, "loss": 0.0126, "step": 141930 }, { "epoch": 90.46526449968133, "grad_norm": 0.21051102876663208, "learning_rate": 5.853455199833238e-05, "loss": 0.0158, "step": 141940 }, { "epoch": 90.47163798597833, "grad_norm": 0.3057529926300049, "learning_rate": 5.852912126486534e-05, "loss": 0.0098, "step": 141950 }, { "epoch": 90.47801147227534, "grad_norm": 0.3566526770591736, "learning_rate": 5.85236904277614e-05, "loss": 0.0123, "step": 141960 }, { "epoch": 90.48438495857233, "grad_norm": 0.28281307220458984, "learning_rate": 5.85182594870866e-05, "loss": 0.0078, "step": 141970 }, { "epoch": 90.49075844486934, "grad_norm": 0.25493302941322327, "learning_rate": 5.851282844290691e-05, "loss": 0.0213, "step": 141980 }, { "epoch": 90.49713193116635, "grad_norm": 0.2636866867542267, "learning_rate": 5.850739729528829e-05, "loss": 0.0173, "step": 141990 }, { "epoch": 90.50350541746336, "grad_norm": 0.13784508407115936, "learning_rate": 5.8501966044296784e-05, "loss": 0.02, "step": 142000 }, { "epoch": 90.50987890376035, "grad_norm": 0.18250297009944916, "learning_rate": 5.849653468999834e-05, "loss": 0.0107, "step": 142010 }, { "epoch": 90.51625239005736, "grad_norm": 0.4454967677593231, "learning_rate": 5.8491103232459e-05, "loss": 0.0146, "step": 142020 }, { "epoch": 90.52262587635437, "grad_norm": 0.511841356754303, "learning_rate": 5.8485671671744726e-05, "loss": 0.0169, "step": 142030 }, { "epoch": 90.52899936265138, "grad_norm": 0.36742061376571655, "learning_rate": 5.848024000792154e-05, "loss": 0.0097, "step": 142040 }, { "epoch": 90.53537284894837, "grad_norm": 0.45918187499046326, "learning_rate": 5.8474808241055426e-05, "loss": 0.0194, "step": 142050 }, { "epoch": 90.54174633524538, "grad_norm": 0.32693660259246826, "learning_rate": 5.846937637121238e-05, "loss": 0.0141, "step": 142060 }, { "epoch": 90.54811982154239, "grad_norm": 0.22170774638652802, "learning_rate": 5.846394439845843e-05, "loss": 0.0139, "step": 142070 }, { "epoch": 90.5544933078394, "grad_norm": 0.23606182634830475, "learning_rate": 5.8458512322859546e-05, "loss": 0.0106, "step": 142080 }, { "epoch": 90.56086679413639, "grad_norm": 0.3069648742675781, "learning_rate": 5.8453080144481774e-05, "loss": 0.0072, "step": 142090 }, { "epoch": 90.5672402804334, "grad_norm": 0.1525389552116394, "learning_rate": 5.844764786339107e-05, "loss": 0.0118, "step": 142100 }, { "epoch": 90.5736137667304, "grad_norm": 0.22297807037830353, "learning_rate": 5.8442215479653496e-05, "loss": 0.0112, "step": 142110 }, { "epoch": 90.57998725302741, "grad_norm": 0.6187921762466431, "learning_rate": 5.843678299333501e-05, "loss": 0.0136, "step": 142120 }, { "epoch": 90.5863607393244, "grad_norm": 0.46292364597320557, "learning_rate": 5.843135040450167e-05, "loss": 0.012, "step": 142130 }, { "epoch": 90.59273422562141, "grad_norm": 0.6930348873138428, "learning_rate": 5.842591771321945e-05, "loss": 0.0134, "step": 142140 }, { "epoch": 90.59910771191842, "grad_norm": 0.2729194760322571, "learning_rate": 5.842048491955436e-05, "loss": 0.0092, "step": 142150 }, { "epoch": 90.60548119821543, "grad_norm": 0.5713133811950684, "learning_rate": 5.841505202357245e-05, "loss": 0.0192, "step": 142160 }, { "epoch": 90.61185468451242, "grad_norm": 0.3286181092262268, "learning_rate": 5.84096190253397e-05, "loss": 0.0084, "step": 142170 }, { "epoch": 90.61822817080943, "grad_norm": 0.4007602035999298, "learning_rate": 5.840418592492215e-05, "loss": 0.0174, "step": 142180 }, { "epoch": 90.62460165710644, "grad_norm": 0.25987479090690613, "learning_rate": 5.83987527223858e-05, "loss": 0.0112, "step": 142190 }, { "epoch": 90.63097514340345, "grad_norm": 0.42849498987197876, "learning_rate": 5.8393319417796664e-05, "loss": 0.0111, "step": 142200 }, { "epoch": 90.63734862970044, "grad_norm": 0.41704168915748596, "learning_rate": 5.838788601122078e-05, "loss": 0.0122, "step": 142210 }, { "epoch": 90.64372211599745, "grad_norm": 0.2689794898033142, "learning_rate": 5.838245250272417e-05, "loss": 0.0143, "step": 142220 }, { "epoch": 90.65009560229446, "grad_norm": 0.2969506084918976, "learning_rate": 5.837701889237285e-05, "loss": 0.0143, "step": 142230 }, { "epoch": 90.65646908859146, "grad_norm": 0.23657868802547455, "learning_rate": 5.8371585180232823e-05, "loss": 0.014, "step": 142240 }, { "epoch": 90.66284257488846, "grad_norm": 0.36705294251441956, "learning_rate": 5.8366151366370145e-05, "loss": 0.0126, "step": 142250 }, { "epoch": 90.66921606118547, "grad_norm": 0.48882240056991577, "learning_rate": 5.836071745085082e-05, "loss": 0.0148, "step": 142260 }, { "epoch": 90.67558954748247, "grad_norm": 0.5141257047653198, "learning_rate": 5.8355283433740893e-05, "loss": 0.0154, "step": 142270 }, { "epoch": 90.68196303377948, "grad_norm": 0.3155660331249237, "learning_rate": 5.8349849315106376e-05, "loss": 0.0131, "step": 142280 }, { "epoch": 90.68833652007648, "grad_norm": 0.26177123188972473, "learning_rate": 5.8344415095013316e-05, "loss": 0.0108, "step": 142290 }, { "epoch": 90.69471000637348, "grad_norm": 0.17189887166023254, "learning_rate": 5.833898077352773e-05, "loss": 0.0093, "step": 142300 }, { "epoch": 90.70108349267049, "grad_norm": 0.1409689486026764, "learning_rate": 5.833354635071565e-05, "loss": 0.0205, "step": 142310 }, { "epoch": 90.7074569789675, "grad_norm": 0.23328454792499542, "learning_rate": 5.8328111826643114e-05, "loss": 0.0277, "step": 142320 }, { "epoch": 90.7138304652645, "grad_norm": 0.3241676986217499, "learning_rate": 5.832267720137616e-05, "loss": 0.0128, "step": 142330 }, { "epoch": 90.7202039515615, "grad_norm": 0.2147231549024582, "learning_rate": 5.8317242474980826e-05, "loss": 0.0117, "step": 142340 }, { "epoch": 90.72657743785851, "grad_norm": 0.2729504406452179, "learning_rate": 5.831180764752313e-05, "loss": 0.0195, "step": 142350 }, { "epoch": 90.73295092415552, "grad_norm": 0.2550337314605713, "learning_rate": 5.830637271906914e-05, "loss": 0.0126, "step": 142360 }, { "epoch": 90.73932441045251, "grad_norm": 0.3214569091796875, "learning_rate": 5.830093768968485e-05, "loss": 0.012, "step": 142370 }, { "epoch": 90.74569789674952, "grad_norm": 0.22962884604930878, "learning_rate": 5.829550255943636e-05, "loss": 0.0094, "step": 142380 }, { "epoch": 90.75207138304653, "grad_norm": 0.39214545488357544, "learning_rate": 5.829006732838966e-05, "loss": 0.0129, "step": 142390 }, { "epoch": 90.75844486934353, "grad_norm": 0.5053321123123169, "learning_rate": 5.8284631996610827e-05, "loss": 0.0144, "step": 142400 }, { "epoch": 90.76481835564053, "grad_norm": 0.45925453305244446, "learning_rate": 5.827919656416589e-05, "loss": 0.0238, "step": 142410 }, { "epoch": 90.77119184193754, "grad_norm": 0.5793464183807373, "learning_rate": 5.8273761031120896e-05, "loss": 0.015, "step": 142420 }, { "epoch": 90.77756532823454, "grad_norm": 0.37298378348350525, "learning_rate": 5.82683253975419e-05, "loss": 0.0117, "step": 142430 }, { "epoch": 90.78393881453155, "grad_norm": 0.2786058187484741, "learning_rate": 5.826288966349493e-05, "loss": 0.0106, "step": 142440 }, { "epoch": 90.79031230082855, "grad_norm": 0.21934247016906738, "learning_rate": 5.8257453829046074e-05, "loss": 0.011, "step": 142450 }, { "epoch": 90.79668578712555, "grad_norm": 0.39184579253196716, "learning_rate": 5.825201789426135e-05, "loss": 0.0191, "step": 142460 }, { "epoch": 90.80305927342256, "grad_norm": 0.20093168318271637, "learning_rate": 5.8246581859206796e-05, "loss": 0.0199, "step": 142470 }, { "epoch": 90.80943275971957, "grad_norm": 0.3172554671764374, "learning_rate": 5.8241145723948506e-05, "loss": 0.0226, "step": 142480 }, { "epoch": 90.81580624601658, "grad_norm": 0.41148945689201355, "learning_rate": 5.8235709488552504e-05, "loss": 0.0127, "step": 142490 }, { "epoch": 90.82217973231357, "grad_norm": 0.28104740381240845, "learning_rate": 5.823027315308486e-05, "loss": 0.0119, "step": 142500 }, { "epoch": 90.82855321861058, "grad_norm": 0.5401007533073425, "learning_rate": 5.822483671761163e-05, "loss": 0.0134, "step": 142510 }, { "epoch": 90.83492670490759, "grad_norm": 0.36608394980430603, "learning_rate": 5.821940018219886e-05, "loss": 0.0108, "step": 142520 }, { "epoch": 90.8413001912046, "grad_norm": 0.5117450952529907, "learning_rate": 5.8213963546912634e-05, "loss": 0.0162, "step": 142530 }, { "epoch": 90.84767367750159, "grad_norm": 0.4670466184616089, "learning_rate": 5.820852681181898e-05, "loss": 0.0133, "step": 142540 }, { "epoch": 90.8540471637986, "grad_norm": 0.3224651515483856, "learning_rate": 5.8203089976983985e-05, "loss": 0.0115, "step": 142550 }, { "epoch": 90.8604206500956, "grad_norm": 0.3221488296985626, "learning_rate": 5.8197653042473686e-05, "loss": 0.0115, "step": 142560 }, { "epoch": 90.86679413639261, "grad_norm": 0.33710774779319763, "learning_rate": 5.819221600835418e-05, "loss": 0.0137, "step": 142570 }, { "epoch": 90.8731676226896, "grad_norm": 0.40489786863327026, "learning_rate": 5.8186778874691504e-05, "loss": 0.0247, "step": 142580 }, { "epoch": 90.87954110898661, "grad_norm": 0.15585626661777496, "learning_rate": 5.8181341641551736e-05, "loss": 0.024, "step": 142590 }, { "epoch": 90.88591459528362, "grad_norm": 0.47592276334762573, "learning_rate": 5.817590430900094e-05, "loss": 0.016, "step": 142600 }, { "epoch": 90.89228808158063, "grad_norm": 0.27935731410980225, "learning_rate": 5.81704668771052e-05, "loss": 0.0124, "step": 142610 }, { "epoch": 90.89866156787762, "grad_norm": 0.3360534608364105, "learning_rate": 5.8165029345930575e-05, "loss": 0.0147, "step": 142620 }, { "epoch": 90.90503505417463, "grad_norm": 0.44825997948646545, "learning_rate": 5.815959171554312e-05, "loss": 0.0128, "step": 142630 }, { "epoch": 90.91140854047164, "grad_norm": 0.4470396041870117, "learning_rate": 5.815415398600893e-05, "loss": 0.0192, "step": 142640 }, { "epoch": 90.91778202676865, "grad_norm": 0.17037133872509003, "learning_rate": 5.814871615739407e-05, "loss": 0.0132, "step": 142650 }, { "epoch": 90.92415551306564, "grad_norm": 0.5636090636253357, "learning_rate": 5.814327822976461e-05, "loss": 0.0126, "step": 142660 }, { "epoch": 90.93052899936265, "grad_norm": 0.36405283212661743, "learning_rate": 5.8137840203186636e-05, "loss": 0.0096, "step": 142670 }, { "epoch": 90.93690248565966, "grad_norm": 0.2567422091960907, "learning_rate": 5.8132402077726215e-05, "loss": 0.0128, "step": 142680 }, { "epoch": 90.94327597195667, "grad_norm": 0.31094878911972046, "learning_rate": 5.8126963853449425e-05, "loss": 0.0123, "step": 142690 }, { "epoch": 90.94964945825366, "grad_norm": 0.4025634229183197, "learning_rate": 5.812152553042237e-05, "loss": 0.0199, "step": 142700 }, { "epoch": 90.95602294455067, "grad_norm": 0.5137103796005249, "learning_rate": 5.8116087108711104e-05, "loss": 0.0264, "step": 142710 }, { "epoch": 90.96239643084768, "grad_norm": 0.48208653926849365, "learning_rate": 5.811064858838171e-05, "loss": 0.0102, "step": 142720 }, { "epoch": 90.96876991714468, "grad_norm": 0.4193313419818878, "learning_rate": 5.810520996950028e-05, "loss": 0.0271, "step": 142730 }, { "epoch": 90.97514340344168, "grad_norm": 0.3766099512577057, "learning_rate": 5.809977125213289e-05, "loss": 0.0166, "step": 142740 }, { "epoch": 90.98151688973869, "grad_norm": 0.23898516595363617, "learning_rate": 5.809433243634565e-05, "loss": 0.0138, "step": 142750 }, { "epoch": 90.9878903760357, "grad_norm": 0.6055013537406921, "learning_rate": 5.8088893522204614e-05, "loss": 0.0221, "step": 142760 }, { "epoch": 90.9942638623327, "grad_norm": 0.3679284155368805, "learning_rate": 5.808345450977589e-05, "loss": 0.0161, "step": 142770 }, { "epoch": 91.0006373486297, "grad_norm": 0.35362154245376587, "learning_rate": 5.8078015399125573e-05, "loss": 0.0123, "step": 142780 }, { "epoch": 91.0070108349267, "grad_norm": 0.2732907235622406, "learning_rate": 5.8072576190319716e-05, "loss": 0.0122, "step": 142790 }, { "epoch": 91.01338432122371, "grad_norm": 0.5536474585533142, "learning_rate": 5.806713688342447e-05, "loss": 0.0158, "step": 142800 }, { "epoch": 91.01975780752072, "grad_norm": 0.6747725605964661, "learning_rate": 5.806169747850586e-05, "loss": 0.0151, "step": 142810 }, { "epoch": 91.02613129381771, "grad_norm": 0.5221205949783325, "learning_rate": 5.8056257975630036e-05, "loss": 0.0175, "step": 142820 }, { "epoch": 91.03250478011472, "grad_norm": 0.5564306378364563, "learning_rate": 5.8050818374863056e-05, "loss": 0.0253, "step": 142830 }, { "epoch": 91.03887826641173, "grad_norm": 0.36881595849990845, "learning_rate": 5.804537867627104e-05, "loss": 0.0124, "step": 142840 }, { "epoch": 91.04525175270874, "grad_norm": 0.16863568127155304, "learning_rate": 5.8039938879920076e-05, "loss": 0.0145, "step": 142850 }, { "epoch": 91.05162523900573, "grad_norm": 0.5783346891403198, "learning_rate": 5.803449898587627e-05, "loss": 0.0152, "step": 142860 }, { "epoch": 91.05799872530274, "grad_norm": 0.2659682333469391, "learning_rate": 5.802905899420571e-05, "loss": 0.013, "step": 142870 }, { "epoch": 91.06437221159975, "grad_norm": 0.4924461245536804, "learning_rate": 5.8023618904974494e-05, "loss": 0.0135, "step": 142880 }, { "epoch": 91.07074569789675, "grad_norm": 0.28429359197616577, "learning_rate": 5.801817871824875e-05, "loss": 0.0119, "step": 142890 }, { "epoch": 91.07711918419375, "grad_norm": 0.543844997882843, "learning_rate": 5.801273843409455e-05, "loss": 0.0135, "step": 142900 }, { "epoch": 91.08349267049076, "grad_norm": 0.4478318691253662, "learning_rate": 5.800729805257803e-05, "loss": 0.0173, "step": 142910 }, { "epoch": 91.08986615678776, "grad_norm": 0.33464542031288147, "learning_rate": 5.800185757376525e-05, "loss": 0.0135, "step": 142920 }, { "epoch": 91.09623964308477, "grad_norm": 0.24073347449302673, "learning_rate": 5.799641699772237e-05, "loss": 0.0148, "step": 142930 }, { "epoch": 91.10261312938177, "grad_norm": 0.5524978041648865, "learning_rate": 5.799097632451547e-05, "loss": 0.019, "step": 142940 }, { "epoch": 91.10898661567877, "grad_norm": 0.47264567017555237, "learning_rate": 5.798553555421066e-05, "loss": 0.0129, "step": 142950 }, { "epoch": 91.11536010197578, "grad_norm": 0.27267247438430786, "learning_rate": 5.7980094686874054e-05, "loss": 0.0156, "step": 142960 }, { "epoch": 91.12173358827279, "grad_norm": 0.40088367462158203, "learning_rate": 5.7974653722571746e-05, "loss": 0.0115, "step": 142970 }, { "epoch": 91.12810707456978, "grad_norm": 0.4748351275920868, "learning_rate": 5.7969212661369885e-05, "loss": 0.0135, "step": 142980 }, { "epoch": 91.13448056086679, "grad_norm": 0.3378336727619171, "learning_rate": 5.796377150333456e-05, "loss": 0.0087, "step": 142990 }, { "epoch": 91.1408540471638, "grad_norm": 0.27643996477127075, "learning_rate": 5.79583302485319e-05, "loss": 0.0117, "step": 143000 }, { "epoch": 91.1472275334608, "grad_norm": 0.2982424199581146, "learning_rate": 5.7952888897028e-05, "loss": 0.0078, "step": 143010 }, { "epoch": 91.1536010197578, "grad_norm": 0.14035022258758545, "learning_rate": 5.7947447448889e-05, "loss": 0.01, "step": 143020 }, { "epoch": 91.15997450605481, "grad_norm": 0.3021611273288727, "learning_rate": 5.7942005904181016e-05, "loss": 0.0193, "step": 143030 }, { "epoch": 91.16634799235182, "grad_norm": 0.3575431704521179, "learning_rate": 5.793656426297015e-05, "loss": 0.0106, "step": 143040 }, { "epoch": 91.17272147864882, "grad_norm": 0.8330222368240356, "learning_rate": 5.793112252532255e-05, "loss": 0.025, "step": 143050 }, { "epoch": 91.17909496494582, "grad_norm": 0.3998124897480011, "learning_rate": 5.7925680691304305e-05, "loss": 0.015, "step": 143060 }, { "epoch": 91.18546845124283, "grad_norm": 0.4637693464756012, "learning_rate": 5.792023876098157e-05, "loss": 0.014, "step": 143070 }, { "epoch": 91.19184193753983, "grad_norm": 0.2698320150375366, "learning_rate": 5.791479673442044e-05, "loss": 0.0073, "step": 143080 }, { "epoch": 91.19821542383684, "grad_norm": 0.3761422038078308, "learning_rate": 5.790935461168707e-05, "loss": 0.0212, "step": 143090 }, { "epoch": 91.20458891013385, "grad_norm": 0.40735912322998047, "learning_rate": 5.790391239284758e-05, "loss": 0.0108, "step": 143100 }, { "epoch": 91.21096239643084, "grad_norm": 0.43654683232307434, "learning_rate": 5.789847007796807e-05, "loss": 0.0228, "step": 143110 }, { "epoch": 91.21733588272785, "grad_norm": 0.22735877335071564, "learning_rate": 5.789302766711471e-05, "loss": 0.0099, "step": 143120 }, { "epoch": 91.22370936902486, "grad_norm": 0.3028746247291565, "learning_rate": 5.78875851603536e-05, "loss": 0.0129, "step": 143130 }, { "epoch": 91.23008285532187, "grad_norm": 0.7635648250579834, "learning_rate": 5.7882142557750895e-05, "loss": 0.0131, "step": 143140 }, { "epoch": 91.23645634161886, "grad_norm": 0.26345154643058777, "learning_rate": 5.7876699859372704e-05, "loss": 0.0102, "step": 143150 }, { "epoch": 91.24282982791587, "grad_norm": 0.531442403793335, "learning_rate": 5.787125706528518e-05, "loss": 0.0156, "step": 143160 }, { "epoch": 91.24920331421288, "grad_norm": 0.36445334553718567, "learning_rate": 5.7865814175554445e-05, "loss": 0.0099, "step": 143170 }, { "epoch": 91.25557680050989, "grad_norm": 0.3424485921859741, "learning_rate": 5.7860371190246656e-05, "loss": 0.0121, "step": 143180 }, { "epoch": 91.26195028680688, "grad_norm": 0.5669740438461304, "learning_rate": 5.785492810942792e-05, "loss": 0.0185, "step": 143190 }, { "epoch": 91.26832377310389, "grad_norm": 0.5091249942779541, "learning_rate": 5.78494849331644e-05, "loss": 0.0114, "step": 143200 }, { "epoch": 91.2746972594009, "grad_norm": 0.38268426060676575, "learning_rate": 5.784404166152222e-05, "loss": 0.0082, "step": 143210 }, { "epoch": 91.2810707456979, "grad_norm": 0.21912887692451477, "learning_rate": 5.7838598294567525e-05, "loss": 0.0207, "step": 143220 }, { "epoch": 91.2874442319949, "grad_norm": 0.47931501269340515, "learning_rate": 5.7833154832366465e-05, "loss": 0.0169, "step": 143230 }, { "epoch": 91.2938177182919, "grad_norm": 0.47381630539894104, "learning_rate": 5.7827711274985186e-05, "loss": 0.015, "step": 143240 }, { "epoch": 91.30019120458891, "grad_norm": 0.2243773490190506, "learning_rate": 5.782226762248981e-05, "loss": 0.0126, "step": 143250 }, { "epoch": 91.30656469088592, "grad_norm": 0.3762310743331909, "learning_rate": 5.781682387494651e-05, "loss": 0.0156, "step": 143260 }, { "epoch": 91.31293817718291, "grad_norm": 0.32886582612991333, "learning_rate": 5.7811380032421405e-05, "loss": 0.0171, "step": 143270 }, { "epoch": 91.31931166347992, "grad_norm": 0.3300354480743408, "learning_rate": 5.780593609498067e-05, "loss": 0.011, "step": 143280 }, { "epoch": 91.32568514977693, "grad_norm": 0.44899001717567444, "learning_rate": 5.7800492062690435e-05, "loss": 0.016, "step": 143290 }, { "epoch": 91.33205863607394, "grad_norm": 0.6902531385421753, "learning_rate": 5.7795047935616866e-05, "loss": 0.0218, "step": 143300 }, { "epoch": 91.33843212237093, "grad_norm": 0.39714664220809937, "learning_rate": 5.778960371382609e-05, "loss": 0.0129, "step": 143310 }, { "epoch": 91.34480560866794, "grad_norm": 0.5821247100830078, "learning_rate": 5.7784159397384283e-05, "loss": 0.0082, "step": 143320 }, { "epoch": 91.35117909496495, "grad_norm": 0.5112342834472656, "learning_rate": 5.777871498635758e-05, "loss": 0.0271, "step": 143330 }, { "epoch": 91.35755258126196, "grad_norm": 0.23704199492931366, "learning_rate": 5.777327048081217e-05, "loss": 0.0189, "step": 143340 }, { "epoch": 91.36392606755895, "grad_norm": 0.3783239722251892, "learning_rate": 5.7767825880814174e-05, "loss": 0.0175, "step": 143350 }, { "epoch": 91.37029955385596, "grad_norm": 0.294452041387558, "learning_rate": 5.7762381186429745e-05, "loss": 0.0219, "step": 143360 }, { "epoch": 91.37667304015297, "grad_norm": 0.3146178126335144, "learning_rate": 5.775693639772507e-05, "loss": 0.012, "step": 143370 }, { "epoch": 91.38304652644997, "grad_norm": 0.20632924139499664, "learning_rate": 5.775149151476629e-05, "loss": 0.0121, "step": 143380 }, { "epoch": 91.38942001274697, "grad_norm": 0.4169023036956787, "learning_rate": 5.774604653761957e-05, "loss": 0.014, "step": 143390 }, { "epoch": 91.39579349904398, "grad_norm": 0.32090768218040466, "learning_rate": 5.774060146635106e-05, "loss": 0.0291, "step": 143400 }, { "epoch": 91.40216698534098, "grad_norm": 0.26907211542129517, "learning_rate": 5.7735156301026946e-05, "loss": 0.0123, "step": 143410 }, { "epoch": 91.40854047163799, "grad_norm": 0.30807265639305115, "learning_rate": 5.772971104171337e-05, "loss": 0.026, "step": 143420 }, { "epoch": 91.41491395793498, "grad_norm": 0.1753760725259781, "learning_rate": 5.772426568847653e-05, "loss": 0.0114, "step": 143430 }, { "epoch": 91.421287444232, "grad_norm": 0.23186221718788147, "learning_rate": 5.771882024138255e-05, "loss": 0.0139, "step": 143440 }, { "epoch": 91.427660930529, "grad_norm": 0.23614908754825592, "learning_rate": 5.771337470049763e-05, "loss": 0.0118, "step": 143450 }, { "epoch": 91.43403441682601, "grad_norm": 0.5569922924041748, "learning_rate": 5.770792906588791e-05, "loss": 0.0165, "step": 143460 }, { "epoch": 91.440407903123, "grad_norm": 0.19641266763210297, "learning_rate": 5.770248333761957e-05, "loss": 0.0111, "step": 143470 }, { "epoch": 91.44678138942001, "grad_norm": 0.2690991461277008, "learning_rate": 5.76970375157588e-05, "loss": 0.0113, "step": 143480 }, { "epoch": 91.45315487571702, "grad_norm": 0.24708661437034607, "learning_rate": 5.769159160037174e-05, "loss": 0.0152, "step": 143490 }, { "epoch": 91.45952836201403, "grad_norm": 0.2555195093154907, "learning_rate": 5.76861455915246e-05, "loss": 0.0169, "step": 143500 }, { "epoch": 91.46590184831102, "grad_norm": 0.3025877773761749, "learning_rate": 5.768069948928353e-05, "loss": 0.0097, "step": 143510 }, { "epoch": 91.47227533460803, "grad_norm": 0.4445381462574005, "learning_rate": 5.76752532937147e-05, "loss": 0.0099, "step": 143520 }, { "epoch": 91.47864882090504, "grad_norm": 0.428090900182724, "learning_rate": 5.7669807004884304e-05, "loss": 0.02, "step": 143530 }, { "epoch": 91.48502230720204, "grad_norm": 0.23505356907844543, "learning_rate": 5.76643606228585e-05, "loss": 0.0118, "step": 143540 }, { "epoch": 91.49139579349904, "grad_norm": 0.15751467645168304, "learning_rate": 5.7658914147703494e-05, "loss": 0.0171, "step": 143550 }, { "epoch": 91.49776927979605, "grad_norm": 0.173712357878685, "learning_rate": 5.765346757948543e-05, "loss": 0.0173, "step": 143560 }, { "epoch": 91.50414276609305, "grad_norm": 0.31655415892601013, "learning_rate": 5.764802091827053e-05, "loss": 0.0198, "step": 143570 }, { "epoch": 91.51051625239006, "grad_norm": 0.19907057285308838, "learning_rate": 5.764257416412494e-05, "loss": 0.0127, "step": 143580 }, { "epoch": 91.51688973868706, "grad_norm": 0.26132693886756897, "learning_rate": 5.763712731711487e-05, "loss": 0.0073, "step": 143590 }, { "epoch": 91.52326322498406, "grad_norm": 0.41726207733154297, "learning_rate": 5.763168037730649e-05, "loss": 0.0124, "step": 143600 }, { "epoch": 91.52963671128107, "grad_norm": 0.5490280985832214, "learning_rate": 5.762623334476598e-05, "loss": 0.0142, "step": 143610 }, { "epoch": 91.53601019757808, "grad_norm": 0.33242860436439514, "learning_rate": 5.762078621955954e-05, "loss": 0.02, "step": 143620 }, { "epoch": 91.54238368387507, "grad_norm": 0.34122318029403687, "learning_rate": 5.761533900175334e-05, "loss": 0.0143, "step": 143630 }, { "epoch": 91.54875717017208, "grad_norm": 0.6027663946151733, "learning_rate": 5.7609891691413595e-05, "loss": 0.011, "step": 143640 }, { "epoch": 91.55513065646909, "grad_norm": 0.3250682055950165, "learning_rate": 5.7604444288606474e-05, "loss": 0.0126, "step": 143650 }, { "epoch": 91.5615041427661, "grad_norm": 0.4440905749797821, "learning_rate": 5.759899679339819e-05, "loss": 0.0146, "step": 143660 }, { "epoch": 91.56787762906309, "grad_norm": 0.5600935816764832, "learning_rate": 5.7593549205854916e-05, "loss": 0.0208, "step": 143670 }, { "epoch": 91.5742511153601, "grad_norm": 0.25952187180519104, "learning_rate": 5.758810152604284e-05, "loss": 0.0098, "step": 143680 }, { "epoch": 91.5806246016571, "grad_norm": 0.4033315181732178, "learning_rate": 5.758265375402817e-05, "loss": 0.0169, "step": 143690 }, { "epoch": 91.58699808795411, "grad_norm": 0.3822246789932251, "learning_rate": 5.757720588987708e-05, "loss": 0.0139, "step": 143700 }, { "epoch": 91.59337157425111, "grad_norm": 0.27464914321899414, "learning_rate": 5.7571757933655814e-05, "loss": 0.0153, "step": 143710 }, { "epoch": 91.59974506054812, "grad_norm": 0.334176629781723, "learning_rate": 5.7566309885430526e-05, "loss": 0.0169, "step": 143720 }, { "epoch": 91.60611854684512, "grad_norm": 0.43456992506980896, "learning_rate": 5.756086174526743e-05, "loss": 0.0189, "step": 143730 }, { "epoch": 91.61249203314213, "grad_norm": 0.3372412919998169, "learning_rate": 5.755541351323272e-05, "loss": 0.0096, "step": 143740 }, { "epoch": 91.61886551943914, "grad_norm": 0.17508728802204132, "learning_rate": 5.754996518939262e-05, "loss": 0.017, "step": 143750 }, { "epoch": 91.62523900573613, "grad_norm": 0.1819859892129898, "learning_rate": 5.754451677381329e-05, "loss": 0.0259, "step": 143760 }, { "epoch": 91.63161249203314, "grad_norm": 0.34899598360061646, "learning_rate": 5.753906826656097e-05, "loss": 0.0182, "step": 143770 }, { "epoch": 91.63798597833015, "grad_norm": 0.21936041116714478, "learning_rate": 5.753361966770187e-05, "loss": 0.0069, "step": 143780 }, { "epoch": 91.64435946462716, "grad_norm": 0.550836980342865, "learning_rate": 5.752817097730214e-05, "loss": 0.0211, "step": 143790 }, { "epoch": 91.65073295092415, "grad_norm": 0.31838512420654297, "learning_rate": 5.7522722195428056e-05, "loss": 0.0115, "step": 143800 }, { "epoch": 91.65710643722116, "grad_norm": 0.15074428915977478, "learning_rate": 5.7517273322145785e-05, "loss": 0.0088, "step": 143810 }, { "epoch": 91.66347992351817, "grad_norm": 0.22498367726802826, "learning_rate": 5.751182435752155e-05, "loss": 0.022, "step": 143820 }, { "epoch": 91.66985340981518, "grad_norm": 0.21611127257347107, "learning_rate": 5.750637530162155e-05, "loss": 0.0095, "step": 143830 }, { "epoch": 91.67622689611217, "grad_norm": 0.31902602314949036, "learning_rate": 5.7500926154512004e-05, "loss": 0.0114, "step": 143840 }, { "epoch": 91.68260038240918, "grad_norm": 0.1984030306339264, "learning_rate": 5.749547691625913e-05, "loss": 0.0077, "step": 143850 }, { "epoch": 91.68897386870619, "grad_norm": 0.6419076323509216, "learning_rate": 5.749002758692912e-05, "loss": 0.0119, "step": 143860 }, { "epoch": 91.6953473550032, "grad_norm": 0.34690120816230774, "learning_rate": 5.748457816658821e-05, "loss": 0.0125, "step": 143870 }, { "epoch": 91.70172084130019, "grad_norm": 0.23911719024181366, "learning_rate": 5.747912865530261e-05, "loss": 0.0135, "step": 143880 }, { "epoch": 91.7080943275972, "grad_norm": 0.271011620759964, "learning_rate": 5.747367905313854e-05, "loss": 0.0083, "step": 143890 }, { "epoch": 91.7144678138942, "grad_norm": 0.47997817397117615, "learning_rate": 5.746822936016221e-05, "loss": 0.0149, "step": 143900 }, { "epoch": 91.72084130019121, "grad_norm": 0.5543095469474792, "learning_rate": 5.746277957643984e-05, "loss": 0.0138, "step": 143910 }, { "epoch": 91.7272147864882, "grad_norm": 0.18243597447872162, "learning_rate": 5.7457329702037646e-05, "loss": 0.013, "step": 143920 }, { "epoch": 91.73358827278521, "grad_norm": 0.2860872149467468, "learning_rate": 5.745187973702186e-05, "loss": 0.0146, "step": 143930 }, { "epoch": 91.73996175908222, "grad_norm": 0.23381422460079193, "learning_rate": 5.74464296814587e-05, "loss": 0.0203, "step": 143940 }, { "epoch": 91.74633524537923, "grad_norm": 0.38907214999198914, "learning_rate": 5.7440979535414395e-05, "loss": 0.0117, "step": 143950 }, { "epoch": 91.75270873167622, "grad_norm": 0.24458523094654083, "learning_rate": 5.743552929895515e-05, "loss": 0.0175, "step": 143960 }, { "epoch": 91.75908221797323, "grad_norm": 0.4146716892719269, "learning_rate": 5.7430078972147226e-05, "loss": 0.011, "step": 143970 }, { "epoch": 91.76545570427024, "grad_norm": 0.2958703339099884, "learning_rate": 5.74246285550568e-05, "loss": 0.0145, "step": 143980 }, { "epoch": 91.77182919056725, "grad_norm": 0.24257995188236237, "learning_rate": 5.741917804775014e-05, "loss": 0.0168, "step": 143990 }, { "epoch": 91.77820267686424, "grad_norm": 0.2928883731365204, "learning_rate": 5.7413727450293465e-05, "loss": 0.0161, "step": 144000 }, { "epoch": 91.78457616316125, "grad_norm": 1.1435242891311646, "learning_rate": 5.7408276762753e-05, "loss": 0.0093, "step": 144010 }, { "epoch": 91.79094964945826, "grad_norm": 0.420482873916626, "learning_rate": 5.740282598519497e-05, "loss": 0.0208, "step": 144020 }, { "epoch": 91.79732313575526, "grad_norm": 0.49920010566711426, "learning_rate": 5.739737511768564e-05, "loss": 0.0181, "step": 144030 }, { "epoch": 91.80369662205226, "grad_norm": 0.30887657403945923, "learning_rate": 5.739192416029119e-05, "loss": 0.0134, "step": 144040 }, { "epoch": 91.81007010834927, "grad_norm": 0.25943177938461304, "learning_rate": 5.73864731130779e-05, "loss": 0.0158, "step": 144050 }, { "epoch": 91.81644359464627, "grad_norm": 0.46177148818969727, "learning_rate": 5.738102197611197e-05, "loss": 0.0226, "step": 144060 }, { "epoch": 91.82281708094328, "grad_norm": 0.15177281200885773, "learning_rate": 5.7375570749459676e-05, "loss": 0.0127, "step": 144070 }, { "epoch": 91.82919056724027, "grad_norm": 0.41283848881721497, "learning_rate": 5.737011943318722e-05, "loss": 0.0116, "step": 144080 }, { "epoch": 91.83556405353728, "grad_norm": 0.280236154794693, "learning_rate": 5.736466802736085e-05, "loss": 0.0084, "step": 144090 }, { "epoch": 91.84193753983429, "grad_norm": 0.3123399317264557, "learning_rate": 5.735921653204682e-05, "loss": 0.0237, "step": 144100 }, { "epoch": 91.8483110261313, "grad_norm": 0.29268351197242737, "learning_rate": 5.735376494731135e-05, "loss": 0.0116, "step": 144110 }, { "epoch": 91.85468451242829, "grad_norm": 0.47951367497444153, "learning_rate": 5.73483132732207e-05, "loss": 0.0182, "step": 144120 }, { "epoch": 91.8610579987253, "grad_norm": 0.43992215394973755, "learning_rate": 5.734286150984111e-05, "loss": 0.0141, "step": 144130 }, { "epoch": 91.86743148502231, "grad_norm": 0.3784279525279999, "learning_rate": 5.733740965723882e-05, "loss": 0.0133, "step": 144140 }, { "epoch": 91.87380497131932, "grad_norm": 0.2926246225833893, "learning_rate": 5.7331957715480065e-05, "loss": 0.0202, "step": 144150 }, { "epoch": 91.88017845761631, "grad_norm": 0.4754764437675476, "learning_rate": 5.7326505684631105e-05, "loss": 0.0163, "step": 144160 }, { "epoch": 91.88655194391332, "grad_norm": 0.3183499574661255, "learning_rate": 5.732105356475819e-05, "loss": 0.0165, "step": 144170 }, { "epoch": 91.89292543021033, "grad_norm": 0.8661971688270569, "learning_rate": 5.731560135592755e-05, "loss": 0.0111, "step": 144180 }, { "epoch": 91.89929891650733, "grad_norm": 0.18645060062408447, "learning_rate": 5.731014905820547e-05, "loss": 0.0104, "step": 144190 }, { "epoch": 91.90567240280433, "grad_norm": 0.28012487292289734, "learning_rate": 5.730469667165814e-05, "loss": 0.0175, "step": 144200 }, { "epoch": 91.91204588910134, "grad_norm": 0.25844624638557434, "learning_rate": 5.729924419635189e-05, "loss": 0.0112, "step": 144210 }, { "epoch": 91.91841937539834, "grad_norm": 0.3714515268802643, "learning_rate": 5.72937916323529e-05, "loss": 0.0149, "step": 144220 }, { "epoch": 91.92479286169535, "grad_norm": 0.33458569645881653, "learning_rate": 5.728833897972747e-05, "loss": 0.0153, "step": 144230 }, { "epoch": 91.93116634799235, "grad_norm": 0.29994329810142517, "learning_rate": 5.7282886238541834e-05, "loss": 0.0118, "step": 144240 }, { "epoch": 91.93753983428935, "grad_norm": 0.7645736336708069, "learning_rate": 5.727743340886226e-05, "loss": 0.0125, "step": 144250 }, { "epoch": 91.94391332058636, "grad_norm": 0.3496071696281433, "learning_rate": 5.7271980490754994e-05, "loss": 0.0195, "step": 144260 }, { "epoch": 91.95028680688337, "grad_norm": 0.24165931344032288, "learning_rate": 5.7266527484286295e-05, "loss": 0.0121, "step": 144270 }, { "epoch": 91.95666029318036, "grad_norm": 0.5784472227096558, "learning_rate": 5.726107438952244e-05, "loss": 0.0175, "step": 144280 }, { "epoch": 91.96303377947737, "grad_norm": 0.5242268443107605, "learning_rate": 5.725562120652965e-05, "loss": 0.0191, "step": 144290 }, { "epoch": 91.96940726577438, "grad_norm": 0.492284893989563, "learning_rate": 5.725016793537424e-05, "loss": 0.0143, "step": 144300 }, { "epoch": 91.97578075207139, "grad_norm": 0.39136508107185364, "learning_rate": 5.724471457612244e-05, "loss": 0.0159, "step": 144310 }, { "epoch": 91.98215423836838, "grad_norm": 0.35635054111480713, "learning_rate": 5.723926112884051e-05, "loss": 0.0096, "step": 144320 }, { "epoch": 91.98852772466539, "grad_norm": 0.6183533072471619, "learning_rate": 5.7233807593594726e-05, "loss": 0.0169, "step": 144330 }, { "epoch": 91.9949012109624, "grad_norm": 0.16442865133285522, "learning_rate": 5.722835397045134e-05, "loss": 0.0114, "step": 144340 }, { "epoch": 92.0012746972594, "grad_norm": 0.23481552302837372, "learning_rate": 5.722290025947664e-05, "loss": 0.0157, "step": 144350 }, { "epoch": 92.00764818355641, "grad_norm": 0.5103223323822021, "learning_rate": 5.721744646073688e-05, "loss": 0.0201, "step": 144360 }, { "epoch": 92.0140216698534, "grad_norm": 0.33418670296669006, "learning_rate": 5.7211992574298325e-05, "loss": 0.0136, "step": 144370 }, { "epoch": 92.02039515615041, "grad_norm": 0.19295544922351837, "learning_rate": 5.720653860022726e-05, "loss": 0.0174, "step": 144380 }, { "epoch": 92.02676864244742, "grad_norm": 0.42627185583114624, "learning_rate": 5.7201084538589945e-05, "loss": 0.0222, "step": 144390 }, { "epoch": 92.03314212874443, "grad_norm": 0.2236996740102768, "learning_rate": 5.719563038945265e-05, "loss": 0.0105, "step": 144400 }, { "epoch": 92.03951561504142, "grad_norm": 0.23806728422641754, "learning_rate": 5.719017615288165e-05, "loss": 0.023, "step": 144410 }, { "epoch": 92.04588910133843, "grad_norm": 0.40482738614082336, "learning_rate": 5.718472182894322e-05, "loss": 0.0102, "step": 144420 }, { "epoch": 92.05226258763544, "grad_norm": 0.18808086216449738, "learning_rate": 5.717926741770364e-05, "loss": 0.0068, "step": 144430 }, { "epoch": 92.05863607393245, "grad_norm": 0.2935611605644226, "learning_rate": 5.717381291922919e-05, "loss": 0.0137, "step": 144440 }, { "epoch": 92.06500956022944, "grad_norm": 0.34073376655578613, "learning_rate": 5.716835833358612e-05, "loss": 0.0105, "step": 144450 }, { "epoch": 92.07138304652645, "grad_norm": 0.4030036926269531, "learning_rate": 5.716290366084075e-05, "loss": 0.0169, "step": 144460 }, { "epoch": 92.07775653282346, "grad_norm": 0.9832319021224976, "learning_rate": 5.715744890105933e-05, "loss": 0.0144, "step": 144470 }, { "epoch": 92.08413001912047, "grad_norm": 0.19579362869262695, "learning_rate": 5.715199405430814e-05, "loss": 0.0065, "step": 144480 }, { "epoch": 92.09050350541746, "grad_norm": 0.16694028675556183, "learning_rate": 5.7146539120653486e-05, "loss": 0.0108, "step": 144490 }, { "epoch": 92.09687699171447, "grad_norm": 0.3932695686817169, "learning_rate": 5.7141084100161616e-05, "loss": 0.0139, "step": 144500 }, { "epoch": 92.10325047801147, "grad_norm": 0.35489794611930847, "learning_rate": 5.713562899289885e-05, "loss": 0.0117, "step": 144510 }, { "epoch": 92.10962396430848, "grad_norm": 0.3094278872013092, "learning_rate": 5.713017379893144e-05, "loss": 0.0079, "step": 144520 }, { "epoch": 92.11599745060548, "grad_norm": 0.2102876454591751, "learning_rate": 5.712471851832569e-05, "loss": 0.0214, "step": 144530 }, { "epoch": 92.12237093690248, "grad_norm": 0.6374598145484924, "learning_rate": 5.7119263151147885e-05, "loss": 0.0148, "step": 144540 }, { "epoch": 92.12874442319949, "grad_norm": 0.39451780915260315, "learning_rate": 5.711380769746432e-05, "loss": 0.0113, "step": 144550 }, { "epoch": 92.1351179094965, "grad_norm": 0.29130491614341736, "learning_rate": 5.710835215734126e-05, "loss": 0.0158, "step": 144560 }, { "epoch": 92.1414913957935, "grad_norm": 0.3967180848121643, "learning_rate": 5.7102896530845015e-05, "loss": 0.014, "step": 144570 }, { "epoch": 92.1478648820905, "grad_norm": 0.2926569879055023, "learning_rate": 5.709744081804187e-05, "loss": 0.0162, "step": 144580 }, { "epoch": 92.15423836838751, "grad_norm": 0.43795520067214966, "learning_rate": 5.7091985018998116e-05, "loss": 0.01, "step": 144590 }, { "epoch": 92.16061185468452, "grad_norm": 0.3280838131904602, "learning_rate": 5.7086529133780054e-05, "loss": 0.0196, "step": 144600 }, { "epoch": 92.16698534098151, "grad_norm": 0.3483107089996338, "learning_rate": 5.708107316245397e-05, "loss": 0.0124, "step": 144610 }, { "epoch": 92.17335882727852, "grad_norm": 0.530800998210907, "learning_rate": 5.7075617105086166e-05, "loss": 0.0183, "step": 144620 }, { "epoch": 92.17973231357553, "grad_norm": 0.39673465490341187, "learning_rate": 5.707016096174291e-05, "loss": 0.0196, "step": 144630 }, { "epoch": 92.18610579987254, "grad_norm": 0.2795568108558655, "learning_rate": 5.706470473249056e-05, "loss": 0.0157, "step": 144640 }, { "epoch": 92.19247928616953, "grad_norm": 0.46249303221702576, "learning_rate": 5.705924841739535e-05, "loss": 0.0149, "step": 144650 }, { "epoch": 92.19885277246654, "grad_norm": 0.2643163204193115, "learning_rate": 5.7053792016523625e-05, "loss": 0.0118, "step": 144660 }, { "epoch": 92.20522625876355, "grad_norm": 0.2976883351802826, "learning_rate": 5.704833552994164e-05, "loss": 0.0158, "step": 144670 }, { "epoch": 92.21159974506055, "grad_norm": 0.4826933741569519, "learning_rate": 5.704287895771575e-05, "loss": 0.0209, "step": 144680 }, { "epoch": 92.21797323135755, "grad_norm": 0.19507089257240295, "learning_rate": 5.7037422299912226e-05, "loss": 0.0109, "step": 144690 }, { "epoch": 92.22434671765456, "grad_norm": 0.22496354579925537, "learning_rate": 5.703196555659737e-05, "loss": 0.0102, "step": 144700 }, { "epoch": 92.23072020395156, "grad_norm": 0.19993332028388977, "learning_rate": 5.702650872783749e-05, "loss": 0.0138, "step": 144710 }, { "epoch": 92.23709369024857, "grad_norm": 0.40232613682746887, "learning_rate": 5.702105181369891e-05, "loss": 0.0123, "step": 144720 }, { "epoch": 92.24346717654556, "grad_norm": 0.1777331531047821, "learning_rate": 5.701559481424791e-05, "loss": 0.0108, "step": 144730 }, { "epoch": 92.24984066284257, "grad_norm": 0.2166650891304016, "learning_rate": 5.701013772955082e-05, "loss": 0.0172, "step": 144740 }, { "epoch": 92.25621414913958, "grad_norm": 0.25684210658073425, "learning_rate": 5.7004680559673915e-05, "loss": 0.0192, "step": 144750 }, { "epoch": 92.26258763543659, "grad_norm": 0.19950206577777863, "learning_rate": 5.699922330468355e-05, "loss": 0.0119, "step": 144760 }, { "epoch": 92.26896112173358, "grad_norm": 0.5421521067619324, "learning_rate": 5.6993765964645993e-05, "loss": 0.0148, "step": 144770 }, { "epoch": 92.27533460803059, "grad_norm": 0.19173504412174225, "learning_rate": 5.69883085396276e-05, "loss": 0.0143, "step": 144780 }, { "epoch": 92.2817080943276, "grad_norm": 0.21388083696365356, "learning_rate": 5.6982851029694636e-05, "loss": 0.0168, "step": 144790 }, { "epoch": 92.2880815806246, "grad_norm": 0.2538701891899109, "learning_rate": 5.697739343491346e-05, "loss": 0.0125, "step": 144800 }, { "epoch": 92.2944550669216, "grad_norm": 0.24677500128746033, "learning_rate": 5.697193575535037e-05, "loss": 0.0146, "step": 144810 }, { "epoch": 92.30082855321861, "grad_norm": 0.20921340584754944, "learning_rate": 5.696647799107165e-05, "loss": 0.0103, "step": 144820 }, { "epoch": 92.30720203951562, "grad_norm": 0.22500020265579224, "learning_rate": 5.6961020142143676e-05, "loss": 0.0101, "step": 144830 }, { "epoch": 92.31357552581262, "grad_norm": 0.23651817440986633, "learning_rate": 5.695556220863271e-05, "loss": 0.0118, "step": 144840 }, { "epoch": 92.31994901210962, "grad_norm": 0.3561020493507385, "learning_rate": 5.695010419060511e-05, "loss": 0.0176, "step": 144850 }, { "epoch": 92.32632249840663, "grad_norm": 0.4764436185359955, "learning_rate": 5.6944646088127174e-05, "loss": 0.0217, "step": 144860 }, { "epoch": 92.33269598470363, "grad_norm": 0.411953330039978, "learning_rate": 5.6939187901265244e-05, "loss": 0.0154, "step": 144870 }, { "epoch": 92.33906947100064, "grad_norm": 0.5255167484283447, "learning_rate": 5.6933729630085617e-05, "loss": 0.0098, "step": 144880 }, { "epoch": 92.34544295729764, "grad_norm": 0.17044448852539062, "learning_rate": 5.692827127465463e-05, "loss": 0.0101, "step": 144890 }, { "epoch": 92.35181644359464, "grad_norm": 0.38473448157310486, "learning_rate": 5.692281283503862e-05, "loss": 0.0115, "step": 144900 }, { "epoch": 92.35818992989165, "grad_norm": 0.40604695677757263, "learning_rate": 5.691735431130387e-05, "loss": 0.0221, "step": 144910 }, { "epoch": 92.36456341618866, "grad_norm": 0.24267427623271942, "learning_rate": 5.691189570351676e-05, "loss": 0.0099, "step": 144920 }, { "epoch": 92.37093690248565, "grad_norm": 0.26977965235710144, "learning_rate": 5.690643701174357e-05, "loss": 0.0149, "step": 144930 }, { "epoch": 92.37731038878266, "grad_norm": 0.4230418801307678, "learning_rate": 5.690097823605067e-05, "loss": 0.0104, "step": 144940 }, { "epoch": 92.38368387507967, "grad_norm": 0.25527092814445496, "learning_rate": 5.689551937650436e-05, "loss": 0.0127, "step": 144950 }, { "epoch": 92.39005736137668, "grad_norm": 0.12630251049995422, "learning_rate": 5.689006043317098e-05, "loss": 0.012, "step": 144960 }, { "epoch": 92.39643084767367, "grad_norm": 0.17016585171222687, "learning_rate": 5.688460140611687e-05, "loss": 0.0229, "step": 144970 }, { "epoch": 92.40280433397068, "grad_norm": 0.4182908535003662, "learning_rate": 5.687914229540833e-05, "loss": 0.0171, "step": 144980 }, { "epoch": 92.40917782026769, "grad_norm": 0.3214331865310669, "learning_rate": 5.687368310111173e-05, "loss": 0.0104, "step": 144990 }, { "epoch": 92.4155513065647, "grad_norm": 0.4221337139606476, "learning_rate": 5.686822382329339e-05, "loss": 0.0149, "step": 145000 }, { "epoch": 92.4219247928617, "grad_norm": 0.18912814557552338, "learning_rate": 5.6862764462019646e-05, "loss": 0.0104, "step": 145010 }, { "epoch": 92.4282982791587, "grad_norm": 0.22305582463741302, "learning_rate": 5.685730501735682e-05, "loss": 0.0099, "step": 145020 }, { "epoch": 92.4346717654557, "grad_norm": 0.317444384098053, "learning_rate": 5.685184548937129e-05, "loss": 0.0084, "step": 145030 }, { "epoch": 92.44104525175271, "grad_norm": 0.285159170627594, "learning_rate": 5.684638587812936e-05, "loss": 0.013, "step": 145040 }, { "epoch": 92.44741873804972, "grad_norm": 0.15300928056240082, "learning_rate": 5.684092618369736e-05, "loss": 0.0107, "step": 145050 }, { "epoch": 92.45379222434671, "grad_norm": 0.40498650074005127, "learning_rate": 5.683546640614167e-05, "loss": 0.0213, "step": 145060 }, { "epoch": 92.46016571064372, "grad_norm": 0.5655746459960938, "learning_rate": 5.6830006545528594e-05, "loss": 0.0141, "step": 145070 }, { "epoch": 92.46653919694073, "grad_norm": 1.1130366325378418, "learning_rate": 5.68245466019245e-05, "loss": 0.0164, "step": 145080 }, { "epoch": 92.47291268323774, "grad_norm": 0.27705803513526917, "learning_rate": 5.681908657539571e-05, "loss": 0.0231, "step": 145090 }, { "epoch": 92.47928616953473, "grad_norm": 0.5756409168243408, "learning_rate": 5.6813626466008585e-05, "loss": 0.0178, "step": 145100 }, { "epoch": 92.48565965583174, "grad_norm": 0.6108702421188354, "learning_rate": 5.6808166273829465e-05, "loss": 0.012, "step": 145110 }, { "epoch": 92.49203314212875, "grad_norm": 0.37920907139778137, "learning_rate": 5.68027059989247e-05, "loss": 0.0312, "step": 145120 }, { "epoch": 92.49840662842576, "grad_norm": 0.6271448135375977, "learning_rate": 5.6797245641360644e-05, "loss": 0.02, "step": 145130 }, { "epoch": 92.50478011472275, "grad_norm": 0.5343686938285828, "learning_rate": 5.6791785201203616e-05, "loss": 0.026, "step": 145140 }, { "epoch": 92.51115360101976, "grad_norm": 0.5083519220352173, "learning_rate": 5.678632467852e-05, "loss": 0.0339, "step": 145150 }, { "epoch": 92.51752708731676, "grad_norm": 0.5379785895347595, "learning_rate": 5.678086407337612e-05, "loss": 0.0093, "step": 145160 }, { "epoch": 92.52390057361377, "grad_norm": 0.37798309326171875, "learning_rate": 5.6775403385838354e-05, "loss": 0.0128, "step": 145170 }, { "epoch": 92.53027405991077, "grad_norm": 0.3203207552433014, "learning_rate": 5.676994261597301e-05, "loss": 0.0097, "step": 145180 }, { "epoch": 92.53664754620777, "grad_norm": 0.21205225586891174, "learning_rate": 5.676448176384651e-05, "loss": 0.0125, "step": 145190 }, { "epoch": 92.54302103250478, "grad_norm": 0.20712727308273315, "learning_rate": 5.6759020829525154e-05, "loss": 0.0141, "step": 145200 }, { "epoch": 92.54939451880179, "grad_norm": 0.1966904103755951, "learning_rate": 5.675355981307531e-05, "loss": 0.008, "step": 145210 }, { "epoch": 92.55576800509878, "grad_norm": 0.22146159410476685, "learning_rate": 5.6748098714563345e-05, "loss": 0.0126, "step": 145220 }, { "epoch": 92.56214149139579, "grad_norm": 1.6382886171340942, "learning_rate": 5.67426375340556e-05, "loss": 0.0095, "step": 145230 }, { "epoch": 92.5685149776928, "grad_norm": 0.27126505970954895, "learning_rate": 5.673717627161844e-05, "loss": 0.0192, "step": 145240 }, { "epoch": 92.57488846398981, "grad_norm": 0.2915521264076233, "learning_rate": 5.6731714927318236e-05, "loss": 0.0129, "step": 145250 }, { "epoch": 92.5812619502868, "grad_norm": 0.31756970286369324, "learning_rate": 5.6726253501221336e-05, "loss": 0.0147, "step": 145260 }, { "epoch": 92.58763543658381, "grad_norm": 0.25484082102775574, "learning_rate": 5.6720791993394104e-05, "loss": 0.0092, "step": 145270 }, { "epoch": 92.59400892288082, "grad_norm": 0.5886210799217224, "learning_rate": 5.671533040390291e-05, "loss": 0.0156, "step": 145280 }, { "epoch": 92.60038240917783, "grad_norm": 0.282459020614624, "learning_rate": 5.670986873281412e-05, "loss": 0.0104, "step": 145290 }, { "epoch": 92.60675589547482, "grad_norm": 0.22436954081058502, "learning_rate": 5.670440698019407e-05, "loss": 0.0179, "step": 145300 }, { "epoch": 92.61312938177183, "grad_norm": 0.4374842345714569, "learning_rate": 5.669894514610916e-05, "loss": 0.0109, "step": 145310 }, { "epoch": 92.61950286806884, "grad_norm": 0.53550124168396, "learning_rate": 5.669348323062572e-05, "loss": 0.0101, "step": 145320 }, { "epoch": 92.62587635436584, "grad_norm": 0.7441917061805725, "learning_rate": 5.6688021233810164e-05, "loss": 0.0306, "step": 145330 }, { "epoch": 92.63224984066284, "grad_norm": 0.4227094352245331, "learning_rate": 5.668255915572882e-05, "loss": 0.0187, "step": 145340 }, { "epoch": 92.63862332695984, "grad_norm": 0.41193634271621704, "learning_rate": 5.667709699644808e-05, "loss": 0.0112, "step": 145350 }, { "epoch": 92.64499681325685, "grad_norm": 0.2775179445743561, "learning_rate": 5.6671634756034295e-05, "loss": 0.0237, "step": 145360 }, { "epoch": 92.65137029955386, "grad_norm": 0.2522629499435425, "learning_rate": 5.666617243455387e-05, "loss": 0.0186, "step": 145370 }, { "epoch": 92.65774378585085, "grad_norm": 0.40172287821769714, "learning_rate": 5.666071003207314e-05, "loss": 0.0076, "step": 145380 }, { "epoch": 92.66411727214786, "grad_norm": 0.18423353135585785, "learning_rate": 5.66552475486585e-05, "loss": 0.0207, "step": 145390 }, { "epoch": 92.67049075844487, "grad_norm": 0.5866109132766724, "learning_rate": 5.664978498437632e-05, "loss": 0.0105, "step": 145400 }, { "epoch": 92.67686424474188, "grad_norm": 0.4495161473751068, "learning_rate": 5.6644322339292975e-05, "loss": 0.0213, "step": 145410 }, { "epoch": 92.68323773103887, "grad_norm": 0.23948319256305695, "learning_rate": 5.6638859613474825e-05, "loss": 0.0111, "step": 145420 }, { "epoch": 92.68961121733588, "grad_norm": 0.48597249388694763, "learning_rate": 5.663339680698828e-05, "loss": 0.0153, "step": 145430 }, { "epoch": 92.69598470363289, "grad_norm": 0.8170632123947144, "learning_rate": 5.662793391989969e-05, "loss": 0.0174, "step": 145440 }, { "epoch": 92.7023581899299, "grad_norm": 0.26335737109184265, "learning_rate": 5.662247095227546e-05, "loss": 0.0227, "step": 145450 }, { "epoch": 92.70873167622689, "grad_norm": 0.24948248267173767, "learning_rate": 5.661700790418194e-05, "loss": 0.0066, "step": 145460 }, { "epoch": 92.7151051625239, "grad_norm": 0.28452831506729126, "learning_rate": 5.661154477568553e-05, "loss": 0.009, "step": 145470 }, { "epoch": 92.7214786488209, "grad_norm": 0.6214900016784668, "learning_rate": 5.660608156685261e-05, "loss": 0.0118, "step": 145480 }, { "epoch": 92.72785213511791, "grad_norm": 0.45220911502838135, "learning_rate": 5.6600618277749575e-05, "loss": 0.0209, "step": 145490 }, { "epoch": 92.73422562141491, "grad_norm": 0.495716392993927, "learning_rate": 5.659515490844277e-05, "loss": 0.0123, "step": 145500 }, { "epoch": 92.74059910771192, "grad_norm": 0.3224101960659027, "learning_rate": 5.658969145899863e-05, "loss": 0.0117, "step": 145510 }, { "epoch": 92.74697259400892, "grad_norm": 0.22997558116912842, "learning_rate": 5.6584227929483514e-05, "loss": 0.0159, "step": 145520 }, { "epoch": 92.75334608030593, "grad_norm": 0.7531261444091797, "learning_rate": 5.65787643199638e-05, "loss": 0.0108, "step": 145530 }, { "epoch": 92.75971956660293, "grad_norm": 0.21900245547294617, "learning_rate": 5.657330063050589e-05, "loss": 0.0098, "step": 145540 }, { "epoch": 92.76609305289993, "grad_norm": 0.4711115062236786, "learning_rate": 5.656783686117617e-05, "loss": 0.0168, "step": 145550 }, { "epoch": 92.77246653919694, "grad_norm": 0.5555914044380188, "learning_rate": 5.656237301204104e-05, "loss": 0.0136, "step": 145560 }, { "epoch": 92.77884002549395, "grad_norm": 0.3505522310733795, "learning_rate": 5.655690908316688e-05, "loss": 0.0113, "step": 145570 }, { "epoch": 92.78521351179094, "grad_norm": 0.292263388633728, "learning_rate": 5.655144507462008e-05, "loss": 0.0135, "step": 145580 }, { "epoch": 92.79158699808795, "grad_norm": 0.2468748241662979, "learning_rate": 5.654598098646703e-05, "loss": 0.0132, "step": 145590 }, { "epoch": 92.79796048438496, "grad_norm": 0.2356233149766922, "learning_rate": 5.654051681877415e-05, "loss": 0.0068, "step": 145600 }, { "epoch": 92.80433397068197, "grad_norm": 0.44373148679733276, "learning_rate": 5.6535052571607806e-05, "loss": 0.0101, "step": 145610 }, { "epoch": 92.81070745697897, "grad_norm": 0.5234602689743042, "learning_rate": 5.6529588245034396e-05, "loss": 0.0112, "step": 145620 }, { "epoch": 92.81708094327597, "grad_norm": 0.38471031188964844, "learning_rate": 5.652412383912034e-05, "loss": 0.014, "step": 145630 }, { "epoch": 92.82345442957298, "grad_norm": 0.22267208993434906, "learning_rate": 5.651865935393199e-05, "loss": 0.0196, "step": 145640 }, { "epoch": 92.82982791586998, "grad_norm": 0.5239781141281128, "learning_rate": 5.6513194789535805e-05, "loss": 0.0131, "step": 145650 }, { "epoch": 92.83620140216699, "grad_norm": 0.56694096326828, "learning_rate": 5.6507730145998126e-05, "loss": 0.0203, "step": 145660 }, { "epoch": 92.84257488846399, "grad_norm": 0.540851354598999, "learning_rate": 5.65022654233854e-05, "loss": 0.0109, "step": 145670 }, { "epoch": 92.848948374761, "grad_norm": 0.3921886384487152, "learning_rate": 5.649680062176401e-05, "loss": 0.0148, "step": 145680 }, { "epoch": 92.855321861058, "grad_norm": 0.32475823163986206, "learning_rate": 5.6491335741200356e-05, "loss": 0.0132, "step": 145690 }, { "epoch": 92.86169534735501, "grad_norm": 0.24148790538311005, "learning_rate": 5.648587078176085e-05, "loss": 0.0182, "step": 145700 }, { "epoch": 92.868068833652, "grad_norm": 0.22038544714450836, "learning_rate": 5.6480405743511876e-05, "loss": 0.0068, "step": 145710 }, { "epoch": 92.87444231994901, "grad_norm": 0.6419162750244141, "learning_rate": 5.6474940626519866e-05, "loss": 0.024, "step": 145720 }, { "epoch": 92.88081580624602, "grad_norm": 0.23162826895713806, "learning_rate": 5.646947543085121e-05, "loss": 0.0169, "step": 145730 }, { "epoch": 92.88718929254303, "grad_norm": 0.1890372931957245, "learning_rate": 5.646401015657232e-05, "loss": 0.0127, "step": 145740 }, { "epoch": 92.89356277884002, "grad_norm": 0.49108946323394775, "learning_rate": 5.645854480374959e-05, "loss": 0.0154, "step": 145750 }, { "epoch": 92.89993626513703, "grad_norm": 1.136120319366455, "learning_rate": 5.6453079372449466e-05, "loss": 0.0216, "step": 145760 }, { "epoch": 92.90630975143404, "grad_norm": 0.5663464665412903, "learning_rate": 5.644761386273833e-05, "loss": 0.0129, "step": 145770 }, { "epoch": 92.91268323773104, "grad_norm": 0.32834672927856445, "learning_rate": 5.6442148274682584e-05, "loss": 0.0117, "step": 145780 }, { "epoch": 92.91905672402804, "grad_norm": 0.25218522548675537, "learning_rate": 5.643668260834867e-05, "loss": 0.0141, "step": 145790 }, { "epoch": 92.92543021032505, "grad_norm": 0.3194790184497833, "learning_rate": 5.643121686380297e-05, "loss": 0.0169, "step": 145800 }, { "epoch": 92.93180369662205, "grad_norm": 0.4805222153663635, "learning_rate": 5.642575104111192e-05, "loss": 0.0267, "step": 145810 }, { "epoch": 92.93817718291906, "grad_norm": 0.2933099865913391, "learning_rate": 5.642028514034192e-05, "loss": 0.0104, "step": 145820 }, { "epoch": 92.94455066921606, "grad_norm": 0.3519473671913147, "learning_rate": 5.641481916155942e-05, "loss": 0.0128, "step": 145830 }, { "epoch": 92.95092415551306, "grad_norm": 0.4094841182231903, "learning_rate": 5.640935310483078e-05, "loss": 0.0107, "step": 145840 }, { "epoch": 92.95729764181007, "grad_norm": 0.34233465790748596, "learning_rate": 5.6403886970222463e-05, "loss": 0.0155, "step": 145850 }, { "epoch": 92.96367112810708, "grad_norm": 0.30390235781669617, "learning_rate": 5.639842075780087e-05, "loss": 0.0098, "step": 145860 }, { "epoch": 92.97004461440407, "grad_norm": 0.17502020299434662, "learning_rate": 5.639295446763242e-05, "loss": 0.0142, "step": 145870 }, { "epoch": 92.97641810070108, "grad_norm": 0.4761334955692291, "learning_rate": 5.6387488099783546e-05, "loss": 0.0183, "step": 145880 }, { "epoch": 92.98279158699809, "grad_norm": 0.25022220611572266, "learning_rate": 5.638202165432066e-05, "loss": 0.0132, "step": 145890 }, { "epoch": 92.9891650732951, "grad_norm": 0.1968190222978592, "learning_rate": 5.637655513131018e-05, "loss": 0.0173, "step": 145900 }, { "epoch": 92.99553855959209, "grad_norm": 0.32935282588005066, "learning_rate": 5.637108853081854e-05, "loss": 0.0125, "step": 145910 }, { "epoch": 93.0019120458891, "grad_norm": 0.3159519135951996, "learning_rate": 5.6365621852912154e-05, "loss": 0.0093, "step": 145920 }, { "epoch": 93.00828553218611, "grad_norm": 0.28224751353263855, "learning_rate": 5.6360155097657464e-05, "loss": 0.0121, "step": 145930 }, { "epoch": 93.01465901848312, "grad_norm": 0.3833948075771332, "learning_rate": 5.635468826512087e-05, "loss": 0.0138, "step": 145940 }, { "epoch": 93.02103250478011, "grad_norm": 0.2712289094924927, "learning_rate": 5.634922135536882e-05, "loss": 0.0115, "step": 145950 }, { "epoch": 93.02740599107712, "grad_norm": 0.17850737273693085, "learning_rate": 5.634375436846773e-05, "loss": 0.0157, "step": 145960 }, { "epoch": 93.03377947737413, "grad_norm": 0.35803720355033875, "learning_rate": 5.633828730448404e-05, "loss": 0.0102, "step": 145970 }, { "epoch": 93.04015296367113, "grad_norm": 0.3269091546535492, "learning_rate": 5.633282016348417e-05, "loss": 0.008, "step": 145980 }, { "epoch": 93.04652644996813, "grad_norm": 0.7712300419807434, "learning_rate": 5.632735294553456e-05, "loss": 0.0294, "step": 145990 }, { "epoch": 93.05289993626513, "grad_norm": 0.37199366092681885, "learning_rate": 5.632188565070163e-05, "loss": 0.0159, "step": 146000 }, { "epoch": 93.05927342256214, "grad_norm": 0.3492395579814911, "learning_rate": 5.6316418279051844e-05, "loss": 0.0114, "step": 146010 }, { "epoch": 93.06564690885915, "grad_norm": 0.359671950340271, "learning_rate": 5.63109508306516e-05, "loss": 0.0135, "step": 146020 }, { "epoch": 93.07202039515614, "grad_norm": 0.457660973072052, "learning_rate": 5.630548330556733e-05, "loss": 0.0113, "step": 146030 }, { "epoch": 93.07839388145315, "grad_norm": 0.2951827645301819, "learning_rate": 5.6300015703865494e-05, "loss": 0.0121, "step": 146040 }, { "epoch": 93.08476736775016, "grad_norm": 0.7867652177810669, "learning_rate": 5.6294548025612514e-05, "loss": 0.0144, "step": 146050 }, { "epoch": 93.09114085404717, "grad_norm": 0.3991543650627136, "learning_rate": 5.628908027087485e-05, "loss": 0.0124, "step": 146060 }, { "epoch": 93.09751434034416, "grad_norm": 0.31761133670806885, "learning_rate": 5.62836124397189e-05, "loss": 0.0186, "step": 146070 }, { "epoch": 93.10388782664117, "grad_norm": 0.3676363229751587, "learning_rate": 5.627814453221115e-05, "loss": 0.0108, "step": 146080 }, { "epoch": 93.11026131293818, "grad_norm": 0.5715415477752686, "learning_rate": 5.6272676548418e-05, "loss": 0.0199, "step": 146090 }, { "epoch": 93.11663479923519, "grad_norm": 0.27619993686676025, "learning_rate": 5.6267208488405924e-05, "loss": 0.0112, "step": 146100 }, { "epoch": 93.12300828553218, "grad_norm": 0.3095812499523163, "learning_rate": 5.626174035224132e-05, "loss": 0.0093, "step": 146110 }, { "epoch": 93.12938177182919, "grad_norm": 0.3883094787597656, "learning_rate": 5.6256272139990676e-05, "loss": 0.015, "step": 146120 }, { "epoch": 93.1357552581262, "grad_norm": 0.3421034514904022, "learning_rate": 5.6250803851720416e-05, "loss": 0.0147, "step": 146130 }, { "epoch": 93.1421287444232, "grad_norm": 0.19505247473716736, "learning_rate": 5.624533548749699e-05, "loss": 0.0132, "step": 146140 }, { "epoch": 93.1485022307202, "grad_norm": 0.7759355902671814, "learning_rate": 5.623986704738684e-05, "loss": 0.0193, "step": 146150 }, { "epoch": 93.1548757170172, "grad_norm": 0.3946567177772522, "learning_rate": 5.623439853145641e-05, "loss": 0.0147, "step": 146160 }, { "epoch": 93.16124920331421, "grad_norm": 0.560217559337616, "learning_rate": 5.622892993977216e-05, "loss": 0.0153, "step": 146170 }, { "epoch": 93.16762268961122, "grad_norm": 0.35469695925712585, "learning_rate": 5.6223461272400525e-05, "loss": 0.0166, "step": 146180 }, { "epoch": 93.17399617590821, "grad_norm": 0.23648107051849365, "learning_rate": 5.6217992529407946e-05, "loss": 0.0132, "step": 146190 }, { "epoch": 93.18036966220522, "grad_norm": 0.32512709498405457, "learning_rate": 5.621252371086091e-05, "loss": 0.0101, "step": 146200 }, { "epoch": 93.18674314850223, "grad_norm": 0.20987388491630554, "learning_rate": 5.6207054816825835e-05, "loss": 0.0102, "step": 146210 }, { "epoch": 93.19311663479924, "grad_norm": 0.682107150554657, "learning_rate": 5.6201585847369185e-05, "loss": 0.0119, "step": 146220 }, { "epoch": 93.19949012109623, "grad_norm": 0.3830803334712982, "learning_rate": 5.619611680255741e-05, "loss": 0.0112, "step": 146230 }, { "epoch": 93.20586360739324, "grad_norm": 1.043789029121399, "learning_rate": 5.6190647682456965e-05, "loss": 0.0294, "step": 146240 }, { "epoch": 93.21223709369025, "grad_norm": 0.5551419258117676, "learning_rate": 5.618517848713432e-05, "loss": 0.0137, "step": 146250 }, { "epoch": 93.21861057998726, "grad_norm": 0.24493442475795746, "learning_rate": 5.6179709216655895e-05, "loss": 0.0143, "step": 146260 }, { "epoch": 93.22498406628426, "grad_norm": 0.6193442940711975, "learning_rate": 5.617423987108818e-05, "loss": 0.0102, "step": 146270 }, { "epoch": 93.23135755258126, "grad_norm": 0.48596128821372986, "learning_rate": 5.616877045049762e-05, "loss": 0.0194, "step": 146280 }, { "epoch": 93.23773103887827, "grad_norm": 0.4406862258911133, "learning_rate": 5.616330095495067e-05, "loss": 0.021, "step": 146290 }, { "epoch": 93.24410452517527, "grad_norm": 0.2438385933637619, "learning_rate": 5.6157831384513796e-05, "loss": 0.0101, "step": 146300 }, { "epoch": 93.25047801147228, "grad_norm": 0.4448620676994324, "learning_rate": 5.615236173925347e-05, "loss": 0.0116, "step": 146310 }, { "epoch": 93.25685149776928, "grad_norm": 0.6925949454307556, "learning_rate": 5.614689201923612e-05, "loss": 0.0211, "step": 146320 }, { "epoch": 93.26322498406628, "grad_norm": 0.5809403657913208, "learning_rate": 5.614142222452824e-05, "loss": 0.0095, "step": 146330 }, { "epoch": 93.26959847036329, "grad_norm": 0.23931095004081726, "learning_rate": 5.613595235519629e-05, "loss": 0.0089, "step": 146340 }, { "epoch": 93.2759719566603, "grad_norm": 0.46331050992012024, "learning_rate": 5.613048241130671e-05, "loss": 0.0227, "step": 146350 }, { "epoch": 93.2823454429573, "grad_norm": 0.23879273235797882, "learning_rate": 5.612501239292598e-05, "loss": 0.0151, "step": 146360 }, { "epoch": 93.2887189292543, "grad_norm": 0.31689977645874023, "learning_rate": 5.611954230012056e-05, "loss": 0.0176, "step": 146370 }, { "epoch": 93.29509241555131, "grad_norm": 0.27960047125816345, "learning_rate": 5.611407213295694e-05, "loss": 0.011, "step": 146380 }, { "epoch": 93.30146590184832, "grad_norm": 0.46159446239471436, "learning_rate": 5.610860189150156e-05, "loss": 0.0247, "step": 146390 }, { "epoch": 93.30783938814531, "grad_norm": 0.8031337857246399, "learning_rate": 5.610313157582091e-05, "loss": 0.0126, "step": 146400 }, { "epoch": 93.31421287444232, "grad_norm": 0.48176252841949463, "learning_rate": 5.6097661185981455e-05, "loss": 0.0148, "step": 146410 }, { "epoch": 93.32058636073933, "grad_norm": 0.5185056328773499, "learning_rate": 5.609219072204964e-05, "loss": 0.0117, "step": 146420 }, { "epoch": 93.32695984703633, "grad_norm": 0.26833078265190125, "learning_rate": 5.6086720184091967e-05, "loss": 0.0075, "step": 146430 }, { "epoch": 93.33333333333333, "grad_norm": 0.4677334129810333, "learning_rate": 5.6081249572174884e-05, "loss": 0.0089, "step": 146440 }, { "epoch": 93.33970681963034, "grad_norm": 0.35330575704574585, "learning_rate": 5.6075778886364904e-05, "loss": 0.0149, "step": 146450 }, { "epoch": 93.34608030592734, "grad_norm": 0.31058207154273987, "learning_rate": 5.607030812672844e-05, "loss": 0.0127, "step": 146460 }, { "epoch": 93.35245379222435, "grad_norm": 0.2615984380245209, "learning_rate": 5.606483729333203e-05, "loss": 0.0136, "step": 146470 }, { "epoch": 93.35882727852135, "grad_norm": 0.18231306970119476, "learning_rate": 5.6059366386242105e-05, "loss": 0.0133, "step": 146480 }, { "epoch": 93.36520076481835, "grad_norm": 0.26447466015815735, "learning_rate": 5.605389540552517e-05, "loss": 0.021, "step": 146490 }, { "epoch": 93.37157425111536, "grad_norm": 0.46362578868865967, "learning_rate": 5.6048424351247686e-05, "loss": 0.0188, "step": 146500 }, { "epoch": 93.37794773741237, "grad_norm": 0.5507728457450867, "learning_rate": 5.604295322347613e-05, "loss": 0.0143, "step": 146510 }, { "epoch": 93.38432122370936, "grad_norm": 0.3092469871044159, "learning_rate": 5.603748202227699e-05, "loss": 0.0159, "step": 146520 }, { "epoch": 93.39069471000637, "grad_norm": 0.2868509590625763, "learning_rate": 5.603201074771674e-05, "loss": 0.0167, "step": 146530 }, { "epoch": 93.39706819630338, "grad_norm": 0.668409526348114, "learning_rate": 5.6026539399861855e-05, "loss": 0.0143, "step": 146540 }, { "epoch": 93.40344168260039, "grad_norm": 0.2760385572910309, "learning_rate": 5.602106797877883e-05, "loss": 0.0109, "step": 146550 }, { "epoch": 93.40981516889738, "grad_norm": 0.22825536131858826, "learning_rate": 5.601559648453416e-05, "loss": 0.0123, "step": 146560 }, { "epoch": 93.41618865519439, "grad_norm": 0.3017338216304779, "learning_rate": 5.6010124917194306e-05, "loss": 0.0172, "step": 146570 }, { "epoch": 93.4225621414914, "grad_norm": 0.2905019521713257, "learning_rate": 5.600465327682576e-05, "loss": 0.0241, "step": 146580 }, { "epoch": 93.4289356277884, "grad_norm": 0.21306301653385162, "learning_rate": 5.599918156349501e-05, "loss": 0.0157, "step": 146590 }, { "epoch": 93.4353091140854, "grad_norm": 0.37282463908195496, "learning_rate": 5.599370977726852e-05, "loss": 0.0119, "step": 146600 }, { "epoch": 93.4416826003824, "grad_norm": 0.19169357419013977, "learning_rate": 5.5988237918212814e-05, "loss": 0.0162, "step": 146610 }, { "epoch": 93.44805608667941, "grad_norm": 0.2760481536388397, "learning_rate": 5.5982765986394356e-05, "loss": 0.0109, "step": 146620 }, { "epoch": 93.45442957297642, "grad_norm": 0.3691377639770508, "learning_rate": 5.5977293981879655e-05, "loss": 0.0267, "step": 146630 }, { "epoch": 93.46080305927342, "grad_norm": 0.26188427209854126, "learning_rate": 5.597182190473518e-05, "loss": 0.0108, "step": 146640 }, { "epoch": 93.46717654557042, "grad_norm": 0.6165068745613098, "learning_rate": 5.5966349755027436e-05, "loss": 0.0154, "step": 146650 }, { "epoch": 93.47355003186743, "grad_norm": 0.2852493226528168, "learning_rate": 5.596087753282291e-05, "loss": 0.0123, "step": 146660 }, { "epoch": 93.47992351816444, "grad_norm": 0.16620545089244843, "learning_rate": 5.595540523818809e-05, "loss": 0.0172, "step": 146670 }, { "epoch": 93.48629700446143, "grad_norm": 0.46186888217926025, "learning_rate": 5.594993287118947e-05, "loss": 0.0234, "step": 146680 }, { "epoch": 93.49267049075844, "grad_norm": 0.5115196108818054, "learning_rate": 5.5944460431893545e-05, "loss": 0.0109, "step": 146690 }, { "epoch": 93.49904397705545, "grad_norm": 0.41353029012680054, "learning_rate": 5.593898792036682e-05, "loss": 0.0189, "step": 146700 }, { "epoch": 93.50541746335246, "grad_norm": 0.5312375426292419, "learning_rate": 5.593351533667578e-05, "loss": 0.0188, "step": 146710 }, { "epoch": 93.51179094964945, "grad_norm": 0.6242886185646057, "learning_rate": 5.592804268088694e-05, "loss": 0.0148, "step": 146720 }, { "epoch": 93.51816443594646, "grad_norm": 0.3480576276779175, "learning_rate": 5.5922569953066775e-05, "loss": 0.0246, "step": 146730 }, { "epoch": 93.52453792224347, "grad_norm": 0.5602370500564575, "learning_rate": 5.5917097153281785e-05, "loss": 0.0177, "step": 146740 }, { "epoch": 93.53091140854048, "grad_norm": 0.303656667470932, "learning_rate": 5.5911624281598496e-05, "loss": 0.0151, "step": 146750 }, { "epoch": 93.53728489483747, "grad_norm": 0.27307626605033875, "learning_rate": 5.5906151338083365e-05, "loss": 0.0118, "step": 146760 }, { "epoch": 93.54365838113448, "grad_norm": 0.492296040058136, "learning_rate": 5.5900678322802946e-05, "loss": 0.016, "step": 146770 }, { "epoch": 93.55003186743149, "grad_norm": 0.2231941819190979, "learning_rate": 5.589520523582369e-05, "loss": 0.0109, "step": 146780 }, { "epoch": 93.5564053537285, "grad_norm": 0.2319934070110321, "learning_rate": 5.5889732077212144e-05, "loss": 0.0125, "step": 146790 }, { "epoch": 93.56277884002549, "grad_norm": 0.33493542671203613, "learning_rate": 5.588425884703478e-05, "loss": 0.0218, "step": 146800 }, { "epoch": 93.5691523263225, "grad_norm": 0.204631507396698, "learning_rate": 5.587878554535813e-05, "loss": 0.0093, "step": 146810 }, { "epoch": 93.5755258126195, "grad_norm": 0.26580896973609924, "learning_rate": 5.587331217224867e-05, "loss": 0.0133, "step": 146820 }, { "epoch": 93.58189929891651, "grad_norm": 0.3853384852409363, "learning_rate": 5.586783872777292e-05, "loss": 0.0125, "step": 146830 }, { "epoch": 93.5882727852135, "grad_norm": 0.2811070382595062, "learning_rate": 5.586236521199739e-05, "loss": 0.0077, "step": 146840 }, { "epoch": 93.59464627151051, "grad_norm": 0.19772961735725403, "learning_rate": 5.5856891624988606e-05, "loss": 0.0141, "step": 146850 }, { "epoch": 93.60101975780752, "grad_norm": 0.27188682556152344, "learning_rate": 5.585141796681305e-05, "loss": 0.0147, "step": 146860 }, { "epoch": 93.60739324410453, "grad_norm": 0.4893603026866913, "learning_rate": 5.584594423753724e-05, "loss": 0.0307, "step": 146870 }, { "epoch": 93.61376673040154, "grad_norm": 0.3177490234375, "learning_rate": 5.5840470437227673e-05, "loss": 0.009, "step": 146880 }, { "epoch": 93.62014021669853, "grad_norm": 0.5333943963050842, "learning_rate": 5.583499656595089e-05, "loss": 0.008, "step": 146890 }, { "epoch": 93.62651370299554, "grad_norm": 0.5907066464424133, "learning_rate": 5.582952262377339e-05, "loss": 0.0128, "step": 146900 }, { "epoch": 93.63288718929255, "grad_norm": 0.3174366354942322, "learning_rate": 5.582404861076169e-05, "loss": 0.0153, "step": 146910 }, { "epoch": 93.63926067558955, "grad_norm": 0.44515666365623474, "learning_rate": 5.581857452698229e-05, "loss": 0.018, "step": 146920 }, { "epoch": 93.64563416188655, "grad_norm": 0.324459969997406, "learning_rate": 5.581310037250173e-05, "loss": 0.014, "step": 146930 }, { "epoch": 93.65200764818356, "grad_norm": 0.2650090754032135, "learning_rate": 5.580762614738649e-05, "loss": 0.0113, "step": 146940 }, { "epoch": 93.65838113448056, "grad_norm": 0.22670666873455048, "learning_rate": 5.580215185170313e-05, "loss": 0.0203, "step": 146950 }, { "epoch": 93.66475462077757, "grad_norm": 0.3930833041667938, "learning_rate": 5.579667748551813e-05, "loss": 0.0096, "step": 146960 }, { "epoch": 93.67112810707457, "grad_norm": 0.3619782328605652, "learning_rate": 5.5791203048898043e-05, "loss": 0.0137, "step": 146970 }, { "epoch": 93.67750159337157, "grad_norm": 0.39673498272895813, "learning_rate": 5.5785728541909366e-05, "loss": 0.0194, "step": 146980 }, { "epoch": 93.68387507966858, "grad_norm": 0.4701694846153259, "learning_rate": 5.578025396461861e-05, "loss": 0.0148, "step": 146990 }, { "epoch": 93.69024856596559, "grad_norm": 0.19050949811935425, "learning_rate": 5.577477931709233e-05, "loss": 0.0132, "step": 147000 }, { "epoch": 93.69662205226258, "grad_norm": 0.35949671268463135, "learning_rate": 5.576930459939702e-05, "loss": 0.0201, "step": 147010 }, { "epoch": 93.70299553855959, "grad_norm": 0.19395531713962555, "learning_rate": 5.576382981159921e-05, "loss": 0.0159, "step": 147020 }, { "epoch": 93.7093690248566, "grad_norm": 0.4551393687725067, "learning_rate": 5.5758354953765426e-05, "loss": 0.0269, "step": 147030 }, { "epoch": 93.7157425111536, "grad_norm": 0.5933775305747986, "learning_rate": 5.575288002596221e-05, "loss": 0.0144, "step": 147040 }, { "epoch": 93.7221159974506, "grad_norm": 0.40691158175468445, "learning_rate": 5.574740502825604e-05, "loss": 0.0161, "step": 147050 }, { "epoch": 93.72848948374761, "grad_norm": 0.3654322028160095, "learning_rate": 5.57419299607135e-05, "loss": 0.0125, "step": 147060 }, { "epoch": 93.73486297004462, "grad_norm": 0.33925312757492065, "learning_rate": 5.573645482340107e-05, "loss": 0.0199, "step": 147070 }, { "epoch": 93.74123645634162, "grad_norm": 0.3975124955177307, "learning_rate": 5.57309796163853e-05, "loss": 0.0196, "step": 147080 }, { "epoch": 93.74760994263862, "grad_norm": 0.280232310295105, "learning_rate": 5.5725504339732736e-05, "loss": 0.0111, "step": 147090 }, { "epoch": 93.75398342893563, "grad_norm": 0.23695294559001923, "learning_rate": 5.5720028993509865e-05, "loss": 0.0106, "step": 147100 }, { "epoch": 93.76035691523263, "grad_norm": 0.23087525367736816, "learning_rate": 5.571455357778326e-05, "loss": 0.0105, "step": 147110 }, { "epoch": 93.76673040152964, "grad_norm": 0.43688225746154785, "learning_rate": 5.570907809261942e-05, "loss": 0.0114, "step": 147120 }, { "epoch": 93.77310388782664, "grad_norm": 0.3735135793685913, "learning_rate": 5.5703602538084895e-05, "loss": 0.0199, "step": 147130 }, { "epoch": 93.77947737412364, "grad_norm": 0.2742200493812561, "learning_rate": 5.569812691424622e-05, "loss": 0.009, "step": 147140 }, { "epoch": 93.78585086042065, "grad_norm": 0.163246750831604, "learning_rate": 5.5692651221169914e-05, "loss": 0.0106, "step": 147150 }, { "epoch": 93.79222434671766, "grad_norm": 0.2281038612127304, "learning_rate": 5.568717545892253e-05, "loss": 0.0173, "step": 147160 }, { "epoch": 93.79859783301465, "grad_norm": 0.2654304504394531, "learning_rate": 5.568169962757058e-05, "loss": 0.0113, "step": 147170 }, { "epoch": 93.80497131931166, "grad_norm": 0.2713688313961029, "learning_rate": 5.5676223727180635e-05, "loss": 0.0082, "step": 147180 }, { "epoch": 93.81134480560867, "grad_norm": 0.42497652769088745, "learning_rate": 5.56707477578192e-05, "loss": 0.01, "step": 147190 }, { "epoch": 93.81771829190568, "grad_norm": 0.2189909666776657, "learning_rate": 5.5665271719552826e-05, "loss": 0.0177, "step": 147200 }, { "epoch": 93.82409177820267, "grad_norm": 0.1999620646238327, "learning_rate": 5.565979561244805e-05, "loss": 0.0178, "step": 147210 }, { "epoch": 93.83046526449968, "grad_norm": 0.30896949768066406, "learning_rate": 5.5654319436571425e-05, "loss": 0.0122, "step": 147220 }, { "epoch": 93.83683875079669, "grad_norm": 0.13463763892650604, "learning_rate": 5.5648843191989476e-05, "loss": 0.0142, "step": 147230 }, { "epoch": 93.8432122370937, "grad_norm": 0.36026543378829956, "learning_rate": 5.5643366878768734e-05, "loss": 0.0213, "step": 147240 }, { "epoch": 93.84958572339069, "grad_norm": 0.26850783824920654, "learning_rate": 5.5637890496975776e-05, "loss": 0.015, "step": 147250 }, { "epoch": 93.8559592096877, "grad_norm": 0.5835054516792297, "learning_rate": 5.5632414046677104e-05, "loss": 0.0162, "step": 147260 }, { "epoch": 93.8623326959847, "grad_norm": 0.2554718852043152, "learning_rate": 5.562693752793931e-05, "loss": 0.0179, "step": 147270 }, { "epoch": 93.86870618228171, "grad_norm": 0.22960048913955688, "learning_rate": 5.562146094082888e-05, "loss": 0.0223, "step": 147280 }, { "epoch": 93.8750796685787, "grad_norm": 0.4155656099319458, "learning_rate": 5.5615984285412414e-05, "loss": 0.0252, "step": 147290 }, { "epoch": 93.88145315487571, "grad_norm": 0.3126201629638672, "learning_rate": 5.561050756175644e-05, "loss": 0.0128, "step": 147300 }, { "epoch": 93.88782664117272, "grad_norm": 0.4329872727394104, "learning_rate": 5.560503076992749e-05, "loss": 0.0155, "step": 147310 }, { "epoch": 93.89420012746973, "grad_norm": 0.3621690273284912, "learning_rate": 5.559955390999213e-05, "loss": 0.0121, "step": 147320 }, { "epoch": 93.90057361376672, "grad_norm": 0.3472231328487396, "learning_rate": 5.5594076982016895e-05, "loss": 0.0103, "step": 147330 }, { "epoch": 93.90694710006373, "grad_norm": 0.26024600863456726, "learning_rate": 5.558859998606836e-05, "loss": 0.0133, "step": 147340 }, { "epoch": 93.91332058636074, "grad_norm": 0.2420089989900589, "learning_rate": 5.558312292221304e-05, "loss": 0.0101, "step": 147350 }, { "epoch": 93.91969407265775, "grad_norm": 0.5097218155860901, "learning_rate": 5.557764579051752e-05, "loss": 0.0235, "step": 147360 }, { "epoch": 93.92606755895474, "grad_norm": 0.3041343688964844, "learning_rate": 5.5572168591048324e-05, "loss": 0.0125, "step": 147370 }, { "epoch": 93.93244104525175, "grad_norm": 0.4220650792121887, "learning_rate": 5.556669132387203e-05, "loss": 0.0155, "step": 147380 }, { "epoch": 93.93881453154876, "grad_norm": 0.3493608832359314, "learning_rate": 5.556121398905518e-05, "loss": 0.0163, "step": 147390 }, { "epoch": 93.94518801784577, "grad_norm": 0.33359768986701965, "learning_rate": 5.555573658666432e-05, "loss": 0.0133, "step": 147400 }, { "epoch": 93.95156150414276, "grad_norm": 0.415194571018219, "learning_rate": 5.5550259116766037e-05, "loss": 0.0152, "step": 147410 }, { "epoch": 93.95793499043977, "grad_norm": 0.4668678939342499, "learning_rate": 5.554478157942684e-05, "loss": 0.0255, "step": 147420 }, { "epoch": 93.96430847673678, "grad_norm": 0.28652423620224, "learning_rate": 5.553930397471332e-05, "loss": 0.011, "step": 147430 }, { "epoch": 93.97068196303378, "grad_norm": 0.3434980809688568, "learning_rate": 5.5533826302692026e-05, "loss": 0.0158, "step": 147440 }, { "epoch": 93.97705544933078, "grad_norm": 0.34546107053756714, "learning_rate": 5.5528348563429524e-05, "loss": 0.0134, "step": 147450 }, { "epoch": 93.98342893562778, "grad_norm": 0.26357752084732056, "learning_rate": 5.5522870756992364e-05, "loss": 0.0089, "step": 147460 }, { "epoch": 93.9898024219248, "grad_norm": 0.35534146428108215, "learning_rate": 5.551739288344711e-05, "loss": 0.0127, "step": 147470 }, { "epoch": 93.9961759082218, "grad_norm": 0.3244878947734833, "learning_rate": 5.5511914942860323e-05, "loss": 0.0243, "step": 147480 }, { "epoch": 94.00254939451881, "grad_norm": 0.19800378382205963, "learning_rate": 5.550643693529856e-05, "loss": 0.0055, "step": 147490 }, { "epoch": 94.0089228808158, "grad_norm": 0.23249919712543488, "learning_rate": 5.550095886082839e-05, "loss": 0.0103, "step": 147500 }, { "epoch": 94.01529636711281, "grad_norm": 0.1489308774471283, "learning_rate": 5.549548071951638e-05, "loss": 0.0109, "step": 147510 }, { "epoch": 94.02166985340982, "grad_norm": 0.2706194221973419, "learning_rate": 5.549000251142909e-05, "loss": 0.0188, "step": 147520 }, { "epoch": 94.02804333970683, "grad_norm": 0.25989216566085815, "learning_rate": 5.548452423663309e-05, "loss": 0.0117, "step": 147530 }, { "epoch": 94.03441682600382, "grad_norm": 0.4431345462799072, "learning_rate": 5.5479045895194934e-05, "loss": 0.0165, "step": 147540 }, { "epoch": 94.04079031230083, "grad_norm": 0.22554750740528107, "learning_rate": 5.54735674871812e-05, "loss": 0.0087, "step": 147550 }, { "epoch": 94.04716379859784, "grad_norm": 0.20739416778087616, "learning_rate": 5.5468089012658455e-05, "loss": 0.0109, "step": 147560 }, { "epoch": 94.05353728489484, "grad_norm": 0.3964749872684479, "learning_rate": 5.5462610471693255e-05, "loss": 0.0091, "step": 147570 }, { "epoch": 94.05991077119184, "grad_norm": 0.3225475251674652, "learning_rate": 5.5457131864352196e-05, "loss": 0.017, "step": 147580 }, { "epoch": 94.06628425748885, "grad_norm": 0.19156919419765472, "learning_rate": 5.5451653190701826e-05, "loss": 0.0245, "step": 147590 }, { "epoch": 94.07265774378585, "grad_norm": 0.3057920038700104, "learning_rate": 5.544617445080873e-05, "loss": 0.0127, "step": 147600 }, { "epoch": 94.07903123008286, "grad_norm": 0.2883702218532562, "learning_rate": 5.544069564473946e-05, "loss": 0.0065, "step": 147610 }, { "epoch": 94.08540471637986, "grad_norm": 0.22612227499485016, "learning_rate": 5.54352167725606e-05, "loss": 0.0113, "step": 147620 }, { "epoch": 94.09177820267686, "grad_norm": 0.4367998242378235, "learning_rate": 5.542973783433873e-05, "loss": 0.0109, "step": 147630 }, { "epoch": 94.09815168897387, "grad_norm": 0.26230156421661377, "learning_rate": 5.542425883014043e-05, "loss": 0.0154, "step": 147640 }, { "epoch": 94.10452517527088, "grad_norm": 0.2477300465106964, "learning_rate": 5.541877976003225e-05, "loss": 0.0152, "step": 147650 }, { "epoch": 94.11089866156787, "grad_norm": 0.32599782943725586, "learning_rate": 5.541330062408079e-05, "loss": 0.0171, "step": 147660 }, { "epoch": 94.11727214786488, "grad_norm": 0.3691728413105011, "learning_rate": 5.54078214223526e-05, "loss": 0.0117, "step": 147670 }, { "epoch": 94.12364563416189, "grad_norm": 0.2239648550748825, "learning_rate": 5.540234215491429e-05, "loss": 0.017, "step": 147680 }, { "epoch": 94.1300191204589, "grad_norm": 0.39215850830078125, "learning_rate": 5.539686282183242e-05, "loss": 0.014, "step": 147690 }, { "epoch": 94.13639260675589, "grad_norm": 0.3220116198062897, "learning_rate": 5.539138342317357e-05, "loss": 0.0158, "step": 147700 }, { "epoch": 94.1427660930529, "grad_norm": 0.35392430424690247, "learning_rate": 5.5385903959004316e-05, "loss": 0.0161, "step": 147710 }, { "epoch": 94.1491395793499, "grad_norm": 1.4880084991455078, "learning_rate": 5.5380424429391245e-05, "loss": 0.0182, "step": 147720 }, { "epoch": 94.15551306564691, "grad_norm": 0.7110500335693359, "learning_rate": 5.5374944834400946e-05, "loss": 0.03, "step": 147730 }, { "epoch": 94.16188655194391, "grad_norm": 0.27056410908699036, "learning_rate": 5.5369465174099975e-05, "loss": 0.0086, "step": 147740 }, { "epoch": 94.16826003824092, "grad_norm": 0.24239632487297058, "learning_rate": 5.536398544855494e-05, "loss": 0.017, "step": 147750 }, { "epoch": 94.17463352453792, "grad_norm": 0.21519123017787933, "learning_rate": 5.535850565783243e-05, "loss": 0.0139, "step": 147760 }, { "epoch": 94.18100701083493, "grad_norm": 0.36156922578811646, "learning_rate": 5.5353025801999e-05, "loss": 0.0099, "step": 147770 }, { "epoch": 94.18738049713193, "grad_norm": 0.41914102435112, "learning_rate": 5.534754588112127e-05, "loss": 0.0129, "step": 147780 }, { "epoch": 94.19375398342893, "grad_norm": 0.2641972303390503, "learning_rate": 5.534206589526579e-05, "loss": 0.0102, "step": 147790 }, { "epoch": 94.20012746972594, "grad_norm": 0.4788120985031128, "learning_rate": 5.533658584449918e-05, "loss": 0.0211, "step": 147800 }, { "epoch": 94.20650095602295, "grad_norm": 0.7230458855628967, "learning_rate": 5.5331105728888e-05, "loss": 0.0083, "step": 147810 }, { "epoch": 94.21287444231994, "grad_norm": 0.25670725107192993, "learning_rate": 5.532562554849886e-05, "loss": 0.0137, "step": 147820 }, { "epoch": 94.21924792861695, "grad_norm": 0.27952322363853455, "learning_rate": 5.532014530339834e-05, "loss": 0.0142, "step": 147830 }, { "epoch": 94.22562141491396, "grad_norm": 0.3252853751182556, "learning_rate": 5.531466499365303e-05, "loss": 0.0115, "step": 147840 }, { "epoch": 94.23199490121097, "grad_norm": 0.130824014544487, "learning_rate": 5.5309184619329524e-05, "loss": 0.0151, "step": 147850 }, { "epoch": 94.23836838750796, "grad_norm": 1.048818588256836, "learning_rate": 5.530370418049442e-05, "loss": 0.0129, "step": 147860 }, { "epoch": 94.24474187380497, "grad_norm": 0.28404906392097473, "learning_rate": 5.52982236772143e-05, "loss": 0.0143, "step": 147870 }, { "epoch": 94.25111536010198, "grad_norm": 0.4501838684082031, "learning_rate": 5.529274310955575e-05, "loss": 0.014, "step": 147880 }, { "epoch": 94.25748884639899, "grad_norm": 0.28795942664146423, "learning_rate": 5.5287262477585376e-05, "loss": 0.0115, "step": 147890 }, { "epoch": 94.26386233269598, "grad_norm": 0.3622882068157196, "learning_rate": 5.528178178136978e-05, "loss": 0.0103, "step": 147900 }, { "epoch": 94.27023581899299, "grad_norm": 0.3344901204109192, "learning_rate": 5.527630102097554e-05, "loss": 0.0179, "step": 147910 }, { "epoch": 94.27660930529, "grad_norm": 0.5198674201965332, "learning_rate": 5.5270820196469266e-05, "loss": 0.0235, "step": 147920 }, { "epoch": 94.282982791587, "grad_norm": 0.7023051977157593, "learning_rate": 5.526533930791755e-05, "loss": 0.0125, "step": 147930 }, { "epoch": 94.289356277884, "grad_norm": 0.5925998687744141, "learning_rate": 5.525985835538699e-05, "loss": 0.0137, "step": 147940 }, { "epoch": 94.295729764181, "grad_norm": 0.20604760944843292, "learning_rate": 5.525437733894417e-05, "loss": 0.0185, "step": 147950 }, { "epoch": 94.30210325047801, "grad_norm": 0.1374204456806183, "learning_rate": 5.524889625865572e-05, "loss": 0.0128, "step": 147960 }, { "epoch": 94.30847673677502, "grad_norm": 0.26614484190940857, "learning_rate": 5.5243415114588224e-05, "loss": 0.016, "step": 147970 }, { "epoch": 94.31485022307201, "grad_norm": 0.5109741687774658, "learning_rate": 5.523793390680829e-05, "loss": 0.0198, "step": 147980 }, { "epoch": 94.32122370936902, "grad_norm": 0.44019779562950134, "learning_rate": 5.523245263538249e-05, "loss": 0.0077, "step": 147990 }, { "epoch": 94.32759719566603, "grad_norm": 0.7418987154960632, "learning_rate": 5.5226971300377486e-05, "loss": 0.0116, "step": 148000 }, { "epoch": 94.33397068196304, "grad_norm": 0.5081104636192322, "learning_rate": 5.5221489901859814e-05, "loss": 0.0207, "step": 148010 }, { "epoch": 94.34034416826003, "grad_norm": 0.19987323880195618, "learning_rate": 5.521600843989613e-05, "loss": 0.0111, "step": 148020 }, { "epoch": 94.34671765455704, "grad_norm": 0.33192336559295654, "learning_rate": 5.521052691455301e-05, "loss": 0.0125, "step": 148030 }, { "epoch": 94.35309114085405, "grad_norm": 0.4478926360607147, "learning_rate": 5.5205045325897065e-05, "loss": 0.0138, "step": 148040 }, { "epoch": 94.35946462715106, "grad_norm": 0.22463195025920868, "learning_rate": 5.51995636739949e-05, "loss": 0.0205, "step": 148050 }, { "epoch": 94.36583811344805, "grad_norm": 0.20886953175067902, "learning_rate": 5.5194081958913134e-05, "loss": 0.0125, "step": 148060 }, { "epoch": 94.37221159974506, "grad_norm": 0.19657589495182037, "learning_rate": 5.518860018071838e-05, "loss": 0.0118, "step": 148070 }, { "epoch": 94.37858508604207, "grad_norm": 0.385936975479126, "learning_rate": 5.518311833947721e-05, "loss": 0.0132, "step": 148080 }, { "epoch": 94.38495857233907, "grad_norm": 0.5278751850128174, "learning_rate": 5.5177636435256286e-05, "loss": 0.0166, "step": 148090 }, { "epoch": 94.39133205863607, "grad_norm": 0.21838858723640442, "learning_rate": 5.517215446812217e-05, "loss": 0.0079, "step": 148100 }, { "epoch": 94.39770554493307, "grad_norm": 0.2558227777481079, "learning_rate": 5.516667243814151e-05, "loss": 0.0196, "step": 148110 }, { "epoch": 94.40407903123008, "grad_norm": 0.4287673234939575, "learning_rate": 5.51611903453809e-05, "loss": 0.0192, "step": 148120 }, { "epoch": 94.41045251752709, "grad_norm": 0.3997235894203186, "learning_rate": 5.515570818990693e-05, "loss": 0.011, "step": 148130 }, { "epoch": 94.4168260038241, "grad_norm": 0.30725377798080444, "learning_rate": 5.5150225971786265e-05, "loss": 0.0234, "step": 148140 }, { "epoch": 94.42319949012109, "grad_norm": 0.397394597530365, "learning_rate": 5.514474369108548e-05, "loss": 0.0095, "step": 148150 }, { "epoch": 94.4295729764181, "grad_norm": 0.14131593704223633, "learning_rate": 5.513926134787121e-05, "loss": 0.01, "step": 148160 }, { "epoch": 94.43594646271511, "grad_norm": 0.34349629282951355, "learning_rate": 5.5133778942210045e-05, "loss": 0.0121, "step": 148170 }, { "epoch": 94.44231994901212, "grad_norm": 0.469112753868103, "learning_rate": 5.512829647416864e-05, "loss": 0.0159, "step": 148180 }, { "epoch": 94.44869343530911, "grad_norm": 0.3486471474170685, "learning_rate": 5.512281394381358e-05, "loss": 0.0215, "step": 148190 }, { "epoch": 94.45506692160612, "grad_norm": 0.2198377251625061, "learning_rate": 5.5117331351211485e-05, "loss": 0.0072, "step": 148200 }, { "epoch": 94.46144040790313, "grad_norm": 0.46365290880203247, "learning_rate": 5.5111848696429e-05, "loss": 0.0193, "step": 148210 }, { "epoch": 94.46781389420013, "grad_norm": 0.2521294355392456, "learning_rate": 5.5106365979532714e-05, "loss": 0.0113, "step": 148220 }, { "epoch": 94.47418738049713, "grad_norm": 0.49250829219818115, "learning_rate": 5.5100883200589274e-05, "loss": 0.0123, "step": 148230 }, { "epoch": 94.48056086679414, "grad_norm": 0.2167847901582718, "learning_rate": 5.5095400359665274e-05, "loss": 0.007, "step": 148240 }, { "epoch": 94.48693435309114, "grad_norm": 0.1605955809354782, "learning_rate": 5.5089917456827355e-05, "loss": 0.0225, "step": 148250 }, { "epoch": 94.49330783938815, "grad_norm": 0.23764413595199585, "learning_rate": 5.5084434492142125e-05, "loss": 0.0091, "step": 148260 }, { "epoch": 94.49968132568515, "grad_norm": 0.28203296661376953, "learning_rate": 5.507895146567622e-05, "loss": 0.0117, "step": 148270 }, { "epoch": 94.50605481198215, "grad_norm": 0.2948940396308899, "learning_rate": 5.5073468377496265e-05, "loss": 0.0133, "step": 148280 }, { "epoch": 94.51242829827916, "grad_norm": 0.5270401835441589, "learning_rate": 5.506798522766887e-05, "loss": 0.0114, "step": 148290 }, { "epoch": 94.51880178457617, "grad_norm": 0.31761911511421204, "learning_rate": 5.506250201626068e-05, "loss": 0.0119, "step": 148300 }, { "epoch": 94.52517527087316, "grad_norm": 0.4103859066963196, "learning_rate": 5.505701874333831e-05, "loss": 0.0102, "step": 148310 }, { "epoch": 94.53154875717017, "grad_norm": 0.2289949208498001, "learning_rate": 5.5051535408968366e-05, "loss": 0.0186, "step": 148320 }, { "epoch": 94.53792224346718, "grad_norm": 0.28073060512542725, "learning_rate": 5.504605201321752e-05, "loss": 0.0158, "step": 148330 }, { "epoch": 94.54429572976419, "grad_norm": 0.3429391384124756, "learning_rate": 5.504056855615237e-05, "loss": 0.019, "step": 148340 }, { "epoch": 94.55066921606118, "grad_norm": 0.20877911150455475, "learning_rate": 5.503508503783956e-05, "loss": 0.0173, "step": 148350 }, { "epoch": 94.55704270235819, "grad_norm": 0.29286444187164307, "learning_rate": 5.5029601458345705e-05, "loss": 0.0172, "step": 148360 }, { "epoch": 94.5634161886552, "grad_norm": 0.31284311413764954, "learning_rate": 5.502411781773745e-05, "loss": 0.0107, "step": 148370 }, { "epoch": 94.5697896749522, "grad_norm": 0.25922414660453796, "learning_rate": 5.5018634116081415e-05, "loss": 0.0125, "step": 148380 }, { "epoch": 94.5761631612492, "grad_norm": 0.452631413936615, "learning_rate": 5.501315035344423e-05, "loss": 0.0214, "step": 148390 }, { "epoch": 94.5825366475462, "grad_norm": 0.19491618871688843, "learning_rate": 5.500766652989254e-05, "loss": 0.0097, "step": 148400 }, { "epoch": 94.58891013384321, "grad_norm": 0.3739275336265564, "learning_rate": 5.500218264549298e-05, "loss": 0.0088, "step": 148410 }, { "epoch": 94.59528362014022, "grad_norm": 0.6323074102401733, "learning_rate": 5.499669870031218e-05, "loss": 0.0219, "step": 148420 }, { "epoch": 94.60165710643722, "grad_norm": 0.6122003197669983, "learning_rate": 5.499121469441677e-05, "loss": 0.0129, "step": 148430 }, { "epoch": 94.60803059273422, "grad_norm": 0.46616896986961365, "learning_rate": 5.498573062787339e-05, "loss": 0.0151, "step": 148440 }, { "epoch": 94.61440407903123, "grad_norm": 0.44036102294921875, "learning_rate": 5.4980246500748665e-05, "loss": 0.0138, "step": 148450 }, { "epoch": 94.62077756532824, "grad_norm": 0.5817452073097229, "learning_rate": 5.497476231310926e-05, "loss": 0.0167, "step": 148460 }, { "epoch": 94.62715105162523, "grad_norm": 0.6669417023658752, "learning_rate": 5.496927806502178e-05, "loss": 0.0173, "step": 148470 }, { "epoch": 94.63352453792224, "grad_norm": 0.2698137164115906, "learning_rate": 5.496379375655289e-05, "loss": 0.0121, "step": 148480 }, { "epoch": 94.63989802421925, "grad_norm": 0.22615528106689453, "learning_rate": 5.495830938776921e-05, "loss": 0.0124, "step": 148490 }, { "epoch": 94.64627151051626, "grad_norm": 0.18765072524547577, "learning_rate": 5.49528249587374e-05, "loss": 0.0178, "step": 148500 }, { "epoch": 94.65264499681325, "grad_norm": 0.22492346167564392, "learning_rate": 5.494734046952409e-05, "loss": 0.0142, "step": 148510 }, { "epoch": 94.65901848311026, "grad_norm": 0.44900962710380554, "learning_rate": 5.494185592019592e-05, "loss": 0.0138, "step": 148520 }, { "epoch": 94.66539196940727, "grad_norm": 0.4493436813354492, "learning_rate": 5.493637131081953e-05, "loss": 0.0142, "step": 148530 }, { "epoch": 94.67176545570427, "grad_norm": 0.37618401646614075, "learning_rate": 5.493088664146156e-05, "loss": 0.0149, "step": 148540 }, { "epoch": 94.67813894200127, "grad_norm": 0.45867666602134705, "learning_rate": 5.492540191218869e-05, "loss": 0.0182, "step": 148550 }, { "epoch": 94.68451242829828, "grad_norm": 0.4445740580558777, "learning_rate": 5.491991712306751e-05, "loss": 0.0127, "step": 148560 }, { "epoch": 94.69088591459528, "grad_norm": 0.28907260298728943, "learning_rate": 5.491443227416471e-05, "loss": 0.0174, "step": 148570 }, { "epoch": 94.69725940089229, "grad_norm": 0.35643860697746277, "learning_rate": 5.4908947365546895e-05, "loss": 0.0249, "step": 148580 }, { "epoch": 94.70363288718929, "grad_norm": 0.19916582107543945, "learning_rate": 5.4903462397280756e-05, "loss": 0.0204, "step": 148590 }, { "epoch": 94.7100063734863, "grad_norm": 0.3528454303741455, "learning_rate": 5.489797736943292e-05, "loss": 0.0191, "step": 148600 }, { "epoch": 94.7163798597833, "grad_norm": 0.5787343978881836, "learning_rate": 5.4892492282070016e-05, "loss": 0.0104, "step": 148610 }, { "epoch": 94.72275334608031, "grad_norm": 0.36331266164779663, "learning_rate": 5.4887007135258724e-05, "loss": 0.015, "step": 148620 }, { "epoch": 94.7291268323773, "grad_norm": 0.34050503373146057, "learning_rate": 5.488152192906567e-05, "loss": 0.0139, "step": 148630 }, { "epoch": 94.73550031867431, "grad_norm": 0.3950363099575043, "learning_rate": 5.4876036663557526e-05, "loss": 0.0124, "step": 148640 }, { "epoch": 94.74187380497132, "grad_norm": 0.22955359518527985, "learning_rate": 5.4870551338800936e-05, "loss": 0.0115, "step": 148650 }, { "epoch": 94.74824729126833, "grad_norm": 0.2201840728521347, "learning_rate": 5.486506595486254e-05, "loss": 0.0107, "step": 148660 }, { "epoch": 94.75462077756532, "grad_norm": 0.26805540919303894, "learning_rate": 5.485958051180902e-05, "loss": 0.0131, "step": 148670 }, { "epoch": 94.76099426386233, "grad_norm": 0.4261666238307953, "learning_rate": 5.485409500970697e-05, "loss": 0.0091, "step": 148680 }, { "epoch": 94.76736775015934, "grad_norm": 0.3640148341655731, "learning_rate": 5.484860944862311e-05, "loss": 0.0175, "step": 148690 }, { "epoch": 94.77374123645635, "grad_norm": 0.46711912751197815, "learning_rate": 5.484312382862407e-05, "loss": 0.0173, "step": 148700 }, { "epoch": 94.78011472275334, "grad_norm": 0.22272107005119324, "learning_rate": 5.4837638149776495e-05, "loss": 0.0198, "step": 148710 }, { "epoch": 94.78648820905035, "grad_norm": 0.19235128164291382, "learning_rate": 5.4832152412147044e-05, "loss": 0.0137, "step": 148720 }, { "epoch": 94.79286169534736, "grad_norm": 0.4468098282814026, "learning_rate": 5.482666661580238e-05, "loss": 0.0175, "step": 148730 }, { "epoch": 94.79923518164436, "grad_norm": 0.341447114944458, "learning_rate": 5.4821180760809155e-05, "loss": 0.0157, "step": 148740 }, { "epoch": 94.80560866794136, "grad_norm": 0.3378106951713562, "learning_rate": 5.481569484723405e-05, "loss": 0.0167, "step": 148750 }, { "epoch": 94.81198215423836, "grad_norm": 0.47642233967781067, "learning_rate": 5.4810208875143695e-05, "loss": 0.0195, "step": 148760 }, { "epoch": 94.81835564053537, "grad_norm": 0.20794612169265747, "learning_rate": 5.480472284460476e-05, "loss": 0.0101, "step": 148770 }, { "epoch": 94.82472912683238, "grad_norm": 0.279787540435791, "learning_rate": 5.4799236755683914e-05, "loss": 0.0198, "step": 148780 }, { "epoch": 94.83110261312939, "grad_norm": 0.2321513146162033, "learning_rate": 5.479375060844779e-05, "loss": 0.0124, "step": 148790 }, { "epoch": 94.83747609942638, "grad_norm": 0.3568318486213684, "learning_rate": 5.47882644029631e-05, "loss": 0.0133, "step": 148800 }, { "epoch": 94.84384958572339, "grad_norm": 0.346884161233902, "learning_rate": 5.4782778139296445e-05, "loss": 0.0139, "step": 148810 }, { "epoch": 94.8502230720204, "grad_norm": 0.2635842561721802, "learning_rate": 5.4777291817514544e-05, "loss": 0.0175, "step": 148820 }, { "epoch": 94.8565965583174, "grad_norm": 0.5264608263969421, "learning_rate": 5.477180543768403e-05, "loss": 0.0146, "step": 148830 }, { "epoch": 94.8629700446144, "grad_norm": 0.692413866519928, "learning_rate": 5.4766318999871566e-05, "loss": 0.0151, "step": 148840 }, { "epoch": 94.86934353091141, "grad_norm": 0.612583339214325, "learning_rate": 5.476083250414383e-05, "loss": 0.0288, "step": 148850 }, { "epoch": 94.87571701720842, "grad_norm": 0.3315301835536957, "learning_rate": 5.475534595056748e-05, "loss": 0.01, "step": 148860 }, { "epoch": 94.88209050350542, "grad_norm": 0.2904511094093323, "learning_rate": 5.474985933920921e-05, "loss": 0.0124, "step": 148870 }, { "epoch": 94.88846398980242, "grad_norm": 0.6543716192245483, "learning_rate": 5.474437267013563e-05, "loss": 0.0086, "step": 148880 }, { "epoch": 94.89483747609943, "grad_norm": 0.3598393499851227, "learning_rate": 5.4738885943413476e-05, "loss": 0.0152, "step": 148890 }, { "epoch": 94.90121096239643, "grad_norm": 0.5844879150390625, "learning_rate": 5.4733399159109356e-05, "loss": 0.0193, "step": 148900 }, { "epoch": 94.90758444869344, "grad_norm": 0.4949098527431488, "learning_rate": 5.4727912317289985e-05, "loss": 0.0127, "step": 148910 }, { "epoch": 94.91395793499044, "grad_norm": 0.26134559512138367, "learning_rate": 5.472242541802202e-05, "loss": 0.0072, "step": 148920 }, { "epoch": 94.92033142128744, "grad_norm": 0.21850579977035522, "learning_rate": 5.4716938461372106e-05, "loss": 0.0192, "step": 148930 }, { "epoch": 94.92670490758445, "grad_norm": 0.6526776552200317, "learning_rate": 5.471145144740695e-05, "loss": 0.0087, "step": 148940 }, { "epoch": 94.93307839388146, "grad_norm": 0.3279213011264801, "learning_rate": 5.47059643761932e-05, "loss": 0.0094, "step": 148950 }, { "epoch": 94.93945188017845, "grad_norm": 0.27727341651916504, "learning_rate": 5.470047724779756e-05, "loss": 0.0114, "step": 148960 }, { "epoch": 94.94582536647546, "grad_norm": 0.2846534848213196, "learning_rate": 5.469499006228665e-05, "loss": 0.0178, "step": 148970 }, { "epoch": 94.95219885277247, "grad_norm": 0.23814576864242554, "learning_rate": 5.468950281972721e-05, "loss": 0.013, "step": 148980 }, { "epoch": 94.95857233906948, "grad_norm": 0.27495288848876953, "learning_rate": 5.468401552018587e-05, "loss": 0.0101, "step": 148990 }, { "epoch": 94.96494582536647, "grad_norm": 0.6820880770683289, "learning_rate": 5.4678528163729326e-05, "loss": 0.0185, "step": 149000 }, { "epoch": 94.97131931166348, "grad_norm": 0.1700764298439026, "learning_rate": 5.467304075042423e-05, "loss": 0.0108, "step": 149010 }, { "epoch": 94.97769279796049, "grad_norm": 0.2720614969730377, "learning_rate": 5.46675532803373e-05, "loss": 0.0161, "step": 149020 }, { "epoch": 94.9840662842575, "grad_norm": 0.23125040531158447, "learning_rate": 5.466206575353518e-05, "loss": 0.0106, "step": 149030 }, { "epoch": 94.99043977055449, "grad_norm": 0.46922576427459717, "learning_rate": 5.465657817008456e-05, "loss": 0.008, "step": 149040 }, { "epoch": 94.9968132568515, "grad_norm": 0.18791848421096802, "learning_rate": 5.4651090530052105e-05, "loss": 0.0124, "step": 149050 }, { "epoch": 95.0031867431485, "grad_norm": 0.376442015171051, "learning_rate": 5.4645602833504526e-05, "loss": 0.014, "step": 149060 }, { "epoch": 95.00956022944551, "grad_norm": 0.27130743861198425, "learning_rate": 5.464011508050848e-05, "loss": 0.0147, "step": 149070 }, { "epoch": 95.0159337157425, "grad_norm": 0.1544101983308792, "learning_rate": 5.4634627271130667e-05, "loss": 0.0102, "step": 149080 }, { "epoch": 95.02230720203951, "grad_norm": 0.2704155147075653, "learning_rate": 5.4629139405437736e-05, "loss": 0.0125, "step": 149090 }, { "epoch": 95.02868068833652, "grad_norm": 0.25516629219055176, "learning_rate": 5.4623651483496404e-05, "loss": 0.0119, "step": 149100 }, { "epoch": 95.03505417463353, "grad_norm": 0.2663826644420624, "learning_rate": 5.4618163505373335e-05, "loss": 0.0093, "step": 149110 }, { "epoch": 95.04142766093052, "grad_norm": 0.30743229389190674, "learning_rate": 5.461267547113522e-05, "loss": 0.0199, "step": 149120 }, { "epoch": 95.04780114722753, "grad_norm": 0.29381394386291504, "learning_rate": 5.4607187380848746e-05, "loss": 0.0166, "step": 149130 }, { "epoch": 95.05417463352454, "grad_norm": 0.28823691606521606, "learning_rate": 5.4601699234580604e-05, "loss": 0.0108, "step": 149140 }, { "epoch": 95.06054811982155, "grad_norm": 0.2896229922771454, "learning_rate": 5.459621103239746e-05, "loss": 0.0086, "step": 149150 }, { "epoch": 95.06692160611854, "grad_norm": 0.2861909568309784, "learning_rate": 5.459072277436601e-05, "loss": 0.0138, "step": 149160 }, { "epoch": 95.07329509241555, "grad_norm": 0.5005630850791931, "learning_rate": 5.4585234460552955e-05, "loss": 0.0093, "step": 149170 }, { "epoch": 95.07966857871256, "grad_norm": 0.36937305331230164, "learning_rate": 5.457974609102497e-05, "loss": 0.007, "step": 149180 }, { "epoch": 95.08604206500956, "grad_norm": 0.30674371123313904, "learning_rate": 5.457425766584874e-05, "loss": 0.0146, "step": 149190 }, { "epoch": 95.09241555130656, "grad_norm": 0.22086775302886963, "learning_rate": 5.456876918509095e-05, "loss": 0.0068, "step": 149200 }, { "epoch": 95.09878903760357, "grad_norm": 0.39696353673934937, "learning_rate": 5.4563280648818325e-05, "loss": 0.011, "step": 149210 }, { "epoch": 95.10516252390057, "grad_norm": 0.35711950063705444, "learning_rate": 5.4557792057097515e-05, "loss": 0.0089, "step": 149220 }, { "epoch": 95.11153601019758, "grad_norm": 0.14046573638916016, "learning_rate": 5.455230340999523e-05, "loss": 0.0187, "step": 149230 }, { "epoch": 95.11790949649458, "grad_norm": 0.36534738540649414, "learning_rate": 5.454681470757818e-05, "loss": 0.0089, "step": 149240 }, { "epoch": 95.12428298279158, "grad_norm": 0.32343366742134094, "learning_rate": 5.4541325949913e-05, "loss": 0.0104, "step": 149250 }, { "epoch": 95.13065646908859, "grad_norm": 0.5546572804450989, "learning_rate": 5.453583713706646e-05, "loss": 0.0161, "step": 149260 }, { "epoch": 95.1370299553856, "grad_norm": 0.32366976141929626, "learning_rate": 5.453034826910519e-05, "loss": 0.0151, "step": 149270 }, { "epoch": 95.1434034416826, "grad_norm": 0.30678093433380127, "learning_rate": 5.4524859346095926e-05, "loss": 0.0145, "step": 149280 }, { "epoch": 95.1497769279796, "grad_norm": 0.3341193199157715, "learning_rate": 5.4519370368105336e-05, "loss": 0.0153, "step": 149290 }, { "epoch": 95.15615041427661, "grad_norm": 0.3049311637878418, "learning_rate": 5.4513881335200144e-05, "loss": 0.0102, "step": 149300 }, { "epoch": 95.16252390057362, "grad_norm": 0.29205048084259033, "learning_rate": 5.4508392247447014e-05, "loss": 0.0157, "step": 149310 }, { "epoch": 95.16889738687061, "grad_norm": 0.23826740682125092, "learning_rate": 5.4502903104912685e-05, "loss": 0.0126, "step": 149320 }, { "epoch": 95.17527087316762, "grad_norm": 0.25184834003448486, "learning_rate": 5.4497413907663816e-05, "loss": 0.0121, "step": 149330 }, { "epoch": 95.18164435946463, "grad_norm": 0.2282467931509018, "learning_rate": 5.4491924655767126e-05, "loss": 0.0102, "step": 149340 }, { "epoch": 95.18801784576164, "grad_norm": 0.3355931341648102, "learning_rate": 5.448643534928931e-05, "loss": 0.0092, "step": 149350 }, { "epoch": 95.19439133205863, "grad_norm": 0.1604497879743576, "learning_rate": 5.448094598829707e-05, "loss": 0.014, "step": 149360 }, { "epoch": 95.20076481835564, "grad_norm": 0.6062396764755249, "learning_rate": 5.447545657285711e-05, "loss": 0.0165, "step": 149370 }, { "epoch": 95.20713830465264, "grad_norm": 0.1915375143289566, "learning_rate": 5.446996710303611e-05, "loss": 0.0066, "step": 149380 }, { "epoch": 95.21351179094965, "grad_norm": 0.29346534609794617, "learning_rate": 5.446447757890081e-05, "loss": 0.0071, "step": 149390 }, { "epoch": 95.21988527724666, "grad_norm": 0.4728187620639801, "learning_rate": 5.4458988000517885e-05, "loss": 0.0151, "step": 149400 }, { "epoch": 95.22625876354365, "grad_norm": 0.1910751610994339, "learning_rate": 5.445349836795404e-05, "loss": 0.0137, "step": 149410 }, { "epoch": 95.23263224984066, "grad_norm": 0.512949526309967, "learning_rate": 5.4448008681275985e-05, "loss": 0.0111, "step": 149420 }, { "epoch": 95.23900573613767, "grad_norm": 0.3026995062828064, "learning_rate": 5.444251894055042e-05, "loss": 0.0118, "step": 149430 }, { "epoch": 95.24537922243468, "grad_norm": 0.3116975426673889, "learning_rate": 5.443702914584408e-05, "loss": 0.0159, "step": 149440 }, { "epoch": 95.25175270873167, "grad_norm": 0.20224547386169434, "learning_rate": 5.443153929722362e-05, "loss": 0.019, "step": 149450 }, { "epoch": 95.25812619502868, "grad_norm": 0.3210320472717285, "learning_rate": 5.4426049394755786e-05, "loss": 0.0143, "step": 149460 }, { "epoch": 95.26449968132569, "grad_norm": 0.36309534311294556, "learning_rate": 5.442055943850727e-05, "loss": 0.0163, "step": 149470 }, { "epoch": 95.2708731676227, "grad_norm": 0.461465060710907, "learning_rate": 5.441506942854478e-05, "loss": 0.0113, "step": 149480 }, { "epoch": 95.27724665391969, "grad_norm": 0.3632928133010864, "learning_rate": 5.4409579364935035e-05, "loss": 0.0127, "step": 149490 }, { "epoch": 95.2836201402167, "grad_norm": 0.31910446286201477, "learning_rate": 5.4404089247744736e-05, "loss": 0.0108, "step": 149500 }, { "epoch": 95.2899936265137, "grad_norm": 0.25607970356941223, "learning_rate": 5.439859907704059e-05, "loss": 0.0158, "step": 149510 }, { "epoch": 95.29636711281071, "grad_norm": 0.20795413851737976, "learning_rate": 5.439310885288932e-05, "loss": 0.0096, "step": 149520 }, { "epoch": 95.30274059910771, "grad_norm": 0.492428719997406, "learning_rate": 5.438761857535762e-05, "loss": 0.0213, "step": 149530 }, { "epoch": 95.30911408540472, "grad_norm": 0.25877267122268677, "learning_rate": 5.438212824451221e-05, "loss": 0.0249, "step": 149540 }, { "epoch": 95.31548757170172, "grad_norm": 0.38214582204818726, "learning_rate": 5.437663786041982e-05, "loss": 0.0176, "step": 149550 }, { "epoch": 95.32186105799873, "grad_norm": 0.34114670753479004, "learning_rate": 5.437114742314714e-05, "loss": 0.0096, "step": 149560 }, { "epoch": 95.32823454429573, "grad_norm": 0.4007178843021393, "learning_rate": 5.436565693276089e-05, "loss": 0.0102, "step": 149570 }, { "epoch": 95.33460803059273, "grad_norm": 0.23746730387210846, "learning_rate": 5.436016638932779e-05, "loss": 0.0101, "step": 149580 }, { "epoch": 95.34098151688974, "grad_norm": 0.40467655658721924, "learning_rate": 5.435467579291454e-05, "loss": 0.0139, "step": 149590 }, { "epoch": 95.34735500318675, "grad_norm": 0.15319301187992096, "learning_rate": 5.434918514358788e-05, "loss": 0.0127, "step": 149600 }, { "epoch": 95.35372848948374, "grad_norm": 0.2824612259864807, "learning_rate": 5.4343694441414507e-05, "loss": 0.024, "step": 149610 }, { "epoch": 95.36010197578075, "grad_norm": 0.3946649432182312, "learning_rate": 5.433820368646115e-05, "loss": 0.0195, "step": 149620 }, { "epoch": 95.36647546207776, "grad_norm": 0.3371158838272095, "learning_rate": 5.433271287879451e-05, "loss": 0.0142, "step": 149630 }, { "epoch": 95.37284894837477, "grad_norm": 0.29002615809440613, "learning_rate": 5.432722201848134e-05, "loss": 0.016, "step": 149640 }, { "epoch": 95.37922243467176, "grad_norm": 0.19360649585723877, "learning_rate": 5.432173110558834e-05, "loss": 0.01, "step": 149650 }, { "epoch": 95.38559592096877, "grad_norm": 0.30188387632369995, "learning_rate": 5.4316240140182194e-05, "loss": 0.0098, "step": 149660 }, { "epoch": 95.39196940726578, "grad_norm": 0.43326354026794434, "learning_rate": 5.4310749122329686e-05, "loss": 0.0129, "step": 149670 }, { "epoch": 95.39834289356278, "grad_norm": 0.3087943494319916, "learning_rate": 5.430525805209749e-05, "loss": 0.0149, "step": 149680 }, { "epoch": 95.40471637985978, "grad_norm": 0.2337004691362381, "learning_rate": 5.4299766929552345e-05, "loss": 0.0146, "step": 149690 }, { "epoch": 95.41108986615679, "grad_norm": 0.5700981616973877, "learning_rate": 5.429427575476097e-05, "loss": 0.0194, "step": 149700 }, { "epoch": 95.4174633524538, "grad_norm": 0.1887144297361374, "learning_rate": 5.4288784527790096e-05, "loss": 0.0175, "step": 149710 }, { "epoch": 95.4238368387508, "grad_norm": 0.28061339259147644, "learning_rate": 5.4283293248706434e-05, "loss": 0.0093, "step": 149720 }, { "epoch": 95.4302103250478, "grad_norm": 0.1816241592168808, "learning_rate": 5.427780191757672e-05, "loss": 0.0126, "step": 149730 }, { "epoch": 95.4365838113448, "grad_norm": 0.55616295337677, "learning_rate": 5.427231053446767e-05, "loss": 0.0176, "step": 149740 }, { "epoch": 95.44295729764181, "grad_norm": 0.37318408489227295, "learning_rate": 5.426681909944602e-05, "loss": 0.0137, "step": 149750 }, { "epoch": 95.44933078393882, "grad_norm": 0.3433413505554199, "learning_rate": 5.4261327612578485e-05, "loss": 0.0135, "step": 149760 }, { "epoch": 95.45570427023581, "grad_norm": 0.29974570870399475, "learning_rate": 5.425583607393181e-05, "loss": 0.0107, "step": 149770 }, { "epoch": 95.46207775653282, "grad_norm": 0.3011167645454407, "learning_rate": 5.425034448357268e-05, "loss": 0.0111, "step": 149780 }, { "epoch": 95.46845124282983, "grad_norm": 0.5191665887832642, "learning_rate": 5.424485284156787e-05, "loss": 0.0256, "step": 149790 }, { "epoch": 95.47482472912684, "grad_norm": 0.30862298607826233, "learning_rate": 5.423936114798409e-05, "loss": 0.0116, "step": 149800 }, { "epoch": 95.48119821542383, "grad_norm": 0.36959734559059143, "learning_rate": 5.423386940288807e-05, "loss": 0.0127, "step": 149810 }, { "epoch": 95.48757170172084, "grad_norm": 0.4745253324508667, "learning_rate": 5.422837760634654e-05, "loss": 0.0175, "step": 149820 }, { "epoch": 95.49394518801785, "grad_norm": 0.2764197289943695, "learning_rate": 5.4222885758426224e-05, "loss": 0.0089, "step": 149830 }, { "epoch": 95.50031867431485, "grad_norm": 0.27762097120285034, "learning_rate": 5.4217393859193855e-05, "loss": 0.0119, "step": 149840 }, { "epoch": 95.50669216061185, "grad_norm": 0.16980919241905212, "learning_rate": 5.421190190871618e-05, "loss": 0.0117, "step": 149850 }, { "epoch": 95.51306564690886, "grad_norm": 0.3427116572856903, "learning_rate": 5.420640990705991e-05, "loss": 0.0115, "step": 149860 }, { "epoch": 95.51943913320586, "grad_norm": 0.20824280381202698, "learning_rate": 5.42009178542918e-05, "loss": 0.0061, "step": 149870 }, { "epoch": 95.52581261950287, "grad_norm": 0.6053286790847778, "learning_rate": 5.4195425750478565e-05, "loss": 0.0152, "step": 149880 }, { "epoch": 95.53218610579987, "grad_norm": 0.48166704177856445, "learning_rate": 5.4189933595686936e-05, "loss": 0.0198, "step": 149890 }, { "epoch": 95.53855959209687, "grad_norm": 0.2955203056335449, "learning_rate": 5.4184441389983674e-05, "loss": 0.012, "step": 149900 }, { "epoch": 95.54493307839388, "grad_norm": 0.2665337026119232, "learning_rate": 5.417894913343548e-05, "loss": 0.0145, "step": 149910 }, { "epoch": 95.55130656469089, "grad_norm": 0.31912973523139954, "learning_rate": 5.417345682610914e-05, "loss": 0.0257, "step": 149920 }, { "epoch": 95.55768005098788, "grad_norm": 0.5165213346481323, "learning_rate": 5.416796446807133e-05, "loss": 0.0215, "step": 149930 }, { "epoch": 95.56405353728489, "grad_norm": 0.3781658709049225, "learning_rate": 5.416247205938883e-05, "loss": 0.0147, "step": 149940 }, { "epoch": 95.5704270235819, "grad_norm": 0.4234388470649719, "learning_rate": 5.415697960012835e-05, "loss": 0.0134, "step": 149950 }, { "epoch": 95.57680050987891, "grad_norm": 0.227054625749588, "learning_rate": 5.4151487090356666e-05, "loss": 0.0095, "step": 149960 }, { "epoch": 95.5831739961759, "grad_norm": 0.2017863243818283, "learning_rate": 5.414599453014049e-05, "loss": 0.0191, "step": 149970 }, { "epoch": 95.58954748247291, "grad_norm": 0.18486297130584717, "learning_rate": 5.414050191954655e-05, "loss": 0.0168, "step": 149980 }, { "epoch": 95.59592096876992, "grad_norm": 0.3358275890350342, "learning_rate": 5.413500925864161e-05, "loss": 0.012, "step": 149990 }, { "epoch": 95.60229445506693, "grad_norm": 0.24106521904468536, "learning_rate": 5.41295165474924e-05, "loss": 0.0128, "step": 150000 }, { "epoch": 95.60866794136393, "grad_norm": 0.3707525432109833, "learning_rate": 5.4124023786165676e-05, "loss": 0.0129, "step": 150010 }, { "epoch": 95.61504142766093, "grad_norm": 0.27248266339302063, "learning_rate": 5.4118530974728154e-05, "loss": 0.0197, "step": 150020 }, { "epoch": 95.62141491395793, "grad_norm": 0.3447757959365845, "learning_rate": 5.41130381132466e-05, "loss": 0.0087, "step": 150030 }, { "epoch": 95.62778840025494, "grad_norm": 0.4385470449924469, "learning_rate": 5.410754520178777e-05, "loss": 0.0177, "step": 150040 }, { "epoch": 95.63416188655195, "grad_norm": 0.5444403290748596, "learning_rate": 5.4102052240418356e-05, "loss": 0.0133, "step": 150050 }, { "epoch": 95.64053537284894, "grad_norm": 0.21968892216682434, "learning_rate": 5.4096559229205145e-05, "loss": 0.009, "step": 150060 }, { "epoch": 95.64690885914595, "grad_norm": 0.7708837389945984, "learning_rate": 5.409106616821486e-05, "loss": 0.014, "step": 150070 }, { "epoch": 95.65328234544296, "grad_norm": 0.2853727340698242, "learning_rate": 5.4085573057514285e-05, "loss": 0.0125, "step": 150080 }, { "epoch": 95.65965583173997, "grad_norm": 0.2546008229255676, "learning_rate": 5.408007989717012e-05, "loss": 0.0179, "step": 150090 }, { "epoch": 95.66602931803696, "grad_norm": 0.15702779591083527, "learning_rate": 5.407458668724914e-05, "loss": 0.0081, "step": 150100 }, { "epoch": 95.67240280433397, "grad_norm": 0.4234316051006317, "learning_rate": 5.406909342781809e-05, "loss": 0.0092, "step": 150110 }, { "epoch": 95.67877629063098, "grad_norm": 0.14478625357151031, "learning_rate": 5.40636001189437e-05, "loss": 0.0138, "step": 150120 }, { "epoch": 95.68514977692799, "grad_norm": 0.241457998752594, "learning_rate": 5.405810676069274e-05, "loss": 0.0107, "step": 150130 }, { "epoch": 95.69152326322498, "grad_norm": 0.23224221169948578, "learning_rate": 5.405261335313194e-05, "loss": 0.0162, "step": 150140 }, { "epoch": 95.69789674952199, "grad_norm": 0.40523940324783325, "learning_rate": 5.404711989632808e-05, "loss": 0.0147, "step": 150150 }, { "epoch": 95.704270235819, "grad_norm": 0.2217174470424652, "learning_rate": 5.4041626390347875e-05, "loss": 0.0155, "step": 150160 }, { "epoch": 95.710643722116, "grad_norm": 0.29583540558815, "learning_rate": 5.4036132835258115e-05, "loss": 0.0179, "step": 150170 }, { "epoch": 95.717017208413, "grad_norm": 0.30985668301582336, "learning_rate": 5.40306392311255e-05, "loss": 0.0113, "step": 150180 }, { "epoch": 95.72339069471, "grad_norm": 0.2038620561361313, "learning_rate": 5.402514557801684e-05, "loss": 0.0109, "step": 150190 }, { "epoch": 95.72976418100701, "grad_norm": 0.6133955717086792, "learning_rate": 5.401965187599886e-05, "loss": 0.014, "step": 150200 }, { "epoch": 95.73613766730402, "grad_norm": 0.33463653922080994, "learning_rate": 5.401415812513829e-05, "loss": 0.018, "step": 150210 }, { "epoch": 95.74251115360101, "grad_norm": 0.35593482851982117, "learning_rate": 5.400866432550193e-05, "loss": 0.0143, "step": 150220 }, { "epoch": 95.74888463989802, "grad_norm": 0.16508427262306213, "learning_rate": 5.40031704771565e-05, "loss": 0.0149, "step": 150230 }, { "epoch": 95.75525812619503, "grad_norm": 0.21315805613994598, "learning_rate": 5.3997676580168774e-05, "loss": 0.0093, "step": 150240 }, { "epoch": 95.76163161249204, "grad_norm": 0.20292431116104126, "learning_rate": 5.3992182634605506e-05, "loss": 0.0194, "step": 150250 }, { "epoch": 95.76800509878903, "grad_norm": 0.4402928054332733, "learning_rate": 5.398668864053345e-05, "loss": 0.0163, "step": 150260 }, { "epoch": 95.77437858508604, "grad_norm": 0.4598844051361084, "learning_rate": 5.3981194598019356e-05, "loss": 0.0116, "step": 150270 }, { "epoch": 95.78075207138305, "grad_norm": 0.21390120685100555, "learning_rate": 5.397570050713e-05, "loss": 0.0098, "step": 150280 }, { "epoch": 95.78712555768006, "grad_norm": 0.23957723379135132, "learning_rate": 5.397020636793213e-05, "loss": 0.0106, "step": 150290 }, { "epoch": 95.79349904397705, "grad_norm": 0.3590390384197235, "learning_rate": 5.39647121804925e-05, "loss": 0.022, "step": 150300 }, { "epoch": 95.79987253027406, "grad_norm": 0.4889942407608032, "learning_rate": 5.395921794487787e-05, "loss": 0.0115, "step": 150310 }, { "epoch": 95.80624601657107, "grad_norm": 0.3299259543418884, "learning_rate": 5.3953723661155e-05, "loss": 0.0109, "step": 150320 }, { "epoch": 95.81261950286807, "grad_norm": 0.35490506887435913, "learning_rate": 5.394822932939067e-05, "loss": 0.0102, "step": 150330 }, { "epoch": 95.81899298916507, "grad_norm": 0.4003136157989502, "learning_rate": 5.3942734949651606e-05, "loss": 0.0068, "step": 150340 }, { "epoch": 95.82536647546208, "grad_norm": 0.3788246512413025, "learning_rate": 5.3937240522004605e-05, "loss": 0.0148, "step": 150350 }, { "epoch": 95.83173996175908, "grad_norm": 0.3574615716934204, "learning_rate": 5.393174604651641e-05, "loss": 0.0096, "step": 150360 }, { "epoch": 95.83811344805609, "grad_norm": 0.2034969925880432, "learning_rate": 5.392625152325378e-05, "loss": 0.0145, "step": 150370 }, { "epoch": 95.84448693435309, "grad_norm": 0.3238828480243683, "learning_rate": 5.39207569522835e-05, "loss": 0.0134, "step": 150380 }, { "epoch": 95.8508604206501, "grad_norm": 0.3959044814109802, "learning_rate": 5.391526233367231e-05, "loss": 0.0191, "step": 150390 }, { "epoch": 95.8572339069471, "grad_norm": 0.18591246008872986, "learning_rate": 5.3909767667487e-05, "loss": 0.0091, "step": 150400 }, { "epoch": 95.86360739324411, "grad_norm": 0.39012715220451355, "learning_rate": 5.3904272953794296e-05, "loss": 0.0133, "step": 150410 }, { "epoch": 95.8699808795411, "grad_norm": 0.6856299638748169, "learning_rate": 5.389877819266102e-05, "loss": 0.0116, "step": 150420 }, { "epoch": 95.87635436583811, "grad_norm": 0.3289375901222229, "learning_rate": 5.389328338415388e-05, "loss": 0.0273, "step": 150430 }, { "epoch": 95.88272785213512, "grad_norm": 0.5589417815208435, "learning_rate": 5.38877885283397e-05, "loss": 0.022, "step": 150440 }, { "epoch": 95.88910133843213, "grad_norm": 0.1940336376428604, "learning_rate": 5.3882293625285195e-05, "loss": 0.0063, "step": 150450 }, { "epoch": 95.89547482472912, "grad_norm": 0.28886333107948303, "learning_rate": 5.387679867505717e-05, "loss": 0.0177, "step": 150460 }, { "epoch": 95.90184831102613, "grad_norm": 0.17628121376037598, "learning_rate": 5.387130367772237e-05, "loss": 0.0153, "step": 150470 }, { "epoch": 95.90822179732314, "grad_norm": 0.18966108560562134, "learning_rate": 5.386580863334759e-05, "loss": 0.0074, "step": 150480 }, { "epoch": 95.91459528362014, "grad_norm": 0.22141148149967194, "learning_rate": 5.386031354199956e-05, "loss": 0.011, "step": 150490 }, { "epoch": 95.92096876991714, "grad_norm": 0.36335840821266174, "learning_rate": 5.385481840374511e-05, "loss": 0.0209, "step": 150500 }, { "epoch": 95.92734225621415, "grad_norm": 0.22734837234020233, "learning_rate": 5.3849323218650946e-05, "loss": 0.007, "step": 150510 }, { "epoch": 95.93371574251115, "grad_norm": 0.2172764241695404, "learning_rate": 5.384382798678389e-05, "loss": 0.0121, "step": 150520 }, { "epoch": 95.94008922880816, "grad_norm": 0.428811252117157, "learning_rate": 5.383833270821068e-05, "loss": 0.0114, "step": 150530 }, { "epoch": 95.94646271510516, "grad_norm": 0.2476511001586914, "learning_rate": 5.383283738299812e-05, "loss": 0.0121, "step": 150540 }, { "epoch": 95.95283620140216, "grad_norm": 0.253643274307251, "learning_rate": 5.3827342011212955e-05, "loss": 0.0132, "step": 150550 }, { "epoch": 95.95920968769917, "grad_norm": 0.16548459231853485, "learning_rate": 5.382184659292198e-05, "loss": 0.008, "step": 150560 }, { "epoch": 95.96558317399618, "grad_norm": 0.3131624758243561, "learning_rate": 5.381635112819195e-05, "loss": 0.0104, "step": 150570 }, { "epoch": 95.97195666029317, "grad_norm": 0.3131071925163269, "learning_rate": 5.381085561708966e-05, "loss": 0.0157, "step": 150580 }, { "epoch": 95.97833014659018, "grad_norm": 0.21524351835250854, "learning_rate": 5.380536005968186e-05, "loss": 0.0153, "step": 150590 }, { "epoch": 95.98470363288719, "grad_norm": 0.16066624224185944, "learning_rate": 5.3799864456035364e-05, "loss": 0.0188, "step": 150600 }, { "epoch": 95.9910771191842, "grad_norm": 0.2235276997089386, "learning_rate": 5.379436880621693e-05, "loss": 0.0142, "step": 150610 }, { "epoch": 95.99745060548119, "grad_norm": 0.1795903891324997, "learning_rate": 5.3788873110293314e-05, "loss": 0.0223, "step": 150620 }, { "epoch": 96.0038240917782, "grad_norm": 0.5296271443367004, "learning_rate": 5.378337736833132e-05, "loss": 0.0089, "step": 150630 }, { "epoch": 96.0101975780752, "grad_norm": 0.27400633692741394, "learning_rate": 5.3777881580397716e-05, "loss": 0.0145, "step": 150640 }, { "epoch": 96.01657106437222, "grad_norm": 0.19077059626579285, "learning_rate": 5.3772385746559294e-05, "loss": 0.0116, "step": 150650 }, { "epoch": 96.02294455066922, "grad_norm": 0.5107924342155457, "learning_rate": 5.376688986688281e-05, "loss": 0.0122, "step": 150660 }, { "epoch": 96.02931803696622, "grad_norm": 0.259936660528183, "learning_rate": 5.3761393941435065e-05, "loss": 0.015, "step": 150670 }, { "epoch": 96.03569152326322, "grad_norm": 0.5798413753509521, "learning_rate": 5.375589797028282e-05, "loss": 0.0188, "step": 150680 }, { "epoch": 96.04206500956023, "grad_norm": 0.4639965295791626, "learning_rate": 5.375040195349289e-05, "loss": 0.0142, "step": 150690 }, { "epoch": 96.04843849585724, "grad_norm": 0.5610266327857971, "learning_rate": 5.3744905891132025e-05, "loss": 0.0166, "step": 150700 }, { "epoch": 96.05481198215423, "grad_norm": 0.35211461782455444, "learning_rate": 5.373940978326701e-05, "loss": 0.017, "step": 150710 }, { "epoch": 96.06118546845124, "grad_norm": 0.320793479681015, "learning_rate": 5.373391362996465e-05, "loss": 0.0234, "step": 150720 }, { "epoch": 96.06755895474825, "grad_norm": 0.3520224988460541, "learning_rate": 5.37284174312917e-05, "loss": 0.025, "step": 150730 }, { "epoch": 96.07393244104526, "grad_norm": 0.4152885973453522, "learning_rate": 5.3722921187314964e-05, "loss": 0.0146, "step": 150740 }, { "epoch": 96.08030592734225, "grad_norm": 0.24834975600242615, "learning_rate": 5.371742489810122e-05, "loss": 0.0121, "step": 150750 }, { "epoch": 96.08667941363926, "grad_norm": 0.17653058469295502, "learning_rate": 5.3711928563717265e-05, "loss": 0.0108, "step": 150760 }, { "epoch": 96.09305289993627, "grad_norm": 0.5355924963951111, "learning_rate": 5.370643218422986e-05, "loss": 0.0178, "step": 150770 }, { "epoch": 96.09942638623328, "grad_norm": 0.24079301953315735, "learning_rate": 5.3700935759705804e-05, "loss": 0.0119, "step": 150780 }, { "epoch": 96.10579987253027, "grad_norm": 0.6802887320518494, "learning_rate": 5.369543929021189e-05, "loss": 0.0109, "step": 150790 }, { "epoch": 96.11217335882728, "grad_norm": 0.2637234926223755, "learning_rate": 5.3689942775814895e-05, "loss": 0.0103, "step": 150800 }, { "epoch": 96.11854684512429, "grad_norm": 0.3673524558544159, "learning_rate": 5.368444621658162e-05, "loss": 0.0177, "step": 150810 }, { "epoch": 96.1249203314213, "grad_norm": 0.4692763388156891, "learning_rate": 5.367894961257882e-05, "loss": 0.0105, "step": 150820 }, { "epoch": 96.13129381771829, "grad_norm": 0.2925211787223816, "learning_rate": 5.367345296387334e-05, "loss": 0.0184, "step": 150830 }, { "epoch": 96.1376673040153, "grad_norm": 0.2561262249946594, "learning_rate": 5.3667956270531916e-05, "loss": 0.018, "step": 150840 }, { "epoch": 96.1440407903123, "grad_norm": 0.24524979293346405, "learning_rate": 5.3662459532621376e-05, "loss": 0.0081, "step": 150850 }, { "epoch": 96.15041427660931, "grad_norm": 0.543901801109314, "learning_rate": 5.36569627502085e-05, "loss": 0.0121, "step": 150860 }, { "epoch": 96.1567877629063, "grad_norm": 0.23777735233306885, "learning_rate": 5.3651465923360045e-05, "loss": 0.0196, "step": 150870 }, { "epoch": 96.16316124920331, "grad_norm": 0.4262204170227051, "learning_rate": 5.364596905214285e-05, "loss": 0.0184, "step": 150880 }, { "epoch": 96.16953473550032, "grad_norm": 0.6023387908935547, "learning_rate": 5.364047213662369e-05, "loss": 0.0213, "step": 150890 }, { "epoch": 96.17590822179733, "grad_norm": 0.5951100587844849, "learning_rate": 5.363497517686935e-05, "loss": 0.0119, "step": 150900 }, { "epoch": 96.18228170809432, "grad_norm": 0.35594406723976135, "learning_rate": 5.362947817294664e-05, "loss": 0.0112, "step": 150910 }, { "epoch": 96.18865519439133, "grad_norm": 0.2705243229866028, "learning_rate": 5.362398112492233e-05, "loss": 0.0103, "step": 150920 }, { "epoch": 96.19502868068834, "grad_norm": 0.2771702706813812, "learning_rate": 5.361848403286325e-05, "loss": 0.0126, "step": 150930 }, { "epoch": 96.20140216698535, "grad_norm": 0.4071270823478699, "learning_rate": 5.3612986896836145e-05, "loss": 0.0147, "step": 150940 }, { "epoch": 96.20777565328234, "grad_norm": 0.3121296763420105, "learning_rate": 5.360748971690786e-05, "loss": 0.0108, "step": 150950 }, { "epoch": 96.21414913957935, "grad_norm": 0.16502408683300018, "learning_rate": 5.360199249314515e-05, "loss": 0.0089, "step": 150960 }, { "epoch": 96.22052262587636, "grad_norm": 0.17367666959762573, "learning_rate": 5.359649522561484e-05, "loss": 0.0153, "step": 150970 }, { "epoch": 96.22689611217336, "grad_norm": 0.5321402549743652, "learning_rate": 5.359099791438371e-05, "loss": 0.0139, "step": 150980 }, { "epoch": 96.23326959847036, "grad_norm": 0.38739535212516785, "learning_rate": 5.3585500559518574e-05, "loss": 0.0085, "step": 150990 }, { "epoch": 96.23964308476737, "grad_norm": 0.19366535544395447, "learning_rate": 5.358000316108621e-05, "loss": 0.0123, "step": 151000 }, { "epoch": 96.24601657106437, "grad_norm": 0.29442232847213745, "learning_rate": 5.357450571915344e-05, "loss": 0.0069, "step": 151010 }, { "epoch": 96.25239005736138, "grad_norm": 0.5773637294769287, "learning_rate": 5.356900823378705e-05, "loss": 0.0171, "step": 151020 }, { "epoch": 96.25876354365838, "grad_norm": 0.31757235527038574, "learning_rate": 5.356351070505383e-05, "loss": 0.0128, "step": 151030 }, { "epoch": 96.26513702995538, "grad_norm": 0.38070523738861084, "learning_rate": 5.3558013133020616e-05, "loss": 0.012, "step": 151040 }, { "epoch": 96.27151051625239, "grad_norm": 0.31348511576652527, "learning_rate": 5.3552515517754155e-05, "loss": 0.0108, "step": 151050 }, { "epoch": 96.2778840025494, "grad_norm": 0.5206337571144104, "learning_rate": 5.354701785932129e-05, "loss": 0.01, "step": 151060 }, { "epoch": 96.2842574888464, "grad_norm": 0.2771928608417511, "learning_rate": 5.354152015778879e-05, "loss": 0.0141, "step": 151070 }, { "epoch": 96.2906309751434, "grad_norm": 0.2808614671230316, "learning_rate": 5.3536022413223496e-05, "loss": 0.0116, "step": 151080 }, { "epoch": 96.29700446144041, "grad_norm": 0.49046412110328674, "learning_rate": 5.353052462569219e-05, "loss": 0.0184, "step": 151090 }, { "epoch": 96.30337794773742, "grad_norm": 0.24321837723255157, "learning_rate": 5.352502679526167e-05, "loss": 0.0171, "step": 151100 }, { "epoch": 96.30975143403441, "grad_norm": 0.36017653346061707, "learning_rate": 5.351952892199875e-05, "loss": 0.0076, "step": 151110 }, { "epoch": 96.31612492033142, "grad_norm": 0.6447550058364868, "learning_rate": 5.3514031005970235e-05, "loss": 0.0121, "step": 151120 }, { "epoch": 96.32249840662843, "grad_norm": 0.19536396861076355, "learning_rate": 5.350853304724291e-05, "loss": 0.007, "step": 151130 }, { "epoch": 96.32887189292543, "grad_norm": 0.43226420879364014, "learning_rate": 5.350303504588361e-05, "loss": 0.0083, "step": 151140 }, { "epoch": 96.33524537922243, "grad_norm": 0.24115608632564545, "learning_rate": 5.349753700195912e-05, "loss": 0.0089, "step": 151150 }, { "epoch": 96.34161886551944, "grad_norm": 0.2465478628873825, "learning_rate": 5.3492038915536255e-05, "loss": 0.0119, "step": 151160 }, { "epoch": 96.34799235181644, "grad_norm": 0.38441002368927, "learning_rate": 5.348654078668183e-05, "loss": 0.0096, "step": 151170 }, { "epoch": 96.35436583811345, "grad_norm": 0.4109472632408142, "learning_rate": 5.3481042615462617e-05, "loss": 0.0273, "step": 151180 }, { "epoch": 96.36073932441045, "grad_norm": 0.25199493765830994, "learning_rate": 5.3475544401945465e-05, "loss": 0.0123, "step": 151190 }, { "epoch": 96.36711281070745, "grad_norm": 0.37932756543159485, "learning_rate": 5.3470046146197176e-05, "loss": 0.013, "step": 151200 }, { "epoch": 96.37348629700446, "grad_norm": 0.666317880153656, "learning_rate": 5.346454784828453e-05, "loss": 0.0088, "step": 151210 }, { "epoch": 96.37985978330147, "grad_norm": 0.2033117115497589, "learning_rate": 5.345904950827437e-05, "loss": 0.0093, "step": 151220 }, { "epoch": 96.38623326959846, "grad_norm": 0.6982336044311523, "learning_rate": 5.3453551126233495e-05, "loss": 0.0144, "step": 151230 }, { "epoch": 96.39260675589547, "grad_norm": 0.37083926796913147, "learning_rate": 5.344805270222871e-05, "loss": 0.0096, "step": 151240 }, { "epoch": 96.39898024219248, "grad_norm": 0.5740245580673218, "learning_rate": 5.344255423632684e-05, "loss": 0.0109, "step": 151250 }, { "epoch": 96.40535372848949, "grad_norm": 0.18541572988033295, "learning_rate": 5.343705572859466e-05, "loss": 0.0077, "step": 151260 }, { "epoch": 96.4117272147865, "grad_norm": 0.5252173542976379, "learning_rate": 5.343155717909903e-05, "loss": 0.0158, "step": 151270 }, { "epoch": 96.41810070108349, "grad_norm": 0.2844577729701996, "learning_rate": 5.342605858790672e-05, "loss": 0.0135, "step": 151280 }, { "epoch": 96.4244741873805, "grad_norm": 0.21786843240261078, "learning_rate": 5.342055995508458e-05, "loss": 0.006, "step": 151290 }, { "epoch": 96.4308476736775, "grad_norm": 0.19830934703350067, "learning_rate": 5.341506128069941e-05, "loss": 0.0147, "step": 151300 }, { "epoch": 96.43722115997451, "grad_norm": 0.45990777015686035, "learning_rate": 5.3409562564818016e-05, "loss": 0.0146, "step": 151310 }, { "epoch": 96.4435946462715, "grad_norm": 0.486273854970932, "learning_rate": 5.340406380750721e-05, "loss": 0.0159, "step": 151320 }, { "epoch": 96.44996813256851, "grad_norm": 0.3230139911174774, "learning_rate": 5.339856500883382e-05, "loss": 0.0107, "step": 151330 }, { "epoch": 96.45634161886552, "grad_norm": 0.23300528526306152, "learning_rate": 5.339306616886466e-05, "loss": 0.0097, "step": 151340 }, { "epoch": 96.46271510516253, "grad_norm": 0.7423639297485352, "learning_rate": 5.338756728766654e-05, "loss": 0.0163, "step": 151350 }, { "epoch": 96.46908859145952, "grad_norm": 0.5449203848838806, "learning_rate": 5.338206836530628e-05, "loss": 0.0118, "step": 151360 }, { "epoch": 96.47546207775653, "grad_norm": 0.1562412977218628, "learning_rate": 5.337656940185069e-05, "loss": 0.0182, "step": 151370 }, { "epoch": 96.48183556405354, "grad_norm": 0.39346566796302795, "learning_rate": 5.3371070397366594e-05, "loss": 0.0139, "step": 151380 }, { "epoch": 96.48820905035055, "grad_norm": 0.40649229288101196, "learning_rate": 5.336557135192082e-05, "loss": 0.0123, "step": 151390 }, { "epoch": 96.49458253664754, "grad_norm": 0.28547659516334534, "learning_rate": 5.336007226558016e-05, "loss": 0.0153, "step": 151400 }, { "epoch": 96.50095602294455, "grad_norm": 0.5884092450141907, "learning_rate": 5.335457313841148e-05, "loss": 0.0103, "step": 151410 }, { "epoch": 96.50732950924156, "grad_norm": 0.17969046533107758, "learning_rate": 5.3349073970481535e-05, "loss": 0.0097, "step": 151420 }, { "epoch": 96.51370299553857, "grad_norm": 0.3536794185638428, "learning_rate": 5.3343574761857195e-05, "loss": 0.0101, "step": 151430 }, { "epoch": 96.52007648183556, "grad_norm": 0.4218703508377075, "learning_rate": 5.333807551260526e-05, "loss": 0.0149, "step": 151440 }, { "epoch": 96.52644996813257, "grad_norm": 0.2818814218044281, "learning_rate": 5.333257622279256e-05, "loss": 0.0114, "step": 151450 }, { "epoch": 96.53282345442958, "grad_norm": 0.1721377819776535, "learning_rate": 5.332707689248591e-05, "loss": 0.0104, "step": 151460 }, { "epoch": 96.53919694072658, "grad_norm": 0.21814203262329102, "learning_rate": 5.332157752175214e-05, "loss": 0.0133, "step": 151470 }, { "epoch": 96.54557042702358, "grad_norm": 0.3677891492843628, "learning_rate": 5.331607811065806e-05, "loss": 0.0161, "step": 151480 }, { "epoch": 96.55194391332059, "grad_norm": 0.2986154854297638, "learning_rate": 5.331057865927051e-05, "loss": 0.012, "step": 151490 }, { "epoch": 96.5583173996176, "grad_norm": 0.6363844275474548, "learning_rate": 5.33050791676563e-05, "loss": 0.0208, "step": 151500 }, { "epoch": 96.5646908859146, "grad_norm": 0.2675909101963043, "learning_rate": 5.3299579635882246e-05, "loss": 0.0108, "step": 151510 }, { "epoch": 96.5710643722116, "grad_norm": 0.37660741806030273, "learning_rate": 5.329408006401519e-05, "loss": 0.0086, "step": 151520 }, { "epoch": 96.5774378585086, "grad_norm": 0.30987033247947693, "learning_rate": 5.328858045212195e-05, "loss": 0.0098, "step": 151530 }, { "epoch": 96.58381134480561, "grad_norm": 0.30794671177864075, "learning_rate": 5.328308080026936e-05, "loss": 0.0166, "step": 151540 }, { "epoch": 96.59018483110262, "grad_norm": 0.26615604758262634, "learning_rate": 5.3277581108524234e-05, "loss": 0.0071, "step": 151550 }, { "epoch": 96.59655831739961, "grad_norm": 0.26090139150619507, "learning_rate": 5.327208137695341e-05, "loss": 0.0072, "step": 151560 }, { "epoch": 96.60293180369662, "grad_norm": 0.2395295351743698, "learning_rate": 5.326658160562369e-05, "loss": 0.0144, "step": 151570 }, { "epoch": 96.60930528999363, "grad_norm": 0.15935613214969635, "learning_rate": 5.326108179460194e-05, "loss": 0.0082, "step": 151580 }, { "epoch": 96.61567877629064, "grad_norm": 0.3425524830818176, "learning_rate": 5.325558194395496e-05, "loss": 0.0201, "step": 151590 }, { "epoch": 96.62205226258763, "grad_norm": 0.3417484164237976, "learning_rate": 5.325008205374957e-05, "loss": 0.0104, "step": 151600 }, { "epoch": 96.62842574888464, "grad_norm": 0.5200260877609253, "learning_rate": 5.324458212405264e-05, "loss": 0.0147, "step": 151610 }, { "epoch": 96.63479923518165, "grad_norm": 0.18435199558734894, "learning_rate": 5.323908215493095e-05, "loss": 0.0164, "step": 151620 }, { "epoch": 96.64117272147865, "grad_norm": 0.2736758589744568, "learning_rate": 5.3233582146451375e-05, "loss": 0.0154, "step": 151630 }, { "epoch": 96.64754620777565, "grad_norm": 0.5757254958152771, "learning_rate": 5.322808209868071e-05, "loss": 0.0213, "step": 151640 }, { "epoch": 96.65391969407266, "grad_norm": 0.2276429384946823, "learning_rate": 5.322258201168581e-05, "loss": 0.0079, "step": 151650 }, { "epoch": 96.66029318036966, "grad_norm": 0.2700411081314087, "learning_rate": 5.321708188553349e-05, "loss": 0.008, "step": 151660 }, { "epoch": 96.66666666666667, "grad_norm": 0.2514929175376892, "learning_rate": 5.321158172029058e-05, "loss": 0.0107, "step": 151670 }, { "epoch": 96.67304015296367, "grad_norm": 0.2724277973175049, "learning_rate": 5.320608151602393e-05, "loss": 0.0161, "step": 151680 }, { "epoch": 96.67941363926067, "grad_norm": 0.16540037095546722, "learning_rate": 5.3200581272800366e-05, "loss": 0.0182, "step": 151690 }, { "epoch": 96.68578712555768, "grad_norm": 0.21694545447826385, "learning_rate": 5.3195080990686706e-05, "loss": 0.0118, "step": 151700 }, { "epoch": 96.69216061185469, "grad_norm": 0.3881620466709137, "learning_rate": 5.3189580669749793e-05, "loss": 0.0089, "step": 151710 }, { "epoch": 96.69853409815168, "grad_norm": 0.5129000544548035, "learning_rate": 5.318408031005648e-05, "loss": 0.0243, "step": 151720 }, { "epoch": 96.70490758444869, "grad_norm": 0.33086904883384705, "learning_rate": 5.317857991167358e-05, "loss": 0.017, "step": 151730 }, { "epoch": 96.7112810707457, "grad_norm": 0.29575425386428833, "learning_rate": 5.317307947466793e-05, "loss": 0.0147, "step": 151740 }, { "epoch": 96.7176545570427, "grad_norm": 0.46261945366859436, "learning_rate": 5.316757899910637e-05, "loss": 0.0131, "step": 151750 }, { "epoch": 96.7240280433397, "grad_norm": 0.4670831859111786, "learning_rate": 5.3162078485055725e-05, "loss": 0.0102, "step": 151760 }, { "epoch": 96.73040152963671, "grad_norm": 0.2701079845428467, "learning_rate": 5.315657793258285e-05, "loss": 0.0128, "step": 151770 }, { "epoch": 96.73677501593372, "grad_norm": 0.3212505280971527, "learning_rate": 5.3151077341754576e-05, "loss": 0.0067, "step": 151780 }, { "epoch": 96.74314850223072, "grad_norm": 0.30156633257865906, "learning_rate": 5.3145576712637735e-05, "loss": 0.0185, "step": 151790 }, { "epoch": 96.74952198852772, "grad_norm": 0.5736785531044006, "learning_rate": 5.314007604529915e-05, "loss": 0.0129, "step": 151800 }, { "epoch": 96.75589547482473, "grad_norm": 0.2238437980413437, "learning_rate": 5.31345753398057e-05, "loss": 0.0122, "step": 151810 }, { "epoch": 96.76226896112173, "grad_norm": 0.13095133006572723, "learning_rate": 5.312907459622418e-05, "loss": 0.0099, "step": 151820 }, { "epoch": 96.76864244741874, "grad_norm": 0.4320639669895172, "learning_rate": 5.312357381462146e-05, "loss": 0.0104, "step": 151830 }, { "epoch": 96.77501593371574, "grad_norm": 0.3958726227283478, "learning_rate": 5.3118072995064375e-05, "loss": 0.0185, "step": 151840 }, { "epoch": 96.78138942001274, "grad_norm": 0.4804927408695221, "learning_rate": 5.3112572137619754e-05, "loss": 0.0241, "step": 151850 }, { "epoch": 96.78776290630975, "grad_norm": 0.3269464671611786, "learning_rate": 5.3107071242354435e-05, "loss": 0.0136, "step": 151860 }, { "epoch": 96.79413639260676, "grad_norm": 0.4299466609954834, "learning_rate": 5.310157030933528e-05, "loss": 0.0083, "step": 151870 }, { "epoch": 96.80050987890377, "grad_norm": 0.2618471086025238, "learning_rate": 5.3096069338629105e-05, "loss": 0.0149, "step": 151880 }, { "epoch": 96.80688336520076, "grad_norm": 0.3140926957130432, "learning_rate": 5.309056833030277e-05, "loss": 0.0119, "step": 151890 }, { "epoch": 96.81325685149777, "grad_norm": 0.4688023328781128, "learning_rate": 5.308506728442311e-05, "loss": 0.0283, "step": 151900 }, { "epoch": 96.81963033779478, "grad_norm": 0.6076087355613708, "learning_rate": 5.307956620105697e-05, "loss": 0.012, "step": 151910 }, { "epoch": 96.82600382409179, "grad_norm": 0.37517812848091125, "learning_rate": 5.3074065080271196e-05, "loss": 0.007, "step": 151920 }, { "epoch": 96.83237731038878, "grad_norm": 0.2919142544269562, "learning_rate": 5.306856392213263e-05, "loss": 0.0155, "step": 151930 }, { "epoch": 96.83875079668579, "grad_norm": 0.1746671497821808, "learning_rate": 5.3063062726708115e-05, "loss": 0.0087, "step": 151940 }, { "epoch": 96.8451242829828, "grad_norm": 0.6914273500442505, "learning_rate": 5.305756149406449e-05, "loss": 0.0198, "step": 151950 }, { "epoch": 96.8514977692798, "grad_norm": 0.2389294058084488, "learning_rate": 5.3052060224268606e-05, "loss": 0.0212, "step": 151960 }, { "epoch": 96.8578712555768, "grad_norm": 0.2459172159433365, "learning_rate": 5.304655891738732e-05, "loss": 0.0116, "step": 151970 }, { "epoch": 96.8642447418738, "grad_norm": 0.2545512318611145, "learning_rate": 5.3041057573487464e-05, "loss": 0.008, "step": 151980 }, { "epoch": 96.87061822817081, "grad_norm": 0.4606887102127075, "learning_rate": 5.303555619263587e-05, "loss": 0.0112, "step": 151990 }, { "epoch": 96.87699171446782, "grad_norm": 0.3768860697746277, "learning_rate": 5.303005477489943e-05, "loss": 0.0232, "step": 152000 }, { "epoch": 96.88336520076481, "grad_norm": 0.2618246376514435, "learning_rate": 5.302455332034494e-05, "loss": 0.0101, "step": 152010 }, { "epoch": 96.88973868706182, "grad_norm": 0.27187323570251465, "learning_rate": 5.301905182903928e-05, "loss": 0.0086, "step": 152020 }, { "epoch": 96.89611217335883, "grad_norm": 0.4436056315898895, "learning_rate": 5.3013550301049286e-05, "loss": 0.0137, "step": 152030 }, { "epoch": 96.90248565965584, "grad_norm": 0.1923888772726059, "learning_rate": 5.3008048736441814e-05, "loss": 0.0128, "step": 152040 }, { "epoch": 96.90885914595283, "grad_norm": 0.1676735281944275, "learning_rate": 5.3002547135283706e-05, "loss": 0.0112, "step": 152050 }, { "epoch": 96.91523263224984, "grad_norm": 0.21163253486156464, "learning_rate": 5.299704549764183e-05, "loss": 0.0155, "step": 152060 }, { "epoch": 96.92160611854685, "grad_norm": 0.1931218057870865, "learning_rate": 5.299154382358301e-05, "loss": 0.0125, "step": 152070 }, { "epoch": 96.92797960484386, "grad_norm": 0.4403468072414398, "learning_rate": 5.2986042113174104e-05, "loss": 0.0132, "step": 152080 }, { "epoch": 96.93435309114085, "grad_norm": 0.4867379665374756, "learning_rate": 5.298054036648198e-05, "loss": 0.0131, "step": 152090 }, { "epoch": 96.94072657743786, "grad_norm": 0.26056769490242004, "learning_rate": 5.297503858357346e-05, "loss": 0.01, "step": 152100 }, { "epoch": 96.94710006373487, "grad_norm": 0.21436373889446259, "learning_rate": 5.2969536764515425e-05, "loss": 0.0086, "step": 152110 }, { "epoch": 96.95347355003187, "grad_norm": 0.5208535194396973, "learning_rate": 5.2964034909374704e-05, "loss": 0.0131, "step": 152120 }, { "epoch": 96.95984703632887, "grad_norm": 0.434498131275177, "learning_rate": 5.295853301821817e-05, "loss": 0.0087, "step": 152130 }, { "epoch": 96.96622052262587, "grad_norm": 0.608322262763977, "learning_rate": 5.295303109111267e-05, "loss": 0.0174, "step": 152140 }, { "epoch": 96.97259400892288, "grad_norm": 0.28252333402633667, "learning_rate": 5.2947529128125036e-05, "loss": 0.0098, "step": 152150 }, { "epoch": 96.97896749521989, "grad_norm": 0.2917349934577942, "learning_rate": 5.294202712932217e-05, "loss": 0.0119, "step": 152160 }, { "epoch": 96.98534098151688, "grad_norm": 0.1386253386735916, "learning_rate": 5.293652509477087e-05, "loss": 0.0078, "step": 152170 }, { "epoch": 96.99171446781389, "grad_norm": 0.42090919613838196, "learning_rate": 5.293102302453803e-05, "loss": 0.0164, "step": 152180 }, { "epoch": 96.9980879541109, "grad_norm": 0.22740305960178375, "learning_rate": 5.2925520918690486e-05, "loss": 0.0111, "step": 152190 }, { "epoch": 97.00446144040791, "grad_norm": 0.47587835788726807, "learning_rate": 5.29200187772951e-05, "loss": 0.0112, "step": 152200 }, { "epoch": 97.0108349267049, "grad_norm": 0.2440553903579712, "learning_rate": 5.291451660041873e-05, "loss": 0.012, "step": 152210 }, { "epoch": 97.01720841300191, "grad_norm": 0.434550940990448, "learning_rate": 5.290901438812824e-05, "loss": 0.0172, "step": 152220 }, { "epoch": 97.02358189929892, "grad_norm": 0.24537333846092224, "learning_rate": 5.2903512140490476e-05, "loss": 0.0154, "step": 152230 }, { "epoch": 97.02995538559593, "grad_norm": 0.24721452593803406, "learning_rate": 5.2898009857572285e-05, "loss": 0.0124, "step": 152240 }, { "epoch": 97.03632887189292, "grad_norm": 0.23098906874656677, "learning_rate": 5.289250753944055e-05, "loss": 0.0226, "step": 152250 }, { "epoch": 97.04270235818993, "grad_norm": 0.3592780828475952, "learning_rate": 5.288700518616212e-05, "loss": 0.0133, "step": 152260 }, { "epoch": 97.04907584448694, "grad_norm": 0.41876697540283203, "learning_rate": 5.288150279780385e-05, "loss": 0.0158, "step": 152270 }, { "epoch": 97.05544933078394, "grad_norm": 0.3980472683906555, "learning_rate": 5.287600037443259e-05, "loss": 0.0092, "step": 152280 }, { "epoch": 97.06182281708094, "grad_norm": 0.4248199760913849, "learning_rate": 5.287049791611521e-05, "loss": 0.0122, "step": 152290 }, { "epoch": 97.06819630337795, "grad_norm": 0.39865219593048096, "learning_rate": 5.2864995422918586e-05, "loss": 0.0225, "step": 152300 }, { "epoch": 97.07456978967495, "grad_norm": 0.19380980730056763, "learning_rate": 5.285949289490955e-05, "loss": 0.0117, "step": 152310 }, { "epoch": 97.08094327597196, "grad_norm": 0.6231576800346375, "learning_rate": 5.285399033215499e-05, "loss": 0.02, "step": 152320 }, { "epoch": 97.08731676226896, "grad_norm": 0.7415942549705505, "learning_rate": 5.284848773472173e-05, "loss": 0.0185, "step": 152330 }, { "epoch": 97.09369024856596, "grad_norm": 0.7736468315124512, "learning_rate": 5.284298510267667e-05, "loss": 0.019, "step": 152340 }, { "epoch": 97.10006373486297, "grad_norm": 0.25295743346214294, "learning_rate": 5.283748243608665e-05, "loss": 0.0133, "step": 152350 }, { "epoch": 97.10643722115998, "grad_norm": 0.2671968638896942, "learning_rate": 5.2831979735018547e-05, "loss": 0.0129, "step": 152360 }, { "epoch": 97.11281070745697, "grad_norm": 0.4490671753883362, "learning_rate": 5.282647699953921e-05, "loss": 0.0109, "step": 152370 }, { "epoch": 97.11918419375398, "grad_norm": 0.25614264607429504, "learning_rate": 5.2820974229715515e-05, "loss": 0.0221, "step": 152380 }, { "epoch": 97.12555768005099, "grad_norm": 0.22738981246948242, "learning_rate": 5.2815471425614313e-05, "loss": 0.0103, "step": 152390 }, { "epoch": 97.131931166348, "grad_norm": 0.1733320951461792, "learning_rate": 5.2809968587302474e-05, "loss": 0.0224, "step": 152400 }, { "epoch": 97.13830465264499, "grad_norm": 0.5330601930618286, "learning_rate": 5.280446571484687e-05, "loss": 0.0208, "step": 152410 }, { "epoch": 97.144678138942, "grad_norm": 0.374647855758667, "learning_rate": 5.2798962808314356e-05, "loss": 0.0127, "step": 152420 }, { "epoch": 97.151051625239, "grad_norm": 0.44461148977279663, "learning_rate": 5.27934598677718e-05, "loss": 0.0107, "step": 152430 }, { "epoch": 97.15742511153601, "grad_norm": 0.26457053422927856, "learning_rate": 5.278795689328607e-05, "loss": 0.0133, "step": 152440 }, { "epoch": 97.16379859783301, "grad_norm": 0.18976230919361115, "learning_rate": 5.2782453884924034e-05, "loss": 0.0151, "step": 152450 }, { "epoch": 97.17017208413002, "grad_norm": 0.6664788722991943, "learning_rate": 5.277695084275256e-05, "loss": 0.0268, "step": 152460 }, { "epoch": 97.17654557042702, "grad_norm": 0.2123701125383377, "learning_rate": 5.27714477668385e-05, "loss": 0.0179, "step": 152470 }, { "epoch": 97.18291905672403, "grad_norm": 0.32728835940361023, "learning_rate": 5.276594465724875e-05, "loss": 0.0109, "step": 152480 }, { "epoch": 97.18929254302103, "grad_norm": 0.39515796303749084, "learning_rate": 5.276044151405014e-05, "loss": 0.0201, "step": 152490 }, { "epoch": 97.19566602931803, "grad_norm": 0.437180757522583, "learning_rate": 5.2754938337309576e-05, "loss": 0.0089, "step": 152500 }, { "epoch": 97.20203951561504, "grad_norm": 0.29451534152030945, "learning_rate": 5.2749435127093895e-05, "loss": 0.0087, "step": 152510 }, { "epoch": 97.20841300191205, "grad_norm": 0.535929799079895, "learning_rate": 5.274393188347e-05, "loss": 0.0232, "step": 152520 }, { "epoch": 97.21478648820906, "grad_norm": 0.3414114713668823, "learning_rate": 5.273842860650472e-05, "loss": 0.0194, "step": 152530 }, { "epoch": 97.22115997450605, "grad_norm": 0.30935218930244446, "learning_rate": 5.2732925296264965e-05, "loss": 0.0144, "step": 152540 }, { "epoch": 97.22753346080306, "grad_norm": 0.4188404977321625, "learning_rate": 5.272742195281758e-05, "loss": 0.0287, "step": 152550 }, { "epoch": 97.23390694710007, "grad_norm": 0.27270421385765076, "learning_rate": 5.272191857622944e-05, "loss": 0.0118, "step": 152560 }, { "epoch": 97.24028043339707, "grad_norm": 0.16882987320423126, "learning_rate": 5.2716415166567426e-05, "loss": 0.0136, "step": 152570 }, { "epoch": 97.24665391969407, "grad_norm": 0.47059738636016846, "learning_rate": 5.2710911723898406e-05, "loss": 0.0118, "step": 152580 }, { "epoch": 97.25302740599108, "grad_norm": 0.33363252878189087, "learning_rate": 5.270540824828924e-05, "loss": 0.007, "step": 152590 }, { "epoch": 97.25940089228808, "grad_norm": 0.3092624843120575, "learning_rate": 5.2699904739806814e-05, "loss": 0.0192, "step": 152600 }, { "epoch": 97.26577437858509, "grad_norm": 0.3244243562221527, "learning_rate": 5.2694401198517985e-05, "loss": 0.013, "step": 152610 }, { "epoch": 97.27214786488209, "grad_norm": 0.25771641731262207, "learning_rate": 5.268889762448965e-05, "loss": 0.014, "step": 152620 }, { "epoch": 97.2785213511791, "grad_norm": 0.44303518533706665, "learning_rate": 5.268339401778867e-05, "loss": 0.0147, "step": 152630 }, { "epoch": 97.2848948374761, "grad_norm": 0.21621668338775635, "learning_rate": 5.267789037848192e-05, "loss": 0.0197, "step": 152640 }, { "epoch": 97.29126832377311, "grad_norm": 0.23631927371025085, "learning_rate": 5.2672386706636256e-05, "loss": 0.0104, "step": 152650 }, { "epoch": 97.2976418100701, "grad_norm": 0.6051642298698425, "learning_rate": 5.26668830023186e-05, "loss": 0.0194, "step": 152660 }, { "epoch": 97.30401529636711, "grad_norm": 0.313846617937088, "learning_rate": 5.2661379265595766e-05, "loss": 0.0102, "step": 152670 }, { "epoch": 97.31038878266412, "grad_norm": 0.36921679973602295, "learning_rate": 5.265587549653469e-05, "loss": 0.0121, "step": 152680 }, { "epoch": 97.31676226896113, "grad_norm": 0.22451713681221008, "learning_rate": 5.26503716952022e-05, "loss": 0.0108, "step": 152690 }, { "epoch": 97.32313575525812, "grad_norm": 0.2924201190471649, "learning_rate": 5.26448678616652e-05, "loss": 0.0124, "step": 152700 }, { "epoch": 97.32950924155513, "grad_norm": 0.2661421000957489, "learning_rate": 5.2639363995990564e-05, "loss": 0.0141, "step": 152710 }, { "epoch": 97.33588272785214, "grad_norm": 0.49959316849708557, "learning_rate": 5.2633860098245144e-05, "loss": 0.013, "step": 152720 }, { "epoch": 97.34225621414915, "grad_norm": 0.24530307948589325, "learning_rate": 5.2628356168495865e-05, "loss": 0.0072, "step": 152730 }, { "epoch": 97.34862970044614, "grad_norm": 0.3386513888835907, "learning_rate": 5.262285220680955e-05, "loss": 0.0092, "step": 152740 }, { "epoch": 97.35500318674315, "grad_norm": 0.3265387713909149, "learning_rate": 5.261734821325313e-05, "loss": 0.0163, "step": 152750 }, { "epoch": 97.36137667304016, "grad_norm": 0.3347891569137573, "learning_rate": 5.261184418789343e-05, "loss": 0.0258, "step": 152760 }, { "epoch": 97.36775015933716, "grad_norm": 0.4003423750400543, "learning_rate": 5.260634013079737e-05, "loss": 0.0259, "step": 152770 }, { "epoch": 97.37412364563416, "grad_norm": 0.2662723660469055, "learning_rate": 5.260083604203182e-05, "loss": 0.0098, "step": 152780 }, { "epoch": 97.38049713193116, "grad_norm": 0.31269729137420654, "learning_rate": 5.259533192166366e-05, "loss": 0.0099, "step": 152790 }, { "epoch": 97.38687061822817, "grad_norm": 0.30887213349342346, "learning_rate": 5.258982776975976e-05, "loss": 0.0202, "step": 152800 }, { "epoch": 97.39324410452518, "grad_norm": 0.3474363684654236, "learning_rate": 5.2584323586387e-05, "loss": 0.0265, "step": 152810 }, { "epoch": 97.39961759082217, "grad_norm": 0.2794405519962311, "learning_rate": 5.2578819371612284e-05, "loss": 0.0092, "step": 152820 }, { "epoch": 97.40599107711918, "grad_norm": 0.5493617057800293, "learning_rate": 5.257331512550247e-05, "loss": 0.0116, "step": 152830 }, { "epoch": 97.41236456341619, "grad_norm": 0.44666212797164917, "learning_rate": 5.256781084812444e-05, "loss": 0.0168, "step": 152840 }, { "epoch": 97.4187380497132, "grad_norm": 0.7042959332466125, "learning_rate": 5.256230653954509e-05, "loss": 0.0309, "step": 152850 }, { "epoch": 97.42511153601019, "grad_norm": 0.5670607686042786, "learning_rate": 5.25568021998313e-05, "loss": 0.0102, "step": 152860 }, { "epoch": 97.4314850223072, "grad_norm": 0.3833569884300232, "learning_rate": 5.255129782904996e-05, "loss": 0.0218, "step": 152870 }, { "epoch": 97.43785850860421, "grad_norm": 0.21072880923748016, "learning_rate": 5.254579342726792e-05, "loss": 0.0211, "step": 152880 }, { "epoch": 97.44423199490122, "grad_norm": 0.7457658648490906, "learning_rate": 5.25402889945521e-05, "loss": 0.0297, "step": 152890 }, { "epoch": 97.45060548119821, "grad_norm": 0.4226810336112976, "learning_rate": 5.253478453096936e-05, "loss": 0.019, "step": 152900 }, { "epoch": 97.45697896749522, "grad_norm": 0.2861146032810211, "learning_rate": 5.25292800365866e-05, "loss": 0.0087, "step": 152910 }, { "epoch": 97.46335245379223, "grad_norm": 0.6532594561576843, "learning_rate": 5.25237755114707e-05, "loss": 0.0147, "step": 152920 }, { "epoch": 97.46972594008923, "grad_norm": 0.1897241324186325, "learning_rate": 5.251827095568855e-05, "loss": 0.0167, "step": 152930 }, { "epoch": 97.47609942638623, "grad_norm": 0.22914575040340424, "learning_rate": 5.251276636930702e-05, "loss": 0.0097, "step": 152940 }, { "epoch": 97.48247291268324, "grad_norm": 0.44922059774398804, "learning_rate": 5.250726175239302e-05, "loss": 0.0151, "step": 152950 }, { "epoch": 97.48884639898024, "grad_norm": 0.34937071800231934, "learning_rate": 5.250175710501342e-05, "loss": 0.0132, "step": 152960 }, { "epoch": 97.49521988527725, "grad_norm": 0.32423436641693115, "learning_rate": 5.2496252427235093e-05, "loss": 0.0132, "step": 152970 }, { "epoch": 97.50159337157424, "grad_norm": 0.29478719830513, "learning_rate": 5.249074771912495e-05, "loss": 0.0099, "step": 152980 }, { "epoch": 97.50796685787125, "grad_norm": 0.20938363671302795, "learning_rate": 5.248524298074987e-05, "loss": 0.0141, "step": 152990 }, { "epoch": 97.51434034416826, "grad_norm": 0.4518100917339325, "learning_rate": 5.2479738212176755e-05, "loss": 0.0215, "step": 153000 }, { "epoch": 97.52071383046527, "grad_norm": 0.31175732612609863, "learning_rate": 5.247423341347246e-05, "loss": 0.007, "step": 153010 }, { "epoch": 97.52708731676226, "grad_norm": 0.25221875309944153, "learning_rate": 5.246872858470391e-05, "loss": 0.0092, "step": 153020 }, { "epoch": 97.53346080305927, "grad_norm": 0.40673667192459106, "learning_rate": 5.246322372593797e-05, "loss": 0.0153, "step": 153030 }, { "epoch": 97.53983428935628, "grad_norm": 0.2953347861766815, "learning_rate": 5.245771883724153e-05, "loss": 0.0216, "step": 153040 }, { "epoch": 97.54620777565329, "grad_norm": 0.3158614933490753, "learning_rate": 5.2452213918681494e-05, "loss": 0.011, "step": 153050 }, { "epoch": 97.55258126195028, "grad_norm": 0.46626871824264526, "learning_rate": 5.2446708970324734e-05, "loss": 0.0153, "step": 153060 }, { "epoch": 97.55895474824729, "grad_norm": 0.6247791647911072, "learning_rate": 5.244120399223816e-05, "loss": 0.0231, "step": 153070 }, { "epoch": 97.5653282345443, "grad_norm": 0.2811007499694824, "learning_rate": 5.2435698984488645e-05, "loss": 0.0124, "step": 153080 }, { "epoch": 97.5717017208413, "grad_norm": 0.1482483148574829, "learning_rate": 5.243019394714309e-05, "loss": 0.0088, "step": 153090 }, { "epoch": 97.5780752071383, "grad_norm": 0.3038245439529419, "learning_rate": 5.242468888026838e-05, "loss": 0.0069, "step": 153100 }, { "epoch": 97.5844486934353, "grad_norm": 0.4128970503807068, "learning_rate": 5.241918378393143e-05, "loss": 0.0162, "step": 153110 }, { "epoch": 97.59082217973231, "grad_norm": 0.3123569190502167, "learning_rate": 5.24136786581991e-05, "loss": 0.0204, "step": 153120 }, { "epoch": 97.59719566602932, "grad_norm": 0.2464195340871811, "learning_rate": 5.240817350313828e-05, "loss": 0.0106, "step": 153130 }, { "epoch": 97.60356915232632, "grad_norm": 0.4462961256504059, "learning_rate": 5.240266831881591e-05, "loss": 0.0141, "step": 153140 }, { "epoch": 97.60994263862332, "grad_norm": 0.40051496028900146, "learning_rate": 5.239716310529882e-05, "loss": 0.019, "step": 153150 }, { "epoch": 97.61631612492033, "grad_norm": 0.3353978395462036, "learning_rate": 5.2391657862653953e-05, "loss": 0.0212, "step": 153160 }, { "epoch": 97.62268961121734, "grad_norm": 0.24019920825958252, "learning_rate": 5.238615259094818e-05, "loss": 0.0079, "step": 153170 }, { "epoch": 97.62906309751435, "grad_norm": 0.26586678624153137, "learning_rate": 5.2380647290248396e-05, "loss": 0.0118, "step": 153180 }, { "epoch": 97.63543658381134, "grad_norm": 0.17551906406879425, "learning_rate": 5.237514196062151e-05, "loss": 0.0187, "step": 153190 }, { "epoch": 97.64181007010835, "grad_norm": 0.45579567551612854, "learning_rate": 5.2369636602134397e-05, "loss": 0.0292, "step": 153200 }, { "epoch": 97.64818355640536, "grad_norm": 0.3279739320278168, "learning_rate": 5.236413121485397e-05, "loss": 0.0127, "step": 153210 }, { "epoch": 97.65455704270236, "grad_norm": 0.23638862371444702, "learning_rate": 5.2358625798847105e-05, "loss": 0.0111, "step": 153220 }, { "epoch": 97.66093052899936, "grad_norm": 0.3217392861843109, "learning_rate": 5.235312035418073e-05, "loss": 0.0149, "step": 153230 }, { "epoch": 97.66730401529637, "grad_norm": 0.19279783964157104, "learning_rate": 5.2347614880921704e-05, "loss": 0.0097, "step": 153240 }, { "epoch": 97.67367750159337, "grad_norm": 0.4712560772895813, "learning_rate": 5.234210937913695e-05, "loss": 0.0146, "step": 153250 }, { "epoch": 97.68005098789038, "grad_norm": 0.45799922943115234, "learning_rate": 5.233660384889334e-05, "loss": 0.0134, "step": 153260 }, { "epoch": 97.68642447418738, "grad_norm": 0.340853214263916, "learning_rate": 5.233109829025781e-05, "loss": 0.024, "step": 153270 }, { "epoch": 97.69279796048438, "grad_norm": 0.3823428153991699, "learning_rate": 5.232559270329721e-05, "loss": 0.0098, "step": 153280 }, { "epoch": 97.69917144678139, "grad_norm": 0.380995512008667, "learning_rate": 5.232008708807848e-05, "loss": 0.0114, "step": 153290 }, { "epoch": 97.7055449330784, "grad_norm": 0.23990286886692047, "learning_rate": 5.231458144466849e-05, "loss": 0.0105, "step": 153300 }, { "epoch": 97.7119184193754, "grad_norm": 0.6793375015258789, "learning_rate": 5.230907577313416e-05, "loss": 0.0114, "step": 153310 }, { "epoch": 97.7182919056724, "grad_norm": 0.2432444542646408, "learning_rate": 5.230357007354237e-05, "loss": 0.0115, "step": 153320 }, { "epoch": 97.72466539196941, "grad_norm": 0.2720632255077362, "learning_rate": 5.229806434596004e-05, "loss": 0.0089, "step": 153330 }, { "epoch": 97.73103887826642, "grad_norm": 0.20387569069862366, "learning_rate": 5.229255859045406e-05, "loss": 0.0076, "step": 153340 }, { "epoch": 97.73741236456341, "grad_norm": 0.31721025705337524, "learning_rate": 5.228705280709132e-05, "loss": 0.0116, "step": 153350 }, { "epoch": 97.74378585086042, "grad_norm": 0.3242476284503937, "learning_rate": 5.228154699593873e-05, "loss": 0.0155, "step": 153360 }, { "epoch": 97.75015933715743, "grad_norm": 0.5140753388404846, "learning_rate": 5.2276041157063196e-05, "loss": 0.011, "step": 153370 }, { "epoch": 97.75653282345444, "grad_norm": 0.3773769736289978, "learning_rate": 5.22705352905316e-05, "loss": 0.0119, "step": 153380 }, { "epoch": 97.76290630975143, "grad_norm": 0.28213873505592346, "learning_rate": 5.226502939641087e-05, "loss": 0.0092, "step": 153390 }, { "epoch": 97.76927979604844, "grad_norm": 0.4631240665912628, "learning_rate": 5.225952347476788e-05, "loss": 0.013, "step": 153400 }, { "epoch": 97.77565328234544, "grad_norm": 0.5073865056037903, "learning_rate": 5.225401752566956e-05, "loss": 0.0181, "step": 153410 }, { "epoch": 97.78202676864245, "grad_norm": 0.3677632808685303, "learning_rate": 5.2248511549182786e-05, "loss": 0.0143, "step": 153420 }, { "epoch": 97.78840025493945, "grad_norm": 0.31847652792930603, "learning_rate": 5.224300554537449e-05, "loss": 0.0165, "step": 153430 }, { "epoch": 97.79477374123645, "grad_norm": 0.36296704411506653, "learning_rate": 5.223749951431155e-05, "loss": 0.0168, "step": 153440 }, { "epoch": 97.80114722753346, "grad_norm": 0.275348037481308, "learning_rate": 5.223199345606088e-05, "loss": 0.0103, "step": 153450 }, { "epoch": 97.80752071383047, "grad_norm": 0.4989657402038574, "learning_rate": 5.222648737068938e-05, "loss": 0.0101, "step": 153460 }, { "epoch": 97.81389420012746, "grad_norm": 0.45165738463401794, "learning_rate": 5.222098125826396e-05, "loss": 0.0134, "step": 153470 }, { "epoch": 97.82026768642447, "grad_norm": 0.33907854557037354, "learning_rate": 5.221547511885152e-05, "loss": 0.0121, "step": 153480 }, { "epoch": 97.82664117272148, "grad_norm": 0.2527565360069275, "learning_rate": 5.220996895251895e-05, "loss": 0.0135, "step": 153490 }, { "epoch": 97.83301465901849, "grad_norm": 0.5439563393592834, "learning_rate": 5.2204462759333196e-05, "loss": 0.0121, "step": 153500 }, { "epoch": 97.83938814531548, "grad_norm": 0.2644340693950653, "learning_rate": 5.219895653936113e-05, "loss": 0.0137, "step": 153510 }, { "epoch": 97.84576163161249, "grad_norm": 0.21048656105995178, "learning_rate": 5.2193450292669654e-05, "loss": 0.0156, "step": 153520 }, { "epoch": 97.8521351179095, "grad_norm": 0.2526402473449707, "learning_rate": 5.218794401932571e-05, "loss": 0.0114, "step": 153530 }, { "epoch": 97.8585086042065, "grad_norm": 0.2728976905345917, "learning_rate": 5.2182437719396146e-05, "loss": 0.0107, "step": 153540 }, { "epoch": 97.8648820905035, "grad_norm": 0.22918735444545746, "learning_rate": 5.217693139294794e-05, "loss": 0.0243, "step": 153550 }, { "epoch": 97.87125557680051, "grad_norm": 0.494890958070755, "learning_rate": 5.217142504004793e-05, "loss": 0.0158, "step": 153560 }, { "epoch": 97.87762906309752, "grad_norm": 0.29522016644477844, "learning_rate": 5.216591866076308e-05, "loss": 0.0224, "step": 153570 }, { "epoch": 97.88400254939452, "grad_norm": 0.3467455804347992, "learning_rate": 5.216041225516025e-05, "loss": 0.017, "step": 153580 }, { "epoch": 97.89037603569152, "grad_norm": 0.45662564039230347, "learning_rate": 5.215490582330639e-05, "loss": 0.0155, "step": 153590 }, { "epoch": 97.89674952198853, "grad_norm": 0.34645354747772217, "learning_rate": 5.214939936526839e-05, "loss": 0.0178, "step": 153600 }, { "epoch": 97.90312300828553, "grad_norm": 0.5467672348022461, "learning_rate": 5.214389288111314e-05, "loss": 0.0246, "step": 153610 }, { "epoch": 97.90949649458254, "grad_norm": 0.283273845911026, "learning_rate": 5.21383863709076e-05, "loss": 0.0172, "step": 153620 }, { "epoch": 97.91586998087953, "grad_norm": 0.181484654545784, "learning_rate": 5.213287983471862e-05, "loss": 0.0104, "step": 153630 }, { "epoch": 97.92224346717654, "grad_norm": 0.35499608516693115, "learning_rate": 5.212737327261315e-05, "loss": 0.0102, "step": 153640 }, { "epoch": 97.92861695347355, "grad_norm": 0.26387104392051697, "learning_rate": 5.212186668465806e-05, "loss": 0.0087, "step": 153650 }, { "epoch": 97.93499043977056, "grad_norm": 0.3699643015861511, "learning_rate": 5.211636007092031e-05, "loss": 0.008, "step": 153660 }, { "epoch": 97.94136392606755, "grad_norm": 0.30858656764030457, "learning_rate": 5.211085343146679e-05, "loss": 0.0095, "step": 153670 }, { "epoch": 97.94773741236456, "grad_norm": 0.4537908434867859, "learning_rate": 5.210534676636439e-05, "loss": 0.0135, "step": 153680 }, { "epoch": 97.95411089866157, "grad_norm": 0.48459410667419434, "learning_rate": 5.2099840075680064e-05, "loss": 0.0208, "step": 153690 }, { "epoch": 97.96048438495858, "grad_norm": 0.23257751762866974, "learning_rate": 5.209433335948067e-05, "loss": 0.0133, "step": 153700 }, { "epoch": 97.96685787125557, "grad_norm": 0.34180504083633423, "learning_rate": 5.208882661783318e-05, "loss": 0.0124, "step": 153710 }, { "epoch": 97.97323135755258, "grad_norm": 0.2348267138004303, "learning_rate": 5.208331985080445e-05, "loss": 0.0162, "step": 153720 }, { "epoch": 97.97960484384959, "grad_norm": 0.6168025135993958, "learning_rate": 5.207781305846142e-05, "loss": 0.0147, "step": 153730 }, { "epoch": 97.9859783301466, "grad_norm": 0.4028720557689667, "learning_rate": 5.207230624087099e-05, "loss": 0.0122, "step": 153740 }, { "epoch": 97.99235181644359, "grad_norm": 0.22823357582092285, "learning_rate": 5.206679939810011e-05, "loss": 0.0129, "step": 153750 }, { "epoch": 97.9987253027406, "grad_norm": 0.5082472562789917, "learning_rate": 5.2061292530215656e-05, "loss": 0.01, "step": 153760 }, { "epoch": 98.0050987890376, "grad_norm": 0.27186208963394165, "learning_rate": 5.205578563728454e-05, "loss": 0.0118, "step": 153770 }, { "epoch": 98.01147227533461, "grad_norm": 0.2829427421092987, "learning_rate": 5.20502787193737e-05, "loss": 0.008, "step": 153780 }, { "epoch": 98.01784576163162, "grad_norm": 0.1906144917011261, "learning_rate": 5.204477177655003e-05, "loss": 0.0183, "step": 153790 }, { "epoch": 98.02421924792861, "grad_norm": 0.4098353385925293, "learning_rate": 5.203926480888045e-05, "loss": 0.0164, "step": 153800 }, { "epoch": 98.03059273422562, "grad_norm": 0.5764397978782654, "learning_rate": 5.203375781643188e-05, "loss": 0.0187, "step": 153810 }, { "epoch": 98.03696622052263, "grad_norm": 0.2680857479572296, "learning_rate": 5.202825079927124e-05, "loss": 0.0113, "step": 153820 }, { "epoch": 98.04333970681964, "grad_norm": 0.20686204731464386, "learning_rate": 5.202274375746543e-05, "loss": 0.0095, "step": 153830 }, { "epoch": 98.04971319311663, "grad_norm": 0.23688280582427979, "learning_rate": 5.2017236691081374e-05, "loss": 0.015, "step": 153840 }, { "epoch": 98.05608667941364, "grad_norm": 0.2610875368118286, "learning_rate": 5.2011729600185986e-05, "loss": 0.007, "step": 153850 }, { "epoch": 98.06246016571065, "grad_norm": 0.27357161045074463, "learning_rate": 5.200622248484619e-05, "loss": 0.0114, "step": 153860 }, { "epoch": 98.06883365200765, "grad_norm": 0.3673422932624817, "learning_rate": 5.200071534512889e-05, "loss": 0.0135, "step": 153870 }, { "epoch": 98.07520713830465, "grad_norm": 0.22578459978103638, "learning_rate": 5.199520818110102e-05, "loss": 0.0125, "step": 153880 }, { "epoch": 98.08158062460166, "grad_norm": 0.3352314829826355, "learning_rate": 5.1989700992829484e-05, "loss": 0.0107, "step": 153890 }, { "epoch": 98.08795411089866, "grad_norm": 0.3152848482131958, "learning_rate": 5.198419378038118e-05, "loss": 0.0115, "step": 153900 }, { "epoch": 98.09432759719567, "grad_norm": 0.32359862327575684, "learning_rate": 5.1978686543823076e-05, "loss": 0.0098, "step": 153910 }, { "epoch": 98.10070108349267, "grad_norm": 0.1897839456796646, "learning_rate": 5.197317928322206e-05, "loss": 0.0094, "step": 153920 }, { "epoch": 98.10707456978967, "grad_norm": 0.47806915640830994, "learning_rate": 5.196767199864504e-05, "loss": 0.0138, "step": 153930 }, { "epoch": 98.11344805608668, "grad_norm": 0.36720162630081177, "learning_rate": 5.196216469015895e-05, "loss": 0.0161, "step": 153940 }, { "epoch": 98.11982154238369, "grad_norm": 0.33704298734664917, "learning_rate": 5.195665735783071e-05, "loss": 0.0148, "step": 153950 }, { "epoch": 98.12619502868068, "grad_norm": 0.29598504304885864, "learning_rate": 5.1951150001727244e-05, "loss": 0.0119, "step": 153960 }, { "epoch": 98.13256851497769, "grad_norm": 0.34717923402786255, "learning_rate": 5.194564262191545e-05, "loss": 0.0184, "step": 153970 }, { "epoch": 98.1389420012747, "grad_norm": 0.4820290207862854, "learning_rate": 5.194013521846226e-05, "loss": 0.0116, "step": 153980 }, { "epoch": 98.14531548757171, "grad_norm": 0.24551202356815338, "learning_rate": 5.1934627791434595e-05, "loss": 0.0144, "step": 153990 }, { "epoch": 98.1516889738687, "grad_norm": 0.5944722294807434, "learning_rate": 5.192912034089939e-05, "loss": 0.0123, "step": 154000 }, { "epoch": 98.15806246016571, "grad_norm": 0.3287455141544342, "learning_rate": 5.192361286692354e-05, "loss": 0.019, "step": 154010 }, { "epoch": 98.16443594646272, "grad_norm": 0.31727486848831177, "learning_rate": 5.191810536957399e-05, "loss": 0.0144, "step": 154020 }, { "epoch": 98.17080943275973, "grad_norm": 0.4623107612133026, "learning_rate": 5.191259784891763e-05, "loss": 0.0333, "step": 154030 }, { "epoch": 98.17718291905672, "grad_norm": 0.5252547264099121, "learning_rate": 5.190709030502141e-05, "loss": 0.0212, "step": 154040 }, { "epoch": 98.18355640535373, "grad_norm": 0.17360080778598785, "learning_rate": 5.1901582737952236e-05, "loss": 0.0156, "step": 154050 }, { "epoch": 98.18992989165073, "grad_norm": 0.33297625184059143, "learning_rate": 5.1896075147777045e-05, "loss": 0.0098, "step": 154060 }, { "epoch": 98.19630337794774, "grad_norm": 0.17485082149505615, "learning_rate": 5.189056753456275e-05, "loss": 0.0158, "step": 154070 }, { "epoch": 98.20267686424474, "grad_norm": 0.5697721242904663, "learning_rate": 5.188505989837627e-05, "loss": 0.0167, "step": 154080 }, { "epoch": 98.20905035054174, "grad_norm": 0.6438561677932739, "learning_rate": 5.187955223928454e-05, "loss": 0.0242, "step": 154090 }, { "epoch": 98.21542383683875, "grad_norm": 0.3070814311504364, "learning_rate": 5.1874044557354475e-05, "loss": 0.0157, "step": 154100 }, { "epoch": 98.22179732313576, "grad_norm": 0.5172650218009949, "learning_rate": 5.186853685265298e-05, "loss": 0.0121, "step": 154110 }, { "epoch": 98.22817080943275, "grad_norm": 0.24384024739265442, "learning_rate": 5.186302912524702e-05, "loss": 0.0067, "step": 154120 }, { "epoch": 98.23454429572976, "grad_norm": 0.33083078265190125, "learning_rate": 5.185752137520349e-05, "loss": 0.0095, "step": 154130 }, { "epoch": 98.24091778202677, "grad_norm": 0.5068547129631042, "learning_rate": 5.185201360258932e-05, "loss": 0.0131, "step": 154140 }, { "epoch": 98.24729126832378, "grad_norm": 0.31531938910484314, "learning_rate": 5.184650580747143e-05, "loss": 0.0112, "step": 154150 }, { "epoch": 98.25366475462077, "grad_norm": 0.3306485712528229, "learning_rate": 5.184099798991676e-05, "loss": 0.0114, "step": 154160 }, { "epoch": 98.26003824091778, "grad_norm": 0.14801965653896332, "learning_rate": 5.1835490149992236e-05, "loss": 0.0085, "step": 154170 }, { "epoch": 98.26641172721479, "grad_norm": 0.273418664932251, "learning_rate": 5.1829982287764754e-05, "loss": 0.0108, "step": 154180 }, { "epoch": 98.2727852135118, "grad_norm": 0.6161956787109375, "learning_rate": 5.1824474403301274e-05, "loss": 0.0094, "step": 154190 }, { "epoch": 98.27915869980879, "grad_norm": 0.2822195291519165, "learning_rate": 5.18189664966687e-05, "loss": 0.0237, "step": 154200 }, { "epoch": 98.2855321861058, "grad_norm": 0.5317549109458923, "learning_rate": 5.181345856793397e-05, "loss": 0.0211, "step": 154210 }, { "epoch": 98.2919056724028, "grad_norm": 0.3027542233467102, "learning_rate": 5.180795061716399e-05, "loss": 0.0102, "step": 154220 }, { "epoch": 98.29827915869981, "grad_norm": 0.24093860387802124, "learning_rate": 5.180244264442573e-05, "loss": 0.0109, "step": 154230 }, { "epoch": 98.3046526449968, "grad_norm": 0.2203730046749115, "learning_rate": 5.179693464978608e-05, "loss": 0.0127, "step": 154240 }, { "epoch": 98.31102613129381, "grad_norm": 0.3250463008880615, "learning_rate": 5.179142663331197e-05, "loss": 0.018, "step": 154250 }, { "epoch": 98.31739961759082, "grad_norm": 0.8564786314964294, "learning_rate": 5.1785918595070337e-05, "loss": 0.0178, "step": 154260 }, { "epoch": 98.32377310388783, "grad_norm": 0.46950680017471313, "learning_rate": 5.17804105351281e-05, "loss": 0.0119, "step": 154270 }, { "epoch": 98.33014659018482, "grad_norm": 0.3172660171985626, "learning_rate": 5.177490245355221e-05, "loss": 0.0166, "step": 154280 }, { "epoch": 98.33652007648183, "grad_norm": 0.4933871924877167, "learning_rate": 5.176939435040957e-05, "loss": 0.0173, "step": 154290 }, { "epoch": 98.34289356277884, "grad_norm": 0.3217349350452423, "learning_rate": 5.176388622576712e-05, "loss": 0.0104, "step": 154300 }, { "epoch": 98.34926704907585, "grad_norm": 0.5009505152702332, "learning_rate": 5.175837807969178e-05, "loss": 0.0109, "step": 154310 }, { "epoch": 98.35564053537284, "grad_norm": 0.17005330324172974, "learning_rate": 5.17528699122505e-05, "loss": 0.0097, "step": 154320 }, { "epoch": 98.36201402166985, "grad_norm": 0.32015231251716614, "learning_rate": 5.17473617235102e-05, "loss": 0.0122, "step": 154330 }, { "epoch": 98.36838750796686, "grad_norm": 0.5221731662750244, "learning_rate": 5.1741853513537785e-05, "loss": 0.0176, "step": 154340 }, { "epoch": 98.37476099426387, "grad_norm": 0.29479095339775085, "learning_rate": 5.173634528240022e-05, "loss": 0.0133, "step": 154350 }, { "epoch": 98.38113448056086, "grad_norm": 0.42271286249160767, "learning_rate": 5.1730837030164406e-05, "loss": 0.0115, "step": 154360 }, { "epoch": 98.38750796685787, "grad_norm": 0.2986777126789093, "learning_rate": 5.1725328756897306e-05, "loss": 0.0052, "step": 154370 }, { "epoch": 98.39388145315488, "grad_norm": 0.41202086210250854, "learning_rate": 5.1719820462665815e-05, "loss": 0.0189, "step": 154380 }, { "epoch": 98.40025493945188, "grad_norm": 0.35112833976745605, "learning_rate": 5.171431214753689e-05, "loss": 0.023, "step": 154390 }, { "epoch": 98.40662842574889, "grad_norm": 0.3189299702644348, "learning_rate": 5.1708803811577466e-05, "loss": 0.015, "step": 154400 }, { "epoch": 98.41300191204589, "grad_norm": 0.1796872615814209, "learning_rate": 5.170329545485444e-05, "loss": 0.0126, "step": 154410 }, { "epoch": 98.4193753983429, "grad_norm": 0.29029276967048645, "learning_rate": 5.169778707743478e-05, "loss": 0.0142, "step": 154420 }, { "epoch": 98.4257488846399, "grad_norm": 0.3529004454612732, "learning_rate": 5.16922786793854e-05, "loss": 0.0138, "step": 154430 }, { "epoch": 98.43212237093691, "grad_norm": 0.39378297328948975, "learning_rate": 5.168677026077324e-05, "loss": 0.0119, "step": 154440 }, { "epoch": 98.4384958572339, "grad_norm": 0.3879932761192322, "learning_rate": 5.1681261821665214e-05, "loss": 0.0151, "step": 154450 }, { "epoch": 98.44486934353091, "grad_norm": 0.5720310211181641, "learning_rate": 5.1675753362128285e-05, "loss": 0.0164, "step": 154460 }, { "epoch": 98.45124282982792, "grad_norm": 0.3142522871494293, "learning_rate": 5.167024488222937e-05, "loss": 0.019, "step": 154470 }, { "epoch": 98.45761631612493, "grad_norm": 0.22419798374176025, "learning_rate": 5.166473638203539e-05, "loss": 0.0126, "step": 154480 }, { "epoch": 98.46398980242192, "grad_norm": 0.43870630860328674, "learning_rate": 5.1659227861613314e-05, "loss": 0.0126, "step": 154490 }, { "epoch": 98.47036328871893, "grad_norm": 0.19178681075572968, "learning_rate": 5.1653719321030036e-05, "loss": 0.0159, "step": 154500 }, { "epoch": 98.47673677501594, "grad_norm": 0.20829389989376068, "learning_rate": 5.1648210760352514e-05, "loss": 0.0178, "step": 154510 }, { "epoch": 98.48311026131294, "grad_norm": 0.29754024744033813, "learning_rate": 5.164270217964766e-05, "loss": 0.0101, "step": 154520 }, { "epoch": 98.48948374760994, "grad_norm": 0.5164246559143066, "learning_rate": 5.163719357898245e-05, "loss": 0.0183, "step": 154530 }, { "epoch": 98.49585723390695, "grad_norm": 0.5189772248268127, "learning_rate": 5.163168495842376e-05, "loss": 0.0183, "step": 154540 }, { "epoch": 98.50223072020395, "grad_norm": 0.11767160892486572, "learning_rate": 5.1626176318038576e-05, "loss": 0.0076, "step": 154550 }, { "epoch": 98.50860420650096, "grad_norm": 0.21246616542339325, "learning_rate": 5.162066765789382e-05, "loss": 0.014, "step": 154560 }, { "epoch": 98.51497769279796, "grad_norm": 0.293222576379776, "learning_rate": 5.161515897805641e-05, "loss": 0.015, "step": 154570 }, { "epoch": 98.52135117909496, "grad_norm": 0.43075063824653625, "learning_rate": 5.16096502785933e-05, "loss": 0.0285, "step": 154580 }, { "epoch": 98.52772466539197, "grad_norm": 0.12363023310899734, "learning_rate": 5.160414155957142e-05, "loss": 0.0128, "step": 154590 }, { "epoch": 98.53409815168898, "grad_norm": 0.5207599997520447, "learning_rate": 5.1598632821057705e-05, "loss": 0.0173, "step": 154600 }, { "epoch": 98.54047163798597, "grad_norm": 0.33351919054985046, "learning_rate": 5.159312406311908e-05, "loss": 0.0162, "step": 154610 }, { "epoch": 98.54684512428298, "grad_norm": 0.4469392001628876, "learning_rate": 5.158761528582251e-05, "loss": 0.0099, "step": 154620 }, { "epoch": 98.55321861057999, "grad_norm": 0.18054184317588806, "learning_rate": 5.158210648923489e-05, "loss": 0.0201, "step": 154630 }, { "epoch": 98.559592096877, "grad_norm": 0.2721685767173767, "learning_rate": 5.157659767342321e-05, "loss": 0.0085, "step": 154640 }, { "epoch": 98.56596558317399, "grad_norm": 0.2600773572921753, "learning_rate": 5.157108883845437e-05, "loss": 0.0089, "step": 154650 }, { "epoch": 98.572339069471, "grad_norm": 0.2348613739013672, "learning_rate": 5.156557998439532e-05, "loss": 0.0069, "step": 154660 }, { "epoch": 98.578712555768, "grad_norm": 0.2292715609073639, "learning_rate": 5.1560071111313e-05, "loss": 0.0152, "step": 154670 }, { "epoch": 98.58508604206502, "grad_norm": 0.4233141541481018, "learning_rate": 5.1554562219274325e-05, "loss": 0.0115, "step": 154680 }, { "epoch": 98.59145952836201, "grad_norm": 0.3240736722946167, "learning_rate": 5.1549053308346273e-05, "loss": 0.014, "step": 154690 }, { "epoch": 98.59783301465902, "grad_norm": 0.19381096959114075, "learning_rate": 5.154354437859574e-05, "loss": 0.0163, "step": 154700 }, { "epoch": 98.60420650095602, "grad_norm": 0.40221551060676575, "learning_rate": 5.15380354300897e-05, "loss": 0.0067, "step": 154710 }, { "epoch": 98.61057998725303, "grad_norm": 0.24813121557235718, "learning_rate": 5.153252646289507e-05, "loss": 0.0092, "step": 154720 }, { "epoch": 98.61695347355003, "grad_norm": 0.36479318141937256, "learning_rate": 5.15270174770788e-05, "loss": 0.0166, "step": 154730 }, { "epoch": 98.62332695984703, "grad_norm": 0.20187965035438538, "learning_rate": 5.152150847270781e-05, "loss": 0.0161, "step": 154740 }, { "epoch": 98.62970044614404, "grad_norm": 0.31552430987358093, "learning_rate": 5.1515999449849085e-05, "loss": 0.0093, "step": 154750 }, { "epoch": 98.63607393244105, "grad_norm": 0.7701984643936157, "learning_rate": 5.1510490408569514e-05, "loss": 0.0144, "step": 154760 }, { "epoch": 98.64244741873804, "grad_norm": 0.1880282163619995, "learning_rate": 5.150498134893607e-05, "loss": 0.0112, "step": 154770 }, { "epoch": 98.64882090503505, "grad_norm": 0.4399813413619995, "learning_rate": 5.1499472271015666e-05, "loss": 0.009, "step": 154780 }, { "epoch": 98.65519439133206, "grad_norm": 0.30926457047462463, "learning_rate": 5.149396317487528e-05, "loss": 0.0094, "step": 154790 }, { "epoch": 98.66156787762907, "grad_norm": 0.2663269340991974, "learning_rate": 5.148845406058182e-05, "loss": 0.0081, "step": 154800 }, { "epoch": 98.66794136392606, "grad_norm": 0.3683769702911377, "learning_rate": 5.148294492820224e-05, "loss": 0.0167, "step": 154810 }, { "epoch": 98.67431485022307, "grad_norm": 0.31261125206947327, "learning_rate": 5.147743577780346e-05, "loss": 0.0106, "step": 154820 }, { "epoch": 98.68068833652008, "grad_norm": 0.47341418266296387, "learning_rate": 5.147192660945247e-05, "loss": 0.0111, "step": 154830 }, { "epoch": 98.68706182281709, "grad_norm": 0.24053357541561127, "learning_rate": 5.146641742321616e-05, "loss": 0.0066, "step": 154840 }, { "epoch": 98.69343530911408, "grad_norm": 0.25421467423439026, "learning_rate": 5.146090821916151e-05, "loss": 0.0113, "step": 154850 }, { "epoch": 98.69980879541109, "grad_norm": 0.39404815435409546, "learning_rate": 5.1455398997355434e-05, "loss": 0.0099, "step": 154860 }, { "epoch": 98.7061822817081, "grad_norm": 0.2070356011390686, "learning_rate": 5.144988975786489e-05, "loss": 0.0175, "step": 154870 }, { "epoch": 98.7125557680051, "grad_norm": 0.4026572108268738, "learning_rate": 5.144438050075683e-05, "loss": 0.0101, "step": 154880 }, { "epoch": 98.7189292543021, "grad_norm": 0.1946873962879181, "learning_rate": 5.143887122609816e-05, "loss": 0.0102, "step": 154890 }, { "epoch": 98.7253027405991, "grad_norm": 0.3395257890224457, "learning_rate": 5.143336193395586e-05, "loss": 0.0091, "step": 154900 }, { "epoch": 98.73167622689611, "grad_norm": 0.3217450678348541, "learning_rate": 5.142785262439685e-05, "loss": 0.0107, "step": 154910 }, { "epoch": 98.73804971319312, "grad_norm": 0.24184224009513855, "learning_rate": 5.1422343297488096e-05, "loss": 0.0146, "step": 154920 }, { "epoch": 98.74442319949011, "grad_norm": 0.2243834286928177, "learning_rate": 5.1416833953296506e-05, "loss": 0.011, "step": 154930 }, { "epoch": 98.75079668578712, "grad_norm": 0.17343956232070923, "learning_rate": 5.1411324591889065e-05, "loss": 0.012, "step": 154940 }, { "epoch": 98.75717017208413, "grad_norm": 0.4316220283508301, "learning_rate": 5.1405815213332684e-05, "loss": 0.0116, "step": 154950 }, { "epoch": 98.76354365838114, "grad_norm": 0.27392634749412537, "learning_rate": 5.140030581769434e-05, "loss": 0.0206, "step": 154960 }, { "epoch": 98.76991714467813, "grad_norm": 0.13710230588912964, "learning_rate": 5.139479640504096e-05, "loss": 0.0233, "step": 154970 }, { "epoch": 98.77629063097514, "grad_norm": 0.6559505462646484, "learning_rate": 5.1389286975439464e-05, "loss": 0.0162, "step": 154980 }, { "epoch": 98.78266411727215, "grad_norm": 0.684042751789093, "learning_rate": 5.1383777528956834e-05, "loss": 0.0177, "step": 154990 }, { "epoch": 98.78903760356916, "grad_norm": 0.2361699938774109, "learning_rate": 5.137826806565998e-05, "loss": 0.0176, "step": 155000 }, { "epoch": 98.79541108986615, "grad_norm": 0.2358238250017166, "learning_rate": 5.13727585856159e-05, "loss": 0.0083, "step": 155010 }, { "epoch": 98.80178457616316, "grad_norm": 0.27023133635520935, "learning_rate": 5.136724908889148e-05, "loss": 0.0103, "step": 155020 }, { "epoch": 98.80815806246017, "grad_norm": 0.6180278658866882, "learning_rate": 5.136173957555371e-05, "loss": 0.0183, "step": 155030 }, { "epoch": 98.81453154875717, "grad_norm": 0.2236386388540268, "learning_rate": 5.135623004566952e-05, "loss": 0.0126, "step": 155040 }, { "epoch": 98.82090503505418, "grad_norm": 0.2548777461051941, "learning_rate": 5.1350720499305835e-05, "loss": 0.0091, "step": 155050 }, { "epoch": 98.82727852135118, "grad_norm": 0.4003305435180664, "learning_rate": 5.134521093652964e-05, "loss": 0.017, "step": 155060 }, { "epoch": 98.83365200764818, "grad_norm": 0.4695996642112732, "learning_rate": 5.133970135740785e-05, "loss": 0.0147, "step": 155070 }, { "epoch": 98.84002549394519, "grad_norm": 0.5772498250007629, "learning_rate": 5.133419176200743e-05, "loss": 0.0157, "step": 155080 }, { "epoch": 98.8463989802422, "grad_norm": 0.6990782618522644, "learning_rate": 5.1328682150395314e-05, "loss": 0.0142, "step": 155090 }, { "epoch": 98.8527724665392, "grad_norm": 0.505440890789032, "learning_rate": 5.1323172522638465e-05, "loss": 0.0122, "step": 155100 }, { "epoch": 98.8591459528362, "grad_norm": 0.31949061155319214, "learning_rate": 5.1317662878803806e-05, "loss": 0.0092, "step": 155110 }, { "epoch": 98.86551943913321, "grad_norm": 0.23776234686374664, "learning_rate": 5.131215321895831e-05, "loss": 0.0246, "step": 155120 }, { "epoch": 98.87189292543022, "grad_norm": 0.4295574426651001, "learning_rate": 5.130664354316893e-05, "loss": 0.013, "step": 155130 }, { "epoch": 98.87826641172721, "grad_norm": 0.28643539547920227, "learning_rate": 5.130113385150257e-05, "loss": 0.0071, "step": 155140 }, { "epoch": 98.88463989802422, "grad_norm": 0.4471513330936432, "learning_rate": 5.129562414402622e-05, "loss": 0.0129, "step": 155150 }, { "epoch": 98.89101338432123, "grad_norm": 0.2521721124649048, "learning_rate": 5.12901144208068e-05, "loss": 0.0214, "step": 155160 }, { "epoch": 98.89738687061823, "grad_norm": 0.46161285042762756, "learning_rate": 5.1284604681911285e-05, "loss": 0.0226, "step": 155170 }, { "epoch": 98.90376035691523, "grad_norm": 0.561133623123169, "learning_rate": 5.12790949274066e-05, "loss": 0.0159, "step": 155180 }, { "epoch": 98.91013384321224, "grad_norm": 0.23896737396717072, "learning_rate": 5.1273585157359716e-05, "loss": 0.014, "step": 155190 }, { "epoch": 98.91650732950924, "grad_norm": 0.39847221970558167, "learning_rate": 5.126807537183755e-05, "loss": 0.0157, "step": 155200 }, { "epoch": 98.92288081580625, "grad_norm": 0.7278721928596497, "learning_rate": 5.1262565570907095e-05, "loss": 0.0119, "step": 155210 }, { "epoch": 98.92925430210325, "grad_norm": 0.25674331188201904, "learning_rate": 5.1257055754635264e-05, "loss": 0.01, "step": 155220 }, { "epoch": 98.93562778840025, "grad_norm": 0.1612807661294937, "learning_rate": 5.125154592308902e-05, "loss": 0.0154, "step": 155230 }, { "epoch": 98.94200127469726, "grad_norm": 0.23101358115673065, "learning_rate": 5.1246036076335316e-05, "loss": 0.0055, "step": 155240 }, { "epoch": 98.94837476099427, "grad_norm": 0.22930501401424408, "learning_rate": 5.124052621444108e-05, "loss": 0.0117, "step": 155250 }, { "epoch": 98.95474824729126, "grad_norm": 0.2103288471698761, "learning_rate": 5.12350163374733e-05, "loss": 0.0102, "step": 155260 }, { "epoch": 98.96112173358827, "grad_norm": 0.3337845504283905, "learning_rate": 5.122950644549889e-05, "loss": 0.0288, "step": 155270 }, { "epoch": 98.96749521988528, "grad_norm": 0.4369223415851593, "learning_rate": 5.122399653858483e-05, "loss": 0.0189, "step": 155280 }, { "epoch": 98.97386870618229, "grad_norm": 0.2841877341270447, "learning_rate": 5.121848661679805e-05, "loss": 0.0124, "step": 155290 }, { "epoch": 98.98024219247928, "grad_norm": 0.2001449316740036, "learning_rate": 5.1212976680205505e-05, "loss": 0.0181, "step": 155300 }, { "epoch": 98.98661567877629, "grad_norm": 0.36649444699287415, "learning_rate": 5.1207466728874156e-05, "loss": 0.0193, "step": 155310 }, { "epoch": 98.9929891650733, "grad_norm": 0.32804813981056213, "learning_rate": 5.120195676287093e-05, "loss": 0.0116, "step": 155320 }, { "epoch": 98.9993626513703, "grad_norm": 0.5633274912834167, "learning_rate": 5.119644678226281e-05, "loss": 0.0144, "step": 155330 }, { "epoch": 99.0057361376673, "grad_norm": 0.4065292477607727, "learning_rate": 5.119093678711672e-05, "loss": 0.0154, "step": 155340 }, { "epoch": 99.0121096239643, "grad_norm": 0.2852652966976166, "learning_rate": 5.118542677749964e-05, "loss": 0.0112, "step": 155350 }, { "epoch": 99.01848311026131, "grad_norm": 0.29619333148002625, "learning_rate": 5.117991675347849e-05, "loss": 0.0158, "step": 155360 }, { "epoch": 99.02485659655832, "grad_norm": 0.14595761895179749, "learning_rate": 5.1174406715120236e-05, "loss": 0.0085, "step": 155370 }, { "epoch": 99.03123008285532, "grad_norm": 0.4411018192768097, "learning_rate": 5.116889666249185e-05, "loss": 0.0221, "step": 155380 }, { "epoch": 99.03760356915232, "grad_norm": 0.27945271134376526, "learning_rate": 5.116338659566025e-05, "loss": 0.0116, "step": 155390 }, { "epoch": 99.04397705544933, "grad_norm": 0.3646855354309082, "learning_rate": 5.11578765146924e-05, "loss": 0.0082, "step": 155400 }, { "epoch": 99.05035054174634, "grad_norm": 0.2193252295255661, "learning_rate": 5.115236641965526e-05, "loss": 0.0104, "step": 155410 }, { "epoch": 99.05672402804333, "grad_norm": 0.1669130027294159, "learning_rate": 5.11468563106158e-05, "loss": 0.0089, "step": 155420 }, { "epoch": 99.06309751434034, "grad_norm": 0.5155532956123352, "learning_rate": 5.114134618764093e-05, "loss": 0.0139, "step": 155430 }, { "epoch": 99.06947100063735, "grad_norm": 0.21704740822315216, "learning_rate": 5.1135836050797634e-05, "loss": 0.0107, "step": 155440 }, { "epoch": 99.07584448693436, "grad_norm": 0.25931113958358765, "learning_rate": 5.113032590015285e-05, "loss": 0.0136, "step": 155450 }, { "epoch": 99.08221797323135, "grad_norm": 0.1668749898672104, "learning_rate": 5.112481573577356e-05, "loss": 0.0059, "step": 155460 }, { "epoch": 99.08859145952836, "grad_norm": 0.3811054825782776, "learning_rate": 5.111930555772668e-05, "loss": 0.0169, "step": 155470 }, { "epoch": 99.09496494582537, "grad_norm": 0.5909488797187805, "learning_rate": 5.111379536607919e-05, "loss": 0.0153, "step": 155480 }, { "epoch": 99.10133843212238, "grad_norm": 0.27697789669036865, "learning_rate": 5.110828516089802e-05, "loss": 0.0128, "step": 155490 }, { "epoch": 99.10771191841937, "grad_norm": 0.5183128118515015, "learning_rate": 5.1102774942250154e-05, "loss": 0.0291, "step": 155500 }, { "epoch": 99.11408540471638, "grad_norm": 0.2583252191543579, "learning_rate": 5.109726471020252e-05, "loss": 0.0135, "step": 155510 }, { "epoch": 99.12045889101339, "grad_norm": 0.312069833278656, "learning_rate": 5.1091754464822095e-05, "loss": 0.009, "step": 155520 }, { "epoch": 99.1268323773104, "grad_norm": 0.211601123213768, "learning_rate": 5.1086244206175815e-05, "loss": 0.0164, "step": 155530 }, { "epoch": 99.13320586360739, "grad_norm": 0.23753590881824493, "learning_rate": 5.108073393433065e-05, "loss": 0.0121, "step": 155540 }, { "epoch": 99.1395793499044, "grad_norm": 0.3479219675064087, "learning_rate": 5.1075223649353534e-05, "loss": 0.0147, "step": 155550 }, { "epoch": 99.1459528362014, "grad_norm": 0.6493948698043823, "learning_rate": 5.106971335131144e-05, "loss": 0.0192, "step": 155560 }, { "epoch": 99.15232632249841, "grad_norm": 0.36732950806617737, "learning_rate": 5.106420304027132e-05, "loss": 0.0116, "step": 155570 }, { "epoch": 99.1586998087954, "grad_norm": 0.34845539927482605, "learning_rate": 5.105869271630014e-05, "loss": 0.0105, "step": 155580 }, { "epoch": 99.16507329509241, "grad_norm": 0.37438589334487915, "learning_rate": 5.105318237946482e-05, "loss": 0.0091, "step": 155590 }, { "epoch": 99.17144678138942, "grad_norm": 0.40019121766090393, "learning_rate": 5.1047672029832364e-05, "loss": 0.011, "step": 155600 }, { "epoch": 99.17782026768643, "grad_norm": 0.28326258063316345, "learning_rate": 5.10421616674697e-05, "loss": 0.0113, "step": 155610 }, { "epoch": 99.18419375398342, "grad_norm": 0.25842636823654175, "learning_rate": 5.1036651292443774e-05, "loss": 0.0071, "step": 155620 }, { "epoch": 99.19056724028043, "grad_norm": 0.2629290223121643, "learning_rate": 5.103114090482156e-05, "loss": 0.0076, "step": 155630 }, { "epoch": 99.19694072657744, "grad_norm": 0.3478277921676636, "learning_rate": 5.102563050467002e-05, "loss": 0.0129, "step": 155640 }, { "epoch": 99.20331421287445, "grad_norm": 0.1825607866048813, "learning_rate": 5.102012009205609e-05, "loss": 0.0099, "step": 155650 }, { "epoch": 99.20968769917145, "grad_norm": 0.11866192519664764, "learning_rate": 5.101460966704674e-05, "loss": 0.0056, "step": 155660 }, { "epoch": 99.21606118546845, "grad_norm": 0.49371254444122314, "learning_rate": 5.1009099229708935e-05, "loss": 0.0192, "step": 155670 }, { "epoch": 99.22243467176546, "grad_norm": 0.1989348828792572, "learning_rate": 5.10035887801096e-05, "loss": 0.0104, "step": 155680 }, { "epoch": 99.22880815806246, "grad_norm": 0.2659602463245392, "learning_rate": 5.0998078318315736e-05, "loss": 0.0123, "step": 155690 }, { "epoch": 99.23518164435947, "grad_norm": 0.4583275616168976, "learning_rate": 5.099256784439428e-05, "loss": 0.011, "step": 155700 }, { "epoch": 99.24155513065647, "grad_norm": 0.23785732686519623, "learning_rate": 5.0987057358412174e-05, "loss": 0.011, "step": 155710 }, { "epoch": 99.24792861695347, "grad_norm": 0.26196473836898804, "learning_rate": 5.098154686043639e-05, "loss": 0.0093, "step": 155720 }, { "epoch": 99.25430210325048, "grad_norm": 0.24107283353805542, "learning_rate": 5.0976036350533885e-05, "loss": 0.0154, "step": 155730 }, { "epoch": 99.26067558954749, "grad_norm": 0.2478841096162796, "learning_rate": 5.097052582877162e-05, "loss": 0.0169, "step": 155740 }, { "epoch": 99.26704907584448, "grad_norm": 0.3470514714717865, "learning_rate": 5.096501529521655e-05, "loss": 0.0088, "step": 155750 }, { "epoch": 99.27342256214149, "grad_norm": 0.17444375157356262, "learning_rate": 5.0959504749935626e-05, "loss": 0.0092, "step": 155760 }, { "epoch": 99.2797960484385, "grad_norm": 0.2638604938983917, "learning_rate": 5.095399419299584e-05, "loss": 0.0148, "step": 155770 }, { "epoch": 99.2861695347355, "grad_norm": 0.21993526816368103, "learning_rate": 5.0948483624464085e-05, "loss": 0.0135, "step": 155780 }, { "epoch": 99.2925430210325, "grad_norm": 0.30740615725517273, "learning_rate": 5.0942973044407386e-05, "loss": 0.0119, "step": 155790 }, { "epoch": 99.29891650732951, "grad_norm": 0.2948124408721924, "learning_rate": 5.093746245289266e-05, "loss": 0.0134, "step": 155800 }, { "epoch": 99.30528999362652, "grad_norm": 0.3926822543144226, "learning_rate": 5.093195184998689e-05, "loss": 0.0079, "step": 155810 }, { "epoch": 99.31166347992352, "grad_norm": 0.4710669219493866, "learning_rate": 5.0926441235757004e-05, "loss": 0.0179, "step": 155820 }, { "epoch": 99.31803696622052, "grad_norm": 0.376798540353775, "learning_rate": 5.092093061027001e-05, "loss": 0.0092, "step": 155830 }, { "epoch": 99.32441045251753, "grad_norm": 0.23420186340808868, "learning_rate": 5.0915419973592825e-05, "loss": 0.0115, "step": 155840 }, { "epoch": 99.33078393881453, "grad_norm": 0.26896411180496216, "learning_rate": 5.090990932579243e-05, "loss": 0.0097, "step": 155850 }, { "epoch": 99.33715742511154, "grad_norm": 0.5857445001602173, "learning_rate": 5.090439866693577e-05, "loss": 0.0164, "step": 155860 }, { "epoch": 99.34353091140854, "grad_norm": 0.2992708384990692, "learning_rate": 5.0898887997089814e-05, "loss": 0.0148, "step": 155870 }, { "epoch": 99.34990439770554, "grad_norm": 0.36370986700057983, "learning_rate": 5.0893377316321534e-05, "loss": 0.0105, "step": 155880 }, { "epoch": 99.35627788400255, "grad_norm": 0.29803603887557983, "learning_rate": 5.0887866624697856e-05, "loss": 0.0068, "step": 155890 }, { "epoch": 99.36265137029956, "grad_norm": 0.5728763937950134, "learning_rate": 5.0882355922285776e-05, "loss": 0.0157, "step": 155900 }, { "epoch": 99.36902485659655, "grad_norm": 0.426893949508667, "learning_rate": 5.087684520915221e-05, "loss": 0.0094, "step": 155910 }, { "epoch": 99.37539834289356, "grad_norm": 0.30479905009269714, "learning_rate": 5.0871334485364185e-05, "loss": 0.0123, "step": 155920 }, { "epoch": 99.38177182919057, "grad_norm": 0.19078660011291504, "learning_rate": 5.0865823750988606e-05, "loss": 0.0106, "step": 155930 }, { "epoch": 99.38814531548758, "grad_norm": 0.27964386343955994, "learning_rate": 5.0860313006092444e-05, "loss": 0.0112, "step": 155940 }, { "epoch": 99.39451880178457, "grad_norm": 0.31125614047050476, "learning_rate": 5.085480225074268e-05, "loss": 0.0156, "step": 155950 }, { "epoch": 99.40089228808158, "grad_norm": 0.2374599128961563, "learning_rate": 5.0849291485006244e-05, "loss": 0.0062, "step": 155960 }, { "epoch": 99.40726577437859, "grad_norm": 0.3531534969806671, "learning_rate": 5.084378070895014e-05, "loss": 0.0106, "step": 155970 }, { "epoch": 99.4136392606756, "grad_norm": 0.12947697937488556, "learning_rate": 5.083826992264128e-05, "loss": 0.0096, "step": 155980 }, { "epoch": 99.42001274697259, "grad_norm": 0.25544652342796326, "learning_rate": 5.0832759126146665e-05, "loss": 0.0111, "step": 155990 }, { "epoch": 99.4263862332696, "grad_norm": 0.3076120913028717, "learning_rate": 5.082724831953323e-05, "loss": 0.0099, "step": 156000 }, { "epoch": 99.4327597195666, "grad_norm": 0.2055576592683792, "learning_rate": 5.082173750286795e-05, "loss": 0.0093, "step": 156010 }, { "epoch": 99.43913320586361, "grad_norm": 0.1709987372159958, "learning_rate": 5.081622667621779e-05, "loss": 0.0143, "step": 156020 }, { "epoch": 99.4455066921606, "grad_norm": 0.3651348054409027, "learning_rate": 5.081071583964969e-05, "loss": 0.0111, "step": 156030 }, { "epoch": 99.45188017845761, "grad_norm": 0.2676248252391815, "learning_rate": 5.080520499323065e-05, "loss": 0.0098, "step": 156040 }, { "epoch": 99.45825366475462, "grad_norm": 0.3292093575000763, "learning_rate": 5.079969413702759e-05, "loss": 0.0178, "step": 156050 }, { "epoch": 99.46462715105163, "grad_norm": 0.48330843448638916, "learning_rate": 5.07941832711075e-05, "loss": 0.0178, "step": 156060 }, { "epoch": 99.47100063734862, "grad_norm": 0.4120030403137207, "learning_rate": 5.0788672395537316e-05, "loss": 0.013, "step": 156070 }, { "epoch": 99.47737412364563, "grad_norm": 0.24776072800159454, "learning_rate": 5.078316151038404e-05, "loss": 0.023, "step": 156080 }, { "epoch": 99.48374760994264, "grad_norm": 0.3588331341743469, "learning_rate": 5.077765061571461e-05, "loss": 0.0229, "step": 156090 }, { "epoch": 99.49012109623965, "grad_norm": 0.3254258632659912, "learning_rate": 5.077213971159598e-05, "loss": 0.0183, "step": 156100 }, { "epoch": 99.49649458253664, "grad_norm": 0.40409037470817566, "learning_rate": 5.076662879809513e-05, "loss": 0.0124, "step": 156110 }, { "epoch": 99.50286806883365, "grad_norm": 0.26359736919403076, "learning_rate": 5.0761117875279007e-05, "loss": 0.0077, "step": 156120 }, { "epoch": 99.50924155513066, "grad_norm": 0.33201324939727783, "learning_rate": 5.0755606943214604e-05, "loss": 0.0088, "step": 156130 }, { "epoch": 99.51561504142767, "grad_norm": 0.27565887570381165, "learning_rate": 5.075009600196884e-05, "loss": 0.0138, "step": 156140 }, { "epoch": 99.52198852772466, "grad_norm": 0.3102879822254181, "learning_rate": 5.0744585051608716e-05, "loss": 0.0114, "step": 156150 }, { "epoch": 99.52836201402167, "grad_norm": 0.30275440216064453, "learning_rate": 5.073907409220117e-05, "loss": 0.0078, "step": 156160 }, { "epoch": 99.53473550031867, "grad_norm": 0.36678338050842285, "learning_rate": 5.073356312381319e-05, "loss": 0.013, "step": 156170 }, { "epoch": 99.54110898661568, "grad_norm": 0.16903693974018097, "learning_rate": 5.0728052146511704e-05, "loss": 0.0165, "step": 156180 }, { "epoch": 99.54748247291268, "grad_norm": 0.12443340569734573, "learning_rate": 5.0722541160363726e-05, "loss": 0.0122, "step": 156190 }, { "epoch": 99.55385595920968, "grad_norm": 0.6098366975784302, "learning_rate": 5.0717030165436164e-05, "loss": 0.0226, "step": 156200 }, { "epoch": 99.56022944550669, "grad_norm": 0.636284589767456, "learning_rate": 5.071151916179603e-05, "loss": 0.0234, "step": 156210 }, { "epoch": 99.5666029318037, "grad_norm": 0.29480957984924316, "learning_rate": 5.070600814951024e-05, "loss": 0.012, "step": 156220 }, { "epoch": 99.5729764181007, "grad_norm": 0.2818854749202728, "learning_rate": 5.0700497128645804e-05, "loss": 0.0157, "step": 156230 }, { "epoch": 99.5793499043977, "grad_norm": 0.15630872547626495, "learning_rate": 5.069498609926967e-05, "loss": 0.0081, "step": 156240 }, { "epoch": 99.58572339069471, "grad_norm": 0.17214764654636383, "learning_rate": 5.0689475061448786e-05, "loss": 0.0095, "step": 156250 }, { "epoch": 99.59209687699172, "grad_norm": 0.48634839057922363, "learning_rate": 5.068396401525014e-05, "loss": 0.0156, "step": 156260 }, { "epoch": 99.59847036328873, "grad_norm": 0.5321304798126221, "learning_rate": 5.0678452960740684e-05, "loss": 0.0157, "step": 156270 }, { "epoch": 99.60484384958572, "grad_norm": 0.45532649755477905, "learning_rate": 5.067294189798737e-05, "loss": 0.0123, "step": 156280 }, { "epoch": 99.61121733588273, "grad_norm": 0.1982566863298416, "learning_rate": 5.06674308270572e-05, "loss": 0.0127, "step": 156290 }, { "epoch": 99.61759082217974, "grad_norm": 0.2460765689611435, "learning_rate": 5.066191974801708e-05, "loss": 0.0165, "step": 156300 }, { "epoch": 99.62396430847674, "grad_norm": 0.34185314178466797, "learning_rate": 5.065640866093404e-05, "loss": 0.0171, "step": 156310 }, { "epoch": 99.63033779477374, "grad_norm": 0.29867738485336304, "learning_rate": 5.065089756587501e-05, "loss": 0.0102, "step": 156320 }, { "epoch": 99.63671128107075, "grad_norm": 0.27603310346603394, "learning_rate": 5.0645386462906965e-05, "loss": 0.0094, "step": 156330 }, { "epoch": 99.64308476736775, "grad_norm": 0.30273663997650146, "learning_rate": 5.063987535209685e-05, "loss": 0.0083, "step": 156340 }, { "epoch": 99.64945825366476, "grad_norm": 0.4556749165058136, "learning_rate": 5.0634364233511655e-05, "loss": 0.0106, "step": 156350 }, { "epoch": 99.65583173996176, "grad_norm": 0.26591700315475464, "learning_rate": 5.0628853107218336e-05, "loss": 0.0098, "step": 156360 }, { "epoch": 99.66220522625876, "grad_norm": 0.16451002657413483, "learning_rate": 5.062334197328386e-05, "loss": 0.0155, "step": 156370 }, { "epoch": 99.66857871255577, "grad_norm": 1.0991692543029785, "learning_rate": 5.0617830831775184e-05, "loss": 0.0083, "step": 156380 }, { "epoch": 99.67495219885278, "grad_norm": 0.43002232909202576, "learning_rate": 5.061231968275927e-05, "loss": 0.0239, "step": 156390 }, { "epoch": 99.68132568514977, "grad_norm": 0.20098745822906494, "learning_rate": 5.0606808526303115e-05, "loss": 0.0162, "step": 156400 }, { "epoch": 99.68769917144678, "grad_norm": 0.4041197597980499, "learning_rate": 5.0601297362473646e-05, "loss": 0.0121, "step": 156410 }, { "epoch": 99.69407265774379, "grad_norm": 0.4311886131763458, "learning_rate": 5.0595786191337854e-05, "loss": 0.0162, "step": 156420 }, { "epoch": 99.7004461440408, "grad_norm": 0.359276682138443, "learning_rate": 5.059027501296271e-05, "loss": 0.0127, "step": 156430 }, { "epoch": 99.70681963033779, "grad_norm": 0.6419093608856201, "learning_rate": 5.058476382741515e-05, "loss": 0.0197, "step": 156440 }, { "epoch": 99.7131931166348, "grad_norm": 0.5533326864242554, "learning_rate": 5.057925263476216e-05, "loss": 0.023, "step": 156450 }, { "epoch": 99.7195666029318, "grad_norm": 0.2725984752178192, "learning_rate": 5.0573741435070707e-05, "loss": 0.0089, "step": 156460 }, { "epoch": 99.72594008922881, "grad_norm": 0.2384340614080429, "learning_rate": 5.056823022840775e-05, "loss": 0.0123, "step": 156470 }, { "epoch": 99.73231357552581, "grad_norm": 0.2153787612915039, "learning_rate": 5.056271901484026e-05, "loss": 0.0236, "step": 156480 }, { "epoch": 99.73868706182282, "grad_norm": 0.29640814661979675, "learning_rate": 5.0557207794435213e-05, "loss": 0.0109, "step": 156490 }, { "epoch": 99.74506054811982, "grad_norm": 0.18337497115135193, "learning_rate": 5.0551696567259566e-05, "loss": 0.0103, "step": 156500 }, { "epoch": 99.75143403441683, "grad_norm": 0.5862196683883667, "learning_rate": 5.054618533338027e-05, "loss": 0.0134, "step": 156510 }, { "epoch": 99.75780752071383, "grad_norm": 0.20478704571723938, "learning_rate": 5.054067409286431e-05, "loss": 0.0102, "step": 156520 }, { "epoch": 99.76418100701083, "grad_norm": 0.3433673679828644, "learning_rate": 5.053516284577865e-05, "loss": 0.0214, "step": 156530 }, { "epoch": 99.77055449330784, "grad_norm": 0.2779102623462677, "learning_rate": 5.052965159219026e-05, "loss": 0.0083, "step": 156540 }, { "epoch": 99.77692797960485, "grad_norm": 0.43407824635505676, "learning_rate": 5.0524140332166094e-05, "loss": 0.0076, "step": 156550 }, { "epoch": 99.78330146590184, "grad_norm": 0.39178207516670227, "learning_rate": 5.051862906577314e-05, "loss": 0.0119, "step": 156560 }, { "epoch": 99.78967495219885, "grad_norm": 0.5111430883407593, "learning_rate": 5.051311779307835e-05, "loss": 0.0239, "step": 156570 }, { "epoch": 99.79604843849586, "grad_norm": 0.4787918031215668, "learning_rate": 5.050760651414869e-05, "loss": 0.0186, "step": 156580 }, { "epoch": 99.80242192479287, "grad_norm": 0.36320891976356506, "learning_rate": 5.050209522905114e-05, "loss": 0.0169, "step": 156590 }, { "epoch": 99.80879541108986, "grad_norm": 0.5774436593055725, "learning_rate": 5.049658393785265e-05, "loss": 0.0116, "step": 156600 }, { "epoch": 99.81516889738687, "grad_norm": 0.22433722019195557, "learning_rate": 5.04910726406202e-05, "loss": 0.0147, "step": 156610 }, { "epoch": 99.82154238368388, "grad_norm": 0.5400145649909973, "learning_rate": 5.048556133742075e-05, "loss": 0.0184, "step": 156620 }, { "epoch": 99.82791586998088, "grad_norm": 1.5766167640686035, "learning_rate": 5.048005002832128e-05, "loss": 0.0133, "step": 156630 }, { "epoch": 99.83428935627788, "grad_norm": 0.2978486120700836, "learning_rate": 5.047453871338873e-05, "loss": 0.0163, "step": 156640 }, { "epoch": 99.84066284257489, "grad_norm": 0.4662528336048126, "learning_rate": 5.046902739269011e-05, "loss": 0.0105, "step": 156650 }, { "epoch": 99.8470363288719, "grad_norm": 0.17132021486759186, "learning_rate": 5.046351606629236e-05, "loss": 0.0138, "step": 156660 }, { "epoch": 99.8534098151689, "grad_norm": 0.1729094237089157, "learning_rate": 5.0458004734262445e-05, "loss": 0.0115, "step": 156670 }, { "epoch": 99.8597833014659, "grad_norm": 0.4941617250442505, "learning_rate": 5.0452493396667345e-05, "loss": 0.0148, "step": 156680 }, { "epoch": 99.8661567877629, "grad_norm": 0.23511789739131927, "learning_rate": 5.044698205357401e-05, "loss": 0.0098, "step": 156690 }, { "epoch": 99.87253027405991, "grad_norm": 0.18659253418445587, "learning_rate": 5.0441470705049435e-05, "loss": 0.0075, "step": 156700 }, { "epoch": 99.87890376035692, "grad_norm": 0.31714093685150146, "learning_rate": 5.0435959351160565e-05, "loss": 0.0136, "step": 156710 }, { "epoch": 99.88527724665391, "grad_norm": 0.24178344011306763, "learning_rate": 5.0430447991974385e-05, "loss": 0.0098, "step": 156720 }, { "epoch": 99.89165073295092, "grad_norm": 0.2010858803987503, "learning_rate": 5.042493662755785e-05, "loss": 0.007, "step": 156730 }, { "epoch": 99.89802421924793, "grad_norm": 0.4463373124599457, "learning_rate": 5.041942525797794e-05, "loss": 0.0116, "step": 156740 }, { "epoch": 99.90439770554494, "grad_norm": 0.24132788181304932, "learning_rate": 5.041391388330161e-05, "loss": 0.0127, "step": 156750 }, { "epoch": 99.91077119184193, "grad_norm": 0.36900588870048523, "learning_rate": 5.0408402503595845e-05, "loss": 0.0096, "step": 156760 }, { "epoch": 99.91714467813894, "grad_norm": 0.284086674451828, "learning_rate": 5.04028911189276e-05, "loss": 0.0086, "step": 156770 }, { "epoch": 99.92351816443595, "grad_norm": 0.2964104413986206, "learning_rate": 5.039737972936384e-05, "loss": 0.0156, "step": 156780 }, { "epoch": 99.92989165073296, "grad_norm": 0.23365163803100586, "learning_rate": 5.0391868334971546e-05, "loss": 0.011, "step": 156790 }, { "epoch": 99.93626513702995, "grad_norm": 0.39755842089653015, "learning_rate": 5.038635693581768e-05, "loss": 0.0199, "step": 156800 }, { "epoch": 99.94263862332696, "grad_norm": 0.2535015344619751, "learning_rate": 5.038084553196921e-05, "loss": 0.0083, "step": 156810 }, { "epoch": 99.94901210962396, "grad_norm": 0.42501240968704224, "learning_rate": 5.0375334123493126e-05, "loss": 0.0155, "step": 156820 }, { "epoch": 99.95538559592097, "grad_norm": 0.4703937768936157, "learning_rate": 5.036982271045635e-05, "loss": 0.015, "step": 156830 }, { "epoch": 99.96175908221797, "grad_norm": 0.5414004921913147, "learning_rate": 5.0364311292925906e-05, "loss": 0.014, "step": 156840 }, { "epoch": 99.96813256851497, "grad_norm": 0.28800731897354126, "learning_rate": 5.035879987096871e-05, "loss": 0.0118, "step": 156850 }, { "epoch": 99.97450605481198, "grad_norm": 0.4995836019515991, "learning_rate": 5.035328844465178e-05, "loss": 0.0166, "step": 156860 }, { "epoch": 99.98087954110899, "grad_norm": 0.4184298813343048, "learning_rate": 5.034777701404204e-05, "loss": 0.0087, "step": 156870 }, { "epoch": 99.98725302740598, "grad_norm": 0.706151008605957, "learning_rate": 5.03422655792065e-05, "loss": 0.0117, "step": 156880 }, { "epoch": 99.99362651370299, "grad_norm": 0.2277713567018509, "learning_rate": 5.03367541402121e-05, "loss": 0.0099, "step": 156890 }, { "epoch": 100.0, "grad_norm": 0.3892098069190979, "learning_rate": 5.033124269712583e-05, "loss": 0.0214, "step": 156900 }, { "epoch": 100.00637348629701, "grad_norm": 0.44853705167770386, "learning_rate": 5.032573125001463e-05, "loss": 0.0223, "step": 156910 }, { "epoch": 100.01274697259402, "grad_norm": 0.1630483865737915, "learning_rate": 5.03202197989455e-05, "loss": 0.0062, "step": 156920 }, { "epoch": 100.01912045889101, "grad_norm": 0.37573421001434326, "learning_rate": 5.031470834398538e-05, "loss": 0.0162, "step": 156930 }, { "epoch": 100.02549394518802, "grad_norm": 0.21648411452770233, "learning_rate": 5.030919688520128e-05, "loss": 0.0096, "step": 156940 }, { "epoch": 100.03186743148503, "grad_norm": 0.2348237931728363, "learning_rate": 5.0303685422660127e-05, "loss": 0.0111, "step": 156950 }, { "epoch": 100.03824091778203, "grad_norm": 0.4069375991821289, "learning_rate": 5.029817395642893e-05, "loss": 0.0105, "step": 156960 }, { "epoch": 100.04461440407903, "grad_norm": 0.2297518253326416, "learning_rate": 5.0292662486574624e-05, "loss": 0.0205, "step": 156970 }, { "epoch": 100.05098789037604, "grad_norm": 0.23388609290122986, "learning_rate": 5.02871510131642e-05, "loss": 0.0095, "step": 156980 }, { "epoch": 100.05736137667304, "grad_norm": 0.28836217522621155, "learning_rate": 5.028163953626461e-05, "loss": 0.0117, "step": 156990 }, { "epoch": 100.06373486297005, "grad_norm": 0.34003782272338867, "learning_rate": 5.027612805594284e-05, "loss": 0.0088, "step": 157000 }, { "epoch": 100.07010834926704, "grad_norm": 0.4059355854988098, "learning_rate": 5.027061657226585e-05, "loss": 0.0219, "step": 157010 }, { "epoch": 100.07648183556405, "grad_norm": 0.11157508194446564, "learning_rate": 5.0265105085300624e-05, "loss": 0.0139, "step": 157020 }, { "epoch": 100.08285532186106, "grad_norm": 0.2745707035064697, "learning_rate": 5.025959359511411e-05, "loss": 0.0087, "step": 157030 }, { "epoch": 100.08922880815807, "grad_norm": 0.5987228155136108, "learning_rate": 5.0254082101773294e-05, "loss": 0.0158, "step": 157040 }, { "epoch": 100.09560229445506, "grad_norm": 0.34071090817451477, "learning_rate": 5.0248570605345137e-05, "loss": 0.0132, "step": 157050 }, { "epoch": 100.10197578075207, "grad_norm": 0.26309025287628174, "learning_rate": 5.0243059105896616e-05, "loss": 0.0132, "step": 157060 }, { "epoch": 100.10834926704908, "grad_norm": 0.2676668167114258, "learning_rate": 5.0237547603494695e-05, "loss": 0.0102, "step": 157070 }, { "epoch": 100.11472275334609, "grad_norm": 0.2476373314857483, "learning_rate": 5.023203609820635e-05, "loss": 0.0083, "step": 157080 }, { "epoch": 100.12109623964308, "grad_norm": 0.3499343693256378, "learning_rate": 5.0226524590098544e-05, "loss": 0.014, "step": 157090 }, { "epoch": 100.12746972594009, "grad_norm": 0.33168771862983704, "learning_rate": 5.0221013079238246e-05, "loss": 0.0103, "step": 157100 }, { "epoch": 100.1338432122371, "grad_norm": 0.1924998164176941, "learning_rate": 5.0215501565692434e-05, "loss": 0.0109, "step": 157110 }, { "epoch": 100.1402166985341, "grad_norm": 0.26140138506889343, "learning_rate": 5.0209990049528064e-05, "loss": 0.0101, "step": 157120 }, { "epoch": 100.1465901848311, "grad_norm": 0.16720938682556152, "learning_rate": 5.020447853081214e-05, "loss": 0.0117, "step": 157130 }, { "epoch": 100.1529636711281, "grad_norm": 0.3973165452480316, "learning_rate": 5.0198967009611596e-05, "loss": 0.0132, "step": 157140 }, { "epoch": 100.15933715742511, "grad_norm": 0.28911349177360535, "learning_rate": 5.019345548599341e-05, "loss": 0.0088, "step": 157150 }, { "epoch": 100.16571064372212, "grad_norm": 0.7732754349708557, "learning_rate": 5.018794396002457e-05, "loss": 0.0116, "step": 157160 }, { "epoch": 100.17208413001912, "grad_norm": 0.34070703387260437, "learning_rate": 5.018243243177202e-05, "loss": 0.0133, "step": 157170 }, { "epoch": 100.17845761631612, "grad_norm": 0.3201432228088379, "learning_rate": 5.017692090130276e-05, "loss": 0.0125, "step": 157180 }, { "epoch": 100.18483110261313, "grad_norm": 0.22271603345870972, "learning_rate": 5.017140936868373e-05, "loss": 0.0126, "step": 157190 }, { "epoch": 100.19120458891014, "grad_norm": 0.45955660939216614, "learning_rate": 5.016589783398192e-05, "loss": 0.0294, "step": 157200 }, { "epoch": 100.19757807520713, "grad_norm": 0.28196701407432556, "learning_rate": 5.0160386297264285e-05, "loss": 0.0092, "step": 157210 }, { "epoch": 100.20395156150414, "grad_norm": 0.21749041974544525, "learning_rate": 5.0154874758597814e-05, "loss": 0.0113, "step": 157220 }, { "epoch": 100.21032504780115, "grad_norm": 0.2574242353439331, "learning_rate": 5.014936321804947e-05, "loss": 0.0098, "step": 157230 }, { "epoch": 100.21669853409816, "grad_norm": 0.3607490360736847, "learning_rate": 5.014385167568622e-05, "loss": 0.0126, "step": 157240 }, { "epoch": 100.22307202039515, "grad_norm": 0.13529694080352783, "learning_rate": 5.0138340131575045e-05, "loss": 0.019, "step": 157250 }, { "epoch": 100.22944550669216, "grad_norm": 0.7141773700714111, "learning_rate": 5.0132828585782886e-05, "loss": 0.0182, "step": 157260 }, { "epoch": 100.23581899298917, "grad_norm": 0.3013504147529602, "learning_rate": 5.0127317038376756e-05, "loss": 0.0136, "step": 157270 }, { "epoch": 100.24219247928617, "grad_norm": 0.207049161195755, "learning_rate": 5.012180548942358e-05, "loss": 0.0132, "step": 157280 }, { "epoch": 100.24856596558317, "grad_norm": 0.49492719769477844, "learning_rate": 5.0116293938990375e-05, "loss": 0.0105, "step": 157290 }, { "epoch": 100.25493945188018, "grad_norm": 0.2923400104045868, "learning_rate": 5.0110782387144085e-05, "loss": 0.0274, "step": 157300 }, { "epoch": 100.26131293817718, "grad_norm": 0.3016882538795471, "learning_rate": 5.010527083395167e-05, "loss": 0.0202, "step": 157310 }, { "epoch": 100.26768642447419, "grad_norm": 0.37741178274154663, "learning_rate": 5.009975927948013e-05, "loss": 0.0151, "step": 157320 }, { "epoch": 100.27405991077119, "grad_norm": 0.1782444715499878, "learning_rate": 5.0094247723796403e-05, "loss": 0.0103, "step": 157330 }, { "epoch": 100.2804333970682, "grad_norm": 0.3573989272117615, "learning_rate": 5.00887361669675e-05, "loss": 0.0157, "step": 157340 }, { "epoch": 100.2868068833652, "grad_norm": 0.4147266447544098, "learning_rate": 5.0083224609060356e-05, "loss": 0.0158, "step": 157350 }, { "epoch": 100.29318036966221, "grad_norm": 0.37958455085754395, "learning_rate": 5.0077713050141964e-05, "loss": 0.0123, "step": 157360 }, { "epoch": 100.2995538559592, "grad_norm": 0.252625972032547, "learning_rate": 5.0072201490279266e-05, "loss": 0.0198, "step": 157370 }, { "epoch": 100.30592734225621, "grad_norm": 0.2822698652744293, "learning_rate": 5.0066689929539265e-05, "loss": 0.0228, "step": 157380 }, { "epoch": 100.31230082855322, "grad_norm": 0.3699684143066406, "learning_rate": 5.0061178367988925e-05, "loss": 0.0126, "step": 157390 }, { "epoch": 100.31867431485023, "grad_norm": 0.3763628900051117, "learning_rate": 5.005566680569519e-05, "loss": 0.0155, "step": 157400 }, { "epoch": 100.32504780114722, "grad_norm": 0.3047430217266083, "learning_rate": 5.005015524272507e-05, "loss": 0.0111, "step": 157410 }, { "epoch": 100.33142128744423, "grad_norm": 0.33715197443962097, "learning_rate": 5.00446436791455e-05, "loss": 0.021, "step": 157420 }, { "epoch": 100.33779477374124, "grad_norm": 0.26300692558288574, "learning_rate": 5.003913211502348e-05, "loss": 0.0171, "step": 157430 }, { "epoch": 100.34416826003824, "grad_norm": 0.24116064608097076, "learning_rate": 5.0033620550425966e-05, "loss": 0.015, "step": 157440 }, { "epoch": 100.35054174633524, "grad_norm": 0.3457689583301544, "learning_rate": 5.002810898541993e-05, "loss": 0.0132, "step": 157450 }, { "epoch": 100.35691523263225, "grad_norm": 0.38945314288139343, "learning_rate": 5.002259742007234e-05, "loss": 0.0097, "step": 157460 }, { "epoch": 100.36328871892925, "grad_norm": 0.30984941124916077, "learning_rate": 5.001708585445016e-05, "loss": 0.0126, "step": 157470 }, { "epoch": 100.36966220522626, "grad_norm": 0.17567123472690582, "learning_rate": 5.00115742886204e-05, "loss": 0.0139, "step": 157480 }, { "epoch": 100.37603569152326, "grad_norm": 0.3123360872268677, "learning_rate": 5.000606272264997e-05, "loss": 0.0098, "step": 157490 }, { "epoch": 100.38240917782026, "grad_norm": 0.37302327156066895, "learning_rate": 5.000055115660589e-05, "loss": 0.0122, "step": 157500 }, { "epoch": 100.38878266411727, "grad_norm": 0.44685643911361694, "learning_rate": 4.9995039590555116e-05, "loss": 0.0107, "step": 157510 }, { "epoch": 100.39515615041428, "grad_norm": 0.19918909668922424, "learning_rate": 4.99895280245646e-05, "loss": 0.01, "step": 157520 }, { "epoch": 100.40152963671127, "grad_norm": 0.3311254680156708, "learning_rate": 4.998401645870133e-05, "loss": 0.0118, "step": 157530 }, { "epoch": 100.40790312300828, "grad_norm": 0.19546686112880707, "learning_rate": 4.997850489303228e-05, "loss": 0.0098, "step": 157540 }, { "epoch": 100.41427660930529, "grad_norm": 0.26555657386779785, "learning_rate": 4.9972993327624434e-05, "loss": 0.016, "step": 157550 }, { "epoch": 100.4206500956023, "grad_norm": 0.3313911259174347, "learning_rate": 4.996748176254472e-05, "loss": 0.0119, "step": 157560 }, { "epoch": 100.4270235818993, "grad_norm": 0.18579529225826263, "learning_rate": 4.996197019786014e-05, "loss": 0.0173, "step": 157570 }, { "epoch": 100.4333970681963, "grad_norm": 0.20862258970737457, "learning_rate": 4.995645863363766e-05, "loss": 0.0079, "step": 157580 }, { "epoch": 100.43977055449331, "grad_norm": 0.2933211922645569, "learning_rate": 4.995094706994425e-05, "loss": 0.0124, "step": 157590 }, { "epoch": 100.44614404079032, "grad_norm": 0.20135386288166046, "learning_rate": 4.994543550684688e-05, "loss": 0.0079, "step": 157600 }, { "epoch": 100.45251752708732, "grad_norm": 0.15696591138839722, "learning_rate": 4.9939923944412506e-05, "loss": 0.0071, "step": 157610 }, { "epoch": 100.45889101338432, "grad_norm": 0.4898512363433838, "learning_rate": 4.993441238270813e-05, "loss": 0.013, "step": 157620 }, { "epoch": 100.46526449968133, "grad_norm": 0.5493180751800537, "learning_rate": 4.9928900821800705e-05, "loss": 0.0127, "step": 157630 }, { "epoch": 100.47163798597833, "grad_norm": 0.4135250151157379, "learning_rate": 4.992338926175719e-05, "loss": 0.0146, "step": 157640 }, { "epoch": 100.47801147227534, "grad_norm": 0.1926726996898651, "learning_rate": 4.991787770264458e-05, "loss": 0.013, "step": 157650 }, { "epoch": 100.48438495857233, "grad_norm": 0.6826784014701843, "learning_rate": 4.991236614452983e-05, "loss": 0.0176, "step": 157660 }, { "epoch": 100.49075844486934, "grad_norm": 0.3423718512058258, "learning_rate": 4.990685458747991e-05, "loss": 0.0109, "step": 157670 }, { "epoch": 100.49713193116635, "grad_norm": 0.3367782235145569, "learning_rate": 4.990134303156179e-05, "loss": 0.0096, "step": 157680 }, { "epoch": 100.50350541746336, "grad_norm": 0.21888087689876556, "learning_rate": 4.989583147684246e-05, "loss": 0.0123, "step": 157690 }, { "epoch": 100.50987890376035, "grad_norm": 0.3280242383480072, "learning_rate": 4.989031992338886e-05, "loss": 0.0075, "step": 157700 }, { "epoch": 100.51625239005736, "grad_norm": 0.41066181659698486, "learning_rate": 4.9884808371267994e-05, "loss": 0.0093, "step": 157710 }, { "epoch": 100.52262587635437, "grad_norm": 0.2032085359096527, "learning_rate": 4.9879296820546814e-05, "loss": 0.0092, "step": 157720 }, { "epoch": 100.52899936265138, "grad_norm": 0.15789374709129333, "learning_rate": 4.9873785271292275e-05, "loss": 0.0095, "step": 157730 }, { "epoch": 100.53537284894837, "grad_norm": 0.3600371479988098, "learning_rate": 4.986827372357136e-05, "loss": 0.0132, "step": 157740 }, { "epoch": 100.54174633524538, "grad_norm": 0.19077013432979584, "learning_rate": 4.986276217745107e-05, "loss": 0.0124, "step": 157750 }, { "epoch": 100.54811982154239, "grad_norm": 0.21295997500419617, "learning_rate": 4.985725063299834e-05, "loss": 0.008, "step": 157760 }, { "epoch": 100.5544933078394, "grad_norm": 0.14275018870830536, "learning_rate": 4.9851739090280135e-05, "loss": 0.0086, "step": 157770 }, { "epoch": 100.56086679413639, "grad_norm": 0.3199930489063263, "learning_rate": 4.984622754936344e-05, "loss": 0.008, "step": 157780 }, { "epoch": 100.5672402804334, "grad_norm": 0.5204578042030334, "learning_rate": 4.984071601031526e-05, "loss": 0.0162, "step": 157790 }, { "epoch": 100.5736137667304, "grad_norm": 0.3460322320461273, "learning_rate": 4.9835204473202496e-05, "loss": 0.0107, "step": 157800 }, { "epoch": 100.57998725302741, "grad_norm": 0.4252386689186096, "learning_rate": 4.982969293809216e-05, "loss": 0.0108, "step": 157810 }, { "epoch": 100.5863607393244, "grad_norm": 0.1835353672504425, "learning_rate": 4.9824181405051214e-05, "loss": 0.0203, "step": 157820 }, { "epoch": 100.59273422562141, "grad_norm": 0.33413606882095337, "learning_rate": 4.981866987414665e-05, "loss": 0.0156, "step": 157830 }, { "epoch": 100.59910771191842, "grad_norm": 0.38861384987831116, "learning_rate": 4.98131583454454e-05, "loss": 0.009, "step": 157840 }, { "epoch": 100.60548119821543, "grad_norm": 0.256673663854599, "learning_rate": 4.9807646819014446e-05, "loss": 0.0108, "step": 157850 }, { "epoch": 100.61185468451242, "grad_norm": 0.33345726132392883, "learning_rate": 4.980213529492077e-05, "loss": 0.0145, "step": 157860 }, { "epoch": 100.61822817080943, "grad_norm": 0.562724769115448, "learning_rate": 4.979662377323135e-05, "loss": 0.0097, "step": 157870 }, { "epoch": 100.62460165710644, "grad_norm": 0.17177513241767883, "learning_rate": 4.979111225401313e-05, "loss": 0.0058, "step": 157880 }, { "epoch": 100.63097514340345, "grad_norm": 0.3648971915245056, "learning_rate": 4.97856007373331e-05, "loss": 0.0092, "step": 157890 }, { "epoch": 100.63734862970044, "grad_norm": 0.3772374093532562, "learning_rate": 4.978008922325824e-05, "loss": 0.0127, "step": 157900 }, { "epoch": 100.64372211599745, "grad_norm": 0.2785877287387848, "learning_rate": 4.9774577711855474e-05, "loss": 0.0138, "step": 157910 }, { "epoch": 100.65009560229446, "grad_norm": 0.32001593708992004, "learning_rate": 4.9769066203191804e-05, "loss": 0.0152, "step": 157920 }, { "epoch": 100.65646908859146, "grad_norm": 0.3175228238105774, "learning_rate": 4.9763554697334206e-05, "loss": 0.0154, "step": 157930 }, { "epoch": 100.66284257488846, "grad_norm": 0.21483607590198517, "learning_rate": 4.9758043194349654e-05, "loss": 0.0152, "step": 157940 }, { "epoch": 100.66921606118547, "grad_norm": 0.29549726843833923, "learning_rate": 4.975253169430509e-05, "loss": 0.0119, "step": 157950 }, { "epoch": 100.67558954748247, "grad_norm": 0.2563040852546692, "learning_rate": 4.974702019726749e-05, "loss": 0.01, "step": 157960 }, { "epoch": 100.68196303377948, "grad_norm": 0.26962998509407043, "learning_rate": 4.974150870330387e-05, "loss": 0.0164, "step": 157970 }, { "epoch": 100.68833652007648, "grad_norm": 0.40166813135147095, "learning_rate": 4.973599721248113e-05, "loss": 0.016, "step": 157980 }, { "epoch": 100.69471000637348, "grad_norm": 0.20739541947841644, "learning_rate": 4.973048572486628e-05, "loss": 0.0123, "step": 157990 }, { "epoch": 100.70108349267049, "grad_norm": 0.3817727565765381, "learning_rate": 4.972497424052628e-05, "loss": 0.0152, "step": 158000 }, { "epoch": 100.7074569789675, "grad_norm": 0.24749785661697388, "learning_rate": 4.971946275952812e-05, "loss": 0.0172, "step": 158010 }, { "epoch": 100.7138304652645, "grad_norm": 0.3845568895339966, "learning_rate": 4.971395128193874e-05, "loss": 0.012, "step": 158020 }, { "epoch": 100.7202039515615, "grad_norm": 0.2710892856121063, "learning_rate": 4.970843980782512e-05, "loss": 0.0111, "step": 158030 }, { "epoch": 100.72657743785851, "grad_norm": 0.3539685606956482, "learning_rate": 4.970292833725424e-05, "loss": 0.0085, "step": 158040 }, { "epoch": 100.73295092415552, "grad_norm": 0.5752547383308411, "learning_rate": 4.9697416870293055e-05, "loss": 0.027, "step": 158050 }, { "epoch": 100.73932441045251, "grad_norm": 0.23035882413387299, "learning_rate": 4.969190540700854e-05, "loss": 0.0129, "step": 158060 }, { "epoch": 100.74569789674952, "grad_norm": 0.26802676916122437, "learning_rate": 4.9686393947467656e-05, "loss": 0.0101, "step": 158070 }, { "epoch": 100.75207138304653, "grad_norm": 0.26258349418640137, "learning_rate": 4.968088249173741e-05, "loss": 0.0129, "step": 158080 }, { "epoch": 100.75844486934353, "grad_norm": 0.5556529760360718, "learning_rate": 4.9675371039884724e-05, "loss": 0.014, "step": 158090 }, { "epoch": 100.76481835564053, "grad_norm": 0.33157309889793396, "learning_rate": 4.966985959197658e-05, "loss": 0.0305, "step": 158100 }, { "epoch": 100.77119184193754, "grad_norm": 0.20796620845794678, "learning_rate": 4.9664348148079976e-05, "loss": 0.0081, "step": 158110 }, { "epoch": 100.77756532823454, "grad_norm": 0.512351393699646, "learning_rate": 4.9658836708261844e-05, "loss": 0.0146, "step": 158120 }, { "epoch": 100.78393881453155, "grad_norm": 0.2049277126789093, "learning_rate": 4.965332527258916e-05, "loss": 0.0145, "step": 158130 }, { "epoch": 100.79031230082855, "grad_norm": 0.2624233663082123, "learning_rate": 4.9647813841128895e-05, "loss": 0.0091, "step": 158140 }, { "epoch": 100.79668578712555, "grad_norm": 0.3363584578037262, "learning_rate": 4.964230241394806e-05, "loss": 0.0108, "step": 158150 }, { "epoch": 100.80305927342256, "grad_norm": 0.3817940652370453, "learning_rate": 4.963679099111356e-05, "loss": 0.0131, "step": 158160 }, { "epoch": 100.80943275971957, "grad_norm": 0.804876446723938, "learning_rate": 4.9631279572692396e-05, "loss": 0.0126, "step": 158170 }, { "epoch": 100.81580624601658, "grad_norm": 0.6905345320701599, "learning_rate": 4.9625768158751526e-05, "loss": 0.0125, "step": 158180 }, { "epoch": 100.82217973231357, "grad_norm": 0.4164600074291229, "learning_rate": 4.9620256749357956e-05, "loss": 0.0137, "step": 158190 }, { "epoch": 100.82855321861058, "grad_norm": 0.2972763776779175, "learning_rate": 4.961474534457861e-05, "loss": 0.0146, "step": 158200 }, { "epoch": 100.83492670490759, "grad_norm": 0.4881385564804077, "learning_rate": 4.960923394448045e-05, "loss": 0.0131, "step": 158210 }, { "epoch": 100.8413001912046, "grad_norm": 0.3344879448413849, "learning_rate": 4.9603722549130505e-05, "loss": 0.0111, "step": 158220 }, { "epoch": 100.84767367750159, "grad_norm": 0.3630744516849518, "learning_rate": 4.959821115859568e-05, "loss": 0.0142, "step": 158230 }, { "epoch": 100.8540471637986, "grad_norm": 0.1970565766096115, "learning_rate": 4.959269977294297e-05, "loss": 0.0103, "step": 158240 }, { "epoch": 100.8604206500956, "grad_norm": 0.21326449513435364, "learning_rate": 4.9587188392239334e-05, "loss": 0.0106, "step": 158250 }, { "epoch": 100.86679413639261, "grad_norm": 0.17816390097141266, "learning_rate": 4.958167701655178e-05, "loss": 0.01, "step": 158260 }, { "epoch": 100.8731676226896, "grad_norm": 0.44448038935661316, "learning_rate": 4.957616564594722e-05, "loss": 0.009, "step": 158270 }, { "epoch": 100.87954110898661, "grad_norm": 0.2308378964662552, "learning_rate": 4.9570654280492654e-05, "loss": 0.0132, "step": 158280 }, { "epoch": 100.88591459528362, "grad_norm": 0.24014340341091156, "learning_rate": 4.956514292025504e-05, "loss": 0.0118, "step": 158290 }, { "epoch": 100.89228808158063, "grad_norm": 0.3047611713409424, "learning_rate": 4.955963156530136e-05, "loss": 0.0164, "step": 158300 }, { "epoch": 100.89866156787762, "grad_norm": 0.3848666846752167, "learning_rate": 4.955412021569856e-05, "loss": 0.025, "step": 158310 }, { "epoch": 100.90503505417463, "grad_norm": 0.2799083888530731, "learning_rate": 4.954860887151362e-05, "loss": 0.0113, "step": 158320 }, { "epoch": 100.91140854047164, "grad_norm": 0.48967838287353516, "learning_rate": 4.954309753281352e-05, "loss": 0.0162, "step": 158330 }, { "epoch": 100.91778202676865, "grad_norm": 0.29974403977394104, "learning_rate": 4.9537586199665215e-05, "loss": 0.0088, "step": 158340 }, { "epoch": 100.92415551306564, "grad_norm": 0.14668375253677368, "learning_rate": 4.9532074872135665e-05, "loss": 0.013, "step": 158350 }, { "epoch": 100.93052899936265, "grad_norm": 0.5282368063926697, "learning_rate": 4.952656355029186e-05, "loss": 0.0103, "step": 158360 }, { "epoch": 100.93690248565966, "grad_norm": 0.3681357800960541, "learning_rate": 4.952105223420074e-05, "loss": 0.0277, "step": 158370 }, { "epoch": 100.94327597195667, "grad_norm": 0.1659453958272934, "learning_rate": 4.95155409239293e-05, "loss": 0.0099, "step": 158380 }, { "epoch": 100.94964945825366, "grad_norm": 0.6451607942581177, "learning_rate": 4.951002961954449e-05, "loss": 0.0169, "step": 158390 }, { "epoch": 100.95602294455067, "grad_norm": 0.41066300868988037, "learning_rate": 4.950451832111328e-05, "loss": 0.0084, "step": 158400 }, { "epoch": 100.96239643084768, "grad_norm": 0.20163960754871368, "learning_rate": 4.949900702870264e-05, "loss": 0.0117, "step": 158410 }, { "epoch": 100.96876991714468, "grad_norm": 0.49582764506340027, "learning_rate": 4.949349574237954e-05, "loss": 0.0131, "step": 158420 }, { "epoch": 100.97514340344168, "grad_norm": 0.23970851302146912, "learning_rate": 4.948798446221095e-05, "loss": 0.0123, "step": 158430 }, { "epoch": 100.98151688973869, "grad_norm": 0.25821059942245483, "learning_rate": 4.9482473188263825e-05, "loss": 0.0081, "step": 158440 }, { "epoch": 100.9878903760357, "grad_norm": 0.19815681874752045, "learning_rate": 4.947696192060514e-05, "loss": 0.0137, "step": 158450 }, { "epoch": 100.9942638623327, "grad_norm": 0.7296509146690369, "learning_rate": 4.947145065930186e-05, "loss": 0.0177, "step": 158460 }, { "epoch": 101.0006373486297, "grad_norm": 0.48671334981918335, "learning_rate": 4.9465939404420966e-05, "loss": 0.0124, "step": 158470 }, { "epoch": 101.0070108349267, "grad_norm": 0.34712937474250793, "learning_rate": 4.94604281560294e-05, "loss": 0.0167, "step": 158480 }, { "epoch": 101.01338432122371, "grad_norm": 0.16602161526679993, "learning_rate": 4.945491691419414e-05, "loss": 0.0104, "step": 158490 }, { "epoch": 101.01975780752072, "grad_norm": 0.3221098780632019, "learning_rate": 4.944940567898216e-05, "loss": 0.0107, "step": 158500 }, { "epoch": 101.02613129381771, "grad_norm": 0.6499330401420593, "learning_rate": 4.944389445046043e-05, "loss": 0.0171, "step": 158510 }, { "epoch": 101.03250478011472, "grad_norm": 0.1671932339668274, "learning_rate": 4.943838322869589e-05, "loss": 0.0077, "step": 158520 }, { "epoch": 101.03887826641173, "grad_norm": 0.2470676600933075, "learning_rate": 4.943287201375552e-05, "loss": 0.0161, "step": 158530 }, { "epoch": 101.04525175270874, "grad_norm": 0.15252146124839783, "learning_rate": 4.9427360805706316e-05, "loss": 0.0108, "step": 158540 }, { "epoch": 101.05162523900573, "grad_norm": 0.42263415455818176, "learning_rate": 4.9421849604615203e-05, "loss": 0.0126, "step": 158550 }, { "epoch": 101.05799872530274, "grad_norm": 0.26890745759010315, "learning_rate": 4.941633841054916e-05, "loss": 0.0102, "step": 158560 }, { "epoch": 101.06437221159975, "grad_norm": 0.30704131722450256, "learning_rate": 4.941082722357515e-05, "loss": 0.0083, "step": 158570 }, { "epoch": 101.07074569789675, "grad_norm": 0.6839407682418823, "learning_rate": 4.940531604376017e-05, "loss": 0.019, "step": 158580 }, { "epoch": 101.07711918419375, "grad_norm": 0.32337674498558044, "learning_rate": 4.939980487117115e-05, "loss": 0.0122, "step": 158590 }, { "epoch": 101.08349267049076, "grad_norm": 0.27603647112846375, "learning_rate": 4.939429370587506e-05, "loss": 0.0112, "step": 158600 }, { "epoch": 101.08986615678776, "grad_norm": 0.31121179461479187, "learning_rate": 4.938878254793889e-05, "loss": 0.0183, "step": 158610 }, { "epoch": 101.09623964308477, "grad_norm": 0.14860500395298004, "learning_rate": 4.938327139742957e-05, "loss": 0.0081, "step": 158620 }, { "epoch": 101.10261312938177, "grad_norm": 0.314699649810791, "learning_rate": 4.937776025441409e-05, "loss": 0.0151, "step": 158630 }, { "epoch": 101.10898661567877, "grad_norm": 0.14895778894424438, "learning_rate": 4.937224911895941e-05, "loss": 0.012, "step": 158640 }, { "epoch": 101.11536010197578, "grad_norm": 0.6030282378196716, "learning_rate": 4.9366737991132514e-05, "loss": 0.0133, "step": 158650 }, { "epoch": 101.12173358827279, "grad_norm": 0.23986336588859558, "learning_rate": 4.936122687100033e-05, "loss": 0.0214, "step": 158660 }, { "epoch": 101.12810707456978, "grad_norm": 0.5861750245094299, "learning_rate": 4.935571575862984e-05, "loss": 0.0113, "step": 158670 }, { "epoch": 101.13448056086679, "grad_norm": 0.42578649520874023, "learning_rate": 4.935020465408804e-05, "loss": 0.0125, "step": 158680 }, { "epoch": 101.1408540471638, "grad_norm": 0.20981234312057495, "learning_rate": 4.934469355744183e-05, "loss": 0.0088, "step": 158690 }, { "epoch": 101.1472275334608, "grad_norm": 0.2188674509525299, "learning_rate": 4.9339182468758225e-05, "loss": 0.0155, "step": 158700 }, { "epoch": 101.1536010197578, "grad_norm": 0.3605059087276459, "learning_rate": 4.933367138810418e-05, "loss": 0.0162, "step": 158710 }, { "epoch": 101.15997450605481, "grad_norm": 0.25567150115966797, "learning_rate": 4.932816031554667e-05, "loss": 0.0132, "step": 158720 }, { "epoch": 101.16634799235182, "grad_norm": 0.3585088551044464, "learning_rate": 4.932264925115263e-05, "loss": 0.0091, "step": 158730 }, { "epoch": 101.17272147864882, "grad_norm": 0.3340933918952942, "learning_rate": 4.931713819498904e-05, "loss": 0.0156, "step": 158740 }, { "epoch": 101.17909496494582, "grad_norm": 0.5086612701416016, "learning_rate": 4.931162714712286e-05, "loss": 0.0135, "step": 158750 }, { "epoch": 101.18546845124283, "grad_norm": 0.2908949851989746, "learning_rate": 4.930611610762109e-05, "loss": 0.0151, "step": 158760 }, { "epoch": 101.19184193753983, "grad_norm": 0.22469575703144073, "learning_rate": 4.930060507655063e-05, "loss": 0.0155, "step": 158770 }, { "epoch": 101.19821542383684, "grad_norm": 0.37347522377967834, "learning_rate": 4.929509405397849e-05, "loss": 0.0117, "step": 158780 }, { "epoch": 101.20458891013385, "grad_norm": 0.44560274481773376, "learning_rate": 4.9289583039971644e-05, "loss": 0.0137, "step": 158790 }, { "epoch": 101.21096239643084, "grad_norm": 0.41266337037086487, "learning_rate": 4.9284072034597015e-05, "loss": 0.0162, "step": 158800 }, { "epoch": 101.21733588272785, "grad_norm": 0.23109526932239532, "learning_rate": 4.9278561037921576e-05, "loss": 0.0186, "step": 158810 }, { "epoch": 101.22370936902486, "grad_norm": 0.280849814414978, "learning_rate": 4.927305005001231e-05, "loss": 0.0125, "step": 158820 }, { "epoch": 101.23008285532187, "grad_norm": 0.7174481153488159, "learning_rate": 4.926753907093619e-05, "loss": 0.0128, "step": 158830 }, { "epoch": 101.23645634161886, "grad_norm": 0.4483683109283447, "learning_rate": 4.9262028100760146e-05, "loss": 0.0121, "step": 158840 }, { "epoch": 101.24282982791587, "grad_norm": 0.24802528321743011, "learning_rate": 4.925651713955115e-05, "loss": 0.0147, "step": 158850 }, { "epoch": 101.24920331421288, "grad_norm": 0.6509584188461304, "learning_rate": 4.92510061873762e-05, "loss": 0.0158, "step": 158860 }, { "epoch": 101.25557680050989, "grad_norm": 0.1778278350830078, "learning_rate": 4.9245495244302206e-05, "loss": 0.0144, "step": 158870 }, { "epoch": 101.26195028680688, "grad_norm": 0.30520114302635193, "learning_rate": 4.9239984310396164e-05, "loss": 0.0121, "step": 158880 }, { "epoch": 101.26832377310389, "grad_norm": 0.4373171925544739, "learning_rate": 4.9234473385725025e-05, "loss": 0.0087, "step": 158890 }, { "epoch": 101.2746972594009, "grad_norm": 0.2759115695953369, "learning_rate": 4.922896247035578e-05, "loss": 0.0098, "step": 158900 }, { "epoch": 101.2810707456979, "grad_norm": 0.2157701849937439, "learning_rate": 4.9223451564355345e-05, "loss": 0.0181, "step": 158910 }, { "epoch": 101.2874442319949, "grad_norm": 0.17884890735149384, "learning_rate": 4.9217940667790704e-05, "loss": 0.014, "step": 158920 }, { "epoch": 101.2938177182919, "grad_norm": 0.3544321656227112, "learning_rate": 4.921242978072885e-05, "loss": 0.0127, "step": 158930 }, { "epoch": 101.30019120458891, "grad_norm": 0.35382798314094543, "learning_rate": 4.92069189032367e-05, "loss": 0.017, "step": 158940 }, { "epoch": 101.30656469088592, "grad_norm": 0.2865203022956848, "learning_rate": 4.9201408035381225e-05, "loss": 0.0174, "step": 158950 }, { "epoch": 101.31293817718291, "grad_norm": 0.4586443305015564, "learning_rate": 4.91958971772294e-05, "loss": 0.0153, "step": 158960 }, { "epoch": 101.31931166347992, "grad_norm": 0.5776457190513611, "learning_rate": 4.9190386328848206e-05, "loss": 0.0145, "step": 158970 }, { "epoch": 101.32568514977693, "grad_norm": 0.20774370431900024, "learning_rate": 4.918487549030456e-05, "loss": 0.009, "step": 158980 }, { "epoch": 101.33205863607394, "grad_norm": 0.22199389338493347, "learning_rate": 4.9179364661665445e-05, "loss": 0.0095, "step": 158990 }, { "epoch": 101.33843212237093, "grad_norm": 0.4233485460281372, "learning_rate": 4.9173853842997836e-05, "loss": 0.0117, "step": 159000 }, { "epoch": 101.34480560866794, "grad_norm": 0.24761441349983215, "learning_rate": 4.9168343034368686e-05, "loss": 0.0121, "step": 159010 }, { "epoch": 101.35117909496495, "grad_norm": 0.2304385006427765, "learning_rate": 4.916283223584494e-05, "loss": 0.0155, "step": 159020 }, { "epoch": 101.35755258126196, "grad_norm": 0.4282248318195343, "learning_rate": 4.915732144749357e-05, "loss": 0.0134, "step": 159030 }, { "epoch": 101.36392606755895, "grad_norm": 0.6278927326202393, "learning_rate": 4.9151810669381556e-05, "loss": 0.0117, "step": 159040 }, { "epoch": 101.37029955385596, "grad_norm": 0.4030623137950897, "learning_rate": 4.9146299901575836e-05, "loss": 0.0125, "step": 159050 }, { "epoch": 101.37667304015297, "grad_norm": 0.23739619553089142, "learning_rate": 4.9140789144143377e-05, "loss": 0.0086, "step": 159060 }, { "epoch": 101.38304652644997, "grad_norm": 0.2578420042991638, "learning_rate": 4.913527839715113e-05, "loss": 0.0224, "step": 159070 }, { "epoch": 101.38942001274697, "grad_norm": 0.5495481491088867, "learning_rate": 4.912976766066609e-05, "loss": 0.0145, "step": 159080 }, { "epoch": 101.39579349904398, "grad_norm": 0.24094492197036743, "learning_rate": 4.912425693475519e-05, "loss": 0.0104, "step": 159090 }, { "epoch": 101.40216698534098, "grad_norm": 0.45346978306770325, "learning_rate": 4.911874621948538e-05, "loss": 0.0138, "step": 159100 }, { "epoch": 101.40854047163799, "grad_norm": 0.35673031210899353, "learning_rate": 4.9113235514923646e-05, "loss": 0.0133, "step": 159110 }, { "epoch": 101.41491395793498, "grad_norm": 0.29190000891685486, "learning_rate": 4.910772482113693e-05, "loss": 0.0072, "step": 159120 }, { "epoch": 101.421287444232, "grad_norm": 0.37847867608070374, "learning_rate": 4.910221413819221e-05, "loss": 0.0121, "step": 159130 }, { "epoch": 101.427660930529, "grad_norm": 0.3218032419681549, "learning_rate": 4.9096703466156426e-05, "loss": 0.0082, "step": 159140 }, { "epoch": 101.43403441682601, "grad_norm": 0.2758829891681671, "learning_rate": 4.9091192805096556e-05, "loss": 0.0142, "step": 159150 }, { "epoch": 101.440407903123, "grad_norm": 0.7258878350257874, "learning_rate": 4.908568215507954e-05, "loss": 0.0167, "step": 159160 }, { "epoch": 101.44678138942001, "grad_norm": 0.23145657777786255, "learning_rate": 4.908017151617237e-05, "loss": 0.0134, "step": 159170 }, { "epoch": 101.45315487571702, "grad_norm": 0.2102966159582138, "learning_rate": 4.907466088844197e-05, "loss": 0.0076, "step": 159180 }, { "epoch": 101.45952836201403, "grad_norm": 0.23591935634613037, "learning_rate": 4.9069150271955314e-05, "loss": 0.0148, "step": 159190 }, { "epoch": 101.46590184831102, "grad_norm": 0.30324897170066833, "learning_rate": 4.906363966677936e-05, "loss": 0.0159, "step": 159200 }, { "epoch": 101.47227533460803, "grad_norm": 0.37397444248199463, "learning_rate": 4.9058129072981073e-05, "loss": 0.0127, "step": 159210 }, { "epoch": 101.47864882090504, "grad_norm": 0.19733403623104095, "learning_rate": 4.905261849062741e-05, "loss": 0.0086, "step": 159220 }, { "epoch": 101.48502230720204, "grad_norm": 0.24837501347064972, "learning_rate": 4.904710791978532e-05, "loss": 0.0119, "step": 159230 }, { "epoch": 101.49139579349904, "grad_norm": 0.5398641228675842, "learning_rate": 4.9041597360521766e-05, "loss": 0.0115, "step": 159240 }, { "epoch": 101.49776927979605, "grad_norm": 0.4353085160255432, "learning_rate": 4.9036086812903725e-05, "loss": 0.0111, "step": 159250 }, { "epoch": 101.50414276609305, "grad_norm": 0.4856848418712616, "learning_rate": 4.903057627699812e-05, "loss": 0.019, "step": 159260 }, { "epoch": 101.51051625239006, "grad_norm": 0.4468322694301605, "learning_rate": 4.902506575287194e-05, "loss": 0.0111, "step": 159270 }, { "epoch": 101.51688973868706, "grad_norm": 0.20470930635929108, "learning_rate": 4.901955524059211e-05, "loss": 0.015, "step": 159280 }, { "epoch": 101.52326322498406, "grad_norm": 0.2900790572166443, "learning_rate": 4.901404474022565e-05, "loss": 0.0103, "step": 159290 }, { "epoch": 101.52963671128107, "grad_norm": 0.2747544050216675, "learning_rate": 4.900853425183945e-05, "loss": 0.0107, "step": 159300 }, { "epoch": 101.53601019757808, "grad_norm": 0.21093440055847168, "learning_rate": 4.900302377550049e-05, "loss": 0.0196, "step": 159310 }, { "epoch": 101.54238368387507, "grad_norm": 0.4392291009426117, "learning_rate": 4.899751331127576e-05, "loss": 0.0108, "step": 159320 }, { "epoch": 101.54875717017208, "grad_norm": 0.17322170734405518, "learning_rate": 4.8992002859232166e-05, "loss": 0.012, "step": 159330 }, { "epoch": 101.55513065646909, "grad_norm": 0.23110805451869965, "learning_rate": 4.898649241943668e-05, "loss": 0.0107, "step": 159340 }, { "epoch": 101.5615041427661, "grad_norm": 0.1611788123846054, "learning_rate": 4.8980981991956285e-05, "loss": 0.0098, "step": 159350 }, { "epoch": 101.56787762906309, "grad_norm": 0.13889075815677643, "learning_rate": 4.897547157685793e-05, "loss": 0.012, "step": 159360 }, { "epoch": 101.5742511153601, "grad_norm": 0.20789049565792084, "learning_rate": 4.896996117420855e-05, "loss": 0.0092, "step": 159370 }, { "epoch": 101.5806246016571, "grad_norm": 0.47007834911346436, "learning_rate": 4.896445078407511e-05, "loss": 0.0156, "step": 159380 }, { "epoch": 101.58699808795411, "grad_norm": 0.476319819688797, "learning_rate": 4.895894040652457e-05, "loss": 0.017, "step": 159390 }, { "epoch": 101.59337157425111, "grad_norm": 0.147711381316185, "learning_rate": 4.895343004162391e-05, "loss": 0.0134, "step": 159400 }, { "epoch": 101.59974506054812, "grad_norm": 0.22649411857128143, "learning_rate": 4.894791968944004e-05, "loss": 0.0126, "step": 159410 }, { "epoch": 101.60611854684512, "grad_norm": 0.2726125717163086, "learning_rate": 4.894240935003994e-05, "loss": 0.0188, "step": 159420 }, { "epoch": 101.61249203314213, "grad_norm": 0.6533278226852417, "learning_rate": 4.893689902349058e-05, "loss": 0.0096, "step": 159430 }, { "epoch": 101.61886551943914, "grad_norm": 0.2564314305782318, "learning_rate": 4.893138870985889e-05, "loss": 0.0264, "step": 159440 }, { "epoch": 101.62523900573613, "grad_norm": 0.37579214572906494, "learning_rate": 4.8925878409211834e-05, "loss": 0.0149, "step": 159450 }, { "epoch": 101.63161249203314, "grad_norm": 0.3079954981803894, "learning_rate": 4.8920368121616366e-05, "loss": 0.0098, "step": 159460 }, { "epoch": 101.63798597833015, "grad_norm": 0.8695127964019775, "learning_rate": 4.891485784713947e-05, "loss": 0.0249, "step": 159470 }, { "epoch": 101.64435946462716, "grad_norm": 0.2032768726348877, "learning_rate": 4.8909347585848046e-05, "loss": 0.0189, "step": 159480 }, { "epoch": 101.65073295092415, "grad_norm": 0.2638571858406067, "learning_rate": 4.890383733780909e-05, "loss": 0.0101, "step": 159490 }, { "epoch": 101.65710643722116, "grad_norm": 0.26856881380081177, "learning_rate": 4.889832710308956e-05, "loss": 0.0123, "step": 159500 }, { "epoch": 101.66347992351817, "grad_norm": 0.5787960290908813, "learning_rate": 4.889281688175637e-05, "loss": 0.0109, "step": 159510 }, { "epoch": 101.66985340981518, "grad_norm": 0.18137876689434052, "learning_rate": 4.88873066738765e-05, "loss": 0.0159, "step": 159520 }, { "epoch": 101.67622689611217, "grad_norm": 0.19935017824172974, "learning_rate": 4.888179647951692e-05, "loss": 0.0095, "step": 159530 }, { "epoch": 101.68260038240918, "grad_norm": 0.35974350571632385, "learning_rate": 4.887628629874458e-05, "loss": 0.014, "step": 159540 }, { "epoch": 101.68897386870619, "grad_norm": 0.3553832173347473, "learning_rate": 4.887077613162639e-05, "loss": 0.0131, "step": 159550 }, { "epoch": 101.6953473550032, "grad_norm": 0.20887073874473572, "learning_rate": 4.886526597822935e-05, "loss": 0.0065, "step": 159560 }, { "epoch": 101.70172084130019, "grad_norm": 0.30238282680511475, "learning_rate": 4.8859755838620426e-05, "loss": 0.0149, "step": 159570 }, { "epoch": 101.7080943275972, "grad_norm": 0.42341160774230957, "learning_rate": 4.885424571286651e-05, "loss": 0.0143, "step": 159580 }, { "epoch": 101.7144678138942, "grad_norm": 0.17351680994033813, "learning_rate": 4.88487356010346e-05, "loss": 0.0091, "step": 159590 }, { "epoch": 101.72084130019121, "grad_norm": 0.27289697527885437, "learning_rate": 4.8843225503191634e-05, "loss": 0.0098, "step": 159600 }, { "epoch": 101.7272147864882, "grad_norm": 0.41122326254844666, "learning_rate": 4.883771541940459e-05, "loss": 0.0114, "step": 159610 }, { "epoch": 101.73358827278521, "grad_norm": 0.30598020553588867, "learning_rate": 4.883220534974038e-05, "loss": 0.0163, "step": 159620 }, { "epoch": 101.73996175908222, "grad_norm": 0.4266617000102997, "learning_rate": 4.8826695294265986e-05, "loss": 0.0171, "step": 159630 }, { "epoch": 101.74633524537923, "grad_norm": 0.18158330023288727, "learning_rate": 4.882118525304837e-05, "loss": 0.0074, "step": 159640 }, { "epoch": 101.75270873167622, "grad_norm": 0.33506888151168823, "learning_rate": 4.881567522615444e-05, "loss": 0.0222, "step": 159650 }, { "epoch": 101.75908221797323, "grad_norm": 0.4794153869152069, "learning_rate": 4.881016521365118e-05, "loss": 0.0168, "step": 159660 }, { "epoch": 101.76545570427024, "grad_norm": 0.19932077825069427, "learning_rate": 4.880465521560553e-05, "loss": 0.0081, "step": 159670 }, { "epoch": 101.77182919056725, "grad_norm": 0.2576580047607422, "learning_rate": 4.879914523208447e-05, "loss": 0.0103, "step": 159680 }, { "epoch": 101.77820267686424, "grad_norm": 0.2680951952934265, "learning_rate": 4.879363526315491e-05, "loss": 0.0112, "step": 159690 }, { "epoch": 101.78457616316125, "grad_norm": 0.4879978597164154, "learning_rate": 4.878812530888382e-05, "loss": 0.0208, "step": 159700 }, { "epoch": 101.79094964945826, "grad_norm": 0.2242877185344696, "learning_rate": 4.878261536933815e-05, "loss": 0.013, "step": 159710 }, { "epoch": 101.79732313575526, "grad_norm": 0.18598125874996185, "learning_rate": 4.877710544458488e-05, "loss": 0.0185, "step": 159720 }, { "epoch": 101.80369662205226, "grad_norm": 0.2805085778236389, "learning_rate": 4.877159553469091e-05, "loss": 0.0166, "step": 159730 }, { "epoch": 101.81007010834927, "grad_norm": 0.28653684258461, "learning_rate": 4.8766085639723205e-05, "loss": 0.0117, "step": 159740 }, { "epoch": 101.81644359464627, "grad_norm": 0.3317650854587555, "learning_rate": 4.8760575759748743e-05, "loss": 0.0068, "step": 159750 }, { "epoch": 101.82281708094328, "grad_norm": 0.20575156807899475, "learning_rate": 4.875506589483445e-05, "loss": 0.0067, "step": 159760 }, { "epoch": 101.82919056724027, "grad_norm": 0.5139166116714478, "learning_rate": 4.8749556045047284e-05, "loss": 0.0167, "step": 159770 }, { "epoch": 101.83556405353728, "grad_norm": 0.3231913447380066, "learning_rate": 4.8744046210454184e-05, "loss": 0.0084, "step": 159780 }, { "epoch": 101.84193753983429, "grad_norm": 0.2729186713695526, "learning_rate": 4.873853639112212e-05, "loss": 0.0092, "step": 159790 }, { "epoch": 101.8483110261313, "grad_norm": 0.44663023948669434, "learning_rate": 4.873302658711803e-05, "loss": 0.022, "step": 159800 }, { "epoch": 101.85468451242829, "grad_norm": 0.22999654710292816, "learning_rate": 4.872751679850886e-05, "loss": 0.0139, "step": 159810 }, { "epoch": 101.8610579987253, "grad_norm": 0.27288082242012024, "learning_rate": 4.872200702536156e-05, "loss": 0.01, "step": 159820 }, { "epoch": 101.86743148502231, "grad_norm": 0.30908042192459106, "learning_rate": 4.8716497267743075e-05, "loss": 0.0183, "step": 159830 }, { "epoch": 101.87380497131932, "grad_norm": 0.284462034702301, "learning_rate": 4.871098752572037e-05, "loss": 0.0056, "step": 159840 }, { "epoch": 101.88017845761631, "grad_norm": 0.2509913146495819, "learning_rate": 4.8705477799360375e-05, "loss": 0.0131, "step": 159850 }, { "epoch": 101.88655194391332, "grad_norm": 0.399878591299057, "learning_rate": 4.869996808873006e-05, "loss": 0.0102, "step": 159860 }, { "epoch": 101.89292543021033, "grad_norm": 0.255874902009964, "learning_rate": 4.869445839389634e-05, "loss": 0.0174, "step": 159870 }, { "epoch": 101.89929891650733, "grad_norm": 0.47876736521720886, "learning_rate": 4.868894871492619e-05, "loss": 0.011, "step": 159880 }, { "epoch": 101.90567240280433, "grad_norm": 0.23007117211818695, "learning_rate": 4.868343905188658e-05, "loss": 0.0074, "step": 159890 }, { "epoch": 101.91204588910134, "grad_norm": 0.23895490169525146, "learning_rate": 4.8677929404844394e-05, "loss": 0.0077, "step": 159900 }, { "epoch": 101.91841937539834, "grad_norm": 0.21515603363513947, "learning_rate": 4.867241977386662e-05, "loss": 0.0166, "step": 159910 }, { "epoch": 101.92479286169535, "grad_norm": 0.27013105154037476, "learning_rate": 4.86669101590202e-05, "loss": 0.0065, "step": 159920 }, { "epoch": 101.93116634799235, "grad_norm": 0.19571048021316528, "learning_rate": 4.86614005603721e-05, "loss": 0.0184, "step": 159930 }, { "epoch": 101.93753983428935, "grad_norm": 0.32473424077033997, "learning_rate": 4.8655890977989224e-05, "loss": 0.007, "step": 159940 }, { "epoch": 101.94391332058636, "grad_norm": 0.36306244134902954, "learning_rate": 4.865038141193854e-05, "loss": 0.012, "step": 159950 }, { "epoch": 101.95028680688337, "grad_norm": 0.49587470293045044, "learning_rate": 4.8644871862287e-05, "loss": 0.0096, "step": 159960 }, { "epoch": 101.95666029318036, "grad_norm": 0.47089633345603943, "learning_rate": 4.863936232910157e-05, "loss": 0.0151, "step": 159970 }, { "epoch": 101.96303377947737, "grad_norm": 0.40586143732070923, "learning_rate": 4.863385281244915e-05, "loss": 0.0191, "step": 159980 }, { "epoch": 101.96940726577438, "grad_norm": 0.27615687251091003, "learning_rate": 4.86283433123967e-05, "loss": 0.0106, "step": 159990 }, { "epoch": 101.97578075207139, "grad_norm": 0.1992880254983902, "learning_rate": 4.8622833829011205e-05, "loss": 0.0066, "step": 160000 }, { "epoch": 101.98215423836838, "grad_norm": 0.28830552101135254, "learning_rate": 4.8617324362359555e-05, "loss": 0.0095, "step": 160010 }, { "epoch": 101.98852772466539, "grad_norm": 0.6334031820297241, "learning_rate": 4.8611814912508715e-05, "loss": 0.0133, "step": 160020 }, { "epoch": 101.9949012109624, "grad_norm": 0.46376141905784607, "learning_rate": 4.8606305479525636e-05, "loss": 0.0163, "step": 160030 }, { "epoch": 102.0012746972594, "grad_norm": 0.3218281865119934, "learning_rate": 4.8600796063477286e-05, "loss": 0.0108, "step": 160040 }, { "epoch": 102.00764818355641, "grad_norm": 0.39789825677871704, "learning_rate": 4.859528666443057e-05, "loss": 0.0098, "step": 160050 }, { "epoch": 102.0140216698534, "grad_norm": 0.23951034247875214, "learning_rate": 4.858977728245244e-05, "loss": 0.0117, "step": 160060 }, { "epoch": 102.02039515615041, "grad_norm": 0.4151245057582855, "learning_rate": 4.8584267917609865e-05, "loss": 0.009, "step": 160070 }, { "epoch": 102.02676864244742, "grad_norm": 0.5605700016021729, "learning_rate": 4.857875856996975e-05, "loss": 0.0106, "step": 160080 }, { "epoch": 102.03314212874443, "grad_norm": 0.19044229388237, "learning_rate": 4.857324923959907e-05, "loss": 0.0157, "step": 160090 }, { "epoch": 102.03951561504142, "grad_norm": 0.21342045068740845, "learning_rate": 4.8567739926564746e-05, "loss": 0.009, "step": 160100 }, { "epoch": 102.04588910133843, "grad_norm": 0.34754765033721924, "learning_rate": 4.8562230630933764e-05, "loss": 0.0071, "step": 160110 }, { "epoch": 102.05226258763544, "grad_norm": 0.1898461878299713, "learning_rate": 4.855672135277301e-05, "loss": 0.006, "step": 160120 }, { "epoch": 102.05863607393245, "grad_norm": 0.2056616246700287, "learning_rate": 4.855121209214947e-05, "loss": 0.011, "step": 160130 }, { "epoch": 102.06500956022944, "grad_norm": 0.41425660252571106, "learning_rate": 4.8545702849130076e-05, "loss": 0.0142, "step": 160140 }, { "epoch": 102.07138304652645, "grad_norm": 0.2573557198047638, "learning_rate": 4.854019362378175e-05, "loss": 0.0119, "step": 160150 }, { "epoch": 102.07775653282346, "grad_norm": 0.3645784258842468, "learning_rate": 4.853468441617145e-05, "loss": 0.0086, "step": 160160 }, { "epoch": 102.08413001912047, "grad_norm": 0.7279101610183716, "learning_rate": 4.8529175226366114e-05, "loss": 0.013, "step": 160170 }, { "epoch": 102.09050350541746, "grad_norm": 0.1766565442085266, "learning_rate": 4.852366605443271e-05, "loss": 0.0073, "step": 160180 }, { "epoch": 102.09687699171447, "grad_norm": 0.2215404063463211, "learning_rate": 4.8518156900438136e-05, "loss": 0.0104, "step": 160190 }, { "epoch": 102.10325047801147, "grad_norm": 0.20686674118041992, "learning_rate": 4.851264776444936e-05, "loss": 0.0074, "step": 160200 }, { "epoch": 102.10962396430848, "grad_norm": 0.27118411660194397, "learning_rate": 4.8507138646533336e-05, "loss": 0.0123, "step": 160210 }, { "epoch": 102.11599745060548, "grad_norm": 0.2677743434906006, "learning_rate": 4.8501629546756965e-05, "loss": 0.0142, "step": 160220 }, { "epoch": 102.12237093690248, "grad_norm": 0.2708987593650818, "learning_rate": 4.849612046518721e-05, "loss": 0.0118, "step": 160230 }, { "epoch": 102.12874442319949, "grad_norm": 0.2191029042005539, "learning_rate": 4.8490611401891014e-05, "loss": 0.0121, "step": 160240 }, { "epoch": 102.1351179094965, "grad_norm": 0.22626811265945435, "learning_rate": 4.8485102356935333e-05, "loss": 0.0209, "step": 160250 }, { "epoch": 102.1414913957935, "grad_norm": 0.37286853790283203, "learning_rate": 4.847959333038706e-05, "loss": 0.0149, "step": 160260 }, { "epoch": 102.1478648820905, "grad_norm": 0.36376023292541504, "learning_rate": 4.8474084322313174e-05, "loss": 0.0132, "step": 160270 }, { "epoch": 102.15423836838751, "grad_norm": 0.40346643328666687, "learning_rate": 4.84685753327806e-05, "loss": 0.0155, "step": 160280 }, { "epoch": 102.16061185468452, "grad_norm": 0.24456103146076202, "learning_rate": 4.8463066361856304e-05, "loss": 0.0105, "step": 160290 }, { "epoch": 102.16698534098151, "grad_norm": 0.15047691762447357, "learning_rate": 4.845755740960719e-05, "loss": 0.0143, "step": 160300 }, { "epoch": 102.17335882727852, "grad_norm": 0.4910038411617279, "learning_rate": 4.845204847610019e-05, "loss": 0.0159, "step": 160310 }, { "epoch": 102.17973231357553, "grad_norm": 0.40927767753601074, "learning_rate": 4.84465395614023e-05, "loss": 0.008, "step": 160320 }, { "epoch": 102.18610579987254, "grad_norm": 0.41988927125930786, "learning_rate": 4.844103066558039e-05, "loss": 0.0206, "step": 160330 }, { "epoch": 102.19247928616953, "grad_norm": 0.45612001419067383, "learning_rate": 4.843552178870144e-05, "loss": 0.0096, "step": 160340 }, { "epoch": 102.19885277246654, "grad_norm": 0.7090678215026855, "learning_rate": 4.843001293083237e-05, "loss": 0.0089, "step": 160350 }, { "epoch": 102.20522625876355, "grad_norm": 0.4166829586029053, "learning_rate": 4.8424504092040146e-05, "loss": 0.01, "step": 160360 }, { "epoch": 102.21159974506055, "grad_norm": 0.34563779830932617, "learning_rate": 4.8418995272391665e-05, "loss": 0.0151, "step": 160370 }, { "epoch": 102.21797323135755, "grad_norm": 0.2805960476398468, "learning_rate": 4.8413486471953885e-05, "loss": 0.0128, "step": 160380 }, { "epoch": 102.22434671765456, "grad_norm": 0.16259920597076416, "learning_rate": 4.840797769079377e-05, "loss": 0.0075, "step": 160390 }, { "epoch": 102.23072020395156, "grad_norm": 0.18202003836631775, "learning_rate": 4.8402468928978196e-05, "loss": 0.008, "step": 160400 }, { "epoch": 102.23709369024857, "grad_norm": 0.5829226970672607, "learning_rate": 4.839696018657414e-05, "loss": 0.0095, "step": 160410 }, { "epoch": 102.24346717654556, "grad_norm": 0.273366242647171, "learning_rate": 4.839145146364853e-05, "loss": 0.0087, "step": 160420 }, { "epoch": 102.24984066284257, "grad_norm": 0.2306845635175705, "learning_rate": 4.838594276026832e-05, "loss": 0.0116, "step": 160430 }, { "epoch": 102.25621414913958, "grad_norm": 0.2722187042236328, "learning_rate": 4.838043407650041e-05, "loss": 0.0102, "step": 160440 }, { "epoch": 102.26258763543659, "grad_norm": 0.36266371607780457, "learning_rate": 4.837492541241176e-05, "loss": 0.0113, "step": 160450 }, { "epoch": 102.26896112173358, "grad_norm": 0.16868118941783905, "learning_rate": 4.8369416768069306e-05, "loss": 0.0122, "step": 160460 }, { "epoch": 102.27533460803059, "grad_norm": 0.21866188943386078, "learning_rate": 4.836390814353997e-05, "loss": 0.0088, "step": 160470 }, { "epoch": 102.2817080943276, "grad_norm": 0.34250086545944214, "learning_rate": 4.835839953889071e-05, "loss": 0.0145, "step": 160480 }, { "epoch": 102.2880815806246, "grad_norm": 0.4960741102695465, "learning_rate": 4.835289095418843e-05, "loss": 0.0164, "step": 160490 }, { "epoch": 102.2944550669216, "grad_norm": 0.25108763575553894, "learning_rate": 4.834738238950009e-05, "loss": 0.0174, "step": 160500 }, { "epoch": 102.30082855321861, "grad_norm": 0.5456917881965637, "learning_rate": 4.83418738448926e-05, "loss": 0.0136, "step": 160510 }, { "epoch": 102.30720203951562, "grad_norm": 0.2003200203180313, "learning_rate": 4.833636532043293e-05, "loss": 0.0086, "step": 160520 }, { "epoch": 102.31357552581262, "grad_norm": 0.41056203842163086, "learning_rate": 4.833085681618798e-05, "loss": 0.0113, "step": 160530 }, { "epoch": 102.31994901210962, "grad_norm": 0.4538958668708801, "learning_rate": 4.832534833222469e-05, "loss": 0.0167, "step": 160540 }, { "epoch": 102.32632249840663, "grad_norm": 0.13247710466384888, "learning_rate": 4.831983986861e-05, "loss": 0.0081, "step": 160550 }, { "epoch": 102.33269598470363, "grad_norm": 0.4717662036418915, "learning_rate": 4.831433142541085e-05, "loss": 0.01, "step": 160560 }, { "epoch": 102.33906947100064, "grad_norm": 0.20559249818325043, "learning_rate": 4.830882300269417e-05, "loss": 0.0146, "step": 160570 }, { "epoch": 102.34544295729764, "grad_norm": 0.17774464190006256, "learning_rate": 4.830331460052687e-05, "loss": 0.017, "step": 160580 }, { "epoch": 102.35181644359464, "grad_norm": 0.3168376684188843, "learning_rate": 4.8297806218975906e-05, "loss": 0.0113, "step": 160590 }, { "epoch": 102.35818992989165, "grad_norm": 0.390688955783844, "learning_rate": 4.829229785810821e-05, "loss": 0.02, "step": 160600 }, { "epoch": 102.36456341618866, "grad_norm": 0.2874755561351776, "learning_rate": 4.82867895179907e-05, "loss": 0.0146, "step": 160610 }, { "epoch": 102.37093690248565, "grad_norm": 0.5484776496887207, "learning_rate": 4.828128119869032e-05, "loss": 0.0112, "step": 160620 }, { "epoch": 102.37731038878266, "grad_norm": 0.3445250689983368, "learning_rate": 4.827577290027399e-05, "loss": 0.0083, "step": 160630 }, { "epoch": 102.38368387507967, "grad_norm": 0.594149112701416, "learning_rate": 4.827026462280866e-05, "loss": 0.0125, "step": 160640 }, { "epoch": 102.39005736137668, "grad_norm": 0.2317543625831604, "learning_rate": 4.826475636636124e-05, "loss": 0.0111, "step": 160650 }, { "epoch": 102.39643084767367, "grad_norm": 0.33107784390449524, "learning_rate": 4.825924813099866e-05, "loss": 0.0085, "step": 160660 }, { "epoch": 102.40280433397068, "grad_norm": 0.5741302967071533, "learning_rate": 4.825373991678786e-05, "loss": 0.0306, "step": 160670 }, { "epoch": 102.40917782026769, "grad_norm": 0.26618492603302, "learning_rate": 4.824823172379579e-05, "loss": 0.0077, "step": 160680 }, { "epoch": 102.4155513065647, "grad_norm": 0.3617550730705261, "learning_rate": 4.824272355208934e-05, "loss": 0.0212, "step": 160690 }, { "epoch": 102.4219247928617, "grad_norm": 0.3229774534702301, "learning_rate": 4.8237215401735456e-05, "loss": 0.0154, "step": 160700 }, { "epoch": 102.4282982791587, "grad_norm": 0.37253737449645996, "learning_rate": 4.823170727280109e-05, "loss": 0.0108, "step": 160710 }, { "epoch": 102.4346717654557, "grad_norm": 0.31006813049316406, "learning_rate": 4.822619916535313e-05, "loss": 0.0147, "step": 160720 }, { "epoch": 102.44104525175271, "grad_norm": 0.35687917470932007, "learning_rate": 4.822069107945853e-05, "loss": 0.014, "step": 160730 }, { "epoch": 102.44741873804972, "grad_norm": 0.24385246634483337, "learning_rate": 4.82151830151842e-05, "loss": 0.0047, "step": 160740 }, { "epoch": 102.45379222434671, "grad_norm": 0.1659015417098999, "learning_rate": 4.8209674972597114e-05, "loss": 0.0158, "step": 160750 }, { "epoch": 102.46016571064372, "grad_norm": 0.22399403154850006, "learning_rate": 4.820416695176414e-05, "loss": 0.0075, "step": 160760 }, { "epoch": 102.46653919694073, "grad_norm": 0.4959851801395416, "learning_rate": 4.8198658952752236e-05, "loss": 0.0099, "step": 160770 }, { "epoch": 102.47291268323774, "grad_norm": 0.2989145815372467, "learning_rate": 4.8193150975628345e-05, "loss": 0.0107, "step": 160780 }, { "epoch": 102.47928616953473, "grad_norm": 0.39706841111183167, "learning_rate": 4.818764302045936e-05, "loss": 0.0164, "step": 160790 }, { "epoch": 102.48565965583174, "grad_norm": 0.19319391250610352, "learning_rate": 4.818213508731222e-05, "loss": 0.0089, "step": 160800 }, { "epoch": 102.49203314212875, "grad_norm": 0.3453865945339203, "learning_rate": 4.817662717625386e-05, "loss": 0.0072, "step": 160810 }, { "epoch": 102.49840662842576, "grad_norm": 0.2378266453742981, "learning_rate": 4.8171119287351213e-05, "loss": 0.0198, "step": 160820 }, { "epoch": 102.50478011472275, "grad_norm": 0.2260582596063614, "learning_rate": 4.8165611420671183e-05, "loss": 0.0088, "step": 160830 }, { "epoch": 102.51115360101976, "grad_norm": 0.5661378502845764, "learning_rate": 4.81601035762807e-05, "loss": 0.0127, "step": 160840 }, { "epoch": 102.51752708731676, "grad_norm": 0.33915063738822937, "learning_rate": 4.815459575424671e-05, "loss": 0.0065, "step": 160850 }, { "epoch": 102.52390057361377, "grad_norm": 0.309775173664093, "learning_rate": 4.814908795463611e-05, "loss": 0.0166, "step": 160860 }, { "epoch": 102.53027405991077, "grad_norm": 0.32971683144569397, "learning_rate": 4.814358017751584e-05, "loss": 0.0116, "step": 160870 }, { "epoch": 102.53664754620777, "grad_norm": 0.4299449026584625, "learning_rate": 4.8138072422952827e-05, "loss": 0.0099, "step": 160880 }, { "epoch": 102.54302103250478, "grad_norm": 0.1779673993587494, "learning_rate": 4.813256469101401e-05, "loss": 0.0139, "step": 160890 }, { "epoch": 102.54939451880179, "grad_norm": 0.42075639963150024, "learning_rate": 4.812705698176627e-05, "loss": 0.008, "step": 160900 }, { "epoch": 102.55576800509878, "grad_norm": 0.24983356893062592, "learning_rate": 4.812154929527656e-05, "loss": 0.0137, "step": 160910 }, { "epoch": 102.56214149139579, "grad_norm": 0.5581455826759338, "learning_rate": 4.81160416316118e-05, "loss": 0.0188, "step": 160920 }, { "epoch": 102.5685149776928, "grad_norm": 0.4217549264431, "learning_rate": 4.811053399083893e-05, "loss": 0.0116, "step": 160930 }, { "epoch": 102.57488846398981, "grad_norm": 0.6228019595146179, "learning_rate": 4.810502637302484e-05, "loss": 0.0104, "step": 160940 }, { "epoch": 102.5812619502868, "grad_norm": 0.3318115174770355, "learning_rate": 4.809951877823646e-05, "loss": 0.0093, "step": 160950 }, { "epoch": 102.58763543658381, "grad_norm": 0.30667775869369507, "learning_rate": 4.809401120654075e-05, "loss": 0.0101, "step": 160960 }, { "epoch": 102.59400892288082, "grad_norm": 0.5851026773452759, "learning_rate": 4.808850365800458e-05, "loss": 0.0148, "step": 160970 }, { "epoch": 102.60038240917783, "grad_norm": 0.43433472514152527, "learning_rate": 4.80829961326949e-05, "loss": 0.0104, "step": 160980 }, { "epoch": 102.60675589547482, "grad_norm": 0.3505825698375702, "learning_rate": 4.807748863067862e-05, "loss": 0.0097, "step": 160990 }, { "epoch": 102.61312938177183, "grad_norm": 0.5138236880302429, "learning_rate": 4.807198115202269e-05, "loss": 0.0143, "step": 161000 }, { "epoch": 102.61950286806884, "grad_norm": 0.18321958184242249, "learning_rate": 4.806647369679399e-05, "loss": 0.0124, "step": 161010 }, { "epoch": 102.62587635436584, "grad_norm": 0.3454436659812927, "learning_rate": 4.806096626505946e-05, "loss": 0.0161, "step": 161020 }, { "epoch": 102.63224984066284, "grad_norm": 0.5520305037498474, "learning_rate": 4.805545885688604e-05, "loss": 0.0115, "step": 161030 }, { "epoch": 102.63862332695984, "grad_norm": 0.5117454528808594, "learning_rate": 4.804995147234061e-05, "loss": 0.0138, "step": 161040 }, { "epoch": 102.64499681325685, "grad_norm": 0.3347387909889221, "learning_rate": 4.804444411149012e-05, "loss": 0.0151, "step": 161050 }, { "epoch": 102.65137029955386, "grad_norm": 0.5335161089897156, "learning_rate": 4.8038936774401475e-05, "loss": 0.0093, "step": 161060 }, { "epoch": 102.65774378585085, "grad_norm": 0.2551365792751312, "learning_rate": 4.8033429461141615e-05, "loss": 0.0085, "step": 161070 }, { "epoch": 102.66411727214786, "grad_norm": 0.19747331738471985, "learning_rate": 4.802792217177743e-05, "loss": 0.0104, "step": 161080 }, { "epoch": 102.67049075844487, "grad_norm": 0.3555409014225006, "learning_rate": 4.8022414906375846e-05, "loss": 0.0118, "step": 161090 }, { "epoch": 102.67686424474188, "grad_norm": 0.29479074478149414, "learning_rate": 4.80169076650038e-05, "loss": 0.0143, "step": 161100 }, { "epoch": 102.68323773103887, "grad_norm": 0.4610753655433655, "learning_rate": 4.8011400447728205e-05, "loss": 0.0155, "step": 161110 }, { "epoch": 102.68961121733588, "grad_norm": 0.5861951112747192, "learning_rate": 4.800589325461595e-05, "loss": 0.0097, "step": 161120 }, { "epoch": 102.69598470363289, "grad_norm": 0.3669946789741516, "learning_rate": 4.800038608573398e-05, "loss": 0.0154, "step": 161130 }, { "epoch": 102.7023581899299, "grad_norm": 0.1857881098985672, "learning_rate": 4.799487894114922e-05, "loss": 0.0076, "step": 161140 }, { "epoch": 102.70873167622689, "grad_norm": 0.299490362405777, "learning_rate": 4.7989371820928583e-05, "loss": 0.01, "step": 161150 }, { "epoch": 102.7151051625239, "grad_norm": 0.32568711042404175, "learning_rate": 4.798386472513895e-05, "loss": 0.0136, "step": 161160 }, { "epoch": 102.7214786488209, "grad_norm": 0.4097404479980469, "learning_rate": 4.7978357653847274e-05, "loss": 0.0093, "step": 161170 }, { "epoch": 102.72785213511791, "grad_norm": 0.17551590502262115, "learning_rate": 4.7972850607120466e-05, "loss": 0.0076, "step": 161180 }, { "epoch": 102.73422562141491, "grad_norm": 0.26183265447616577, "learning_rate": 4.796734358502544e-05, "loss": 0.0072, "step": 161190 }, { "epoch": 102.74059910771192, "grad_norm": 0.25000274181365967, "learning_rate": 4.79618365876291e-05, "loss": 0.0134, "step": 161200 }, { "epoch": 102.74697259400892, "grad_norm": 0.4910023808479309, "learning_rate": 4.7956329614998374e-05, "loss": 0.0163, "step": 161210 }, { "epoch": 102.75334608030593, "grad_norm": 0.3583442270755768, "learning_rate": 4.7950822667200163e-05, "loss": 0.0137, "step": 161220 }, { "epoch": 102.75971956660293, "grad_norm": 0.178937628865242, "learning_rate": 4.794531574430141e-05, "loss": 0.0167, "step": 161230 }, { "epoch": 102.76609305289993, "grad_norm": 0.2818681597709656, "learning_rate": 4.793980884636899e-05, "loss": 0.009, "step": 161240 }, { "epoch": 102.77246653919694, "grad_norm": 0.27229198813438416, "learning_rate": 4.7934301973469856e-05, "loss": 0.0097, "step": 161250 }, { "epoch": 102.77884002549395, "grad_norm": 0.24641257524490356, "learning_rate": 4.792879512567089e-05, "loss": 0.0213, "step": 161260 }, { "epoch": 102.78521351179094, "grad_norm": 0.16774334013462067, "learning_rate": 4.792328830303903e-05, "loss": 0.0139, "step": 161270 }, { "epoch": 102.79158699808795, "grad_norm": 0.24208953976631165, "learning_rate": 4.7917781505641175e-05, "loss": 0.0206, "step": 161280 }, { "epoch": 102.79796048438496, "grad_norm": 0.36212119460105896, "learning_rate": 4.7912274733544235e-05, "loss": 0.0172, "step": 161290 }, { "epoch": 102.80433397068197, "grad_norm": 0.5789155960083008, "learning_rate": 4.790676798681512e-05, "loss": 0.0117, "step": 161300 }, { "epoch": 102.81070745697897, "grad_norm": 0.2004670351743698, "learning_rate": 4.790126126552076e-05, "loss": 0.014, "step": 161310 }, { "epoch": 102.81708094327597, "grad_norm": 0.48232749104499817, "learning_rate": 4.7895754569728066e-05, "loss": 0.0159, "step": 161320 }, { "epoch": 102.82345442957298, "grad_norm": 0.35024845600128174, "learning_rate": 4.7890247899503934e-05, "loss": 0.0158, "step": 161330 }, { "epoch": 102.82982791586998, "grad_norm": 0.28597691655158997, "learning_rate": 4.7884741254915275e-05, "loss": 0.0097, "step": 161340 }, { "epoch": 102.83620140216699, "grad_norm": 0.5747855305671692, "learning_rate": 4.7879234636029024e-05, "loss": 0.0093, "step": 161350 }, { "epoch": 102.84257488846399, "grad_norm": 0.18493810296058655, "learning_rate": 4.7873728042912057e-05, "loss": 0.0152, "step": 161360 }, { "epoch": 102.848948374761, "grad_norm": 0.5143281817436218, "learning_rate": 4.78682214756313e-05, "loss": 0.0178, "step": 161370 }, { "epoch": 102.855321861058, "grad_norm": 0.32889223098754883, "learning_rate": 4.7862714934253675e-05, "loss": 0.0183, "step": 161380 }, { "epoch": 102.86169534735501, "grad_norm": 0.4850592613220215, "learning_rate": 4.785720841884609e-05, "loss": 0.0111, "step": 161390 }, { "epoch": 102.868068833652, "grad_norm": 0.5089235305786133, "learning_rate": 4.785170192947543e-05, "loss": 0.0154, "step": 161400 }, { "epoch": 102.87444231994901, "grad_norm": 0.18198640644550323, "learning_rate": 4.7846195466208624e-05, "loss": 0.0103, "step": 161410 }, { "epoch": 102.88081580624602, "grad_norm": 0.6050041317939758, "learning_rate": 4.784068902911259e-05, "loss": 0.0219, "step": 161420 }, { "epoch": 102.88718929254303, "grad_norm": 0.20819583535194397, "learning_rate": 4.78351826182542e-05, "loss": 0.012, "step": 161430 }, { "epoch": 102.89356277884002, "grad_norm": 0.1928359419107437, "learning_rate": 4.7829676233700396e-05, "loss": 0.0154, "step": 161440 }, { "epoch": 102.89993626513703, "grad_norm": 0.5031909346580505, "learning_rate": 4.7824169875518066e-05, "loss": 0.0118, "step": 161450 }, { "epoch": 102.90630975143404, "grad_norm": 0.3712919056415558, "learning_rate": 4.781866354377415e-05, "loss": 0.011, "step": 161460 }, { "epoch": 102.91268323773104, "grad_norm": 0.4128875136375427, "learning_rate": 4.781315723853551e-05, "loss": 0.0109, "step": 161470 }, { "epoch": 102.91905672402804, "grad_norm": 0.1656031459569931, "learning_rate": 4.7807650959869075e-05, "loss": 0.0094, "step": 161480 }, { "epoch": 102.92543021032505, "grad_norm": 0.2182260900735855, "learning_rate": 4.780214470784175e-05, "loss": 0.0106, "step": 161490 }, { "epoch": 102.93180369662205, "grad_norm": 0.6182793974876404, "learning_rate": 4.779663848252047e-05, "loss": 0.0201, "step": 161500 }, { "epoch": 102.93817718291906, "grad_norm": 0.4642295837402344, "learning_rate": 4.779113228397208e-05, "loss": 0.01, "step": 161510 }, { "epoch": 102.94455066921606, "grad_norm": 0.7785695195198059, "learning_rate": 4.778562611226353e-05, "loss": 0.0104, "step": 161520 }, { "epoch": 102.95092415551306, "grad_norm": 0.6141444444656372, "learning_rate": 4.7780119967461726e-05, "loss": 0.0156, "step": 161530 }, { "epoch": 102.95729764181007, "grad_norm": 0.5237746238708496, "learning_rate": 4.7774613849633546e-05, "loss": 0.0135, "step": 161540 }, { "epoch": 102.96367112810708, "grad_norm": 0.3427233099937439, "learning_rate": 4.7769107758845896e-05, "loss": 0.0147, "step": 161550 }, { "epoch": 102.97004461440407, "grad_norm": 0.35747992992401123, "learning_rate": 4.776360169516571e-05, "loss": 0.0114, "step": 161560 }, { "epoch": 102.97641810070108, "grad_norm": 0.4637719988822937, "learning_rate": 4.7758095658659886e-05, "loss": 0.0106, "step": 161570 }, { "epoch": 102.98279158699809, "grad_norm": 0.6037371158599854, "learning_rate": 4.775258964939529e-05, "loss": 0.0199, "step": 161580 }, { "epoch": 102.9891650732951, "grad_norm": 0.22933916747570038, "learning_rate": 4.7747083667438865e-05, "loss": 0.0084, "step": 161590 }, { "epoch": 102.99553855959209, "grad_norm": 0.40032637119293213, "learning_rate": 4.774157771285751e-05, "loss": 0.0151, "step": 161600 }, { "epoch": 103.0019120458891, "grad_norm": 0.20968526601791382, "learning_rate": 4.773607178571811e-05, "loss": 0.0145, "step": 161610 }, { "epoch": 103.00828553218611, "grad_norm": 0.21258476376533508, "learning_rate": 4.773056588608756e-05, "loss": 0.0115, "step": 161620 }, { "epoch": 103.01465901848312, "grad_norm": 0.1612934172153473, "learning_rate": 4.772506001403279e-05, "loss": 0.0103, "step": 161630 }, { "epoch": 103.02103250478011, "grad_norm": 0.4011053144931793, "learning_rate": 4.77195541696207e-05, "loss": 0.01, "step": 161640 }, { "epoch": 103.02740599107712, "grad_norm": 0.1648450493812561, "learning_rate": 4.7714048352918164e-05, "loss": 0.0123, "step": 161650 }, { "epoch": 103.03377947737413, "grad_norm": 0.26446348428726196, "learning_rate": 4.77085425639921e-05, "loss": 0.0122, "step": 161660 }, { "epoch": 103.04015296367113, "grad_norm": 0.29119884967803955, "learning_rate": 4.770303680290942e-05, "loss": 0.0147, "step": 161670 }, { "epoch": 103.04652644996813, "grad_norm": 0.3828341066837311, "learning_rate": 4.7697531069737e-05, "loss": 0.0148, "step": 161680 }, { "epoch": 103.05289993626513, "grad_norm": 0.2496977597475052, "learning_rate": 4.769202536454175e-05, "loss": 0.0136, "step": 161690 }, { "epoch": 103.05927342256214, "grad_norm": 0.38429462909698486, "learning_rate": 4.768651968739057e-05, "loss": 0.0145, "step": 161700 }, { "epoch": 103.06564690885915, "grad_norm": 0.5108785033226013, "learning_rate": 4.768101403835038e-05, "loss": 0.0174, "step": 161710 }, { "epoch": 103.07202039515614, "grad_norm": 0.3186868727207184, "learning_rate": 4.7675508417488035e-05, "loss": 0.0186, "step": 161720 }, { "epoch": 103.07839388145315, "grad_norm": 0.19903123378753662, "learning_rate": 4.767000282487046e-05, "loss": 0.0135, "step": 161730 }, { "epoch": 103.08476736775016, "grad_norm": 0.4094665050506592, "learning_rate": 4.7664497260564576e-05, "loss": 0.0208, "step": 161740 }, { "epoch": 103.09114085404717, "grad_norm": 0.30501917004585266, "learning_rate": 4.765899172463723e-05, "loss": 0.0107, "step": 161750 }, { "epoch": 103.09751434034416, "grad_norm": 0.8201461434364319, "learning_rate": 4.765348621715534e-05, "loss": 0.0112, "step": 161760 }, { "epoch": 103.10388782664117, "grad_norm": 0.245188906788826, "learning_rate": 4.764798073818582e-05, "loss": 0.0114, "step": 161770 }, { "epoch": 103.11026131293818, "grad_norm": 0.17055878043174744, "learning_rate": 4.7642475287795566e-05, "loss": 0.0189, "step": 161780 }, { "epoch": 103.11663479923519, "grad_norm": 0.5497379899024963, "learning_rate": 4.7636969866051446e-05, "loss": 0.015, "step": 161790 }, { "epoch": 103.12300828553218, "grad_norm": 0.3761061728000641, "learning_rate": 4.7631464473020373e-05, "loss": 0.0178, "step": 161800 }, { "epoch": 103.12938177182919, "grad_norm": 0.2553998827934265, "learning_rate": 4.7625959108769245e-05, "loss": 0.0067, "step": 161810 }, { "epoch": 103.1357552581262, "grad_norm": 0.461221843957901, "learning_rate": 4.762045377336497e-05, "loss": 0.0176, "step": 161820 }, { "epoch": 103.1421287444232, "grad_norm": 0.4923268258571625, "learning_rate": 4.761494846687442e-05, "loss": 0.0122, "step": 161830 }, { "epoch": 103.1485022307202, "grad_norm": 0.19892701506614685, "learning_rate": 4.7609443189364484e-05, "loss": 0.0115, "step": 161840 }, { "epoch": 103.1548757170172, "grad_norm": 0.27158528566360474, "learning_rate": 4.760393794090208e-05, "loss": 0.0181, "step": 161850 }, { "epoch": 103.16124920331421, "grad_norm": 0.2994951009750366, "learning_rate": 4.7598432721554106e-05, "loss": 0.014, "step": 161860 }, { "epoch": 103.16762268961122, "grad_norm": 0.29259705543518066, "learning_rate": 4.7592927531387424e-05, "loss": 0.0129, "step": 161870 }, { "epoch": 103.17399617590821, "grad_norm": 0.46032026410102844, "learning_rate": 4.758742237046894e-05, "loss": 0.0125, "step": 161880 }, { "epoch": 103.18036966220522, "grad_norm": 0.4078543186187744, "learning_rate": 4.758191723886557e-05, "loss": 0.0123, "step": 161890 }, { "epoch": 103.18674314850223, "grad_norm": 0.3366329073905945, "learning_rate": 4.757641213664418e-05, "loss": 0.0121, "step": 161900 }, { "epoch": 103.19311663479924, "grad_norm": 0.7535768747329712, "learning_rate": 4.7570907063871666e-05, "loss": 0.0092, "step": 161910 }, { "epoch": 103.19949012109623, "grad_norm": 0.1490851491689682, "learning_rate": 4.7565402020614935e-05, "loss": 0.0121, "step": 161920 }, { "epoch": 103.20586360739324, "grad_norm": 0.26145508885383606, "learning_rate": 4.755989700694086e-05, "loss": 0.0159, "step": 161930 }, { "epoch": 103.21223709369025, "grad_norm": 0.385468453168869, "learning_rate": 4.755439202291635e-05, "loss": 0.019, "step": 161940 }, { "epoch": 103.21861057998726, "grad_norm": 0.28733786940574646, "learning_rate": 4.754888706860827e-05, "loss": 0.0126, "step": 161950 }, { "epoch": 103.22498406628426, "grad_norm": 0.19460803270339966, "learning_rate": 4.7543382144083544e-05, "loss": 0.0186, "step": 161960 }, { "epoch": 103.23135755258126, "grad_norm": 0.2615986764431, "learning_rate": 4.753787724940903e-05, "loss": 0.0089, "step": 161970 }, { "epoch": 103.23773103887827, "grad_norm": 0.3115973472595215, "learning_rate": 4.753237238465164e-05, "loss": 0.0105, "step": 161980 }, { "epoch": 103.24410452517527, "grad_norm": 0.17191392183303833, "learning_rate": 4.752686754987826e-05, "loss": 0.005, "step": 161990 }, { "epoch": 103.25047801147228, "grad_norm": 0.28229019045829773, "learning_rate": 4.752136274515576e-05, "loss": 0.0151, "step": 162000 }, { "epoch": 103.25685149776928, "grad_norm": 0.43926888704299927, "learning_rate": 4.7515857970551045e-05, "loss": 0.0113, "step": 162010 }, { "epoch": 103.26322498406628, "grad_norm": 0.323093056678772, "learning_rate": 4.7510353226131014e-05, "loss": 0.0125, "step": 162020 }, { "epoch": 103.26959847036329, "grad_norm": 0.300505667924881, "learning_rate": 4.750484851196254e-05, "loss": 0.0136, "step": 162030 }, { "epoch": 103.2759719566603, "grad_norm": 0.7028771042823792, "learning_rate": 4.74993438281125e-05, "loss": 0.0154, "step": 162040 }, { "epoch": 103.2823454429573, "grad_norm": 0.2168089598417282, "learning_rate": 4.7493839174647795e-05, "loss": 0.0112, "step": 162050 }, { "epoch": 103.2887189292543, "grad_norm": 0.17862096428871155, "learning_rate": 4.748833455163532e-05, "loss": 0.0114, "step": 162060 }, { "epoch": 103.29509241555131, "grad_norm": 0.24041149020195007, "learning_rate": 4.748282995914196e-05, "loss": 0.0115, "step": 162070 }, { "epoch": 103.30146590184832, "grad_norm": 0.3039979338645935, "learning_rate": 4.7477325397234575e-05, "loss": 0.0113, "step": 162080 }, { "epoch": 103.30783938814531, "grad_norm": 0.3010715842247009, "learning_rate": 4.747182086598007e-05, "loss": 0.0148, "step": 162090 }, { "epoch": 103.31421287444232, "grad_norm": 0.3142956495285034, "learning_rate": 4.746631636544534e-05, "loss": 0.0123, "step": 162100 }, { "epoch": 103.32058636073933, "grad_norm": 0.24023082852363586, "learning_rate": 4.746081189569724e-05, "loss": 0.015, "step": 162110 }, { "epoch": 103.32695984703633, "grad_norm": 0.19582749903202057, "learning_rate": 4.7455307456802685e-05, "loss": 0.0119, "step": 162120 }, { "epoch": 103.33333333333333, "grad_norm": 0.2020477056503296, "learning_rate": 4.744980304882854e-05, "loss": 0.0089, "step": 162130 }, { "epoch": 103.33970681963034, "grad_norm": 0.7932565212249756, "learning_rate": 4.7444298671841716e-05, "loss": 0.014, "step": 162140 }, { "epoch": 103.34608030592734, "grad_norm": 0.4346770644187927, "learning_rate": 4.743879432590905e-05, "loss": 0.0166, "step": 162150 }, { "epoch": 103.35245379222435, "grad_norm": 0.4493076503276825, "learning_rate": 4.743329001109746e-05, "loss": 0.0118, "step": 162160 }, { "epoch": 103.35882727852135, "grad_norm": 0.3177502155303955, "learning_rate": 4.742778572747384e-05, "loss": 0.0102, "step": 162170 }, { "epoch": 103.36520076481835, "grad_norm": 0.38104623556137085, "learning_rate": 4.7422281475105025e-05, "loss": 0.0151, "step": 162180 }, { "epoch": 103.37157425111536, "grad_norm": 0.38845157623291016, "learning_rate": 4.7416777254057924e-05, "loss": 0.0113, "step": 162190 }, { "epoch": 103.37794773741237, "grad_norm": 0.557138979434967, "learning_rate": 4.7411273064399424e-05, "loss": 0.0095, "step": 162200 }, { "epoch": 103.38432122370936, "grad_norm": 0.18401755392551422, "learning_rate": 4.7405768906196415e-05, "loss": 0.0143, "step": 162210 }, { "epoch": 103.39069471000637, "grad_norm": 0.270990788936615, "learning_rate": 4.740026477951575e-05, "loss": 0.0146, "step": 162220 }, { "epoch": 103.39706819630338, "grad_norm": 0.3797665536403656, "learning_rate": 4.739476068442432e-05, "loss": 0.008, "step": 162230 }, { "epoch": 103.40344168260039, "grad_norm": 0.33494284749031067, "learning_rate": 4.738925662098902e-05, "loss": 0.0086, "step": 162240 }, { "epoch": 103.40981516889738, "grad_norm": 0.20402690768241882, "learning_rate": 4.738375258927671e-05, "loss": 0.0138, "step": 162250 }, { "epoch": 103.41618865519439, "grad_norm": 0.4336712658405304, "learning_rate": 4.737824858935427e-05, "loss": 0.0121, "step": 162260 }, { "epoch": 103.4225621414914, "grad_norm": 0.3058012127876282, "learning_rate": 4.737274462128858e-05, "loss": 0.0122, "step": 162270 }, { "epoch": 103.4289356277884, "grad_norm": 0.3188810348510742, "learning_rate": 4.7367240685146546e-05, "loss": 0.0083, "step": 162280 }, { "epoch": 103.4353091140854, "grad_norm": 0.21657811105251312, "learning_rate": 4.736173678099501e-05, "loss": 0.0128, "step": 162290 }, { "epoch": 103.4416826003824, "grad_norm": 0.4373418986797333, "learning_rate": 4.735623290890086e-05, "loss": 0.0091, "step": 162300 }, { "epoch": 103.44805608667941, "grad_norm": 0.192437544465065, "learning_rate": 4.735072906893099e-05, "loss": 0.0075, "step": 162310 }, { "epoch": 103.45442957297642, "grad_norm": 0.26501983404159546, "learning_rate": 4.734522526115225e-05, "loss": 0.0126, "step": 162320 }, { "epoch": 103.46080305927342, "grad_norm": 0.283331036567688, "learning_rate": 4.733972148563153e-05, "loss": 0.0149, "step": 162330 }, { "epoch": 103.46717654557042, "grad_norm": 0.36072292923927307, "learning_rate": 4.73342177424357e-05, "loss": 0.0117, "step": 162340 }, { "epoch": 103.47355003186743, "grad_norm": 0.39541301131248474, "learning_rate": 4.732871403163166e-05, "loss": 0.0162, "step": 162350 }, { "epoch": 103.47992351816444, "grad_norm": 0.4319038391113281, "learning_rate": 4.732321035328625e-05, "loss": 0.0116, "step": 162360 }, { "epoch": 103.48629700446143, "grad_norm": 0.5304919481277466, "learning_rate": 4.7317706707466365e-05, "loss": 0.0173, "step": 162370 }, { "epoch": 103.49267049075844, "grad_norm": 0.4010530412197113, "learning_rate": 4.731220309423887e-05, "loss": 0.0101, "step": 162380 }, { "epoch": 103.49904397705545, "grad_norm": 0.29238998889923096, "learning_rate": 4.730669951367067e-05, "loss": 0.0164, "step": 162390 }, { "epoch": 103.50541746335246, "grad_norm": 0.38149189949035645, "learning_rate": 4.730119596582859e-05, "loss": 0.0195, "step": 162400 }, { "epoch": 103.51179094964945, "grad_norm": 0.44886744022369385, "learning_rate": 4.729569245077953e-05, "loss": 0.0305, "step": 162410 }, { "epoch": 103.51816443594646, "grad_norm": 0.35733655095100403, "learning_rate": 4.7290188968590385e-05, "loss": 0.0126, "step": 162420 }, { "epoch": 103.52453792224347, "grad_norm": 0.2423417568206787, "learning_rate": 4.728468551932798e-05, "loss": 0.01, "step": 162430 }, { "epoch": 103.53091140854048, "grad_norm": 0.17571163177490234, "learning_rate": 4.72791821030592e-05, "loss": 0.0078, "step": 162440 }, { "epoch": 103.53728489483747, "grad_norm": 0.19115567207336426, "learning_rate": 4.727367871985093e-05, "loss": 0.0124, "step": 162450 }, { "epoch": 103.54365838113448, "grad_norm": 0.2977018654346466, "learning_rate": 4.7268175369770066e-05, "loss": 0.0098, "step": 162460 }, { "epoch": 103.55003186743149, "grad_norm": 0.2215782254934311, "learning_rate": 4.7262672052883425e-05, "loss": 0.0133, "step": 162470 }, { "epoch": 103.5564053537285, "grad_norm": 0.37608829140663147, "learning_rate": 4.72571687692579e-05, "loss": 0.017, "step": 162480 }, { "epoch": 103.56277884002549, "grad_norm": 0.549576461315155, "learning_rate": 4.725166551896039e-05, "loss": 0.0159, "step": 162490 }, { "epoch": 103.5691523263225, "grad_norm": 0.3844356834888458, "learning_rate": 4.724616230205772e-05, "loss": 0.0153, "step": 162500 }, { "epoch": 103.5755258126195, "grad_norm": 0.31201502680778503, "learning_rate": 4.724065911861677e-05, "loss": 0.014, "step": 162510 }, { "epoch": 103.58189929891651, "grad_norm": 0.204884335398674, "learning_rate": 4.723515596870442e-05, "loss": 0.0106, "step": 162520 }, { "epoch": 103.5882727852135, "grad_norm": 0.6026825308799744, "learning_rate": 4.722965285238756e-05, "loss": 0.01, "step": 162530 }, { "epoch": 103.59464627151051, "grad_norm": 0.31415772438049316, "learning_rate": 4.7224149769733007e-05, "loss": 0.0143, "step": 162540 }, { "epoch": 103.60101975780752, "grad_norm": 0.3759268522262573, "learning_rate": 4.721864672080766e-05, "loss": 0.01, "step": 162550 }, { "epoch": 103.60739324410453, "grad_norm": 0.10192456096410751, "learning_rate": 4.721314370567839e-05, "loss": 0.0104, "step": 162560 }, { "epoch": 103.61376673040154, "grad_norm": 0.4336622953414917, "learning_rate": 4.720764072441205e-05, "loss": 0.0087, "step": 162570 }, { "epoch": 103.62014021669853, "grad_norm": 0.5918996334075928, "learning_rate": 4.720213777707551e-05, "loss": 0.0098, "step": 162580 }, { "epoch": 103.62651370299554, "grad_norm": 0.25846999883651733, "learning_rate": 4.7196634863735625e-05, "loss": 0.0091, "step": 162590 }, { "epoch": 103.63288718929255, "grad_norm": 0.6275224089622498, "learning_rate": 4.7191131984459294e-05, "loss": 0.0123, "step": 162600 }, { "epoch": 103.63926067558955, "grad_norm": 0.19211213290691376, "learning_rate": 4.718562913931336e-05, "loss": 0.011, "step": 162610 }, { "epoch": 103.64563416188655, "grad_norm": 0.16391561925411224, "learning_rate": 4.718012632836468e-05, "loss": 0.0161, "step": 162620 }, { "epoch": 103.65200764818356, "grad_norm": 0.24184256792068481, "learning_rate": 4.717462355168014e-05, "loss": 0.0156, "step": 162630 }, { "epoch": 103.65838113448056, "grad_norm": 0.2622442841529846, "learning_rate": 4.7169120809326574e-05, "loss": 0.0091, "step": 162640 }, { "epoch": 103.66475462077757, "grad_norm": 0.1922922432422638, "learning_rate": 4.7163618101370876e-05, "loss": 0.0094, "step": 162650 }, { "epoch": 103.67112810707457, "grad_norm": 0.18954165279865265, "learning_rate": 4.715811542787989e-05, "loss": 0.0087, "step": 162660 }, { "epoch": 103.67750159337157, "grad_norm": 0.2838003635406494, "learning_rate": 4.715261278892049e-05, "loss": 0.0164, "step": 162670 }, { "epoch": 103.68387507966858, "grad_norm": 0.6376625299453735, "learning_rate": 4.714711018455952e-05, "loss": 0.0103, "step": 162680 }, { "epoch": 103.69024856596559, "grad_norm": 0.2629677355289459, "learning_rate": 4.714160761486387e-05, "loss": 0.0121, "step": 162690 }, { "epoch": 103.69662205226258, "grad_norm": 0.36456209421157837, "learning_rate": 4.713610507990038e-05, "loss": 0.0172, "step": 162700 }, { "epoch": 103.70299553855959, "grad_norm": 0.22207891941070557, "learning_rate": 4.7130602579735923e-05, "loss": 0.0127, "step": 162710 }, { "epoch": 103.7093690248566, "grad_norm": 0.24512243270874023, "learning_rate": 4.7125100114437337e-05, "loss": 0.0107, "step": 162720 }, { "epoch": 103.7157425111536, "grad_norm": 0.34807249903678894, "learning_rate": 4.711959768407152e-05, "loss": 0.0145, "step": 162730 }, { "epoch": 103.7221159974506, "grad_norm": 0.6480016112327576, "learning_rate": 4.711409528870531e-05, "loss": 0.017, "step": 162740 }, { "epoch": 103.72848948374761, "grad_norm": 0.2528587877750397, "learning_rate": 4.710859292840555e-05, "loss": 0.0263, "step": 162750 }, { "epoch": 103.73486297004462, "grad_norm": 0.28874415159225464, "learning_rate": 4.710309060323912e-05, "loss": 0.013, "step": 162760 }, { "epoch": 103.74123645634162, "grad_norm": 0.5083698034286499, "learning_rate": 4.709758831327288e-05, "loss": 0.0145, "step": 162770 }, { "epoch": 103.74760994263862, "grad_norm": 0.21517349779605865, "learning_rate": 4.709208605857369e-05, "loss": 0.0127, "step": 162780 }, { "epoch": 103.75398342893563, "grad_norm": 0.334418386220932, "learning_rate": 4.708658383920838e-05, "loss": 0.0096, "step": 162790 }, { "epoch": 103.76035691523263, "grad_norm": 0.6698747873306274, "learning_rate": 4.708108165524383e-05, "loss": 0.0167, "step": 162800 }, { "epoch": 103.76673040152964, "grad_norm": 0.2113809436559677, "learning_rate": 4.7075579506746917e-05, "loss": 0.0151, "step": 162810 }, { "epoch": 103.77310388782664, "grad_norm": 0.24581065773963928, "learning_rate": 4.707007739378446e-05, "loss": 0.0112, "step": 162820 }, { "epoch": 103.77947737412364, "grad_norm": 0.21179941296577454, "learning_rate": 4.706457531642332e-05, "loss": 0.009, "step": 162830 }, { "epoch": 103.78585086042065, "grad_norm": 0.46842673420906067, "learning_rate": 4.705907327473036e-05, "loss": 0.0165, "step": 162840 }, { "epoch": 103.79222434671766, "grad_norm": 0.21695515513420105, "learning_rate": 4.705357126877246e-05, "loss": 0.009, "step": 162850 }, { "epoch": 103.79859783301465, "grad_norm": 0.4122987687587738, "learning_rate": 4.7048069298616434e-05, "loss": 0.0081, "step": 162860 }, { "epoch": 103.80497131931166, "grad_norm": 0.7367067337036133, "learning_rate": 4.704256736432915e-05, "loss": 0.0127, "step": 162870 }, { "epoch": 103.81134480560867, "grad_norm": 0.16973592340946198, "learning_rate": 4.7037065465977485e-05, "loss": 0.007, "step": 162880 }, { "epoch": 103.81771829190568, "grad_norm": 0.7852079272270203, "learning_rate": 4.7031563603628256e-05, "loss": 0.0271, "step": 162890 }, { "epoch": 103.82409177820267, "grad_norm": 0.21821367740631104, "learning_rate": 4.7026061777348326e-05, "loss": 0.0107, "step": 162900 }, { "epoch": 103.83046526449968, "grad_norm": 0.7841814756393433, "learning_rate": 4.702055998720456e-05, "loss": 0.0256, "step": 162910 }, { "epoch": 103.83683875079669, "grad_norm": 0.4153786599636078, "learning_rate": 4.7015058233263815e-05, "loss": 0.0106, "step": 162920 }, { "epoch": 103.8432122370937, "grad_norm": 0.8632116913795471, "learning_rate": 4.700955651559291e-05, "loss": 0.0114, "step": 162930 }, { "epoch": 103.84958572339069, "grad_norm": 0.27162495255470276, "learning_rate": 4.700405483425872e-05, "loss": 0.0123, "step": 162940 }, { "epoch": 103.8559592096877, "grad_norm": 0.3202527165412903, "learning_rate": 4.699855318932811e-05, "loss": 0.0126, "step": 162950 }, { "epoch": 103.8623326959847, "grad_norm": 0.2499465048313141, "learning_rate": 4.69930515808679e-05, "loss": 0.025, "step": 162960 }, { "epoch": 103.86870618228171, "grad_norm": 0.3889995217323303, "learning_rate": 4.698755000894495e-05, "loss": 0.0158, "step": 162970 }, { "epoch": 103.8750796685787, "grad_norm": 0.36741530895233154, "learning_rate": 4.698204847362611e-05, "loss": 0.0093, "step": 162980 }, { "epoch": 103.88145315487571, "grad_norm": 0.2680562436580658, "learning_rate": 4.6976546974978256e-05, "loss": 0.0126, "step": 162990 }, { "epoch": 103.88782664117272, "grad_norm": 0.6116255521774292, "learning_rate": 4.6971045513068176e-05, "loss": 0.0124, "step": 163000 }, { "epoch": 103.89420012746973, "grad_norm": 0.5712509155273438, "learning_rate": 4.696554408796276e-05, "loss": 0.0165, "step": 163010 }, { "epoch": 103.90057361376672, "grad_norm": 0.47062456607818604, "learning_rate": 4.6960042699728855e-05, "loss": 0.0141, "step": 163020 }, { "epoch": 103.90694710006373, "grad_norm": 0.22751207649707794, "learning_rate": 4.6954541348433316e-05, "loss": 0.0168, "step": 163030 }, { "epoch": 103.91332058636074, "grad_norm": 0.27102142572402954, "learning_rate": 4.694904003414296e-05, "loss": 0.0167, "step": 163040 }, { "epoch": 103.91969407265775, "grad_norm": 0.2611536681652069, "learning_rate": 4.6943538756924647e-05, "loss": 0.0078, "step": 163050 }, { "epoch": 103.92606755895474, "grad_norm": 0.19170933961868286, "learning_rate": 4.693803751684524e-05, "loss": 0.0151, "step": 163060 }, { "epoch": 103.93244104525175, "grad_norm": 0.2793982923030853, "learning_rate": 4.693253631397155e-05, "loss": 0.0217, "step": 163070 }, { "epoch": 103.93881453154876, "grad_norm": 0.29252564907073975, "learning_rate": 4.6927035148370434e-05, "loss": 0.0216, "step": 163080 }, { "epoch": 103.94518801784577, "grad_norm": 0.18855851888656616, "learning_rate": 4.6921534020108746e-05, "loss": 0.0092, "step": 163090 }, { "epoch": 103.95156150414276, "grad_norm": 0.18608902394771576, "learning_rate": 4.691603292925335e-05, "loss": 0.0076, "step": 163100 }, { "epoch": 103.95793499043977, "grad_norm": 0.3767905533313751, "learning_rate": 4.691053187587105e-05, "loss": 0.0091, "step": 163110 }, { "epoch": 103.96430847673678, "grad_norm": 0.32889220118522644, "learning_rate": 4.690503086002869e-05, "loss": 0.0067, "step": 163120 }, { "epoch": 103.97068196303378, "grad_norm": 0.15630222856998444, "learning_rate": 4.689952988179316e-05, "loss": 0.0144, "step": 163130 }, { "epoch": 103.97705544933078, "grad_norm": 0.23755958676338196, "learning_rate": 4.6894028941231236e-05, "loss": 0.0167, "step": 163140 }, { "epoch": 103.98342893562778, "grad_norm": 0.31133267283439636, "learning_rate": 4.688852803840981e-05, "loss": 0.01, "step": 163150 }, { "epoch": 103.9898024219248, "grad_norm": 0.43789103627204895, "learning_rate": 4.688302717339569e-05, "loss": 0.0123, "step": 163160 }, { "epoch": 103.9961759082218, "grad_norm": 0.24969522655010223, "learning_rate": 4.6877526346255763e-05, "loss": 0.02, "step": 163170 }, { "epoch": 104.00254939451881, "grad_norm": 0.2224799394607544, "learning_rate": 4.687202555705682e-05, "loss": 0.0145, "step": 163180 }, { "epoch": 104.0089228808158, "grad_norm": 0.4773961007595062, "learning_rate": 4.686652480586571e-05, "loss": 0.0152, "step": 163190 }, { "epoch": 104.01529636711281, "grad_norm": 0.4377918243408203, "learning_rate": 4.686102409274931e-05, "loss": 0.0118, "step": 163200 }, { "epoch": 104.02166985340982, "grad_norm": 0.18768438696861267, "learning_rate": 4.685552341777441e-05, "loss": 0.0113, "step": 163210 }, { "epoch": 104.02804333970683, "grad_norm": 0.21357093751430511, "learning_rate": 4.685002278100787e-05, "loss": 0.0085, "step": 163220 }, { "epoch": 104.03441682600382, "grad_norm": 0.25045812129974365, "learning_rate": 4.684452218251652e-05, "loss": 0.0066, "step": 163230 }, { "epoch": 104.04079031230083, "grad_norm": 0.31524619460105896, "learning_rate": 4.683902162236723e-05, "loss": 0.02, "step": 163240 }, { "epoch": 104.04716379859784, "grad_norm": 0.4239996671676636, "learning_rate": 4.683352110062679e-05, "loss": 0.0155, "step": 163250 }, { "epoch": 104.05353728489484, "grad_norm": 0.3828640580177307, "learning_rate": 4.682802061736206e-05, "loss": 0.0066, "step": 163260 }, { "epoch": 104.05991077119184, "grad_norm": 0.2043066769838333, "learning_rate": 4.682252017263988e-05, "loss": 0.0159, "step": 163270 }, { "epoch": 104.06628425748885, "grad_norm": 0.28804323077201843, "learning_rate": 4.6817019766527095e-05, "loss": 0.0121, "step": 163280 }, { "epoch": 104.07265774378585, "grad_norm": 0.28990641236305237, "learning_rate": 4.6811519399090506e-05, "loss": 0.0115, "step": 163290 }, { "epoch": 104.07903123008286, "grad_norm": 0.22326548397541046, "learning_rate": 4.680601907039696e-05, "loss": 0.0064, "step": 163300 }, { "epoch": 104.08540471637986, "grad_norm": 0.21274957060813904, "learning_rate": 4.680051878051331e-05, "loss": 0.0078, "step": 163310 }, { "epoch": 104.09177820267686, "grad_norm": 0.16248098015785217, "learning_rate": 4.679501852950638e-05, "loss": 0.0113, "step": 163320 }, { "epoch": 104.09815168897387, "grad_norm": 0.26222631335258484, "learning_rate": 4.6789518317443e-05, "loss": 0.0084, "step": 163330 }, { "epoch": 104.10452517527088, "grad_norm": 0.3327101469039917, "learning_rate": 4.6784018144389994e-05, "loss": 0.0155, "step": 163340 }, { "epoch": 104.11089866156787, "grad_norm": 0.2782753109931946, "learning_rate": 4.677851801041421e-05, "loss": 0.0164, "step": 163350 }, { "epoch": 104.11727214786488, "grad_norm": 0.19989416003227234, "learning_rate": 4.6773017915582484e-05, "loss": 0.0089, "step": 163360 }, { "epoch": 104.12364563416189, "grad_norm": 0.40651217103004456, "learning_rate": 4.676751785996162e-05, "loss": 0.0121, "step": 163370 }, { "epoch": 104.1300191204589, "grad_norm": 0.20534822344779968, "learning_rate": 4.676201784361848e-05, "loss": 0.0107, "step": 163380 }, { "epoch": 104.13639260675589, "grad_norm": 0.2533700168132782, "learning_rate": 4.6756517866619864e-05, "loss": 0.0079, "step": 163390 }, { "epoch": 104.1427660930529, "grad_norm": 0.2691226601600647, "learning_rate": 4.6751017929032635e-05, "loss": 0.012, "step": 163400 }, { "epoch": 104.1491395793499, "grad_norm": 0.2907792627811432, "learning_rate": 4.67455180309236e-05, "loss": 0.0073, "step": 163410 }, { "epoch": 104.15551306564691, "grad_norm": 0.27021995186805725, "learning_rate": 4.674001817235959e-05, "loss": 0.0139, "step": 163420 }, { "epoch": 104.16188655194391, "grad_norm": 0.24711008369922638, "learning_rate": 4.6734518353407435e-05, "loss": 0.0125, "step": 163430 }, { "epoch": 104.16826003824092, "grad_norm": 0.2666178047657013, "learning_rate": 4.672901857413398e-05, "loss": 0.0108, "step": 163440 }, { "epoch": 104.17463352453792, "grad_norm": 0.2100420743227005, "learning_rate": 4.672351883460603e-05, "loss": 0.0093, "step": 163450 }, { "epoch": 104.18100701083493, "grad_norm": 0.21634916961193085, "learning_rate": 4.671801913489041e-05, "loss": 0.0135, "step": 163460 }, { "epoch": 104.18738049713193, "grad_norm": 0.23923730850219727, "learning_rate": 4.671251947505396e-05, "loss": 0.0101, "step": 163470 }, { "epoch": 104.19375398342893, "grad_norm": 0.6870054602622986, "learning_rate": 4.6707019855163505e-05, "loss": 0.0135, "step": 163480 }, { "epoch": 104.20012746972594, "grad_norm": 0.49115434288978577, "learning_rate": 4.670152027528588e-05, "loss": 0.007, "step": 163490 }, { "epoch": 104.20650095602295, "grad_norm": 0.22706547379493713, "learning_rate": 4.669602073548788e-05, "loss": 0.0086, "step": 163500 }, { "epoch": 104.21287444231994, "grad_norm": 0.23440125584602356, "learning_rate": 4.669052123583634e-05, "loss": 0.0206, "step": 163510 }, { "epoch": 104.21924792861695, "grad_norm": 0.34490495920181274, "learning_rate": 4.6685021776398126e-05, "loss": 0.019, "step": 163520 }, { "epoch": 104.22562141491396, "grad_norm": 0.2755185663700104, "learning_rate": 4.667952235724e-05, "loss": 0.0078, "step": 163530 }, { "epoch": 104.23199490121097, "grad_norm": 0.21842850744724274, "learning_rate": 4.667402297842882e-05, "loss": 0.0076, "step": 163540 }, { "epoch": 104.23836838750796, "grad_norm": 0.56785649061203, "learning_rate": 4.666852364003139e-05, "loss": 0.0146, "step": 163550 }, { "epoch": 104.24474187380497, "grad_norm": 0.44322529435157776, "learning_rate": 4.666302434211457e-05, "loss": 0.0132, "step": 163560 }, { "epoch": 104.25111536010198, "grad_norm": 0.16166920959949493, "learning_rate": 4.665752508474513e-05, "loss": 0.0109, "step": 163570 }, { "epoch": 104.25748884639899, "grad_norm": 0.48840954899787903, "learning_rate": 4.665202586798992e-05, "loss": 0.0094, "step": 163580 }, { "epoch": 104.26386233269598, "grad_norm": 0.2522217333316803, "learning_rate": 4.6646526691915755e-05, "loss": 0.0091, "step": 163590 }, { "epoch": 104.27023581899299, "grad_norm": 0.36070939898490906, "learning_rate": 4.664102755658948e-05, "loss": 0.0184, "step": 163600 }, { "epoch": 104.27660930529, "grad_norm": 0.2944757044315338, "learning_rate": 4.6635528462077866e-05, "loss": 0.012, "step": 163610 }, { "epoch": 104.282982791587, "grad_norm": 0.22832657396793365, "learning_rate": 4.663002940844776e-05, "loss": 0.008, "step": 163620 }, { "epoch": 104.289356277884, "grad_norm": 0.1444467455148697, "learning_rate": 4.6624530395766e-05, "loss": 0.0129, "step": 163630 }, { "epoch": 104.295729764181, "grad_norm": 0.15508335828781128, "learning_rate": 4.661903142409936e-05, "loss": 0.0106, "step": 163640 }, { "epoch": 104.30210325047801, "grad_norm": 0.1880432516336441, "learning_rate": 4.6613532493514685e-05, "loss": 0.0092, "step": 163650 }, { "epoch": 104.30847673677502, "grad_norm": 0.20938415825366974, "learning_rate": 4.660803360407878e-05, "loss": 0.0117, "step": 163660 }, { "epoch": 104.31485022307201, "grad_norm": 0.3033745586872101, "learning_rate": 4.66025347558585e-05, "loss": 0.0179, "step": 163670 }, { "epoch": 104.32122370936902, "grad_norm": 0.21578897535800934, "learning_rate": 4.6597035948920606e-05, "loss": 0.0128, "step": 163680 }, { "epoch": 104.32759719566603, "grad_norm": 0.4568313956260681, "learning_rate": 4.659153718333194e-05, "loss": 0.0156, "step": 163690 }, { "epoch": 104.33397068196304, "grad_norm": 0.2867836654186249, "learning_rate": 4.658603845915933e-05, "loss": 0.0113, "step": 163700 }, { "epoch": 104.34034416826003, "grad_norm": 0.14783108234405518, "learning_rate": 4.658053977646956e-05, "loss": 0.0073, "step": 163710 }, { "epoch": 104.34671765455704, "grad_norm": 0.12087563425302505, "learning_rate": 4.657504113532946e-05, "loss": 0.0083, "step": 163720 }, { "epoch": 104.35309114085405, "grad_norm": 0.2584158182144165, "learning_rate": 4.656954253580584e-05, "loss": 0.012, "step": 163730 }, { "epoch": 104.35946462715106, "grad_norm": 0.214347705245018, "learning_rate": 4.656404397796554e-05, "loss": 0.0127, "step": 163740 }, { "epoch": 104.36583811344805, "grad_norm": 0.3729230761528015, "learning_rate": 4.6558545461875324e-05, "loss": 0.0098, "step": 163750 }, { "epoch": 104.37221159974506, "grad_norm": 0.19144423305988312, "learning_rate": 4.6553046987602035e-05, "loss": 0.0228, "step": 163760 }, { "epoch": 104.37858508604207, "grad_norm": 0.21907897293567657, "learning_rate": 4.6547548555212505e-05, "loss": 0.0075, "step": 163770 }, { "epoch": 104.38495857233907, "grad_norm": 0.4117269814014435, "learning_rate": 4.6542050164773495e-05, "loss": 0.0112, "step": 163780 }, { "epoch": 104.39133205863607, "grad_norm": 0.26180461049079895, "learning_rate": 4.653655181635184e-05, "loss": 0.0123, "step": 163790 }, { "epoch": 104.39770554493307, "grad_norm": 0.2856093943119049, "learning_rate": 4.6531053510014344e-05, "loss": 0.0182, "step": 163800 }, { "epoch": 104.40407903123008, "grad_norm": 0.22326496243476868, "learning_rate": 4.652555524582785e-05, "loss": 0.0166, "step": 163810 }, { "epoch": 104.41045251752709, "grad_norm": 0.30212801694869995, "learning_rate": 4.652005702385912e-05, "loss": 0.0183, "step": 163820 }, { "epoch": 104.4168260038241, "grad_norm": 0.3986416459083557, "learning_rate": 4.6514558844174986e-05, "loss": 0.0183, "step": 163830 }, { "epoch": 104.42319949012109, "grad_norm": 0.305001437664032, "learning_rate": 4.650906070684227e-05, "loss": 0.013, "step": 163840 }, { "epoch": 104.4295729764181, "grad_norm": 0.19282875955104828, "learning_rate": 4.6503562611927734e-05, "loss": 0.0086, "step": 163850 }, { "epoch": 104.43594646271511, "grad_norm": 0.2991209626197815, "learning_rate": 4.6498064559498224e-05, "loss": 0.0133, "step": 163860 }, { "epoch": 104.44231994901212, "grad_norm": 0.3888024389743805, "learning_rate": 4.649256654962053e-05, "loss": 0.0226, "step": 163870 }, { "epoch": 104.44869343530911, "grad_norm": 0.44788652658462524, "learning_rate": 4.6487068582361494e-05, "loss": 0.0105, "step": 163880 }, { "epoch": 104.45506692160612, "grad_norm": 0.29970017075538635, "learning_rate": 4.648157065778786e-05, "loss": 0.0113, "step": 163890 }, { "epoch": 104.46144040790313, "grad_norm": 0.25438860058784485, "learning_rate": 4.6476072775966474e-05, "loss": 0.0106, "step": 163900 }, { "epoch": 104.46781389420013, "grad_norm": 0.3109745681285858, "learning_rate": 4.647057493696412e-05, "loss": 0.0123, "step": 163910 }, { "epoch": 104.47418738049713, "grad_norm": 0.26081159710884094, "learning_rate": 4.646507714084764e-05, "loss": 0.0077, "step": 163920 }, { "epoch": 104.48056086679414, "grad_norm": 0.32104024291038513, "learning_rate": 4.645957938768379e-05, "loss": 0.0073, "step": 163930 }, { "epoch": 104.48693435309114, "grad_norm": 0.18379820883274078, "learning_rate": 4.645408167753939e-05, "loss": 0.0127, "step": 163940 }, { "epoch": 104.49330783938815, "grad_norm": 0.3320101201534271, "learning_rate": 4.644858401048127e-05, "loss": 0.0089, "step": 163950 }, { "epoch": 104.49968132568515, "grad_norm": 0.234152689576149, "learning_rate": 4.6443086386576185e-05, "loss": 0.0149, "step": 163960 }, { "epoch": 104.50605481198215, "grad_norm": 0.34539395570755005, "learning_rate": 4.6437588805890956e-05, "loss": 0.0105, "step": 163970 }, { "epoch": 104.51242829827916, "grad_norm": 0.22773750126361847, "learning_rate": 4.643209126849239e-05, "loss": 0.0072, "step": 163980 }, { "epoch": 104.51880178457617, "grad_norm": 0.45707055926322937, "learning_rate": 4.64265937744473e-05, "loss": 0.0149, "step": 163990 }, { "epoch": 104.52517527087316, "grad_norm": 0.14401966333389282, "learning_rate": 4.6421096323822455e-05, "loss": 0.008, "step": 164000 }, { "epoch": 104.53154875717017, "grad_norm": 0.383672833442688, "learning_rate": 4.641559891668467e-05, "loss": 0.0181, "step": 164010 }, { "epoch": 104.53792224346718, "grad_norm": 0.2964261472225189, "learning_rate": 4.6410101553100745e-05, "loss": 0.008, "step": 164020 }, { "epoch": 104.54429572976419, "grad_norm": 0.46871957182884216, "learning_rate": 4.6404604233137474e-05, "loss": 0.0099, "step": 164030 }, { "epoch": 104.55066921606118, "grad_norm": 0.7012743949890137, "learning_rate": 4.6399106956861656e-05, "loss": 0.0098, "step": 164040 }, { "epoch": 104.55704270235819, "grad_norm": 0.1887843757867813, "learning_rate": 4.639360972434008e-05, "loss": 0.009, "step": 164050 }, { "epoch": 104.5634161886552, "grad_norm": 0.14248818159103394, "learning_rate": 4.638811253563956e-05, "loss": 0.0096, "step": 164060 }, { "epoch": 104.5697896749522, "grad_norm": 0.2364845871925354, "learning_rate": 4.63826153908269e-05, "loss": 0.0187, "step": 164070 }, { "epoch": 104.5761631612492, "grad_norm": 0.21870701014995575, "learning_rate": 4.637711828996885e-05, "loss": 0.0086, "step": 164080 }, { "epoch": 104.5825366475462, "grad_norm": 0.36938804388046265, "learning_rate": 4.637162123313225e-05, "loss": 0.009, "step": 164090 }, { "epoch": 104.58891013384321, "grad_norm": 0.49443310499191284, "learning_rate": 4.636612422038387e-05, "loss": 0.0083, "step": 164100 }, { "epoch": 104.59528362014022, "grad_norm": 0.1594291627407074, "learning_rate": 4.636062725179052e-05, "loss": 0.0126, "step": 164110 }, { "epoch": 104.60165710643722, "grad_norm": 0.29700562357902527, "learning_rate": 4.635513032741897e-05, "loss": 0.013, "step": 164120 }, { "epoch": 104.60803059273422, "grad_norm": 0.2815585434436798, "learning_rate": 4.634963344733604e-05, "loss": 0.0118, "step": 164130 }, { "epoch": 104.61440407903123, "grad_norm": 0.5758978128433228, "learning_rate": 4.63441366116085e-05, "loss": 0.0094, "step": 164140 }, { "epoch": 104.62077756532824, "grad_norm": 0.2055065929889679, "learning_rate": 4.6338639820303166e-05, "loss": 0.0108, "step": 164150 }, { "epoch": 104.62715105162523, "grad_norm": 0.3285977244377136, "learning_rate": 4.633314307348681e-05, "loss": 0.0147, "step": 164160 }, { "epoch": 104.63352453792224, "grad_norm": 0.1710968315601349, "learning_rate": 4.632764637122622e-05, "loss": 0.007, "step": 164170 }, { "epoch": 104.63989802421925, "grad_norm": 0.35291457176208496, "learning_rate": 4.6322149713588194e-05, "loss": 0.0104, "step": 164180 }, { "epoch": 104.64627151051626, "grad_norm": 0.2864823639392853, "learning_rate": 4.631665310063953e-05, "loss": 0.0102, "step": 164190 }, { "epoch": 104.65264499681325, "grad_norm": 0.26937997341156006, "learning_rate": 4.6311156532447016e-05, "loss": 0.0097, "step": 164200 }, { "epoch": 104.65901848311026, "grad_norm": 0.2572837769985199, "learning_rate": 4.630566000907741e-05, "loss": 0.0169, "step": 164210 }, { "epoch": 104.66539196940727, "grad_norm": 0.13870076835155487, "learning_rate": 4.630016353059753e-05, "loss": 0.0147, "step": 164220 }, { "epoch": 104.67176545570427, "grad_norm": 0.3163048326969147, "learning_rate": 4.629466709707417e-05, "loss": 0.0102, "step": 164230 }, { "epoch": 104.67813894200127, "grad_norm": 1.9252128601074219, "learning_rate": 4.62891707085741e-05, "loss": 0.0159, "step": 164240 }, { "epoch": 104.68451242829828, "grad_norm": 0.7248860597610474, "learning_rate": 4.62836743651641e-05, "loss": 0.0157, "step": 164250 }, { "epoch": 104.69088591459528, "grad_norm": 0.32870733737945557, "learning_rate": 4.627817806691096e-05, "loss": 0.0112, "step": 164260 }, { "epoch": 104.69725940089229, "grad_norm": 0.1962704360485077, "learning_rate": 4.62726818138815e-05, "loss": 0.0088, "step": 164270 }, { "epoch": 104.70363288718929, "grad_norm": 0.22991816699504852, "learning_rate": 4.6267185606142444e-05, "loss": 0.0092, "step": 164280 }, { "epoch": 104.7100063734863, "grad_norm": 0.21324129402637482, "learning_rate": 4.6261689443760616e-05, "loss": 0.0132, "step": 164290 }, { "epoch": 104.7163798597833, "grad_norm": 0.7409241795539856, "learning_rate": 4.625619332680278e-05, "loss": 0.0208, "step": 164300 }, { "epoch": 104.72275334608031, "grad_norm": 0.2520051896572113, "learning_rate": 4.6250697255335764e-05, "loss": 0.012, "step": 164310 }, { "epoch": 104.7291268323773, "grad_norm": 0.3420650362968445, "learning_rate": 4.6245201229426285e-05, "loss": 0.0091, "step": 164320 }, { "epoch": 104.73550031867431, "grad_norm": 0.38106369972229004, "learning_rate": 4.6239705249141166e-05, "loss": 0.0077, "step": 164330 }, { "epoch": 104.74187380497132, "grad_norm": 0.22097742557525635, "learning_rate": 4.623420931454719e-05, "loss": 0.0115, "step": 164340 }, { "epoch": 104.74824729126833, "grad_norm": 0.28782784938812256, "learning_rate": 4.622871342571111e-05, "loss": 0.0144, "step": 164350 }, { "epoch": 104.75462077756532, "grad_norm": 0.22496461868286133, "learning_rate": 4.622321758269972e-05, "loss": 0.0128, "step": 164360 }, { "epoch": 104.76099426386233, "grad_norm": 0.3653685450553894, "learning_rate": 4.6217721785579806e-05, "loss": 0.0097, "step": 164370 }, { "epoch": 104.76736775015934, "grad_norm": 0.23268035054206848, "learning_rate": 4.621222603441816e-05, "loss": 0.012, "step": 164380 }, { "epoch": 104.77374123645635, "grad_norm": 0.3711617887020111, "learning_rate": 4.6206730329281525e-05, "loss": 0.0108, "step": 164390 }, { "epoch": 104.78011472275334, "grad_norm": 0.3092097342014313, "learning_rate": 4.62012346702367e-05, "loss": 0.0156, "step": 164400 }, { "epoch": 104.78648820905035, "grad_norm": 0.2653959095478058, "learning_rate": 4.619573905735048e-05, "loss": 0.0097, "step": 164410 }, { "epoch": 104.79286169534736, "grad_norm": 0.19696390628814697, "learning_rate": 4.61902434906896e-05, "loss": 0.0148, "step": 164420 }, { "epoch": 104.79923518164436, "grad_norm": 0.31459084153175354, "learning_rate": 4.618474797032086e-05, "loss": 0.0113, "step": 164430 }, { "epoch": 104.80560866794136, "grad_norm": 0.5098999738693237, "learning_rate": 4.6179252496311036e-05, "loss": 0.0103, "step": 164440 }, { "epoch": 104.81198215423836, "grad_norm": 0.35473471879959106, "learning_rate": 4.617375706872692e-05, "loss": 0.011, "step": 164450 }, { "epoch": 104.81835564053537, "grad_norm": 0.2533298432826996, "learning_rate": 4.616826168763525e-05, "loss": 0.0101, "step": 164460 }, { "epoch": 104.82472912683238, "grad_norm": 0.20615263283252716, "learning_rate": 4.616276635310281e-05, "loss": 0.0119, "step": 164470 }, { "epoch": 104.83110261312939, "grad_norm": 0.3131899833679199, "learning_rate": 4.61572710651964e-05, "loss": 0.0172, "step": 164480 }, { "epoch": 104.83747609942638, "grad_norm": 0.3791768252849579, "learning_rate": 4.615177582398278e-05, "loss": 0.0112, "step": 164490 }, { "epoch": 104.84384958572339, "grad_norm": 0.5734468102455139, "learning_rate": 4.61462806295287e-05, "loss": 0.0126, "step": 164500 }, { "epoch": 104.8502230720204, "grad_norm": 0.36565130949020386, "learning_rate": 4.614078548190095e-05, "loss": 0.0109, "step": 164510 }, { "epoch": 104.8565965583174, "grad_norm": 0.3347036838531494, "learning_rate": 4.613529038116632e-05, "loss": 0.0171, "step": 164520 }, { "epoch": 104.8629700446144, "grad_norm": 0.2848658263683319, "learning_rate": 4.6129795327391536e-05, "loss": 0.0111, "step": 164530 }, { "epoch": 104.86934353091141, "grad_norm": 0.21215978264808655, "learning_rate": 4.61243003206434e-05, "loss": 0.0148, "step": 164540 }, { "epoch": 104.87571701720842, "grad_norm": 0.3046610951423645, "learning_rate": 4.611880536098867e-05, "loss": 0.0174, "step": 164550 }, { "epoch": 104.88209050350542, "grad_norm": 0.20114891231060028, "learning_rate": 4.611331044849414e-05, "loss": 0.0079, "step": 164560 }, { "epoch": 104.88846398980242, "grad_norm": 0.2391107976436615, "learning_rate": 4.6107815583226535e-05, "loss": 0.0095, "step": 164570 }, { "epoch": 104.89483747609943, "grad_norm": 0.34429430961608887, "learning_rate": 4.610232076525265e-05, "loss": 0.0092, "step": 164580 }, { "epoch": 104.90121096239643, "grad_norm": 0.179554745554924, "learning_rate": 4.6096825994639264e-05, "loss": 0.0141, "step": 164590 }, { "epoch": 104.90758444869344, "grad_norm": 0.6349259614944458, "learning_rate": 4.609133127145311e-05, "loss": 0.0294, "step": 164600 }, { "epoch": 104.91395793499044, "grad_norm": 0.36973434686660767, "learning_rate": 4.608583659576097e-05, "loss": 0.0224, "step": 164610 }, { "epoch": 104.92033142128744, "grad_norm": 0.34637245535850525, "learning_rate": 4.6080341967629614e-05, "loss": 0.0105, "step": 164620 }, { "epoch": 104.92670490758445, "grad_norm": 0.4223269522190094, "learning_rate": 4.607484738712583e-05, "loss": 0.0146, "step": 164630 }, { "epoch": 104.93307839388146, "grad_norm": 0.1324138343334198, "learning_rate": 4.6069352854316324e-05, "loss": 0.01, "step": 164640 }, { "epoch": 104.93945188017845, "grad_norm": 0.11673499643802643, "learning_rate": 4.606385836926789e-05, "loss": 0.016, "step": 164650 }, { "epoch": 104.94582536647546, "grad_norm": 0.24793964624404907, "learning_rate": 4.6058363932047325e-05, "loss": 0.0138, "step": 164660 }, { "epoch": 104.95219885277247, "grad_norm": 0.319110244512558, "learning_rate": 4.6052869542721334e-05, "loss": 0.0118, "step": 164670 }, { "epoch": 104.95857233906948, "grad_norm": 0.4385717511177063, "learning_rate": 4.60473752013567e-05, "loss": 0.0117, "step": 164680 }, { "epoch": 104.96494582536647, "grad_norm": 0.5065938830375671, "learning_rate": 4.6041880908020196e-05, "loss": 0.0111, "step": 164690 }, { "epoch": 104.97131931166348, "grad_norm": 0.38248303532600403, "learning_rate": 4.603638666277859e-05, "loss": 0.0125, "step": 164700 }, { "epoch": 104.97769279796049, "grad_norm": 0.2089262753725052, "learning_rate": 4.603089246569862e-05, "loss": 0.0134, "step": 164710 }, { "epoch": 104.9840662842575, "grad_norm": 0.15473276376724243, "learning_rate": 4.602539831684704e-05, "loss": 0.0118, "step": 164720 }, { "epoch": 104.99043977055449, "grad_norm": 0.24449460208415985, "learning_rate": 4.6019904216290636e-05, "loss": 0.0122, "step": 164730 }, { "epoch": 104.9968132568515, "grad_norm": 0.10775483399629593, "learning_rate": 4.601441016409616e-05, "loss": 0.0181, "step": 164740 }, { "epoch": 105.0031867431485, "grad_norm": 0.45335689187049866, "learning_rate": 4.600891616033035e-05, "loss": 0.0079, "step": 164750 }, { "epoch": 105.00956022944551, "grad_norm": 0.14054228365421295, "learning_rate": 4.6003422205059976e-05, "loss": 0.0044, "step": 164760 }, { "epoch": 105.0159337157425, "grad_norm": 0.1543091982603073, "learning_rate": 4.599792829835181e-05, "loss": 0.0095, "step": 164770 }, { "epoch": 105.02230720203951, "grad_norm": 0.15779204666614532, "learning_rate": 4.599243444027259e-05, "loss": 0.0113, "step": 164780 }, { "epoch": 105.02868068833652, "grad_norm": 0.19850915670394897, "learning_rate": 4.5986940630889066e-05, "loss": 0.0079, "step": 164790 }, { "epoch": 105.03505417463353, "grad_norm": 0.3490373492240906, "learning_rate": 4.5981446870268e-05, "loss": 0.0147, "step": 164800 }, { "epoch": 105.04142766093052, "grad_norm": 0.253221720457077, "learning_rate": 4.597595315847616e-05, "loss": 0.0119, "step": 164810 }, { "epoch": 105.04780114722753, "grad_norm": 0.31387749314308167, "learning_rate": 4.597045949558029e-05, "loss": 0.0167, "step": 164820 }, { "epoch": 105.05417463352454, "grad_norm": 0.5084949135780334, "learning_rate": 4.5964965881647134e-05, "loss": 0.0146, "step": 164830 }, { "epoch": 105.06054811982155, "grad_norm": 0.4515387713909149, "learning_rate": 4.595947231674346e-05, "loss": 0.0082, "step": 164840 }, { "epoch": 105.06692160611854, "grad_norm": 0.26384785771369934, "learning_rate": 4.5953978800936e-05, "loss": 0.0234, "step": 164850 }, { "epoch": 105.07329509241555, "grad_norm": 0.4057203531265259, "learning_rate": 4.594848533429153e-05, "loss": 0.0082, "step": 164860 }, { "epoch": 105.07966857871256, "grad_norm": 0.3421948552131653, "learning_rate": 4.5942991916876776e-05, "loss": 0.0116, "step": 164870 }, { "epoch": 105.08604206500956, "grad_norm": 0.2277587503194809, "learning_rate": 4.5937498548758514e-05, "loss": 0.0122, "step": 164880 }, { "epoch": 105.09241555130656, "grad_norm": 0.26995304226875305, "learning_rate": 4.593200523000347e-05, "loss": 0.016, "step": 164890 }, { "epoch": 105.09878903760357, "grad_norm": 0.3554224371910095, "learning_rate": 4.592651196067841e-05, "loss": 0.0116, "step": 164900 }, { "epoch": 105.10516252390057, "grad_norm": 0.40067335963249207, "learning_rate": 4.5921018740850075e-05, "loss": 0.0157, "step": 164910 }, { "epoch": 105.11153601019758, "grad_norm": 0.18783152103424072, "learning_rate": 4.5915525570585205e-05, "loss": 0.0099, "step": 164920 }, { "epoch": 105.11790949649458, "grad_norm": 0.24049372971057892, "learning_rate": 4.5910032449950556e-05, "loss": 0.0096, "step": 164930 }, { "epoch": 105.12428298279158, "grad_norm": 0.49375441670417786, "learning_rate": 4.590453937901289e-05, "loss": 0.0129, "step": 164940 }, { "epoch": 105.13065646908859, "grad_norm": 0.38762179017066956, "learning_rate": 4.589904635783893e-05, "loss": 0.0107, "step": 164950 }, { "epoch": 105.1370299553856, "grad_norm": 0.4207230508327484, "learning_rate": 4.589355338649542e-05, "loss": 0.0146, "step": 164960 }, { "epoch": 105.1434034416826, "grad_norm": 0.4865589141845703, "learning_rate": 4.588806046504912e-05, "loss": 0.0148, "step": 164970 }, { "epoch": 105.1497769279796, "grad_norm": 0.277074933052063, "learning_rate": 4.588256759356678e-05, "loss": 0.0086, "step": 164980 }, { "epoch": 105.15615041427661, "grad_norm": 0.13391909003257751, "learning_rate": 4.5877074772115116e-05, "loss": 0.0073, "step": 164990 }, { "epoch": 105.16252390057362, "grad_norm": 0.3552212119102478, "learning_rate": 4.587158200076087e-05, "loss": 0.0108, "step": 165000 }, { "epoch": 105.16889738687061, "grad_norm": 0.16519656777381897, "learning_rate": 4.586608927957082e-05, "loss": 0.0198, "step": 165010 }, { "epoch": 105.17527087316762, "grad_norm": 0.17419427633285522, "learning_rate": 4.586059660861171e-05, "loss": 0.0046, "step": 165020 }, { "epoch": 105.18164435946463, "grad_norm": 0.37047427892684937, "learning_rate": 4.585510398795023e-05, "loss": 0.0171, "step": 165030 }, { "epoch": 105.18801784576164, "grad_norm": 0.33010709285736084, "learning_rate": 4.584961141765315e-05, "loss": 0.0074, "step": 165040 }, { "epoch": 105.19439133205863, "grad_norm": 0.19728612899780273, "learning_rate": 4.584411889778723e-05, "loss": 0.0078, "step": 165050 }, { "epoch": 105.20076481835564, "grad_norm": 0.4565852880477905, "learning_rate": 4.583862642841917e-05, "loss": 0.0178, "step": 165060 }, { "epoch": 105.20713830465264, "grad_norm": 0.3124496042728424, "learning_rate": 4.583313400961573e-05, "loss": 0.0127, "step": 165070 }, { "epoch": 105.21351179094965, "grad_norm": 0.34568503499031067, "learning_rate": 4.5827641641443645e-05, "loss": 0.0097, "step": 165080 }, { "epoch": 105.21988527724666, "grad_norm": 0.2898944914340973, "learning_rate": 4.5822149323969676e-05, "loss": 0.01, "step": 165090 }, { "epoch": 105.22625876354365, "grad_norm": 0.2628885805606842, "learning_rate": 4.581665705726051e-05, "loss": 0.0065, "step": 165100 }, { "epoch": 105.23263224984066, "grad_norm": 0.3364340364933014, "learning_rate": 4.581116484138291e-05, "loss": 0.0123, "step": 165110 }, { "epoch": 105.23900573613767, "grad_norm": 0.32400092482566833, "learning_rate": 4.580567267640362e-05, "loss": 0.0111, "step": 165120 }, { "epoch": 105.24537922243468, "grad_norm": 0.2918822169303894, "learning_rate": 4.580018056238939e-05, "loss": 0.0157, "step": 165130 }, { "epoch": 105.25175270873167, "grad_norm": 0.270231157541275, "learning_rate": 4.57946884994069e-05, "loss": 0.0099, "step": 165140 }, { "epoch": 105.25812619502868, "grad_norm": 0.3671479821205139, "learning_rate": 4.5789196487522926e-05, "loss": 0.0168, "step": 165150 }, { "epoch": 105.26449968132569, "grad_norm": 0.2999049723148346, "learning_rate": 4.5783704526804204e-05, "loss": 0.0061, "step": 165160 }, { "epoch": 105.2708731676227, "grad_norm": 0.1813345104455948, "learning_rate": 4.577821261731744e-05, "loss": 0.0115, "step": 165170 }, { "epoch": 105.27724665391969, "grad_norm": 0.1951654553413391, "learning_rate": 4.577272075912937e-05, "loss": 0.013, "step": 165180 }, { "epoch": 105.2836201402167, "grad_norm": 0.178595170378685, "learning_rate": 4.576722895230673e-05, "loss": 0.0058, "step": 165190 }, { "epoch": 105.2899936265137, "grad_norm": 0.3757530152797699, "learning_rate": 4.576173719691628e-05, "loss": 0.0101, "step": 165200 }, { "epoch": 105.29636711281071, "grad_norm": 0.15660516917705536, "learning_rate": 4.5756245493024706e-05, "loss": 0.0083, "step": 165210 }, { "epoch": 105.30274059910771, "grad_norm": 0.30324986577033997, "learning_rate": 4.5750753840698754e-05, "loss": 0.0108, "step": 165220 }, { "epoch": 105.30911408540472, "grad_norm": 0.26270732283592224, "learning_rate": 4.574526224000518e-05, "loss": 0.0143, "step": 165230 }, { "epoch": 105.31548757170172, "grad_norm": 0.37147343158721924, "learning_rate": 4.573977069101066e-05, "loss": 0.015, "step": 165240 }, { "epoch": 105.32186105799873, "grad_norm": 0.2600480616092682, "learning_rate": 4.573427919378196e-05, "loss": 0.0082, "step": 165250 }, { "epoch": 105.32823454429573, "grad_norm": 0.32712408900260925, "learning_rate": 4.572878774838578e-05, "loss": 0.009, "step": 165260 }, { "epoch": 105.33460803059273, "grad_norm": 0.3003917336463928, "learning_rate": 4.5723296354888894e-05, "loss": 0.0073, "step": 165270 }, { "epoch": 105.34098151688974, "grad_norm": 0.2143411487340927, "learning_rate": 4.571780501335797e-05, "loss": 0.0249, "step": 165280 }, { "epoch": 105.34735500318675, "grad_norm": 0.4603065252304077, "learning_rate": 4.5712313723859756e-05, "loss": 0.0115, "step": 165290 }, { "epoch": 105.35372848948374, "grad_norm": 0.2059091031551361, "learning_rate": 4.5706822486461004e-05, "loss": 0.0099, "step": 165300 }, { "epoch": 105.36010197578075, "grad_norm": 0.23932264745235443, "learning_rate": 4.570133130122839e-05, "loss": 0.0089, "step": 165310 }, { "epoch": 105.36647546207776, "grad_norm": 0.36996400356292725, "learning_rate": 4.569584016822865e-05, "loss": 0.0132, "step": 165320 }, { "epoch": 105.37284894837477, "grad_norm": 0.20208922028541565, "learning_rate": 4.569034908752853e-05, "loss": 0.0165, "step": 165330 }, { "epoch": 105.37922243467176, "grad_norm": 0.5816066265106201, "learning_rate": 4.5684858059194754e-05, "loss": 0.0123, "step": 165340 }, { "epoch": 105.38559592096877, "grad_norm": 0.6565074920654297, "learning_rate": 4.5679367083294006e-05, "loss": 0.0096, "step": 165350 }, { "epoch": 105.39196940726578, "grad_norm": 0.3003045618534088, "learning_rate": 4.5673876159893025e-05, "loss": 0.0099, "step": 165360 }, { "epoch": 105.39834289356278, "grad_norm": 0.22809910774230957, "learning_rate": 4.566838528905855e-05, "loss": 0.0078, "step": 165370 }, { "epoch": 105.40471637985978, "grad_norm": 0.42286980152130127, "learning_rate": 4.566289447085727e-05, "loss": 0.0094, "step": 165380 }, { "epoch": 105.41108986615679, "grad_norm": 0.19107605516910553, "learning_rate": 4.565740370535591e-05, "loss": 0.0114, "step": 165390 }, { "epoch": 105.4174633524538, "grad_norm": 0.22961324453353882, "learning_rate": 4.56519129926212e-05, "loss": 0.0096, "step": 165400 }, { "epoch": 105.4238368387508, "grad_norm": 0.7763230204582214, "learning_rate": 4.564642233271987e-05, "loss": 0.01, "step": 165410 }, { "epoch": 105.4302103250478, "grad_norm": 0.20595932006835938, "learning_rate": 4.564093172571861e-05, "loss": 0.0153, "step": 165420 }, { "epoch": 105.4365838113448, "grad_norm": 0.2659761607646942, "learning_rate": 4.563544117168413e-05, "loss": 0.0118, "step": 165430 }, { "epoch": 105.44295729764181, "grad_norm": 0.28715261816978455, "learning_rate": 4.5629950670683166e-05, "loss": 0.0169, "step": 165440 }, { "epoch": 105.44933078393882, "grad_norm": 0.17832614481449127, "learning_rate": 4.562446022278245e-05, "loss": 0.0096, "step": 165450 }, { "epoch": 105.45570427023581, "grad_norm": 0.30551865696907043, "learning_rate": 4.5618969828048655e-05, "loss": 0.017, "step": 165460 }, { "epoch": 105.46207775653282, "grad_norm": 0.23358982801437378, "learning_rate": 4.561347948654851e-05, "loss": 0.0116, "step": 165470 }, { "epoch": 105.46845124282983, "grad_norm": 0.36578696966171265, "learning_rate": 4.560798919834874e-05, "loss": 0.0098, "step": 165480 }, { "epoch": 105.47482472912684, "grad_norm": 0.31599462032318115, "learning_rate": 4.560249896351605e-05, "loss": 0.017, "step": 165490 }, { "epoch": 105.48119821542383, "grad_norm": 0.22916024923324585, "learning_rate": 4.5597008782117134e-05, "loss": 0.0074, "step": 165500 }, { "epoch": 105.48757170172084, "grad_norm": 0.5234941840171814, "learning_rate": 4.5591518654218726e-05, "loss": 0.0174, "step": 165510 }, { "epoch": 105.49394518801785, "grad_norm": 0.45782601833343506, "learning_rate": 4.558602857988753e-05, "loss": 0.0157, "step": 165520 }, { "epoch": 105.50031867431485, "grad_norm": 0.34328165650367737, "learning_rate": 4.558053855919026e-05, "loss": 0.0239, "step": 165530 }, { "epoch": 105.50669216061185, "grad_norm": 0.17353937029838562, "learning_rate": 4.5575048592193606e-05, "loss": 0.0076, "step": 165540 }, { "epoch": 105.51306564690886, "grad_norm": 0.22570361196994781, "learning_rate": 4.55695586789643e-05, "loss": 0.0071, "step": 165550 }, { "epoch": 105.51943913320586, "grad_norm": 0.330072283744812, "learning_rate": 4.5564068819569034e-05, "loss": 0.0112, "step": 165560 }, { "epoch": 105.52581261950287, "grad_norm": 0.22611400485038757, "learning_rate": 4.555857901407453e-05, "loss": 0.019, "step": 165570 }, { "epoch": 105.53218610579987, "grad_norm": 0.2677900493144989, "learning_rate": 4.555308926254747e-05, "loss": 0.0144, "step": 165580 }, { "epoch": 105.53855959209687, "grad_norm": 0.2223922163248062, "learning_rate": 4.554759956505458e-05, "loss": 0.0181, "step": 165590 }, { "epoch": 105.54493307839388, "grad_norm": 0.11938492953777313, "learning_rate": 4.554210992166256e-05, "loss": 0.0114, "step": 165600 }, { "epoch": 105.55130656469089, "grad_norm": 0.5961727499961853, "learning_rate": 4.553662033243812e-05, "loss": 0.0148, "step": 165610 }, { "epoch": 105.55768005098788, "grad_norm": 0.14510346949100494, "learning_rate": 4.5531130797447954e-05, "loss": 0.0093, "step": 165620 }, { "epoch": 105.56405353728489, "grad_norm": 0.24920956790447235, "learning_rate": 4.5525641316758754e-05, "loss": 0.0134, "step": 165630 }, { "epoch": 105.5704270235819, "grad_norm": 0.4684341251850128, "learning_rate": 4.5520151890437244e-05, "loss": 0.0108, "step": 165640 }, { "epoch": 105.57680050987891, "grad_norm": 0.26735353469848633, "learning_rate": 4.551466251855013e-05, "loss": 0.0193, "step": 165650 }, { "epoch": 105.5831739961759, "grad_norm": 0.11165406554937363, "learning_rate": 4.55091732011641e-05, "loss": 0.0072, "step": 165660 }, { "epoch": 105.58954748247291, "grad_norm": 0.5248162150382996, "learning_rate": 4.550368393834584e-05, "loss": 0.017, "step": 165670 }, { "epoch": 105.59592096876992, "grad_norm": 0.4059060513973236, "learning_rate": 4.5498194730162066e-05, "loss": 0.0134, "step": 165680 }, { "epoch": 105.60229445506693, "grad_norm": 0.20850513875484467, "learning_rate": 4.5492705576679484e-05, "loss": 0.0118, "step": 165690 }, { "epoch": 105.60866794136393, "grad_norm": 0.24125045537948608, "learning_rate": 4.5487216477964786e-05, "loss": 0.0073, "step": 165700 }, { "epoch": 105.61504142766093, "grad_norm": 0.2573881149291992, "learning_rate": 4.548172743408466e-05, "loss": 0.0233, "step": 165710 }, { "epoch": 105.62141491395793, "grad_norm": 0.17981478571891785, "learning_rate": 4.547623844510582e-05, "loss": 0.0084, "step": 165720 }, { "epoch": 105.62778840025494, "grad_norm": 0.24646705389022827, "learning_rate": 4.547074951109497e-05, "loss": 0.0204, "step": 165730 }, { "epoch": 105.63416188655195, "grad_norm": 0.4881674349308014, "learning_rate": 4.5465260632118764e-05, "loss": 0.0153, "step": 165740 }, { "epoch": 105.64053537284894, "grad_norm": 0.48513883352279663, "learning_rate": 4.545977180824393e-05, "loss": 0.0127, "step": 165750 }, { "epoch": 105.64690885914595, "grad_norm": 0.5598253011703491, "learning_rate": 4.545428303953715e-05, "loss": 0.0222, "step": 165760 }, { "epoch": 105.65328234544296, "grad_norm": 0.5572965145111084, "learning_rate": 4.5448794326065145e-05, "loss": 0.009, "step": 165770 }, { "epoch": 105.65965583173997, "grad_norm": 0.3365209996700287, "learning_rate": 4.5443305667894565e-05, "loss": 0.0142, "step": 165780 }, { "epoch": 105.66602931803696, "grad_norm": 1.1334960460662842, "learning_rate": 4.543781706509212e-05, "loss": 0.019, "step": 165790 }, { "epoch": 105.67240280433397, "grad_norm": 0.4801350235939026, "learning_rate": 4.5432328517724534e-05, "loss": 0.0194, "step": 165800 }, { "epoch": 105.67877629063098, "grad_norm": 0.444461852312088, "learning_rate": 4.542684002585844e-05, "loss": 0.0138, "step": 165810 }, { "epoch": 105.68514977692799, "grad_norm": 0.29055944085121155, "learning_rate": 4.5421351589560564e-05, "loss": 0.0191, "step": 165820 }, { "epoch": 105.69152326322498, "grad_norm": 0.21023231744766235, "learning_rate": 4.541586320889759e-05, "loss": 0.0129, "step": 165830 }, { "epoch": 105.69789674952199, "grad_norm": 0.4426427483558655, "learning_rate": 4.541037488393622e-05, "loss": 0.015, "step": 165840 }, { "epoch": 105.704270235819, "grad_norm": 0.16825765371322632, "learning_rate": 4.540488661474311e-05, "loss": 0.0085, "step": 165850 }, { "epoch": 105.710643722116, "grad_norm": 0.3711811304092407, "learning_rate": 4.5399398401384965e-05, "loss": 0.0089, "step": 165860 }, { "epoch": 105.717017208413, "grad_norm": 0.31117525696754456, "learning_rate": 4.5393910243928494e-05, "loss": 0.009, "step": 165870 }, { "epoch": 105.72339069471, "grad_norm": 0.32970720529556274, "learning_rate": 4.538842214244035e-05, "loss": 0.014, "step": 165880 }, { "epoch": 105.72976418100701, "grad_norm": 0.21114099025726318, "learning_rate": 4.538293409698722e-05, "loss": 0.0131, "step": 165890 }, { "epoch": 105.73613766730402, "grad_norm": 0.2937590181827545, "learning_rate": 4.53774461076358e-05, "loss": 0.0083, "step": 165900 }, { "epoch": 105.74251115360101, "grad_norm": 0.1986055225133896, "learning_rate": 4.53719581744528e-05, "loss": 0.0101, "step": 165910 }, { "epoch": 105.74888463989802, "grad_norm": 0.18887728452682495, "learning_rate": 4.536647029750486e-05, "loss": 0.0103, "step": 165920 }, { "epoch": 105.75525812619503, "grad_norm": 0.2097914069890976, "learning_rate": 4.536098247685868e-05, "loss": 0.0214, "step": 165930 }, { "epoch": 105.76163161249204, "grad_norm": 0.26166197657585144, "learning_rate": 4.535549471258096e-05, "loss": 0.0072, "step": 165940 }, { "epoch": 105.76800509878903, "grad_norm": 0.39259180426597595, "learning_rate": 4.535000700473834e-05, "loss": 0.0071, "step": 165950 }, { "epoch": 105.77437858508604, "grad_norm": 0.15137693285942078, "learning_rate": 4.534451935339753e-05, "loss": 0.0061, "step": 165960 }, { "epoch": 105.78075207138305, "grad_norm": 0.5244869589805603, "learning_rate": 4.5339031758625204e-05, "loss": 0.0176, "step": 165970 }, { "epoch": 105.78712555768006, "grad_norm": 0.42969587445259094, "learning_rate": 4.5333544220488066e-05, "loss": 0.0119, "step": 165980 }, { "epoch": 105.79349904397705, "grad_norm": 0.35874873399734497, "learning_rate": 4.532805673905275e-05, "loss": 0.0119, "step": 165990 }, { "epoch": 105.79987253027406, "grad_norm": 0.222822368144989, "learning_rate": 4.5322569314385955e-05, "loss": 0.0119, "step": 166000 }, { "epoch": 105.80624601657107, "grad_norm": 0.470676451921463, "learning_rate": 4.531708194655436e-05, "loss": 0.0105, "step": 166010 }, { "epoch": 105.81261950286807, "grad_norm": 0.39115992188453674, "learning_rate": 4.531159463562466e-05, "loss": 0.0103, "step": 166020 }, { "epoch": 105.81899298916507, "grad_norm": 0.3121834695339203, "learning_rate": 4.5306107381663494e-05, "loss": 0.0115, "step": 166030 }, { "epoch": 105.82536647546208, "grad_norm": 0.30424734950065613, "learning_rate": 4.530062018473755e-05, "loss": 0.0124, "step": 166040 }, { "epoch": 105.83173996175908, "grad_norm": 0.3106956481933594, "learning_rate": 4.529513304491353e-05, "loss": 0.0103, "step": 166050 }, { "epoch": 105.83811344805609, "grad_norm": 0.38186269998550415, "learning_rate": 4.528964596225806e-05, "loss": 0.0075, "step": 166060 }, { "epoch": 105.84448693435309, "grad_norm": 0.2761509418487549, "learning_rate": 4.528415893683785e-05, "loss": 0.0064, "step": 166070 }, { "epoch": 105.8508604206501, "grad_norm": 0.26422494649887085, "learning_rate": 4.5278671968719555e-05, "loss": 0.0153, "step": 166080 }, { "epoch": 105.8572339069471, "grad_norm": 0.26956841349601746, "learning_rate": 4.527318505796988e-05, "loss": 0.012, "step": 166090 }, { "epoch": 105.86360739324411, "grad_norm": 0.30586665868759155, "learning_rate": 4.526769820465544e-05, "loss": 0.0091, "step": 166100 }, { "epoch": 105.8699808795411, "grad_norm": 0.2055150270462036, "learning_rate": 4.526221140884294e-05, "loss": 0.0071, "step": 166110 }, { "epoch": 105.87635436583811, "grad_norm": 0.36304736137390137, "learning_rate": 4.525672467059906e-05, "loss": 0.0114, "step": 166120 }, { "epoch": 105.88272785213512, "grad_norm": 0.5406454801559448, "learning_rate": 4.5251237989990444e-05, "loss": 0.014, "step": 166130 }, { "epoch": 105.88910133843213, "grad_norm": 0.5882258415222168, "learning_rate": 4.524575136708376e-05, "loss": 0.0208, "step": 166140 }, { "epoch": 105.89547482472912, "grad_norm": 0.30486977100372314, "learning_rate": 4.524026480194569e-05, "loss": 0.0122, "step": 166150 }, { "epoch": 105.90184831102613, "grad_norm": 1.7114167213439941, "learning_rate": 4.5234778294642914e-05, "loss": 0.0178, "step": 166160 }, { "epoch": 105.90822179732314, "grad_norm": 0.15974929928779602, "learning_rate": 4.522929184524207e-05, "loss": 0.0058, "step": 166170 }, { "epoch": 105.91459528362014, "grad_norm": 0.4643283188343048, "learning_rate": 4.522380545380982e-05, "loss": 0.0132, "step": 166180 }, { "epoch": 105.92096876991714, "grad_norm": 0.4504989683628082, "learning_rate": 4.521831912041286e-05, "loss": 0.0116, "step": 166190 }, { "epoch": 105.92734225621415, "grad_norm": 0.4237154424190521, "learning_rate": 4.5212832845117836e-05, "loss": 0.009, "step": 166200 }, { "epoch": 105.93371574251115, "grad_norm": 0.16112928092479706, "learning_rate": 4.52073466279914e-05, "loss": 0.0147, "step": 166210 }, { "epoch": 105.94008922880816, "grad_norm": 0.20877911150455475, "learning_rate": 4.520186046910024e-05, "loss": 0.0162, "step": 166220 }, { "epoch": 105.94646271510516, "grad_norm": 0.42304062843322754, "learning_rate": 4.5196374368511e-05, "loss": 0.0076, "step": 166230 }, { "epoch": 105.95283620140216, "grad_norm": 0.20353613793849945, "learning_rate": 4.519088832629036e-05, "loss": 0.0156, "step": 166240 }, { "epoch": 105.95920968769917, "grad_norm": 0.22104640305042267, "learning_rate": 4.518540234250495e-05, "loss": 0.0117, "step": 166250 }, { "epoch": 105.96558317399618, "grad_norm": 0.3754919469356537, "learning_rate": 4.5179916417221455e-05, "loss": 0.0156, "step": 166260 }, { "epoch": 105.97195666029317, "grad_norm": 0.21416927874088287, "learning_rate": 4.517443055050652e-05, "loss": 0.0087, "step": 166270 }, { "epoch": 105.97833014659018, "grad_norm": 0.23531128466129303, "learning_rate": 4.516894474242682e-05, "loss": 0.0094, "step": 166280 }, { "epoch": 105.98470363288719, "grad_norm": 0.1957581788301468, "learning_rate": 4.5163458993048994e-05, "loss": 0.0108, "step": 166290 }, { "epoch": 105.9910771191842, "grad_norm": 0.3626924157142639, "learning_rate": 4.515797330243972e-05, "loss": 0.0104, "step": 166300 }, { "epoch": 105.99745060548119, "grad_norm": 0.2936982810497284, "learning_rate": 4.515248767066562e-05, "loss": 0.0131, "step": 166310 }, { "epoch": 106.0038240917782, "grad_norm": 0.5649032592773438, "learning_rate": 4.5147002097793395e-05, "loss": 0.0116, "step": 166320 }, { "epoch": 106.0101975780752, "grad_norm": 0.3118785321712494, "learning_rate": 4.514151658388966e-05, "loss": 0.0061, "step": 166330 }, { "epoch": 106.01657106437222, "grad_norm": 0.19442786276340485, "learning_rate": 4.5136031129021096e-05, "loss": 0.0098, "step": 166340 }, { "epoch": 106.02294455066922, "grad_norm": 0.36763107776641846, "learning_rate": 4.513054573325434e-05, "loss": 0.0141, "step": 166350 }, { "epoch": 106.02931803696622, "grad_norm": 0.25034889578819275, "learning_rate": 4.512506039665605e-05, "loss": 0.0078, "step": 166360 }, { "epoch": 106.03569152326322, "grad_norm": 0.33479043841362, "learning_rate": 4.5119575119292894e-05, "loss": 0.0081, "step": 166370 }, { "epoch": 106.04206500956023, "grad_norm": 0.18534307181835175, "learning_rate": 4.511408990123149e-05, "loss": 0.0043, "step": 166380 }, { "epoch": 106.04843849585724, "grad_norm": 0.23896019160747528, "learning_rate": 4.5108604742538505e-05, "loss": 0.0121, "step": 166390 }, { "epoch": 106.05481198215423, "grad_norm": 0.5372748970985413, "learning_rate": 4.5103119643280605e-05, "loss": 0.0086, "step": 166400 }, { "epoch": 106.06118546845124, "grad_norm": 0.23789270222187042, "learning_rate": 4.5097634603524425e-05, "loss": 0.0109, "step": 166410 }, { "epoch": 106.06755895474825, "grad_norm": 0.2284580022096634, "learning_rate": 4.5092149623336596e-05, "loss": 0.0099, "step": 166420 }, { "epoch": 106.07393244104526, "grad_norm": 0.3142921030521393, "learning_rate": 4.508666470278379e-05, "loss": 0.0172, "step": 166430 }, { "epoch": 106.08030592734225, "grad_norm": 0.2592828869819641, "learning_rate": 4.508117984193267e-05, "loss": 0.0074, "step": 166440 }, { "epoch": 106.08667941363926, "grad_norm": 0.3298291862010956, "learning_rate": 4.507569504084983e-05, "loss": 0.0089, "step": 166450 }, { "epoch": 106.09305289993627, "grad_norm": 0.24507883191108704, "learning_rate": 4.507021029960195e-05, "loss": 0.0103, "step": 166460 }, { "epoch": 106.09942638623328, "grad_norm": 0.196636363863945, "learning_rate": 4.506472561825567e-05, "loss": 0.0097, "step": 166470 }, { "epoch": 106.10579987253027, "grad_norm": 0.19393733143806458, "learning_rate": 4.505924099687765e-05, "loss": 0.0153, "step": 166480 }, { "epoch": 106.11217335882728, "grad_norm": 0.4515109062194824, "learning_rate": 4.50537564355345e-05, "loss": 0.0195, "step": 166490 }, { "epoch": 106.11854684512429, "grad_norm": 0.23030348122119904, "learning_rate": 4.504827193429287e-05, "loss": 0.0111, "step": 166500 }, { "epoch": 106.1249203314213, "grad_norm": 0.23016777634620667, "learning_rate": 4.5042787493219445e-05, "loss": 0.0156, "step": 166510 }, { "epoch": 106.13129381771829, "grad_norm": 0.3052428066730499, "learning_rate": 4.50373031123808e-05, "loss": 0.0192, "step": 166520 }, { "epoch": 106.1376673040153, "grad_norm": 0.42862796783447266, "learning_rate": 4.5031818791843615e-05, "loss": 0.0099, "step": 166530 }, { "epoch": 106.1440407903123, "grad_norm": 0.1738714575767517, "learning_rate": 4.502633453167452e-05, "loss": 0.008, "step": 166540 }, { "epoch": 106.15041427660931, "grad_norm": 0.22859561443328857, "learning_rate": 4.502085033194018e-05, "loss": 0.0065, "step": 166550 }, { "epoch": 106.1567877629063, "grad_norm": 0.2447023093700409, "learning_rate": 4.501536619270719e-05, "loss": 0.0101, "step": 166560 }, { "epoch": 106.16316124920331, "grad_norm": 0.23319442570209503, "learning_rate": 4.50098821140422e-05, "loss": 0.0127, "step": 166570 }, { "epoch": 106.16953473550032, "grad_norm": 0.6549547910690308, "learning_rate": 4.500439809601188e-05, "loss": 0.0171, "step": 166580 }, { "epoch": 106.17590822179733, "grad_norm": 0.4816547930240631, "learning_rate": 4.499891413868282e-05, "loss": 0.013, "step": 166590 }, { "epoch": 106.18228170809432, "grad_norm": 0.18699151277542114, "learning_rate": 4.499343024212167e-05, "loss": 0.0068, "step": 166600 }, { "epoch": 106.18865519439133, "grad_norm": 0.5287861227989197, "learning_rate": 4.4987946406395074e-05, "loss": 0.02, "step": 166610 }, { "epoch": 106.19502868068834, "grad_norm": 0.450363427400589, "learning_rate": 4.4982462631569674e-05, "loss": 0.0117, "step": 166620 }, { "epoch": 106.20140216698535, "grad_norm": 0.39426571130752563, "learning_rate": 4.497697891771208e-05, "loss": 0.0097, "step": 166630 }, { "epoch": 106.20777565328234, "grad_norm": 0.19618472456932068, "learning_rate": 4.497149526488893e-05, "loss": 0.0106, "step": 166640 }, { "epoch": 106.21414913957935, "grad_norm": 0.3545568883419037, "learning_rate": 4.4966011673166864e-05, "loss": 0.0097, "step": 166650 }, { "epoch": 106.22052262587636, "grad_norm": 0.32601475715637207, "learning_rate": 4.4960528142612526e-05, "loss": 0.0192, "step": 166660 }, { "epoch": 106.22689611217336, "grad_norm": 0.3436446785926819, "learning_rate": 4.495504467329251e-05, "loss": 0.0084, "step": 166670 }, { "epoch": 106.23326959847036, "grad_norm": 0.20044559240341187, "learning_rate": 4.494956126527346e-05, "loss": 0.0081, "step": 166680 }, { "epoch": 106.23964308476737, "grad_norm": 0.6896430253982544, "learning_rate": 4.494407791862204e-05, "loss": 0.0103, "step": 166690 }, { "epoch": 106.24601657106437, "grad_norm": 0.28889375925064087, "learning_rate": 4.4938594633404815e-05, "loss": 0.006, "step": 166700 }, { "epoch": 106.25239005736138, "grad_norm": 0.22264464199543, "learning_rate": 4.4933111409688446e-05, "loss": 0.0109, "step": 166710 }, { "epoch": 106.25876354365838, "grad_norm": 0.4796260595321655, "learning_rate": 4.4927628247539554e-05, "loss": 0.0088, "step": 166720 }, { "epoch": 106.26513702995538, "grad_norm": 0.22293870151042938, "learning_rate": 4.49221451470248e-05, "loss": 0.0082, "step": 166730 }, { "epoch": 106.27151051625239, "grad_norm": 0.19948141276836395, "learning_rate": 4.491666210821075e-05, "loss": 0.0125, "step": 166740 }, { "epoch": 106.2778840025494, "grad_norm": 0.39523401856422424, "learning_rate": 4.491117913116405e-05, "loss": 0.0129, "step": 166750 }, { "epoch": 106.2842574888464, "grad_norm": 0.18973372876644135, "learning_rate": 4.490569621595136e-05, "loss": 0.0085, "step": 166760 }, { "epoch": 106.2906309751434, "grad_norm": 0.29409515857696533, "learning_rate": 4.490021336263924e-05, "loss": 0.0182, "step": 166770 }, { "epoch": 106.29700446144041, "grad_norm": 0.38938915729522705, "learning_rate": 4.4894730571294345e-05, "loss": 0.0112, "step": 166780 }, { "epoch": 106.30337794773742, "grad_norm": 0.34609100222587585, "learning_rate": 4.4889247841983295e-05, "loss": 0.0137, "step": 166790 }, { "epoch": 106.30975143403441, "grad_norm": 0.390376478433609, "learning_rate": 4.488376517477273e-05, "loss": 0.0189, "step": 166800 }, { "epoch": 106.31612492033142, "grad_norm": 0.4284866750240326, "learning_rate": 4.487828256972923e-05, "loss": 0.0057, "step": 166810 }, { "epoch": 106.32249840662843, "grad_norm": 0.3541337251663208, "learning_rate": 4.4872800026919434e-05, "loss": 0.0086, "step": 166820 }, { "epoch": 106.32887189292543, "grad_norm": 0.26559585332870483, "learning_rate": 4.486731754640997e-05, "loss": 0.0093, "step": 166830 }, { "epoch": 106.33524537922243, "grad_norm": 0.3511841297149658, "learning_rate": 4.486183512826743e-05, "loss": 0.0084, "step": 166840 }, { "epoch": 106.34161886551944, "grad_norm": 0.24728809297084808, "learning_rate": 4.4856352772558444e-05, "loss": 0.0137, "step": 166850 }, { "epoch": 106.34799235181644, "grad_norm": 0.2493736892938614, "learning_rate": 4.4850870479349625e-05, "loss": 0.0105, "step": 166860 }, { "epoch": 106.35436583811345, "grad_norm": 0.2585674822330475, "learning_rate": 4.484538824870762e-05, "loss": 0.0094, "step": 166870 }, { "epoch": 106.36073932441045, "grad_norm": 0.7416593432426453, "learning_rate": 4.4839906080698984e-05, "loss": 0.0106, "step": 166880 }, { "epoch": 106.36711281070745, "grad_norm": 0.34846919775009155, "learning_rate": 4.4834423975390364e-05, "loss": 0.0107, "step": 166890 }, { "epoch": 106.37348629700446, "grad_norm": 0.28626659512519836, "learning_rate": 4.4828941932848375e-05, "loss": 0.0128, "step": 166900 }, { "epoch": 106.37985978330147, "grad_norm": 0.33101746439933777, "learning_rate": 4.482345995313964e-05, "loss": 0.0131, "step": 166910 }, { "epoch": 106.38623326959846, "grad_norm": 0.41533342003822327, "learning_rate": 4.481797803633073e-05, "loss": 0.016, "step": 166920 }, { "epoch": 106.39260675589547, "grad_norm": 0.15967024862766266, "learning_rate": 4.481249618248828e-05, "loss": 0.0185, "step": 166930 }, { "epoch": 106.39898024219248, "grad_norm": 0.2933744490146637, "learning_rate": 4.480701439167892e-05, "loss": 0.0115, "step": 166940 }, { "epoch": 106.40535372848949, "grad_norm": 0.15521451830863953, "learning_rate": 4.4801532663969234e-05, "loss": 0.011, "step": 166950 }, { "epoch": 106.4117272147865, "grad_norm": 0.33595025539398193, "learning_rate": 4.479605099942582e-05, "loss": 0.0083, "step": 166960 }, { "epoch": 106.41810070108349, "grad_norm": 0.24172012507915497, "learning_rate": 4.4790569398115304e-05, "loss": 0.0182, "step": 166970 }, { "epoch": 106.4244741873805, "grad_norm": 0.3407798707485199, "learning_rate": 4.478508786010429e-05, "loss": 0.0104, "step": 166980 }, { "epoch": 106.4308476736775, "grad_norm": 0.19785688817501068, "learning_rate": 4.477960638545939e-05, "loss": 0.0125, "step": 166990 }, { "epoch": 106.43722115997451, "grad_norm": 0.35379427671432495, "learning_rate": 4.477412497424719e-05, "loss": 0.0065, "step": 167000 }, { "epoch": 106.4435946462715, "grad_norm": 0.1793152093887329, "learning_rate": 4.4768643626534314e-05, "loss": 0.012, "step": 167010 }, { "epoch": 106.44996813256851, "grad_norm": 0.26411184668540955, "learning_rate": 4.476316234238734e-05, "loss": 0.0095, "step": 167020 }, { "epoch": 106.45634161886552, "grad_norm": 0.2677430212497711, "learning_rate": 4.475768112187291e-05, "loss": 0.0116, "step": 167030 }, { "epoch": 106.46271510516253, "grad_norm": 0.15941005945205688, "learning_rate": 4.475219996505759e-05, "loss": 0.0117, "step": 167040 }, { "epoch": 106.46908859145952, "grad_norm": 0.411973774433136, "learning_rate": 4.4746718872008004e-05, "loss": 0.0121, "step": 167050 }, { "epoch": 106.47546207775653, "grad_norm": 0.2478920966386795, "learning_rate": 4.4741237842790734e-05, "loss": 0.0116, "step": 167060 }, { "epoch": 106.48183556405354, "grad_norm": 0.3998870849609375, "learning_rate": 4.4735756877472403e-05, "loss": 0.0066, "step": 167070 }, { "epoch": 106.48820905035055, "grad_norm": 0.4296642243862152, "learning_rate": 4.473027597611959e-05, "loss": 0.0092, "step": 167080 }, { "epoch": 106.49458253664754, "grad_norm": 0.417406290769577, "learning_rate": 4.4724795138798893e-05, "loss": 0.0134, "step": 167090 }, { "epoch": 106.50095602294455, "grad_norm": 0.42621976137161255, "learning_rate": 4.4719314365576915e-05, "loss": 0.0174, "step": 167100 }, { "epoch": 106.50732950924156, "grad_norm": 0.21771240234375, "learning_rate": 4.471383365652026e-05, "loss": 0.0078, "step": 167110 }, { "epoch": 106.51370299553857, "grad_norm": 0.23992468416690826, "learning_rate": 4.4708353011695524e-05, "loss": 0.0068, "step": 167120 }, { "epoch": 106.52007648183556, "grad_norm": 0.3137378692626953, "learning_rate": 4.470287243116928e-05, "loss": 0.0126, "step": 167130 }, { "epoch": 106.52644996813257, "grad_norm": 0.34193524718284607, "learning_rate": 4.469739191500813e-05, "loss": 0.0103, "step": 167140 }, { "epoch": 106.53282345442958, "grad_norm": 0.49141815304756165, "learning_rate": 4.4691911463278716e-05, "loss": 0.0133, "step": 167150 }, { "epoch": 106.53919694072658, "grad_norm": 0.38300827145576477, "learning_rate": 4.468643107604755e-05, "loss": 0.0166, "step": 167160 }, { "epoch": 106.54557042702358, "grad_norm": 0.16822749376296997, "learning_rate": 4.468095075338127e-05, "loss": 0.0115, "step": 167170 }, { "epoch": 106.55194391332059, "grad_norm": 0.3027929961681366, "learning_rate": 4.467547049534646e-05, "loss": 0.0106, "step": 167180 }, { "epoch": 106.5583173996176, "grad_norm": 0.4251919984817505, "learning_rate": 4.4669990302009736e-05, "loss": 0.0119, "step": 167190 }, { "epoch": 106.5646908859146, "grad_norm": 0.4137505888938904, "learning_rate": 4.466451017343763e-05, "loss": 0.0113, "step": 167200 }, { "epoch": 106.5710643722116, "grad_norm": 0.42540422081947327, "learning_rate": 4.465903010969677e-05, "loss": 0.0166, "step": 167210 }, { "epoch": 106.5774378585086, "grad_norm": 0.20113803446292877, "learning_rate": 4.4653550110853735e-05, "loss": 0.0094, "step": 167220 }, { "epoch": 106.58381134480561, "grad_norm": 0.3649429976940155, "learning_rate": 4.464807017697514e-05, "loss": 0.0101, "step": 167230 }, { "epoch": 106.59018483110262, "grad_norm": 0.43077871203422546, "learning_rate": 4.464259030812752e-05, "loss": 0.0106, "step": 167240 }, { "epoch": 106.59655831739961, "grad_norm": 0.2757472097873688, "learning_rate": 4.4637110504377485e-05, "loss": 0.0122, "step": 167250 }, { "epoch": 106.60293180369662, "grad_norm": 0.2103421539068222, "learning_rate": 4.463163076579165e-05, "loss": 0.0099, "step": 167260 }, { "epoch": 106.60930528999363, "grad_norm": 0.21765978634357452, "learning_rate": 4.462615109243654e-05, "loss": 0.0173, "step": 167270 }, { "epoch": 106.61567877629064, "grad_norm": 0.40600234270095825, "learning_rate": 4.462067148437877e-05, "loss": 0.0126, "step": 167280 }, { "epoch": 106.62205226258763, "grad_norm": 0.2591892182826996, "learning_rate": 4.4615191941684916e-05, "loss": 0.0133, "step": 167290 }, { "epoch": 106.62842574888464, "grad_norm": 0.22667638957500458, "learning_rate": 4.460971246442159e-05, "loss": 0.0089, "step": 167300 }, { "epoch": 106.63479923518165, "grad_norm": 0.7190356254577637, "learning_rate": 4.460423305265533e-05, "loss": 0.0132, "step": 167310 }, { "epoch": 106.64117272147865, "grad_norm": 0.7051494121551514, "learning_rate": 4.459875370645273e-05, "loss": 0.0111, "step": 167320 }, { "epoch": 106.64754620777565, "grad_norm": 0.46592023968696594, "learning_rate": 4.4593274425880395e-05, "loss": 0.012, "step": 167330 }, { "epoch": 106.65391969407266, "grad_norm": 0.2580007016658783, "learning_rate": 4.458779521100485e-05, "loss": 0.0277, "step": 167340 }, { "epoch": 106.66029318036966, "grad_norm": 0.5094923973083496, "learning_rate": 4.458231606189271e-05, "loss": 0.0083, "step": 167350 }, { "epoch": 106.66666666666667, "grad_norm": 0.33632123470306396, "learning_rate": 4.4576836978610546e-05, "loss": 0.0171, "step": 167360 }, { "epoch": 106.67304015296367, "grad_norm": 0.1763758361339569, "learning_rate": 4.457135796122495e-05, "loss": 0.0081, "step": 167370 }, { "epoch": 106.67941363926067, "grad_norm": 0.3989444971084595, "learning_rate": 4.4565879009802464e-05, "loss": 0.0129, "step": 167380 }, { "epoch": 106.68578712555768, "grad_norm": 0.18257509171962738, "learning_rate": 4.456040012440967e-05, "loss": 0.0101, "step": 167390 }, { "epoch": 106.69216061185469, "grad_norm": 0.17717529833316803, "learning_rate": 4.455492130511319e-05, "loss": 0.0098, "step": 167400 }, { "epoch": 106.69853409815168, "grad_norm": 0.3526151776313782, "learning_rate": 4.454944255197952e-05, "loss": 0.0094, "step": 167410 }, { "epoch": 106.70490758444869, "grad_norm": 0.19713515043258667, "learning_rate": 4.4543963865075275e-05, "loss": 0.0084, "step": 167420 }, { "epoch": 106.7112810707457, "grad_norm": 0.35611751675605774, "learning_rate": 4.453848524446702e-05, "loss": 0.0101, "step": 167430 }, { "epoch": 106.7176545570427, "grad_norm": 0.24356567859649658, "learning_rate": 4.453300669022134e-05, "loss": 0.0217, "step": 167440 }, { "epoch": 106.7240280433397, "grad_norm": 0.2159564048051834, "learning_rate": 4.452752820240477e-05, "loss": 0.012, "step": 167450 }, { "epoch": 106.73040152963671, "grad_norm": 0.23949627578258514, "learning_rate": 4.452204978108391e-05, "loss": 0.0158, "step": 167460 }, { "epoch": 106.73677501593372, "grad_norm": 0.4460504353046417, "learning_rate": 4.451657142632533e-05, "loss": 0.0135, "step": 167470 }, { "epoch": 106.74314850223072, "grad_norm": 0.18424226343631744, "learning_rate": 4.451109313819556e-05, "loss": 0.0081, "step": 167480 }, { "epoch": 106.74952198852772, "grad_norm": 0.34652596712112427, "learning_rate": 4.45056149167612e-05, "loss": 0.0163, "step": 167490 }, { "epoch": 106.75589547482473, "grad_norm": 0.33046039938926697, "learning_rate": 4.45001367620888e-05, "loss": 0.0241, "step": 167500 }, { "epoch": 106.76226896112173, "grad_norm": 0.3314240872859955, "learning_rate": 4.449465867424496e-05, "loss": 0.0081, "step": 167510 }, { "epoch": 106.76864244741874, "grad_norm": 0.7736092805862427, "learning_rate": 4.448918065329619e-05, "loss": 0.0133, "step": 167520 }, { "epoch": 106.77501593371574, "grad_norm": 0.45830294489860535, "learning_rate": 4.4483702699309074e-05, "loss": 0.013, "step": 167530 }, { "epoch": 106.78138942001274, "grad_norm": 0.20568916201591492, "learning_rate": 4.4478224812350176e-05, "loss": 0.0083, "step": 167540 }, { "epoch": 106.78776290630975, "grad_norm": 0.48441657423973083, "learning_rate": 4.447274699248609e-05, "loss": 0.012, "step": 167550 }, { "epoch": 106.79413639260676, "grad_norm": 0.25239691138267517, "learning_rate": 4.446726923978332e-05, "loss": 0.0084, "step": 167560 }, { "epoch": 106.80050987890377, "grad_norm": 0.3024815320968628, "learning_rate": 4.4461791554308454e-05, "loss": 0.0101, "step": 167570 }, { "epoch": 106.80688336520076, "grad_norm": 0.24919156730175018, "learning_rate": 4.445631393612807e-05, "loss": 0.0208, "step": 167580 }, { "epoch": 106.81325685149777, "grad_norm": 0.44242337346076965, "learning_rate": 4.445083638530869e-05, "loss": 0.013, "step": 167590 }, { "epoch": 106.81963033779478, "grad_norm": 0.4889225661754608, "learning_rate": 4.4445358901916886e-05, "loss": 0.0137, "step": 167600 }, { "epoch": 106.82600382409179, "grad_norm": 0.674306333065033, "learning_rate": 4.443988148601921e-05, "loss": 0.0253, "step": 167610 }, { "epoch": 106.83237731038878, "grad_norm": 0.1818370521068573, "learning_rate": 4.4434404137682245e-05, "loss": 0.007, "step": 167620 }, { "epoch": 106.83875079668579, "grad_norm": 0.22868144512176514, "learning_rate": 4.442892685697251e-05, "loss": 0.0153, "step": 167630 }, { "epoch": 106.8451242829828, "grad_norm": 0.3443567454814911, "learning_rate": 4.442344964395657e-05, "loss": 0.007, "step": 167640 }, { "epoch": 106.8514977692798, "grad_norm": 0.42343518137931824, "learning_rate": 4.4417972498700986e-05, "loss": 0.008, "step": 167650 }, { "epoch": 106.8578712555768, "grad_norm": 0.20259499549865723, "learning_rate": 4.441249542127231e-05, "loss": 0.01, "step": 167660 }, { "epoch": 106.8642447418738, "grad_norm": 0.22341018915176392, "learning_rate": 4.4407018411737084e-05, "loss": 0.0143, "step": 167670 }, { "epoch": 106.87061822817081, "grad_norm": 0.20292797684669495, "learning_rate": 4.4401541470161864e-05, "loss": 0.0134, "step": 167680 }, { "epoch": 106.87699171446782, "grad_norm": 0.2521200180053711, "learning_rate": 4.439606459661321e-05, "loss": 0.0106, "step": 167690 }, { "epoch": 106.88336520076481, "grad_norm": 0.33086097240448, "learning_rate": 4.439058779115766e-05, "loss": 0.0191, "step": 167700 }, { "epoch": 106.88973868706182, "grad_norm": 0.31251874566078186, "learning_rate": 4.438511105386176e-05, "loss": 0.0132, "step": 167710 }, { "epoch": 106.89611217335883, "grad_norm": 0.4000931680202484, "learning_rate": 4.437963438479207e-05, "loss": 0.0175, "step": 167720 }, { "epoch": 106.90248565965584, "grad_norm": 0.2689957022666931, "learning_rate": 4.4374157784015114e-05, "loss": 0.0248, "step": 167730 }, { "epoch": 106.90885914595283, "grad_norm": 0.21612469851970673, "learning_rate": 4.4368681251597465e-05, "loss": 0.0153, "step": 167740 }, { "epoch": 106.91523263224984, "grad_norm": 0.22122327983379364, "learning_rate": 4.436320478760565e-05, "loss": 0.0167, "step": 167750 }, { "epoch": 106.92160611854685, "grad_norm": 0.2540423572063446, "learning_rate": 4.435772839210622e-05, "loss": 0.0118, "step": 167760 }, { "epoch": 106.92797960484386, "grad_norm": 0.40485936403274536, "learning_rate": 4.4352252065165715e-05, "loss": 0.0099, "step": 167770 }, { "epoch": 106.93435309114085, "grad_norm": 0.5184552669525146, "learning_rate": 4.4346775806850685e-05, "loss": 0.0208, "step": 167780 }, { "epoch": 106.94072657743786, "grad_norm": 0.16791628301143646, "learning_rate": 4.434129961722767e-05, "loss": 0.0076, "step": 167790 }, { "epoch": 106.94710006373487, "grad_norm": 0.4175131320953369, "learning_rate": 4.43358234963632e-05, "loss": 0.0127, "step": 167800 }, { "epoch": 106.95347355003187, "grad_norm": 0.41521573066711426, "learning_rate": 4.433034744432382e-05, "loss": 0.0136, "step": 167810 }, { "epoch": 106.95984703632887, "grad_norm": 0.31920406222343445, "learning_rate": 4.4324871461176074e-05, "loss": 0.0129, "step": 167820 }, { "epoch": 106.96622052262587, "grad_norm": 0.41104865074157715, "learning_rate": 4.431939554698651e-05, "loss": 0.0176, "step": 167830 }, { "epoch": 106.97259400892288, "grad_norm": 0.3219737410545349, "learning_rate": 4.431391970182164e-05, "loss": 0.0138, "step": 167840 }, { "epoch": 106.97896749521989, "grad_norm": 0.4090932011604309, "learning_rate": 4.4308443925748005e-05, "loss": 0.0164, "step": 167850 }, { "epoch": 106.98534098151688, "grad_norm": 0.2128528505563736, "learning_rate": 4.4302968218832166e-05, "loss": 0.0106, "step": 167860 }, { "epoch": 106.99171446781389, "grad_norm": 0.22206652164459229, "learning_rate": 4.429749258114065e-05, "loss": 0.0139, "step": 167870 }, { "epoch": 106.9980879541109, "grad_norm": 0.17705927789211273, "learning_rate": 4.429201701273996e-05, "loss": 0.0091, "step": 167880 }, { "epoch": 107.00446144040791, "grad_norm": 0.541948676109314, "learning_rate": 4.428654151369666e-05, "loss": 0.0162, "step": 167890 }, { "epoch": 107.0108349267049, "grad_norm": 0.1894221007823944, "learning_rate": 4.4281066084077296e-05, "loss": 0.0142, "step": 167900 }, { "epoch": 107.01720841300191, "grad_norm": 0.2504042387008667, "learning_rate": 4.427559072394836e-05, "loss": 0.0091, "step": 167910 }, { "epoch": 107.02358189929892, "grad_norm": 0.1910182535648346, "learning_rate": 4.4270115433376394e-05, "loss": 0.0086, "step": 167920 }, { "epoch": 107.02995538559593, "grad_norm": 0.46944186091423035, "learning_rate": 4.426464021242794e-05, "loss": 0.0169, "step": 167930 }, { "epoch": 107.03632887189292, "grad_norm": 0.3773885667324066, "learning_rate": 4.425916506116954e-05, "loss": 0.0085, "step": 167940 }, { "epoch": 107.04270235818993, "grad_norm": 0.1787872165441513, "learning_rate": 4.4253689979667684e-05, "loss": 0.0081, "step": 167950 }, { "epoch": 107.04907584448694, "grad_norm": 0.3650685250759125, "learning_rate": 4.424821496798892e-05, "loss": 0.0132, "step": 167960 }, { "epoch": 107.05544933078394, "grad_norm": 0.40530431270599365, "learning_rate": 4.4242740026199805e-05, "loss": 0.0144, "step": 167970 }, { "epoch": 107.06182281708094, "grad_norm": 0.5342543125152588, "learning_rate": 4.42372651543668e-05, "loss": 0.0119, "step": 167980 }, { "epoch": 107.06819630337795, "grad_norm": 0.26608729362487793, "learning_rate": 4.4231790352556476e-05, "loss": 0.007, "step": 167990 }, { "epoch": 107.07456978967495, "grad_norm": 0.1591409593820572, "learning_rate": 4.4226315620835334e-05, "loss": 0.012, "step": 168000 }, { "epoch": 107.08094327597196, "grad_norm": 0.20532108843326569, "learning_rate": 4.422084095926994e-05, "loss": 0.012, "step": 168010 }, { "epoch": 107.08731676226896, "grad_norm": 0.2948269248008728, "learning_rate": 4.4215366367926756e-05, "loss": 0.0069, "step": 168020 }, { "epoch": 107.09369024856596, "grad_norm": 0.14121031761169434, "learning_rate": 4.420989184687234e-05, "loss": 0.011, "step": 168030 }, { "epoch": 107.10006373486297, "grad_norm": 0.18527691066265106, "learning_rate": 4.420441739617322e-05, "loss": 0.0144, "step": 168040 }, { "epoch": 107.10643722115998, "grad_norm": 0.21744918823242188, "learning_rate": 4.419894301589588e-05, "loss": 0.0096, "step": 168050 }, { "epoch": 107.11281070745697, "grad_norm": 0.23117676377296448, "learning_rate": 4.419346870610686e-05, "loss": 0.01, "step": 168060 }, { "epoch": 107.11918419375398, "grad_norm": 0.45043131709098816, "learning_rate": 4.4187994466872676e-05, "loss": 0.0086, "step": 168070 }, { "epoch": 107.12555768005099, "grad_norm": 0.23421920835971832, "learning_rate": 4.418252029825987e-05, "loss": 0.0168, "step": 168080 }, { "epoch": 107.131931166348, "grad_norm": 0.26247796416282654, "learning_rate": 4.417704620033491e-05, "loss": 0.0145, "step": 168090 }, { "epoch": 107.13830465264499, "grad_norm": 0.2640664279460907, "learning_rate": 4.4171572173164345e-05, "loss": 0.0102, "step": 168100 }, { "epoch": 107.144678138942, "grad_norm": 0.46251633763313293, "learning_rate": 4.4166098216814676e-05, "loss": 0.0118, "step": 168110 }, { "epoch": 107.151051625239, "grad_norm": 0.23944881558418274, "learning_rate": 4.416062433135244e-05, "loss": 0.0119, "step": 168120 }, { "epoch": 107.15742511153601, "grad_norm": 0.4349523186683655, "learning_rate": 4.415515051684412e-05, "loss": 0.0094, "step": 168130 }, { "epoch": 107.16379859783301, "grad_norm": 0.5165630578994751, "learning_rate": 4.414967677335623e-05, "loss": 0.0151, "step": 168140 }, { "epoch": 107.17017208413002, "grad_norm": 0.3525736331939697, "learning_rate": 4.4144203100955314e-05, "loss": 0.0141, "step": 168150 }, { "epoch": 107.17654557042702, "grad_norm": 0.190159872174263, "learning_rate": 4.4138729499707844e-05, "loss": 0.0096, "step": 168160 }, { "epoch": 107.18291905672403, "grad_norm": 0.274755597114563, "learning_rate": 4.413325596968034e-05, "loss": 0.0108, "step": 168170 }, { "epoch": 107.18929254302103, "grad_norm": 0.4033345580101013, "learning_rate": 4.4127782510939305e-05, "loss": 0.0137, "step": 168180 }, { "epoch": 107.19566602931803, "grad_norm": 0.5576061606407166, "learning_rate": 4.412230912355129e-05, "loss": 0.021, "step": 168190 }, { "epoch": 107.20203951561504, "grad_norm": 0.4414364695549011, "learning_rate": 4.411683580758275e-05, "loss": 0.01, "step": 168200 }, { "epoch": 107.20841300191205, "grad_norm": 0.3954830467700958, "learning_rate": 4.41113625631002e-05, "loss": 0.0092, "step": 168210 }, { "epoch": 107.21478648820906, "grad_norm": 0.3221707046031952, "learning_rate": 4.4105889390170186e-05, "loss": 0.0105, "step": 168220 }, { "epoch": 107.22115997450605, "grad_norm": 0.6310611367225647, "learning_rate": 4.410041628885915e-05, "loss": 0.017, "step": 168230 }, { "epoch": 107.22753346080306, "grad_norm": 0.2606613337993622, "learning_rate": 4.4094943259233635e-05, "loss": 0.0107, "step": 168240 }, { "epoch": 107.23390694710007, "grad_norm": 0.25502848625183105, "learning_rate": 4.408947030136012e-05, "loss": 0.0084, "step": 168250 }, { "epoch": 107.24028043339707, "grad_norm": 0.22630298137664795, "learning_rate": 4.408399741530516e-05, "loss": 0.0131, "step": 168260 }, { "epoch": 107.24665391969407, "grad_norm": 0.36780521273612976, "learning_rate": 4.4078524601135184e-05, "loss": 0.0155, "step": 168270 }, { "epoch": 107.25302740599108, "grad_norm": 0.26027587056159973, "learning_rate": 4.4073051858916734e-05, "loss": 0.0095, "step": 168280 }, { "epoch": 107.25940089228808, "grad_norm": 0.46419423818588257, "learning_rate": 4.406757918871631e-05, "loss": 0.0157, "step": 168290 }, { "epoch": 107.26577437858509, "grad_norm": 0.47098323702812195, "learning_rate": 4.4062106590600386e-05, "loss": 0.021, "step": 168300 }, { "epoch": 107.27214786488209, "grad_norm": 0.45686835050582886, "learning_rate": 4.405663406463548e-05, "loss": 0.0139, "step": 168310 }, { "epoch": 107.2785213511791, "grad_norm": 0.6317861080169678, "learning_rate": 4.405116161088807e-05, "loss": 0.0129, "step": 168320 }, { "epoch": 107.2848948374761, "grad_norm": 0.2511119544506073, "learning_rate": 4.404568922942469e-05, "loss": 0.0086, "step": 168330 }, { "epoch": 107.29126832377311, "grad_norm": 0.5460946559906006, "learning_rate": 4.404021692031178e-05, "loss": 0.0134, "step": 168340 }, { "epoch": 107.2976418100701, "grad_norm": 0.41660410165786743, "learning_rate": 4.403474468361587e-05, "loss": 0.0121, "step": 168350 }, { "epoch": 107.30401529636711, "grad_norm": 0.17875921726226807, "learning_rate": 4.4029272519403455e-05, "loss": 0.0107, "step": 168360 }, { "epoch": 107.31038878266412, "grad_norm": 0.2599368095397949, "learning_rate": 4.402380042774101e-05, "loss": 0.0127, "step": 168370 }, { "epoch": 107.31676226896113, "grad_norm": 0.4019927978515625, "learning_rate": 4.401832840869502e-05, "loss": 0.0105, "step": 168380 }, { "epoch": 107.32313575525812, "grad_norm": 0.33844152092933655, "learning_rate": 4.401285646233199e-05, "loss": 0.0101, "step": 168390 }, { "epoch": 107.32950924155513, "grad_norm": 0.4590342938899994, "learning_rate": 4.400738458871842e-05, "loss": 0.0116, "step": 168400 }, { "epoch": 107.33588272785214, "grad_norm": 0.15378929674625397, "learning_rate": 4.4001912787920784e-05, "loss": 0.0069, "step": 168410 }, { "epoch": 107.34225621414915, "grad_norm": 0.27457475662231445, "learning_rate": 4.399644106000556e-05, "loss": 0.0092, "step": 168420 }, { "epoch": 107.34862970044614, "grad_norm": 0.1798926740884781, "learning_rate": 4.3990969405039244e-05, "loss": 0.0127, "step": 168430 }, { "epoch": 107.35500318674315, "grad_norm": 0.42825260758399963, "learning_rate": 4.398549782308834e-05, "loss": 0.011, "step": 168440 }, { "epoch": 107.36137667304016, "grad_norm": 0.26509198546409607, "learning_rate": 4.3980026314219306e-05, "loss": 0.017, "step": 168450 }, { "epoch": 107.36775015933716, "grad_norm": 0.25537216663360596, "learning_rate": 4.3974554878498634e-05, "loss": 0.0121, "step": 168460 }, { "epoch": 107.37412364563416, "grad_norm": 0.2869643270969391, "learning_rate": 4.396908351599281e-05, "loss": 0.0109, "step": 168470 }, { "epoch": 107.38049713193116, "grad_norm": 0.16179266571998596, "learning_rate": 4.396361222676832e-05, "loss": 0.0067, "step": 168480 }, { "epoch": 107.38687061822817, "grad_norm": 0.3926658630371094, "learning_rate": 4.395814101089164e-05, "loss": 0.0095, "step": 168490 }, { "epoch": 107.39324410452518, "grad_norm": 0.17013344168663025, "learning_rate": 4.395266986842924e-05, "loss": 0.0073, "step": 168500 }, { "epoch": 107.39961759082217, "grad_norm": 0.26261672377586365, "learning_rate": 4.3947198799447625e-05, "loss": 0.0227, "step": 168510 }, { "epoch": 107.40599107711918, "grad_norm": 0.25777313113212585, "learning_rate": 4.394172780401325e-05, "loss": 0.0143, "step": 168520 }, { "epoch": 107.41236456341619, "grad_norm": 0.2292126566171646, "learning_rate": 4.393625688219261e-05, "loss": 0.0118, "step": 168530 }, { "epoch": 107.4187380497132, "grad_norm": 0.32905644178390503, "learning_rate": 4.393078603405218e-05, "loss": 0.0085, "step": 168540 }, { "epoch": 107.42511153601019, "grad_norm": 0.18481527268886566, "learning_rate": 4.392531525965841e-05, "loss": 0.0075, "step": 168550 }, { "epoch": 107.4314850223072, "grad_norm": 0.19198228418827057, "learning_rate": 4.39198445590778e-05, "loss": 0.0063, "step": 168560 }, { "epoch": 107.43785850860421, "grad_norm": 0.2440151870250702, "learning_rate": 4.3914373932376816e-05, "loss": 0.0087, "step": 168570 }, { "epoch": 107.44423199490122, "grad_norm": 0.7701067924499512, "learning_rate": 4.390890337962196e-05, "loss": 0.0096, "step": 168580 }, { "epoch": 107.45060548119821, "grad_norm": 0.40999963879585266, "learning_rate": 4.3903432900879656e-05, "loss": 0.0175, "step": 168590 }, { "epoch": 107.45697896749522, "grad_norm": 0.16197143495082855, "learning_rate": 4.38979624962164e-05, "loss": 0.0088, "step": 168600 }, { "epoch": 107.46335245379223, "grad_norm": 0.4681512415409088, "learning_rate": 4.3892492165698684e-05, "loss": 0.0079, "step": 168610 }, { "epoch": 107.46972594008923, "grad_norm": 0.3463338613510132, "learning_rate": 4.388702190939293e-05, "loss": 0.0207, "step": 168620 }, { "epoch": 107.47609942638623, "grad_norm": 0.3508475124835968, "learning_rate": 4.3881551727365626e-05, "loss": 0.0163, "step": 168630 }, { "epoch": 107.48247291268324, "grad_norm": 0.31784185767173767, "learning_rate": 4.387608161968325e-05, "loss": 0.0099, "step": 168640 }, { "epoch": 107.48884639898024, "grad_norm": 0.45470115542411804, "learning_rate": 4.387061158641229e-05, "loss": 0.0102, "step": 168650 }, { "epoch": 107.49521988527725, "grad_norm": 0.31578347086906433, "learning_rate": 4.386514162761916e-05, "loss": 0.0134, "step": 168660 }, { "epoch": 107.50159337157424, "grad_norm": 0.24687528610229492, "learning_rate": 4.3859671743370355e-05, "loss": 0.006, "step": 168670 }, { "epoch": 107.50796685787125, "grad_norm": 0.185679093003273, "learning_rate": 4.3854201933732353e-05, "loss": 0.0083, "step": 168680 }, { "epoch": 107.51434034416826, "grad_norm": 0.21940624713897705, "learning_rate": 4.384873219877158e-05, "loss": 0.0118, "step": 168690 }, { "epoch": 107.52071383046527, "grad_norm": 0.22988533973693848, "learning_rate": 4.384326253855452e-05, "loss": 0.0112, "step": 168700 }, { "epoch": 107.52708731676226, "grad_norm": 0.27603474259376526, "learning_rate": 4.3837792953147636e-05, "loss": 0.0102, "step": 168710 }, { "epoch": 107.53346080305927, "grad_norm": 0.16200973093509674, "learning_rate": 4.38323234426174e-05, "loss": 0.0132, "step": 168720 }, { "epoch": 107.53983428935628, "grad_norm": 0.1570654660463333, "learning_rate": 4.382685400703024e-05, "loss": 0.007, "step": 168730 }, { "epoch": 107.54620777565329, "grad_norm": 0.5780215859413147, "learning_rate": 4.3821384646452635e-05, "loss": 0.0122, "step": 168740 }, { "epoch": 107.55258126195028, "grad_norm": 0.21443845331668854, "learning_rate": 4.381591536095103e-05, "loss": 0.0112, "step": 168750 }, { "epoch": 107.55895474824729, "grad_norm": 0.1632094532251358, "learning_rate": 4.381044615059192e-05, "loss": 0.0115, "step": 168760 }, { "epoch": 107.5653282345443, "grad_norm": 0.24409353733062744, "learning_rate": 4.380497701544172e-05, "loss": 0.009, "step": 168770 }, { "epoch": 107.5717017208413, "grad_norm": 0.5020456910133362, "learning_rate": 4.379950795556689e-05, "loss": 0.01, "step": 168780 }, { "epoch": 107.5780752071383, "grad_norm": 0.3329426646232605, "learning_rate": 4.3794038971033914e-05, "loss": 0.0129, "step": 168790 }, { "epoch": 107.5844486934353, "grad_norm": 0.2761944830417633, "learning_rate": 4.37885700619092e-05, "loss": 0.0105, "step": 168800 }, { "epoch": 107.59082217973231, "grad_norm": 0.24435168504714966, "learning_rate": 4.3783101228259225e-05, "loss": 0.0103, "step": 168810 }, { "epoch": 107.59719566602932, "grad_norm": 0.33291956782341003, "learning_rate": 4.377763247015044e-05, "loss": 0.009, "step": 168820 }, { "epoch": 107.60356915232632, "grad_norm": 0.2135736644268036, "learning_rate": 4.3772163787649324e-05, "loss": 0.0102, "step": 168830 }, { "epoch": 107.60994263862332, "grad_norm": 0.33483514189720154, "learning_rate": 4.376669518082227e-05, "loss": 0.0167, "step": 168840 }, { "epoch": 107.61631612492033, "grad_norm": 0.23057053983211517, "learning_rate": 4.376122664973575e-05, "loss": 0.0061, "step": 168850 }, { "epoch": 107.62268961121734, "grad_norm": 0.22614842653274536, "learning_rate": 4.375575819445625e-05, "loss": 0.0106, "step": 168860 }, { "epoch": 107.62906309751435, "grad_norm": 0.33006319403648376, "learning_rate": 4.375028981505015e-05, "loss": 0.0084, "step": 168870 }, { "epoch": 107.63543658381134, "grad_norm": 0.11983554810285568, "learning_rate": 4.3744821511583945e-05, "loss": 0.0178, "step": 168880 }, { "epoch": 107.64181007010835, "grad_norm": 0.33226528763771057, "learning_rate": 4.373935328412405e-05, "loss": 0.0126, "step": 168890 }, { "epoch": 107.64818355640536, "grad_norm": 0.382074773311615, "learning_rate": 4.373388513273695e-05, "loss": 0.0094, "step": 168900 }, { "epoch": 107.65455704270236, "grad_norm": 0.22729197144508362, "learning_rate": 4.372841705748905e-05, "loss": 0.0077, "step": 168910 }, { "epoch": 107.66093052899936, "grad_norm": 0.19280017912387848, "learning_rate": 4.3722949058446786e-05, "loss": 0.007, "step": 168920 }, { "epoch": 107.66730401529637, "grad_norm": 0.1817396730184555, "learning_rate": 4.371748113567665e-05, "loss": 0.0105, "step": 168930 }, { "epoch": 107.67367750159337, "grad_norm": 0.2667253017425537, "learning_rate": 4.371201328924502e-05, "loss": 0.0064, "step": 168940 }, { "epoch": 107.68005098789038, "grad_norm": 0.20557133853435516, "learning_rate": 4.370654551921838e-05, "loss": 0.0082, "step": 168950 }, { "epoch": 107.68642447418738, "grad_norm": 0.1444830596446991, "learning_rate": 4.370107782566314e-05, "loss": 0.0087, "step": 168960 }, { "epoch": 107.69279796048438, "grad_norm": 0.26510506868362427, "learning_rate": 4.369561020864578e-05, "loss": 0.0168, "step": 168970 }, { "epoch": 107.69917144678139, "grad_norm": 0.1604451984167099, "learning_rate": 4.3690142668232686e-05, "loss": 0.0172, "step": 168980 }, { "epoch": 107.7055449330784, "grad_norm": 0.4496043622493744, "learning_rate": 4.368467520449032e-05, "loss": 0.0135, "step": 168990 }, { "epoch": 107.7119184193754, "grad_norm": 0.22099965810775757, "learning_rate": 4.3679207817485124e-05, "loss": 0.0136, "step": 169000 }, { "epoch": 107.7182919056724, "grad_norm": 0.33396217226982117, "learning_rate": 4.36737405072835e-05, "loss": 0.0128, "step": 169010 }, { "epoch": 107.72466539196941, "grad_norm": 0.30377447605133057, "learning_rate": 4.366827327395191e-05, "loss": 0.0247, "step": 169020 }, { "epoch": 107.73103887826642, "grad_norm": 0.284218430519104, "learning_rate": 4.366280611755678e-05, "loss": 0.0099, "step": 169030 }, { "epoch": 107.73741236456341, "grad_norm": 0.6005522608757019, "learning_rate": 4.365733903816455e-05, "loss": 0.0146, "step": 169040 }, { "epoch": 107.74378585086042, "grad_norm": 0.20171748101711273, "learning_rate": 4.365187203584162e-05, "loss": 0.0085, "step": 169050 }, { "epoch": 107.75015933715743, "grad_norm": 0.2714349031448364, "learning_rate": 4.364640511065445e-05, "loss": 0.0076, "step": 169060 }, { "epoch": 107.75653282345444, "grad_norm": 0.37475326657295227, "learning_rate": 4.3640938262669445e-05, "loss": 0.0091, "step": 169070 }, { "epoch": 107.76290630975143, "grad_norm": 0.2824309170246124, "learning_rate": 4.363547149195307e-05, "loss": 0.0165, "step": 169080 }, { "epoch": 107.76927979604844, "grad_norm": 0.2836514413356781, "learning_rate": 4.36300047985717e-05, "loss": 0.0115, "step": 169090 }, { "epoch": 107.77565328234544, "grad_norm": 0.5598851442337036, "learning_rate": 4.362453818259179e-05, "loss": 0.0094, "step": 169100 }, { "epoch": 107.78202676864245, "grad_norm": 0.1470775157213211, "learning_rate": 4.3619071644079765e-05, "loss": 0.0111, "step": 169110 }, { "epoch": 107.78840025493945, "grad_norm": 0.2937297821044922, "learning_rate": 4.3613605183102054e-05, "loss": 0.0057, "step": 169120 }, { "epoch": 107.79477374123645, "grad_norm": 0.5934840440750122, "learning_rate": 4.360813879972505e-05, "loss": 0.0156, "step": 169130 }, { "epoch": 107.80114722753346, "grad_norm": 0.10643921047449112, "learning_rate": 4.36026724940152e-05, "loss": 0.0128, "step": 169140 }, { "epoch": 107.80752071383047, "grad_norm": 0.21196448802947998, "learning_rate": 4.3597206266038933e-05, "loss": 0.0118, "step": 169150 }, { "epoch": 107.81389420012746, "grad_norm": 0.2918565273284912, "learning_rate": 4.3591740115862636e-05, "loss": 0.0092, "step": 169160 }, { "epoch": 107.82026768642447, "grad_norm": 0.2131723165512085, "learning_rate": 4.358627404355276e-05, "loss": 0.0116, "step": 169170 }, { "epoch": 107.82664117272148, "grad_norm": 0.23555244505405426, "learning_rate": 4.3580808049175714e-05, "loss": 0.0188, "step": 169180 }, { "epoch": 107.83301465901849, "grad_norm": 0.3175053596496582, "learning_rate": 4.35753421327979e-05, "loss": 0.0085, "step": 169190 }, { "epoch": 107.83938814531548, "grad_norm": 0.26014643907546997, "learning_rate": 4.356987629448574e-05, "loss": 0.0132, "step": 169200 }, { "epoch": 107.84576163161249, "grad_norm": 0.3561245799064636, "learning_rate": 4.3564410534305666e-05, "loss": 0.0132, "step": 169210 }, { "epoch": 107.8521351179095, "grad_norm": 0.3332255482673645, "learning_rate": 4.355894485232408e-05, "loss": 0.0153, "step": 169220 }, { "epoch": 107.8585086042065, "grad_norm": 0.47792279720306396, "learning_rate": 4.355347924860739e-05, "loss": 0.0129, "step": 169230 }, { "epoch": 107.8648820905035, "grad_norm": 0.34003570675849915, "learning_rate": 4.354801372322201e-05, "loss": 0.0135, "step": 169240 }, { "epoch": 107.87125557680051, "grad_norm": 0.2978619635105133, "learning_rate": 4.3542548276234374e-05, "loss": 0.0083, "step": 169250 }, { "epoch": 107.87762906309752, "grad_norm": 0.4650380611419678, "learning_rate": 4.3537082907710855e-05, "loss": 0.0085, "step": 169260 }, { "epoch": 107.88400254939452, "grad_norm": 0.31560760736465454, "learning_rate": 4.353161761771788e-05, "loss": 0.015, "step": 169270 }, { "epoch": 107.89037603569152, "grad_norm": 0.3394320607185364, "learning_rate": 4.352615240632186e-05, "loss": 0.0176, "step": 169280 }, { "epoch": 107.89674952198853, "grad_norm": 0.2864569127559662, "learning_rate": 4.352068727358922e-05, "loss": 0.0088, "step": 169290 }, { "epoch": 107.90312300828553, "grad_norm": 0.23367980122566223, "learning_rate": 4.3515222219586325e-05, "loss": 0.0079, "step": 169300 }, { "epoch": 107.90949649458254, "grad_norm": 0.2513819634914398, "learning_rate": 4.35097572443796e-05, "loss": 0.0146, "step": 169310 }, { "epoch": 107.91586998087953, "grad_norm": 0.32742807269096375, "learning_rate": 4.350429234803546e-05, "loss": 0.0122, "step": 169320 }, { "epoch": 107.92224346717654, "grad_norm": 0.18341760337352753, "learning_rate": 4.349882753062032e-05, "loss": 0.0108, "step": 169330 }, { "epoch": 107.92861695347355, "grad_norm": 0.18861381709575653, "learning_rate": 4.3493362792200536e-05, "loss": 0.0102, "step": 169340 }, { "epoch": 107.93499043977056, "grad_norm": 0.31750673055648804, "learning_rate": 4.3487898132842544e-05, "loss": 0.0138, "step": 169350 }, { "epoch": 107.94136392606755, "grad_norm": 0.3758854568004608, "learning_rate": 4.348243355261276e-05, "loss": 0.0103, "step": 169360 }, { "epoch": 107.94773741236456, "grad_norm": 0.14941057562828064, "learning_rate": 4.347696905157753e-05, "loss": 0.0081, "step": 169370 }, { "epoch": 107.95411089866157, "grad_norm": 0.36138200759887695, "learning_rate": 4.3471504629803294e-05, "loss": 0.0117, "step": 169380 }, { "epoch": 107.96048438495858, "grad_norm": 0.3289087414741516, "learning_rate": 4.346604028735644e-05, "loss": 0.0148, "step": 169390 }, { "epoch": 107.96685787125557, "grad_norm": 0.27919766306877136, "learning_rate": 4.3460576024303384e-05, "loss": 0.0101, "step": 169400 }, { "epoch": 107.97323135755258, "grad_norm": 0.2684210240840912, "learning_rate": 4.3455111840710483e-05, "loss": 0.007, "step": 169410 }, { "epoch": 107.97960484384959, "grad_norm": 0.17993927001953125, "learning_rate": 4.344964773664416e-05, "loss": 0.0077, "step": 169420 }, { "epoch": 107.9859783301466, "grad_norm": 0.3651474416255951, "learning_rate": 4.3444183712170814e-05, "loss": 0.0103, "step": 169430 }, { "epoch": 107.99235181644359, "grad_norm": 0.5867128968238831, "learning_rate": 4.3438719767356807e-05, "loss": 0.0102, "step": 169440 }, { "epoch": 107.9987253027406, "grad_norm": 0.33620065450668335, "learning_rate": 4.343325590226855e-05, "loss": 0.0117, "step": 169450 }, { "epoch": 108.0050987890376, "grad_norm": 0.41417253017425537, "learning_rate": 4.3427792116972434e-05, "loss": 0.0136, "step": 169460 }, { "epoch": 108.01147227533461, "grad_norm": 0.19463954865932465, "learning_rate": 4.342232841153487e-05, "loss": 0.0169, "step": 169470 }, { "epoch": 108.01784576163162, "grad_norm": 0.4086342751979828, "learning_rate": 4.34168647860222e-05, "loss": 0.0094, "step": 169480 }, { "epoch": 108.02421924792861, "grad_norm": 0.4564129412174225, "learning_rate": 4.341140124050085e-05, "loss": 0.0114, "step": 169490 }, { "epoch": 108.03059273422562, "grad_norm": 0.19648173451423645, "learning_rate": 4.340593777503721e-05, "loss": 0.0112, "step": 169500 }, { "epoch": 108.03696622052263, "grad_norm": 0.4616908133029938, "learning_rate": 4.340047438969763e-05, "loss": 0.0106, "step": 169510 }, { "epoch": 108.04333970681964, "grad_norm": 0.29557517170906067, "learning_rate": 4.3395011084548515e-05, "loss": 0.0147, "step": 169520 }, { "epoch": 108.04971319311663, "grad_norm": 0.29123538732528687, "learning_rate": 4.3389547859656256e-05, "loss": 0.011, "step": 169530 }, { "epoch": 108.05608667941364, "grad_norm": 0.42378583550453186, "learning_rate": 4.338408471508725e-05, "loss": 0.0205, "step": 169540 }, { "epoch": 108.06246016571065, "grad_norm": 0.1504535675048828, "learning_rate": 4.337862165090784e-05, "loss": 0.0111, "step": 169550 }, { "epoch": 108.06883365200765, "grad_norm": 0.19076324999332428, "learning_rate": 4.3373158667184425e-05, "loss": 0.0162, "step": 169560 }, { "epoch": 108.07520713830465, "grad_norm": 0.2480001002550125, "learning_rate": 4.336769576398342e-05, "loss": 0.0127, "step": 169570 }, { "epoch": 108.08158062460166, "grad_norm": 0.5998677611351013, "learning_rate": 4.3362232941371144e-05, "loss": 0.0163, "step": 169580 }, { "epoch": 108.08795411089866, "grad_norm": 0.25222447514533997, "learning_rate": 4.335677019941401e-05, "loss": 0.0115, "step": 169590 }, { "epoch": 108.09432759719567, "grad_norm": 0.22998031973838806, "learning_rate": 4.335130753817839e-05, "loss": 0.0057, "step": 169600 }, { "epoch": 108.10070108349267, "grad_norm": 0.17723073065280914, "learning_rate": 4.3345844957730675e-05, "loss": 0.0095, "step": 169610 }, { "epoch": 108.10707456978967, "grad_norm": 0.19583043456077576, "learning_rate": 4.334038245813721e-05, "loss": 0.0101, "step": 169620 }, { "epoch": 108.11344805608668, "grad_norm": 0.32471615076065063, "learning_rate": 4.333492003946439e-05, "loss": 0.0094, "step": 169630 }, { "epoch": 108.11982154238369, "grad_norm": 0.339629590511322, "learning_rate": 4.332945770177858e-05, "loss": 0.0099, "step": 169640 }, { "epoch": 108.12619502868068, "grad_norm": 0.15896373987197876, "learning_rate": 4.332399544514618e-05, "loss": 0.009, "step": 169650 }, { "epoch": 108.13256851497769, "grad_norm": 0.2657043933868408, "learning_rate": 4.331853326963352e-05, "loss": 0.0116, "step": 169660 }, { "epoch": 108.1389420012747, "grad_norm": 0.14209134876728058, "learning_rate": 4.331307117530698e-05, "loss": 0.0077, "step": 169670 }, { "epoch": 108.14531548757171, "grad_norm": 0.3936816453933716, "learning_rate": 4.330760916223296e-05, "loss": 0.0123, "step": 169680 }, { "epoch": 108.1516889738687, "grad_norm": 0.3351946771144867, "learning_rate": 4.3302147230477796e-05, "loss": 0.0082, "step": 169690 }, { "epoch": 108.15806246016571, "grad_norm": 0.27615928649902344, "learning_rate": 4.329668538010786e-05, "loss": 0.0118, "step": 169700 }, { "epoch": 108.16443594646272, "grad_norm": 0.43189314007759094, "learning_rate": 4.329122361118953e-05, "loss": 0.0191, "step": 169710 }, { "epoch": 108.17080943275973, "grad_norm": 0.16405583918094635, "learning_rate": 4.328576192378919e-05, "loss": 0.0071, "step": 169720 }, { "epoch": 108.17718291905672, "grad_norm": 0.2693024277687073, "learning_rate": 4.328030031797315e-05, "loss": 0.0075, "step": 169730 }, { "epoch": 108.18355640535373, "grad_norm": 0.29222580790519714, "learning_rate": 4.3274838793807814e-05, "loss": 0.0099, "step": 169740 }, { "epoch": 108.18992989165073, "grad_norm": 0.26303380727767944, "learning_rate": 4.3269377351359546e-05, "loss": 0.0084, "step": 169750 }, { "epoch": 108.19630337794774, "grad_norm": 0.353554368019104, "learning_rate": 4.32639159906947e-05, "loss": 0.0125, "step": 169760 }, { "epoch": 108.20267686424474, "grad_norm": 0.3440670073032379, "learning_rate": 4.325845471187962e-05, "loss": 0.013, "step": 169770 }, { "epoch": 108.20905035054174, "grad_norm": 0.2490426003932953, "learning_rate": 4.325299351498068e-05, "loss": 0.0076, "step": 169780 }, { "epoch": 108.21542383683875, "grad_norm": 0.3207528591156006, "learning_rate": 4.324753240006425e-05, "loss": 0.0104, "step": 169790 }, { "epoch": 108.22179732313576, "grad_norm": 0.5038672089576721, "learning_rate": 4.324207136719668e-05, "loss": 0.0081, "step": 169800 }, { "epoch": 108.22817080943275, "grad_norm": 0.21652023494243622, "learning_rate": 4.323661041644431e-05, "loss": 0.0178, "step": 169810 }, { "epoch": 108.23454429572976, "grad_norm": 0.21275198459625244, "learning_rate": 4.323114954787352e-05, "loss": 0.0085, "step": 169820 }, { "epoch": 108.24091778202677, "grad_norm": 0.26170581579208374, "learning_rate": 4.322568876155065e-05, "loss": 0.0079, "step": 169830 }, { "epoch": 108.24729126832378, "grad_norm": 0.26703304052352905, "learning_rate": 4.322022805754206e-05, "loss": 0.0101, "step": 169840 }, { "epoch": 108.25366475462077, "grad_norm": 0.34508857131004333, "learning_rate": 4.321476743591409e-05, "loss": 0.0207, "step": 169850 }, { "epoch": 108.26003824091778, "grad_norm": 0.24186117947101593, "learning_rate": 4.3209306896733117e-05, "loss": 0.0128, "step": 169860 }, { "epoch": 108.26641172721479, "grad_norm": 0.6387977600097656, "learning_rate": 4.320384644006546e-05, "loss": 0.0098, "step": 169870 }, { "epoch": 108.2727852135118, "grad_norm": 0.6288864612579346, "learning_rate": 4.31983860659775e-05, "loss": 0.0196, "step": 169880 }, { "epoch": 108.27915869980879, "grad_norm": 0.7174990773200989, "learning_rate": 4.319292577453557e-05, "loss": 0.0131, "step": 169890 }, { "epoch": 108.2855321861058, "grad_norm": 0.3065948188304901, "learning_rate": 4.3187465565806005e-05, "loss": 0.0085, "step": 169900 }, { "epoch": 108.2919056724028, "grad_norm": 0.18736563622951508, "learning_rate": 4.318200543985517e-05, "loss": 0.0085, "step": 169910 }, { "epoch": 108.29827915869981, "grad_norm": 0.5465612411499023, "learning_rate": 4.317654539674942e-05, "loss": 0.0095, "step": 169920 }, { "epoch": 108.3046526449968, "grad_norm": 0.5014667510986328, "learning_rate": 4.317108543655508e-05, "loss": 0.0131, "step": 169930 }, { "epoch": 108.31102613129381, "grad_norm": 0.3926471769809723, "learning_rate": 4.316562555933849e-05, "loss": 0.0107, "step": 169940 }, { "epoch": 108.31739961759082, "grad_norm": 0.35734668374061584, "learning_rate": 4.3160165765166005e-05, "loss": 0.0108, "step": 169950 }, { "epoch": 108.32377310388783, "grad_norm": 0.2349957525730133, "learning_rate": 4.315470605410397e-05, "loss": 0.026, "step": 169960 }, { "epoch": 108.33014659018482, "grad_norm": 0.12414094060659409, "learning_rate": 4.314924642621873e-05, "loss": 0.0072, "step": 169970 }, { "epoch": 108.33652007648183, "grad_norm": 0.39034000039100647, "learning_rate": 4.3143786881576584e-05, "loss": 0.0073, "step": 169980 }, { "epoch": 108.34289356277884, "grad_norm": 0.28555285930633545, "learning_rate": 4.313832742024392e-05, "loss": 0.0136, "step": 169990 }, { "epoch": 108.34926704907585, "grad_norm": 0.20723187923431396, "learning_rate": 4.3132868042287064e-05, "loss": 0.0131, "step": 170000 }, { "epoch": 108.35564053537284, "grad_norm": 0.2090780884027481, "learning_rate": 4.312740874777233e-05, "loss": 0.0064, "step": 170010 }, { "epoch": 108.36201402166985, "grad_norm": 0.1732848435640335, "learning_rate": 4.312194953676607e-05, "loss": 0.01, "step": 170020 }, { "epoch": 108.36838750796686, "grad_norm": 0.28313109278678894, "learning_rate": 4.311649040933461e-05, "loss": 0.0114, "step": 170030 }, { "epoch": 108.37476099426387, "grad_norm": 0.3119018077850342, "learning_rate": 4.311103136554432e-05, "loss": 0.0121, "step": 170040 }, { "epoch": 108.38113448056086, "grad_norm": 0.48149117827415466, "learning_rate": 4.310557240546148e-05, "loss": 0.0096, "step": 170050 }, { "epoch": 108.38750796685787, "grad_norm": 0.2264803797006607, "learning_rate": 4.310011352915244e-05, "loss": 0.0064, "step": 170060 }, { "epoch": 108.39388145315488, "grad_norm": 0.5079030990600586, "learning_rate": 4.309465473668356e-05, "loss": 0.0087, "step": 170070 }, { "epoch": 108.40025493945188, "grad_norm": 0.5987603068351746, "learning_rate": 4.3089196028121124e-05, "loss": 0.0176, "step": 170080 }, { "epoch": 108.40662842574889, "grad_norm": 0.22685036063194275, "learning_rate": 4.308373740353147e-05, "loss": 0.0092, "step": 170090 }, { "epoch": 108.41300191204589, "grad_norm": 0.3347600996494293, "learning_rate": 4.307827886298096e-05, "loss": 0.0087, "step": 170100 }, { "epoch": 108.4193753983429, "grad_norm": 0.3077840805053711, "learning_rate": 4.30728204065359e-05, "loss": 0.0073, "step": 170110 }, { "epoch": 108.4257488846399, "grad_norm": 0.6328039765357971, "learning_rate": 4.30673620342626e-05, "loss": 0.0088, "step": 170120 }, { "epoch": 108.43212237093691, "grad_norm": 0.44154542684555054, "learning_rate": 4.30619037462274e-05, "loss": 0.0124, "step": 170130 }, { "epoch": 108.4384958572339, "grad_norm": 0.2862395644187927, "learning_rate": 4.305644554249662e-05, "loss": 0.0104, "step": 170140 }, { "epoch": 108.44486934353091, "grad_norm": 0.28740987181663513, "learning_rate": 4.305098742313658e-05, "loss": 0.0148, "step": 170150 }, { "epoch": 108.45124282982792, "grad_norm": 0.44796013832092285, "learning_rate": 4.30455293882136e-05, "loss": 0.0103, "step": 170160 }, { "epoch": 108.45761631612493, "grad_norm": 0.34499454498291016, "learning_rate": 4.3040071437794e-05, "loss": 0.0087, "step": 170170 }, { "epoch": 108.46398980242192, "grad_norm": 0.38056108355522156, "learning_rate": 4.303461357194413e-05, "loss": 0.0097, "step": 170180 }, { "epoch": 108.47036328871893, "grad_norm": 0.19683057069778442, "learning_rate": 4.3029155790730255e-05, "loss": 0.0074, "step": 170190 }, { "epoch": 108.47673677501594, "grad_norm": 0.2343364655971527, "learning_rate": 4.302369809421873e-05, "loss": 0.0078, "step": 170200 }, { "epoch": 108.48311026131294, "grad_norm": 0.4842488169670105, "learning_rate": 4.3018240482475866e-05, "loss": 0.0091, "step": 170210 }, { "epoch": 108.48948374760994, "grad_norm": 0.4980618953704834, "learning_rate": 4.301278295556797e-05, "loss": 0.0174, "step": 170220 }, { "epoch": 108.49585723390695, "grad_norm": 0.5956015586853027, "learning_rate": 4.300732551356133e-05, "loss": 0.0096, "step": 170230 }, { "epoch": 108.50223072020395, "grad_norm": 0.20001919567584991, "learning_rate": 4.3001868156522315e-05, "loss": 0.0092, "step": 170240 }, { "epoch": 108.50860420650096, "grad_norm": 0.23201513290405273, "learning_rate": 4.299641088451721e-05, "loss": 0.0125, "step": 170250 }, { "epoch": 108.51497769279796, "grad_norm": 0.46366021037101746, "learning_rate": 4.299095369761232e-05, "loss": 0.0108, "step": 170260 }, { "epoch": 108.52135117909496, "grad_norm": 0.3104013502597809, "learning_rate": 4.2985496595873956e-05, "loss": 0.0087, "step": 170270 }, { "epoch": 108.52772466539197, "grad_norm": 0.18646173179149628, "learning_rate": 4.298003957936843e-05, "loss": 0.0129, "step": 170280 }, { "epoch": 108.53409815168898, "grad_norm": 0.19372393190860748, "learning_rate": 4.2974582648162064e-05, "loss": 0.0124, "step": 170290 }, { "epoch": 108.54047163798597, "grad_norm": 0.30049243569374084, "learning_rate": 4.296912580232114e-05, "loss": 0.0117, "step": 170300 }, { "epoch": 108.54684512428298, "grad_norm": 0.2139739841222763, "learning_rate": 4.296366904191197e-05, "loss": 0.0129, "step": 170310 }, { "epoch": 108.55321861057999, "grad_norm": 0.34110790491104126, "learning_rate": 4.295821236700089e-05, "loss": 0.0134, "step": 170320 }, { "epoch": 108.559592096877, "grad_norm": 0.499602347612381, "learning_rate": 4.295275577765415e-05, "loss": 0.0153, "step": 170330 }, { "epoch": 108.56596558317399, "grad_norm": 0.3195493817329407, "learning_rate": 4.294729927393809e-05, "loss": 0.0096, "step": 170340 }, { "epoch": 108.572339069471, "grad_norm": 0.3643755614757538, "learning_rate": 4.2941842855919e-05, "loss": 0.0154, "step": 170350 }, { "epoch": 108.578712555768, "grad_norm": 0.2838767468929291, "learning_rate": 4.2936386523663205e-05, "loss": 0.0098, "step": 170360 }, { "epoch": 108.58508604206502, "grad_norm": 0.3810317814350128, "learning_rate": 4.293093027723696e-05, "loss": 0.016, "step": 170370 }, { "epoch": 108.59145952836201, "grad_norm": 0.3853338360786438, "learning_rate": 4.292547411670658e-05, "loss": 0.0173, "step": 170380 }, { "epoch": 108.59783301465902, "grad_norm": 0.5438846945762634, "learning_rate": 4.2920018042138397e-05, "loss": 0.0108, "step": 170390 }, { "epoch": 108.60420650095602, "grad_norm": 0.29155635833740234, "learning_rate": 4.291456205359866e-05, "loss": 0.0078, "step": 170400 }, { "epoch": 108.61057998725303, "grad_norm": 0.43992745876312256, "learning_rate": 4.290910615115368e-05, "loss": 0.0086, "step": 170410 }, { "epoch": 108.61695347355003, "grad_norm": 0.14736612141132355, "learning_rate": 4.290365033486976e-05, "loss": 0.0058, "step": 170420 }, { "epoch": 108.62332695984703, "grad_norm": 0.2849392890930176, "learning_rate": 4.289819460481321e-05, "loss": 0.009, "step": 170430 }, { "epoch": 108.62970044614404, "grad_norm": 0.3303876519203186, "learning_rate": 4.289273896105027e-05, "loss": 0.0068, "step": 170440 }, { "epoch": 108.63607393244105, "grad_norm": 0.4552581012248993, "learning_rate": 4.2887283403647275e-05, "loss": 0.0195, "step": 170450 }, { "epoch": 108.64244741873804, "grad_norm": 0.20947322249412537, "learning_rate": 4.2881827932670506e-05, "loss": 0.0288, "step": 170460 }, { "epoch": 108.64882090503505, "grad_norm": 0.3589184284210205, "learning_rate": 4.287637254818625e-05, "loss": 0.0107, "step": 170470 }, { "epoch": 108.65519439133206, "grad_norm": 0.4011816680431366, "learning_rate": 4.287091725026078e-05, "loss": 0.0099, "step": 170480 }, { "epoch": 108.66156787762907, "grad_norm": 0.288547545671463, "learning_rate": 4.28654620389604e-05, "loss": 0.0125, "step": 170490 }, { "epoch": 108.66794136392606, "grad_norm": 0.20802952349185944, "learning_rate": 4.2860006914351395e-05, "loss": 0.009, "step": 170500 }, { "epoch": 108.67431485022307, "grad_norm": 0.49950510263442993, "learning_rate": 4.285455187650005e-05, "loss": 0.0174, "step": 170510 }, { "epoch": 108.68068833652008, "grad_norm": 0.23715883493423462, "learning_rate": 4.284909692547264e-05, "loss": 0.0096, "step": 170520 }, { "epoch": 108.68706182281709, "grad_norm": 0.38710451126098633, "learning_rate": 4.284364206133545e-05, "loss": 0.0114, "step": 170530 }, { "epoch": 108.69343530911408, "grad_norm": 0.32314783334732056, "learning_rate": 4.2838187284154784e-05, "loss": 0.0095, "step": 170540 }, { "epoch": 108.69980879541109, "grad_norm": 0.5360276699066162, "learning_rate": 4.28327325939969e-05, "loss": 0.0087, "step": 170550 }, { "epoch": 108.7061822817081, "grad_norm": 0.36965176463127136, "learning_rate": 4.282727799092807e-05, "loss": 0.0099, "step": 170560 }, { "epoch": 108.7125557680051, "grad_norm": 0.42684370279312134, "learning_rate": 4.282182347501459e-05, "loss": 0.0105, "step": 170570 }, { "epoch": 108.7189292543021, "grad_norm": 0.23568502068519592, "learning_rate": 4.281636904632273e-05, "loss": 0.0069, "step": 170580 }, { "epoch": 108.7253027405991, "grad_norm": 0.1915809065103531, "learning_rate": 4.2810914704918774e-05, "loss": 0.0156, "step": 170590 }, { "epoch": 108.73167622689611, "grad_norm": 0.315685898065567, "learning_rate": 4.280546045086898e-05, "loss": 0.0081, "step": 170600 }, { "epoch": 108.73804971319312, "grad_norm": 0.22328370809555054, "learning_rate": 4.280000628423966e-05, "loss": 0.0058, "step": 170610 }, { "epoch": 108.74442319949011, "grad_norm": 0.22390682995319366, "learning_rate": 4.2794552205097035e-05, "loss": 0.0169, "step": 170620 }, { "epoch": 108.75079668578712, "grad_norm": 0.16001178324222565, "learning_rate": 4.2789098213507415e-05, "loss": 0.0167, "step": 170630 }, { "epoch": 108.75717017208413, "grad_norm": 0.3616292178630829, "learning_rate": 4.2783644309537066e-05, "loss": 0.012, "step": 170640 }, { "epoch": 108.76354365838114, "grad_norm": 0.4070604145526886, "learning_rate": 4.277819049325224e-05, "loss": 0.0103, "step": 170650 }, { "epoch": 108.76991714467813, "grad_norm": 0.5809354186058044, "learning_rate": 4.277273676471922e-05, "loss": 0.0094, "step": 170660 }, { "epoch": 108.77629063097514, "grad_norm": 0.14609187841415405, "learning_rate": 4.2767283124004275e-05, "loss": 0.0083, "step": 170670 }, { "epoch": 108.78266411727215, "grad_norm": 0.20084534585475922, "learning_rate": 4.276182957117367e-05, "loss": 0.007, "step": 170680 }, { "epoch": 108.78903760356916, "grad_norm": 0.1522175520658493, "learning_rate": 4.275637610629367e-05, "loss": 0.012, "step": 170690 }, { "epoch": 108.79541108986615, "grad_norm": 0.16896946728229523, "learning_rate": 4.275092272943053e-05, "loss": 0.011, "step": 170700 }, { "epoch": 108.80178457616316, "grad_norm": 0.3650156557559967, "learning_rate": 4.274546944065055e-05, "loss": 0.0152, "step": 170710 }, { "epoch": 108.80815806246017, "grad_norm": 0.37917962670326233, "learning_rate": 4.2740016240019936e-05, "loss": 0.0178, "step": 170720 }, { "epoch": 108.81453154875717, "grad_norm": 0.34003975987434387, "learning_rate": 4.2734563127604986e-05, "loss": 0.0144, "step": 170730 }, { "epoch": 108.82090503505418, "grad_norm": 0.3570193946361542, "learning_rate": 4.272911010347195e-05, "loss": 0.0187, "step": 170740 }, { "epoch": 108.82727852135118, "grad_norm": 0.30827629566192627, "learning_rate": 4.2723657167687114e-05, "loss": 0.0102, "step": 170750 }, { "epoch": 108.83365200764818, "grad_norm": 0.3422752022743225, "learning_rate": 4.271820432031669e-05, "loss": 0.0128, "step": 170760 }, { "epoch": 108.84002549394519, "grad_norm": 0.42872658371925354, "learning_rate": 4.2712751561426964e-05, "loss": 0.0071, "step": 170770 }, { "epoch": 108.8463989802422, "grad_norm": 0.30568158626556396, "learning_rate": 4.270729889108421e-05, "loss": 0.0129, "step": 170780 }, { "epoch": 108.8527724665392, "grad_norm": 0.3439638316631317, "learning_rate": 4.270184630935463e-05, "loss": 0.0089, "step": 170790 }, { "epoch": 108.8591459528362, "grad_norm": 0.33820635080337524, "learning_rate": 4.269639381630452e-05, "loss": 0.0146, "step": 170800 }, { "epoch": 108.86551943913321, "grad_norm": 0.20687250792980194, "learning_rate": 4.26909414120001e-05, "loss": 0.0138, "step": 170810 }, { "epoch": 108.87189292543022, "grad_norm": 0.13651686906814575, "learning_rate": 4.268548909650768e-05, "loss": 0.0127, "step": 170820 }, { "epoch": 108.87826641172721, "grad_norm": 0.16999384760856628, "learning_rate": 4.268003686989345e-05, "loss": 0.0182, "step": 170830 }, { "epoch": 108.88463989802422, "grad_norm": 0.27678462862968445, "learning_rate": 4.2674584732223685e-05, "loss": 0.0122, "step": 170840 }, { "epoch": 108.89101338432123, "grad_norm": 0.16495956480503082, "learning_rate": 4.266913268356463e-05, "loss": 0.0136, "step": 170850 }, { "epoch": 108.89738687061823, "grad_norm": 0.22773180902004242, "learning_rate": 4.266368072398255e-05, "loss": 0.01, "step": 170860 }, { "epoch": 108.90376035691523, "grad_norm": 0.44807153940200806, "learning_rate": 4.265822885354366e-05, "loss": 0.012, "step": 170870 }, { "epoch": 108.91013384321224, "grad_norm": 0.2984979450702667, "learning_rate": 4.265277707231422e-05, "loss": 0.0081, "step": 170880 }, { "epoch": 108.91650732950924, "grad_norm": 0.3954803943634033, "learning_rate": 4.26473253803605e-05, "loss": 0.0147, "step": 170890 }, { "epoch": 108.92288081580625, "grad_norm": 0.2637425363063812, "learning_rate": 4.264187377774869e-05, "loss": 0.0091, "step": 170900 }, { "epoch": 108.92925430210325, "grad_norm": 0.6172242164611816, "learning_rate": 4.263642226454507e-05, "loss": 0.022, "step": 170910 }, { "epoch": 108.93562778840025, "grad_norm": 0.33556893467903137, "learning_rate": 4.2630970840815865e-05, "loss": 0.0152, "step": 170920 }, { "epoch": 108.94200127469726, "grad_norm": 0.6509662866592407, "learning_rate": 4.2625519506627345e-05, "loss": 0.0124, "step": 170930 }, { "epoch": 108.94837476099427, "grad_norm": 0.31472650170326233, "learning_rate": 4.26200682620457e-05, "loss": 0.0115, "step": 170940 }, { "epoch": 108.95474824729126, "grad_norm": 0.2834939658641815, "learning_rate": 4.26146171071372e-05, "loss": 0.0085, "step": 170950 }, { "epoch": 108.96112173358827, "grad_norm": 0.17114341259002686, "learning_rate": 4.2609166041968096e-05, "loss": 0.0089, "step": 170960 }, { "epoch": 108.96749521988528, "grad_norm": 0.2519650459289551, "learning_rate": 4.260371506660458e-05, "loss": 0.022, "step": 170970 }, { "epoch": 108.97386870618229, "grad_norm": 0.5140575766563416, "learning_rate": 4.2598264181112907e-05, "loss": 0.0117, "step": 170980 }, { "epoch": 108.98024219247928, "grad_norm": 0.3042718470096588, "learning_rate": 4.259281338555931e-05, "loss": 0.0105, "step": 170990 }, { "epoch": 108.98661567877629, "grad_norm": 0.2868387997150421, "learning_rate": 4.258736268001006e-05, "loss": 0.0123, "step": 171000 }, { "epoch": 108.9929891650733, "grad_norm": 0.4460926949977875, "learning_rate": 4.258191206453131e-05, "loss": 0.0104, "step": 171010 }, { "epoch": 108.9993626513703, "grad_norm": 0.3334428668022156, "learning_rate": 4.257646153918935e-05, "loss": 0.01, "step": 171020 }, { "epoch": 109.0057361376673, "grad_norm": 0.382476270198822, "learning_rate": 4.257101110405041e-05, "loss": 0.0114, "step": 171030 }, { "epoch": 109.0121096239643, "grad_norm": 0.37844476103782654, "learning_rate": 4.256556075918067e-05, "loss": 0.0095, "step": 171040 }, { "epoch": 109.01848311026131, "grad_norm": 0.1845501810312271, "learning_rate": 4.256011050464639e-05, "loss": 0.0076, "step": 171050 }, { "epoch": 109.02485659655832, "grad_norm": 0.20060916244983673, "learning_rate": 4.255466034051379e-05, "loss": 0.0101, "step": 171060 }, { "epoch": 109.03123008285532, "grad_norm": 0.24900072813034058, "learning_rate": 4.2549210266849114e-05, "loss": 0.0099, "step": 171070 }, { "epoch": 109.03760356915232, "grad_norm": 0.18433108925819397, "learning_rate": 4.254376028371856e-05, "loss": 0.0057, "step": 171080 }, { "epoch": 109.04397705544933, "grad_norm": 0.16389451920986176, "learning_rate": 4.253831039118835e-05, "loss": 0.0041, "step": 171090 }, { "epoch": 109.05035054174634, "grad_norm": 0.37560272216796875, "learning_rate": 4.253286058932473e-05, "loss": 0.0117, "step": 171100 }, { "epoch": 109.05672402804333, "grad_norm": 0.3737509250640869, "learning_rate": 4.252741087819389e-05, "loss": 0.011, "step": 171110 }, { "epoch": 109.06309751434034, "grad_norm": 0.2653356194496155, "learning_rate": 4.252196125786206e-05, "loss": 0.0089, "step": 171120 }, { "epoch": 109.06947100063735, "grad_norm": 0.28585270047187805, "learning_rate": 4.251651172839546e-05, "loss": 0.0071, "step": 171130 }, { "epoch": 109.07584448693436, "grad_norm": 0.2396208941936493, "learning_rate": 4.251106228986033e-05, "loss": 0.0107, "step": 171140 }, { "epoch": 109.08221797323135, "grad_norm": 0.18799270689487457, "learning_rate": 4.250561294232285e-05, "loss": 0.01, "step": 171150 }, { "epoch": 109.08859145952836, "grad_norm": 0.3141668438911438, "learning_rate": 4.2500163685849234e-05, "loss": 0.0086, "step": 171160 }, { "epoch": 109.09496494582537, "grad_norm": 0.20227903127670288, "learning_rate": 4.249471452050573e-05, "loss": 0.0087, "step": 171170 }, { "epoch": 109.10133843212238, "grad_norm": 0.2893296480178833, "learning_rate": 4.248926544635854e-05, "loss": 0.0076, "step": 171180 }, { "epoch": 109.10771191841937, "grad_norm": 0.5557260513305664, "learning_rate": 4.248381646347384e-05, "loss": 0.0104, "step": 171190 }, { "epoch": 109.11408540471638, "grad_norm": 0.3308170735836029, "learning_rate": 4.247836757191788e-05, "loss": 0.0256, "step": 171200 }, { "epoch": 109.12045889101339, "grad_norm": 0.23273631930351257, "learning_rate": 4.247291877175685e-05, "loss": 0.0081, "step": 171210 }, { "epoch": 109.1268323773104, "grad_norm": 0.16978636384010315, "learning_rate": 4.246747006305697e-05, "loss": 0.0107, "step": 171220 }, { "epoch": 109.13320586360739, "grad_norm": 0.25699687004089355, "learning_rate": 4.246202144588444e-05, "loss": 0.0129, "step": 171230 }, { "epoch": 109.1395793499044, "grad_norm": 0.23132507503032684, "learning_rate": 4.245657292030546e-05, "loss": 0.0085, "step": 171240 }, { "epoch": 109.1459528362014, "grad_norm": 0.3716508746147156, "learning_rate": 4.2451124486386244e-05, "loss": 0.0128, "step": 171250 }, { "epoch": 109.15232632249841, "grad_norm": 0.5993147492408752, "learning_rate": 4.2445676144193e-05, "loss": 0.0143, "step": 171260 }, { "epoch": 109.1586998087954, "grad_norm": 0.21676838397979736, "learning_rate": 4.244022789379191e-05, "loss": 0.0079, "step": 171270 }, { "epoch": 109.16507329509241, "grad_norm": 0.2732655107975006, "learning_rate": 4.24347797352492e-05, "loss": 0.0082, "step": 171280 }, { "epoch": 109.17144678138942, "grad_norm": 0.17327693104743958, "learning_rate": 4.242933166863105e-05, "loss": 0.0102, "step": 171290 }, { "epoch": 109.17782026768643, "grad_norm": 0.3781842589378357, "learning_rate": 4.242388369400368e-05, "loss": 0.0216, "step": 171300 }, { "epoch": 109.18419375398342, "grad_norm": 0.2239922136068344, "learning_rate": 4.2418435811433267e-05, "loss": 0.0211, "step": 171310 }, { "epoch": 109.19056724028043, "grad_norm": 0.37924936413764954, "learning_rate": 4.241298802098602e-05, "loss": 0.0123, "step": 171320 }, { "epoch": 109.19694072657744, "grad_norm": 0.17974121868610382, "learning_rate": 4.240754032272813e-05, "loss": 0.0088, "step": 171330 }, { "epoch": 109.20331421287445, "grad_norm": 0.5045853853225708, "learning_rate": 4.24020927167258e-05, "loss": 0.0131, "step": 171340 }, { "epoch": 109.20968769917145, "grad_norm": 0.4807288348674774, "learning_rate": 4.239664520304522e-05, "loss": 0.0117, "step": 171350 }, { "epoch": 109.21606118546845, "grad_norm": 0.22471381723880768, "learning_rate": 4.239119778175256e-05, "loss": 0.0127, "step": 171360 }, { "epoch": 109.22243467176546, "grad_norm": 0.6168084740638733, "learning_rate": 4.238575045291404e-05, "loss": 0.0219, "step": 171370 }, { "epoch": 109.22880815806246, "grad_norm": 0.48330914974212646, "learning_rate": 4.2380303216595845e-05, "loss": 0.0176, "step": 171380 }, { "epoch": 109.23518164435947, "grad_norm": 0.2258267104625702, "learning_rate": 4.2374856072864166e-05, "loss": 0.0095, "step": 171390 }, { "epoch": 109.24155513065647, "grad_norm": 0.4041142463684082, "learning_rate": 4.236940902178517e-05, "loss": 0.0117, "step": 171400 }, { "epoch": 109.24792861695347, "grad_norm": 0.271754652261734, "learning_rate": 4.236396206342506e-05, "loss": 0.0064, "step": 171410 }, { "epoch": 109.25430210325048, "grad_norm": 0.3129040002822876, "learning_rate": 4.2358515197850036e-05, "loss": 0.0143, "step": 171420 }, { "epoch": 109.26067558954749, "grad_norm": 0.24160872399806976, "learning_rate": 4.235306842512626e-05, "loss": 0.0102, "step": 171430 }, { "epoch": 109.26704907584448, "grad_norm": 0.2752612829208374, "learning_rate": 4.2347621745319915e-05, "loss": 0.0148, "step": 171440 }, { "epoch": 109.27342256214149, "grad_norm": 0.494932621717453, "learning_rate": 4.234217515849719e-05, "loss": 0.0152, "step": 171450 }, { "epoch": 109.2797960484385, "grad_norm": 0.14983505010604858, "learning_rate": 4.233672866472429e-05, "loss": 0.0085, "step": 171460 }, { "epoch": 109.2861695347355, "grad_norm": 0.4194854497909546, "learning_rate": 4.233128226406735e-05, "loss": 0.0115, "step": 171470 }, { "epoch": 109.2925430210325, "grad_norm": 0.4901782274246216, "learning_rate": 4.2325835956592576e-05, "loss": 0.0131, "step": 171480 }, { "epoch": 109.29891650732951, "grad_norm": 0.3674180209636688, "learning_rate": 4.2320389742366135e-05, "loss": 0.0073, "step": 171490 }, { "epoch": 109.30528999362652, "grad_norm": 0.1500648409128189, "learning_rate": 4.231494362145423e-05, "loss": 0.0113, "step": 171500 }, { "epoch": 109.31166347992352, "grad_norm": 0.16180400550365448, "learning_rate": 4.2309497593923e-05, "loss": 0.0088, "step": 171510 }, { "epoch": 109.31803696622052, "grad_norm": 0.21947318315505981, "learning_rate": 4.2304051659838625e-05, "loss": 0.0127, "step": 171520 }, { "epoch": 109.32441045251753, "grad_norm": 0.44120922684669495, "learning_rate": 4.229860581926732e-05, "loss": 0.0212, "step": 171530 }, { "epoch": 109.33078393881453, "grad_norm": 0.25522419810295105, "learning_rate": 4.2293160072275194e-05, "loss": 0.0083, "step": 171540 }, { "epoch": 109.33715742511154, "grad_norm": 0.32502225041389465, "learning_rate": 4.228771441892846e-05, "loss": 0.0097, "step": 171550 }, { "epoch": 109.34353091140854, "grad_norm": 0.3405592739582062, "learning_rate": 4.2282268859293274e-05, "loss": 0.0083, "step": 171560 }, { "epoch": 109.34990439770554, "grad_norm": 0.39621198177337646, "learning_rate": 4.2276823393435824e-05, "loss": 0.0111, "step": 171570 }, { "epoch": 109.35627788400255, "grad_norm": 0.2664855122566223, "learning_rate": 4.2271378021422246e-05, "loss": 0.0098, "step": 171580 }, { "epoch": 109.36265137029956, "grad_norm": 0.24460317194461823, "learning_rate": 4.226593274331873e-05, "loss": 0.0103, "step": 171590 }, { "epoch": 109.36902485659655, "grad_norm": 0.1879783719778061, "learning_rate": 4.2260487559191444e-05, "loss": 0.0102, "step": 171600 }, { "epoch": 109.37539834289356, "grad_norm": 0.800293505191803, "learning_rate": 4.225504246910652e-05, "loss": 0.0117, "step": 171610 }, { "epoch": 109.38177182919057, "grad_norm": 0.26299846172332764, "learning_rate": 4.224959747313014e-05, "loss": 0.0141, "step": 171620 }, { "epoch": 109.38814531548758, "grad_norm": 0.21143440902233124, "learning_rate": 4.224415257132848e-05, "loss": 0.0119, "step": 171630 }, { "epoch": 109.39451880178457, "grad_norm": 0.26259496808052063, "learning_rate": 4.2238707763767696e-05, "loss": 0.014, "step": 171640 }, { "epoch": 109.40089228808158, "grad_norm": 0.8171494603157043, "learning_rate": 4.223326305051393e-05, "loss": 0.0102, "step": 171650 }, { "epoch": 109.40726577437859, "grad_norm": 0.3419456481933594, "learning_rate": 4.2227818431633345e-05, "loss": 0.0048, "step": 171660 }, { "epoch": 109.4136392606756, "grad_norm": 0.3192119002342224, "learning_rate": 4.2222373907192124e-05, "loss": 0.0166, "step": 171670 }, { "epoch": 109.42001274697259, "grad_norm": 0.4798201620578766, "learning_rate": 4.221692947725639e-05, "loss": 0.0128, "step": 171680 }, { "epoch": 109.4263862332696, "grad_norm": 0.204572394490242, "learning_rate": 4.22114851418923e-05, "loss": 0.0078, "step": 171690 }, { "epoch": 109.4327597195666, "grad_norm": 0.2679121196269989, "learning_rate": 4.2206040901166035e-05, "loss": 0.0101, "step": 171700 }, { "epoch": 109.43913320586361, "grad_norm": 0.3925614356994629, "learning_rate": 4.220059675514374e-05, "loss": 0.0135, "step": 171710 }, { "epoch": 109.4455066921606, "grad_norm": 0.23752984404563904, "learning_rate": 4.219515270389154e-05, "loss": 0.0155, "step": 171720 }, { "epoch": 109.45188017845761, "grad_norm": 0.22955745458602905, "learning_rate": 4.218970874747561e-05, "loss": 0.0122, "step": 171730 }, { "epoch": 109.45825366475462, "grad_norm": 0.46520721912384033, "learning_rate": 4.218426488596209e-05, "loss": 0.0096, "step": 171740 }, { "epoch": 109.46462715105163, "grad_norm": 0.48901501297950745, "learning_rate": 4.2178821119417145e-05, "loss": 0.0112, "step": 171750 }, { "epoch": 109.47100063734862, "grad_norm": 0.18841320276260376, "learning_rate": 4.217337744790689e-05, "loss": 0.0096, "step": 171760 }, { "epoch": 109.47737412364563, "grad_norm": 0.24760234355926514, "learning_rate": 4.216793387149749e-05, "loss": 0.0174, "step": 171770 }, { "epoch": 109.48374760994264, "grad_norm": 0.2376897633075714, "learning_rate": 4.216249039025511e-05, "loss": 0.0261, "step": 171780 }, { "epoch": 109.49012109623965, "grad_norm": 0.3901434540748596, "learning_rate": 4.2157047004245856e-05, "loss": 0.0133, "step": 171790 }, { "epoch": 109.49649458253664, "grad_norm": 0.460152804851532, "learning_rate": 4.215160371353588e-05, "loss": 0.0103, "step": 171800 }, { "epoch": 109.50286806883365, "grad_norm": 0.24992209672927856, "learning_rate": 4.214616051819132e-05, "loss": 0.008, "step": 171810 }, { "epoch": 109.50924155513066, "grad_norm": 0.25282537937164307, "learning_rate": 4.2140717418278355e-05, "loss": 0.0112, "step": 171820 }, { "epoch": 109.51561504142767, "grad_norm": 1.7654428482055664, "learning_rate": 4.2135274413863075e-05, "loss": 0.0115, "step": 171830 }, { "epoch": 109.52198852772466, "grad_norm": 0.18119685351848602, "learning_rate": 4.212983150501164e-05, "loss": 0.008, "step": 171840 }, { "epoch": 109.52836201402167, "grad_norm": 0.21903249621391296, "learning_rate": 4.212438869179019e-05, "loss": 0.0113, "step": 171850 }, { "epoch": 109.53473550031867, "grad_norm": 0.21433235704898834, "learning_rate": 4.211894597426484e-05, "loss": 0.0078, "step": 171860 }, { "epoch": 109.54110898661568, "grad_norm": 0.17953987419605255, "learning_rate": 4.211350335250174e-05, "loss": 0.0119, "step": 171870 }, { "epoch": 109.54748247291268, "grad_norm": 0.546501874923706, "learning_rate": 4.210806082656702e-05, "loss": 0.0147, "step": 171880 }, { "epoch": 109.55385595920968, "grad_norm": 0.39115792512893677, "learning_rate": 4.210261839652683e-05, "loss": 0.0228, "step": 171890 }, { "epoch": 109.56022944550669, "grad_norm": 0.2340622991323471, "learning_rate": 4.209717606244726e-05, "loss": 0.0107, "step": 171900 }, { "epoch": 109.5666029318037, "grad_norm": 0.28470170497894287, "learning_rate": 4.2091733824394466e-05, "loss": 0.0136, "step": 171910 }, { "epoch": 109.5729764181007, "grad_norm": 0.24246437847614288, "learning_rate": 4.2086291682434584e-05, "loss": 0.0124, "step": 171920 }, { "epoch": 109.5793499043977, "grad_norm": 0.3209938406944275, "learning_rate": 4.208084963663372e-05, "loss": 0.0107, "step": 171930 }, { "epoch": 109.58572339069471, "grad_norm": 0.316974937915802, "learning_rate": 4.207540768705801e-05, "loss": 0.0113, "step": 171940 }, { "epoch": 109.59209687699172, "grad_norm": 0.32699984312057495, "learning_rate": 4.206996583377358e-05, "loss": 0.007, "step": 171950 }, { "epoch": 109.59847036328873, "grad_norm": 0.2589528262615204, "learning_rate": 4.206452407684656e-05, "loss": 0.0118, "step": 171960 }, { "epoch": 109.60484384958572, "grad_norm": 2.6415398120880127, "learning_rate": 4.205908241634306e-05, "loss": 0.014, "step": 171970 }, { "epoch": 109.61121733588273, "grad_norm": 0.18904365599155426, "learning_rate": 4.20536408523292e-05, "loss": 0.0058, "step": 171980 }, { "epoch": 109.61759082217974, "grad_norm": 0.4905429184436798, "learning_rate": 4.2048199384871116e-05, "loss": 0.0119, "step": 171990 }, { "epoch": 109.62396430847674, "grad_norm": 0.2168460488319397, "learning_rate": 4.2042758014034904e-05, "loss": 0.0099, "step": 172000 }, { "epoch": 109.63033779477374, "grad_norm": 0.5997743606567383, "learning_rate": 4.203731673988671e-05, "loss": 0.0106, "step": 172010 }, { "epoch": 109.63671128107075, "grad_norm": 0.3258243501186371, "learning_rate": 4.203187556249262e-05, "loss": 0.012, "step": 172020 }, { "epoch": 109.64308476736775, "grad_norm": 0.19657760858535767, "learning_rate": 4.2026434481918786e-05, "loss": 0.0093, "step": 172030 }, { "epoch": 109.64945825366476, "grad_norm": 0.14571408927440643, "learning_rate": 4.202099349823128e-05, "loss": 0.0127, "step": 172040 }, { "epoch": 109.65583173996176, "grad_norm": 0.3674352169036865, "learning_rate": 4.201555261149625e-05, "loss": 0.0067, "step": 172050 }, { "epoch": 109.66220522625876, "grad_norm": 0.21744151413440704, "learning_rate": 4.201011182177979e-05, "loss": 0.0088, "step": 172060 }, { "epoch": 109.66857871255577, "grad_norm": 0.2496325522661209, "learning_rate": 4.2004671129148017e-05, "loss": 0.0145, "step": 172070 }, { "epoch": 109.67495219885278, "grad_norm": 0.5369653105735779, "learning_rate": 4.1999230533667035e-05, "loss": 0.0149, "step": 172080 }, { "epoch": 109.68132568514977, "grad_norm": 0.5687335133552551, "learning_rate": 4.199379003540296e-05, "loss": 0.011, "step": 172090 }, { "epoch": 109.68769917144678, "grad_norm": 0.5148007869720459, "learning_rate": 4.19883496344219e-05, "loss": 0.0082, "step": 172100 }, { "epoch": 109.69407265774379, "grad_norm": 0.3046034574508667, "learning_rate": 4.198290933078994e-05, "loss": 0.0112, "step": 172110 }, { "epoch": 109.7004461440408, "grad_norm": 0.307907372713089, "learning_rate": 4.19774691245732e-05, "loss": 0.0189, "step": 172120 }, { "epoch": 109.70681963033779, "grad_norm": 0.24587257206439972, "learning_rate": 4.1972029015837805e-05, "loss": 0.0154, "step": 172130 }, { "epoch": 109.7131931166348, "grad_norm": 0.2344251275062561, "learning_rate": 4.1966589004649834e-05, "loss": 0.0164, "step": 172140 }, { "epoch": 109.7195666029318, "grad_norm": 0.18727250397205353, "learning_rate": 4.1961149091075376e-05, "loss": 0.0109, "step": 172150 }, { "epoch": 109.72594008922881, "grad_norm": 0.4394564926624298, "learning_rate": 4.195570927518055e-05, "loss": 0.0072, "step": 172160 }, { "epoch": 109.73231357552581, "grad_norm": 0.6318947672843933, "learning_rate": 4.195026955703147e-05, "loss": 0.02, "step": 172170 }, { "epoch": 109.73868706182282, "grad_norm": 0.2984684407711029, "learning_rate": 4.1944829936694195e-05, "loss": 0.0061, "step": 172180 }, { "epoch": 109.74506054811982, "grad_norm": 0.494525671005249, "learning_rate": 4.193939041423484e-05, "loss": 0.0113, "step": 172190 }, { "epoch": 109.75143403441683, "grad_norm": 0.5724154114723206, "learning_rate": 4.193395098971951e-05, "loss": 0.0261, "step": 172200 }, { "epoch": 109.75780752071383, "grad_norm": 0.3066841661930084, "learning_rate": 4.19285116632143e-05, "loss": 0.008, "step": 172210 }, { "epoch": 109.76418100701083, "grad_norm": 0.28918513655662537, "learning_rate": 4.192307243478528e-05, "loss": 0.011, "step": 172220 }, { "epoch": 109.77055449330784, "grad_norm": 0.24969854950904846, "learning_rate": 4.191763330449856e-05, "loss": 0.0083, "step": 172230 }, { "epoch": 109.77692797960485, "grad_norm": 0.530104398727417, "learning_rate": 4.191219427242024e-05, "loss": 0.0152, "step": 172240 }, { "epoch": 109.78330146590184, "grad_norm": 0.1586245894432068, "learning_rate": 4.1906755338616376e-05, "loss": 0.0079, "step": 172250 }, { "epoch": 109.78967495219885, "grad_norm": 0.17156627774238586, "learning_rate": 4.190131650315308e-05, "loss": 0.0122, "step": 172260 }, { "epoch": 109.79604843849586, "grad_norm": 0.2004336416721344, "learning_rate": 4.189587776609643e-05, "loss": 0.0091, "step": 172270 }, { "epoch": 109.80242192479287, "grad_norm": 0.15056149661540985, "learning_rate": 4.1890439127512544e-05, "loss": 0.016, "step": 172280 }, { "epoch": 109.80879541108986, "grad_norm": 0.5163003206253052, "learning_rate": 4.188500058746745e-05, "loss": 0.0184, "step": 172290 }, { "epoch": 109.81516889738687, "grad_norm": 0.2417282611131668, "learning_rate": 4.187956214602726e-05, "loss": 0.0119, "step": 172300 }, { "epoch": 109.82154238368388, "grad_norm": 0.21514086425304413, "learning_rate": 4.187412380325809e-05, "loss": 0.0114, "step": 172310 }, { "epoch": 109.82791586998088, "grad_norm": 0.1511971652507782, "learning_rate": 4.186868555922596e-05, "loss": 0.0085, "step": 172320 }, { "epoch": 109.83428935627788, "grad_norm": 0.17077092826366425, "learning_rate": 4.186324741399697e-05, "loss": 0.0158, "step": 172330 }, { "epoch": 109.84066284257489, "grad_norm": 0.2133258730173111, "learning_rate": 4.185780936763722e-05, "loss": 0.0115, "step": 172340 }, { "epoch": 109.8470363288719, "grad_norm": 0.2053299993276596, "learning_rate": 4.185237142021279e-05, "loss": 0.0075, "step": 172350 }, { "epoch": 109.8534098151689, "grad_norm": 0.21862468123435974, "learning_rate": 4.184693357178972e-05, "loss": 0.0099, "step": 172360 }, { "epoch": 109.8597833014659, "grad_norm": 0.1859138011932373, "learning_rate": 4.1841495822434096e-05, "loss": 0.009, "step": 172370 }, { "epoch": 109.8661567877629, "grad_norm": 0.6480569839477539, "learning_rate": 4.1836058172212006e-05, "loss": 0.0173, "step": 172380 }, { "epoch": 109.87253027405991, "grad_norm": 0.536190390586853, "learning_rate": 4.183062062118954e-05, "loss": 0.0122, "step": 172390 }, { "epoch": 109.87890376035692, "grad_norm": 0.5370351672172546, "learning_rate": 4.1825183169432726e-05, "loss": 0.0119, "step": 172400 }, { "epoch": 109.88527724665391, "grad_norm": 0.3432498872280121, "learning_rate": 4.181974581700765e-05, "loss": 0.0126, "step": 172410 }, { "epoch": 109.89165073295092, "grad_norm": 0.16557182371616364, "learning_rate": 4.181430856398041e-05, "loss": 0.0094, "step": 172420 }, { "epoch": 109.89802421924793, "grad_norm": 0.4331972002983093, "learning_rate": 4.180887141041703e-05, "loss": 0.0098, "step": 172430 }, { "epoch": 109.90439770554494, "grad_norm": 0.18901489675045013, "learning_rate": 4.18034343563836e-05, "loss": 0.0098, "step": 172440 }, { "epoch": 109.91077119184193, "grad_norm": 0.6216018795967102, "learning_rate": 4.179799740194617e-05, "loss": 0.0097, "step": 172450 }, { "epoch": 109.91714467813894, "grad_norm": 0.2868194878101349, "learning_rate": 4.1792560547170844e-05, "loss": 0.011, "step": 172460 }, { "epoch": 109.92351816443595, "grad_norm": 0.45318156480789185, "learning_rate": 4.178712379212363e-05, "loss": 0.011, "step": 172470 }, { "epoch": 109.92989165073296, "grad_norm": 0.2538149356842041, "learning_rate": 4.178168713687062e-05, "loss": 0.0103, "step": 172480 }, { "epoch": 109.93626513702995, "grad_norm": 0.6081294417381287, "learning_rate": 4.177625058147789e-05, "loss": 0.0198, "step": 172490 }, { "epoch": 109.94263862332696, "grad_norm": 0.2344064712524414, "learning_rate": 4.1770814126011474e-05, "loss": 0.0093, "step": 172500 }, { "epoch": 109.94901210962396, "grad_norm": 0.3216239809989929, "learning_rate": 4.176537777053742e-05, "loss": 0.0132, "step": 172510 }, { "epoch": 109.95538559592097, "grad_norm": 0.49401551485061646, "learning_rate": 4.17599415151218e-05, "loss": 0.0086, "step": 172520 }, { "epoch": 109.96175908221797, "grad_norm": 0.17153014242649078, "learning_rate": 4.17545053598307e-05, "loss": 0.0074, "step": 172530 }, { "epoch": 109.96813256851497, "grad_norm": 0.411493182182312, "learning_rate": 4.1749069304730116e-05, "loss": 0.0117, "step": 172540 }, { "epoch": 109.97450605481198, "grad_norm": 0.2201618254184723, "learning_rate": 4.174363334988613e-05, "loss": 0.0174, "step": 172550 }, { "epoch": 109.98087954110899, "grad_norm": 0.24185654520988464, "learning_rate": 4.1738197495364815e-05, "loss": 0.0117, "step": 172560 }, { "epoch": 109.98725302740598, "grad_norm": 0.2820224165916443, "learning_rate": 4.1732761741232184e-05, "loss": 0.011, "step": 172570 }, { "epoch": 109.99362651370299, "grad_norm": 0.23259374499320984, "learning_rate": 4.1727326087554307e-05, "loss": 0.0079, "step": 172580 }, { "epoch": 110.0, "grad_norm": 0.4759277105331421, "learning_rate": 4.1721890534397224e-05, "loss": 0.0075, "step": 172590 }, { "epoch": 110.00637348629701, "grad_norm": 0.4233880043029785, "learning_rate": 4.171645508182701e-05, "loss": 0.008, "step": 172600 }, { "epoch": 110.01274697259402, "grad_norm": 0.34708428382873535, "learning_rate": 4.171101972990966e-05, "loss": 0.0088, "step": 172610 }, { "epoch": 110.01912045889101, "grad_norm": 0.3664065897464752, "learning_rate": 4.170558447871126e-05, "loss": 0.0187, "step": 172620 }, { "epoch": 110.02549394518802, "grad_norm": 0.21647362411022186, "learning_rate": 4.170014932829782e-05, "loss": 0.0154, "step": 172630 }, { "epoch": 110.03186743148503, "grad_norm": 0.35520806908607483, "learning_rate": 4.169471427873543e-05, "loss": 0.0076, "step": 172640 }, { "epoch": 110.03824091778203, "grad_norm": 0.17340098321437836, "learning_rate": 4.16892793300901e-05, "loss": 0.0061, "step": 172650 }, { "epoch": 110.04461440407903, "grad_norm": 0.26088181138038635, "learning_rate": 4.1683844482427855e-05, "loss": 0.0081, "step": 172660 }, { "epoch": 110.05098789037604, "grad_norm": 0.4854391813278198, "learning_rate": 4.1678409735814765e-05, "loss": 0.0192, "step": 172670 }, { "epoch": 110.05736137667304, "grad_norm": 0.3228088319301605, "learning_rate": 4.167297509031686e-05, "loss": 0.0102, "step": 172680 }, { "epoch": 110.06373486297005, "grad_norm": 0.21064162254333496, "learning_rate": 4.166754054600015e-05, "loss": 0.0156, "step": 172690 }, { "epoch": 110.07010834926704, "grad_norm": 0.19387248158454895, "learning_rate": 4.166210610293069e-05, "loss": 0.0076, "step": 172700 }, { "epoch": 110.07648183556405, "grad_norm": 0.5341606736183167, "learning_rate": 4.165667176117453e-05, "loss": 0.0169, "step": 172710 }, { "epoch": 110.08285532186106, "grad_norm": 0.4119935631752014, "learning_rate": 4.165123752079768e-05, "loss": 0.007, "step": 172720 }, { "epoch": 110.08922880815807, "grad_norm": 0.4524100720882416, "learning_rate": 4.1645803381866165e-05, "loss": 0.0093, "step": 172730 }, { "epoch": 110.09560229445506, "grad_norm": 0.38652774691581726, "learning_rate": 4.164036934444604e-05, "loss": 0.0116, "step": 172740 }, { "epoch": 110.10197578075207, "grad_norm": 0.40483057498931885, "learning_rate": 4.163493540860331e-05, "loss": 0.0119, "step": 172750 }, { "epoch": 110.10834926704908, "grad_norm": 0.5596733689308167, "learning_rate": 4.162950157440403e-05, "loss": 0.0156, "step": 172760 }, { "epoch": 110.11472275334609, "grad_norm": 0.1670064479112625, "learning_rate": 4.162406784191419e-05, "loss": 0.0103, "step": 172770 }, { "epoch": 110.12109623964308, "grad_norm": 0.3147008419036865, "learning_rate": 4.161863421119985e-05, "loss": 0.0074, "step": 172780 }, { "epoch": 110.12746972594009, "grad_norm": 0.6475774645805359, "learning_rate": 4.1613200682326994e-05, "loss": 0.0123, "step": 172790 }, { "epoch": 110.1338432122371, "grad_norm": 0.18585817515850067, "learning_rate": 4.160776725536169e-05, "loss": 0.01, "step": 172800 }, { "epoch": 110.1402166985341, "grad_norm": 0.2975488007068634, "learning_rate": 4.1602333930369934e-05, "loss": 0.0093, "step": 172810 }, { "epoch": 110.1465901848311, "grad_norm": 0.13970054686069489, "learning_rate": 4.159690070741774e-05, "loss": 0.0149, "step": 172820 }, { "epoch": 110.1529636711281, "grad_norm": 0.6077752709388733, "learning_rate": 4.159146758657114e-05, "loss": 0.0148, "step": 172830 }, { "epoch": 110.15933715742511, "grad_norm": 0.3972182869911194, "learning_rate": 4.158603456789615e-05, "loss": 0.0132, "step": 172840 }, { "epoch": 110.16571064372212, "grad_norm": 0.23251689970493317, "learning_rate": 4.1580601651458795e-05, "loss": 0.0093, "step": 172850 }, { "epoch": 110.17208413001912, "grad_norm": 0.36706361174583435, "learning_rate": 4.157516883732506e-05, "loss": 0.0091, "step": 172860 }, { "epoch": 110.17845761631612, "grad_norm": 0.34928786754608154, "learning_rate": 4.156973612556098e-05, "loss": 0.0073, "step": 172870 }, { "epoch": 110.18483110261313, "grad_norm": 0.441378653049469, "learning_rate": 4.156430351623259e-05, "loss": 0.012, "step": 172880 }, { "epoch": 110.19120458891014, "grad_norm": 0.34413984417915344, "learning_rate": 4.1558871009405854e-05, "loss": 0.0123, "step": 172890 }, { "epoch": 110.19757807520713, "grad_norm": 0.19374719262123108, "learning_rate": 4.15534386051468e-05, "loss": 0.0056, "step": 172900 }, { "epoch": 110.20395156150414, "grad_norm": 0.40389153361320496, "learning_rate": 4.154800630352145e-05, "loss": 0.0138, "step": 172910 }, { "epoch": 110.21032504780115, "grad_norm": 0.18452785909175873, "learning_rate": 4.154257410459582e-05, "loss": 0.0193, "step": 172920 }, { "epoch": 110.21669853409816, "grad_norm": 0.8509377241134644, "learning_rate": 4.153714200843587e-05, "loss": 0.0124, "step": 172930 }, { "epoch": 110.22307202039515, "grad_norm": 0.20215463638305664, "learning_rate": 4.153171001510765e-05, "loss": 0.0202, "step": 172940 }, { "epoch": 110.22944550669216, "grad_norm": 0.41150638461112976, "learning_rate": 4.152627812467714e-05, "loss": 0.0145, "step": 172950 }, { "epoch": 110.23581899298917, "grad_norm": 0.14795047044754028, "learning_rate": 4.152084633721038e-05, "loss": 0.0067, "step": 172960 }, { "epoch": 110.24219247928617, "grad_norm": 0.18055403232574463, "learning_rate": 4.151541465277332e-05, "loss": 0.011, "step": 172970 }, { "epoch": 110.24856596558317, "grad_norm": 0.34487468004226685, "learning_rate": 4.1509983071431977e-05, "loss": 0.0094, "step": 172980 }, { "epoch": 110.25493945188018, "grad_norm": 0.49562814831733704, "learning_rate": 4.150455159325238e-05, "loss": 0.0101, "step": 172990 }, { "epoch": 110.26131293817718, "grad_norm": 0.11693304777145386, "learning_rate": 4.149912021830048e-05, "loss": 0.0073, "step": 173000 }, { "epoch": 110.26768642447419, "grad_norm": 0.23640048503875732, "learning_rate": 4.149368894664229e-05, "loss": 0.0085, "step": 173010 }, { "epoch": 110.27405991077119, "grad_norm": 0.347206175327301, "learning_rate": 4.148825777834382e-05, "loss": 0.012, "step": 173020 }, { "epoch": 110.2804333970682, "grad_norm": 0.49558237195014954, "learning_rate": 4.148282671347107e-05, "loss": 0.0127, "step": 173030 }, { "epoch": 110.2868068833652, "grad_norm": 0.37004128098487854, "learning_rate": 4.1477395752089997e-05, "loss": 0.014, "step": 173040 }, { "epoch": 110.29318036966221, "grad_norm": 0.1632647067308426, "learning_rate": 4.1471964894266615e-05, "loss": 0.0141, "step": 173050 }, { "epoch": 110.2995538559592, "grad_norm": 0.22519782185554504, "learning_rate": 4.1466534140066927e-05, "loss": 0.0082, "step": 173060 }, { "epoch": 110.30592734225621, "grad_norm": 0.2857963740825653, "learning_rate": 4.146110348955689e-05, "loss": 0.0098, "step": 173070 }, { "epoch": 110.31230082855322, "grad_norm": 0.26260387897491455, "learning_rate": 4.1455672942802504e-05, "loss": 0.0082, "step": 173080 }, { "epoch": 110.31867431485023, "grad_norm": 0.29855671525001526, "learning_rate": 4.1450242499869765e-05, "loss": 0.0113, "step": 173090 }, { "epoch": 110.32504780114722, "grad_norm": 0.5619910359382629, "learning_rate": 4.144481216082466e-05, "loss": 0.0128, "step": 173100 }, { "epoch": 110.33142128744423, "grad_norm": 0.19362685084342957, "learning_rate": 4.143938192573315e-05, "loss": 0.0117, "step": 173110 }, { "epoch": 110.33779477374124, "grad_norm": 0.3830394744873047, "learning_rate": 4.143395179466124e-05, "loss": 0.009, "step": 173120 }, { "epoch": 110.34416826003824, "grad_norm": 0.28222838044166565, "learning_rate": 4.142852176767492e-05, "loss": 0.0073, "step": 173130 }, { "epoch": 110.35054174633524, "grad_norm": 0.35965415835380554, "learning_rate": 4.142309184484014e-05, "loss": 0.0166, "step": 173140 }, { "epoch": 110.35691523263225, "grad_norm": 0.7128010988235474, "learning_rate": 4.1417662026222884e-05, "loss": 0.0135, "step": 173150 }, { "epoch": 110.36328871892925, "grad_norm": 0.4735579192638397, "learning_rate": 4.141223231188913e-05, "loss": 0.0088, "step": 173160 }, { "epoch": 110.36966220522626, "grad_norm": 0.28517580032348633, "learning_rate": 4.140680270190489e-05, "loss": 0.0071, "step": 173170 }, { "epoch": 110.37603569152326, "grad_norm": 0.1983332335948944, "learning_rate": 4.140137319633609e-05, "loss": 0.0081, "step": 173180 }, { "epoch": 110.38240917782026, "grad_norm": 0.22564129531383514, "learning_rate": 4.139594379524872e-05, "loss": 0.0114, "step": 173190 }, { "epoch": 110.38878266411727, "grad_norm": 0.261639803647995, "learning_rate": 4.139051449870878e-05, "loss": 0.0072, "step": 173200 }, { "epoch": 110.39515615041428, "grad_norm": 0.3976747989654541, "learning_rate": 4.13850853067822e-05, "loss": 0.0135, "step": 173210 }, { "epoch": 110.40152963671127, "grad_norm": 0.1832655519247055, "learning_rate": 4.137965621953497e-05, "loss": 0.0134, "step": 173220 }, { "epoch": 110.40790312300828, "grad_norm": 0.15438468754291534, "learning_rate": 4.1374227237033044e-05, "loss": 0.0069, "step": 173230 }, { "epoch": 110.41427660930529, "grad_norm": 0.18552778661251068, "learning_rate": 4.136879835934243e-05, "loss": 0.012, "step": 173240 }, { "epoch": 110.4206500956023, "grad_norm": 0.21409502625465393, "learning_rate": 4.136336958652904e-05, "loss": 0.0119, "step": 173250 }, { "epoch": 110.4270235818993, "grad_norm": 0.24258585274219513, "learning_rate": 4.135794091865887e-05, "loss": 0.0114, "step": 173260 }, { "epoch": 110.4333970681963, "grad_norm": 0.36190447211265564, "learning_rate": 4.135251235579787e-05, "loss": 0.0063, "step": 173270 }, { "epoch": 110.43977055449331, "grad_norm": 0.5185433030128479, "learning_rate": 4.134708389801204e-05, "loss": 0.0172, "step": 173280 }, { "epoch": 110.44614404079032, "grad_norm": 0.11410941928625107, "learning_rate": 4.134165554536728e-05, "loss": 0.0119, "step": 173290 }, { "epoch": 110.45251752708732, "grad_norm": 0.2789957523345947, "learning_rate": 4.133622729792959e-05, "loss": 0.0086, "step": 173300 }, { "epoch": 110.45889101338432, "grad_norm": 0.40417179465293884, "learning_rate": 4.1330799155764925e-05, "loss": 0.0076, "step": 173310 }, { "epoch": 110.46526449968133, "grad_norm": 0.6420514583587646, "learning_rate": 4.1325371118939224e-05, "loss": 0.0182, "step": 173320 }, { "epoch": 110.47163798597833, "grad_norm": 0.22557948529720306, "learning_rate": 4.131994318751845e-05, "loss": 0.0104, "step": 173330 }, { "epoch": 110.47801147227534, "grad_norm": 0.2452649027109146, "learning_rate": 4.131451536156857e-05, "loss": 0.0069, "step": 173340 }, { "epoch": 110.48438495857233, "grad_norm": 0.3359207808971405, "learning_rate": 4.130908764115554e-05, "loss": 0.006, "step": 173350 }, { "epoch": 110.49075844486934, "grad_norm": 0.1981654316186905, "learning_rate": 4.130366002634528e-05, "loss": 0.0163, "step": 173360 }, { "epoch": 110.49713193116635, "grad_norm": 0.4589751958847046, "learning_rate": 4.1298232517203764e-05, "loss": 0.0096, "step": 173370 }, { "epoch": 110.50350541746336, "grad_norm": 0.24223503470420837, "learning_rate": 4.129280511379695e-05, "loss": 0.0128, "step": 173380 }, { "epoch": 110.50987890376035, "grad_norm": 0.11424202471971512, "learning_rate": 4.128737781619078e-05, "loss": 0.0092, "step": 173390 }, { "epoch": 110.51625239005736, "grad_norm": 0.41829782724380493, "learning_rate": 4.128195062445118e-05, "loss": 0.0143, "step": 173400 }, { "epoch": 110.52262587635437, "grad_norm": 0.10103035718202591, "learning_rate": 4.127652353864411e-05, "loss": 0.007, "step": 173410 }, { "epoch": 110.52899936265138, "grad_norm": 0.47525259852409363, "learning_rate": 4.1271096558835524e-05, "loss": 0.0133, "step": 173420 }, { "epoch": 110.53537284894837, "grad_norm": 0.3729122281074524, "learning_rate": 4.126566968509136e-05, "loss": 0.0139, "step": 173430 }, { "epoch": 110.54174633524538, "grad_norm": 0.3420347571372986, "learning_rate": 4.126024291747755e-05, "loss": 0.009, "step": 173440 }, { "epoch": 110.54811982154239, "grad_norm": 0.43644970655441284, "learning_rate": 4.125481625606004e-05, "loss": 0.0128, "step": 173450 }, { "epoch": 110.5544933078394, "grad_norm": 0.17394432425498962, "learning_rate": 4.124938970090477e-05, "loss": 0.0093, "step": 173460 }, { "epoch": 110.56086679413639, "grad_norm": 0.3915448486804962, "learning_rate": 4.1243963252077675e-05, "loss": 0.0085, "step": 173470 }, { "epoch": 110.5672402804334, "grad_norm": 0.5635557174682617, "learning_rate": 4.1238536909644694e-05, "loss": 0.0131, "step": 173480 }, { "epoch": 110.5736137667304, "grad_norm": 0.4040292203426361, "learning_rate": 4.123311067367177e-05, "loss": 0.0138, "step": 173490 }, { "epoch": 110.57998725302741, "grad_norm": 0.29873955249786377, "learning_rate": 4.1227684544224816e-05, "loss": 0.007, "step": 173500 }, { "epoch": 110.5863607393244, "grad_norm": 0.2269936501979828, "learning_rate": 4.1222258521369784e-05, "loss": 0.0055, "step": 173510 }, { "epoch": 110.59273422562141, "grad_norm": 0.3973376154899597, "learning_rate": 4.1216832605172605e-05, "loss": 0.0094, "step": 173520 }, { "epoch": 110.59910771191842, "grad_norm": 0.2770066261291504, "learning_rate": 4.1211406795699186e-05, "loss": 0.0066, "step": 173530 }, { "epoch": 110.60548119821543, "grad_norm": 0.30403897166252136, "learning_rate": 4.120598109301548e-05, "loss": 0.0102, "step": 173540 }, { "epoch": 110.61185468451242, "grad_norm": 0.4077003002166748, "learning_rate": 4.120055549718741e-05, "loss": 0.0084, "step": 173550 }, { "epoch": 110.61822817080943, "grad_norm": 0.38037192821502686, "learning_rate": 4.11951300082809e-05, "loss": 0.0093, "step": 173560 }, { "epoch": 110.62460165710644, "grad_norm": 0.30052316188812256, "learning_rate": 4.118970462636187e-05, "loss": 0.0115, "step": 173570 }, { "epoch": 110.63097514340345, "grad_norm": 0.1857544332742691, "learning_rate": 4.118427935149624e-05, "loss": 0.0245, "step": 173580 }, { "epoch": 110.63734862970044, "grad_norm": 0.17546743154525757, "learning_rate": 4.117885418374995e-05, "loss": 0.0089, "step": 173590 }, { "epoch": 110.64372211599745, "grad_norm": 0.2551175057888031, "learning_rate": 4.117342912318891e-05, "loss": 0.0138, "step": 173600 }, { "epoch": 110.65009560229446, "grad_norm": 0.237527534365654, "learning_rate": 4.116800416987903e-05, "loss": 0.0163, "step": 173610 }, { "epoch": 110.65646908859146, "grad_norm": 0.19116747379302979, "learning_rate": 4.116257932388624e-05, "loss": 0.0084, "step": 173620 }, { "epoch": 110.66284257488846, "grad_norm": 0.2677864134311676, "learning_rate": 4.1157154585276484e-05, "loss": 0.0089, "step": 173630 }, { "epoch": 110.66921606118547, "grad_norm": 0.16322354972362518, "learning_rate": 4.115172995411561e-05, "loss": 0.0071, "step": 173640 }, { "epoch": 110.67558954748247, "grad_norm": 0.24484673142433167, "learning_rate": 4.114630543046958e-05, "loss": 0.0173, "step": 173650 }, { "epoch": 110.68196303377948, "grad_norm": 0.2014627456665039, "learning_rate": 4.11408810144043e-05, "loss": 0.0051, "step": 173660 }, { "epoch": 110.68833652007648, "grad_norm": 0.537994921207428, "learning_rate": 4.1135456705985704e-05, "loss": 0.0138, "step": 173670 }, { "epoch": 110.69471000637348, "grad_norm": 0.21087758243083954, "learning_rate": 4.1130032505279644e-05, "loss": 0.0067, "step": 173680 }, { "epoch": 110.70108349267049, "grad_norm": 0.29069098830223083, "learning_rate": 4.1124608412352076e-05, "loss": 0.01, "step": 173690 }, { "epoch": 110.7074569789675, "grad_norm": 0.39698049426078796, "learning_rate": 4.111918442726892e-05, "loss": 0.0125, "step": 173700 }, { "epoch": 110.7138304652645, "grad_norm": 0.3344404697418213, "learning_rate": 4.1113760550096023e-05, "loss": 0.0149, "step": 173710 }, { "epoch": 110.7202039515615, "grad_norm": 0.1885492503643036, "learning_rate": 4.110833678089934e-05, "loss": 0.0107, "step": 173720 }, { "epoch": 110.72657743785851, "grad_norm": 0.21057160198688507, "learning_rate": 4.1102913119744754e-05, "loss": 0.0065, "step": 173730 }, { "epoch": 110.73295092415552, "grad_norm": 0.35860171914100647, "learning_rate": 4.1097489566698185e-05, "loss": 0.0153, "step": 173740 }, { "epoch": 110.73932441045251, "grad_norm": 0.32900235056877136, "learning_rate": 4.1092066121825516e-05, "loss": 0.0094, "step": 173750 }, { "epoch": 110.74569789674952, "grad_norm": 0.21436460316181183, "learning_rate": 4.108664278519265e-05, "loss": 0.0245, "step": 173760 }, { "epoch": 110.75207138304653, "grad_norm": 0.13988441228866577, "learning_rate": 4.1081219556865515e-05, "loss": 0.0167, "step": 173770 }, { "epoch": 110.75844486934353, "grad_norm": 0.2826445400714874, "learning_rate": 4.1075796436909955e-05, "loss": 0.0124, "step": 173780 }, { "epoch": 110.76481835564053, "grad_norm": 0.27706053853034973, "learning_rate": 4.10703734253919e-05, "loss": 0.0098, "step": 173790 }, { "epoch": 110.77119184193754, "grad_norm": 0.31398582458496094, "learning_rate": 4.106495052237723e-05, "loss": 0.0112, "step": 173800 }, { "epoch": 110.77756532823454, "grad_norm": 0.16802220046520233, "learning_rate": 4.105952772793187e-05, "loss": 0.0125, "step": 173810 }, { "epoch": 110.78393881453155, "grad_norm": 0.4520033299922943, "learning_rate": 4.105410504212167e-05, "loss": 0.0102, "step": 173820 }, { "epoch": 110.79031230082855, "grad_norm": 0.31608840823173523, "learning_rate": 4.1048682465012546e-05, "loss": 0.0108, "step": 173830 }, { "epoch": 110.79668578712555, "grad_norm": 0.23395900428295135, "learning_rate": 4.104325999667037e-05, "loss": 0.0123, "step": 173840 }, { "epoch": 110.80305927342256, "grad_norm": 0.28533056378364563, "learning_rate": 4.1037837637161076e-05, "loss": 0.0096, "step": 173850 }, { "epoch": 110.80943275971957, "grad_norm": 0.2525097131729126, "learning_rate": 4.103241538655049e-05, "loss": 0.0149, "step": 173860 }, { "epoch": 110.81580624601658, "grad_norm": 0.2730976939201355, "learning_rate": 4.102699324490452e-05, "loss": 0.0137, "step": 173870 }, { "epoch": 110.82217973231357, "grad_norm": 0.35257846117019653, "learning_rate": 4.102157121228908e-05, "loss": 0.0142, "step": 173880 }, { "epoch": 110.82855321861058, "grad_norm": 0.6145070195198059, "learning_rate": 4.101614928877001e-05, "loss": 0.0115, "step": 173890 }, { "epoch": 110.83492670490759, "grad_norm": 0.2778039872646332, "learning_rate": 4.10107274744132e-05, "loss": 0.0084, "step": 173900 }, { "epoch": 110.8413001912046, "grad_norm": 0.2601245939731598, "learning_rate": 4.100530576928454e-05, "loss": 0.0095, "step": 173910 }, { "epoch": 110.84767367750159, "grad_norm": 0.3198692202568054, "learning_rate": 4.099988417344993e-05, "loss": 0.0178, "step": 173920 }, { "epoch": 110.8540471637986, "grad_norm": 0.4306259751319885, "learning_rate": 4.09944626869752e-05, "loss": 0.0127, "step": 173930 }, { "epoch": 110.8604206500956, "grad_norm": 0.2733701169490814, "learning_rate": 4.0989041309926254e-05, "loss": 0.0067, "step": 173940 }, { "epoch": 110.86679413639261, "grad_norm": 0.3376750946044922, "learning_rate": 4.098362004236899e-05, "loss": 0.0093, "step": 173950 }, { "epoch": 110.8731676226896, "grad_norm": 0.3247934877872467, "learning_rate": 4.0978198884369226e-05, "loss": 0.0083, "step": 173960 }, { "epoch": 110.87954110898661, "grad_norm": 0.38264867663383484, "learning_rate": 4.097277783599287e-05, "loss": 0.0093, "step": 173970 }, { "epoch": 110.88591459528362, "grad_norm": 0.43965867161750793, "learning_rate": 4.096735689730577e-05, "loss": 0.0257, "step": 173980 }, { "epoch": 110.89228808158063, "grad_norm": 0.3879861533641815, "learning_rate": 4.096193606837385e-05, "loss": 0.0151, "step": 173990 }, { "epoch": 110.89866156787762, "grad_norm": 0.30821120738983154, "learning_rate": 4.095651534926292e-05, "loss": 0.0152, "step": 174000 }, { "epoch": 110.90503505417463, "grad_norm": 0.17563052475452423, "learning_rate": 4.095109474003886e-05, "loss": 0.0092, "step": 174010 }, { "epoch": 110.91140854047164, "grad_norm": 0.17665360867977142, "learning_rate": 4.094567424076756e-05, "loss": 0.0056, "step": 174020 }, { "epoch": 110.91778202676865, "grad_norm": 0.28957098722457886, "learning_rate": 4.094025385151485e-05, "loss": 0.0129, "step": 174030 }, { "epoch": 110.92415551306564, "grad_norm": 0.30644601583480835, "learning_rate": 4.093483357234662e-05, "loss": 0.0151, "step": 174040 }, { "epoch": 110.93052899936265, "grad_norm": 0.24226413667201996, "learning_rate": 4.092941340332871e-05, "loss": 0.0213, "step": 174050 }, { "epoch": 110.93690248565966, "grad_norm": 0.48290953040122986, "learning_rate": 4.092399334452701e-05, "loss": 0.0113, "step": 174060 }, { "epoch": 110.94327597195667, "grad_norm": 0.15204881131649017, "learning_rate": 4.091857339600734e-05, "loss": 0.0086, "step": 174070 }, { "epoch": 110.94964945825366, "grad_norm": 0.2425958812236786, "learning_rate": 4.091315355783558e-05, "loss": 0.0134, "step": 174080 }, { "epoch": 110.95602294455067, "grad_norm": 0.2802336513996124, "learning_rate": 4.09077338300776e-05, "loss": 0.0126, "step": 174090 }, { "epoch": 110.96239643084768, "grad_norm": 0.23078888654708862, "learning_rate": 4.090231421279923e-05, "loss": 0.0109, "step": 174100 }, { "epoch": 110.96876991714468, "grad_norm": 0.4412590265274048, "learning_rate": 4.0896894706066324e-05, "loss": 0.0163, "step": 174110 }, { "epoch": 110.97514340344168, "grad_norm": 0.5573789477348328, "learning_rate": 4.089147530994474e-05, "loss": 0.0121, "step": 174120 }, { "epoch": 110.98151688973869, "grad_norm": 0.19921037554740906, "learning_rate": 4.0886056024500355e-05, "loss": 0.009, "step": 174130 }, { "epoch": 110.9878903760357, "grad_norm": 0.1172899678349495, "learning_rate": 4.088063684979898e-05, "loss": 0.0111, "step": 174140 }, { "epoch": 110.9942638623327, "grad_norm": 0.26665762066841125, "learning_rate": 4.087521778590647e-05, "loss": 0.0073, "step": 174150 }, { "epoch": 111.0006373486297, "grad_norm": 0.2004757821559906, "learning_rate": 4.086979883288867e-05, "loss": 0.0061, "step": 174160 }, { "epoch": 111.0070108349267, "grad_norm": 0.295284628868103, "learning_rate": 4.086437999081146e-05, "loss": 0.0123, "step": 174170 }, { "epoch": 111.01338432122371, "grad_norm": 0.2269737869501114, "learning_rate": 4.085896125974065e-05, "loss": 0.0163, "step": 174180 }, { "epoch": 111.01975780752072, "grad_norm": 0.19258292019367218, "learning_rate": 4.085354263974209e-05, "loss": 0.0118, "step": 174190 }, { "epoch": 111.02613129381771, "grad_norm": 0.5036726593971252, "learning_rate": 4.0848124130881625e-05, "loss": 0.0077, "step": 174200 }, { "epoch": 111.03250478011472, "grad_norm": 0.26809072494506836, "learning_rate": 4.084270573322509e-05, "loss": 0.0089, "step": 174210 }, { "epoch": 111.03887826641173, "grad_norm": 0.5431151986122131, "learning_rate": 4.083728744683833e-05, "loss": 0.0111, "step": 174220 }, { "epoch": 111.04525175270874, "grad_norm": 0.3863450288772583, "learning_rate": 4.0831869271787157e-05, "loss": 0.0193, "step": 174230 }, { "epoch": 111.05162523900573, "grad_norm": 0.2613341510295868, "learning_rate": 4.0826451208137454e-05, "loss": 0.0071, "step": 174240 }, { "epoch": 111.05799872530274, "grad_norm": 0.4046638607978821, "learning_rate": 4.082103325595502e-05, "loss": 0.0077, "step": 174250 }, { "epoch": 111.06437221159975, "grad_norm": 0.27210232615470886, "learning_rate": 4.08156154153057e-05, "loss": 0.0085, "step": 174260 }, { "epoch": 111.07074569789675, "grad_norm": 0.4678545892238617, "learning_rate": 4.081019768625533e-05, "loss": 0.0161, "step": 174270 }, { "epoch": 111.07711918419375, "grad_norm": 0.26334813237190247, "learning_rate": 4.0804780068869726e-05, "loss": 0.018, "step": 174280 }, { "epoch": 111.08349267049076, "grad_norm": 0.252785861492157, "learning_rate": 4.079936256321472e-05, "loss": 0.0071, "step": 174290 }, { "epoch": 111.08986615678776, "grad_norm": 0.3081073462963104, "learning_rate": 4.079394516935616e-05, "loss": 0.0076, "step": 174300 }, { "epoch": 111.09623964308477, "grad_norm": 0.1896023005247116, "learning_rate": 4.078852788735986e-05, "loss": 0.0076, "step": 174310 }, { "epoch": 111.10261312938177, "grad_norm": 0.25839897990226746, "learning_rate": 4.078311071729163e-05, "loss": 0.0079, "step": 174320 }, { "epoch": 111.10898661567877, "grad_norm": 0.24875114858150482, "learning_rate": 4.077769365921731e-05, "loss": 0.0063, "step": 174330 }, { "epoch": 111.11536010197578, "grad_norm": 0.3962564766407013, "learning_rate": 4.0772276713202735e-05, "loss": 0.0093, "step": 174340 }, { "epoch": 111.12173358827279, "grad_norm": 0.4850078523159027, "learning_rate": 4.076685987931369e-05, "loss": 0.012, "step": 174350 }, { "epoch": 111.12810707456978, "grad_norm": 0.2075565606355667, "learning_rate": 4.0761443157616016e-05, "loss": 0.0124, "step": 174360 }, { "epoch": 111.13448056086679, "grad_norm": 0.5085026025772095, "learning_rate": 4.075602654817554e-05, "loss": 0.0112, "step": 174370 }, { "epoch": 111.1408540471638, "grad_norm": 0.11526487022638321, "learning_rate": 4.075061005105807e-05, "loss": 0.0126, "step": 174380 }, { "epoch": 111.1472275334608, "grad_norm": 0.3309567868709564, "learning_rate": 4.074519366632942e-05, "loss": 0.0091, "step": 174390 }, { "epoch": 111.1536010197578, "grad_norm": 0.18009887635707855, "learning_rate": 4.07397773940554e-05, "loss": 0.007, "step": 174400 }, { "epoch": 111.15997450605481, "grad_norm": 0.27830857038497925, "learning_rate": 4.0734361234301837e-05, "loss": 0.0149, "step": 174410 }, { "epoch": 111.16634799235182, "grad_norm": 0.22171884775161743, "learning_rate": 4.072894518713452e-05, "loss": 0.0098, "step": 174420 }, { "epoch": 111.17272147864882, "grad_norm": 0.13336969912052155, "learning_rate": 4.072352925261927e-05, "loss": 0.0103, "step": 174430 }, { "epoch": 111.17909496494582, "grad_norm": 0.40126973390579224, "learning_rate": 4.0718113430821904e-05, "loss": 0.0092, "step": 174440 }, { "epoch": 111.18546845124283, "grad_norm": 0.1856762021780014, "learning_rate": 4.071269772180824e-05, "loss": 0.0058, "step": 174450 }, { "epoch": 111.19184193753983, "grad_norm": 0.2363644689321518, "learning_rate": 4.070728212564405e-05, "loss": 0.0148, "step": 174460 }, { "epoch": 111.19821542383684, "grad_norm": 0.37764424085617065, "learning_rate": 4.070186664239515e-05, "loss": 0.0179, "step": 174470 }, { "epoch": 111.20458891013385, "grad_norm": 0.5003322958946228, "learning_rate": 4.069645127212736e-05, "loss": 0.0082, "step": 174480 }, { "epoch": 111.21096239643084, "grad_norm": 0.4906109869480133, "learning_rate": 4.069103601490648e-05, "loss": 0.0101, "step": 174490 }, { "epoch": 111.21733588272785, "grad_norm": 0.3006395399570465, "learning_rate": 4.068562087079828e-05, "loss": 0.0094, "step": 174500 }, { "epoch": 111.22370936902486, "grad_norm": 0.4113823175430298, "learning_rate": 4.068020583986859e-05, "loss": 0.0099, "step": 174510 }, { "epoch": 111.23008285532187, "grad_norm": 0.21026413142681122, "learning_rate": 4.067479092218321e-05, "loss": 0.0171, "step": 174520 }, { "epoch": 111.23645634161886, "grad_norm": 0.13853681087493896, "learning_rate": 4.066937611780791e-05, "loss": 0.0116, "step": 174530 }, { "epoch": 111.24282982791587, "grad_norm": 0.3767240345478058, "learning_rate": 4.06639614268085e-05, "loss": 0.0074, "step": 174540 }, { "epoch": 111.24920331421288, "grad_norm": 0.4122052490711212, "learning_rate": 4.0658546849250774e-05, "loss": 0.0185, "step": 174550 }, { "epoch": 111.25557680050989, "grad_norm": 0.2512182593345642, "learning_rate": 4.0653132385200545e-05, "loss": 0.0091, "step": 174560 }, { "epoch": 111.26195028680688, "grad_norm": 0.31465861201286316, "learning_rate": 4.0647718034723556e-05, "loss": 0.0129, "step": 174570 }, { "epoch": 111.26832377310389, "grad_norm": 0.4239789545536041, "learning_rate": 4.0642303797885626e-05, "loss": 0.0181, "step": 174580 }, { "epoch": 111.2746972594009, "grad_norm": 0.3654109239578247, "learning_rate": 4.063688967475256e-05, "loss": 0.0085, "step": 174590 }, { "epoch": 111.2810707456979, "grad_norm": 0.16224771738052368, "learning_rate": 4.06314756653901e-05, "loss": 0.0191, "step": 174600 }, { "epoch": 111.2874442319949, "grad_norm": 0.6480357050895691, "learning_rate": 4.0626061769864065e-05, "loss": 0.0124, "step": 174610 }, { "epoch": 111.2938177182919, "grad_norm": 0.15860216319561005, "learning_rate": 4.062064798824022e-05, "loss": 0.0072, "step": 174620 }, { "epoch": 111.30019120458891, "grad_norm": 0.5867629051208496, "learning_rate": 4.061523432058439e-05, "loss": 0.0136, "step": 174630 }, { "epoch": 111.30656469088592, "grad_norm": 0.6278953552246094, "learning_rate": 4.06098207669623e-05, "loss": 0.0177, "step": 174640 }, { "epoch": 111.31293817718291, "grad_norm": 0.21906651556491852, "learning_rate": 4.060440732743975e-05, "loss": 0.0119, "step": 174650 }, { "epoch": 111.31931166347992, "grad_norm": 0.39359354972839355, "learning_rate": 4.059899400208254e-05, "loss": 0.0101, "step": 174660 }, { "epoch": 111.32568514977693, "grad_norm": 0.5734029412269592, "learning_rate": 4.059358079095641e-05, "loss": 0.0166, "step": 174670 }, { "epoch": 111.33205863607394, "grad_norm": 0.3298412263393402, "learning_rate": 4.058816769412716e-05, "loss": 0.0115, "step": 174680 }, { "epoch": 111.33843212237093, "grad_norm": 0.2708350718021393, "learning_rate": 4.058275471166055e-05, "loss": 0.0097, "step": 174690 }, { "epoch": 111.34480560866794, "grad_norm": 0.3750899136066437, "learning_rate": 4.057734184362239e-05, "loss": 0.0153, "step": 174700 }, { "epoch": 111.35117909496495, "grad_norm": 0.27463197708129883, "learning_rate": 4.05719290900784e-05, "loss": 0.0088, "step": 174710 }, { "epoch": 111.35755258126196, "grad_norm": 0.2922426462173462, "learning_rate": 4.0566516451094364e-05, "loss": 0.0095, "step": 174720 }, { "epoch": 111.36392606755895, "grad_norm": 0.6567325592041016, "learning_rate": 4.056110392673609e-05, "loss": 0.0122, "step": 174730 }, { "epoch": 111.37029955385596, "grad_norm": 0.2330668866634369, "learning_rate": 4.0555691517069295e-05, "loss": 0.0137, "step": 174740 }, { "epoch": 111.37667304015297, "grad_norm": 0.21467433869838715, "learning_rate": 4.055027922215977e-05, "loss": 0.0091, "step": 174750 }, { "epoch": 111.38304652644997, "grad_norm": 0.7934666275978088, "learning_rate": 4.054486704207326e-05, "loss": 0.0172, "step": 174760 }, { "epoch": 111.38942001274697, "grad_norm": 0.21955260634422302, "learning_rate": 4.053945497687557e-05, "loss": 0.0132, "step": 174770 }, { "epoch": 111.39579349904398, "grad_norm": 0.33279702067375183, "learning_rate": 4.0534043026632416e-05, "loss": 0.0203, "step": 174780 }, { "epoch": 111.40216698534098, "grad_norm": 0.7015087008476257, "learning_rate": 4.052863119140957e-05, "loss": 0.0242, "step": 174790 }, { "epoch": 111.40854047163799, "grad_norm": 0.259308785200119, "learning_rate": 4.0523219471272805e-05, "loss": 0.0072, "step": 174800 }, { "epoch": 111.41491395793498, "grad_norm": 0.16835501790046692, "learning_rate": 4.0517807866287896e-05, "loss": 0.0113, "step": 174810 }, { "epoch": 111.421287444232, "grad_norm": 0.27920112013816833, "learning_rate": 4.051239637652054e-05, "loss": 0.0068, "step": 174820 }, { "epoch": 111.427660930529, "grad_norm": 0.40618813037872314, "learning_rate": 4.0506985002036534e-05, "loss": 0.0093, "step": 174830 }, { "epoch": 111.43403441682601, "grad_norm": 0.5462485551834106, "learning_rate": 4.0501573742901635e-05, "loss": 0.0253, "step": 174840 }, { "epoch": 111.440407903123, "grad_norm": 0.3404155373573303, "learning_rate": 4.0496162599181576e-05, "loss": 0.0134, "step": 174850 }, { "epoch": 111.44678138942001, "grad_norm": 0.2710186839103699, "learning_rate": 4.049075157094211e-05, "loss": 0.0103, "step": 174860 }, { "epoch": 111.45315487571702, "grad_norm": 0.27113085985183716, "learning_rate": 4.048534065824898e-05, "loss": 0.0136, "step": 174870 }, { "epoch": 111.45952836201403, "grad_norm": 0.24910013377666473, "learning_rate": 4.047992986116797e-05, "loss": 0.0098, "step": 174880 }, { "epoch": 111.46590184831102, "grad_norm": 0.25256815552711487, "learning_rate": 4.047451917976479e-05, "loss": 0.0104, "step": 174890 }, { "epoch": 111.47227533460803, "grad_norm": 0.3062129020690918, "learning_rate": 4.0469108614105186e-05, "loss": 0.0082, "step": 174900 }, { "epoch": 111.47864882090504, "grad_norm": 0.3178284466266632, "learning_rate": 4.0463698164254926e-05, "loss": 0.0121, "step": 174910 }, { "epoch": 111.48502230720204, "grad_norm": 0.19649584591388702, "learning_rate": 4.045828783027973e-05, "loss": 0.0134, "step": 174920 }, { "epoch": 111.49139579349904, "grad_norm": 0.2591179609298706, "learning_rate": 4.045287761224535e-05, "loss": 0.0144, "step": 174930 }, { "epoch": 111.49776927979605, "grad_norm": 0.2426328957080841, "learning_rate": 4.0447467510217504e-05, "loss": 0.0098, "step": 174940 }, { "epoch": 111.50414276609305, "grad_norm": 0.25797712802886963, "learning_rate": 4.0442057524261975e-05, "loss": 0.015, "step": 174950 }, { "epoch": 111.51051625239006, "grad_norm": 0.23340006172657013, "learning_rate": 4.043664765444445e-05, "loss": 0.0109, "step": 174960 }, { "epoch": 111.51688973868706, "grad_norm": 0.2747410237789154, "learning_rate": 4.0431237900830694e-05, "loss": 0.0106, "step": 174970 }, { "epoch": 111.52326322498406, "grad_norm": 0.26155680418014526, "learning_rate": 4.042582826348643e-05, "loss": 0.0144, "step": 174980 }, { "epoch": 111.52963671128107, "grad_norm": 0.23910878598690033, "learning_rate": 4.042041874247739e-05, "loss": 0.0105, "step": 174990 }, { "epoch": 111.53601019757808, "grad_norm": 0.1787520796060562, "learning_rate": 4.04150093378693e-05, "loss": 0.0127, "step": 175000 }, { "epoch": 111.54238368387507, "grad_norm": 0.3940960764884949, "learning_rate": 4.0409600049727915e-05, "loss": 0.0102, "step": 175010 }, { "epoch": 111.54875717017208, "grad_norm": 0.2469814419746399, "learning_rate": 4.0404190878118936e-05, "loss": 0.0135, "step": 175020 }, { "epoch": 111.55513065646909, "grad_norm": 0.21339267492294312, "learning_rate": 4.0398781823108086e-05, "loss": 0.0084, "step": 175030 }, { "epoch": 111.5615041427661, "grad_norm": 0.2226686030626297, "learning_rate": 4.039337288476111e-05, "loss": 0.0125, "step": 175040 }, { "epoch": 111.56787762906309, "grad_norm": 0.3818334639072418, "learning_rate": 4.0387964063143726e-05, "loss": 0.01, "step": 175050 }, { "epoch": 111.5742511153601, "grad_norm": 0.22918444871902466, "learning_rate": 4.038255535832166e-05, "loss": 0.0123, "step": 175060 }, { "epoch": 111.5806246016571, "grad_norm": 0.600343644618988, "learning_rate": 4.037714677036062e-05, "loss": 0.0107, "step": 175070 }, { "epoch": 111.58699808795411, "grad_norm": 0.3764926493167877, "learning_rate": 4.037173829932632e-05, "loss": 0.0079, "step": 175080 }, { "epoch": 111.59337157425111, "grad_norm": 0.4205983579158783, "learning_rate": 4.036632994528452e-05, "loss": 0.0136, "step": 175090 }, { "epoch": 111.59974506054812, "grad_norm": 0.5099342465400696, "learning_rate": 4.036092170830089e-05, "loss": 0.0125, "step": 175100 }, { "epoch": 111.60611854684512, "grad_norm": 0.2527737319469452, "learning_rate": 4.035551358844115e-05, "loss": 0.0182, "step": 175110 }, { "epoch": 111.61249203314213, "grad_norm": 0.2515697479248047, "learning_rate": 4.0350105585771034e-05, "loss": 0.0075, "step": 175120 }, { "epoch": 111.61886551943914, "grad_norm": 0.2630361020565033, "learning_rate": 4.034469770035627e-05, "loss": 0.0068, "step": 175130 }, { "epoch": 111.62523900573613, "grad_norm": 0.48258763551712036, "learning_rate": 4.033928993226251e-05, "loss": 0.0069, "step": 175140 }, { "epoch": 111.63161249203314, "grad_norm": 0.1900261789560318, "learning_rate": 4.0333882281555514e-05, "loss": 0.0122, "step": 175150 }, { "epoch": 111.63798597833015, "grad_norm": 0.4577310085296631, "learning_rate": 4.0328474748300985e-05, "loss": 0.0189, "step": 175160 }, { "epoch": 111.64435946462716, "grad_norm": 0.4591793119907379, "learning_rate": 4.03230673325646e-05, "loss": 0.0127, "step": 175170 }, { "epoch": 111.65073295092415, "grad_norm": 0.22874733805656433, "learning_rate": 4.031766003441209e-05, "loss": 0.0083, "step": 175180 }, { "epoch": 111.65710643722116, "grad_norm": 0.15694664418697357, "learning_rate": 4.0312252853909146e-05, "loss": 0.0092, "step": 175190 }, { "epoch": 111.66347992351817, "grad_norm": 0.29540151357650757, "learning_rate": 4.03068457911215e-05, "loss": 0.0077, "step": 175200 }, { "epoch": 111.66985340981518, "grad_norm": 0.6849016547203064, "learning_rate": 4.030143884611481e-05, "loss": 0.0157, "step": 175210 }, { "epoch": 111.67622689611217, "grad_norm": 0.3555660843849182, "learning_rate": 4.029603201895479e-05, "loss": 0.01, "step": 175220 }, { "epoch": 111.68260038240918, "grad_norm": 0.5894003510475159, "learning_rate": 4.029062530970717e-05, "loss": 0.0152, "step": 175230 }, { "epoch": 111.68897386870619, "grad_norm": 0.46851426362991333, "learning_rate": 4.028521871843759e-05, "loss": 0.0088, "step": 175240 }, { "epoch": 111.6953473550032, "grad_norm": 0.2603919208049774, "learning_rate": 4.027981224521178e-05, "loss": 0.007, "step": 175250 }, { "epoch": 111.70172084130019, "grad_norm": 0.5473756790161133, "learning_rate": 4.0274405890095434e-05, "loss": 0.014, "step": 175260 }, { "epoch": 111.7080943275972, "grad_norm": 0.45269736647605896, "learning_rate": 4.0268999653154255e-05, "loss": 0.0083, "step": 175270 }, { "epoch": 111.7144678138942, "grad_norm": 0.27014854550361633, "learning_rate": 4.026359353445389e-05, "loss": 0.0128, "step": 175280 }, { "epoch": 111.72084130019121, "grad_norm": 0.23327229917049408, "learning_rate": 4.0258187534060065e-05, "loss": 0.0087, "step": 175290 }, { "epoch": 111.7272147864882, "grad_norm": 0.3196769058704376, "learning_rate": 4.0252781652038485e-05, "loss": 0.0111, "step": 175300 }, { "epoch": 111.73358827278521, "grad_norm": 0.38244491815567017, "learning_rate": 4.024737588845478e-05, "loss": 0.0135, "step": 175310 }, { "epoch": 111.73996175908222, "grad_norm": 0.4752812087535858, "learning_rate": 4.024197024337467e-05, "loss": 0.0205, "step": 175320 }, { "epoch": 111.74633524537923, "grad_norm": 0.1644783765077591, "learning_rate": 4.023656471686383e-05, "loss": 0.0142, "step": 175330 }, { "epoch": 111.75270873167622, "grad_norm": 0.3740662634372711, "learning_rate": 4.023115930898797e-05, "loss": 0.0074, "step": 175340 }, { "epoch": 111.75908221797323, "grad_norm": 0.1652674823999405, "learning_rate": 4.0225754019812735e-05, "loss": 0.0103, "step": 175350 }, { "epoch": 111.76545570427024, "grad_norm": 0.17166903614997864, "learning_rate": 4.02203488494038e-05, "loss": 0.0083, "step": 175360 }, { "epoch": 111.77182919056725, "grad_norm": 0.5038650035858154, "learning_rate": 4.021494379782688e-05, "loss": 0.0081, "step": 175370 }, { "epoch": 111.77820267686424, "grad_norm": 0.14139890670776367, "learning_rate": 4.020953886514764e-05, "loss": 0.01, "step": 175380 }, { "epoch": 111.78457616316125, "grad_norm": 0.16757066547870636, "learning_rate": 4.020413405143173e-05, "loss": 0.0126, "step": 175390 }, { "epoch": 111.79094964945826, "grad_norm": 0.24506807327270508, "learning_rate": 4.019872935674484e-05, "loss": 0.0145, "step": 175400 }, { "epoch": 111.79732313575526, "grad_norm": 0.48308929800987244, "learning_rate": 4.0193324781152655e-05, "loss": 0.0107, "step": 175410 }, { "epoch": 111.80369662205226, "grad_norm": 0.3951493203639984, "learning_rate": 4.0187920324720824e-05, "loss": 0.0102, "step": 175420 }, { "epoch": 111.81007010834927, "grad_norm": 0.35153549909591675, "learning_rate": 4.018251598751502e-05, "loss": 0.0118, "step": 175430 }, { "epoch": 111.81644359464627, "grad_norm": 0.3353115916252136, "learning_rate": 4.017711176960092e-05, "loss": 0.0165, "step": 175440 }, { "epoch": 111.82281708094328, "grad_norm": 0.2945771813392639, "learning_rate": 4.01717076710442e-05, "loss": 0.0127, "step": 175450 }, { "epoch": 111.82919056724027, "grad_norm": 0.385074645280838, "learning_rate": 4.016630369191049e-05, "loss": 0.0115, "step": 175460 }, { "epoch": 111.83556405353728, "grad_norm": 0.5766335725784302, "learning_rate": 4.016089983226548e-05, "loss": 0.0179, "step": 175470 }, { "epoch": 111.84193753983429, "grad_norm": 0.3882495164871216, "learning_rate": 4.0155496092174845e-05, "loss": 0.019, "step": 175480 }, { "epoch": 111.8483110261313, "grad_norm": 0.1418733298778534, "learning_rate": 4.0150092471704204e-05, "loss": 0.0129, "step": 175490 }, { "epoch": 111.85468451242829, "grad_norm": 0.8369735479354858, "learning_rate": 4.014468897091924e-05, "loss": 0.0178, "step": 175500 }, { "epoch": 111.8610579987253, "grad_norm": 0.5654796957969666, "learning_rate": 4.013928558988562e-05, "loss": 0.0127, "step": 175510 }, { "epoch": 111.86743148502231, "grad_norm": 0.23696261644363403, "learning_rate": 4.013388232866901e-05, "loss": 0.0073, "step": 175520 }, { "epoch": 111.87380497131932, "grad_norm": 0.2580890357494354, "learning_rate": 4.0128479187335014e-05, "loss": 0.0111, "step": 175530 }, { "epoch": 111.88017845761631, "grad_norm": 0.23214469850063324, "learning_rate": 4.0123076165949316e-05, "loss": 0.0084, "step": 175540 }, { "epoch": 111.88655194391332, "grad_norm": 0.2302417904138565, "learning_rate": 4.011767326457758e-05, "loss": 0.0085, "step": 175550 }, { "epoch": 111.89292543021033, "grad_norm": 0.31375178694725037, "learning_rate": 4.0112270483285446e-05, "loss": 0.0146, "step": 175560 }, { "epoch": 111.89929891650733, "grad_norm": 0.20101439952850342, "learning_rate": 4.010686782213855e-05, "loss": 0.0123, "step": 175570 }, { "epoch": 111.90567240280433, "grad_norm": 0.3931977152824402, "learning_rate": 4.010146528120256e-05, "loss": 0.012, "step": 175580 }, { "epoch": 111.91204588910134, "grad_norm": 0.18691349029541016, "learning_rate": 4.00960628605431e-05, "loss": 0.0092, "step": 175590 }, { "epoch": 111.91841937539834, "grad_norm": 0.1841801106929779, "learning_rate": 4.0090660560225846e-05, "loss": 0.0084, "step": 175600 }, { "epoch": 111.92479286169535, "grad_norm": 0.14194723963737488, "learning_rate": 4.008525838031641e-05, "loss": 0.0116, "step": 175610 }, { "epoch": 111.93116634799235, "grad_norm": 0.32521161437034607, "learning_rate": 4.0079856320880457e-05, "loss": 0.0138, "step": 175620 }, { "epoch": 111.93753983428935, "grad_norm": 0.6390588283538818, "learning_rate": 4.00744543819836e-05, "loss": 0.0112, "step": 175630 }, { "epoch": 111.94391332058636, "grad_norm": 0.24557508528232574, "learning_rate": 4.00690525636915e-05, "loss": 0.0098, "step": 175640 }, { "epoch": 111.95028680688337, "grad_norm": 0.30528050661087036, "learning_rate": 4.006365086606979e-05, "loss": 0.0106, "step": 175650 }, { "epoch": 111.95666029318036, "grad_norm": 0.2345661222934723, "learning_rate": 4.0058249289184105e-05, "loss": 0.0102, "step": 175660 }, { "epoch": 111.96303377947737, "grad_norm": 0.2306177020072937, "learning_rate": 4.0052847833100074e-05, "loss": 0.0135, "step": 175670 }, { "epoch": 111.96940726577438, "grad_norm": 0.35815784335136414, "learning_rate": 4.0047446497883335e-05, "loss": 0.0106, "step": 175680 }, { "epoch": 111.97578075207139, "grad_norm": 0.1586458534002304, "learning_rate": 4.004204528359951e-05, "loss": 0.0081, "step": 175690 }, { "epoch": 111.98215423836838, "grad_norm": 0.20372185111045837, "learning_rate": 4.003664419031425e-05, "loss": 0.0133, "step": 175700 }, { "epoch": 111.98852772466539, "grad_norm": 0.317695677280426, "learning_rate": 4.0031243218093164e-05, "loss": 0.013, "step": 175710 }, { "epoch": 111.9949012109624, "grad_norm": 0.24991784989833832, "learning_rate": 4.002584236700189e-05, "loss": 0.0113, "step": 175720 }, { "epoch": 112.0012746972594, "grad_norm": 0.2828601002693176, "learning_rate": 4.002044163710604e-05, "loss": 0.0083, "step": 175730 }, { "epoch": 112.00764818355641, "grad_norm": 0.4789600968360901, "learning_rate": 4.0015041028471245e-05, "loss": 0.0206, "step": 175740 }, { "epoch": 112.0140216698534, "grad_norm": 0.45655596256256104, "learning_rate": 4.000964054116313e-05, "loss": 0.0126, "step": 175750 }, { "epoch": 112.02039515615041, "grad_norm": 0.41869810223579407, "learning_rate": 4.0004240175247324e-05, "loss": 0.0125, "step": 175760 }, { "epoch": 112.02676864244742, "grad_norm": 0.22168120741844177, "learning_rate": 3.9998839930789436e-05, "loss": 0.011, "step": 175770 }, { "epoch": 112.03314212874443, "grad_norm": 0.4799032211303711, "learning_rate": 3.9993439807855074e-05, "loss": 0.0132, "step": 175780 }, { "epoch": 112.03951561504142, "grad_norm": 0.3762616217136383, "learning_rate": 3.998803980650987e-05, "loss": 0.0134, "step": 175790 }, { "epoch": 112.04588910133843, "grad_norm": 0.15323702991008759, "learning_rate": 3.998263992681945e-05, "loss": 0.0111, "step": 175800 }, { "epoch": 112.05226258763544, "grad_norm": 0.6325253248214722, "learning_rate": 3.99772401688494e-05, "loss": 0.0132, "step": 175810 }, { "epoch": 112.05863607393245, "grad_norm": 0.445814847946167, "learning_rate": 3.997184053266534e-05, "loss": 0.0182, "step": 175820 }, { "epoch": 112.06500956022944, "grad_norm": 0.23984266817569733, "learning_rate": 3.996644101833289e-05, "loss": 0.0108, "step": 175830 }, { "epoch": 112.07138304652645, "grad_norm": 0.18162307143211365, "learning_rate": 3.996104162591767e-05, "loss": 0.0162, "step": 175840 }, { "epoch": 112.07775653282346, "grad_norm": 0.19113872945308685, "learning_rate": 3.9955642355485256e-05, "loss": 0.0053, "step": 175850 }, { "epoch": 112.08413001912047, "grad_norm": 0.25689274072647095, "learning_rate": 3.995024320710127e-05, "loss": 0.0111, "step": 175860 }, { "epoch": 112.09050350541746, "grad_norm": 0.2453933209180832, "learning_rate": 3.9944844180831334e-05, "loss": 0.0121, "step": 175870 }, { "epoch": 112.09687699171447, "grad_norm": 0.260588139295578, "learning_rate": 3.9939445276741014e-05, "loss": 0.0069, "step": 175880 }, { "epoch": 112.10325047801147, "grad_norm": 0.5374392867088318, "learning_rate": 3.993404649489594e-05, "loss": 0.0122, "step": 175890 }, { "epoch": 112.10962396430848, "grad_norm": 0.36996886134147644, "learning_rate": 3.99286478353617e-05, "loss": 0.0095, "step": 175900 }, { "epoch": 112.11599745060548, "grad_norm": 0.2303079068660736, "learning_rate": 3.9923249298203924e-05, "loss": 0.0154, "step": 175910 }, { "epoch": 112.12237093690248, "grad_norm": 0.3484993577003479, "learning_rate": 3.991785088348816e-05, "loss": 0.0106, "step": 175920 }, { "epoch": 112.12874442319949, "grad_norm": 0.2574855387210846, "learning_rate": 3.991245259128002e-05, "loss": 0.0085, "step": 175930 }, { "epoch": 112.1351179094965, "grad_norm": 0.33944135904312134, "learning_rate": 3.990705442164513e-05, "loss": 0.0123, "step": 175940 }, { "epoch": 112.1414913957935, "grad_norm": 0.18148933351039886, "learning_rate": 3.990165637464904e-05, "loss": 0.0103, "step": 175950 }, { "epoch": 112.1478648820905, "grad_norm": 0.29624223709106445, "learning_rate": 3.989625845035736e-05, "loss": 0.0119, "step": 175960 }, { "epoch": 112.15423836838751, "grad_norm": 0.25997522473335266, "learning_rate": 3.989086064883567e-05, "loss": 0.0162, "step": 175970 }, { "epoch": 112.16061185468452, "grad_norm": 0.22992074489593506, "learning_rate": 3.98854629701496e-05, "loss": 0.0118, "step": 175980 }, { "epoch": 112.16698534098151, "grad_norm": 0.15468871593475342, "learning_rate": 3.988006541436468e-05, "loss": 0.0134, "step": 175990 }, { "epoch": 112.17335882727852, "grad_norm": 0.20667187869548798, "learning_rate": 3.987466798154652e-05, "loss": 0.0122, "step": 176000 }, { "epoch": 112.17973231357553, "grad_norm": 0.15763306617736816, "learning_rate": 3.98692706717607e-05, "loss": 0.0108, "step": 176010 }, { "epoch": 112.18610579987254, "grad_norm": 0.14893756806850433, "learning_rate": 3.9863873485072834e-05, "loss": 0.0047, "step": 176020 }, { "epoch": 112.19247928616953, "grad_norm": 0.33191773295402527, "learning_rate": 3.985847642154845e-05, "loss": 0.0125, "step": 176030 }, { "epoch": 112.19885277246654, "grad_norm": 0.46343904733657837, "learning_rate": 3.9853079481253155e-05, "loss": 0.0167, "step": 176040 }, { "epoch": 112.20522625876355, "grad_norm": 0.20505450665950775, "learning_rate": 3.984768266425254e-05, "loss": 0.0099, "step": 176050 }, { "epoch": 112.21159974506055, "grad_norm": 0.3005519509315491, "learning_rate": 3.984228597061215e-05, "loss": 0.0063, "step": 176060 }, { "epoch": 112.21797323135755, "grad_norm": 0.5672322511672974, "learning_rate": 3.983688940039758e-05, "loss": 0.0073, "step": 176070 }, { "epoch": 112.22434671765456, "grad_norm": 0.3914397656917572, "learning_rate": 3.9831492953674386e-05, "loss": 0.0092, "step": 176080 }, { "epoch": 112.23072020395156, "grad_norm": 0.32005712389945984, "learning_rate": 3.982609663050819e-05, "loss": 0.0123, "step": 176090 }, { "epoch": 112.23709369024857, "grad_norm": 0.2084474414587021, "learning_rate": 3.982070043096449e-05, "loss": 0.0074, "step": 176100 }, { "epoch": 112.24346717654556, "grad_norm": 0.4391154646873474, "learning_rate": 3.98153043551089e-05, "loss": 0.0147, "step": 176110 }, { "epoch": 112.24984066284257, "grad_norm": 0.3643544912338257, "learning_rate": 3.9809908403006996e-05, "loss": 0.0135, "step": 176120 }, { "epoch": 112.25621414913958, "grad_norm": 0.21197518706321716, "learning_rate": 3.9804512574724304e-05, "loss": 0.0085, "step": 176130 }, { "epoch": 112.26258763543659, "grad_norm": 0.5537325739860535, "learning_rate": 3.9799116870326415e-05, "loss": 0.011, "step": 176140 }, { "epoch": 112.26896112173358, "grad_norm": 0.3631308674812317, "learning_rate": 3.979372128987888e-05, "loss": 0.0139, "step": 176150 }, { "epoch": 112.27533460803059, "grad_norm": 0.235083669424057, "learning_rate": 3.97883258334473e-05, "loss": 0.0088, "step": 176160 }, { "epoch": 112.2817080943276, "grad_norm": 0.23662562668323517, "learning_rate": 3.9782930501097174e-05, "loss": 0.0101, "step": 176170 }, { "epoch": 112.2880815806246, "grad_norm": 0.2409946322441101, "learning_rate": 3.9777535292894085e-05, "loss": 0.0084, "step": 176180 }, { "epoch": 112.2944550669216, "grad_norm": 0.2213013470172882, "learning_rate": 3.977214020890362e-05, "loss": 0.0086, "step": 176190 }, { "epoch": 112.30082855321861, "grad_norm": 0.2705506384372711, "learning_rate": 3.976674524919129e-05, "loss": 0.0211, "step": 176200 }, { "epoch": 112.30720203951562, "grad_norm": 0.26386579871177673, "learning_rate": 3.976135041382266e-05, "loss": 0.0128, "step": 176210 }, { "epoch": 112.31357552581262, "grad_norm": 0.30293941497802734, "learning_rate": 3.975595570286329e-05, "loss": 0.0157, "step": 176220 }, { "epoch": 112.31994901210962, "grad_norm": 0.2469511479139328, "learning_rate": 3.975056111637876e-05, "loss": 0.012, "step": 176230 }, { "epoch": 112.32632249840663, "grad_norm": 0.40474623441696167, "learning_rate": 3.974516665443456e-05, "loss": 0.0132, "step": 176240 }, { "epoch": 112.33269598470363, "grad_norm": 0.6492941975593567, "learning_rate": 3.973977231709627e-05, "loss": 0.0157, "step": 176250 }, { "epoch": 112.33906947100064, "grad_norm": 0.3254856765270233, "learning_rate": 3.973437810442943e-05, "loss": 0.0089, "step": 176260 }, { "epoch": 112.34544295729764, "grad_norm": 0.22973796725273132, "learning_rate": 3.9728984016499614e-05, "loss": 0.0089, "step": 176270 }, { "epoch": 112.35181644359464, "grad_norm": 0.2581694722175598, "learning_rate": 3.972359005337232e-05, "loss": 0.0097, "step": 176280 }, { "epoch": 112.35818992989165, "grad_norm": 0.12121628224849701, "learning_rate": 3.971819621511311e-05, "loss": 0.0093, "step": 176290 }, { "epoch": 112.36456341618866, "grad_norm": 0.3599086403846741, "learning_rate": 3.9712802501787525e-05, "loss": 0.0137, "step": 176300 }, { "epoch": 112.37093690248565, "grad_norm": 0.22389383614063263, "learning_rate": 3.970740891346111e-05, "loss": 0.0067, "step": 176310 }, { "epoch": 112.37731038878266, "grad_norm": 0.15556426346302032, "learning_rate": 3.970201545019938e-05, "loss": 0.0122, "step": 176320 }, { "epoch": 112.38368387507967, "grad_norm": 0.3315950632095337, "learning_rate": 3.9696622112067896e-05, "loss": 0.0136, "step": 176330 }, { "epoch": 112.39005736137668, "grad_norm": 0.25917601585388184, "learning_rate": 3.9691228899132176e-05, "loss": 0.0085, "step": 176340 }, { "epoch": 112.39643084767367, "grad_norm": 0.4703764319419861, "learning_rate": 3.968583581145777e-05, "loss": 0.0078, "step": 176350 }, { "epoch": 112.40280433397068, "grad_norm": 0.5336405038833618, "learning_rate": 3.968044284911018e-05, "loss": 0.0064, "step": 176360 }, { "epoch": 112.40917782026769, "grad_norm": 0.4752900302410126, "learning_rate": 3.967505001215496e-05, "loss": 0.0168, "step": 176370 }, { "epoch": 112.4155513065647, "grad_norm": 0.20913881063461304, "learning_rate": 3.966965730065763e-05, "loss": 0.0063, "step": 176380 }, { "epoch": 112.4219247928617, "grad_norm": 0.166171595454216, "learning_rate": 3.9664264714683725e-05, "loss": 0.0057, "step": 176390 }, { "epoch": 112.4282982791587, "grad_norm": 0.28224676847457886, "learning_rate": 3.965887225429874e-05, "loss": 0.0111, "step": 176400 }, { "epoch": 112.4346717654557, "grad_norm": 0.39853253960609436, "learning_rate": 3.9653479919568246e-05, "loss": 0.012, "step": 176410 }, { "epoch": 112.44104525175271, "grad_norm": 0.11685539036989212, "learning_rate": 3.964808771055773e-05, "loss": 0.0083, "step": 176420 }, { "epoch": 112.44741873804972, "grad_norm": 0.39877042174339294, "learning_rate": 3.9642695627332725e-05, "loss": 0.0131, "step": 176430 }, { "epoch": 112.45379222434671, "grad_norm": 0.2852799892425537, "learning_rate": 3.963730366995875e-05, "loss": 0.0088, "step": 176440 }, { "epoch": 112.46016571064372, "grad_norm": 0.19582001864910126, "learning_rate": 3.9631911838501305e-05, "loss": 0.0117, "step": 176450 }, { "epoch": 112.46653919694073, "grad_norm": 0.2981765568256378, "learning_rate": 3.962652013302593e-05, "loss": 0.006, "step": 176460 }, { "epoch": 112.47291268323774, "grad_norm": 0.44252827763557434, "learning_rate": 3.9621128553598124e-05, "loss": 0.011, "step": 176470 }, { "epoch": 112.47928616953473, "grad_norm": 0.342862606048584, "learning_rate": 3.961573710028342e-05, "loss": 0.009, "step": 176480 }, { "epoch": 112.48565965583174, "grad_norm": 0.23468968272209167, "learning_rate": 3.961034577314729e-05, "loss": 0.0076, "step": 176490 }, { "epoch": 112.49203314212875, "grad_norm": 0.28373149037361145, "learning_rate": 3.960495457225528e-05, "loss": 0.0103, "step": 176500 }, { "epoch": 112.49840662842576, "grad_norm": 0.10728014260530472, "learning_rate": 3.959956349767291e-05, "loss": 0.006, "step": 176510 }, { "epoch": 112.50478011472275, "grad_norm": 0.4037874937057495, "learning_rate": 3.9594172549465636e-05, "loss": 0.0095, "step": 176520 }, { "epoch": 112.51115360101976, "grad_norm": 0.2723773121833801, "learning_rate": 3.9588781727698986e-05, "loss": 0.0095, "step": 176530 }, { "epoch": 112.51752708731676, "grad_norm": 0.12891915440559387, "learning_rate": 3.9583391032438475e-05, "loss": 0.0069, "step": 176540 }, { "epoch": 112.52390057361377, "grad_norm": 0.44940486550331116, "learning_rate": 3.957800046374962e-05, "loss": 0.0104, "step": 176550 }, { "epoch": 112.53027405991077, "grad_norm": 0.3461102545261383, "learning_rate": 3.957261002169787e-05, "loss": 0.0107, "step": 176560 }, { "epoch": 112.53664754620777, "grad_norm": 0.2631368339061737, "learning_rate": 3.956721970634876e-05, "loss": 0.008, "step": 176570 }, { "epoch": 112.54302103250478, "grad_norm": 0.23533326387405396, "learning_rate": 3.956182951776778e-05, "loss": 0.021, "step": 176580 }, { "epoch": 112.54939451880179, "grad_norm": 0.25835904479026794, "learning_rate": 3.955643945602044e-05, "loss": 0.0081, "step": 176590 }, { "epoch": 112.55576800509878, "grad_norm": 0.23050934076309204, "learning_rate": 3.955104952117221e-05, "loss": 0.0143, "step": 176600 }, { "epoch": 112.56214149139579, "grad_norm": 0.3984905779361725, "learning_rate": 3.954565971328859e-05, "loss": 0.015, "step": 176610 }, { "epoch": 112.5685149776928, "grad_norm": 0.22648601233959198, "learning_rate": 3.9540270032435103e-05, "loss": 0.0135, "step": 176620 }, { "epoch": 112.57488846398981, "grad_norm": 0.2227528989315033, "learning_rate": 3.9534880478677175e-05, "loss": 0.0118, "step": 176630 }, { "epoch": 112.5812619502868, "grad_norm": 0.22381730377674103, "learning_rate": 3.952949105208034e-05, "loss": 0.0066, "step": 176640 }, { "epoch": 112.58763543658381, "grad_norm": 0.5559903979301453, "learning_rate": 3.9524101752710075e-05, "loss": 0.0109, "step": 176650 }, { "epoch": 112.59400892288082, "grad_norm": 0.13281039893627167, "learning_rate": 3.951871258063188e-05, "loss": 0.0169, "step": 176660 }, { "epoch": 112.60038240917783, "grad_norm": 0.24557512998580933, "learning_rate": 3.951332353591121e-05, "loss": 0.0096, "step": 176670 }, { "epoch": 112.60675589547482, "grad_norm": 0.40994203090667725, "learning_rate": 3.950793461861355e-05, "loss": 0.0092, "step": 176680 }, { "epoch": 112.61312938177183, "grad_norm": 0.26557308435440063, "learning_rate": 3.950254582880442e-05, "loss": 0.0106, "step": 176690 }, { "epoch": 112.61950286806884, "grad_norm": 0.267691433429718, "learning_rate": 3.949715716654925e-05, "loss": 0.0108, "step": 176700 }, { "epoch": 112.62587635436584, "grad_norm": 0.43069326877593994, "learning_rate": 3.949176863191353e-05, "loss": 0.0093, "step": 176710 }, { "epoch": 112.63224984066284, "grad_norm": 0.23487722873687744, "learning_rate": 3.948638022496274e-05, "loss": 0.0116, "step": 176720 }, { "epoch": 112.63862332695984, "grad_norm": 0.5730644464492798, "learning_rate": 3.948099194576239e-05, "loss": 0.011, "step": 176730 }, { "epoch": 112.64499681325685, "grad_norm": 0.20394375920295715, "learning_rate": 3.947560379437788e-05, "loss": 0.0094, "step": 176740 }, { "epoch": 112.65137029955386, "grad_norm": 0.7272995710372925, "learning_rate": 3.947021577087473e-05, "loss": 0.0135, "step": 176750 }, { "epoch": 112.65774378585085, "grad_norm": 0.2305154651403427, "learning_rate": 3.946482787531843e-05, "loss": 0.0122, "step": 176760 }, { "epoch": 112.66411727214786, "grad_norm": 0.39077767729759216, "learning_rate": 3.9459440107774385e-05, "loss": 0.0131, "step": 176770 }, { "epoch": 112.67049075844487, "grad_norm": 0.2412169873714447, "learning_rate": 3.94540524683081e-05, "loss": 0.0082, "step": 176780 }, { "epoch": 112.67686424474188, "grad_norm": 0.39440709352493286, "learning_rate": 3.9448664956985036e-05, "loss": 0.015, "step": 176790 }, { "epoch": 112.68323773103887, "grad_norm": 0.2390725314617157, "learning_rate": 3.9443277573870676e-05, "loss": 0.0081, "step": 176800 }, { "epoch": 112.68961121733588, "grad_norm": 0.19182558357715607, "learning_rate": 3.9437890319030436e-05, "loss": 0.0065, "step": 176810 }, { "epoch": 112.69598470363289, "grad_norm": 0.23392806947231293, "learning_rate": 3.9432503192529803e-05, "loss": 0.0112, "step": 176820 }, { "epoch": 112.7023581899299, "grad_norm": 0.3226105570793152, "learning_rate": 3.9427116194434256e-05, "loss": 0.0096, "step": 176830 }, { "epoch": 112.70873167622689, "grad_norm": 0.3875381648540497, "learning_rate": 3.9421729324809205e-05, "loss": 0.009, "step": 176840 }, { "epoch": 112.7151051625239, "grad_norm": 0.3304571211338043, "learning_rate": 3.941634258372013e-05, "loss": 0.0132, "step": 176850 }, { "epoch": 112.7214786488209, "grad_norm": 0.3273222744464874, "learning_rate": 3.941095597123249e-05, "loss": 0.0083, "step": 176860 }, { "epoch": 112.72785213511791, "grad_norm": 0.2871202230453491, "learning_rate": 3.940556948741175e-05, "loss": 0.0136, "step": 176870 }, { "epoch": 112.73422562141491, "grad_norm": 0.15424677729606628, "learning_rate": 3.940018313232332e-05, "loss": 0.0097, "step": 176880 }, { "epoch": 112.74059910771192, "grad_norm": 0.48276692628860474, "learning_rate": 3.9394796906032675e-05, "loss": 0.0197, "step": 176890 }, { "epoch": 112.74697259400892, "grad_norm": 0.38393157720565796, "learning_rate": 3.9389410808605265e-05, "loss": 0.0101, "step": 176900 }, { "epoch": 112.75334608030593, "grad_norm": 0.13515637814998627, "learning_rate": 3.938402484010655e-05, "loss": 0.0108, "step": 176910 }, { "epoch": 112.75971956660293, "grad_norm": 0.4304657280445099, "learning_rate": 3.937863900060194e-05, "loss": 0.0122, "step": 176920 }, { "epoch": 112.76609305289993, "grad_norm": 0.23692315816879272, "learning_rate": 3.937325329015688e-05, "loss": 0.0101, "step": 176930 }, { "epoch": 112.77246653919694, "grad_norm": 0.31349995732307434, "learning_rate": 3.936786770883686e-05, "loss": 0.0081, "step": 176940 }, { "epoch": 112.77884002549395, "grad_norm": 0.3492848873138428, "learning_rate": 3.936248225670726e-05, "loss": 0.0124, "step": 176950 }, { "epoch": 112.78521351179094, "grad_norm": 0.14434875547885895, "learning_rate": 3.935709693383354e-05, "loss": 0.0074, "step": 176960 }, { "epoch": 112.79158699808795, "grad_norm": 0.4324643611907959, "learning_rate": 3.935171174028115e-05, "loss": 0.012, "step": 176970 }, { "epoch": 112.79796048438496, "grad_norm": 0.37414222955703735, "learning_rate": 3.934632667611553e-05, "loss": 0.0081, "step": 176980 }, { "epoch": 112.80433397068197, "grad_norm": 0.2502996027469635, "learning_rate": 3.9340941741402084e-05, "loss": 0.0082, "step": 176990 }, { "epoch": 112.81070745697897, "grad_norm": 0.21053850650787354, "learning_rate": 3.9335556936206255e-05, "loss": 0.0109, "step": 177000 }, { "epoch": 112.81708094327597, "grad_norm": 0.22470057010650635, "learning_rate": 3.933017226059349e-05, "loss": 0.0079, "step": 177010 }, { "epoch": 112.82345442957298, "grad_norm": 0.23982469737529755, "learning_rate": 3.932478771462921e-05, "loss": 0.0098, "step": 177020 }, { "epoch": 112.82982791586998, "grad_norm": 0.5987948775291443, "learning_rate": 3.931940329837882e-05, "loss": 0.0094, "step": 177030 }, { "epoch": 112.83620140216699, "grad_norm": 0.3401685357093811, "learning_rate": 3.9314019011907774e-05, "loss": 0.0185, "step": 177040 }, { "epoch": 112.84257488846399, "grad_norm": 0.1945200264453888, "learning_rate": 3.930863485528149e-05, "loss": 0.011, "step": 177050 }, { "epoch": 112.848948374761, "grad_norm": 0.39179906249046326, "learning_rate": 3.930325082856538e-05, "loss": 0.0103, "step": 177060 }, { "epoch": 112.855321861058, "grad_norm": 0.2695293128490448, "learning_rate": 3.929786693182487e-05, "loss": 0.0129, "step": 177070 }, { "epoch": 112.86169534735501, "grad_norm": 0.3253249526023865, "learning_rate": 3.929248316512539e-05, "loss": 0.0074, "step": 177080 }, { "epoch": 112.868068833652, "grad_norm": 0.1434982866048813, "learning_rate": 3.9287099528532346e-05, "loss": 0.009, "step": 177090 }, { "epoch": 112.87444231994901, "grad_norm": 0.1870257407426834, "learning_rate": 3.928171602211116e-05, "loss": 0.0101, "step": 177100 }, { "epoch": 112.88081580624602, "grad_norm": 0.14129093289375305, "learning_rate": 3.927633264592724e-05, "loss": 0.0063, "step": 177110 }, { "epoch": 112.88718929254303, "grad_norm": 0.643645703792572, "learning_rate": 3.9270949400046e-05, "loss": 0.0099, "step": 177120 }, { "epoch": 112.89356277884002, "grad_norm": 0.18891631066799164, "learning_rate": 3.9265566284532854e-05, "loss": 0.0127, "step": 177130 }, { "epoch": 112.89993626513703, "grad_norm": 0.3154776990413666, "learning_rate": 3.9260183299453226e-05, "loss": 0.0111, "step": 177140 }, { "epoch": 112.90630975143404, "grad_norm": 0.23492243885993958, "learning_rate": 3.925480044487251e-05, "loss": 0.0062, "step": 177150 }, { "epoch": 112.91268323773104, "grad_norm": 0.5032705664634705, "learning_rate": 3.92494177208561e-05, "loss": 0.015, "step": 177160 }, { "epoch": 112.91905672402804, "grad_norm": 0.383256196975708, "learning_rate": 3.9244035127469414e-05, "loss": 0.0123, "step": 177170 }, { "epoch": 112.92543021032505, "grad_norm": 0.1383005678653717, "learning_rate": 3.923865266477787e-05, "loss": 0.0107, "step": 177180 }, { "epoch": 112.93180369662205, "grad_norm": 0.17325665056705475, "learning_rate": 3.923327033284685e-05, "loss": 0.0071, "step": 177190 }, { "epoch": 112.93817718291906, "grad_norm": 0.3167622983455658, "learning_rate": 3.9227888131741756e-05, "loss": 0.0075, "step": 177200 }, { "epoch": 112.94455066921606, "grad_norm": 0.2963675558567047, "learning_rate": 3.9222506061527995e-05, "loss": 0.0056, "step": 177210 }, { "epoch": 112.95092415551306, "grad_norm": 0.3826880156993866, "learning_rate": 3.921712412227097e-05, "loss": 0.0091, "step": 177220 }, { "epoch": 112.95729764181007, "grad_norm": 0.5483189821243286, "learning_rate": 3.921174231403606e-05, "loss": 0.0303, "step": 177230 }, { "epoch": 112.96367112810708, "grad_norm": 0.19116197526454926, "learning_rate": 3.920636063688866e-05, "loss": 0.0068, "step": 177240 }, { "epoch": 112.97004461440407, "grad_norm": 0.425224244594574, "learning_rate": 3.9200979090894166e-05, "loss": 0.01, "step": 177250 }, { "epoch": 112.97641810070108, "grad_norm": 0.6669591069221497, "learning_rate": 3.9195597676118e-05, "loss": 0.0129, "step": 177260 }, { "epoch": 112.98279158699809, "grad_norm": 0.2592135965824127, "learning_rate": 3.91902163926255e-05, "loss": 0.0122, "step": 177270 }, { "epoch": 112.9891650732951, "grad_norm": 0.2151811718940735, "learning_rate": 3.918483524048207e-05, "loss": 0.0193, "step": 177280 }, { "epoch": 112.99553855959209, "grad_norm": 0.6674246191978455, "learning_rate": 3.917945421975312e-05, "loss": 0.0159, "step": 177290 }, { "epoch": 113.0019120458891, "grad_norm": 0.30354902148246765, "learning_rate": 3.9174073330504024e-05, "loss": 0.008, "step": 177300 }, { "epoch": 113.00828553218611, "grad_norm": 0.2737705409526825, "learning_rate": 3.916869257280015e-05, "loss": 0.0119, "step": 177310 }, { "epoch": 113.01465901848312, "grad_norm": 0.4548632800579071, "learning_rate": 3.9163311946706873e-05, "loss": 0.0087, "step": 177320 }, { "epoch": 113.02103250478011, "grad_norm": 0.181422621011734, "learning_rate": 3.915793145228961e-05, "loss": 0.0076, "step": 177330 }, { "epoch": 113.02740599107712, "grad_norm": 0.19409604370594025, "learning_rate": 3.91525510896137e-05, "loss": 0.0089, "step": 177340 }, { "epoch": 113.03377947737413, "grad_norm": 0.42699384689331055, "learning_rate": 3.914717085874454e-05, "loss": 0.0149, "step": 177350 }, { "epoch": 113.04015296367113, "grad_norm": 0.23341527581214905, "learning_rate": 3.91417907597475e-05, "loss": 0.0057, "step": 177360 }, { "epoch": 113.04652644996813, "grad_norm": 0.10493051260709763, "learning_rate": 3.913641079268797e-05, "loss": 0.0073, "step": 177370 }, { "epoch": 113.05289993626513, "grad_norm": 0.16039562225341797, "learning_rate": 3.913103095763129e-05, "loss": 0.0087, "step": 177380 }, { "epoch": 113.05927342256214, "grad_norm": 0.38405564427375793, "learning_rate": 3.912565125464284e-05, "loss": 0.0093, "step": 177390 }, { "epoch": 113.06564690885915, "grad_norm": 0.4758222997188568, "learning_rate": 3.912027168378802e-05, "loss": 0.0148, "step": 177400 }, { "epoch": 113.07202039515614, "grad_norm": 0.22674499452114105, "learning_rate": 3.911489224513215e-05, "loss": 0.0059, "step": 177410 }, { "epoch": 113.07839388145315, "grad_norm": 0.3080192804336548, "learning_rate": 3.910951293874061e-05, "loss": 0.0111, "step": 177420 }, { "epoch": 113.08476736775016, "grad_norm": 0.17929930984973907, "learning_rate": 3.9104133764678774e-05, "loss": 0.007, "step": 177430 }, { "epoch": 113.09114085404717, "grad_norm": 0.5105981230735779, "learning_rate": 3.909875472301203e-05, "loss": 0.009, "step": 177440 }, { "epoch": 113.09751434034416, "grad_norm": 0.4830833971500397, "learning_rate": 3.909337581380568e-05, "loss": 0.0118, "step": 177450 }, { "epoch": 113.10388782664117, "grad_norm": 0.23640567064285278, "learning_rate": 3.908799703712511e-05, "loss": 0.0084, "step": 177460 }, { "epoch": 113.11026131293818, "grad_norm": 0.17790113389492035, "learning_rate": 3.9082618393035675e-05, "loss": 0.012, "step": 177470 }, { "epoch": 113.11663479923519, "grad_norm": 0.48914456367492676, "learning_rate": 3.907723988160276e-05, "loss": 0.0119, "step": 177480 }, { "epoch": 113.12300828553218, "grad_norm": 0.25785472989082336, "learning_rate": 3.907186150289167e-05, "loss": 0.0119, "step": 177490 }, { "epoch": 113.12938177182919, "grad_norm": 0.47855836153030396, "learning_rate": 3.906648325696778e-05, "loss": 0.0082, "step": 177500 }, { "epoch": 113.1357552581262, "grad_norm": 0.4979599714279175, "learning_rate": 3.906110514389647e-05, "loss": 0.0132, "step": 177510 }, { "epoch": 113.1421287444232, "grad_norm": 0.3126462697982788, "learning_rate": 3.9055727163743036e-05, "loss": 0.0091, "step": 177520 }, { "epoch": 113.1485022307202, "grad_norm": 0.23720131814479828, "learning_rate": 3.905034931657285e-05, "loss": 0.0056, "step": 177530 }, { "epoch": 113.1548757170172, "grad_norm": 0.5200250744819641, "learning_rate": 3.904497160245125e-05, "loss": 0.018, "step": 177540 }, { "epoch": 113.16124920331421, "grad_norm": 0.33366942405700684, "learning_rate": 3.903959402144362e-05, "loss": 0.0052, "step": 177550 }, { "epoch": 113.16762268961122, "grad_norm": 0.15940651297569275, "learning_rate": 3.903421657361525e-05, "loss": 0.0101, "step": 177560 }, { "epoch": 113.17399617590821, "grad_norm": 0.20114070177078247, "learning_rate": 3.902883925903149e-05, "loss": 0.0078, "step": 177570 }, { "epoch": 113.18036966220522, "grad_norm": 0.4036467373371124, "learning_rate": 3.902346207775772e-05, "loss": 0.0114, "step": 177580 }, { "epoch": 113.18674314850223, "grad_norm": 0.5293850898742676, "learning_rate": 3.901808502985924e-05, "loss": 0.0145, "step": 177590 }, { "epoch": 113.19311663479924, "grad_norm": 0.2610442340373993, "learning_rate": 3.901270811540138e-05, "loss": 0.0101, "step": 177600 }, { "epoch": 113.19949012109623, "grad_norm": 0.21868306398391724, "learning_rate": 3.9007331334449495e-05, "loss": 0.0123, "step": 177610 }, { "epoch": 113.20586360739324, "grad_norm": 0.47242847084999084, "learning_rate": 3.900195468706893e-05, "loss": 0.0095, "step": 177620 }, { "epoch": 113.21223709369025, "grad_norm": 0.37297186255455017, "learning_rate": 3.899657817332498e-05, "loss": 0.0109, "step": 177630 }, { "epoch": 113.21861057998726, "grad_norm": 0.5897127985954285, "learning_rate": 3.8991201793283e-05, "loss": 0.014, "step": 177640 }, { "epoch": 113.22498406628426, "grad_norm": 0.13258925080299377, "learning_rate": 3.898582554700833e-05, "loss": 0.0129, "step": 177650 }, { "epoch": 113.23135755258126, "grad_norm": 0.3616393506526947, "learning_rate": 3.898044943456626e-05, "loss": 0.0106, "step": 177660 }, { "epoch": 113.23773103887827, "grad_norm": 0.1644676923751831, "learning_rate": 3.897507345602213e-05, "loss": 0.0077, "step": 177670 }, { "epoch": 113.24410452517527, "grad_norm": 0.6157948970794678, "learning_rate": 3.896969761144126e-05, "loss": 0.016, "step": 177680 }, { "epoch": 113.25047801147228, "grad_norm": 0.462326318025589, "learning_rate": 3.896432190088901e-05, "loss": 0.0197, "step": 177690 }, { "epoch": 113.25685149776928, "grad_norm": 0.5568975806236267, "learning_rate": 3.8958946324430644e-05, "loss": 0.0164, "step": 177700 }, { "epoch": 113.26322498406628, "grad_norm": 0.6567326784133911, "learning_rate": 3.895357088213151e-05, "loss": 0.0087, "step": 177710 }, { "epoch": 113.26959847036329, "grad_norm": 0.3245493769645691, "learning_rate": 3.894819557405692e-05, "loss": 0.0129, "step": 177720 }, { "epoch": 113.2759719566603, "grad_norm": 0.3998517692089081, "learning_rate": 3.894282040027219e-05, "loss": 0.0185, "step": 177730 }, { "epoch": 113.2823454429573, "grad_norm": 0.44359371066093445, "learning_rate": 3.893744536084263e-05, "loss": 0.012, "step": 177740 }, { "epoch": 113.2887189292543, "grad_norm": 0.2962329089641571, "learning_rate": 3.893207045583355e-05, "loss": 0.0077, "step": 177750 }, { "epoch": 113.29509241555131, "grad_norm": 0.18598218262195587, "learning_rate": 3.892669568531027e-05, "loss": 0.0149, "step": 177760 }, { "epoch": 113.30146590184832, "grad_norm": 0.2188878357410431, "learning_rate": 3.89213210493381e-05, "loss": 0.0067, "step": 177770 }, { "epoch": 113.30783938814531, "grad_norm": 0.21227772533893585, "learning_rate": 3.891594654798232e-05, "loss": 0.0097, "step": 177780 }, { "epoch": 113.31421287444232, "grad_norm": 0.14507830142974854, "learning_rate": 3.891057218130826e-05, "loss": 0.0088, "step": 177790 }, { "epoch": 113.32058636073933, "grad_norm": 0.18394312262535095, "learning_rate": 3.8905197949381226e-05, "loss": 0.0109, "step": 177800 }, { "epoch": 113.32695984703633, "grad_norm": 0.3679178059101105, "learning_rate": 3.889982385226651e-05, "loss": 0.0111, "step": 177810 }, { "epoch": 113.33333333333333, "grad_norm": 0.18628866970539093, "learning_rate": 3.88944498900294e-05, "loss": 0.0082, "step": 177820 }, { "epoch": 113.33970681963034, "grad_norm": 0.15942008793354034, "learning_rate": 3.888907606273523e-05, "loss": 0.0081, "step": 177830 }, { "epoch": 113.34608030592734, "grad_norm": 0.1695641428232193, "learning_rate": 3.8883702370449264e-05, "loss": 0.0092, "step": 177840 }, { "epoch": 113.35245379222435, "grad_norm": 0.164969339966774, "learning_rate": 3.887832881323681e-05, "loss": 0.008, "step": 177850 }, { "epoch": 113.35882727852135, "grad_norm": 0.3371752202510834, "learning_rate": 3.887295539116317e-05, "loss": 0.0119, "step": 177860 }, { "epoch": 113.36520076481835, "grad_norm": 0.31977206468582153, "learning_rate": 3.8867582104293626e-05, "loss": 0.0095, "step": 177870 }, { "epoch": 113.37157425111536, "grad_norm": 0.1552468091249466, "learning_rate": 3.886220895269346e-05, "loss": 0.0118, "step": 177880 }, { "epoch": 113.37794773741237, "grad_norm": 0.26571428775787354, "learning_rate": 3.885683593642797e-05, "loss": 0.0086, "step": 177890 }, { "epoch": 113.38432122370936, "grad_norm": 0.3011552095413208, "learning_rate": 3.885146305556247e-05, "loss": 0.0124, "step": 177900 }, { "epoch": 113.39069471000637, "grad_norm": 0.34179943799972534, "learning_rate": 3.8846090310162195e-05, "loss": 0.013, "step": 177910 }, { "epoch": 113.39706819630338, "grad_norm": 0.2159290313720703, "learning_rate": 3.8840717700292453e-05, "loss": 0.0091, "step": 177920 }, { "epoch": 113.40344168260039, "grad_norm": 0.1432826966047287, "learning_rate": 3.883534522601853e-05, "loss": 0.013, "step": 177930 }, { "epoch": 113.40981516889738, "grad_norm": 0.17767688632011414, "learning_rate": 3.882997288740573e-05, "loss": 0.006, "step": 177940 }, { "epoch": 113.41618865519439, "grad_norm": 0.40304291248321533, "learning_rate": 3.8824600684519276e-05, "loss": 0.0068, "step": 177950 }, { "epoch": 113.4225621414914, "grad_norm": 0.2382969856262207, "learning_rate": 3.881922861742447e-05, "loss": 0.0102, "step": 177960 }, { "epoch": 113.4289356277884, "grad_norm": 0.1809762418270111, "learning_rate": 3.8813856686186626e-05, "loss": 0.0064, "step": 177970 }, { "epoch": 113.4353091140854, "grad_norm": 0.39773765206336975, "learning_rate": 3.8808484890870964e-05, "loss": 0.0099, "step": 177980 }, { "epoch": 113.4416826003824, "grad_norm": 0.3777170777320862, "learning_rate": 3.880311323154277e-05, "loss": 0.0084, "step": 177990 }, { "epoch": 113.44805608667941, "grad_norm": 0.29593056440353394, "learning_rate": 3.879774170826732e-05, "loss": 0.0137, "step": 178000 }, { "epoch": 113.45442957297642, "grad_norm": 0.7261837720870972, "learning_rate": 3.8792370321109915e-05, "loss": 0.0154, "step": 178010 }, { "epoch": 113.46080305927342, "grad_norm": 0.5233850479125977, "learning_rate": 3.878699907013577e-05, "loss": 0.0109, "step": 178020 }, { "epoch": 113.46717654557042, "grad_norm": 0.5035421848297119, "learning_rate": 3.8781627955410164e-05, "loss": 0.0127, "step": 178030 }, { "epoch": 113.47355003186743, "grad_norm": 0.29035869240760803, "learning_rate": 3.87762569769984e-05, "loss": 0.0096, "step": 178040 }, { "epoch": 113.47992351816444, "grad_norm": 0.14091888070106506, "learning_rate": 3.877088613496568e-05, "loss": 0.0104, "step": 178050 }, { "epoch": 113.48629700446143, "grad_norm": 0.4467627704143524, "learning_rate": 3.8765515429377306e-05, "loss": 0.0071, "step": 178060 }, { "epoch": 113.49267049075844, "grad_norm": 0.371452271938324, "learning_rate": 3.876014486029852e-05, "loss": 0.0071, "step": 178070 }, { "epoch": 113.49904397705545, "grad_norm": 0.30368518829345703, "learning_rate": 3.875477442779461e-05, "loss": 0.0076, "step": 178080 }, { "epoch": 113.50541746335246, "grad_norm": 0.18611398339271545, "learning_rate": 3.874940413193079e-05, "loss": 0.0133, "step": 178090 }, { "epoch": 113.51179094964945, "grad_norm": 0.1612868309020996, "learning_rate": 3.874403397277233e-05, "loss": 0.0121, "step": 178100 }, { "epoch": 113.51816443594646, "grad_norm": 0.12281367927789688, "learning_rate": 3.873866395038448e-05, "loss": 0.0098, "step": 178110 }, { "epoch": 113.52453792224347, "grad_norm": 0.2679552733898163, "learning_rate": 3.873329406483252e-05, "loss": 0.0096, "step": 178120 }, { "epoch": 113.53091140854048, "grad_norm": 0.3196175992488861, "learning_rate": 3.872792431618165e-05, "loss": 0.0093, "step": 178130 }, { "epoch": 113.53728489483747, "grad_norm": 0.15305033326148987, "learning_rate": 3.872255470449715e-05, "loss": 0.0173, "step": 178140 }, { "epoch": 113.54365838113448, "grad_norm": 0.28334808349609375, "learning_rate": 3.8717185229844276e-05, "loss": 0.0089, "step": 178150 }, { "epoch": 113.55003186743149, "grad_norm": 0.2573261857032776, "learning_rate": 3.8711815892288236e-05, "loss": 0.0086, "step": 178160 }, { "epoch": 113.5564053537285, "grad_norm": 0.32117554545402527, "learning_rate": 3.870644669189428e-05, "loss": 0.0129, "step": 178170 }, { "epoch": 113.56277884002549, "grad_norm": 0.2827949821949005, "learning_rate": 3.8701077628727666e-05, "loss": 0.0105, "step": 178180 }, { "epoch": 113.5691523263225, "grad_norm": 0.4611082971096039, "learning_rate": 3.869570870285365e-05, "loss": 0.0128, "step": 178190 }, { "epoch": 113.5755258126195, "grad_norm": 0.250137060880661, "learning_rate": 3.8690339914337424e-05, "loss": 0.0086, "step": 178200 }, { "epoch": 113.58189929891651, "grad_norm": 0.11999135464429855, "learning_rate": 3.868497126324425e-05, "loss": 0.0076, "step": 178210 }, { "epoch": 113.5882727852135, "grad_norm": 0.1431588977575302, "learning_rate": 3.867960274963937e-05, "loss": 0.0069, "step": 178220 }, { "epoch": 113.59464627151051, "grad_norm": 0.19370244443416595, "learning_rate": 3.8674234373587984e-05, "loss": 0.0129, "step": 178230 }, { "epoch": 113.60101975780752, "grad_norm": 0.2576071619987488, "learning_rate": 3.8668866135155354e-05, "loss": 0.0182, "step": 178240 }, { "epoch": 113.60739324410453, "grad_norm": 0.3125888407230377, "learning_rate": 3.866349803440669e-05, "loss": 0.0107, "step": 178250 }, { "epoch": 113.61376673040154, "grad_norm": 0.3781757652759552, "learning_rate": 3.865813007140725e-05, "loss": 0.0195, "step": 178260 }, { "epoch": 113.62014021669853, "grad_norm": 0.24083822965621948, "learning_rate": 3.8652762246222225e-05, "loss": 0.0098, "step": 178270 }, { "epoch": 113.62651370299554, "grad_norm": 0.3546588122844696, "learning_rate": 3.864739455891685e-05, "loss": 0.0083, "step": 178280 }, { "epoch": 113.63288718929255, "grad_norm": 0.1812039464712143, "learning_rate": 3.8642027009556355e-05, "loss": 0.0061, "step": 178290 }, { "epoch": 113.63926067558955, "grad_norm": 0.40673306584358215, "learning_rate": 3.863665959820596e-05, "loss": 0.0112, "step": 178300 }, { "epoch": 113.64563416188655, "grad_norm": 0.3006138503551483, "learning_rate": 3.863129232493086e-05, "loss": 0.0119, "step": 178310 }, { "epoch": 113.65200764818356, "grad_norm": 0.18279005587100983, "learning_rate": 3.86259251897963e-05, "loss": 0.0067, "step": 178320 }, { "epoch": 113.65838113448056, "grad_norm": 0.1580856740474701, "learning_rate": 3.8620558192867504e-05, "loss": 0.0078, "step": 178330 }, { "epoch": 113.66475462077757, "grad_norm": 0.16663475334644318, "learning_rate": 3.8615191334209654e-05, "loss": 0.0102, "step": 178340 }, { "epoch": 113.67112810707457, "grad_norm": 0.19586816430091858, "learning_rate": 3.860982461388798e-05, "loss": 0.0099, "step": 178350 }, { "epoch": 113.67750159337157, "grad_norm": 0.16156989336013794, "learning_rate": 3.8604458031967706e-05, "loss": 0.0105, "step": 178360 }, { "epoch": 113.68387507966858, "grad_norm": 0.5907993912696838, "learning_rate": 3.859909158851401e-05, "loss": 0.0167, "step": 178370 }, { "epoch": 113.69024856596559, "grad_norm": 0.31679147481918335, "learning_rate": 3.859372528359211e-05, "loss": 0.0115, "step": 178380 }, { "epoch": 113.69662205226258, "grad_norm": 0.2571215331554413, "learning_rate": 3.8588359117267225e-05, "loss": 0.026, "step": 178390 }, { "epoch": 113.70299553855959, "grad_norm": 0.46128228306770325, "learning_rate": 3.858299308960456e-05, "loss": 0.0186, "step": 178400 }, { "epoch": 113.7093690248566, "grad_norm": 0.4084472358226776, "learning_rate": 3.85776272006693e-05, "loss": 0.0094, "step": 178410 }, { "epoch": 113.7157425111536, "grad_norm": 0.3620920777320862, "learning_rate": 3.857226145052665e-05, "loss": 0.0067, "step": 178420 }, { "epoch": 113.7221159974506, "grad_norm": 0.19337086379528046, "learning_rate": 3.8566895839241814e-05, "loss": 0.0091, "step": 178430 }, { "epoch": 113.72848948374761, "grad_norm": 0.22092066705226898, "learning_rate": 3.856153036688e-05, "loss": 0.0084, "step": 178440 }, { "epoch": 113.73486297004462, "grad_norm": 0.2368648797273636, "learning_rate": 3.855616503350638e-05, "loss": 0.0102, "step": 178450 }, { "epoch": 113.74123645634162, "grad_norm": 0.4419950246810913, "learning_rate": 3.855079983918616e-05, "loss": 0.0086, "step": 178460 }, { "epoch": 113.74760994263862, "grad_norm": 0.5289159417152405, "learning_rate": 3.854543478398454e-05, "loss": 0.0109, "step": 178470 }, { "epoch": 113.75398342893563, "grad_norm": 0.12728483974933624, "learning_rate": 3.854006986796669e-05, "loss": 0.0071, "step": 178480 }, { "epoch": 113.76035691523263, "grad_norm": 0.24976487457752228, "learning_rate": 3.8534705091197824e-05, "loss": 0.0197, "step": 178490 }, { "epoch": 113.76673040152964, "grad_norm": 0.2149515002965927, "learning_rate": 3.852934045374311e-05, "loss": 0.015, "step": 178500 }, { "epoch": 113.77310388782664, "grad_norm": 0.27912646532058716, "learning_rate": 3.852397595566774e-05, "loss": 0.0061, "step": 178510 }, { "epoch": 113.77947737412364, "grad_norm": 0.28804078698158264, "learning_rate": 3.851861159703689e-05, "loss": 0.0128, "step": 178520 }, { "epoch": 113.78585086042065, "grad_norm": 0.17266957461833954, "learning_rate": 3.851324737791576e-05, "loss": 0.0102, "step": 178530 }, { "epoch": 113.79222434671766, "grad_norm": 0.7046707272529602, "learning_rate": 3.850788329836952e-05, "loss": 0.0158, "step": 178540 }, { "epoch": 113.79859783301465, "grad_norm": 0.17101147770881653, "learning_rate": 3.850251935846334e-05, "loss": 0.0078, "step": 178550 }, { "epoch": 113.80497131931166, "grad_norm": 0.21005263924598694, "learning_rate": 3.849715555826241e-05, "loss": 0.0132, "step": 178560 }, { "epoch": 113.81134480560867, "grad_norm": 0.12465547770261765, "learning_rate": 3.8491791897831903e-05, "loss": 0.0068, "step": 178570 }, { "epoch": 113.81771829190568, "grad_norm": 0.14444604516029358, "learning_rate": 3.8486428377236993e-05, "loss": 0.0144, "step": 178580 }, { "epoch": 113.82409177820267, "grad_norm": 0.6627892851829529, "learning_rate": 3.8481064996542836e-05, "loss": 0.0257, "step": 178590 }, { "epoch": 113.83046526449968, "grad_norm": 0.44390496611595154, "learning_rate": 3.847570175581462e-05, "loss": 0.0137, "step": 178600 }, { "epoch": 113.83683875079669, "grad_norm": 0.16301792860031128, "learning_rate": 3.847033865511752e-05, "loss": 0.0088, "step": 178610 }, { "epoch": 113.8432122370937, "grad_norm": 0.2258462756872177, "learning_rate": 3.846497569451668e-05, "loss": 0.0093, "step": 178620 }, { "epoch": 113.84958572339069, "grad_norm": 0.3123023211956024, "learning_rate": 3.845961287407727e-05, "loss": 0.0234, "step": 178630 }, { "epoch": 113.8559592096877, "grad_norm": 0.3745649456977844, "learning_rate": 3.845425019386446e-05, "loss": 0.0134, "step": 178640 }, { "epoch": 113.8623326959847, "grad_norm": 0.23728370666503906, "learning_rate": 3.844888765394343e-05, "loss": 0.0158, "step": 178650 }, { "epoch": 113.86870618228171, "grad_norm": 0.3770734965801239, "learning_rate": 3.844352525437931e-05, "loss": 0.0089, "step": 178660 }, { "epoch": 113.8750796685787, "grad_norm": 0.28146758675575256, "learning_rate": 3.843816299523726e-05, "loss": 0.0168, "step": 178670 }, { "epoch": 113.88145315487571, "grad_norm": 0.5108863711357117, "learning_rate": 3.843280087658245e-05, "loss": 0.008, "step": 178680 }, { "epoch": 113.88782664117272, "grad_norm": 0.3393782675266266, "learning_rate": 3.842743889848005e-05, "loss": 0.009, "step": 178690 }, { "epoch": 113.89420012746973, "grad_norm": 0.5606580376625061, "learning_rate": 3.842207706099518e-05, "loss": 0.0199, "step": 178700 }, { "epoch": 113.90057361376672, "grad_norm": 0.3028365671634674, "learning_rate": 3.8416715364193e-05, "loss": 0.0117, "step": 178710 }, { "epoch": 113.90694710006373, "grad_norm": 0.3422660827636719, "learning_rate": 3.841135380813868e-05, "loss": 0.0106, "step": 178720 }, { "epoch": 113.91332058636074, "grad_norm": 0.29288652539253235, "learning_rate": 3.840599239289733e-05, "loss": 0.0098, "step": 178730 }, { "epoch": 113.91969407265775, "grad_norm": 0.32392579317092896, "learning_rate": 3.8400631118534134e-05, "loss": 0.0079, "step": 178740 }, { "epoch": 113.92606755895474, "grad_norm": 0.3739319443702698, "learning_rate": 3.839526998511421e-05, "loss": 0.0139, "step": 178750 }, { "epoch": 113.93244104525175, "grad_norm": 0.5105111002922058, "learning_rate": 3.838990899270273e-05, "loss": 0.0121, "step": 178760 }, { "epoch": 113.93881453154876, "grad_norm": 0.4340144097805023, "learning_rate": 3.838454814136481e-05, "loss": 0.0135, "step": 178770 }, { "epoch": 113.94518801784577, "grad_norm": 0.3000466823577881, "learning_rate": 3.8379187431165586e-05, "loss": 0.008, "step": 178780 }, { "epoch": 113.95156150414276, "grad_norm": 0.18842081725597382, "learning_rate": 3.8373826862170225e-05, "loss": 0.0086, "step": 178790 }, { "epoch": 113.95793499043977, "grad_norm": 0.2754532992839813, "learning_rate": 3.836846643444383e-05, "loss": 0.0124, "step": 178800 }, { "epoch": 113.96430847673678, "grad_norm": 0.3731856048107147, "learning_rate": 3.8363106148051556e-05, "loss": 0.0231, "step": 178810 }, { "epoch": 113.97068196303378, "grad_norm": 0.28526633977890015, "learning_rate": 3.8357746003058514e-05, "loss": 0.01, "step": 178820 }, { "epoch": 113.97705544933078, "grad_norm": 0.17754250764846802, "learning_rate": 3.835238599952988e-05, "loss": 0.0095, "step": 178830 }, { "epoch": 113.98342893562778, "grad_norm": 0.14922389388084412, "learning_rate": 3.8347026137530736e-05, "loss": 0.0193, "step": 178840 }, { "epoch": 113.9898024219248, "grad_norm": 0.38445183634757996, "learning_rate": 3.834166641712622e-05, "loss": 0.0171, "step": 178850 }, { "epoch": 113.9961759082218, "grad_norm": 0.2785264253616333, "learning_rate": 3.833630683838149e-05, "loss": 0.0084, "step": 178860 }, { "epoch": 114.00254939451881, "grad_norm": 0.5824851989746094, "learning_rate": 3.833094740136162e-05, "loss": 0.0109, "step": 178870 }, { "epoch": 114.0089228808158, "grad_norm": 0.3256286084651947, "learning_rate": 3.832558810613176e-05, "loss": 0.0099, "step": 178880 }, { "epoch": 114.01529636711281, "grad_norm": 0.39145731925964355, "learning_rate": 3.832022895275702e-05, "loss": 0.0082, "step": 178890 }, { "epoch": 114.02166985340982, "grad_norm": 0.4356032609939575, "learning_rate": 3.8314869941302554e-05, "loss": 0.0055, "step": 178900 }, { "epoch": 114.02804333970683, "grad_norm": 0.16026492416858673, "learning_rate": 3.830951107183343e-05, "loss": 0.0063, "step": 178910 }, { "epoch": 114.03441682600382, "grad_norm": 0.6214796900749207, "learning_rate": 3.830415234441478e-05, "loss": 0.0149, "step": 178920 }, { "epoch": 114.04079031230083, "grad_norm": 0.21397894620895386, "learning_rate": 3.829879375911174e-05, "loss": 0.0205, "step": 178930 }, { "epoch": 114.04716379859784, "grad_norm": 0.2647179663181305, "learning_rate": 3.829343531598939e-05, "loss": 0.0094, "step": 178940 }, { "epoch": 114.05353728489484, "grad_norm": 0.4886988401412964, "learning_rate": 3.828807701511285e-05, "loss": 0.0097, "step": 178950 }, { "epoch": 114.05991077119184, "grad_norm": 0.26467233896255493, "learning_rate": 3.828271885654723e-05, "loss": 0.008, "step": 178960 }, { "epoch": 114.06628425748885, "grad_norm": 0.33581987023353577, "learning_rate": 3.827736084035766e-05, "loss": 0.0106, "step": 178970 }, { "epoch": 114.07265774378585, "grad_norm": 0.1675119549036026, "learning_rate": 3.827200296660921e-05, "loss": 0.0141, "step": 178980 }, { "epoch": 114.07903123008286, "grad_norm": 0.36346864700317383, "learning_rate": 3.8266645235367e-05, "loss": 0.0143, "step": 178990 }, { "epoch": 114.08540471637986, "grad_norm": 0.14711464941501617, "learning_rate": 3.8261287646696106e-05, "loss": 0.0132, "step": 179000 }, { "epoch": 114.09177820267686, "grad_norm": 0.2562950551509857, "learning_rate": 3.8255930200661695e-05, "loss": 0.0069, "step": 179010 }, { "epoch": 114.09815168897387, "grad_norm": 0.18231964111328125, "learning_rate": 3.825057289732879e-05, "loss": 0.0095, "step": 179020 }, { "epoch": 114.10452517527088, "grad_norm": 0.26838287711143494, "learning_rate": 3.824521573676251e-05, "loss": 0.0153, "step": 179030 }, { "epoch": 114.11089866156787, "grad_norm": 0.2383211851119995, "learning_rate": 3.8239858719027985e-05, "loss": 0.0098, "step": 179040 }, { "epoch": 114.11727214786488, "grad_norm": 0.22020594775676727, "learning_rate": 3.823450184419026e-05, "loss": 0.0094, "step": 179050 }, { "epoch": 114.12364563416189, "grad_norm": 0.2467908412218094, "learning_rate": 3.8229145112314454e-05, "loss": 0.0097, "step": 179060 }, { "epoch": 114.1300191204589, "grad_norm": 0.2797517478466034, "learning_rate": 3.8223788523465635e-05, "loss": 0.0087, "step": 179070 }, { "epoch": 114.13639260675589, "grad_norm": 0.418698787689209, "learning_rate": 3.821843207770893e-05, "loss": 0.0102, "step": 179080 }, { "epoch": 114.1427660930529, "grad_norm": 0.26065585017204285, "learning_rate": 3.821307577510938e-05, "loss": 0.0074, "step": 179090 }, { "epoch": 114.1491395793499, "grad_norm": 0.1309475153684616, "learning_rate": 3.8207719615732084e-05, "loss": 0.0105, "step": 179100 }, { "epoch": 114.15551306564691, "grad_norm": 0.16495440900325775, "learning_rate": 3.8202363599642146e-05, "loss": 0.0044, "step": 179110 }, { "epoch": 114.16188655194391, "grad_norm": 0.3254731595516205, "learning_rate": 3.819700772690462e-05, "loss": 0.0124, "step": 179120 }, { "epoch": 114.16826003824092, "grad_norm": 0.11331923305988312, "learning_rate": 3.8191651997584585e-05, "loss": 0.0208, "step": 179130 }, { "epoch": 114.17463352453792, "grad_norm": 0.39009782671928406, "learning_rate": 3.8186296411747133e-05, "loss": 0.0145, "step": 179140 }, { "epoch": 114.18100701083493, "grad_norm": 0.214320570230484, "learning_rate": 3.8180940969457345e-05, "loss": 0.0122, "step": 179150 }, { "epoch": 114.18738049713193, "grad_norm": 0.22692890465259552, "learning_rate": 3.817558567078028e-05, "loss": 0.0177, "step": 179160 }, { "epoch": 114.19375398342893, "grad_norm": 0.26568034291267395, "learning_rate": 3.8170230515781006e-05, "loss": 0.0066, "step": 179170 }, { "epoch": 114.20012746972594, "grad_norm": 0.1930224746465683, "learning_rate": 3.81648755045246e-05, "loss": 0.0079, "step": 179180 }, { "epoch": 114.20650095602295, "grad_norm": 0.21789012849330902, "learning_rate": 3.815952063707613e-05, "loss": 0.0211, "step": 179190 }, { "epoch": 114.21287444231994, "grad_norm": 0.3592105507850647, "learning_rate": 3.8154165913500665e-05, "loss": 0.0091, "step": 179200 }, { "epoch": 114.21924792861695, "grad_norm": 0.189369335770607, "learning_rate": 3.814881133386327e-05, "loss": 0.0117, "step": 179210 }, { "epoch": 114.22562141491396, "grad_norm": 0.24307966232299805, "learning_rate": 3.8143456898229006e-05, "loss": 0.018, "step": 179220 }, { "epoch": 114.23199490121097, "grad_norm": 0.3254278600215912, "learning_rate": 3.813810260666293e-05, "loss": 0.0057, "step": 179230 }, { "epoch": 114.23836838750796, "grad_norm": 0.13742564618587494, "learning_rate": 3.813274845923012e-05, "loss": 0.0142, "step": 179240 }, { "epoch": 114.24474187380497, "grad_norm": 0.11412722617387772, "learning_rate": 3.812739445599562e-05, "loss": 0.0095, "step": 179250 }, { "epoch": 114.25111536010198, "grad_norm": 0.47138461470603943, "learning_rate": 3.812204059702447e-05, "loss": 0.0144, "step": 179260 }, { "epoch": 114.25748884639899, "grad_norm": 0.14753375947475433, "learning_rate": 3.811668688238175e-05, "loss": 0.0102, "step": 179270 }, { "epoch": 114.26386233269598, "grad_norm": 0.19574135541915894, "learning_rate": 3.811133331213251e-05, "loss": 0.012, "step": 179280 }, { "epoch": 114.27023581899299, "grad_norm": 0.16649290919303894, "learning_rate": 3.8105979886341795e-05, "loss": 0.007, "step": 179290 }, { "epoch": 114.27660930529, "grad_norm": 0.3042882978916168, "learning_rate": 3.810062660507465e-05, "loss": 0.0239, "step": 179300 }, { "epoch": 114.282982791587, "grad_norm": 0.14687716960906982, "learning_rate": 3.8095273468396115e-05, "loss": 0.0118, "step": 179310 }, { "epoch": 114.289356277884, "grad_norm": 0.12676042318344116, "learning_rate": 3.8089920476371266e-05, "loss": 0.0122, "step": 179320 }, { "epoch": 114.295729764181, "grad_norm": 0.6552348732948303, "learning_rate": 3.8084567629065126e-05, "loss": 0.0155, "step": 179330 }, { "epoch": 114.30210325047801, "grad_norm": 0.2478150576353073, "learning_rate": 3.807921492654273e-05, "loss": 0.0145, "step": 179340 }, { "epoch": 114.30847673677502, "grad_norm": 0.3388139009475708, "learning_rate": 3.8073862368869124e-05, "loss": 0.0093, "step": 179350 }, { "epoch": 114.31485022307201, "grad_norm": 0.1648576408624649, "learning_rate": 3.806850995610937e-05, "loss": 0.011, "step": 179360 }, { "epoch": 114.32122370936902, "grad_norm": 0.3395031690597534, "learning_rate": 3.806315768832847e-05, "loss": 0.0115, "step": 179370 }, { "epoch": 114.32759719566603, "grad_norm": 0.21888487040996552, "learning_rate": 3.805780556559148e-05, "loss": 0.0081, "step": 179380 }, { "epoch": 114.33397068196304, "grad_norm": 0.16564522683620453, "learning_rate": 3.805245358796341e-05, "loss": 0.0108, "step": 179390 }, { "epoch": 114.34034416826003, "grad_norm": 0.2561747431755066, "learning_rate": 3.804710175550934e-05, "loss": 0.0086, "step": 179400 }, { "epoch": 114.34671765455704, "grad_norm": 0.5474560260772705, "learning_rate": 3.804175006829425e-05, "loss": 0.0169, "step": 179410 }, { "epoch": 114.35309114085405, "grad_norm": 0.11777635663747787, "learning_rate": 3.8036398526383186e-05, "loss": 0.0079, "step": 179420 }, { "epoch": 114.35946462715106, "grad_norm": 0.3301403224468231, "learning_rate": 3.80310471298412e-05, "loss": 0.0101, "step": 179430 }, { "epoch": 114.36583811344805, "grad_norm": 0.41252005100250244, "learning_rate": 3.802569587873326e-05, "loss": 0.0104, "step": 179440 }, { "epoch": 114.37221159974506, "grad_norm": 0.4045006036758423, "learning_rate": 3.802034477312443e-05, "loss": 0.009, "step": 179450 }, { "epoch": 114.37858508604207, "grad_norm": 0.11355694383382797, "learning_rate": 3.801499381307972e-05, "loss": 0.006, "step": 179460 }, { "epoch": 114.38495857233907, "grad_norm": 0.3029007315635681, "learning_rate": 3.8009642998664175e-05, "loss": 0.0132, "step": 179470 }, { "epoch": 114.39133205863607, "grad_norm": 0.2872891426086426, "learning_rate": 3.800429232994276e-05, "loss": 0.006, "step": 179480 }, { "epoch": 114.39770554493307, "grad_norm": 0.3044799864292145, "learning_rate": 3.799894180698052e-05, "loss": 0.0122, "step": 179490 }, { "epoch": 114.40407903123008, "grad_norm": 0.4049661159515381, "learning_rate": 3.79935914298425e-05, "loss": 0.0205, "step": 179500 }, { "epoch": 114.41045251752709, "grad_norm": 0.1834016889333725, "learning_rate": 3.7988241198593646e-05, "loss": 0.0119, "step": 179510 }, { "epoch": 114.4168260038241, "grad_norm": 0.13463814556598663, "learning_rate": 3.7982891113299e-05, "loss": 0.0151, "step": 179520 }, { "epoch": 114.42319949012109, "grad_norm": 0.30290406942367554, "learning_rate": 3.797754117402358e-05, "loss": 0.0083, "step": 179530 }, { "epoch": 114.4295729764181, "grad_norm": 0.23396538197994232, "learning_rate": 3.7972191380832414e-05, "loss": 0.0111, "step": 179540 }, { "epoch": 114.43594646271511, "grad_norm": 0.29909875988960266, "learning_rate": 3.7966841733790445e-05, "loss": 0.0137, "step": 179550 }, { "epoch": 114.44231994901212, "grad_norm": 0.26335540413856506, "learning_rate": 3.796149223296272e-05, "loss": 0.0106, "step": 179560 }, { "epoch": 114.44869343530911, "grad_norm": 0.620797336101532, "learning_rate": 3.7956142878414246e-05, "loss": 0.0243, "step": 179570 }, { "epoch": 114.45506692160612, "grad_norm": 0.22951249778270721, "learning_rate": 3.795079367020999e-05, "loss": 0.0174, "step": 179580 }, { "epoch": 114.46144040790313, "grad_norm": 0.5688453912734985, "learning_rate": 3.794544460841496e-05, "loss": 0.0145, "step": 179590 }, { "epoch": 114.46781389420013, "grad_norm": 0.2536564767360687, "learning_rate": 3.794009569309417e-05, "loss": 0.0096, "step": 179600 }, { "epoch": 114.47418738049713, "grad_norm": 0.4843027889728546, "learning_rate": 3.7934746924312615e-05, "loss": 0.0184, "step": 179610 }, { "epoch": 114.48056086679414, "grad_norm": 0.3170766532421112, "learning_rate": 3.792939830213526e-05, "loss": 0.0098, "step": 179620 }, { "epoch": 114.48693435309114, "grad_norm": 0.1328769326210022, "learning_rate": 3.792404982662711e-05, "loss": 0.0118, "step": 179630 }, { "epoch": 114.49330783938815, "grad_norm": 0.1797766536474228, "learning_rate": 3.791870149785316e-05, "loss": 0.0044, "step": 179640 }, { "epoch": 114.49968132568515, "grad_norm": 0.5185573697090149, "learning_rate": 3.7913353315878407e-05, "loss": 0.0079, "step": 179650 }, { "epoch": 114.50605481198215, "grad_norm": 0.29857420921325684, "learning_rate": 3.790800528076781e-05, "loss": 0.0088, "step": 179660 }, { "epoch": 114.51242829827916, "grad_norm": 0.6270190477371216, "learning_rate": 3.790265739258636e-05, "loss": 0.0114, "step": 179670 }, { "epoch": 114.51880178457617, "grad_norm": 0.19420453906059265, "learning_rate": 3.789730965139907e-05, "loss": 0.0186, "step": 179680 }, { "epoch": 114.52517527087316, "grad_norm": 0.13794712722301483, "learning_rate": 3.789196205727088e-05, "loss": 0.0059, "step": 179690 }, { "epoch": 114.53154875717017, "grad_norm": 0.1921842098236084, "learning_rate": 3.788661461026677e-05, "loss": 0.0101, "step": 179700 }, { "epoch": 114.53792224346718, "grad_norm": 0.4337097108364105, "learning_rate": 3.7881267310451726e-05, "loss": 0.022, "step": 179710 }, { "epoch": 114.54429572976419, "grad_norm": 0.27192944288253784, "learning_rate": 3.787592015789076e-05, "loss": 0.011, "step": 179720 }, { "epoch": 114.55066921606118, "grad_norm": 0.2075824737548828, "learning_rate": 3.7870573152648784e-05, "loss": 0.0125, "step": 179730 }, { "epoch": 114.55704270235819, "grad_norm": 0.14515624940395355, "learning_rate": 3.78652262947908e-05, "loss": 0.0082, "step": 179740 }, { "epoch": 114.5634161886552, "grad_norm": 0.23110423982143402, "learning_rate": 3.785987958438179e-05, "loss": 0.0106, "step": 179750 }, { "epoch": 114.5697896749522, "grad_norm": 0.3400672376155853, "learning_rate": 3.785453302148669e-05, "loss": 0.0065, "step": 179760 }, { "epoch": 114.5761631612492, "grad_norm": 0.20488505065441132, "learning_rate": 3.7849186606170474e-05, "loss": 0.0091, "step": 179770 }, { "epoch": 114.5825366475462, "grad_norm": 0.24144333600997925, "learning_rate": 3.784384033849812e-05, "loss": 0.0091, "step": 179780 }, { "epoch": 114.58891013384321, "grad_norm": 0.23610354959964752, "learning_rate": 3.78384942185346e-05, "loss": 0.0106, "step": 179790 }, { "epoch": 114.59528362014022, "grad_norm": 0.27359089255332947, "learning_rate": 3.7833148246344835e-05, "loss": 0.0089, "step": 179800 }, { "epoch": 114.60165710643722, "grad_norm": 0.4806097447872162, "learning_rate": 3.782780242199381e-05, "loss": 0.0076, "step": 179810 }, { "epoch": 114.60803059273422, "grad_norm": 0.3290145695209503, "learning_rate": 3.7822456745546486e-05, "loss": 0.006, "step": 179820 }, { "epoch": 114.61440407903123, "grad_norm": 0.16267041862010956, "learning_rate": 3.7817111217067816e-05, "loss": 0.0162, "step": 179830 }, { "epoch": 114.62077756532824, "grad_norm": 0.24568794667720795, "learning_rate": 3.781176583662272e-05, "loss": 0.0078, "step": 179840 }, { "epoch": 114.62715105162523, "grad_norm": 0.2681775987148285, "learning_rate": 3.780642060427619e-05, "loss": 0.0085, "step": 179850 }, { "epoch": 114.63352453792224, "grad_norm": 0.2566984295845032, "learning_rate": 3.7801075520093176e-05, "loss": 0.0072, "step": 179860 }, { "epoch": 114.63989802421925, "grad_norm": 0.15820126235485077, "learning_rate": 3.77957305841386e-05, "loss": 0.0123, "step": 179870 }, { "epoch": 114.64627151051626, "grad_norm": 0.21655522286891937, "learning_rate": 3.779038579647742e-05, "loss": 0.0106, "step": 179880 }, { "epoch": 114.65264499681325, "grad_norm": 0.19644831120967865, "learning_rate": 3.7785041157174584e-05, "loss": 0.0076, "step": 179890 }, { "epoch": 114.65901848311026, "grad_norm": 0.19673460721969604, "learning_rate": 3.7779696666295025e-05, "loss": 0.0071, "step": 179900 }, { "epoch": 114.66539196940727, "grad_norm": 0.25686290860176086, "learning_rate": 3.77743523239037e-05, "loss": 0.0069, "step": 179910 }, { "epoch": 114.67176545570427, "grad_norm": 0.26764389872550964, "learning_rate": 3.776900813006553e-05, "loss": 0.0061, "step": 179920 }, { "epoch": 114.67813894200127, "grad_norm": 0.1584281027317047, "learning_rate": 3.776366408484547e-05, "loss": 0.0069, "step": 179930 }, { "epoch": 114.68451242829828, "grad_norm": 0.25838735699653625, "learning_rate": 3.775832018830843e-05, "loss": 0.0087, "step": 179940 }, { "epoch": 114.69088591459528, "grad_norm": 0.3144531846046448, "learning_rate": 3.775297644051936e-05, "loss": 0.005, "step": 179950 }, { "epoch": 114.69725940089229, "grad_norm": 0.20965544879436493, "learning_rate": 3.774763284154319e-05, "loss": 0.0147, "step": 179960 }, { "epoch": 114.70363288718929, "grad_norm": 0.3269522488117218, "learning_rate": 3.7742289391444855e-05, "loss": 0.006, "step": 179970 }, { "epoch": 114.7100063734863, "grad_norm": 0.3857893645763397, "learning_rate": 3.773694609028927e-05, "loss": 0.0175, "step": 179980 }, { "epoch": 114.7163798597833, "grad_norm": 0.35083895921707153, "learning_rate": 3.773160293814138e-05, "loss": 0.0124, "step": 179990 }, { "epoch": 114.72275334608031, "grad_norm": 0.30971619486808777, "learning_rate": 3.772625993506609e-05, "loss": 0.0056, "step": 180000 }, { "epoch": 114.7291268323773, "grad_norm": 0.3121204376220703, "learning_rate": 3.772091708112832e-05, "loss": 0.0192, "step": 180010 }, { "epoch": 114.73550031867431, "grad_norm": 0.3081899583339691, "learning_rate": 3.7715574376393006e-05, "loss": 0.0146, "step": 180020 }, { "epoch": 114.74187380497132, "grad_norm": 0.2197681963443756, "learning_rate": 3.771023182092507e-05, "loss": 0.0055, "step": 180030 }, { "epoch": 114.74824729126833, "grad_norm": 0.2648077607154846, "learning_rate": 3.770488941478942e-05, "loss": 0.0152, "step": 180040 }, { "epoch": 114.75462077756532, "grad_norm": 0.33885523676872253, "learning_rate": 3.769954715805096e-05, "loss": 0.0055, "step": 180050 }, { "epoch": 114.76099426386233, "grad_norm": 0.3850075900554657, "learning_rate": 3.769420505077462e-05, "loss": 0.0077, "step": 180060 }, { "epoch": 114.76736775015934, "grad_norm": 0.42408859729766846, "learning_rate": 3.768886309302533e-05, "loss": 0.0139, "step": 180070 }, { "epoch": 114.77374123645635, "grad_norm": 0.17810271680355072, "learning_rate": 3.768352128486795e-05, "loss": 0.005, "step": 180080 }, { "epoch": 114.78011472275334, "grad_norm": 0.33928370475769043, "learning_rate": 3.7678179626367424e-05, "loss": 0.0126, "step": 180090 }, { "epoch": 114.78648820905035, "grad_norm": 0.1717829555273056, "learning_rate": 3.767283811758864e-05, "loss": 0.0116, "step": 180100 }, { "epoch": 114.79286169534736, "grad_norm": 0.26002660393714905, "learning_rate": 3.7667496758596535e-05, "loss": 0.0099, "step": 180110 }, { "epoch": 114.79923518164436, "grad_norm": 0.1512378752231598, "learning_rate": 3.766215554945597e-05, "loss": 0.018, "step": 180120 }, { "epoch": 114.80560866794136, "grad_norm": 0.42949575185775757, "learning_rate": 3.7656814490231865e-05, "loss": 0.0123, "step": 180130 }, { "epoch": 114.81198215423836, "grad_norm": 0.46737971901893616, "learning_rate": 3.765147358098913e-05, "loss": 0.0134, "step": 180140 }, { "epoch": 114.81835564053537, "grad_norm": 0.3011957108974457, "learning_rate": 3.764613282179263e-05, "loss": 0.021, "step": 180150 }, { "epoch": 114.82472912683238, "grad_norm": 0.32599112391471863, "learning_rate": 3.764079221270729e-05, "loss": 0.0136, "step": 180160 }, { "epoch": 114.83110261312939, "grad_norm": 0.5745822191238403, "learning_rate": 3.763545175379799e-05, "loss": 0.0081, "step": 180170 }, { "epoch": 114.83747609942638, "grad_norm": 0.26554885506629944, "learning_rate": 3.7630111445129645e-05, "loss": 0.0096, "step": 180180 }, { "epoch": 114.84384958572339, "grad_norm": 0.1504933089017868, "learning_rate": 3.7624771286767104e-05, "loss": 0.0142, "step": 180190 }, { "epoch": 114.8502230720204, "grad_norm": 0.22188213467597961, "learning_rate": 3.761943127877527e-05, "loss": 0.0137, "step": 180200 }, { "epoch": 114.8565965583174, "grad_norm": 0.15192896127700806, "learning_rate": 3.761409142121905e-05, "loss": 0.0119, "step": 180210 }, { "epoch": 114.8629700446144, "grad_norm": 0.1591348946094513, "learning_rate": 3.7608751714163325e-05, "loss": 0.0131, "step": 180220 }, { "epoch": 114.86934353091141, "grad_norm": 0.27088218927383423, "learning_rate": 3.7603412157672946e-05, "loss": 0.0084, "step": 180230 }, { "epoch": 114.87571701720842, "grad_norm": 0.31736212968826294, "learning_rate": 3.7598072751812815e-05, "loss": 0.0088, "step": 180240 }, { "epoch": 114.88209050350542, "grad_norm": 0.2168317437171936, "learning_rate": 3.759273349664783e-05, "loss": 0.0069, "step": 180250 }, { "epoch": 114.88846398980242, "grad_norm": 0.16308878362178802, "learning_rate": 3.758739439224282e-05, "loss": 0.0143, "step": 180260 }, { "epoch": 114.89483747609943, "grad_norm": 0.15993016958236694, "learning_rate": 3.7582055438662694e-05, "loss": 0.0076, "step": 180270 }, { "epoch": 114.90121096239643, "grad_norm": 0.15241117775440216, "learning_rate": 3.757671663597231e-05, "loss": 0.008, "step": 180280 }, { "epoch": 114.90758444869344, "grad_norm": 0.6792674660682678, "learning_rate": 3.7571377984236585e-05, "loss": 0.0093, "step": 180290 }, { "epoch": 114.91395793499044, "grad_norm": 0.2150624692440033, "learning_rate": 3.756603948352032e-05, "loss": 0.0217, "step": 180300 }, { "epoch": 114.92033142128744, "grad_norm": 0.3117007613182068, "learning_rate": 3.756070113388842e-05, "loss": 0.0104, "step": 180310 }, { "epoch": 114.92670490758445, "grad_norm": 0.13045284152030945, "learning_rate": 3.7555362935405764e-05, "loss": 0.01, "step": 180320 }, { "epoch": 114.93307839388146, "grad_norm": 0.1526012122631073, "learning_rate": 3.755002488813718e-05, "loss": 0.0188, "step": 180330 }, { "epoch": 114.93945188017845, "grad_norm": 0.18327409029006958, "learning_rate": 3.754468699214754e-05, "loss": 0.0059, "step": 180340 }, { "epoch": 114.94582536647546, "grad_norm": 0.2618985176086426, "learning_rate": 3.753934924750172e-05, "loss": 0.0059, "step": 180350 }, { "epoch": 114.95219885277247, "grad_norm": 0.34466907382011414, "learning_rate": 3.753401165426459e-05, "loss": 0.0132, "step": 180360 }, { "epoch": 114.95857233906948, "grad_norm": 0.29125386476516724, "learning_rate": 3.752867421250097e-05, "loss": 0.0124, "step": 180370 }, { "epoch": 114.96494582536647, "grad_norm": 0.38835835456848145, "learning_rate": 3.752333692227573e-05, "loss": 0.0169, "step": 180380 }, { "epoch": 114.97131931166348, "grad_norm": 0.36682847142219543, "learning_rate": 3.7517999783653754e-05, "loss": 0.0116, "step": 180390 }, { "epoch": 114.97769279796049, "grad_norm": 0.2789430320262909, "learning_rate": 3.751266279669984e-05, "loss": 0.0099, "step": 180400 }, { "epoch": 114.9840662842575, "grad_norm": 0.3850061893463135, "learning_rate": 3.750732596147886e-05, "loss": 0.0113, "step": 180410 }, { "epoch": 114.99043977055449, "grad_norm": 0.3722115457057953, "learning_rate": 3.750198927805567e-05, "loss": 0.0166, "step": 180420 }, { "epoch": 114.9968132568515, "grad_norm": 0.22111797332763672, "learning_rate": 3.749665274649513e-05, "loss": 0.007, "step": 180430 }, { "epoch": 115.0031867431485, "grad_norm": 0.2410888671875, "learning_rate": 3.749131636686204e-05, "loss": 0.0141, "step": 180440 }, { "epoch": 115.00956022944551, "grad_norm": 0.4016611874103546, "learning_rate": 3.748598013922127e-05, "loss": 0.007, "step": 180450 }, { "epoch": 115.0159337157425, "grad_norm": 0.23522984981536865, "learning_rate": 3.748064406363767e-05, "loss": 0.0114, "step": 180460 }, { "epoch": 115.02230720203951, "grad_norm": 0.5208233594894409, "learning_rate": 3.747530814017605e-05, "loss": 0.0105, "step": 180470 }, { "epoch": 115.02868068833652, "grad_norm": 0.5982618927955627, "learning_rate": 3.746997236890126e-05, "loss": 0.011, "step": 180480 }, { "epoch": 115.03505417463353, "grad_norm": 0.19013425707817078, "learning_rate": 3.746463674987814e-05, "loss": 0.0069, "step": 180490 }, { "epoch": 115.04142766093052, "grad_norm": 0.2739647924900055, "learning_rate": 3.7459301283171536e-05, "loss": 0.0133, "step": 180500 }, { "epoch": 115.04780114722753, "grad_norm": 0.41112104058265686, "learning_rate": 3.745396596884624e-05, "loss": 0.0102, "step": 180510 }, { "epoch": 115.05417463352454, "grad_norm": 0.23712483048439026, "learning_rate": 3.744863080696711e-05, "loss": 0.0065, "step": 180520 }, { "epoch": 115.06054811982155, "grad_norm": 0.1679321825504303, "learning_rate": 3.744329579759895e-05, "loss": 0.0154, "step": 180530 }, { "epoch": 115.06692160611854, "grad_norm": 0.1697796732187271, "learning_rate": 3.743796094080664e-05, "loss": 0.0151, "step": 180540 }, { "epoch": 115.07329509241555, "grad_norm": 0.18816380202770233, "learning_rate": 3.7432626236654936e-05, "loss": 0.0235, "step": 180550 }, { "epoch": 115.07966857871256, "grad_norm": 0.39988449215888977, "learning_rate": 3.7427291685208694e-05, "loss": 0.0204, "step": 180560 }, { "epoch": 115.08604206500956, "grad_norm": 0.6214980483055115, "learning_rate": 3.742195728653273e-05, "loss": 0.0115, "step": 180570 }, { "epoch": 115.09241555130656, "grad_norm": 0.28837141394615173, "learning_rate": 3.741662304069187e-05, "loss": 0.0148, "step": 180580 }, { "epoch": 115.09878903760357, "grad_norm": 0.21566234529018402, "learning_rate": 3.7411288947750903e-05, "loss": 0.0123, "step": 180590 }, { "epoch": 115.10516252390057, "grad_norm": 0.28951627016067505, "learning_rate": 3.740595500777466e-05, "loss": 0.0142, "step": 180600 }, { "epoch": 115.11153601019758, "grad_norm": 0.16818267107009888, "learning_rate": 3.7400621220827974e-05, "loss": 0.0088, "step": 180610 }, { "epoch": 115.11790949649458, "grad_norm": 0.1770419329404831, "learning_rate": 3.7395287586975633e-05, "loss": 0.009, "step": 180620 }, { "epoch": 115.12428298279158, "grad_norm": 0.19353865087032318, "learning_rate": 3.7389954106282435e-05, "loss": 0.0137, "step": 180630 }, { "epoch": 115.13065646908859, "grad_norm": 0.5801668763160706, "learning_rate": 3.7384620778813214e-05, "loss": 0.0115, "step": 180640 }, { "epoch": 115.1370299553856, "grad_norm": 0.19017276167869568, "learning_rate": 3.737928760463275e-05, "loss": 0.0055, "step": 180650 }, { "epoch": 115.1434034416826, "grad_norm": 0.3181699812412262, "learning_rate": 3.7373954583805864e-05, "loss": 0.01, "step": 180660 }, { "epoch": 115.1497769279796, "grad_norm": 0.34334778785705566, "learning_rate": 3.7368621716397344e-05, "loss": 0.0107, "step": 180670 }, { "epoch": 115.15615041427661, "grad_norm": 0.23459181189537048, "learning_rate": 3.7363289002472e-05, "loss": 0.0132, "step": 180680 }, { "epoch": 115.16252390057362, "grad_norm": 0.18372194468975067, "learning_rate": 3.735795644209463e-05, "loss": 0.0104, "step": 180690 }, { "epoch": 115.16889738687061, "grad_norm": 0.5120564103126526, "learning_rate": 3.735262403533002e-05, "loss": 0.0129, "step": 180700 }, { "epoch": 115.17527087316762, "grad_norm": 0.25479695200920105, "learning_rate": 3.734729178224298e-05, "loss": 0.0076, "step": 180710 }, { "epoch": 115.18164435946463, "grad_norm": 0.9311663508415222, "learning_rate": 3.7341959682898274e-05, "loss": 0.012, "step": 180720 }, { "epoch": 115.18801784576164, "grad_norm": 0.3108302354812622, "learning_rate": 3.733662773736071e-05, "loss": 0.0094, "step": 180730 }, { "epoch": 115.19439133205863, "grad_norm": 0.46902865171432495, "learning_rate": 3.7331295945695087e-05, "loss": 0.0121, "step": 180740 }, { "epoch": 115.20076481835564, "grad_norm": 0.2681962251663208, "learning_rate": 3.732596430796618e-05, "loss": 0.0065, "step": 180750 }, { "epoch": 115.20713830465264, "grad_norm": 0.5987251996994019, "learning_rate": 3.7320632824238766e-05, "loss": 0.0138, "step": 180760 }, { "epoch": 115.21351179094965, "grad_norm": 0.2055082619190216, "learning_rate": 3.7315301494577626e-05, "loss": 0.0088, "step": 180770 }, { "epoch": 115.21988527724666, "grad_norm": 0.25420084595680237, "learning_rate": 3.730997031904757e-05, "loss": 0.0115, "step": 180780 }, { "epoch": 115.22625876354365, "grad_norm": 0.32305777072906494, "learning_rate": 3.730463929771335e-05, "loss": 0.009, "step": 180790 }, { "epoch": 115.23263224984066, "grad_norm": 0.28831952810287476, "learning_rate": 3.7299308430639735e-05, "loss": 0.011, "step": 180800 }, { "epoch": 115.23900573613767, "grad_norm": 0.2398931235074997, "learning_rate": 3.7293977717891515e-05, "loss": 0.0085, "step": 180810 }, { "epoch": 115.24537922243468, "grad_norm": 0.45463427901268005, "learning_rate": 3.728864715953349e-05, "loss": 0.01, "step": 180820 }, { "epoch": 115.25175270873167, "grad_norm": 0.3590075969696045, "learning_rate": 3.7283316755630374e-05, "loss": 0.0078, "step": 180830 }, { "epoch": 115.25812619502868, "grad_norm": 0.5264917612075806, "learning_rate": 3.7277986506246974e-05, "loss": 0.0098, "step": 180840 }, { "epoch": 115.26449968132569, "grad_norm": 0.4237397611141205, "learning_rate": 3.727265641144805e-05, "loss": 0.0112, "step": 180850 }, { "epoch": 115.2708731676227, "grad_norm": 0.24038569629192352, "learning_rate": 3.726732647129838e-05, "loss": 0.0139, "step": 180860 }, { "epoch": 115.27724665391969, "grad_norm": 0.3369421362876892, "learning_rate": 3.72619966858627e-05, "loss": 0.0312, "step": 180870 }, { "epoch": 115.2836201402167, "grad_norm": 0.34847381711006165, "learning_rate": 3.725666705520579e-05, "loss": 0.0114, "step": 180880 }, { "epoch": 115.2899936265137, "grad_norm": 0.647676408290863, "learning_rate": 3.725133757939242e-05, "loss": 0.0118, "step": 180890 }, { "epoch": 115.29636711281071, "grad_norm": 0.41969218850135803, "learning_rate": 3.7246008258487314e-05, "loss": 0.0089, "step": 180900 }, { "epoch": 115.30274059910771, "grad_norm": 0.31101441383361816, "learning_rate": 3.7240679092555256e-05, "loss": 0.0098, "step": 180910 }, { "epoch": 115.30911408540472, "grad_norm": 0.5063458681106567, "learning_rate": 3.7235350081660985e-05, "loss": 0.0096, "step": 180920 }, { "epoch": 115.31548757170172, "grad_norm": 0.19532141089439392, "learning_rate": 3.723002122586929e-05, "loss": 0.0187, "step": 180930 }, { "epoch": 115.32186105799873, "grad_norm": 0.2082308679819107, "learning_rate": 3.722469252524487e-05, "loss": 0.0071, "step": 180940 }, { "epoch": 115.32823454429573, "grad_norm": 0.23499709367752075, "learning_rate": 3.72193639798525e-05, "loss": 0.0098, "step": 180950 }, { "epoch": 115.33460803059273, "grad_norm": 0.247036874294281, "learning_rate": 3.7214035589756944e-05, "loss": 0.0081, "step": 180960 }, { "epoch": 115.34098151688974, "grad_norm": 0.2813511788845062, "learning_rate": 3.720870735502291e-05, "loss": 0.015, "step": 180970 }, { "epoch": 115.34735500318675, "grad_norm": 0.1931372582912445, "learning_rate": 3.720337927571515e-05, "loss": 0.0137, "step": 180980 }, { "epoch": 115.35372848948374, "grad_norm": 0.3731168210506439, "learning_rate": 3.719805135189842e-05, "loss": 0.0132, "step": 180990 }, { "epoch": 115.36010197578075, "grad_norm": 0.21545180678367615, "learning_rate": 3.719272358363748e-05, "loss": 0.0098, "step": 181000 }, { "epoch": 115.36647546207776, "grad_norm": 0.3766053318977356, "learning_rate": 3.7187395970997014e-05, "loss": 0.0071, "step": 181010 }, { "epoch": 115.37284894837477, "grad_norm": 0.18712681531906128, "learning_rate": 3.718206851404178e-05, "loss": 0.0061, "step": 181020 }, { "epoch": 115.37922243467176, "grad_norm": 0.29028376936912537, "learning_rate": 3.717674121283654e-05, "loss": 0.0107, "step": 181030 }, { "epoch": 115.38559592096877, "grad_norm": 0.1425355076789856, "learning_rate": 3.717141406744599e-05, "loss": 0.0213, "step": 181040 }, { "epoch": 115.39196940726578, "grad_norm": 0.23351071774959564, "learning_rate": 3.716608707793485e-05, "loss": 0.019, "step": 181050 }, { "epoch": 115.39834289356278, "grad_norm": 0.15930454432964325, "learning_rate": 3.716076024436789e-05, "loss": 0.0104, "step": 181060 }, { "epoch": 115.40471637985978, "grad_norm": 0.2625936269760132, "learning_rate": 3.715543356680982e-05, "loss": 0.0113, "step": 181070 }, { "epoch": 115.41108986615679, "grad_norm": 0.6321964859962463, "learning_rate": 3.715010704532535e-05, "loss": 0.0095, "step": 181080 }, { "epoch": 115.4174633524538, "grad_norm": 0.2586086094379425, "learning_rate": 3.714478067997921e-05, "loss": 0.0072, "step": 181090 }, { "epoch": 115.4238368387508, "grad_norm": 0.09259876608848572, "learning_rate": 3.713945447083612e-05, "loss": 0.0131, "step": 181100 }, { "epoch": 115.4302103250478, "grad_norm": 0.32159924507141113, "learning_rate": 3.7134128417960816e-05, "loss": 0.013, "step": 181110 }, { "epoch": 115.4365838113448, "grad_norm": 0.14496684074401855, "learning_rate": 3.712880252141798e-05, "loss": 0.0083, "step": 181120 }, { "epoch": 115.44295729764181, "grad_norm": 0.38036513328552246, "learning_rate": 3.712347678127235e-05, "loss": 0.0101, "step": 181130 }, { "epoch": 115.44933078393882, "grad_norm": 0.1970447450876236, "learning_rate": 3.7118151197588655e-05, "loss": 0.0073, "step": 181140 }, { "epoch": 115.45570427023581, "grad_norm": 0.2213088423013687, "learning_rate": 3.7112825770431567e-05, "loss": 0.0099, "step": 181150 }, { "epoch": 115.46207775653282, "grad_norm": 0.2644363343715668, "learning_rate": 3.710750049986581e-05, "loss": 0.0124, "step": 181160 }, { "epoch": 115.46845124282983, "grad_norm": 0.27851349115371704, "learning_rate": 3.7102175385956095e-05, "loss": 0.0086, "step": 181170 }, { "epoch": 115.47482472912684, "grad_norm": 0.14991919696331024, "learning_rate": 3.7096850428767147e-05, "loss": 0.0197, "step": 181180 }, { "epoch": 115.48119821542383, "grad_norm": 0.2690393924713135, "learning_rate": 3.709152562836363e-05, "loss": 0.0148, "step": 181190 }, { "epoch": 115.48757170172084, "grad_norm": 0.23588962852954865, "learning_rate": 3.708620098481026e-05, "loss": 0.018, "step": 181200 }, { "epoch": 115.49394518801785, "grad_norm": 0.34782683849334717, "learning_rate": 3.708087649817177e-05, "loss": 0.0086, "step": 181210 }, { "epoch": 115.50031867431485, "grad_norm": 0.32915642857551575, "learning_rate": 3.7075552168512804e-05, "loss": 0.0126, "step": 181220 }, { "epoch": 115.50669216061185, "grad_norm": 0.1662992537021637, "learning_rate": 3.7070227995898076e-05, "loss": 0.0116, "step": 181230 }, { "epoch": 115.51306564690886, "grad_norm": 0.2794587314128876, "learning_rate": 3.706490398039229e-05, "loss": 0.0098, "step": 181240 }, { "epoch": 115.51943913320586, "grad_norm": 0.24081909656524658, "learning_rate": 3.705958012206016e-05, "loss": 0.0148, "step": 181250 }, { "epoch": 115.52581261950287, "grad_norm": 0.4775119125843048, "learning_rate": 3.7054256420966317e-05, "loss": 0.0153, "step": 181260 }, { "epoch": 115.53218610579987, "grad_norm": 0.20602594316005707, "learning_rate": 3.704893287717548e-05, "loss": 0.008, "step": 181270 }, { "epoch": 115.53855959209687, "grad_norm": 0.18081866204738617, "learning_rate": 3.704360949075235e-05, "loss": 0.0081, "step": 181280 }, { "epoch": 115.54493307839388, "grad_norm": 0.2920440435409546, "learning_rate": 3.703828626176161e-05, "loss": 0.008, "step": 181290 }, { "epoch": 115.55130656469089, "grad_norm": 0.42640209197998047, "learning_rate": 3.703296319026791e-05, "loss": 0.0114, "step": 181300 }, { "epoch": 115.55768005098788, "grad_norm": 0.2618543803691864, "learning_rate": 3.702764027633595e-05, "loss": 0.0072, "step": 181310 }, { "epoch": 115.56405353728489, "grad_norm": 0.3202635943889618, "learning_rate": 3.7022317520030416e-05, "loss": 0.0122, "step": 181320 }, { "epoch": 115.5704270235819, "grad_norm": 0.28097787499427795, "learning_rate": 3.7016994921415984e-05, "loss": 0.0058, "step": 181330 }, { "epoch": 115.57680050987891, "grad_norm": 0.792365312576294, "learning_rate": 3.7011672480557306e-05, "loss": 0.0359, "step": 181340 }, { "epoch": 115.5831739961759, "grad_norm": 0.2414252609014511, "learning_rate": 3.700635019751908e-05, "loss": 0.0104, "step": 181350 }, { "epoch": 115.58954748247291, "grad_norm": 0.38622841238975525, "learning_rate": 3.700102807236596e-05, "loss": 0.0167, "step": 181360 }, { "epoch": 115.59592096876992, "grad_norm": 0.25104761123657227, "learning_rate": 3.699570610516263e-05, "loss": 0.0153, "step": 181370 }, { "epoch": 115.60229445506693, "grad_norm": 0.21203306317329407, "learning_rate": 3.6990384295973736e-05, "loss": 0.0072, "step": 181380 }, { "epoch": 115.60866794136393, "grad_norm": 0.1302487552165985, "learning_rate": 3.698506264486397e-05, "loss": 0.0126, "step": 181390 }, { "epoch": 115.61504142766093, "grad_norm": 0.19655723869800568, "learning_rate": 3.697974115189797e-05, "loss": 0.0074, "step": 181400 }, { "epoch": 115.62141491395793, "grad_norm": 0.1546267718076706, "learning_rate": 3.6974419817140424e-05, "loss": 0.0041, "step": 181410 }, { "epoch": 115.62778840025494, "grad_norm": 0.5550719499588013, "learning_rate": 3.696909864065596e-05, "loss": 0.0111, "step": 181420 }, { "epoch": 115.63416188655195, "grad_norm": 0.15130937099456787, "learning_rate": 3.696377762250926e-05, "loss": 0.0099, "step": 181430 }, { "epoch": 115.64053537284894, "grad_norm": 0.18788674473762512, "learning_rate": 3.695845676276496e-05, "loss": 0.008, "step": 181440 }, { "epoch": 115.64690885914595, "grad_norm": 0.45777225494384766, "learning_rate": 3.6953136061487737e-05, "loss": 0.0139, "step": 181450 }, { "epoch": 115.65328234544296, "grad_norm": 0.2713438868522644, "learning_rate": 3.694781551874222e-05, "loss": 0.0138, "step": 181460 }, { "epoch": 115.65965583173997, "grad_norm": 0.3295477032661438, "learning_rate": 3.694249513459307e-05, "loss": 0.0141, "step": 181470 }, { "epoch": 115.66602931803696, "grad_norm": 0.2132573425769806, "learning_rate": 3.693717490910493e-05, "loss": 0.0071, "step": 181480 }, { "epoch": 115.67240280433397, "grad_norm": 0.3887552320957184, "learning_rate": 3.693185484234245e-05, "loss": 0.0125, "step": 181490 }, { "epoch": 115.67877629063098, "grad_norm": 0.24029071629047394, "learning_rate": 3.692653493437028e-05, "loss": 0.0076, "step": 181500 }, { "epoch": 115.68514977692799, "grad_norm": 0.147639200091362, "learning_rate": 3.692121518525305e-05, "loss": 0.0097, "step": 181510 }, { "epoch": 115.69152326322498, "grad_norm": 0.5957785844802856, "learning_rate": 3.691589559505538e-05, "loss": 0.017, "step": 181520 }, { "epoch": 115.69789674952199, "grad_norm": 0.20489494502544403, "learning_rate": 3.6910576163841976e-05, "loss": 0.0118, "step": 181530 }, { "epoch": 115.704270235819, "grad_norm": 0.45662403106689453, "learning_rate": 3.69052568916774e-05, "loss": 0.0345, "step": 181540 }, { "epoch": 115.710643722116, "grad_norm": 0.42946484684944153, "learning_rate": 3.689993777862631e-05, "loss": 0.0087, "step": 181550 }, { "epoch": 115.717017208413, "grad_norm": 0.13882233202457428, "learning_rate": 3.689461882475336e-05, "loss": 0.0066, "step": 181560 }, { "epoch": 115.72339069471, "grad_norm": 0.3927915394306183, "learning_rate": 3.6889300030123174e-05, "loss": 0.0072, "step": 181570 }, { "epoch": 115.72976418100701, "grad_norm": 0.3365705609321594, "learning_rate": 3.688398139480036e-05, "loss": 0.0092, "step": 181580 }, { "epoch": 115.73613766730402, "grad_norm": 0.5106320977210999, "learning_rate": 3.687866291884955e-05, "loss": 0.0138, "step": 181590 }, { "epoch": 115.74251115360101, "grad_norm": 0.1940380036830902, "learning_rate": 3.687334460233539e-05, "loss": 0.0136, "step": 181600 }, { "epoch": 115.74888463989802, "grad_norm": 0.20328262448310852, "learning_rate": 3.686802644532247e-05, "loss": 0.0065, "step": 181610 }, { "epoch": 115.75525812619503, "grad_norm": 0.2799372673034668, "learning_rate": 3.6862708447875424e-05, "loss": 0.0242, "step": 181620 }, { "epoch": 115.76163161249204, "grad_norm": 0.24162384867668152, "learning_rate": 3.685739061005888e-05, "loss": 0.0077, "step": 181630 }, { "epoch": 115.76800509878903, "grad_norm": 0.4999712407588959, "learning_rate": 3.685207293193747e-05, "loss": 0.0101, "step": 181640 }, { "epoch": 115.77437858508604, "grad_norm": 0.40922266244888306, "learning_rate": 3.684675541357576e-05, "loss": 0.0116, "step": 181650 }, { "epoch": 115.78075207138305, "grad_norm": 0.17487871646881104, "learning_rate": 3.6841438055038405e-05, "loss": 0.0087, "step": 181660 }, { "epoch": 115.78712555768006, "grad_norm": 0.22431102395057678, "learning_rate": 3.6836120856390004e-05, "loss": 0.0054, "step": 181670 }, { "epoch": 115.79349904397705, "grad_norm": 0.2555784285068512, "learning_rate": 3.6830803817695156e-05, "loss": 0.0129, "step": 181680 }, { "epoch": 115.79987253027406, "grad_norm": 0.20146261155605316, "learning_rate": 3.6825486939018466e-05, "loss": 0.0219, "step": 181690 }, { "epoch": 115.80624601657107, "grad_norm": 0.3329574465751648, "learning_rate": 3.682017022042456e-05, "loss": 0.0075, "step": 181700 }, { "epoch": 115.81261950286807, "grad_norm": 0.30672529339790344, "learning_rate": 3.681485366197803e-05, "loss": 0.009, "step": 181710 }, { "epoch": 115.81899298916507, "grad_norm": 0.29456862807273865, "learning_rate": 3.6809537263743476e-05, "loss": 0.0091, "step": 181720 }, { "epoch": 115.82536647546208, "grad_norm": 0.4135054349899292, "learning_rate": 3.680422102578549e-05, "loss": 0.0183, "step": 181730 }, { "epoch": 115.83173996175908, "grad_norm": 0.3635327219963074, "learning_rate": 3.6798904948168676e-05, "loss": 0.0121, "step": 181740 }, { "epoch": 115.83811344805609, "grad_norm": 0.27099987864494324, "learning_rate": 3.6793589030957654e-05, "loss": 0.0134, "step": 181750 }, { "epoch": 115.84448693435309, "grad_norm": 0.23416276276111603, "learning_rate": 3.678827327421697e-05, "loss": 0.0075, "step": 181760 }, { "epoch": 115.8508604206501, "grad_norm": 0.26648658514022827, "learning_rate": 3.678295767801124e-05, "loss": 0.0074, "step": 181770 }, { "epoch": 115.8572339069471, "grad_norm": 0.3718636929988861, "learning_rate": 3.677764224240509e-05, "loss": 0.0135, "step": 181780 }, { "epoch": 115.86360739324411, "grad_norm": 0.24047234654426575, "learning_rate": 3.6772326967463026e-05, "loss": 0.0091, "step": 181790 }, { "epoch": 115.8699808795411, "grad_norm": 0.08357301354408264, "learning_rate": 3.676701185324969e-05, "loss": 0.0093, "step": 181800 }, { "epoch": 115.87635436583811, "grad_norm": 0.3176027536392212, "learning_rate": 3.6761696899829656e-05, "loss": 0.0087, "step": 181810 }, { "epoch": 115.88272785213512, "grad_norm": 0.35016316175460815, "learning_rate": 3.675638210726752e-05, "loss": 0.0098, "step": 181820 }, { "epoch": 115.88910133843213, "grad_norm": 0.2151004672050476, "learning_rate": 3.675106747562783e-05, "loss": 0.0087, "step": 181830 }, { "epoch": 115.89547482472912, "grad_norm": 0.15078310668468475, "learning_rate": 3.674575300497517e-05, "loss": 0.0095, "step": 181840 }, { "epoch": 115.90184831102613, "grad_norm": 0.15212382376194, "learning_rate": 3.674043869537416e-05, "loss": 0.0121, "step": 181850 }, { "epoch": 115.90822179732314, "grad_norm": 0.46663108468055725, "learning_rate": 3.67351245468893e-05, "loss": 0.0146, "step": 181860 }, { "epoch": 115.91459528362014, "grad_norm": 0.3257564604282379, "learning_rate": 3.672981055958521e-05, "loss": 0.0071, "step": 181870 }, { "epoch": 115.92096876991714, "grad_norm": 0.2028525322675705, "learning_rate": 3.672449673352645e-05, "loss": 0.0076, "step": 181880 }, { "epoch": 115.92734225621415, "grad_norm": 0.15567398071289062, "learning_rate": 3.671918306877762e-05, "loss": 0.0129, "step": 181890 }, { "epoch": 115.93371574251115, "grad_norm": 0.36891257762908936, "learning_rate": 3.6713869565403216e-05, "loss": 0.0059, "step": 181900 }, { "epoch": 115.94008922880816, "grad_norm": 0.1508672833442688, "learning_rate": 3.6708556223467846e-05, "loss": 0.011, "step": 181910 }, { "epoch": 115.94646271510516, "grad_norm": 0.14062173664569855, "learning_rate": 3.670324304303609e-05, "loss": 0.0121, "step": 181920 }, { "epoch": 115.95283620140216, "grad_norm": 0.28327304124832153, "learning_rate": 3.669793002417248e-05, "loss": 0.014, "step": 181930 }, { "epoch": 115.95920968769917, "grad_norm": 0.31458964943885803, "learning_rate": 3.669261716694156e-05, "loss": 0.011, "step": 181940 }, { "epoch": 115.96558317399618, "grad_norm": 0.24992597103118896, "learning_rate": 3.6687304471407914e-05, "loss": 0.018, "step": 181950 }, { "epoch": 115.97195666029317, "grad_norm": 0.20323170721530914, "learning_rate": 3.6681991937636105e-05, "loss": 0.014, "step": 181960 }, { "epoch": 115.97833014659018, "grad_norm": 0.20548538863658905, "learning_rate": 3.6676679565690645e-05, "loss": 0.008, "step": 181970 }, { "epoch": 115.98470363288719, "grad_norm": 0.44621288776397705, "learning_rate": 3.667136735563611e-05, "loss": 0.0083, "step": 181980 }, { "epoch": 115.9910771191842, "grad_norm": 0.3030773997306824, "learning_rate": 3.666605530753705e-05, "loss": 0.0106, "step": 181990 }, { "epoch": 115.99745060548119, "grad_norm": 0.1387702226638794, "learning_rate": 3.6660743421458025e-05, "loss": 0.0115, "step": 182000 }, { "epoch": 116.0038240917782, "grad_norm": 0.4176076352596283, "learning_rate": 3.665543169746354e-05, "loss": 0.0102, "step": 182010 }, { "epoch": 116.0101975780752, "grad_norm": 0.5052423477172852, "learning_rate": 3.665012013561816e-05, "loss": 0.0178, "step": 182020 }, { "epoch": 116.01657106437222, "grad_norm": 0.2153177112340927, "learning_rate": 3.6644808735986434e-05, "loss": 0.0075, "step": 182030 }, { "epoch": 116.02294455066922, "grad_norm": 0.26722297072410583, "learning_rate": 3.66394974986329e-05, "loss": 0.0067, "step": 182040 }, { "epoch": 116.02931803696622, "grad_norm": 0.16796183586120605, "learning_rate": 3.6634186423622065e-05, "loss": 0.0186, "step": 182050 }, { "epoch": 116.03569152326322, "grad_norm": 0.39388349652290344, "learning_rate": 3.662887551101849e-05, "loss": 0.0106, "step": 182060 }, { "epoch": 116.04206500956023, "grad_norm": 0.23234467208385468, "learning_rate": 3.662356476088671e-05, "loss": 0.0136, "step": 182070 }, { "epoch": 116.04843849585724, "grad_norm": 0.6052452921867371, "learning_rate": 3.661825417329125e-05, "loss": 0.0195, "step": 182080 }, { "epoch": 116.05481198215423, "grad_norm": 0.2592047452926636, "learning_rate": 3.6612943748296614e-05, "loss": 0.0097, "step": 182090 }, { "epoch": 116.06118546845124, "grad_norm": 0.33900296688079834, "learning_rate": 3.6607633485967375e-05, "loss": 0.0124, "step": 182100 }, { "epoch": 116.06755895474825, "grad_norm": 0.23788926005363464, "learning_rate": 3.660232338636802e-05, "loss": 0.0105, "step": 182110 }, { "epoch": 116.07393244104526, "grad_norm": 0.23591916263103485, "learning_rate": 3.659701344956309e-05, "loss": 0.0088, "step": 182120 }, { "epoch": 116.08030592734225, "grad_norm": 0.38827142119407654, "learning_rate": 3.659170367561709e-05, "loss": 0.0105, "step": 182130 }, { "epoch": 116.08667941363926, "grad_norm": 0.4132991433143616, "learning_rate": 3.658639406459457e-05, "loss": 0.0088, "step": 182140 }, { "epoch": 116.09305289993627, "grad_norm": 0.22736185789108276, "learning_rate": 3.6581084616560016e-05, "loss": 0.0127, "step": 182150 }, { "epoch": 116.09942638623328, "grad_norm": 0.19817228615283966, "learning_rate": 3.6575775331577954e-05, "loss": 0.022, "step": 182160 }, { "epoch": 116.10579987253027, "grad_norm": 0.19522061944007874, "learning_rate": 3.657046620971291e-05, "loss": 0.007, "step": 182170 }, { "epoch": 116.11217335882728, "grad_norm": 0.3856891095638275, "learning_rate": 3.656515725102936e-05, "loss": 0.0083, "step": 182180 }, { "epoch": 116.11854684512429, "grad_norm": 0.12309333682060242, "learning_rate": 3.6559848455591836e-05, "loss": 0.0062, "step": 182190 }, { "epoch": 116.1249203314213, "grad_norm": 0.8189927935600281, "learning_rate": 3.655453982346485e-05, "loss": 0.0254, "step": 182200 }, { "epoch": 116.13129381771829, "grad_norm": 0.3614010512828827, "learning_rate": 3.654923135471291e-05, "loss": 0.0224, "step": 182210 }, { "epoch": 116.1376673040153, "grad_norm": 0.18367530405521393, "learning_rate": 3.654392304940048e-05, "loss": 0.0064, "step": 182220 }, { "epoch": 116.1440407903123, "grad_norm": 0.11403508484363556, "learning_rate": 3.653861490759211e-05, "loss": 0.0135, "step": 182230 }, { "epoch": 116.15041427660931, "grad_norm": 0.30252736806869507, "learning_rate": 3.653330692935229e-05, "loss": 0.0105, "step": 182240 }, { "epoch": 116.1567877629063, "grad_norm": 0.18244382739067078, "learning_rate": 3.6527999114745484e-05, "loss": 0.0085, "step": 182250 }, { "epoch": 116.16316124920331, "grad_norm": 0.1885964423418045, "learning_rate": 3.6522691463836215e-05, "loss": 0.0105, "step": 182260 }, { "epoch": 116.16953473550032, "grad_norm": 0.18998399376869202, "learning_rate": 3.651738397668896e-05, "loss": 0.0057, "step": 182270 }, { "epoch": 116.17590822179733, "grad_norm": 0.36828845739364624, "learning_rate": 3.651207665336825e-05, "loss": 0.012, "step": 182280 }, { "epoch": 116.18228170809432, "grad_norm": 0.28418052196502686, "learning_rate": 3.650676949393851e-05, "loss": 0.007, "step": 182290 }, { "epoch": 116.18865519439133, "grad_norm": 0.24278542399406433, "learning_rate": 3.650146249846427e-05, "loss": 0.0104, "step": 182300 }, { "epoch": 116.19502868068834, "grad_norm": 0.531290590763092, "learning_rate": 3.649615566701001e-05, "loss": 0.0119, "step": 182310 }, { "epoch": 116.20140216698535, "grad_norm": 0.19582025706768036, "learning_rate": 3.6490848999640215e-05, "loss": 0.01, "step": 182320 }, { "epoch": 116.20777565328234, "grad_norm": 0.16500650346279144, "learning_rate": 3.648554249641934e-05, "loss": 0.0082, "step": 182330 }, { "epoch": 116.21414913957935, "grad_norm": 0.27867811918258667, "learning_rate": 3.6480236157411893e-05, "loss": 0.0135, "step": 182340 }, { "epoch": 116.22052262587636, "grad_norm": 0.26813867688179016, "learning_rate": 3.647492998268236e-05, "loss": 0.0106, "step": 182350 }, { "epoch": 116.22689611217336, "grad_norm": 0.3234195113182068, "learning_rate": 3.646962397229518e-05, "loss": 0.0111, "step": 182360 }, { "epoch": 116.23326959847036, "grad_norm": 0.5656688213348389, "learning_rate": 3.646431812631484e-05, "loss": 0.0105, "step": 182370 }, { "epoch": 116.23964308476737, "grad_norm": 0.2282572090625763, "learning_rate": 3.6459012444805804e-05, "loss": 0.0125, "step": 182380 }, { "epoch": 116.24601657106437, "grad_norm": 0.36844053864479065, "learning_rate": 3.645370692783259e-05, "loss": 0.0091, "step": 182390 }, { "epoch": 116.25239005736138, "grad_norm": 0.28880324959754944, "learning_rate": 3.644840157545959e-05, "loss": 0.0075, "step": 182400 }, { "epoch": 116.25876354365838, "grad_norm": 0.31185275316238403, "learning_rate": 3.644309638775132e-05, "loss": 0.0122, "step": 182410 }, { "epoch": 116.26513702995538, "grad_norm": 0.4512365460395813, "learning_rate": 3.6437791364772246e-05, "loss": 0.0083, "step": 182420 }, { "epoch": 116.27151051625239, "grad_norm": 0.2698647081851959, "learning_rate": 3.6432486506586786e-05, "loss": 0.0097, "step": 182430 }, { "epoch": 116.2778840025494, "grad_norm": 0.1661635935306549, "learning_rate": 3.642718181325942e-05, "loss": 0.0048, "step": 182440 }, { "epoch": 116.2842574888464, "grad_norm": 0.25335392355918884, "learning_rate": 3.642187728485462e-05, "loss": 0.0136, "step": 182450 }, { "epoch": 116.2906309751434, "grad_norm": 0.18582351505756378, "learning_rate": 3.6416572921436846e-05, "loss": 0.0095, "step": 182460 }, { "epoch": 116.29700446144041, "grad_norm": 0.15394610166549683, "learning_rate": 3.6411268723070516e-05, "loss": 0.0111, "step": 182470 }, { "epoch": 116.30337794773742, "grad_norm": 0.21234002709388733, "learning_rate": 3.640596468982011e-05, "loss": 0.014, "step": 182480 }, { "epoch": 116.30975143403441, "grad_norm": 0.22314737737178802, "learning_rate": 3.6400660821750073e-05, "loss": 0.0075, "step": 182490 }, { "epoch": 116.31612492033142, "grad_norm": 0.281870573759079, "learning_rate": 3.639535711892483e-05, "loss": 0.016, "step": 182500 }, { "epoch": 116.32249840662843, "grad_norm": 0.09356390684843063, "learning_rate": 3.639005358140886e-05, "loss": 0.0062, "step": 182510 }, { "epoch": 116.32887189292543, "grad_norm": 0.30023953318595886, "learning_rate": 3.638475020926657e-05, "loss": 0.0117, "step": 182520 }, { "epoch": 116.33524537922243, "grad_norm": 0.29158082604408264, "learning_rate": 3.6379447002562445e-05, "loss": 0.0171, "step": 182530 }, { "epoch": 116.34161886551944, "grad_norm": 0.23474740982055664, "learning_rate": 3.637414396136088e-05, "loss": 0.0061, "step": 182540 }, { "epoch": 116.34799235181644, "grad_norm": 0.45756301283836365, "learning_rate": 3.636884108572633e-05, "loss": 0.0093, "step": 182550 }, { "epoch": 116.35436583811345, "grad_norm": 0.28556856513023376, "learning_rate": 3.636353837572325e-05, "loss": 0.0085, "step": 182560 }, { "epoch": 116.36073932441045, "grad_norm": 0.23882292211055756, "learning_rate": 3.6358235831416034e-05, "loss": 0.0097, "step": 182570 }, { "epoch": 116.36711281070745, "grad_norm": 0.23092330992221832, "learning_rate": 3.635293345286913e-05, "loss": 0.0068, "step": 182580 }, { "epoch": 116.37348629700446, "grad_norm": 0.2606753706932068, "learning_rate": 3.634763124014698e-05, "loss": 0.0094, "step": 182590 }, { "epoch": 116.37985978330147, "grad_norm": 0.18915528059005737, "learning_rate": 3.634232919331401e-05, "loss": 0.0088, "step": 182600 }, { "epoch": 116.38623326959846, "grad_norm": 0.24986712634563446, "learning_rate": 3.633702731243462e-05, "loss": 0.0058, "step": 182610 }, { "epoch": 116.39260675589547, "grad_norm": 0.14513686299324036, "learning_rate": 3.6331725597573245e-05, "loss": 0.007, "step": 182620 }, { "epoch": 116.39898024219248, "grad_norm": 0.1558392196893692, "learning_rate": 3.632642404879431e-05, "loss": 0.0064, "step": 182630 }, { "epoch": 116.40535372848949, "grad_norm": 0.18674138188362122, "learning_rate": 3.6321122666162255e-05, "loss": 0.0089, "step": 182640 }, { "epoch": 116.4117272147865, "grad_norm": 0.35371100902557373, "learning_rate": 3.6315821449741454e-05, "loss": 0.0051, "step": 182650 }, { "epoch": 116.41810070108349, "grad_norm": 0.21712808310985565, "learning_rate": 3.631052039959634e-05, "loss": 0.0139, "step": 182660 }, { "epoch": 116.4244741873805, "grad_norm": 0.29072728753089905, "learning_rate": 3.630521951579135e-05, "loss": 0.0064, "step": 182670 }, { "epoch": 116.4308476736775, "grad_norm": 0.25794005393981934, "learning_rate": 3.629991879839085e-05, "loss": 0.0084, "step": 182680 }, { "epoch": 116.43722115997451, "grad_norm": 0.1468011736869812, "learning_rate": 3.629461824745928e-05, "loss": 0.0064, "step": 182690 }, { "epoch": 116.4435946462715, "grad_norm": 0.16447393596172333, "learning_rate": 3.628931786306103e-05, "loss": 0.0091, "step": 182700 }, { "epoch": 116.44996813256851, "grad_norm": 0.4099271595478058, "learning_rate": 3.628401764526054e-05, "loss": 0.009, "step": 182710 }, { "epoch": 116.45634161886552, "grad_norm": 0.1598593294620514, "learning_rate": 3.627871759412216e-05, "loss": 0.0071, "step": 182720 }, { "epoch": 116.46271510516253, "grad_norm": 0.2015240341424942, "learning_rate": 3.627341770971031e-05, "loss": 0.013, "step": 182730 }, { "epoch": 116.46908859145952, "grad_norm": 0.33778971433639526, "learning_rate": 3.626811799208942e-05, "loss": 0.0091, "step": 182740 }, { "epoch": 116.47546207775653, "grad_norm": 0.3167140483856201, "learning_rate": 3.6262818441323854e-05, "loss": 0.0064, "step": 182750 }, { "epoch": 116.48183556405354, "grad_norm": 0.34092453122138977, "learning_rate": 3.6257519057478004e-05, "loss": 0.0121, "step": 182760 }, { "epoch": 116.48820905035055, "grad_norm": 0.20340125262737274, "learning_rate": 3.625221984061627e-05, "loss": 0.0109, "step": 182770 }, { "epoch": 116.49458253664754, "grad_norm": 0.2016984224319458, "learning_rate": 3.624692079080306e-05, "loss": 0.0081, "step": 182780 }, { "epoch": 116.50095602294455, "grad_norm": 0.26342034339904785, "learning_rate": 3.624162190810274e-05, "loss": 0.0096, "step": 182790 }, { "epoch": 116.50732950924156, "grad_norm": 0.2300245761871338, "learning_rate": 3.6236323192579704e-05, "loss": 0.0132, "step": 182800 }, { "epoch": 116.51370299553857, "grad_norm": 0.22942237555980682, "learning_rate": 3.623102464429834e-05, "loss": 0.011, "step": 182810 }, { "epoch": 116.52007648183556, "grad_norm": 0.1986566185951233, "learning_rate": 3.622572626332301e-05, "loss": 0.0059, "step": 182820 }, { "epoch": 116.52644996813257, "grad_norm": 0.19300344586372375, "learning_rate": 3.622042804971814e-05, "loss": 0.0134, "step": 182830 }, { "epoch": 116.53282345442958, "grad_norm": 0.2644278109073639, "learning_rate": 3.6215130003548045e-05, "loss": 0.009, "step": 182840 }, { "epoch": 116.53919694072658, "grad_norm": 0.3406103849411011, "learning_rate": 3.620983212487716e-05, "loss": 0.0101, "step": 182850 }, { "epoch": 116.54557042702358, "grad_norm": 0.211698517203331, "learning_rate": 3.620453441376982e-05, "loss": 0.0092, "step": 182860 }, { "epoch": 116.55194391332059, "grad_norm": 0.27847009897232056, "learning_rate": 3.6199236870290435e-05, "loss": 0.0088, "step": 182870 }, { "epoch": 116.5583173996176, "grad_norm": 0.18222706019878387, "learning_rate": 3.6193939494503336e-05, "loss": 0.0156, "step": 182880 }, { "epoch": 116.5646908859146, "grad_norm": 0.17528828978538513, "learning_rate": 3.618864228647291e-05, "loss": 0.0083, "step": 182890 }, { "epoch": 116.5710643722116, "grad_norm": 0.09210845828056335, "learning_rate": 3.618334524626351e-05, "loss": 0.0109, "step": 182900 }, { "epoch": 116.5774378585086, "grad_norm": 0.22470079362392426, "learning_rate": 3.617804837393952e-05, "loss": 0.0063, "step": 182910 }, { "epoch": 116.58381134480561, "grad_norm": 0.22986513376235962, "learning_rate": 3.6172751669565294e-05, "loss": 0.0083, "step": 182920 }, { "epoch": 116.59018483110262, "grad_norm": 0.2141786813735962, "learning_rate": 3.6167455133205186e-05, "loss": 0.0096, "step": 182930 }, { "epoch": 116.59655831739961, "grad_norm": 0.13977062702178955, "learning_rate": 3.616215876492356e-05, "loss": 0.0073, "step": 182940 }, { "epoch": 116.60293180369662, "grad_norm": 0.432239830493927, "learning_rate": 3.615686256478477e-05, "loss": 0.0193, "step": 182950 }, { "epoch": 116.60930528999363, "grad_norm": 0.21225184202194214, "learning_rate": 3.6151566532853175e-05, "loss": 0.0073, "step": 182960 }, { "epoch": 116.61567877629064, "grad_norm": 0.43196573853492737, "learning_rate": 3.6146270669193116e-05, "loss": 0.0131, "step": 182970 }, { "epoch": 116.62205226258763, "grad_norm": 0.19476017355918884, "learning_rate": 3.614097497386894e-05, "loss": 0.0096, "step": 182980 }, { "epoch": 116.62842574888464, "grad_norm": 0.45479974150657654, "learning_rate": 3.613567944694503e-05, "loss": 0.011, "step": 182990 }, { "epoch": 116.63479923518165, "grad_norm": 0.19712135195732117, "learning_rate": 3.613038408848568e-05, "loss": 0.0161, "step": 183000 }, { "epoch": 116.64117272147865, "grad_norm": 0.36786895990371704, "learning_rate": 3.612508889855526e-05, "loss": 0.0104, "step": 183010 }, { "epoch": 116.64754620777565, "grad_norm": 0.25546994805336, "learning_rate": 3.611979387721811e-05, "loss": 0.0165, "step": 183020 }, { "epoch": 116.65391969407266, "grad_norm": 0.29083535075187683, "learning_rate": 3.611449902453859e-05, "loss": 0.0112, "step": 183030 }, { "epoch": 116.66029318036966, "grad_norm": 0.19710178673267365, "learning_rate": 3.610920434058099e-05, "loss": 0.0074, "step": 183040 }, { "epoch": 116.66666666666667, "grad_norm": 0.2545722424983978, "learning_rate": 3.610390982540969e-05, "loss": 0.0135, "step": 183050 }, { "epoch": 116.67304015296367, "grad_norm": 0.1485583633184433, "learning_rate": 3.609861547908901e-05, "loss": 0.0117, "step": 183060 }, { "epoch": 116.67941363926067, "grad_norm": 0.1966640204191208, "learning_rate": 3.6093321301683266e-05, "loss": 0.007, "step": 183070 }, { "epoch": 116.68578712555768, "grad_norm": 0.212001234292984, "learning_rate": 3.608802729325679e-05, "loss": 0.0131, "step": 183080 }, { "epoch": 116.69216061185469, "grad_norm": 0.28404489159584045, "learning_rate": 3.608273345387392e-05, "loss": 0.0063, "step": 183090 }, { "epoch": 116.69853409815168, "grad_norm": 0.22347094118595123, "learning_rate": 3.6077439783599003e-05, "loss": 0.0132, "step": 183100 }, { "epoch": 116.70490758444869, "grad_norm": 0.19914786517620087, "learning_rate": 3.607214628249631e-05, "loss": 0.0093, "step": 183110 }, { "epoch": 116.7112810707457, "grad_norm": 0.2869023382663727, "learning_rate": 3.60668529506302e-05, "loss": 0.0117, "step": 183120 }, { "epoch": 116.7176545570427, "grad_norm": 0.42599910497665405, "learning_rate": 3.6061559788064995e-05, "loss": 0.0125, "step": 183130 }, { "epoch": 116.7240280433397, "grad_norm": 0.5865877270698547, "learning_rate": 3.605626679486498e-05, "loss": 0.0065, "step": 183140 }, { "epoch": 116.73040152963671, "grad_norm": 0.4254571795463562, "learning_rate": 3.605097397109449e-05, "loss": 0.0081, "step": 183150 }, { "epoch": 116.73677501593372, "grad_norm": 0.4161181151866913, "learning_rate": 3.6045681316817835e-05, "loss": 0.0086, "step": 183160 }, { "epoch": 116.74314850223072, "grad_norm": 0.23528461158275604, "learning_rate": 3.6040388832099345e-05, "loss": 0.0159, "step": 183170 }, { "epoch": 116.74952198852772, "grad_norm": 0.22027219831943512, "learning_rate": 3.6035096517003296e-05, "loss": 0.0113, "step": 183180 }, { "epoch": 116.75589547482473, "grad_norm": 0.4948532283306122, "learning_rate": 3.6029804371594e-05, "loss": 0.0137, "step": 183190 }, { "epoch": 116.76226896112173, "grad_norm": 0.440176397562027, "learning_rate": 3.6024512395935787e-05, "loss": 0.0082, "step": 183200 }, { "epoch": 116.76864244741874, "grad_norm": 0.26595380902290344, "learning_rate": 3.6019220590092954e-05, "loss": 0.0207, "step": 183210 }, { "epoch": 116.77501593371574, "grad_norm": 0.32157281041145325, "learning_rate": 3.601392895412976e-05, "loss": 0.011, "step": 183220 }, { "epoch": 116.78138942001274, "grad_norm": 0.4798450469970703, "learning_rate": 3.600863748811055e-05, "loss": 0.009, "step": 183230 }, { "epoch": 116.78776290630975, "grad_norm": 0.2754303514957428, "learning_rate": 3.600334619209962e-05, "loss": 0.0069, "step": 183240 }, { "epoch": 116.79413639260676, "grad_norm": 0.22297583520412445, "learning_rate": 3.599805506616122e-05, "loss": 0.007, "step": 183250 }, { "epoch": 116.80050987890377, "grad_norm": 0.3724551200866699, "learning_rate": 3.5992764110359686e-05, "loss": 0.0129, "step": 183260 }, { "epoch": 116.80688336520076, "grad_norm": 0.22629554569721222, "learning_rate": 3.598747332475928e-05, "loss": 0.0108, "step": 183270 }, { "epoch": 116.81325685149777, "grad_norm": 0.2372407764196396, "learning_rate": 3.598218270942433e-05, "loss": 0.0138, "step": 183280 }, { "epoch": 116.81963033779478, "grad_norm": 0.15179343521595, "learning_rate": 3.597689226441908e-05, "loss": 0.0073, "step": 183290 }, { "epoch": 116.82600382409179, "grad_norm": 0.18295016884803772, "learning_rate": 3.597160198980782e-05, "loss": 0.0144, "step": 183300 }, { "epoch": 116.83237731038878, "grad_norm": 0.28695178031921387, "learning_rate": 3.596631188565486e-05, "loss": 0.0093, "step": 183310 }, { "epoch": 116.83875079668579, "grad_norm": 0.3124050498008728, "learning_rate": 3.596102195202445e-05, "loss": 0.0072, "step": 183320 }, { "epoch": 116.8451242829828, "grad_norm": 0.24255068600177765, "learning_rate": 3.595573218898087e-05, "loss": 0.0065, "step": 183330 }, { "epoch": 116.8514977692798, "grad_norm": 0.20546770095825195, "learning_rate": 3.595044259658841e-05, "loss": 0.0115, "step": 183340 }, { "epoch": 116.8578712555768, "grad_norm": 0.23578770458698273, "learning_rate": 3.594515317491135e-05, "loss": 0.0078, "step": 183350 }, { "epoch": 116.8642447418738, "grad_norm": 0.22378695011138916, "learning_rate": 3.5939863924013934e-05, "loss": 0.0071, "step": 183360 }, { "epoch": 116.87061822817081, "grad_norm": 0.14181478321552277, "learning_rate": 3.593457484396044e-05, "loss": 0.0217, "step": 183370 }, { "epoch": 116.87699171446782, "grad_norm": 0.32342666387557983, "learning_rate": 3.5929285934815174e-05, "loss": 0.0125, "step": 183380 }, { "epoch": 116.88336520076481, "grad_norm": 0.3616009056568146, "learning_rate": 3.592399719664234e-05, "loss": 0.0138, "step": 183390 }, { "epoch": 116.88973868706182, "grad_norm": 0.21283037960529327, "learning_rate": 3.5918708629506234e-05, "loss": 0.0146, "step": 183400 }, { "epoch": 116.89611217335883, "grad_norm": 0.4254257082939148, "learning_rate": 3.591342023347111e-05, "loss": 0.0102, "step": 183410 }, { "epoch": 116.90248565965584, "grad_norm": 0.22595837712287903, "learning_rate": 3.590813200860125e-05, "loss": 0.0053, "step": 183420 }, { "epoch": 116.90885914595283, "grad_norm": 0.2719130516052246, "learning_rate": 3.590284395496087e-05, "loss": 0.0072, "step": 183430 }, { "epoch": 116.91523263224984, "grad_norm": 0.3247969448566437, "learning_rate": 3.5897556072614244e-05, "loss": 0.0193, "step": 183440 }, { "epoch": 116.92160611854685, "grad_norm": 0.17105084657669067, "learning_rate": 3.589226836162565e-05, "loss": 0.0086, "step": 183450 }, { "epoch": 116.92797960484386, "grad_norm": 0.25759652256965637, "learning_rate": 3.58869808220593e-05, "loss": 0.0172, "step": 183460 }, { "epoch": 116.93435309114085, "grad_norm": 0.20567137002944946, "learning_rate": 3.588169345397945e-05, "loss": 0.0197, "step": 183470 }, { "epoch": 116.94072657743786, "grad_norm": 0.3531315326690674, "learning_rate": 3.587640625745036e-05, "loss": 0.0123, "step": 183480 }, { "epoch": 116.94710006373487, "grad_norm": 0.14483243227005005, "learning_rate": 3.5871119232536275e-05, "loss": 0.0075, "step": 183490 }, { "epoch": 116.95347355003187, "grad_norm": 0.7979691028594971, "learning_rate": 3.586583237930144e-05, "loss": 0.0243, "step": 183500 }, { "epoch": 116.95984703632887, "grad_norm": 0.5077809691429138, "learning_rate": 3.5860545697810065e-05, "loss": 0.0132, "step": 183510 }, { "epoch": 116.96622052262587, "grad_norm": 0.5172356963157654, "learning_rate": 3.5855259188126414e-05, "loss": 0.0142, "step": 183520 }, { "epoch": 116.97259400892288, "grad_norm": 0.28277266025543213, "learning_rate": 3.5849972850314726e-05, "loss": 0.0097, "step": 183530 }, { "epoch": 116.97896749521989, "grad_norm": 0.35381314158439636, "learning_rate": 3.5844686684439234e-05, "loss": 0.0069, "step": 183540 }, { "epoch": 116.98534098151688, "grad_norm": 0.20428846776485443, "learning_rate": 3.583940069056415e-05, "loss": 0.0075, "step": 183550 }, { "epoch": 116.99171446781389, "grad_norm": 0.14816954731941223, "learning_rate": 3.583411486875373e-05, "loss": 0.0139, "step": 183560 }, { "epoch": 116.9980879541109, "grad_norm": 0.267801970243454, "learning_rate": 3.582882921907218e-05, "loss": 0.011, "step": 183570 }, { "epoch": 117.00446144040791, "grad_norm": 0.2137143909931183, "learning_rate": 3.582354374158374e-05, "loss": 0.0072, "step": 183580 }, { "epoch": 117.0108349267049, "grad_norm": 0.3088333308696747, "learning_rate": 3.581825843635262e-05, "loss": 0.0101, "step": 183590 }, { "epoch": 117.01720841300191, "grad_norm": 0.2980770766735077, "learning_rate": 3.581297330344306e-05, "loss": 0.0061, "step": 183600 }, { "epoch": 117.02358189929892, "grad_norm": 0.26755765080451965, "learning_rate": 3.5807688342919256e-05, "loss": 0.0068, "step": 183610 }, { "epoch": 117.02995538559593, "grad_norm": 0.40261298418045044, "learning_rate": 3.580240355484545e-05, "loss": 0.0082, "step": 183620 }, { "epoch": 117.03632887189292, "grad_norm": 0.14348362386226654, "learning_rate": 3.579711893928585e-05, "loss": 0.0083, "step": 183630 }, { "epoch": 117.04270235818993, "grad_norm": 0.19899854063987732, "learning_rate": 3.5791834496304656e-05, "loss": 0.012, "step": 183640 }, { "epoch": 117.04907584448694, "grad_norm": 0.18727731704711914, "learning_rate": 3.578655022596607e-05, "loss": 0.0099, "step": 183650 }, { "epoch": 117.05544933078394, "grad_norm": 0.4348612427711487, "learning_rate": 3.578126612833435e-05, "loss": 0.0189, "step": 183660 }, { "epoch": 117.06182281708094, "grad_norm": 0.09908061474561691, "learning_rate": 3.577598220347366e-05, "loss": 0.0088, "step": 183670 }, { "epoch": 117.06819630337795, "grad_norm": 0.13749700784683228, "learning_rate": 3.577069845144821e-05, "loss": 0.0077, "step": 183680 }, { "epoch": 117.07456978967495, "grad_norm": 0.19148074090480804, "learning_rate": 3.5765414872322213e-05, "loss": 0.0065, "step": 183690 }, { "epoch": 117.08094327597196, "grad_norm": 0.30518242716789246, "learning_rate": 3.5760131466159875e-05, "loss": 0.0104, "step": 183700 }, { "epoch": 117.08731676226896, "grad_norm": 0.25926119089126587, "learning_rate": 3.575484823302537e-05, "loss": 0.0081, "step": 183710 }, { "epoch": 117.09369024856596, "grad_norm": 0.2605375647544861, "learning_rate": 3.57495651729829e-05, "loss": 0.0074, "step": 183720 }, { "epoch": 117.10006373486297, "grad_norm": 0.11709193885326385, "learning_rate": 3.574428228609668e-05, "loss": 0.0121, "step": 183730 }, { "epoch": 117.10643722115998, "grad_norm": 0.23402097821235657, "learning_rate": 3.573899957243091e-05, "loss": 0.0078, "step": 183740 }, { "epoch": 117.11281070745697, "grad_norm": 0.10447090864181519, "learning_rate": 3.573371703204973e-05, "loss": 0.007, "step": 183750 }, { "epoch": 117.11918419375398, "grad_norm": 0.3060356676578522, "learning_rate": 3.5728434665017366e-05, "loss": 0.0121, "step": 183760 }, { "epoch": 117.12555768005099, "grad_norm": 0.27230924367904663, "learning_rate": 3.5723152471398005e-05, "loss": 0.011, "step": 183770 }, { "epoch": 117.131931166348, "grad_norm": 0.2558720111846924, "learning_rate": 3.571787045125581e-05, "loss": 0.0135, "step": 183780 }, { "epoch": 117.13830465264499, "grad_norm": 0.22620974481105804, "learning_rate": 3.571258860465497e-05, "loss": 0.0044, "step": 183790 }, { "epoch": 117.144678138942, "grad_norm": 0.264216810464859, "learning_rate": 3.570730693165967e-05, "loss": 0.0142, "step": 183800 }, { "epoch": 117.151051625239, "grad_norm": 0.3716268539428711, "learning_rate": 3.570202543233411e-05, "loss": 0.0114, "step": 183810 }, { "epoch": 117.15742511153601, "grad_norm": 0.14392738044261932, "learning_rate": 3.569674410674241e-05, "loss": 0.0138, "step": 183820 }, { "epoch": 117.16379859783301, "grad_norm": 0.40930166840553284, "learning_rate": 3.569146295494878e-05, "loss": 0.0078, "step": 183830 }, { "epoch": 117.17017208413002, "grad_norm": 0.44085967540740967, "learning_rate": 3.568618197701739e-05, "loss": 0.0094, "step": 183840 }, { "epoch": 117.17654557042702, "grad_norm": 0.2533520460128784, "learning_rate": 3.568090117301241e-05, "loss": 0.0187, "step": 183850 }, { "epoch": 117.18291905672403, "grad_norm": 0.23109546303749084, "learning_rate": 3.567562054299799e-05, "loss": 0.0126, "step": 183860 }, { "epoch": 117.18929254302103, "grad_norm": 0.24303504824638367, "learning_rate": 3.56703400870383e-05, "loss": 0.0087, "step": 183870 }, { "epoch": 117.19566602931803, "grad_norm": 0.14673489332199097, "learning_rate": 3.566505980519753e-05, "loss": 0.0059, "step": 183880 }, { "epoch": 117.20203951561504, "grad_norm": 0.1498374342918396, "learning_rate": 3.5659779697539794e-05, "loss": 0.0172, "step": 183890 }, { "epoch": 117.20841300191205, "grad_norm": 0.38739436864852905, "learning_rate": 3.565449976412928e-05, "loss": 0.0125, "step": 183900 }, { "epoch": 117.21478648820906, "grad_norm": 0.22682788968086243, "learning_rate": 3.5649220005030134e-05, "loss": 0.0108, "step": 183910 }, { "epoch": 117.22115997450605, "grad_norm": 0.2048673927783966, "learning_rate": 3.564394042030654e-05, "loss": 0.0131, "step": 183920 }, { "epoch": 117.22753346080306, "grad_norm": 0.32880741357803345, "learning_rate": 3.5638661010022604e-05, "loss": 0.0071, "step": 183930 }, { "epoch": 117.23390694710007, "grad_norm": 0.23784612119197845, "learning_rate": 3.5633381774242485e-05, "loss": 0.0066, "step": 183940 }, { "epoch": 117.24028043339707, "grad_norm": 0.26258978247642517, "learning_rate": 3.562810271303037e-05, "loss": 0.0156, "step": 183950 }, { "epoch": 117.24665391969407, "grad_norm": 0.21251565217971802, "learning_rate": 3.5622823826450355e-05, "loss": 0.0139, "step": 183960 }, { "epoch": 117.25302740599108, "grad_norm": 0.11920630931854248, "learning_rate": 3.5617545114566606e-05, "loss": 0.0113, "step": 183970 }, { "epoch": 117.25940089228808, "grad_norm": 0.38329586386680603, "learning_rate": 3.5612266577443256e-05, "loss": 0.0256, "step": 183980 }, { "epoch": 117.26577437858509, "grad_norm": 0.1369016170501709, "learning_rate": 3.5606988215144485e-05, "loss": 0.0079, "step": 183990 }, { "epoch": 117.27214786488209, "grad_norm": 0.3981565237045288, "learning_rate": 3.560171002773437e-05, "loss": 0.0081, "step": 184000 }, { "epoch": 117.2785213511791, "grad_norm": 0.3336315453052521, "learning_rate": 3.559643201527707e-05, "loss": 0.0096, "step": 184010 }, { "epoch": 117.2848948374761, "grad_norm": 0.35822951793670654, "learning_rate": 3.5591154177836736e-05, "loss": 0.0081, "step": 184020 }, { "epoch": 117.29126832377311, "grad_norm": 0.1904619187116623, "learning_rate": 3.5585876515477474e-05, "loss": 0.0077, "step": 184030 }, { "epoch": 117.2976418100701, "grad_norm": 0.3020016551017761, "learning_rate": 3.5580599028263416e-05, "loss": 0.0119, "step": 184040 }, { "epoch": 117.30401529636711, "grad_norm": 0.16556720435619354, "learning_rate": 3.557532171625869e-05, "loss": 0.0166, "step": 184050 }, { "epoch": 117.31038878266412, "grad_norm": 0.13608086109161377, "learning_rate": 3.557004457952745e-05, "loss": 0.0074, "step": 184060 }, { "epoch": 117.31676226896113, "grad_norm": 0.22783540189266205, "learning_rate": 3.556476761813377e-05, "loss": 0.0093, "step": 184070 }, { "epoch": 117.32313575525812, "grad_norm": 0.19064447283744812, "learning_rate": 3.555949083214179e-05, "loss": 0.0076, "step": 184080 }, { "epoch": 117.32950924155513, "grad_norm": 0.6048488616943359, "learning_rate": 3.555421422161566e-05, "loss": 0.0116, "step": 184090 }, { "epoch": 117.33588272785214, "grad_norm": 0.18983782827854156, "learning_rate": 3.554893778661944e-05, "loss": 0.0087, "step": 184100 }, { "epoch": 117.34225621414915, "grad_norm": 0.3322193920612335, "learning_rate": 3.554366152721727e-05, "loss": 0.0129, "step": 184110 }, { "epoch": 117.34862970044614, "grad_norm": 0.11977183073759079, "learning_rate": 3.553838544347326e-05, "loss": 0.0058, "step": 184120 }, { "epoch": 117.35500318674315, "grad_norm": 0.24557416141033173, "learning_rate": 3.553310953545155e-05, "loss": 0.0092, "step": 184130 }, { "epoch": 117.36137667304016, "grad_norm": 0.3907131850719452, "learning_rate": 3.552783380321618e-05, "loss": 0.0215, "step": 184140 }, { "epoch": 117.36775015933716, "grad_norm": 0.20278136432170868, "learning_rate": 3.5522558246831306e-05, "loss": 0.0077, "step": 184150 }, { "epoch": 117.37412364563416, "grad_norm": 0.2145073264837265, "learning_rate": 3.5517282866361016e-05, "loss": 0.0084, "step": 184160 }, { "epoch": 117.38049713193116, "grad_norm": 0.180498406291008, "learning_rate": 3.551200766186944e-05, "loss": 0.0118, "step": 184170 }, { "epoch": 117.38687061822817, "grad_norm": 0.260321706533432, "learning_rate": 3.550673263342062e-05, "loss": 0.0086, "step": 184180 }, { "epoch": 117.39324410452518, "grad_norm": 0.29199644923210144, "learning_rate": 3.550145778107869e-05, "loss": 0.0089, "step": 184190 }, { "epoch": 117.39961759082217, "grad_norm": 0.1471734642982483, "learning_rate": 3.5496183104907735e-05, "loss": 0.0104, "step": 184200 }, { "epoch": 117.40599107711918, "grad_norm": 0.28078213334083557, "learning_rate": 3.549090860497186e-05, "loss": 0.0092, "step": 184210 }, { "epoch": 117.41236456341619, "grad_norm": 0.25631871819496155, "learning_rate": 3.548563428133513e-05, "loss": 0.0091, "step": 184220 }, { "epoch": 117.4187380497132, "grad_norm": 0.18355116248130798, "learning_rate": 3.548036013406165e-05, "loss": 0.0139, "step": 184230 }, { "epoch": 117.42511153601019, "grad_norm": 0.2862338423728943, "learning_rate": 3.547508616321551e-05, "loss": 0.0086, "step": 184240 }, { "epoch": 117.4314850223072, "grad_norm": 0.3827865421772003, "learning_rate": 3.546981236886079e-05, "loss": 0.0138, "step": 184250 }, { "epoch": 117.43785850860421, "grad_norm": 0.1395380049943924, "learning_rate": 3.5464538751061556e-05, "loss": 0.0181, "step": 184260 }, { "epoch": 117.44423199490122, "grad_norm": 0.1469859927892685, "learning_rate": 3.545926530988191e-05, "loss": 0.0086, "step": 184270 }, { "epoch": 117.45060548119821, "grad_norm": 0.2128150314092636, "learning_rate": 3.545399204538591e-05, "loss": 0.0062, "step": 184280 }, { "epoch": 117.45697896749522, "grad_norm": 0.17391391098499298, "learning_rate": 3.544871895763765e-05, "loss": 0.0059, "step": 184290 }, { "epoch": 117.46335245379223, "grad_norm": 0.23513998091220856, "learning_rate": 3.544344604670118e-05, "loss": 0.0099, "step": 184300 }, { "epoch": 117.46972594008923, "grad_norm": 0.1345917135477066, "learning_rate": 3.54381733126406e-05, "loss": 0.0165, "step": 184310 }, { "epoch": 117.47609942638623, "grad_norm": 0.22889557480812073, "learning_rate": 3.5432900755519946e-05, "loss": 0.0115, "step": 184320 }, { "epoch": 117.48247291268324, "grad_norm": 0.1556912511587143, "learning_rate": 3.5427628375403324e-05, "loss": 0.0096, "step": 184330 }, { "epoch": 117.48884639898024, "grad_norm": 0.2046397179365158, "learning_rate": 3.542235617235477e-05, "loss": 0.0163, "step": 184340 }, { "epoch": 117.49521988527725, "grad_norm": 0.22679877281188965, "learning_rate": 3.541708414643834e-05, "loss": 0.0085, "step": 184350 }, { "epoch": 117.50159337157424, "grad_norm": 0.28306305408477783, "learning_rate": 3.5411812297718106e-05, "loss": 0.0141, "step": 184360 }, { "epoch": 117.50796685787125, "grad_norm": 0.15279459953308105, "learning_rate": 3.5406540626258136e-05, "loss": 0.0077, "step": 184370 }, { "epoch": 117.51434034416826, "grad_norm": 0.4707605540752411, "learning_rate": 3.5401269132122474e-05, "loss": 0.0123, "step": 184380 }, { "epoch": 117.52071383046527, "grad_norm": 0.41590416431427, "learning_rate": 3.539599781537517e-05, "loss": 0.0157, "step": 184390 }, { "epoch": 117.52708731676226, "grad_norm": 0.20515629649162292, "learning_rate": 3.5390726676080276e-05, "loss": 0.0102, "step": 184400 }, { "epoch": 117.53346080305927, "grad_norm": 0.5803506374359131, "learning_rate": 3.538545571430186e-05, "loss": 0.0162, "step": 184410 }, { "epoch": 117.53983428935628, "grad_norm": 0.5085460543632507, "learning_rate": 3.5380184930103955e-05, "loss": 0.0109, "step": 184420 }, { "epoch": 117.54620777565329, "grad_norm": 0.3118307590484619, "learning_rate": 3.53749143235506e-05, "loss": 0.0056, "step": 184430 }, { "epoch": 117.55258126195028, "grad_norm": 0.44879382848739624, "learning_rate": 3.536964389470584e-05, "loss": 0.0179, "step": 184440 }, { "epoch": 117.55895474824729, "grad_norm": 0.19678275287151337, "learning_rate": 3.536437364363374e-05, "loss": 0.0123, "step": 184450 }, { "epoch": 117.5653282345443, "grad_norm": 0.19354206323623657, "learning_rate": 3.53591035703983e-05, "loss": 0.0086, "step": 184460 }, { "epoch": 117.5717017208413, "grad_norm": 0.09767621010541916, "learning_rate": 3.535383367506357e-05, "loss": 0.01, "step": 184470 }, { "epoch": 117.5780752071383, "grad_norm": 0.6197047233581543, "learning_rate": 3.5348563957693595e-05, "loss": 0.0131, "step": 184480 }, { "epoch": 117.5844486934353, "grad_norm": 0.2948933243751526, "learning_rate": 3.534329441835242e-05, "loss": 0.0094, "step": 184490 }, { "epoch": 117.59082217973231, "grad_norm": 0.1393466293811798, "learning_rate": 3.533802505710403e-05, "loss": 0.0112, "step": 184500 }, { "epoch": 117.59719566602932, "grad_norm": 0.4873338043689728, "learning_rate": 3.533275587401249e-05, "loss": 0.0121, "step": 184510 }, { "epoch": 117.60356915232632, "grad_norm": 0.45667463541030884, "learning_rate": 3.532748686914183e-05, "loss": 0.0138, "step": 184520 }, { "epoch": 117.60994263862332, "grad_norm": 0.4023742079734802, "learning_rate": 3.5322218042556034e-05, "loss": 0.0092, "step": 184530 }, { "epoch": 117.61631612492033, "grad_norm": 0.1650409698486328, "learning_rate": 3.531694939431914e-05, "loss": 0.0059, "step": 184540 }, { "epoch": 117.62268961121734, "grad_norm": 0.42557641863822937, "learning_rate": 3.531168092449519e-05, "loss": 0.0121, "step": 184550 }, { "epoch": 117.62906309751435, "grad_norm": 0.329184353351593, "learning_rate": 3.53064126331482e-05, "loss": 0.009, "step": 184560 }, { "epoch": 117.63543658381134, "grad_norm": 0.1788938045501709, "learning_rate": 3.5301144520342145e-05, "loss": 0.0167, "step": 184570 }, { "epoch": 117.64181007010835, "grad_norm": 0.24022994935512543, "learning_rate": 3.529587658614106e-05, "loss": 0.0059, "step": 184580 }, { "epoch": 117.64818355640536, "grad_norm": 0.16473785042762756, "learning_rate": 3.5290608830608984e-05, "loss": 0.0103, "step": 184590 }, { "epoch": 117.65455704270236, "grad_norm": 0.140604630112648, "learning_rate": 3.5285341253809877e-05, "loss": 0.0097, "step": 184600 }, { "epoch": 117.66093052899936, "grad_norm": 0.22847436368465424, "learning_rate": 3.528007385580777e-05, "loss": 0.0111, "step": 184610 }, { "epoch": 117.66730401529637, "grad_norm": 0.27357766032218933, "learning_rate": 3.5274806636666656e-05, "loss": 0.0071, "step": 184620 }, { "epoch": 117.67367750159337, "grad_norm": 0.12225200980901718, "learning_rate": 3.5269539596450574e-05, "loss": 0.0118, "step": 184630 }, { "epoch": 117.68005098789038, "grad_norm": 0.4226360619068146, "learning_rate": 3.5264272735223464e-05, "loss": 0.0056, "step": 184640 }, { "epoch": 117.68642447418738, "grad_norm": 0.49245667457580566, "learning_rate": 3.525900605304936e-05, "loss": 0.0244, "step": 184650 }, { "epoch": 117.69279796048438, "grad_norm": 0.4172997772693634, "learning_rate": 3.5253739549992266e-05, "loss": 0.0127, "step": 184660 }, { "epoch": 117.69917144678139, "grad_norm": 0.24364133179187775, "learning_rate": 3.524847322611614e-05, "loss": 0.013, "step": 184670 }, { "epoch": 117.7055449330784, "grad_norm": 0.1929270327091217, "learning_rate": 3.524320708148499e-05, "loss": 0.0097, "step": 184680 }, { "epoch": 117.7119184193754, "grad_norm": 0.17081287503242493, "learning_rate": 3.523794111616281e-05, "loss": 0.0213, "step": 184690 }, { "epoch": 117.7182919056724, "grad_norm": 0.38591572642326355, "learning_rate": 3.52326753302136e-05, "loss": 0.0142, "step": 184700 }, { "epoch": 117.72466539196941, "grad_norm": 0.32528558373451233, "learning_rate": 3.522740972370131e-05, "loss": 0.0078, "step": 184710 }, { "epoch": 117.73103887826642, "grad_norm": 0.49323540925979614, "learning_rate": 3.522214429668994e-05, "loss": 0.0109, "step": 184720 }, { "epoch": 117.73741236456341, "grad_norm": 0.3174719512462616, "learning_rate": 3.521687904924347e-05, "loss": 0.0056, "step": 184730 }, { "epoch": 117.74378585086042, "grad_norm": 0.1315196305513382, "learning_rate": 3.5211613981425894e-05, "loss": 0.0044, "step": 184740 }, { "epoch": 117.75015933715743, "grad_norm": 0.27856627106666565, "learning_rate": 3.5206349093301156e-05, "loss": 0.008, "step": 184750 }, { "epoch": 117.75653282345444, "grad_norm": 0.2177686244249344, "learning_rate": 3.5201084384933236e-05, "loss": 0.0072, "step": 184760 }, { "epoch": 117.76290630975143, "grad_norm": 0.17786045372486115, "learning_rate": 3.519581985638614e-05, "loss": 0.0105, "step": 184770 }, { "epoch": 117.76927979604844, "grad_norm": 0.5664252638816833, "learning_rate": 3.5190555507723785e-05, "loss": 0.0108, "step": 184780 }, { "epoch": 117.77565328234544, "grad_norm": 0.13033819198608398, "learning_rate": 3.518529133901017e-05, "loss": 0.0099, "step": 184790 }, { "epoch": 117.78202676864245, "grad_norm": 0.2957131564617157, "learning_rate": 3.5180027350309245e-05, "loss": 0.0069, "step": 184800 }, { "epoch": 117.78840025493945, "grad_norm": 0.4164651036262512, "learning_rate": 3.517476354168501e-05, "loss": 0.02, "step": 184810 }, { "epoch": 117.79477374123645, "grad_norm": 0.18850840628147125, "learning_rate": 3.5169499913201365e-05, "loss": 0.0123, "step": 184820 }, { "epoch": 117.80114722753346, "grad_norm": 0.24208050966262817, "learning_rate": 3.516423646492231e-05, "loss": 0.0124, "step": 184830 }, { "epoch": 117.80752071383047, "grad_norm": 0.5703145265579224, "learning_rate": 3.5158973196911795e-05, "loss": 0.0179, "step": 184840 }, { "epoch": 117.81389420012746, "grad_norm": 0.14970898628234863, "learning_rate": 3.515371010923375e-05, "loss": 0.0083, "step": 184850 }, { "epoch": 117.82026768642447, "grad_norm": 0.4447116255760193, "learning_rate": 3.514844720195215e-05, "loss": 0.0168, "step": 184860 }, { "epoch": 117.82664117272148, "grad_norm": 0.40637949109077454, "learning_rate": 3.5143184475130943e-05, "loss": 0.0154, "step": 184870 }, { "epoch": 117.83301465901849, "grad_norm": 0.25334104895591736, "learning_rate": 3.513792192883408e-05, "loss": 0.0094, "step": 184880 }, { "epoch": 117.83938814531548, "grad_norm": 0.3040623962879181, "learning_rate": 3.513265956312549e-05, "loss": 0.0112, "step": 184890 }, { "epoch": 117.84576163161249, "grad_norm": 0.2310255467891693, "learning_rate": 3.5127397378069114e-05, "loss": 0.0088, "step": 184900 }, { "epoch": 117.8521351179095, "grad_norm": 0.1896960288286209, "learning_rate": 3.5122135373728917e-05, "loss": 0.0088, "step": 184910 }, { "epoch": 117.8585086042065, "grad_norm": 0.2286655157804489, "learning_rate": 3.511687355016882e-05, "loss": 0.0072, "step": 184920 }, { "epoch": 117.8648820905035, "grad_norm": 0.27220141887664795, "learning_rate": 3.511161190745275e-05, "loss": 0.0097, "step": 184930 }, { "epoch": 117.87125557680051, "grad_norm": 0.2807601988315582, "learning_rate": 3.510635044564466e-05, "loss": 0.0111, "step": 184940 }, { "epoch": 117.87762906309752, "grad_norm": 0.2937157452106476, "learning_rate": 3.510108916480847e-05, "loss": 0.0105, "step": 184950 }, { "epoch": 117.88400254939452, "grad_norm": 0.3572637438774109, "learning_rate": 3.509582806500813e-05, "loss": 0.021, "step": 184960 }, { "epoch": 117.89037603569152, "grad_norm": 0.3246728181838989, "learning_rate": 3.509056714630753e-05, "loss": 0.0112, "step": 184970 }, { "epoch": 117.89674952198853, "grad_norm": 0.45086485147476196, "learning_rate": 3.508530640877063e-05, "loss": 0.01, "step": 184980 }, { "epoch": 117.90312300828553, "grad_norm": 0.4139746129512787, "learning_rate": 3.508004585246133e-05, "loss": 0.0075, "step": 184990 }, { "epoch": 117.90949649458254, "grad_norm": 0.2589375078678131, "learning_rate": 3.507478547744356e-05, "loss": 0.0091, "step": 185000 }, { "epoch": 117.91586998087953, "grad_norm": 0.50481116771698, "learning_rate": 3.5069525283781233e-05, "loss": 0.0208, "step": 185010 }, { "epoch": 117.92224346717654, "grad_norm": 0.20482321083545685, "learning_rate": 3.5064265271538285e-05, "loss": 0.006, "step": 185020 }, { "epoch": 117.92861695347355, "grad_norm": 0.13600504398345947, "learning_rate": 3.505900544077861e-05, "loss": 0.0067, "step": 185030 }, { "epoch": 117.93499043977056, "grad_norm": 0.19485348463058472, "learning_rate": 3.505374579156613e-05, "loss": 0.01, "step": 185040 }, { "epoch": 117.94136392606755, "grad_norm": 0.39747166633605957, "learning_rate": 3.5048486323964735e-05, "loss": 0.0122, "step": 185050 }, { "epoch": 117.94773741236456, "grad_norm": 0.3925333619117737, "learning_rate": 3.504322703803836e-05, "loss": 0.0059, "step": 185060 }, { "epoch": 117.95411089866157, "grad_norm": 0.12267609685659409, "learning_rate": 3.503796793385089e-05, "loss": 0.0107, "step": 185070 }, { "epoch": 117.96048438495858, "grad_norm": 0.3638955056667328, "learning_rate": 3.503270901146625e-05, "loss": 0.0066, "step": 185080 }, { "epoch": 117.96685787125557, "grad_norm": 0.39019784331321716, "learning_rate": 3.502745027094832e-05, "loss": 0.0138, "step": 185090 }, { "epoch": 117.97323135755258, "grad_norm": 0.20588916540145874, "learning_rate": 3.5022191712361004e-05, "loss": 0.0119, "step": 185100 }, { "epoch": 117.97960484384959, "grad_norm": 0.3928082585334778, "learning_rate": 3.5016933335768194e-05, "loss": 0.0071, "step": 185110 }, { "epoch": 117.9859783301466, "grad_norm": 0.15076501667499542, "learning_rate": 3.50116751412338e-05, "loss": 0.0127, "step": 185120 }, { "epoch": 117.99235181644359, "grad_norm": 0.32047757506370544, "learning_rate": 3.500641712882171e-05, "loss": 0.0083, "step": 185130 }, { "epoch": 117.9987253027406, "grad_norm": 0.29454323649406433, "learning_rate": 3.5001159298595784e-05, "loss": 0.0086, "step": 185140 }, { "epoch": 118.0050987890376, "grad_norm": 0.1851719170808792, "learning_rate": 3.499590165061994e-05, "loss": 0.0047, "step": 185150 }, { "epoch": 118.01147227533461, "grad_norm": 0.34186315536499023, "learning_rate": 3.499064418495809e-05, "loss": 0.0088, "step": 185160 }, { "epoch": 118.01784576163162, "grad_norm": 0.2001643031835556, "learning_rate": 3.4985386901674054e-05, "loss": 0.0078, "step": 185170 }, { "epoch": 118.02421924792861, "grad_norm": 0.23685024678707123, "learning_rate": 3.4980129800831745e-05, "loss": 0.0134, "step": 185180 }, { "epoch": 118.03059273422562, "grad_norm": 0.25642335414886475, "learning_rate": 3.497487288249504e-05, "loss": 0.0136, "step": 185190 }, { "epoch": 118.03696622052263, "grad_norm": 0.6058100461959839, "learning_rate": 3.496961614672783e-05, "loss": 0.0208, "step": 185200 }, { "epoch": 118.04333970681964, "grad_norm": 0.16048094630241394, "learning_rate": 3.496435959359396e-05, "loss": 0.0061, "step": 185210 }, { "epoch": 118.04971319311663, "grad_norm": 0.1955181211233139, "learning_rate": 3.495910322315732e-05, "loss": 0.0172, "step": 185220 }, { "epoch": 118.05608667941364, "grad_norm": 0.42803531885147095, "learning_rate": 3.495384703548179e-05, "loss": 0.0104, "step": 185230 }, { "epoch": 118.06246016571065, "grad_norm": 0.1623987853527069, "learning_rate": 3.494859103063121e-05, "loss": 0.0187, "step": 185240 }, { "epoch": 118.06883365200765, "grad_norm": 0.8235846757888794, "learning_rate": 3.494333520866946e-05, "loss": 0.0128, "step": 185250 }, { "epoch": 118.07520713830465, "grad_norm": 0.7114983797073364, "learning_rate": 3.49380795696604e-05, "loss": 0.0186, "step": 185260 }, { "epoch": 118.08158062460166, "grad_norm": 0.1732538938522339, "learning_rate": 3.4932824113667916e-05, "loss": 0.016, "step": 185270 }, { "epoch": 118.08795411089866, "grad_norm": 0.19939090311527252, "learning_rate": 3.492756884075582e-05, "loss": 0.0096, "step": 185280 }, { "epoch": 118.09432759719567, "grad_norm": 0.2350160926580429, "learning_rate": 3.4922313750988e-05, "loss": 0.0072, "step": 185290 }, { "epoch": 118.10070108349267, "grad_norm": 0.181279256939888, "learning_rate": 3.491705884442832e-05, "loss": 0.0071, "step": 185300 }, { "epoch": 118.10707456978967, "grad_norm": 0.14886313676834106, "learning_rate": 3.49118041211406e-05, "loss": 0.0071, "step": 185310 }, { "epoch": 118.11344805608668, "grad_norm": 0.5979096293449402, "learning_rate": 3.49065495811887e-05, "loss": 0.0188, "step": 185320 }, { "epoch": 118.11982154238369, "grad_norm": 0.26690876483917236, "learning_rate": 3.4901295224636475e-05, "loss": 0.0136, "step": 185330 }, { "epoch": 118.12619502868068, "grad_norm": 0.10212548077106476, "learning_rate": 3.489604105154779e-05, "loss": 0.0074, "step": 185340 }, { "epoch": 118.13256851497769, "grad_norm": 0.24630597233772278, "learning_rate": 3.489078706198644e-05, "loss": 0.0078, "step": 185350 }, { "epoch": 118.1389420012747, "grad_norm": 0.1641557663679123, "learning_rate": 3.48855332560163e-05, "loss": 0.0095, "step": 185360 }, { "epoch": 118.14531548757171, "grad_norm": 0.22123931348323822, "learning_rate": 3.48802796337012e-05, "loss": 0.0089, "step": 185370 }, { "epoch": 118.1516889738687, "grad_norm": 0.48041006922721863, "learning_rate": 3.4875026195105e-05, "loss": 0.0143, "step": 185380 }, { "epoch": 118.15806246016571, "grad_norm": 0.2020798921585083, "learning_rate": 3.486977294029149e-05, "loss": 0.0104, "step": 185390 }, { "epoch": 118.16443594646272, "grad_norm": 0.24526594579219818, "learning_rate": 3.486451986932452e-05, "loss": 0.0122, "step": 185400 }, { "epoch": 118.17080943275973, "grad_norm": 0.2763153314590454, "learning_rate": 3.4859266982267944e-05, "loss": 0.0089, "step": 185410 }, { "epoch": 118.17718291905672, "grad_norm": 0.23906771838665009, "learning_rate": 3.4854014279185556e-05, "loss": 0.0088, "step": 185420 }, { "epoch": 118.18355640535373, "grad_norm": 0.15337081253528595, "learning_rate": 3.484876176014119e-05, "loss": 0.0111, "step": 185430 }, { "epoch": 118.18992989165073, "grad_norm": 0.3555262088775635, "learning_rate": 3.484350942519867e-05, "loss": 0.0148, "step": 185440 }, { "epoch": 118.19630337794774, "grad_norm": 0.2352401316165924, "learning_rate": 3.483825727442185e-05, "loss": 0.011, "step": 185450 }, { "epoch": 118.20267686424474, "grad_norm": 0.35865530371665955, "learning_rate": 3.48330053078745e-05, "loss": 0.0108, "step": 185460 }, { "epoch": 118.20905035054174, "grad_norm": 0.3317022919654846, "learning_rate": 3.4827753525620444e-05, "loss": 0.0092, "step": 185470 }, { "epoch": 118.21542383683875, "grad_norm": 0.43637457489967346, "learning_rate": 3.482250192772354e-05, "loss": 0.0098, "step": 185480 }, { "epoch": 118.22179732313576, "grad_norm": 0.3195885121822357, "learning_rate": 3.481725051424754e-05, "loss": 0.0059, "step": 185490 }, { "epoch": 118.22817080943275, "grad_norm": 0.18523383140563965, "learning_rate": 3.481199928525628e-05, "loss": 0.0075, "step": 185500 }, { "epoch": 118.23454429572976, "grad_norm": 0.545591413974762, "learning_rate": 3.480674824081357e-05, "loss": 0.0159, "step": 185510 }, { "epoch": 118.24091778202677, "grad_norm": 0.34231629967689514, "learning_rate": 3.4801497380983237e-05, "loss": 0.0065, "step": 185520 }, { "epoch": 118.24729126832378, "grad_norm": 0.2900826334953308, "learning_rate": 3.479624670582904e-05, "loss": 0.0086, "step": 185530 }, { "epoch": 118.25366475462077, "grad_norm": 0.3679477870464325, "learning_rate": 3.479099621541479e-05, "loss": 0.0082, "step": 185540 }, { "epoch": 118.26003824091778, "grad_norm": 0.554314374923706, "learning_rate": 3.4785745909804334e-05, "loss": 0.0119, "step": 185550 }, { "epoch": 118.26641172721479, "grad_norm": 0.20194552838802338, "learning_rate": 3.47804957890614e-05, "loss": 0.0116, "step": 185560 }, { "epoch": 118.2727852135118, "grad_norm": 0.11401593685150146, "learning_rate": 3.477524585324981e-05, "loss": 0.01, "step": 185570 }, { "epoch": 118.27915869980879, "grad_norm": 0.3749147355556488, "learning_rate": 3.476999610243336e-05, "loss": 0.008, "step": 185580 }, { "epoch": 118.2855321861058, "grad_norm": 0.219307541847229, "learning_rate": 3.4764746536675856e-05, "loss": 0.0151, "step": 185590 }, { "epoch": 118.2919056724028, "grad_norm": 0.18842892348766327, "learning_rate": 3.475949715604104e-05, "loss": 0.0098, "step": 185600 }, { "epoch": 118.29827915869981, "grad_norm": 0.19374814629554749, "learning_rate": 3.475424796059273e-05, "loss": 0.0055, "step": 185610 }, { "epoch": 118.3046526449968, "grad_norm": 0.20797668397426605, "learning_rate": 3.474899895039471e-05, "loss": 0.0073, "step": 185620 }, { "epoch": 118.31102613129381, "grad_norm": 0.36050015687942505, "learning_rate": 3.474375012551076e-05, "loss": 0.0084, "step": 185630 }, { "epoch": 118.31739961759082, "grad_norm": 0.33513885736465454, "learning_rate": 3.473850148600463e-05, "loss": 0.0099, "step": 185640 }, { "epoch": 118.32377310388783, "grad_norm": 0.5175278186798096, "learning_rate": 3.473325303194012e-05, "loss": 0.01, "step": 185650 }, { "epoch": 118.33014659018482, "grad_norm": 0.15602609515190125, "learning_rate": 3.472800476338101e-05, "loss": 0.006, "step": 185660 }, { "epoch": 118.33652007648183, "grad_norm": 0.4912247657775879, "learning_rate": 3.472275668039106e-05, "loss": 0.0085, "step": 185670 }, { "epoch": 118.34289356277884, "grad_norm": 0.1890733540058136, "learning_rate": 3.4717508783034034e-05, "loss": 0.0045, "step": 185680 }, { "epoch": 118.34926704907585, "grad_norm": 0.48674264550209045, "learning_rate": 3.4712261071373696e-05, "loss": 0.0077, "step": 185690 }, { "epoch": 118.35564053537284, "grad_norm": 0.3529042899608612, "learning_rate": 3.470701354547384e-05, "loss": 0.0074, "step": 185700 }, { "epoch": 118.36201402166985, "grad_norm": 0.7561231255531311, "learning_rate": 3.4701766205398206e-05, "loss": 0.0056, "step": 185710 }, { "epoch": 118.36838750796686, "grad_norm": 0.2430102825164795, "learning_rate": 3.469651905121054e-05, "loss": 0.0123, "step": 185720 }, { "epoch": 118.37476099426387, "grad_norm": 0.20303311944007874, "learning_rate": 3.469127208297464e-05, "loss": 0.0086, "step": 185730 }, { "epoch": 118.38113448056086, "grad_norm": 0.14399820566177368, "learning_rate": 3.4686025300754215e-05, "loss": 0.0103, "step": 185740 }, { "epoch": 118.38750796685787, "grad_norm": 0.2024751603603363, "learning_rate": 3.468077870461306e-05, "loss": 0.0094, "step": 185750 }, { "epoch": 118.39388145315488, "grad_norm": 0.3112424612045288, "learning_rate": 3.467553229461489e-05, "loss": 0.0082, "step": 185760 }, { "epoch": 118.40025493945188, "grad_norm": 0.21319830417633057, "learning_rate": 3.467028607082349e-05, "loss": 0.0089, "step": 185770 }, { "epoch": 118.40662842574889, "grad_norm": 0.18369799852371216, "learning_rate": 3.466504003330257e-05, "loss": 0.0058, "step": 185780 }, { "epoch": 118.41300191204589, "grad_norm": 0.1140410378575325, "learning_rate": 3.46597941821159e-05, "loss": 0.0188, "step": 185790 }, { "epoch": 118.4193753983429, "grad_norm": 0.24418394267559052, "learning_rate": 3.4654548517327225e-05, "loss": 0.0089, "step": 185800 }, { "epoch": 118.4257488846399, "grad_norm": 0.12627282738685608, "learning_rate": 3.464930303900026e-05, "loss": 0.0059, "step": 185810 }, { "epoch": 118.43212237093691, "grad_norm": 0.15333257615566254, "learning_rate": 3.464405774719875e-05, "loss": 0.0073, "step": 185820 }, { "epoch": 118.4384958572339, "grad_norm": 0.2079312950372696, "learning_rate": 3.4638812641986454e-05, "loss": 0.0061, "step": 185830 }, { "epoch": 118.44486934353091, "grad_norm": 0.34945082664489746, "learning_rate": 3.4633567723427086e-05, "loss": 0.0112, "step": 185840 }, { "epoch": 118.45124282982792, "grad_norm": 0.5590543150901794, "learning_rate": 3.462832299158436e-05, "loss": 0.012, "step": 185850 }, { "epoch": 118.45761631612493, "grad_norm": 0.25993865728378296, "learning_rate": 3.462307844652204e-05, "loss": 0.0124, "step": 185860 }, { "epoch": 118.46398980242192, "grad_norm": 0.3778176009654999, "learning_rate": 3.461783408830384e-05, "loss": 0.0073, "step": 185870 }, { "epoch": 118.47036328871893, "grad_norm": 0.28749197721481323, "learning_rate": 3.461258991699346e-05, "loss": 0.0085, "step": 185880 }, { "epoch": 118.47673677501594, "grad_norm": 0.29997846484184265, "learning_rate": 3.4607345932654656e-05, "loss": 0.0099, "step": 185890 }, { "epoch": 118.48311026131294, "grad_norm": 0.44967544078826904, "learning_rate": 3.460210213535112e-05, "loss": 0.0102, "step": 185900 }, { "epoch": 118.48948374760994, "grad_norm": 0.2372714877128601, "learning_rate": 3.45968585251466e-05, "loss": 0.0085, "step": 185910 }, { "epoch": 118.49585723390695, "grad_norm": 0.26029831171035767, "learning_rate": 3.459161510210478e-05, "loss": 0.0103, "step": 185920 }, { "epoch": 118.50223072020395, "grad_norm": 0.2771143615245819, "learning_rate": 3.4586371866289366e-05, "loss": 0.0077, "step": 185930 }, { "epoch": 118.50860420650096, "grad_norm": 0.30843386054039, "learning_rate": 3.4581128817764107e-05, "loss": 0.011, "step": 185940 }, { "epoch": 118.51497769279796, "grad_norm": 0.18804869055747986, "learning_rate": 3.45758859565927e-05, "loss": 0.0122, "step": 185950 }, { "epoch": 118.52135117909496, "grad_norm": 0.6335318684577942, "learning_rate": 3.4570643282838824e-05, "loss": 0.0135, "step": 185960 }, { "epoch": 118.52772466539197, "grad_norm": 0.24816115200519562, "learning_rate": 3.45654007965662e-05, "loss": 0.0149, "step": 185970 }, { "epoch": 118.53409815168898, "grad_norm": 0.138414204120636, "learning_rate": 3.456015849783856e-05, "loss": 0.0119, "step": 185980 }, { "epoch": 118.54047163798597, "grad_norm": 0.18016496300697327, "learning_rate": 3.455491638671954e-05, "loss": 0.0071, "step": 185990 }, { "epoch": 118.54684512428298, "grad_norm": 0.32132551074028015, "learning_rate": 3.454967446327286e-05, "loss": 0.0082, "step": 186000 }, { "epoch": 118.55321861057999, "grad_norm": 0.1981860101222992, "learning_rate": 3.454443272756224e-05, "loss": 0.0055, "step": 186010 }, { "epoch": 118.559592096877, "grad_norm": 0.5462917685508728, "learning_rate": 3.453919117965136e-05, "loss": 0.0146, "step": 186020 }, { "epoch": 118.56596558317399, "grad_norm": 0.11586359143257141, "learning_rate": 3.453394981960389e-05, "loss": 0.0071, "step": 186030 }, { "epoch": 118.572339069471, "grad_norm": 0.2332872450351715, "learning_rate": 3.452870864748354e-05, "loss": 0.0065, "step": 186040 }, { "epoch": 118.578712555768, "grad_norm": 0.26679080724716187, "learning_rate": 3.4523467663354e-05, "loss": 0.0167, "step": 186050 }, { "epoch": 118.58508604206502, "grad_norm": 0.21424280107021332, "learning_rate": 3.451822686727893e-05, "loss": 0.0069, "step": 186060 }, { "epoch": 118.59145952836201, "grad_norm": 0.21609346568584442, "learning_rate": 3.4512986259322005e-05, "loss": 0.006, "step": 186070 }, { "epoch": 118.59783301465902, "grad_norm": 0.22801411151885986, "learning_rate": 3.450774583954692e-05, "loss": 0.0101, "step": 186080 }, { "epoch": 118.60420650095602, "grad_norm": 0.16960832476615906, "learning_rate": 3.450250560801739e-05, "loss": 0.0123, "step": 186090 }, { "epoch": 118.61057998725303, "grad_norm": 0.20928922295570374, "learning_rate": 3.449726556479701e-05, "loss": 0.0092, "step": 186100 }, { "epoch": 118.61695347355003, "grad_norm": 0.13442380726337433, "learning_rate": 3.449202570994949e-05, "loss": 0.0077, "step": 186110 }, { "epoch": 118.62332695984703, "grad_norm": 0.3333491086959839, "learning_rate": 3.4486786043538535e-05, "loss": 0.017, "step": 186120 }, { "epoch": 118.62970044614404, "grad_norm": 0.7096704840660095, "learning_rate": 3.448154656562774e-05, "loss": 0.0117, "step": 186130 }, { "epoch": 118.63607393244105, "grad_norm": 0.3781968355178833, "learning_rate": 3.4476307276280815e-05, "loss": 0.012, "step": 186140 }, { "epoch": 118.64244741873804, "grad_norm": 0.24280154705047607, "learning_rate": 3.447106817556141e-05, "loss": 0.0095, "step": 186150 }, { "epoch": 118.64882090503505, "grad_norm": 0.3726077675819397, "learning_rate": 3.446582926353321e-05, "loss": 0.0113, "step": 186160 }, { "epoch": 118.65519439133206, "grad_norm": 0.24608908593654633, "learning_rate": 3.446059054025983e-05, "loss": 0.0114, "step": 186170 }, { "epoch": 118.66156787762907, "grad_norm": 0.23718662559986115, "learning_rate": 3.445535200580495e-05, "loss": 0.0094, "step": 186180 }, { "epoch": 118.66794136392606, "grad_norm": 0.27885672450065613, "learning_rate": 3.4450113660232244e-05, "loss": 0.0052, "step": 186190 }, { "epoch": 118.67431485022307, "grad_norm": 0.24220582842826843, "learning_rate": 3.444487550360531e-05, "loss": 0.0113, "step": 186200 }, { "epoch": 118.68068833652008, "grad_norm": 0.37253955006599426, "learning_rate": 3.443963753598782e-05, "loss": 0.0073, "step": 186210 }, { "epoch": 118.68706182281709, "grad_norm": 0.15054307878017426, "learning_rate": 3.443439975744344e-05, "loss": 0.0041, "step": 186220 }, { "epoch": 118.69343530911408, "grad_norm": 0.4196837842464447, "learning_rate": 3.442916216803581e-05, "loss": 0.0174, "step": 186230 }, { "epoch": 118.69980879541109, "grad_norm": 0.3447287976741791, "learning_rate": 3.442392476782854e-05, "loss": 0.0222, "step": 186240 }, { "epoch": 118.7061822817081, "grad_norm": 0.28224271535873413, "learning_rate": 3.441868755688529e-05, "loss": 0.0081, "step": 186250 }, { "epoch": 118.7125557680051, "grad_norm": 0.28383398056030273, "learning_rate": 3.44134505352697e-05, "loss": 0.0081, "step": 186260 }, { "epoch": 118.7189292543021, "grad_norm": 0.4879746735095978, "learning_rate": 3.440821370304542e-05, "loss": 0.0187, "step": 186270 }, { "epoch": 118.7253027405991, "grad_norm": 0.4323327839374542, "learning_rate": 3.440297706027606e-05, "loss": 0.0115, "step": 186280 }, { "epoch": 118.73167622689611, "grad_norm": 0.18524613976478577, "learning_rate": 3.4397740607025234e-05, "loss": 0.0081, "step": 186290 }, { "epoch": 118.73804971319312, "grad_norm": 0.45097506046295166, "learning_rate": 3.439250434335662e-05, "loss": 0.0117, "step": 186300 }, { "epoch": 118.74442319949011, "grad_norm": 0.3374060392379761, "learning_rate": 3.438726826933379e-05, "loss": 0.0074, "step": 186310 }, { "epoch": 118.75079668578712, "grad_norm": 0.15724848210811615, "learning_rate": 3.438203238502039e-05, "loss": 0.0116, "step": 186320 }, { "epoch": 118.75717017208413, "grad_norm": 0.3579750657081604, "learning_rate": 3.437679669048005e-05, "loss": 0.012, "step": 186330 }, { "epoch": 118.76354365838114, "grad_norm": 0.41897153854370117, "learning_rate": 3.43715611857764e-05, "loss": 0.0112, "step": 186340 }, { "epoch": 118.76991714467813, "grad_norm": 0.2090025246143341, "learning_rate": 3.436632587097302e-05, "loss": 0.0069, "step": 186350 }, { "epoch": 118.77629063097514, "grad_norm": 0.3959963321685791, "learning_rate": 3.436109074613353e-05, "loss": 0.0105, "step": 186360 }, { "epoch": 118.78266411727215, "grad_norm": 0.22894705832004547, "learning_rate": 3.435585581132157e-05, "loss": 0.0054, "step": 186370 }, { "epoch": 118.78903760356916, "grad_norm": 0.24185827374458313, "learning_rate": 3.435062106660073e-05, "loss": 0.0161, "step": 186380 }, { "epoch": 118.79541108986615, "grad_norm": 0.33043184876441956, "learning_rate": 3.434538651203461e-05, "loss": 0.0097, "step": 186390 }, { "epoch": 118.80178457616316, "grad_norm": 0.3998052179813385, "learning_rate": 3.4340152147686824e-05, "loss": 0.0066, "step": 186400 }, { "epoch": 118.80815806246017, "grad_norm": 0.4439808428287506, "learning_rate": 3.4334917973620987e-05, "loss": 0.0101, "step": 186410 }, { "epoch": 118.81453154875717, "grad_norm": 0.14668023586273193, "learning_rate": 3.4329683989900684e-05, "loss": 0.0179, "step": 186420 }, { "epoch": 118.82090503505418, "grad_norm": 0.8477301001548767, "learning_rate": 3.4324450196589506e-05, "loss": 0.0077, "step": 186430 }, { "epoch": 118.82727852135118, "grad_norm": 0.33219608664512634, "learning_rate": 3.431921659375107e-05, "loss": 0.0119, "step": 186440 }, { "epoch": 118.83365200764818, "grad_norm": 0.21581824123859406, "learning_rate": 3.431398318144895e-05, "loss": 0.0084, "step": 186450 }, { "epoch": 118.84002549394519, "grad_norm": 0.2606217861175537, "learning_rate": 3.430874995974675e-05, "loss": 0.0124, "step": 186460 }, { "epoch": 118.8463989802422, "grad_norm": 0.17745821177959442, "learning_rate": 3.430351692870804e-05, "loss": 0.0199, "step": 186470 }, { "epoch": 118.8527724665392, "grad_norm": 0.327726811170578, "learning_rate": 3.429828408839644e-05, "loss": 0.016, "step": 186480 }, { "epoch": 118.8591459528362, "grad_norm": 0.4861826002597809, "learning_rate": 3.429305143887549e-05, "loss": 0.0169, "step": 186490 }, { "epoch": 118.86551943913321, "grad_norm": 0.30830278992652893, "learning_rate": 3.428781898020882e-05, "loss": 0.013, "step": 186500 }, { "epoch": 118.87189292543022, "grad_norm": 0.17246249318122864, "learning_rate": 3.4282586712459976e-05, "loss": 0.0077, "step": 186510 }, { "epoch": 118.87826641172721, "grad_norm": 0.23478013277053833, "learning_rate": 3.427735463569253e-05, "loss": 0.0101, "step": 186520 }, { "epoch": 118.88463989802422, "grad_norm": 0.30013060569763184, "learning_rate": 3.427212274997007e-05, "loss": 0.0104, "step": 186530 }, { "epoch": 118.89101338432123, "grad_norm": 0.3314497172832489, "learning_rate": 3.426689105535617e-05, "loss": 0.0106, "step": 186540 }, { "epoch": 118.89738687061823, "grad_norm": 0.41326606273651123, "learning_rate": 3.426165955191442e-05, "loss": 0.0073, "step": 186550 }, { "epoch": 118.90376035691523, "grad_norm": 0.12219265848398209, "learning_rate": 3.425642823970835e-05, "loss": 0.0039, "step": 186560 }, { "epoch": 118.91013384321224, "grad_norm": 0.40051066875457764, "learning_rate": 3.4251197118801534e-05, "loss": 0.0068, "step": 186570 }, { "epoch": 118.91650732950924, "grad_norm": 0.26783910393714905, "learning_rate": 3.4245966189257536e-05, "loss": 0.0079, "step": 186580 }, { "epoch": 118.92288081580625, "grad_norm": 1.0861859321594238, "learning_rate": 3.424073545113996e-05, "loss": 0.021, "step": 186590 }, { "epoch": 118.92925430210325, "grad_norm": 0.5669177174568176, "learning_rate": 3.42355049045123e-05, "loss": 0.0087, "step": 186600 }, { "epoch": 118.93562778840025, "grad_norm": 0.26558059453964233, "learning_rate": 3.423027454943813e-05, "loss": 0.0075, "step": 186610 }, { "epoch": 118.94200127469726, "grad_norm": 0.3407991826534271, "learning_rate": 3.4225044385981054e-05, "loss": 0.0143, "step": 186620 }, { "epoch": 118.94837476099427, "grad_norm": 0.5569522380828857, "learning_rate": 3.4219814414204545e-05, "loss": 0.0098, "step": 186630 }, { "epoch": 118.95474824729126, "grad_norm": 0.41362687945365906, "learning_rate": 3.42145846341722e-05, "loss": 0.0067, "step": 186640 }, { "epoch": 118.96112173358827, "grad_norm": 0.30409371852874756, "learning_rate": 3.420935504594754e-05, "loss": 0.0059, "step": 186650 }, { "epoch": 118.96749521988528, "grad_norm": 0.10038065910339355, "learning_rate": 3.420412564959416e-05, "loss": 0.0075, "step": 186660 }, { "epoch": 118.97386870618229, "grad_norm": 0.12926946580410004, "learning_rate": 3.419889644517555e-05, "loss": 0.0079, "step": 186670 }, { "epoch": 118.98024219247928, "grad_norm": 0.20123912394046783, "learning_rate": 3.419366743275526e-05, "loss": 0.0068, "step": 186680 }, { "epoch": 118.98661567877629, "grad_norm": 0.28920841217041016, "learning_rate": 3.4188438612396855e-05, "loss": 0.0124, "step": 186690 }, { "epoch": 118.9929891650733, "grad_norm": 0.43725818395614624, "learning_rate": 3.418320998416383e-05, "loss": 0.0139, "step": 186700 }, { "epoch": 118.9993626513703, "grad_norm": 0.256305068731308, "learning_rate": 3.417798154811973e-05, "loss": 0.0122, "step": 186710 }, { "epoch": 119.0057361376673, "grad_norm": 0.34797191619873047, "learning_rate": 3.417275330432811e-05, "loss": 0.0185, "step": 186720 }, { "epoch": 119.0121096239643, "grad_norm": 0.17991113662719727, "learning_rate": 3.4167525252852484e-05, "loss": 0.0068, "step": 186730 }, { "epoch": 119.01848311026131, "grad_norm": 0.24203404784202576, "learning_rate": 3.416229739375636e-05, "loss": 0.0105, "step": 186740 }, { "epoch": 119.02485659655832, "grad_norm": 0.3297567069530487, "learning_rate": 3.4157069727103274e-05, "loss": 0.0121, "step": 186750 }, { "epoch": 119.03123008285532, "grad_norm": 0.5707392692565918, "learning_rate": 3.415184225295677e-05, "loss": 0.0115, "step": 186760 }, { "epoch": 119.03760356915232, "grad_norm": 0.1268906593322754, "learning_rate": 3.414661497138033e-05, "loss": 0.0057, "step": 186770 }, { "epoch": 119.04397705544933, "grad_norm": 0.1822340041399002, "learning_rate": 3.4141387882437484e-05, "loss": 0.0162, "step": 186780 }, { "epoch": 119.05035054174634, "grad_norm": 0.21467739343643188, "learning_rate": 3.4136160986191744e-05, "loss": 0.006, "step": 186790 }, { "epoch": 119.05672402804333, "grad_norm": 0.35893014073371887, "learning_rate": 3.4130934282706653e-05, "loss": 0.0101, "step": 186800 }, { "epoch": 119.06309751434034, "grad_norm": 0.40892520546913147, "learning_rate": 3.4125707772045666e-05, "loss": 0.0089, "step": 186810 }, { "epoch": 119.06947100063735, "grad_norm": 0.20948047935962677, "learning_rate": 3.412048145427232e-05, "loss": 0.0096, "step": 186820 }, { "epoch": 119.07584448693436, "grad_norm": 0.2953930199146271, "learning_rate": 3.411525532945012e-05, "loss": 0.015, "step": 186830 }, { "epoch": 119.08221797323135, "grad_norm": 0.29375675320625305, "learning_rate": 3.411002939764258e-05, "loss": 0.0068, "step": 186840 }, { "epoch": 119.08859145952836, "grad_norm": 0.18204820156097412, "learning_rate": 3.410480365891318e-05, "loss": 0.007, "step": 186850 }, { "epoch": 119.09496494582537, "grad_norm": 0.2587236762046814, "learning_rate": 3.4099578113325406e-05, "loss": 0.0097, "step": 186860 }, { "epoch": 119.10133843212238, "grad_norm": 0.21637944877147675, "learning_rate": 3.40943527609428e-05, "loss": 0.0129, "step": 186870 }, { "epoch": 119.10771191841937, "grad_norm": 0.19657178223133087, "learning_rate": 3.40891276018288e-05, "loss": 0.0062, "step": 186880 }, { "epoch": 119.11408540471638, "grad_norm": 0.4627445936203003, "learning_rate": 3.4083902636046926e-05, "loss": 0.0144, "step": 186890 }, { "epoch": 119.12045889101339, "grad_norm": 0.34977972507476807, "learning_rate": 3.407867786366066e-05, "loss": 0.0112, "step": 186900 }, { "epoch": 119.1268323773104, "grad_norm": 0.2439834475517273, "learning_rate": 3.407345328473351e-05, "loss": 0.0076, "step": 186910 }, { "epoch": 119.13320586360739, "grad_norm": 0.2001470923423767, "learning_rate": 3.406822889932892e-05, "loss": 0.0107, "step": 186920 }, { "epoch": 119.1395793499044, "grad_norm": 0.23699060082435608, "learning_rate": 3.406300470751039e-05, "loss": 0.0089, "step": 186930 }, { "epoch": 119.1459528362014, "grad_norm": 0.11575308442115784, "learning_rate": 3.405778070934142e-05, "loss": 0.0127, "step": 186940 }, { "epoch": 119.15232632249841, "grad_norm": 0.22917543351650238, "learning_rate": 3.4052556904885446e-05, "loss": 0.0119, "step": 186950 }, { "epoch": 119.1586998087954, "grad_norm": 0.2246921807527542, "learning_rate": 3.404733329420596e-05, "loss": 0.0066, "step": 186960 }, { "epoch": 119.16507329509241, "grad_norm": 0.21541133522987366, "learning_rate": 3.4042109877366446e-05, "loss": 0.0127, "step": 186970 }, { "epoch": 119.17144678138942, "grad_norm": 0.7591056227684021, "learning_rate": 3.4036886654430375e-05, "loss": 0.0221, "step": 186980 }, { "epoch": 119.17782026768643, "grad_norm": 0.23754924535751343, "learning_rate": 3.403166362546119e-05, "loss": 0.0065, "step": 186990 }, { "epoch": 119.18419375398342, "grad_norm": 0.1698738932609558, "learning_rate": 3.4026440790522364e-05, "loss": 0.0075, "step": 187000 }, { "epoch": 119.19056724028043, "grad_norm": 0.21271902322769165, "learning_rate": 3.402121814967738e-05, "loss": 0.0085, "step": 187010 }, { "epoch": 119.19694072657744, "grad_norm": 0.15121783316135406, "learning_rate": 3.401599570298967e-05, "loss": 0.0068, "step": 187020 }, { "epoch": 119.20331421287445, "grad_norm": 0.21890173852443695, "learning_rate": 3.40107734505227e-05, "loss": 0.0102, "step": 187030 }, { "epoch": 119.20968769917145, "grad_norm": 0.1671801656484604, "learning_rate": 3.4005551392339926e-05, "loss": 0.0075, "step": 187040 }, { "epoch": 119.21606118546845, "grad_norm": 0.16397535800933838, "learning_rate": 3.4000329528504825e-05, "loss": 0.0101, "step": 187050 }, { "epoch": 119.22243467176546, "grad_norm": 0.46584552526474, "learning_rate": 3.3995107859080814e-05, "loss": 0.0073, "step": 187060 }, { "epoch": 119.22880815806246, "grad_norm": 0.37249866127967834, "learning_rate": 3.398988638413134e-05, "loss": 0.012, "step": 187070 }, { "epoch": 119.23518164435947, "grad_norm": 0.15465621650218964, "learning_rate": 3.398466510371988e-05, "loss": 0.0069, "step": 187080 }, { "epoch": 119.24155513065647, "grad_norm": 0.1882016807794571, "learning_rate": 3.397944401790987e-05, "loss": 0.0057, "step": 187090 }, { "epoch": 119.24792861695347, "grad_norm": 0.2635177969932556, "learning_rate": 3.3974223126764715e-05, "loss": 0.0076, "step": 187100 }, { "epoch": 119.25430210325048, "grad_norm": 0.22089619934558868, "learning_rate": 3.396900243034789e-05, "loss": 0.0108, "step": 187110 }, { "epoch": 119.26067558954749, "grad_norm": 0.4386845529079437, "learning_rate": 3.396378192872283e-05, "loss": 0.0106, "step": 187120 }, { "epoch": 119.26704907584448, "grad_norm": 0.17326079308986664, "learning_rate": 3.3958561621952946e-05, "loss": 0.0149, "step": 187130 }, { "epoch": 119.27342256214149, "grad_norm": 0.2176916003227234, "learning_rate": 3.395334151010169e-05, "loss": 0.0067, "step": 187140 }, { "epoch": 119.2797960484385, "grad_norm": 0.15622110664844513, "learning_rate": 3.3948121593232486e-05, "loss": 0.0069, "step": 187150 }, { "epoch": 119.2861695347355, "grad_norm": 0.2844868004322052, "learning_rate": 3.394290187140876e-05, "loss": 0.0094, "step": 187160 }, { "epoch": 119.2925430210325, "grad_norm": 0.35404226183891296, "learning_rate": 3.3937682344693935e-05, "loss": 0.0093, "step": 187170 }, { "epoch": 119.29891650732951, "grad_norm": 0.3368358016014099, "learning_rate": 3.393246301315144e-05, "loss": 0.008, "step": 187180 }, { "epoch": 119.30528999362652, "grad_norm": 0.40355923771858215, "learning_rate": 3.3927243876844686e-05, "loss": 0.01, "step": 187190 }, { "epoch": 119.31166347992352, "grad_norm": 0.2722325921058655, "learning_rate": 3.392202493583709e-05, "loss": 0.0106, "step": 187200 }, { "epoch": 119.31803696622052, "grad_norm": 0.14830701053142548, "learning_rate": 3.391680619019208e-05, "loss": 0.0059, "step": 187210 }, { "epoch": 119.32441045251753, "grad_norm": 0.2513669431209564, "learning_rate": 3.391158763997305e-05, "loss": 0.0112, "step": 187220 }, { "epoch": 119.33078393881453, "grad_norm": 0.2679120898246765, "learning_rate": 3.3906369285243434e-05, "loss": 0.0079, "step": 187230 }, { "epoch": 119.33715742511154, "grad_norm": 0.4495238661766052, "learning_rate": 3.390115112606661e-05, "loss": 0.0108, "step": 187240 }, { "epoch": 119.34353091140854, "grad_norm": 0.1537891924381256, "learning_rate": 3.3895933162506e-05, "loss": 0.008, "step": 187250 }, { "epoch": 119.34990439770554, "grad_norm": 0.4033711552619934, "learning_rate": 3.3890715394625025e-05, "loss": 0.0073, "step": 187260 }, { "epoch": 119.35627788400255, "grad_norm": 0.20228245854377747, "learning_rate": 3.3885497822487055e-05, "loss": 0.0169, "step": 187270 }, { "epoch": 119.36265137029956, "grad_norm": 0.30219709873199463, "learning_rate": 3.3880280446155496e-05, "loss": 0.0112, "step": 187280 }, { "epoch": 119.36902485659655, "grad_norm": 0.4655941128730774, "learning_rate": 3.387506326569375e-05, "loss": 0.0089, "step": 187290 }, { "epoch": 119.37539834289356, "grad_norm": 0.14007039368152618, "learning_rate": 3.386984628116523e-05, "loss": 0.0102, "step": 187300 }, { "epoch": 119.38177182919057, "grad_norm": 0.25724709033966064, "learning_rate": 3.386462949263329e-05, "loss": 0.0071, "step": 187310 }, { "epoch": 119.38814531548758, "grad_norm": 0.1471368819475174, "learning_rate": 3.3859412900161336e-05, "loss": 0.0116, "step": 187320 }, { "epoch": 119.39451880178457, "grad_norm": 0.35443663597106934, "learning_rate": 3.385419650381277e-05, "loss": 0.006, "step": 187330 }, { "epoch": 119.40089228808158, "grad_norm": 0.30422523617744446, "learning_rate": 3.384898030365094e-05, "loss": 0.0075, "step": 187340 }, { "epoch": 119.40726577437859, "grad_norm": 0.4362550675868988, "learning_rate": 3.3843764299739255e-05, "loss": 0.0188, "step": 187350 }, { "epoch": 119.4136392606756, "grad_norm": 0.19495521485805511, "learning_rate": 3.383854849214108e-05, "loss": 0.0107, "step": 187360 }, { "epoch": 119.42001274697259, "grad_norm": 0.17926853895187378, "learning_rate": 3.383333288091983e-05, "loss": 0.0152, "step": 187370 }, { "epoch": 119.4263862332696, "grad_norm": 0.4186975657939911, "learning_rate": 3.382811746613882e-05, "loss": 0.012, "step": 187380 }, { "epoch": 119.4327597195666, "grad_norm": 0.367779016494751, "learning_rate": 3.3822902247861454e-05, "loss": 0.0119, "step": 187390 }, { "epoch": 119.43913320586361, "grad_norm": 0.18053476512432098, "learning_rate": 3.381768722615112e-05, "loss": 0.0084, "step": 187400 }, { "epoch": 119.4455066921606, "grad_norm": 0.20297333598136902, "learning_rate": 3.381247240107114e-05, "loss": 0.0098, "step": 187410 }, { "epoch": 119.45188017845761, "grad_norm": 0.22191482782363892, "learning_rate": 3.380725777268491e-05, "loss": 0.0073, "step": 187420 }, { "epoch": 119.45825366475462, "grad_norm": 0.49856117367744446, "learning_rate": 3.380204334105579e-05, "loss": 0.0171, "step": 187430 }, { "epoch": 119.46462715105163, "grad_norm": 0.12751837074756622, "learning_rate": 3.379682910624715e-05, "loss": 0.013, "step": 187440 }, { "epoch": 119.47100063734862, "grad_norm": 0.20724013447761536, "learning_rate": 3.379161506832231e-05, "loss": 0.0094, "step": 187450 }, { "epoch": 119.47737412364563, "grad_norm": 0.4933573305606842, "learning_rate": 3.378640122734466e-05, "loss": 0.0097, "step": 187460 }, { "epoch": 119.48374760994264, "grad_norm": 0.3024183213710785, "learning_rate": 3.3781187583377536e-05, "loss": 0.0188, "step": 187470 }, { "epoch": 119.49012109623965, "grad_norm": 0.1735636293888092, "learning_rate": 3.377597413648432e-05, "loss": 0.0052, "step": 187480 }, { "epoch": 119.49649458253664, "grad_norm": 0.23630887269973755, "learning_rate": 3.377076088672832e-05, "loss": 0.0069, "step": 187490 }, { "epoch": 119.50286806883365, "grad_norm": 0.22041195631027222, "learning_rate": 3.3765547834172896e-05, "loss": 0.0057, "step": 187500 }, { "epoch": 119.50924155513066, "grad_norm": 0.21709054708480835, "learning_rate": 3.3760334978881414e-05, "loss": 0.0146, "step": 187510 }, { "epoch": 119.51561504142767, "grad_norm": 0.12923133373260498, "learning_rate": 3.3755122320917173e-05, "loss": 0.0093, "step": 187520 }, { "epoch": 119.52198852772466, "grad_norm": 0.1649313122034073, "learning_rate": 3.374990986034353e-05, "loss": 0.0058, "step": 187530 }, { "epoch": 119.52836201402167, "grad_norm": 0.3241010010242462, "learning_rate": 3.374469759722383e-05, "loss": 0.0099, "step": 187540 }, { "epoch": 119.53473550031867, "grad_norm": 0.19262367486953735, "learning_rate": 3.3739485531621415e-05, "loss": 0.0095, "step": 187550 }, { "epoch": 119.54110898661568, "grad_norm": 2.8936164379119873, "learning_rate": 3.373427366359959e-05, "loss": 0.0124, "step": 187560 }, { "epoch": 119.54748247291268, "grad_norm": 0.334114134311676, "learning_rate": 3.37290619932217e-05, "loss": 0.0065, "step": 187570 }, { "epoch": 119.55385595920968, "grad_norm": 0.28935644030570984, "learning_rate": 3.372385052055108e-05, "loss": 0.0079, "step": 187580 }, { "epoch": 119.56022944550669, "grad_norm": 0.19699762761592865, "learning_rate": 3.3718639245651026e-05, "loss": 0.0074, "step": 187590 }, { "epoch": 119.5666029318037, "grad_norm": 0.20759056508541107, "learning_rate": 3.3713428168584884e-05, "loss": 0.0066, "step": 187600 }, { "epoch": 119.5729764181007, "grad_norm": 0.20662526786327362, "learning_rate": 3.370821728941596e-05, "loss": 0.0062, "step": 187610 }, { "epoch": 119.5793499043977, "grad_norm": 0.2458811104297638, "learning_rate": 3.37030066082076e-05, "loss": 0.0093, "step": 187620 }, { "epoch": 119.58572339069471, "grad_norm": 0.23415426909923553, "learning_rate": 3.3697796125023076e-05, "loss": 0.0077, "step": 187630 }, { "epoch": 119.59209687699172, "grad_norm": 0.2552855908870697, "learning_rate": 3.3692585839925726e-05, "loss": 0.0219, "step": 187640 }, { "epoch": 119.59847036328873, "grad_norm": 0.18189485371112823, "learning_rate": 3.3687375752978867e-05, "loss": 0.0088, "step": 187650 }, { "epoch": 119.60484384958572, "grad_norm": 0.29844921827316284, "learning_rate": 3.368216586424578e-05, "loss": 0.012, "step": 187660 }, { "epoch": 119.61121733588273, "grad_norm": 0.3938780426979065, "learning_rate": 3.3676956173789784e-05, "loss": 0.0085, "step": 187670 }, { "epoch": 119.61759082217974, "grad_norm": 0.24606908857822418, "learning_rate": 3.3671746681674176e-05, "loss": 0.0147, "step": 187680 }, { "epoch": 119.62396430847674, "grad_norm": 0.22645191848278046, "learning_rate": 3.366653738796229e-05, "loss": 0.0075, "step": 187690 }, { "epoch": 119.63033779477374, "grad_norm": 0.542824923992157, "learning_rate": 3.366132829271737e-05, "loss": 0.0213, "step": 187700 }, { "epoch": 119.63671128107075, "grad_norm": 0.2740172743797302, "learning_rate": 3.365611939600274e-05, "loss": 0.0113, "step": 187710 }, { "epoch": 119.64308476736775, "grad_norm": 0.18111124634742737, "learning_rate": 3.36509106978817e-05, "loss": 0.0181, "step": 187720 }, { "epoch": 119.64945825366476, "grad_norm": 0.2292851060628891, "learning_rate": 3.364570219841753e-05, "loss": 0.0053, "step": 187730 }, { "epoch": 119.65583173996176, "grad_norm": 0.28144291043281555, "learning_rate": 3.364049389767351e-05, "loss": 0.0158, "step": 187740 }, { "epoch": 119.66220522625876, "grad_norm": 0.14677006006240845, "learning_rate": 3.3635285795712934e-05, "loss": 0.0184, "step": 187750 }, { "epoch": 119.66857871255577, "grad_norm": 0.2728338837623596, "learning_rate": 3.36300778925991e-05, "loss": 0.0064, "step": 187760 }, { "epoch": 119.67495219885278, "grad_norm": 0.6350629329681396, "learning_rate": 3.3624870188395274e-05, "loss": 0.0142, "step": 187770 }, { "epoch": 119.68132568514977, "grad_norm": 0.19698944687843323, "learning_rate": 3.361966268316473e-05, "loss": 0.0086, "step": 187780 }, { "epoch": 119.68769917144678, "grad_norm": 0.18060649931430817, "learning_rate": 3.3614455376970744e-05, "loss": 0.0181, "step": 187790 }, { "epoch": 119.69407265774379, "grad_norm": 0.2385033369064331, "learning_rate": 3.3609248269876604e-05, "loss": 0.009, "step": 187800 }, { "epoch": 119.7004461440408, "grad_norm": 0.2971429228782654, "learning_rate": 3.360404136194558e-05, "loss": 0.0089, "step": 187810 }, { "epoch": 119.70681963033779, "grad_norm": 0.1928960531949997, "learning_rate": 3.359883465324092e-05, "loss": 0.0152, "step": 187820 }, { "epoch": 119.7131931166348, "grad_norm": 0.374929279088974, "learning_rate": 3.3593628143825916e-05, "loss": 0.0171, "step": 187830 }, { "epoch": 119.7195666029318, "grad_norm": 0.2181718498468399, "learning_rate": 3.3588421833763806e-05, "loss": 0.0074, "step": 187840 }, { "epoch": 119.72594008922881, "grad_norm": 0.2760421633720398, "learning_rate": 3.358321572311788e-05, "loss": 0.0136, "step": 187850 }, { "epoch": 119.73231357552581, "grad_norm": 0.2938764989376068, "learning_rate": 3.357800981195137e-05, "loss": 0.0134, "step": 187860 }, { "epoch": 119.73868706182282, "grad_norm": 0.36232468485832214, "learning_rate": 3.357280410032756e-05, "loss": 0.0096, "step": 187870 }, { "epoch": 119.74506054811982, "grad_norm": 0.16065604984760284, "learning_rate": 3.3567598588309676e-05, "loss": 0.0113, "step": 187880 }, { "epoch": 119.75143403441683, "grad_norm": 0.1583741456270218, "learning_rate": 3.356239327596099e-05, "loss": 0.0078, "step": 187890 }, { "epoch": 119.75780752071383, "grad_norm": 0.10757773369550705, "learning_rate": 3.355718816334475e-05, "loss": 0.0095, "step": 187900 }, { "epoch": 119.76418100701083, "grad_norm": 0.279744029045105, "learning_rate": 3.355198325052419e-05, "loss": 0.0149, "step": 187910 }, { "epoch": 119.77055449330784, "grad_norm": 0.16854852437973022, "learning_rate": 3.3546778537562564e-05, "loss": 0.009, "step": 187920 }, { "epoch": 119.77692797960485, "grad_norm": 0.21179279685020447, "learning_rate": 3.354157402452312e-05, "loss": 0.0068, "step": 187930 }, { "epoch": 119.78330146590184, "grad_norm": 0.34994834661483765, "learning_rate": 3.353636971146909e-05, "loss": 0.0087, "step": 187940 }, { "epoch": 119.78967495219885, "grad_norm": 0.44888773560523987, "learning_rate": 3.35311655984637e-05, "loss": 0.008, "step": 187950 }, { "epoch": 119.79604843849586, "grad_norm": 0.2748924493789673, "learning_rate": 3.352596168557021e-05, "loss": 0.0074, "step": 187960 }, { "epoch": 119.80242192479287, "grad_norm": 0.7034946084022522, "learning_rate": 3.3520757972851846e-05, "loss": 0.0079, "step": 187970 }, { "epoch": 119.80879541108986, "grad_norm": 0.24560752511024475, "learning_rate": 3.3515554460371815e-05, "loss": 0.0141, "step": 187980 }, { "epoch": 119.81516889738687, "grad_norm": 0.46593162417411804, "learning_rate": 3.3510351148193364e-05, "loss": 0.0112, "step": 187990 }, { "epoch": 119.82154238368388, "grad_norm": 0.25347769260406494, "learning_rate": 3.3505148036379716e-05, "loss": 0.0087, "step": 188000 }, { "epoch": 119.82791586998088, "grad_norm": 0.5385751128196716, "learning_rate": 3.3499945124994115e-05, "loss": 0.0121, "step": 188010 }, { "epoch": 119.83428935627788, "grad_norm": 0.280311644077301, "learning_rate": 3.3494742414099735e-05, "loss": 0.0124, "step": 188020 }, { "epoch": 119.84066284257489, "grad_norm": 0.2528514862060547, "learning_rate": 3.348953990375981e-05, "loss": 0.0091, "step": 188030 }, { "epoch": 119.8470363288719, "grad_norm": 0.2615147531032562, "learning_rate": 3.348433759403757e-05, "loss": 0.014, "step": 188040 }, { "epoch": 119.8534098151689, "grad_norm": 0.21469804644584656, "learning_rate": 3.347913548499625e-05, "loss": 0.0072, "step": 188050 }, { "epoch": 119.8597833014659, "grad_norm": 0.7054173350334167, "learning_rate": 3.347393357669901e-05, "loss": 0.012, "step": 188060 }, { "epoch": 119.8661567877629, "grad_norm": 0.2275954931974411, "learning_rate": 3.346873186920907e-05, "loss": 0.008, "step": 188070 }, { "epoch": 119.87253027405991, "grad_norm": 0.20051677525043488, "learning_rate": 3.3463530362589676e-05, "loss": 0.0074, "step": 188080 }, { "epoch": 119.87890376035692, "grad_norm": 0.39606907963752747, "learning_rate": 3.3458329056903975e-05, "loss": 0.0155, "step": 188090 }, { "epoch": 119.88527724665391, "grad_norm": 0.6015409827232361, "learning_rate": 3.3453127952215195e-05, "loss": 0.0161, "step": 188100 }, { "epoch": 119.89165073295092, "grad_norm": 0.2017737627029419, "learning_rate": 3.344792704858654e-05, "loss": 0.0067, "step": 188110 }, { "epoch": 119.89802421924793, "grad_norm": 0.2273612767457962, "learning_rate": 3.344272634608121e-05, "loss": 0.0044, "step": 188120 }, { "epoch": 119.90439770554494, "grad_norm": 0.5112003684043884, "learning_rate": 3.343752584476237e-05, "loss": 0.0106, "step": 188130 }, { "epoch": 119.91077119184193, "grad_norm": 0.22517727315425873, "learning_rate": 3.3432325544693234e-05, "loss": 0.0103, "step": 188140 }, { "epoch": 119.91714467813894, "grad_norm": 0.2774238884449005, "learning_rate": 3.3427125445936994e-05, "loss": 0.0097, "step": 188150 }, { "epoch": 119.92351816443595, "grad_norm": 0.35551977157592773, "learning_rate": 3.342192554855682e-05, "loss": 0.013, "step": 188160 }, { "epoch": 119.92989165073296, "grad_norm": 0.23363439738750458, "learning_rate": 3.3416725852615894e-05, "loss": 0.0069, "step": 188170 }, { "epoch": 119.93626513702995, "grad_norm": 0.3116872012615204, "learning_rate": 3.341152635817741e-05, "loss": 0.012, "step": 188180 }, { "epoch": 119.94263862332696, "grad_norm": 0.21089811623096466, "learning_rate": 3.340632706530456e-05, "loss": 0.0094, "step": 188190 }, { "epoch": 119.94901210962396, "grad_norm": 0.23486478626728058, "learning_rate": 3.340112797406049e-05, "loss": 0.01, "step": 188200 }, { "epoch": 119.95538559592097, "grad_norm": 0.3332694172859192, "learning_rate": 3.339592908450838e-05, "loss": 0.0149, "step": 188210 }, { "epoch": 119.96175908221797, "grad_norm": 0.14492540061473846, "learning_rate": 3.3390730396711426e-05, "loss": 0.0096, "step": 188220 }, { "epoch": 119.96813256851497, "grad_norm": 0.2841528058052063, "learning_rate": 3.3385531910732765e-05, "loss": 0.0084, "step": 188230 }, { "epoch": 119.97450605481198, "grad_norm": 0.22337578237056732, "learning_rate": 3.338033362663557e-05, "loss": 0.0206, "step": 188240 }, { "epoch": 119.98087954110899, "grad_norm": 0.22531069815158844, "learning_rate": 3.3375135544483025e-05, "loss": 0.0177, "step": 188250 }, { "epoch": 119.98725302740598, "grad_norm": 0.2163085788488388, "learning_rate": 3.3369937664338296e-05, "loss": 0.0129, "step": 188260 }, { "epoch": 119.99362651370299, "grad_norm": 0.14921049773693085, "learning_rate": 3.3364739986264506e-05, "loss": 0.0094, "step": 188270 }, { "epoch": 120.0, "grad_norm": 0.6551525592803955, "learning_rate": 3.335954251032483e-05, "loss": 0.0098, "step": 188280 }, { "epoch": 120.00637348629701, "grad_norm": 0.16660258173942566, "learning_rate": 3.335434523658244e-05, "loss": 0.0082, "step": 188290 }, { "epoch": 120.01274697259402, "grad_norm": 0.3989761173725128, "learning_rate": 3.334914816510046e-05, "loss": 0.0143, "step": 188300 }, { "epoch": 120.01912045889101, "grad_norm": 0.275054007768631, "learning_rate": 3.334395129594205e-05, "loss": 0.0071, "step": 188310 }, { "epoch": 120.02549394518802, "grad_norm": 0.11462637037038803, "learning_rate": 3.333875462917034e-05, "loss": 0.0153, "step": 188320 }, { "epoch": 120.03186743148503, "grad_norm": 0.24614694714546204, "learning_rate": 3.333355816484853e-05, "loss": 0.0072, "step": 188330 }, { "epoch": 120.03824091778203, "grad_norm": 0.24718917906284332, "learning_rate": 3.33283619030397e-05, "loss": 0.0073, "step": 188340 }, { "epoch": 120.04461440407903, "grad_norm": 0.8104550838470459, "learning_rate": 3.3323165843807016e-05, "loss": 0.02, "step": 188350 }, { "epoch": 120.05098789037604, "grad_norm": 0.2500157058238983, "learning_rate": 3.3317969987213604e-05, "loss": 0.0191, "step": 188360 }, { "epoch": 120.05736137667304, "grad_norm": 0.4607402980327606, "learning_rate": 3.3312774333322636e-05, "loss": 0.0092, "step": 188370 }, { "epoch": 120.06373486297005, "grad_norm": 0.34678754210472107, "learning_rate": 3.330757888219719e-05, "loss": 0.0177, "step": 188380 }, { "epoch": 120.07010834926704, "grad_norm": 0.15992899239063263, "learning_rate": 3.3302383633900425e-05, "loss": 0.0094, "step": 188390 }, { "epoch": 120.07648183556405, "grad_norm": 0.5779041647911072, "learning_rate": 3.329718858849548e-05, "loss": 0.019, "step": 188400 }, { "epoch": 120.08285532186106, "grad_norm": 0.5011159181594849, "learning_rate": 3.3291993746045444e-05, "loss": 0.0134, "step": 188410 }, { "epoch": 120.08922880815807, "grad_norm": 0.3445175588130951, "learning_rate": 3.328679910661347e-05, "loss": 0.0097, "step": 188420 }, { "epoch": 120.09560229445506, "grad_norm": 0.275940865278244, "learning_rate": 3.328160467026265e-05, "loss": 0.0085, "step": 188430 }, { "epoch": 120.10197578075207, "grad_norm": 0.35533657670021057, "learning_rate": 3.3276410437056146e-05, "loss": 0.0067, "step": 188440 }, { "epoch": 120.10834926704908, "grad_norm": 0.14644405245780945, "learning_rate": 3.327121640705702e-05, "loss": 0.0048, "step": 188450 }, { "epoch": 120.11472275334609, "grad_norm": 0.16997653245925903, "learning_rate": 3.326602258032842e-05, "loss": 0.0087, "step": 188460 }, { "epoch": 120.12109623964308, "grad_norm": 0.3418639600276947, "learning_rate": 3.326082895693344e-05, "loss": 0.0105, "step": 188470 }, { "epoch": 120.12746972594009, "grad_norm": 0.27759894728660583, "learning_rate": 3.3255635536935205e-05, "loss": 0.0098, "step": 188480 }, { "epoch": 120.1338432122371, "grad_norm": 0.3056199848651886, "learning_rate": 3.325044232039679e-05, "loss": 0.0223, "step": 188490 }, { "epoch": 120.1402166985341, "grad_norm": 0.3715665936470032, "learning_rate": 3.324524930738132e-05, "loss": 0.0133, "step": 188500 }, { "epoch": 120.1465901848311, "grad_norm": 0.5071526169776917, "learning_rate": 3.32400564979519e-05, "loss": 0.008, "step": 188510 }, { "epoch": 120.1529636711281, "grad_norm": 0.2189367562532425, "learning_rate": 3.323486389217161e-05, "loss": 0.0098, "step": 188520 }, { "epoch": 120.15933715742511, "grad_norm": 0.3964100480079651, "learning_rate": 3.3229671490103555e-05, "loss": 0.0115, "step": 188530 }, { "epoch": 120.16571064372212, "grad_norm": 0.3108729422092438, "learning_rate": 3.322447929181083e-05, "loss": 0.0182, "step": 188540 }, { "epoch": 120.17208413001912, "grad_norm": 0.14961139857769012, "learning_rate": 3.3219287297356507e-05, "loss": 0.0214, "step": 188550 }, { "epoch": 120.17845761631612, "grad_norm": 0.22873151302337646, "learning_rate": 3.32140955068037e-05, "loss": 0.007, "step": 188560 }, { "epoch": 120.18483110261313, "grad_norm": 0.2813740372657776, "learning_rate": 3.320890392021547e-05, "loss": 0.0108, "step": 188570 }, { "epoch": 120.19120458891014, "grad_norm": 0.3272632956504822, "learning_rate": 3.320371253765492e-05, "loss": 0.0097, "step": 188580 }, { "epoch": 120.19757807520713, "grad_norm": 0.429794579744339, "learning_rate": 3.319852135918511e-05, "loss": 0.0056, "step": 188590 }, { "epoch": 120.20395156150414, "grad_norm": 0.1290864199399948, "learning_rate": 3.319333038486914e-05, "loss": 0.0135, "step": 188600 }, { "epoch": 120.21032504780115, "grad_norm": 0.2731611132621765, "learning_rate": 3.318813961477008e-05, "loss": 0.0113, "step": 188610 }, { "epoch": 120.21669853409816, "grad_norm": 0.2881918251514435, "learning_rate": 3.3182949048950974e-05, "loss": 0.0108, "step": 188620 }, { "epoch": 120.22307202039515, "grad_norm": 0.08280456811189651, "learning_rate": 3.317775868747492e-05, "loss": 0.0086, "step": 188630 }, { "epoch": 120.22944550669216, "grad_norm": 0.16812627017498016, "learning_rate": 3.3172568530405e-05, "loss": 0.0119, "step": 188640 }, { "epoch": 120.23581899298917, "grad_norm": 0.2431262880563736, "learning_rate": 3.3167378577804254e-05, "loss": 0.008, "step": 188650 }, { "epoch": 120.24219247928617, "grad_norm": 0.4048732817173004, "learning_rate": 3.316218882973574e-05, "loss": 0.0129, "step": 188660 }, { "epoch": 120.24856596558317, "grad_norm": 0.41510042548179626, "learning_rate": 3.315699928626252e-05, "loss": 0.0118, "step": 188670 }, { "epoch": 120.25493945188018, "grad_norm": 0.2578164339065552, "learning_rate": 3.315180994744769e-05, "loss": 0.0127, "step": 188680 }, { "epoch": 120.26131293817718, "grad_norm": 0.14519841969013214, "learning_rate": 3.314662081335427e-05, "loss": 0.0143, "step": 188690 }, { "epoch": 120.26768642447419, "grad_norm": 0.3896768391132355, "learning_rate": 3.3141431884045304e-05, "loss": 0.0089, "step": 188700 }, { "epoch": 120.27405991077119, "grad_norm": 0.6244348883628845, "learning_rate": 3.313624315958386e-05, "loss": 0.0127, "step": 188710 }, { "epoch": 120.2804333970682, "grad_norm": 0.2636599838733673, "learning_rate": 3.3131054640033e-05, "loss": 0.0065, "step": 188720 }, { "epoch": 120.2868068833652, "grad_norm": 0.2599717080593109, "learning_rate": 3.3125866325455735e-05, "loss": 0.0061, "step": 188730 }, { "epoch": 120.29318036966221, "grad_norm": 0.31164905428886414, "learning_rate": 3.312067821591513e-05, "loss": 0.0109, "step": 188740 }, { "epoch": 120.2995538559592, "grad_norm": 0.255028635263443, "learning_rate": 3.311549031147422e-05, "loss": 0.0073, "step": 188750 }, { "epoch": 120.30592734225621, "grad_norm": 0.267743319272995, "learning_rate": 3.3110302612196064e-05, "loss": 0.0133, "step": 188760 }, { "epoch": 120.31230082855322, "grad_norm": 0.18918585777282715, "learning_rate": 3.310511511814365e-05, "loss": 0.009, "step": 188770 }, { "epoch": 120.31867431485023, "grad_norm": 0.4594974219799042, "learning_rate": 3.3099927829380054e-05, "loss": 0.0108, "step": 188780 }, { "epoch": 120.32504780114722, "grad_norm": 0.3726285696029663, "learning_rate": 3.30947407459683e-05, "loss": 0.0181, "step": 188790 }, { "epoch": 120.33142128744423, "grad_norm": 0.4048573076725006, "learning_rate": 3.308955386797139e-05, "loss": 0.0077, "step": 188800 }, { "epoch": 120.33779477374124, "grad_norm": 0.20446771383285522, "learning_rate": 3.308436719545237e-05, "loss": 0.0098, "step": 188810 }, { "epoch": 120.34416826003824, "grad_norm": 0.30722278356552124, "learning_rate": 3.307918072847425e-05, "loss": 0.0083, "step": 188820 }, { "epoch": 120.35054174633524, "grad_norm": 0.30671000480651855, "learning_rate": 3.307399446710009e-05, "loss": 0.0108, "step": 188830 }, { "epoch": 120.35691523263225, "grad_norm": 0.18679538369178772, "learning_rate": 3.306880841139286e-05, "loss": 0.0065, "step": 188840 }, { "epoch": 120.36328871892925, "grad_norm": 0.37540921568870544, "learning_rate": 3.3063622561415585e-05, "loss": 0.0122, "step": 188850 }, { "epoch": 120.36966220522626, "grad_norm": 0.12350878864526749, "learning_rate": 3.3058436917231304e-05, "loss": 0.0072, "step": 188860 }, { "epoch": 120.37603569152326, "grad_norm": 0.5308313369750977, "learning_rate": 3.3053251478903e-05, "loss": 0.0215, "step": 188870 }, { "epoch": 120.38240917782026, "grad_norm": 0.17996403574943542, "learning_rate": 3.304806624649368e-05, "loss": 0.0101, "step": 188880 }, { "epoch": 120.38878266411727, "grad_norm": 0.1650937795639038, "learning_rate": 3.3042881220066366e-05, "loss": 0.0087, "step": 188890 }, { "epoch": 120.39515615041428, "grad_norm": 0.22438663244247437, "learning_rate": 3.303769639968408e-05, "loss": 0.0073, "step": 188900 }, { "epoch": 120.40152963671127, "grad_norm": 0.2626059949398041, "learning_rate": 3.303251178540978e-05, "loss": 0.0078, "step": 188910 }, { "epoch": 120.40790312300828, "grad_norm": 0.4496532380580902, "learning_rate": 3.302732737730647e-05, "loss": 0.0121, "step": 188920 }, { "epoch": 120.41427660930529, "grad_norm": 0.5225582718849182, "learning_rate": 3.302214317543716e-05, "loss": 0.0132, "step": 188930 }, { "epoch": 120.4206500956023, "grad_norm": 0.3899118900299072, "learning_rate": 3.301695917986487e-05, "loss": 0.0182, "step": 188940 }, { "epoch": 120.4270235818993, "grad_norm": 0.31477466225624084, "learning_rate": 3.301177539065253e-05, "loss": 0.0085, "step": 188950 }, { "epoch": 120.4333970681963, "grad_norm": 0.43945494294166565, "learning_rate": 3.300659180786316e-05, "loss": 0.0121, "step": 188960 }, { "epoch": 120.43977055449331, "grad_norm": 0.15269364416599274, "learning_rate": 3.3001408431559764e-05, "loss": 0.0071, "step": 188970 }, { "epoch": 120.44614404079032, "grad_norm": 0.20248101651668549, "learning_rate": 3.299622526180529e-05, "loss": 0.0132, "step": 188980 }, { "epoch": 120.45251752708732, "grad_norm": 0.3558047413825989, "learning_rate": 3.299104229866273e-05, "loss": 0.0086, "step": 188990 }, { "epoch": 120.45889101338432, "grad_norm": 0.19397175312042236, "learning_rate": 3.2985859542195066e-05, "loss": 0.0098, "step": 189000 }, { "epoch": 120.46526449968133, "grad_norm": 0.29003316164016724, "learning_rate": 3.2980676992465294e-05, "loss": 0.0083, "step": 189010 }, { "epoch": 120.47163798597833, "grad_norm": 0.16612893342971802, "learning_rate": 3.297549464953634e-05, "loss": 0.0067, "step": 189020 }, { "epoch": 120.47801147227534, "grad_norm": 0.27966901659965515, "learning_rate": 3.29703125134712e-05, "loss": 0.012, "step": 189030 }, { "epoch": 120.48438495857233, "grad_norm": 0.3138236701488495, "learning_rate": 3.296513058433287e-05, "loss": 0.0115, "step": 189040 }, { "epoch": 120.49075844486934, "grad_norm": 0.5053485035896301, "learning_rate": 3.295994886218427e-05, "loss": 0.0096, "step": 189050 }, { "epoch": 120.49713193116635, "grad_norm": 0.33276039361953735, "learning_rate": 3.295476734708837e-05, "loss": 0.0057, "step": 189060 }, { "epoch": 120.50350541746336, "grad_norm": 0.24217873811721802, "learning_rate": 3.294958603910814e-05, "loss": 0.0126, "step": 189070 }, { "epoch": 120.50987890376035, "grad_norm": 0.10726860910654068, "learning_rate": 3.2944404938306555e-05, "loss": 0.0054, "step": 189080 }, { "epoch": 120.51625239005736, "grad_norm": 0.8169994354248047, "learning_rate": 3.293922404474654e-05, "loss": 0.0133, "step": 189090 }, { "epoch": 120.52262587635437, "grad_norm": 0.24763253331184387, "learning_rate": 3.293404335849106e-05, "loss": 0.0095, "step": 189100 }, { "epoch": 120.52899936265138, "grad_norm": 0.4284667670726776, "learning_rate": 3.2928862879603084e-05, "loss": 0.0073, "step": 189110 }, { "epoch": 120.53537284894837, "grad_norm": 0.23058278858661652, "learning_rate": 3.292368260814552e-05, "loss": 0.0078, "step": 189120 }, { "epoch": 120.54174633524538, "grad_norm": 0.3078266680240631, "learning_rate": 3.291850254418134e-05, "loss": 0.0079, "step": 189130 }, { "epoch": 120.54811982154239, "grad_norm": 0.09405229985713959, "learning_rate": 3.2913322687773476e-05, "loss": 0.0109, "step": 189140 }, { "epoch": 120.5544933078394, "grad_norm": 0.2232741415500641, "learning_rate": 3.29081430389849e-05, "loss": 0.0115, "step": 189150 }, { "epoch": 120.56086679413639, "grad_norm": 0.31511273980140686, "learning_rate": 3.29029635978785e-05, "loss": 0.006, "step": 189160 }, { "epoch": 120.5672402804334, "grad_norm": 0.5032060742378235, "learning_rate": 3.289778436451724e-05, "loss": 0.008, "step": 189170 }, { "epoch": 120.5736137667304, "grad_norm": 0.2979862689971924, "learning_rate": 3.2892605338964043e-05, "loss": 0.008, "step": 189180 }, { "epoch": 120.57998725302741, "grad_norm": 0.45467257499694824, "learning_rate": 3.288742652128186e-05, "loss": 0.0092, "step": 189190 }, { "epoch": 120.5863607393244, "grad_norm": 0.14605708420276642, "learning_rate": 3.288224791153358e-05, "loss": 0.0068, "step": 189200 }, { "epoch": 120.59273422562141, "grad_norm": 0.2454758733510971, "learning_rate": 3.287706950978216e-05, "loss": 0.0111, "step": 189210 }, { "epoch": 120.59910771191842, "grad_norm": 0.17699694633483887, "learning_rate": 3.287189131609051e-05, "loss": 0.0107, "step": 189220 }, { "epoch": 120.60548119821543, "grad_norm": 0.2575676739215851, "learning_rate": 3.286671333052156e-05, "loss": 0.0124, "step": 189230 }, { "epoch": 120.61185468451242, "grad_norm": 0.215260848402977, "learning_rate": 3.286153555313821e-05, "loss": 0.0177, "step": 189240 }, { "epoch": 120.61822817080943, "grad_norm": 0.3105042278766632, "learning_rate": 3.285635798400338e-05, "loss": 0.0107, "step": 189250 }, { "epoch": 120.62460165710644, "grad_norm": 0.3478354513645172, "learning_rate": 3.285118062318e-05, "loss": 0.0089, "step": 189260 }, { "epoch": 120.63097514340345, "grad_norm": 0.42088383436203003, "learning_rate": 3.284600347073097e-05, "loss": 0.0112, "step": 189270 }, { "epoch": 120.63734862970044, "grad_norm": 0.1896929293870926, "learning_rate": 3.2840826526719175e-05, "loss": 0.0137, "step": 189280 }, { "epoch": 120.64372211599745, "grad_norm": 0.495261013507843, "learning_rate": 3.2835649791207555e-05, "loss": 0.0094, "step": 189290 }, { "epoch": 120.65009560229446, "grad_norm": 0.4366725981235504, "learning_rate": 3.283047326425898e-05, "loss": 0.0186, "step": 189300 }, { "epoch": 120.65646908859146, "grad_norm": 0.16328462958335876, "learning_rate": 3.282529694593638e-05, "loss": 0.0133, "step": 189310 }, { "epoch": 120.66284257488846, "grad_norm": 0.19392052292823792, "learning_rate": 3.2820120836302635e-05, "loss": 0.0141, "step": 189320 }, { "epoch": 120.66921606118547, "grad_norm": 0.583411455154419, "learning_rate": 3.281494493542064e-05, "loss": 0.0167, "step": 189330 }, { "epoch": 120.67558954748247, "grad_norm": 0.21840845048427582, "learning_rate": 3.2809769243353285e-05, "loss": 0.0149, "step": 189340 }, { "epoch": 120.68196303377948, "grad_norm": 0.32084140181541443, "learning_rate": 3.280459376016347e-05, "loss": 0.0075, "step": 189350 }, { "epoch": 120.68833652007648, "grad_norm": 0.5047751069068909, "learning_rate": 3.279941848591408e-05, "loss": 0.0056, "step": 189360 }, { "epoch": 120.69471000637348, "grad_norm": 0.47803550958633423, "learning_rate": 3.2794243420667986e-05, "loss": 0.0168, "step": 189370 }, { "epoch": 120.70108349267049, "grad_norm": 0.13995715975761414, "learning_rate": 3.278906856448808e-05, "loss": 0.0056, "step": 189380 }, { "epoch": 120.7074569789675, "grad_norm": 0.19488872587680817, "learning_rate": 3.2783893917437246e-05, "loss": 0.0096, "step": 189390 }, { "epoch": 120.7138304652645, "grad_norm": 0.2848728597164154, "learning_rate": 3.2778719479578356e-05, "loss": 0.01, "step": 189400 }, { "epoch": 120.7202039515615, "grad_norm": 0.22702884674072266, "learning_rate": 3.277354525097428e-05, "loss": 0.008, "step": 189410 }, { "epoch": 120.72657743785851, "grad_norm": 0.15193040668964386, "learning_rate": 3.276837123168789e-05, "loss": 0.009, "step": 189420 }, { "epoch": 120.73295092415552, "grad_norm": 0.2551667392253876, "learning_rate": 3.276319742178208e-05, "loss": 0.0056, "step": 189430 }, { "epoch": 120.73932441045251, "grad_norm": 0.11099129170179367, "learning_rate": 3.275802382131967e-05, "loss": 0.0058, "step": 189440 }, { "epoch": 120.74569789674952, "grad_norm": 0.22366078197956085, "learning_rate": 3.2752850430363555e-05, "loss": 0.0082, "step": 189450 }, { "epoch": 120.75207138304653, "grad_norm": 0.4199567437171936, "learning_rate": 3.2747677248976595e-05, "loss": 0.0074, "step": 189460 }, { "epoch": 120.75844486934353, "grad_norm": 0.18794773519039154, "learning_rate": 3.274250427722165e-05, "loss": 0.0063, "step": 189470 }, { "epoch": 120.76481835564053, "grad_norm": 0.30803021788597107, "learning_rate": 3.273733151516157e-05, "loss": 0.0082, "step": 189480 }, { "epoch": 120.77119184193754, "grad_norm": 0.3313135802745819, "learning_rate": 3.2732158962859204e-05, "loss": 0.0117, "step": 189490 }, { "epoch": 120.77756532823454, "grad_norm": 0.3008878827095032, "learning_rate": 3.272698662037743e-05, "loss": 0.0087, "step": 189500 }, { "epoch": 120.78393881453155, "grad_norm": 0.4917187988758087, "learning_rate": 3.2721814487779055e-05, "loss": 0.0103, "step": 189510 }, { "epoch": 120.79031230082855, "grad_norm": 0.17467522621154785, "learning_rate": 3.2716642565126944e-05, "loss": 0.0056, "step": 189520 }, { "epoch": 120.79668578712555, "grad_norm": 0.5044125318527222, "learning_rate": 3.271147085248395e-05, "loss": 0.0091, "step": 189530 }, { "epoch": 120.80305927342256, "grad_norm": 0.20918110013008118, "learning_rate": 3.270629934991293e-05, "loss": 0.0085, "step": 189540 }, { "epoch": 120.80943275971957, "grad_norm": 0.22176536917686462, "learning_rate": 3.2701128057476675e-05, "loss": 0.0091, "step": 189550 }, { "epoch": 120.81580624601658, "grad_norm": 0.47140610218048096, "learning_rate": 3.269595697523805e-05, "loss": 0.0108, "step": 189560 }, { "epoch": 120.82217973231357, "grad_norm": 0.24336494505405426, "learning_rate": 3.2690786103259886e-05, "loss": 0.0058, "step": 189570 }, { "epoch": 120.82855321861058, "grad_norm": 0.2160656750202179, "learning_rate": 3.268561544160503e-05, "loss": 0.0102, "step": 189580 }, { "epoch": 120.83492670490759, "grad_norm": 0.17127110064029694, "learning_rate": 3.268044499033628e-05, "loss": 0.0067, "step": 189590 }, { "epoch": 120.8413001912046, "grad_norm": 0.20308060944080353, "learning_rate": 3.2675274749516474e-05, "loss": 0.01, "step": 189600 }, { "epoch": 120.84767367750159, "grad_norm": 0.4036678969860077, "learning_rate": 3.267010471920845e-05, "loss": 0.0111, "step": 189610 }, { "epoch": 120.8540471637986, "grad_norm": 0.19663740694522858, "learning_rate": 3.2664934899475006e-05, "loss": 0.0156, "step": 189620 }, { "epoch": 120.8604206500956, "grad_norm": 0.17143882811069489, "learning_rate": 3.265976529037896e-05, "loss": 0.0142, "step": 189630 }, { "epoch": 120.86679413639261, "grad_norm": 0.3473621904850006, "learning_rate": 3.265459589198314e-05, "loss": 0.0078, "step": 189640 }, { "epoch": 120.8731676226896, "grad_norm": 0.18540671467781067, "learning_rate": 3.264942670435038e-05, "loss": 0.0141, "step": 189650 }, { "epoch": 120.87954110898661, "grad_norm": 0.5740290284156799, "learning_rate": 3.264425772754345e-05, "loss": 0.008, "step": 189660 }, { "epoch": 120.88591459528362, "grad_norm": 0.3977932035923004, "learning_rate": 3.2639088961625166e-05, "loss": 0.0065, "step": 189670 }, { "epoch": 120.89228808158063, "grad_norm": 0.2807222306728363, "learning_rate": 3.263392040665838e-05, "loss": 0.0069, "step": 189680 }, { "epoch": 120.89866156787762, "grad_norm": 0.32095250487327576, "learning_rate": 3.262875206270582e-05, "loss": 0.0121, "step": 189690 }, { "epoch": 120.90503505417463, "grad_norm": 0.4541400969028473, "learning_rate": 3.262358392983033e-05, "loss": 0.0098, "step": 189700 }, { "epoch": 120.91140854047164, "grad_norm": 0.578545093536377, "learning_rate": 3.26184160080947e-05, "loss": 0.0139, "step": 189710 }, { "epoch": 120.91778202676865, "grad_norm": 0.27426379919052124, "learning_rate": 3.261324829756175e-05, "loss": 0.0108, "step": 189720 }, { "epoch": 120.92415551306564, "grad_norm": 0.3596566915512085, "learning_rate": 3.260808079829422e-05, "loss": 0.0097, "step": 189730 }, { "epoch": 120.93052899936265, "grad_norm": 0.12744906544685364, "learning_rate": 3.2602913510354937e-05, "loss": 0.0202, "step": 189740 }, { "epoch": 120.93690248565966, "grad_norm": 0.2818721532821655, "learning_rate": 3.25977464338067e-05, "loss": 0.0084, "step": 189750 }, { "epoch": 120.94327597195667, "grad_norm": 0.68265300989151, "learning_rate": 3.259257956871225e-05, "loss": 0.0086, "step": 189760 }, { "epoch": 120.94964945825366, "grad_norm": 0.2996887266635895, "learning_rate": 3.2587412915134395e-05, "loss": 0.0146, "step": 189770 }, { "epoch": 120.95602294455067, "grad_norm": 0.3648566007614136, "learning_rate": 3.2582246473135915e-05, "loss": 0.0093, "step": 189780 }, { "epoch": 120.96239643084768, "grad_norm": 0.2279108166694641, "learning_rate": 3.257708024277961e-05, "loss": 0.0054, "step": 189790 }, { "epoch": 120.96876991714468, "grad_norm": 0.22171273827552795, "learning_rate": 3.2571914224128195e-05, "loss": 0.0072, "step": 189800 }, { "epoch": 120.97514340344168, "grad_norm": 0.2027374655008316, "learning_rate": 3.256674841724449e-05, "loss": 0.0071, "step": 189810 }, { "epoch": 120.98151688973869, "grad_norm": 0.24963192641735077, "learning_rate": 3.2561582822191274e-05, "loss": 0.008, "step": 189820 }, { "epoch": 120.9878903760357, "grad_norm": 0.19162005186080933, "learning_rate": 3.255641743903126e-05, "loss": 0.0085, "step": 189830 }, { "epoch": 120.9942638623327, "grad_norm": 0.38558250665664673, "learning_rate": 3.255125226782725e-05, "loss": 0.0085, "step": 189840 }, { "epoch": 121.0006373486297, "grad_norm": 0.1895996332168579, "learning_rate": 3.2546087308641995e-05, "loss": 0.0105, "step": 189850 }, { "epoch": 121.0070108349267, "grad_norm": 0.11493214219808578, "learning_rate": 3.254092256153829e-05, "loss": 0.0156, "step": 189860 }, { "epoch": 121.01338432122371, "grad_norm": 0.2745555639266968, "learning_rate": 3.253575802657883e-05, "loss": 0.0079, "step": 189870 }, { "epoch": 121.01975780752072, "grad_norm": 0.2919732332229614, "learning_rate": 3.253059370382639e-05, "loss": 0.014, "step": 189880 }, { "epoch": 121.02613129381771, "grad_norm": 0.45334863662719727, "learning_rate": 3.2525429593343745e-05, "loss": 0.0186, "step": 189890 }, { "epoch": 121.03250478011472, "grad_norm": 0.2029789835214615, "learning_rate": 3.252026569519365e-05, "loss": 0.0078, "step": 189900 }, { "epoch": 121.03887826641173, "grad_norm": 0.31936517357826233, "learning_rate": 3.25151020094388e-05, "loss": 0.0108, "step": 189910 }, { "epoch": 121.04525175270874, "grad_norm": 0.426304429769516, "learning_rate": 3.250993853614198e-05, "loss": 0.0111, "step": 189920 }, { "epoch": 121.05162523900573, "grad_norm": 0.21515685319900513, "learning_rate": 3.250477527536593e-05, "loss": 0.0072, "step": 189930 }, { "epoch": 121.05799872530274, "grad_norm": 0.5360755920410156, "learning_rate": 3.249961222717337e-05, "loss": 0.0157, "step": 189940 }, { "epoch": 121.06437221159975, "grad_norm": 0.23490971326828003, "learning_rate": 3.249444939162704e-05, "loss": 0.0083, "step": 189950 }, { "epoch": 121.07074569789675, "grad_norm": 0.2921140491962433, "learning_rate": 3.248928676878967e-05, "loss": 0.0082, "step": 189960 }, { "epoch": 121.07711918419375, "grad_norm": 0.11394108831882477, "learning_rate": 3.2484124358724025e-05, "loss": 0.0083, "step": 189970 }, { "epoch": 121.08349267049076, "grad_norm": 0.3120347857475281, "learning_rate": 3.247896216149279e-05, "loss": 0.0248, "step": 189980 }, { "epoch": 121.08986615678776, "grad_norm": 0.31823426485061646, "learning_rate": 3.2473800177158703e-05, "loss": 0.0138, "step": 189990 }, { "epoch": 121.09623964308477, "grad_norm": 0.4938390254974365, "learning_rate": 3.24686384057845e-05, "loss": 0.0085, "step": 190000 }, { "epoch": 121.10261312938177, "grad_norm": 0.20276954770088196, "learning_rate": 3.246347684743288e-05, "loss": 0.0089, "step": 190010 }, { "epoch": 121.10898661567877, "grad_norm": 0.167133167386055, "learning_rate": 3.2458315502166595e-05, "loss": 0.0091, "step": 190020 }, { "epoch": 121.11536010197578, "grad_norm": 0.13486017286777496, "learning_rate": 3.245315437004832e-05, "loss": 0.007, "step": 190030 }, { "epoch": 121.12173358827279, "grad_norm": 0.14882870018482208, "learning_rate": 3.24479934511408e-05, "loss": 0.0156, "step": 190040 }, { "epoch": 121.12810707456978, "grad_norm": 0.16338451206684113, "learning_rate": 3.2442832745506726e-05, "loss": 0.0083, "step": 190050 }, { "epoch": 121.13448056086679, "grad_norm": 0.17933426797389984, "learning_rate": 3.24376722532088e-05, "loss": 0.0185, "step": 190060 }, { "epoch": 121.1408540471638, "grad_norm": 0.14097245037555695, "learning_rate": 3.2432511974309766e-05, "loss": 0.0096, "step": 190070 }, { "epoch": 121.1472275334608, "grad_norm": 0.8473345041275024, "learning_rate": 3.2427351908872275e-05, "loss": 0.0088, "step": 190080 }, { "epoch": 121.1536010197578, "grad_norm": 0.2407538890838623, "learning_rate": 3.242219205695904e-05, "loss": 0.0073, "step": 190090 }, { "epoch": 121.15997450605481, "grad_norm": 0.16800837218761444, "learning_rate": 3.2417032418632786e-05, "loss": 0.0083, "step": 190100 }, { "epoch": 121.16634799235182, "grad_norm": 0.5004050135612488, "learning_rate": 3.2411872993956196e-05, "loss": 0.015, "step": 190110 }, { "epoch": 121.17272147864882, "grad_norm": 0.2521511912345886, "learning_rate": 3.2406713782991935e-05, "loss": 0.0149, "step": 190120 }, { "epoch": 121.17909496494582, "grad_norm": 0.1356671303510666, "learning_rate": 3.240155478580272e-05, "loss": 0.0064, "step": 190130 }, { "epoch": 121.18546845124283, "grad_norm": 0.2224314957857132, "learning_rate": 3.2396396002451235e-05, "loss": 0.0057, "step": 190140 }, { "epoch": 121.19184193753983, "grad_norm": 0.23800532519817352, "learning_rate": 3.239123743300016e-05, "loss": 0.0086, "step": 190150 }, { "epoch": 121.19821542383684, "grad_norm": 0.33634305000305176, "learning_rate": 3.238607907751217e-05, "loss": 0.0066, "step": 190160 }, { "epoch": 121.20458891013385, "grad_norm": 0.13611754775047302, "learning_rate": 3.238092093604994e-05, "loss": 0.0066, "step": 190170 }, { "epoch": 121.21096239643084, "grad_norm": 0.31762099266052246, "learning_rate": 3.237576300867619e-05, "loss": 0.0168, "step": 190180 }, { "epoch": 121.21733588272785, "grad_norm": 0.2731436789035797, "learning_rate": 3.237060529545353e-05, "loss": 0.0133, "step": 190190 }, { "epoch": 121.22370936902486, "grad_norm": 0.3460673391819, "learning_rate": 3.236544779644466e-05, "loss": 0.0145, "step": 190200 }, { "epoch": 121.23008285532187, "grad_norm": 0.25570452213287354, "learning_rate": 3.236029051171224e-05, "loss": 0.0094, "step": 190210 }, { "epoch": 121.23645634161886, "grad_norm": 0.47507309913635254, "learning_rate": 3.235513344131899e-05, "loss": 0.0111, "step": 190220 }, { "epoch": 121.24282982791587, "grad_norm": 0.49623173475265503, "learning_rate": 3.234997658532748e-05, "loss": 0.0098, "step": 190230 }, { "epoch": 121.24920331421288, "grad_norm": 0.39205530285835266, "learning_rate": 3.234481994380044e-05, "loss": 0.0115, "step": 190240 }, { "epoch": 121.25557680050989, "grad_norm": 0.19014345109462738, "learning_rate": 3.2339663516800514e-05, "loss": 0.0071, "step": 190250 }, { "epoch": 121.26195028680688, "grad_norm": 0.3061116933822632, "learning_rate": 3.2334507304390335e-05, "loss": 0.0068, "step": 190260 }, { "epoch": 121.26832377310389, "grad_norm": 0.4722488522529602, "learning_rate": 3.2329351306632576e-05, "loss": 0.0103, "step": 190270 }, { "epoch": 121.2746972594009, "grad_norm": 0.7860873341560364, "learning_rate": 3.232419552358988e-05, "loss": 0.0131, "step": 190280 }, { "epoch": 121.2810707456979, "grad_norm": 0.3792169392108917, "learning_rate": 3.231903995532492e-05, "loss": 0.0073, "step": 190290 }, { "epoch": 121.2874442319949, "grad_norm": 0.3496931493282318, "learning_rate": 3.231388460190029e-05, "loss": 0.0091, "step": 190300 }, { "epoch": 121.2938177182919, "grad_norm": 0.22113098204135895, "learning_rate": 3.230872946337866e-05, "loss": 0.0109, "step": 190310 }, { "epoch": 121.30019120458891, "grad_norm": 0.3632507026195526, "learning_rate": 3.23035745398227e-05, "loss": 0.0131, "step": 190320 }, { "epoch": 121.30656469088592, "grad_norm": 0.37488695979118347, "learning_rate": 3.229841983129499e-05, "loss": 0.0067, "step": 190330 }, { "epoch": 121.31293817718291, "grad_norm": 0.2551261782646179, "learning_rate": 3.229326533785819e-05, "loss": 0.0043, "step": 190340 }, { "epoch": 121.31931166347992, "grad_norm": 0.16720423102378845, "learning_rate": 3.228811105957494e-05, "loss": 0.0052, "step": 190350 }, { "epoch": 121.32568514977693, "grad_norm": 0.10799352824687958, "learning_rate": 3.2282956996507885e-05, "loss": 0.0075, "step": 190360 }, { "epoch": 121.33205863607394, "grad_norm": 0.2300741821527481, "learning_rate": 3.227780314871961e-05, "loss": 0.0082, "step": 190370 }, { "epoch": 121.33843212237093, "grad_norm": 0.2057197540998459, "learning_rate": 3.227264951627275e-05, "loss": 0.0052, "step": 190380 }, { "epoch": 121.34480560866794, "grad_norm": 0.48943790793418884, "learning_rate": 3.226749609922997e-05, "loss": 0.0122, "step": 190390 }, { "epoch": 121.35117909496495, "grad_norm": 0.14015232026576996, "learning_rate": 3.226234289765382e-05, "loss": 0.0078, "step": 190400 }, { "epoch": 121.35755258126196, "grad_norm": 0.3022996485233307, "learning_rate": 3.2257189911606955e-05, "loss": 0.0078, "step": 190410 }, { "epoch": 121.36392606755895, "grad_norm": 0.36741816997528076, "learning_rate": 3.225203714115199e-05, "loss": 0.0097, "step": 190420 }, { "epoch": 121.37029955385596, "grad_norm": 0.9161378145217896, "learning_rate": 3.224688458635155e-05, "loss": 0.0147, "step": 190430 }, { "epoch": 121.37667304015297, "grad_norm": 0.09896966814994812, "learning_rate": 3.22417322472682e-05, "loss": 0.0095, "step": 190440 }, { "epoch": 121.38304652644997, "grad_norm": 0.3202853500843048, "learning_rate": 3.2236580123964574e-05, "loss": 0.0138, "step": 190450 }, { "epoch": 121.38942001274697, "grad_norm": 0.302489697933197, "learning_rate": 3.223142821650327e-05, "loss": 0.0119, "step": 190460 }, { "epoch": 121.39579349904398, "grad_norm": 0.30568182468414307, "learning_rate": 3.222627652494691e-05, "loss": 0.008, "step": 190470 }, { "epoch": 121.40216698534098, "grad_norm": 0.45658907294273376, "learning_rate": 3.222112504935805e-05, "loss": 0.0123, "step": 190480 }, { "epoch": 121.40854047163799, "grad_norm": 0.3091672658920288, "learning_rate": 3.22159737897993e-05, "loss": 0.006, "step": 190490 }, { "epoch": 121.41491395793498, "grad_norm": 0.5199173092842102, "learning_rate": 3.2210822746333286e-05, "loss": 0.0121, "step": 190500 }, { "epoch": 121.421287444232, "grad_norm": 0.37445497512817383, "learning_rate": 3.220567191902255e-05, "loss": 0.0144, "step": 190510 }, { "epoch": 121.427660930529, "grad_norm": 0.34181442856788635, "learning_rate": 3.2200521307929695e-05, "loss": 0.014, "step": 190520 }, { "epoch": 121.43403441682601, "grad_norm": 0.24285399913787842, "learning_rate": 3.219537091311733e-05, "loss": 0.0086, "step": 190530 }, { "epoch": 121.440407903123, "grad_norm": 0.18420186638832092, "learning_rate": 3.219022073464802e-05, "loss": 0.0091, "step": 190540 }, { "epoch": 121.44678138942001, "grad_norm": 0.18804460763931274, "learning_rate": 3.218507077258434e-05, "loss": 0.0097, "step": 190550 }, { "epoch": 121.45315487571702, "grad_norm": 0.6118188500404358, "learning_rate": 3.217992102698885e-05, "loss": 0.0123, "step": 190560 }, { "epoch": 121.45952836201403, "grad_norm": 0.253313809633255, "learning_rate": 3.217477149792418e-05, "loss": 0.0132, "step": 190570 }, { "epoch": 121.46590184831102, "grad_norm": 0.2540372908115387, "learning_rate": 3.216962218545284e-05, "loss": 0.0073, "step": 190580 }, { "epoch": 121.47227533460803, "grad_norm": 0.31723904609680176, "learning_rate": 3.2164473089637424e-05, "loss": 0.0149, "step": 190590 }, { "epoch": 121.47864882090504, "grad_norm": 0.20547857880592346, "learning_rate": 3.21593242105405e-05, "loss": 0.0119, "step": 190600 }, { "epoch": 121.48502230720204, "grad_norm": 0.17479686439037323, "learning_rate": 3.215417554822465e-05, "loss": 0.0088, "step": 190610 }, { "epoch": 121.49139579349904, "grad_norm": 0.2779425382614136, "learning_rate": 3.21490271027524e-05, "loss": 0.0114, "step": 190620 }, { "epoch": 121.49776927979605, "grad_norm": 0.5749782919883728, "learning_rate": 3.2143878874186325e-05, "loss": 0.0142, "step": 190630 }, { "epoch": 121.50414276609305, "grad_norm": 0.2483048439025879, "learning_rate": 3.213873086258899e-05, "loss": 0.0089, "step": 190640 }, { "epoch": 121.51051625239006, "grad_norm": 0.14831894636154175, "learning_rate": 3.2133583068022946e-05, "loss": 0.0087, "step": 190650 }, { "epoch": 121.51688973868706, "grad_norm": 0.22823864221572876, "learning_rate": 3.212843549055071e-05, "loss": 0.0093, "step": 190660 }, { "epoch": 121.52326322498406, "grad_norm": 0.2387973964214325, "learning_rate": 3.212328813023487e-05, "loss": 0.0165, "step": 190670 }, { "epoch": 121.52963671128107, "grad_norm": 0.23259121179580688, "learning_rate": 3.211814098713796e-05, "loss": 0.0049, "step": 190680 }, { "epoch": 121.53601019757808, "grad_norm": 0.3874776363372803, "learning_rate": 3.211299406132253e-05, "loss": 0.0102, "step": 190690 }, { "epoch": 121.54238368387507, "grad_norm": 0.436784029006958, "learning_rate": 3.210784735285109e-05, "loss": 0.0083, "step": 190700 }, { "epoch": 121.54875717017208, "grad_norm": 0.2504998445510864, "learning_rate": 3.210270086178621e-05, "loss": 0.0071, "step": 190710 }, { "epoch": 121.55513065646909, "grad_norm": 0.12307441979646683, "learning_rate": 3.20975545881904e-05, "loss": 0.0154, "step": 190720 }, { "epoch": 121.5615041427661, "grad_norm": 0.19603724777698517, "learning_rate": 3.209240853212622e-05, "loss": 0.0083, "step": 190730 }, { "epoch": 121.56787762906309, "grad_norm": 0.3852204382419586, "learning_rate": 3.2087262693656175e-05, "loss": 0.0101, "step": 190740 }, { "epoch": 121.5742511153601, "grad_norm": 0.389456182718277, "learning_rate": 3.208211707284281e-05, "loss": 0.0122, "step": 190750 }, { "epoch": 121.5806246016571, "grad_norm": 0.3820416331291199, "learning_rate": 3.2076971669748625e-05, "loss": 0.0073, "step": 190760 }, { "epoch": 121.58699808795411, "grad_norm": 0.33429497480392456, "learning_rate": 3.2071826484436165e-05, "loss": 0.0076, "step": 190770 }, { "epoch": 121.59337157425111, "grad_norm": 0.217999666929245, "learning_rate": 3.2066681516967935e-05, "loss": 0.0124, "step": 190780 }, { "epoch": 121.59974506054812, "grad_norm": 0.42937639355659485, "learning_rate": 3.2061536767406466e-05, "loss": 0.0148, "step": 190790 }, { "epoch": 121.60611854684512, "grad_norm": 0.30357274413108826, "learning_rate": 3.2056392235814255e-05, "loss": 0.0089, "step": 190800 }, { "epoch": 121.61249203314213, "grad_norm": 0.4208036959171295, "learning_rate": 3.205124792225382e-05, "loss": 0.0173, "step": 190810 }, { "epoch": 121.61886551943914, "grad_norm": 0.19151806831359863, "learning_rate": 3.2046103826787674e-05, "loss": 0.0076, "step": 190820 }, { "epoch": 121.62523900573613, "grad_norm": 0.12502740323543549, "learning_rate": 3.204095994947831e-05, "loss": 0.0062, "step": 190830 }, { "epoch": 121.63161249203314, "grad_norm": 0.6558022499084473, "learning_rate": 3.2035816290388244e-05, "loss": 0.0159, "step": 190840 }, { "epoch": 121.63798597833015, "grad_norm": 0.22881723940372467, "learning_rate": 3.203067284957997e-05, "loss": 0.0068, "step": 190850 }, { "epoch": 121.64435946462716, "grad_norm": 0.22540636360645294, "learning_rate": 3.2025529627115994e-05, "loss": 0.0125, "step": 190860 }, { "epoch": 121.65073295092415, "grad_norm": 0.3552665114402771, "learning_rate": 3.20203866230588e-05, "loss": 0.0117, "step": 190870 }, { "epoch": 121.65710643722116, "grad_norm": 0.1510961651802063, "learning_rate": 3.201524383747086e-05, "loss": 0.0122, "step": 190880 }, { "epoch": 121.66347992351817, "grad_norm": 0.24718697369098663, "learning_rate": 3.2010101270414736e-05, "loss": 0.0064, "step": 190890 }, { "epoch": 121.66985340981518, "grad_norm": 0.26390647888183594, "learning_rate": 3.200495892195283e-05, "loss": 0.0151, "step": 190900 }, { "epoch": 121.67622689611217, "grad_norm": 0.18901044130325317, "learning_rate": 3.199981679214767e-05, "loss": 0.0181, "step": 190910 }, { "epoch": 121.68260038240918, "grad_norm": 0.3208310902118683, "learning_rate": 3.1994674881061726e-05, "loss": 0.0044, "step": 190920 }, { "epoch": 121.68897386870619, "grad_norm": 0.3415158689022064, "learning_rate": 3.1989533188757504e-05, "loss": 0.0092, "step": 190930 }, { "epoch": 121.6953473550032, "grad_norm": 0.22101488709449768, "learning_rate": 3.1984391715297444e-05, "loss": 0.0081, "step": 190940 }, { "epoch": 121.70172084130019, "grad_norm": 0.13634154200553894, "learning_rate": 3.197925046074402e-05, "loss": 0.0084, "step": 190950 }, { "epoch": 121.7080943275972, "grad_norm": 0.23671482503414154, "learning_rate": 3.197410942515975e-05, "loss": 0.0071, "step": 190960 }, { "epoch": 121.7144678138942, "grad_norm": 0.380031019449234, "learning_rate": 3.1968968608607055e-05, "loss": 0.0066, "step": 190970 }, { "epoch": 121.72084130019121, "grad_norm": 0.26170721650123596, "learning_rate": 3.19638280111484e-05, "loss": 0.0106, "step": 190980 }, { "epoch": 121.7272147864882, "grad_norm": 0.26962563395500183, "learning_rate": 3.1958687632846274e-05, "loss": 0.0089, "step": 190990 }, { "epoch": 121.73358827278521, "grad_norm": 0.37805595993995667, "learning_rate": 3.195354747376314e-05, "loss": 0.0093, "step": 191000 }, { "epoch": 121.73996175908222, "grad_norm": 0.16654866933822632, "learning_rate": 3.1948407533961425e-05, "loss": 0.0063, "step": 191010 }, { "epoch": 121.74633524537923, "grad_norm": 0.2189732789993286, "learning_rate": 3.194326781350361e-05, "loss": 0.0081, "step": 191020 }, { "epoch": 121.75270873167622, "grad_norm": 0.27719733119010925, "learning_rate": 3.1938128312452146e-05, "loss": 0.0079, "step": 191030 }, { "epoch": 121.75908221797323, "grad_norm": 0.2638314366340637, "learning_rate": 3.193298903086946e-05, "loss": 0.0127, "step": 191040 }, { "epoch": 121.76545570427024, "grad_norm": 0.17376403510570526, "learning_rate": 3.192784996881803e-05, "loss": 0.0081, "step": 191050 }, { "epoch": 121.77182919056725, "grad_norm": 0.5191220045089722, "learning_rate": 3.192271112636027e-05, "loss": 0.0083, "step": 191060 }, { "epoch": 121.77820267686424, "grad_norm": 0.553070068359375, "learning_rate": 3.191757250355866e-05, "loss": 0.0144, "step": 191070 }, { "epoch": 121.78457616316125, "grad_norm": 0.21482007205486298, "learning_rate": 3.19124341004756e-05, "loss": 0.0073, "step": 191080 }, { "epoch": 121.79094964945826, "grad_norm": 0.18981888890266418, "learning_rate": 3.190729591717354e-05, "loss": 0.0083, "step": 191090 }, { "epoch": 121.79732313575526, "grad_norm": 0.33346742391586304, "learning_rate": 3.1902157953714914e-05, "loss": 0.0115, "step": 191100 }, { "epoch": 121.80369662205226, "grad_norm": 0.18084721267223358, "learning_rate": 3.189702021016219e-05, "loss": 0.0126, "step": 191110 }, { "epoch": 121.81007010834927, "grad_norm": 0.18609294295310974, "learning_rate": 3.189188268657773e-05, "loss": 0.0042, "step": 191120 }, { "epoch": 121.81644359464627, "grad_norm": 0.2490134984254837, "learning_rate": 3.1886745383024e-05, "loss": 0.0092, "step": 191130 }, { "epoch": 121.82281708094328, "grad_norm": 0.44245412945747375, "learning_rate": 3.1881608299563426e-05, "loss": 0.0127, "step": 191140 }, { "epoch": 121.82919056724027, "grad_norm": 0.35648787021636963, "learning_rate": 3.18764714362584e-05, "loss": 0.0065, "step": 191150 }, { "epoch": 121.83556405353728, "grad_norm": 0.23611457645893097, "learning_rate": 3.1871334793171356e-05, "loss": 0.0075, "step": 191160 }, { "epoch": 121.84193753983429, "grad_norm": 0.28248587250709534, "learning_rate": 3.186619837036472e-05, "loss": 0.0082, "step": 191170 }, { "epoch": 121.8483110261313, "grad_norm": 0.23177571594715118, "learning_rate": 3.186106216790091e-05, "loss": 0.0103, "step": 191180 }, { "epoch": 121.85468451242829, "grad_norm": 0.6437267065048218, "learning_rate": 3.18559261858423e-05, "loss": 0.0114, "step": 191190 }, { "epoch": 121.8610579987253, "grad_norm": 0.46833574771881104, "learning_rate": 3.185079042425132e-05, "loss": 0.0091, "step": 191200 }, { "epoch": 121.86743148502231, "grad_norm": 0.16829079389572144, "learning_rate": 3.184565488319039e-05, "loss": 0.0122, "step": 191210 }, { "epoch": 121.87380497131932, "grad_norm": 0.26821383833885193, "learning_rate": 3.184051956272187e-05, "loss": 0.0079, "step": 191220 }, { "epoch": 121.88017845761631, "grad_norm": 0.24650458991527557, "learning_rate": 3.183538446290819e-05, "loss": 0.0069, "step": 191230 }, { "epoch": 121.88655194391332, "grad_norm": 0.1399299055337906, "learning_rate": 3.183024958381173e-05, "loss": 0.0101, "step": 191240 }, { "epoch": 121.89292543021033, "grad_norm": 0.31378862261772156, "learning_rate": 3.182511492549492e-05, "loss": 0.0118, "step": 191250 }, { "epoch": 121.89929891650733, "grad_norm": 0.1516398787498474, "learning_rate": 3.181998048802011e-05, "loss": 0.0081, "step": 191260 }, { "epoch": 121.90567240280433, "grad_norm": 0.22072970867156982, "learning_rate": 3.1814846271449684e-05, "loss": 0.0083, "step": 191270 }, { "epoch": 121.91204588910134, "grad_norm": 0.22433353960514069, "learning_rate": 3.180971227584608e-05, "loss": 0.0107, "step": 191280 }, { "epoch": 121.91841937539834, "grad_norm": 0.4574781656265259, "learning_rate": 3.180457850127162e-05, "loss": 0.0066, "step": 191290 }, { "epoch": 121.92479286169535, "grad_norm": 0.10487018525600433, "learning_rate": 3.1799444947788715e-05, "loss": 0.0065, "step": 191300 }, { "epoch": 121.93116634799235, "grad_norm": 0.13583903014659882, "learning_rate": 3.1794311615459735e-05, "loss": 0.0112, "step": 191310 }, { "epoch": 121.93753983428935, "grad_norm": 0.18589630722999573, "learning_rate": 3.178917850434707e-05, "loss": 0.0093, "step": 191320 }, { "epoch": 121.94391332058636, "grad_norm": 0.18893694877624512, "learning_rate": 3.178404561451307e-05, "loss": 0.0049, "step": 191330 }, { "epoch": 121.95028680688337, "grad_norm": 0.23754443228244781, "learning_rate": 3.177891294602011e-05, "loss": 0.0099, "step": 191340 }, { "epoch": 121.95666029318036, "grad_norm": 0.3899127244949341, "learning_rate": 3.1773780498930564e-05, "loss": 0.0074, "step": 191350 }, { "epoch": 121.96303377947737, "grad_norm": 0.23206734657287598, "learning_rate": 3.176864827330682e-05, "loss": 0.0076, "step": 191360 }, { "epoch": 121.96940726577438, "grad_norm": 0.12076902389526367, "learning_rate": 3.176351626921118e-05, "loss": 0.0059, "step": 191370 }, { "epoch": 121.97578075207139, "grad_norm": 0.39656561613082886, "learning_rate": 3.1758384486706047e-05, "loss": 0.0115, "step": 191380 }, { "epoch": 121.98215423836838, "grad_norm": 0.31528592109680176, "learning_rate": 3.175325292585376e-05, "loss": 0.0074, "step": 191390 }, { "epoch": 121.98852772466539, "grad_norm": 0.1170879527926445, "learning_rate": 3.1748121586716696e-05, "loss": 0.0093, "step": 191400 }, { "epoch": 121.9949012109624, "grad_norm": 0.5670944452285767, "learning_rate": 3.174299046935716e-05, "loss": 0.0105, "step": 191410 }, { "epoch": 122.0012746972594, "grad_norm": 0.1496770977973938, "learning_rate": 3.173785957383754e-05, "loss": 0.0064, "step": 191420 }, { "epoch": 122.00764818355641, "grad_norm": 0.2418300360441208, "learning_rate": 3.173272890022018e-05, "loss": 0.0072, "step": 191430 }, { "epoch": 122.0140216698534, "grad_norm": 0.19376665353775024, "learning_rate": 3.17275984485674e-05, "loss": 0.0064, "step": 191440 }, { "epoch": 122.02039515615041, "grad_norm": 0.13113898038864136, "learning_rate": 3.172246821894155e-05, "loss": 0.0076, "step": 191450 }, { "epoch": 122.02676864244742, "grad_norm": 0.4618385136127472, "learning_rate": 3.171733821140497e-05, "loss": 0.0092, "step": 191460 }, { "epoch": 122.03314212874443, "grad_norm": 0.1928713023662567, "learning_rate": 3.1712208426019984e-05, "loss": 0.0097, "step": 191470 }, { "epoch": 122.03951561504142, "grad_norm": 0.2887822389602661, "learning_rate": 3.170707886284895e-05, "loss": 0.0172, "step": 191480 }, { "epoch": 122.04588910133843, "grad_norm": 0.445858895778656, "learning_rate": 3.1701949521954167e-05, "loss": 0.0081, "step": 191490 }, { "epoch": 122.05226258763544, "grad_norm": 0.2721354365348816, "learning_rate": 3.169682040339798e-05, "loss": 0.0075, "step": 191500 }, { "epoch": 122.05863607393245, "grad_norm": 0.3867131471633911, "learning_rate": 3.1691691507242694e-05, "loss": 0.0072, "step": 191510 }, { "epoch": 122.06500956022944, "grad_norm": 0.22803454101085663, "learning_rate": 3.1686562833550655e-05, "loss": 0.0117, "step": 191520 }, { "epoch": 122.07138304652645, "grad_norm": 0.30015984177589417, "learning_rate": 3.1681434382384165e-05, "loss": 0.0053, "step": 191530 }, { "epoch": 122.07775653282346, "grad_norm": 0.3624720573425293, "learning_rate": 3.1676306153805533e-05, "loss": 0.0119, "step": 191540 }, { "epoch": 122.08413001912047, "grad_norm": 0.24954655766487122, "learning_rate": 3.167117814787709e-05, "loss": 0.0088, "step": 191550 }, { "epoch": 122.09050350541746, "grad_norm": 0.1653566062450409, "learning_rate": 3.1666050364661136e-05, "loss": 0.0067, "step": 191560 }, { "epoch": 122.09687699171447, "grad_norm": 0.20406509935855865, "learning_rate": 3.166092280421999e-05, "loss": 0.0089, "step": 191570 }, { "epoch": 122.10325047801147, "grad_norm": 0.31978994607925415, "learning_rate": 3.165579546661593e-05, "loss": 0.0085, "step": 191580 }, { "epoch": 122.10962396430848, "grad_norm": 0.4087543487548828, "learning_rate": 3.165066835191128e-05, "loss": 0.0084, "step": 191590 }, { "epoch": 122.11599745060548, "grad_norm": 0.17339015007019043, "learning_rate": 3.164554146016836e-05, "loss": 0.0169, "step": 191600 }, { "epoch": 122.12237093690248, "grad_norm": 0.2015334814786911, "learning_rate": 3.164041479144941e-05, "loss": 0.0084, "step": 191610 }, { "epoch": 122.12874442319949, "grad_norm": 0.1409757137298584, "learning_rate": 3.163528834581676e-05, "loss": 0.0041, "step": 191620 }, { "epoch": 122.1351179094965, "grad_norm": 0.3084770441055298, "learning_rate": 3.163016212333269e-05, "loss": 0.0163, "step": 191630 }, { "epoch": 122.1414913957935, "grad_norm": 0.23285652697086334, "learning_rate": 3.1625036124059517e-05, "loss": 0.0118, "step": 191640 }, { "epoch": 122.1478648820905, "grad_norm": 0.3191241919994354, "learning_rate": 3.161991034805949e-05, "loss": 0.013, "step": 191650 }, { "epoch": 122.15423836838751, "grad_norm": 0.2298610955476761, "learning_rate": 3.1614784795394895e-05, "loss": 0.0076, "step": 191660 }, { "epoch": 122.16061185468452, "grad_norm": 0.1909330040216446, "learning_rate": 3.160965946612803e-05, "loss": 0.0158, "step": 191670 }, { "epoch": 122.16698534098151, "grad_norm": 0.19119928777217865, "learning_rate": 3.160453436032119e-05, "loss": 0.0189, "step": 191680 }, { "epoch": 122.17335882727852, "grad_norm": 0.19194428622722626, "learning_rate": 3.1599409478036595e-05, "loss": 0.0079, "step": 191690 }, { "epoch": 122.17973231357553, "grad_norm": 0.33839625120162964, "learning_rate": 3.159428481933655e-05, "loss": 0.0107, "step": 191700 }, { "epoch": 122.18610579987254, "grad_norm": 0.3177562355995178, "learning_rate": 3.1589160384283354e-05, "loss": 0.0135, "step": 191710 }, { "epoch": 122.19247928616953, "grad_norm": 0.29654914140701294, "learning_rate": 3.158403617293921e-05, "loss": 0.0082, "step": 191720 }, { "epoch": 122.19885277246654, "grad_norm": 0.529808759689331, "learning_rate": 3.1578912185366414e-05, "loss": 0.0095, "step": 191730 }, { "epoch": 122.20522625876355, "grad_norm": 0.1975748986005783, "learning_rate": 3.157378842162724e-05, "loss": 0.0091, "step": 191740 }, { "epoch": 122.21159974506055, "grad_norm": 0.2107285112142563, "learning_rate": 3.156866488178394e-05, "loss": 0.0191, "step": 191750 }, { "epoch": 122.21797323135755, "grad_norm": 0.1802995204925537, "learning_rate": 3.1563541565898744e-05, "loss": 0.0101, "step": 191760 }, { "epoch": 122.22434671765456, "grad_norm": 0.3057723939418793, "learning_rate": 3.1558418474033934e-05, "loss": 0.0134, "step": 191770 }, { "epoch": 122.23072020395156, "grad_norm": 0.3117622137069702, "learning_rate": 3.1553295606251765e-05, "loss": 0.0098, "step": 191780 }, { "epoch": 122.23709369024857, "grad_norm": 0.16348618268966675, "learning_rate": 3.1548172962614456e-05, "loss": 0.0072, "step": 191790 }, { "epoch": 122.24346717654556, "grad_norm": 0.16773521900177002, "learning_rate": 3.154305054318426e-05, "loss": 0.005, "step": 191800 }, { "epoch": 122.24984066284257, "grad_norm": 0.14496788382530212, "learning_rate": 3.1537928348023434e-05, "loss": 0.0099, "step": 191810 }, { "epoch": 122.25621414913958, "grad_norm": 0.2239914834499359, "learning_rate": 3.153280637719422e-05, "loss": 0.0062, "step": 191820 }, { "epoch": 122.26258763543659, "grad_norm": 0.345546692609787, "learning_rate": 3.1527684630758835e-05, "loss": 0.0127, "step": 191830 }, { "epoch": 122.26896112173358, "grad_norm": 0.2944988012313843, "learning_rate": 3.1522563108779516e-05, "loss": 0.012, "step": 191840 }, { "epoch": 122.27533460803059, "grad_norm": 0.3045710325241089, "learning_rate": 3.1517441811318524e-05, "loss": 0.0096, "step": 191850 }, { "epoch": 122.2817080943276, "grad_norm": 0.3690038025379181, "learning_rate": 3.151232073843804e-05, "loss": 0.0059, "step": 191860 }, { "epoch": 122.2880815806246, "grad_norm": 0.5275442600250244, "learning_rate": 3.150719989020032e-05, "loss": 0.0141, "step": 191870 }, { "epoch": 122.2944550669216, "grad_norm": 0.32228419184684753, "learning_rate": 3.1502079266667575e-05, "loss": 0.0071, "step": 191880 }, { "epoch": 122.30082855321861, "grad_norm": 0.09898168593645096, "learning_rate": 3.149695886790206e-05, "loss": 0.0096, "step": 191890 }, { "epoch": 122.30720203951562, "grad_norm": 0.29126015305519104, "learning_rate": 3.1491838693965935e-05, "loss": 0.0116, "step": 191900 }, { "epoch": 122.31357552581262, "grad_norm": 0.2573452293872833, "learning_rate": 3.148671874492145e-05, "loss": 0.0177, "step": 191910 }, { "epoch": 122.31994901210962, "grad_norm": 0.184273362159729, "learning_rate": 3.1481599020830835e-05, "loss": 0.0059, "step": 191920 }, { "epoch": 122.32632249840663, "grad_norm": 0.12840577960014343, "learning_rate": 3.1476479521756255e-05, "loss": 0.0075, "step": 191930 }, { "epoch": 122.33269598470363, "grad_norm": 0.5673059821128845, "learning_rate": 3.147136024775993e-05, "loss": 0.0129, "step": 191940 }, { "epoch": 122.33906947100064, "grad_norm": 0.8235641717910767, "learning_rate": 3.146624119890408e-05, "loss": 0.0138, "step": 191950 }, { "epoch": 122.34544295729764, "grad_norm": 0.35750681161880493, "learning_rate": 3.146112237525092e-05, "loss": 0.0051, "step": 191960 }, { "epoch": 122.35181644359464, "grad_norm": 0.2374793440103531, "learning_rate": 3.1456003776862597e-05, "loss": 0.0074, "step": 191970 }, { "epoch": 122.35818992989165, "grad_norm": 0.2289259433746338, "learning_rate": 3.145088540380134e-05, "loss": 0.007, "step": 191980 }, { "epoch": 122.36456341618866, "grad_norm": 0.2027670443058014, "learning_rate": 3.1445767256129346e-05, "loss": 0.0125, "step": 191990 }, { "epoch": 122.37093690248565, "grad_norm": 0.3460586667060852, "learning_rate": 3.14406493339088e-05, "loss": 0.0089, "step": 192000 }, { "epoch": 122.37731038878266, "grad_norm": 0.19149789214134216, "learning_rate": 3.1435531637201886e-05, "loss": 0.0067, "step": 192010 }, { "epoch": 122.38368387507967, "grad_norm": 0.43750131130218506, "learning_rate": 3.1430414166070785e-05, "loss": 0.0132, "step": 192020 }, { "epoch": 122.39005736137668, "grad_norm": 0.1590079516172409, "learning_rate": 3.14252969205777e-05, "loss": 0.0057, "step": 192030 }, { "epoch": 122.39643084767367, "grad_norm": 0.1748356968164444, "learning_rate": 3.142017990078478e-05, "loss": 0.0119, "step": 192040 }, { "epoch": 122.40280433397068, "grad_norm": 0.125773087143898, "learning_rate": 3.1415063106754214e-05, "loss": 0.0231, "step": 192050 }, { "epoch": 122.40917782026769, "grad_norm": 0.30482372641563416, "learning_rate": 3.140994653854818e-05, "loss": 0.016, "step": 192060 }, { "epoch": 122.4155513065647, "grad_norm": 0.27082836627960205, "learning_rate": 3.140483019622887e-05, "loss": 0.0143, "step": 192070 }, { "epoch": 122.4219247928617, "grad_norm": 0.1534324735403061, "learning_rate": 3.13997140798584e-05, "loss": 0.0081, "step": 192080 }, { "epoch": 122.4282982791587, "grad_norm": 0.4091256260871887, "learning_rate": 3.139459818949898e-05, "loss": 0.0151, "step": 192090 }, { "epoch": 122.4346717654557, "grad_norm": 0.16286826133728027, "learning_rate": 3.138948252521275e-05, "loss": 0.0123, "step": 192100 }, { "epoch": 122.44104525175271, "grad_norm": 0.24311619997024536, "learning_rate": 3.1384367087061896e-05, "loss": 0.0094, "step": 192110 }, { "epoch": 122.44741873804972, "grad_norm": 0.2648502588272095, "learning_rate": 3.137925187510854e-05, "loss": 0.0154, "step": 192120 }, { "epoch": 122.45379222434671, "grad_norm": 0.2282494306564331, "learning_rate": 3.137413688941485e-05, "loss": 0.0048, "step": 192130 }, { "epoch": 122.46016571064372, "grad_norm": 0.26723799109458923, "learning_rate": 3.1369022130043e-05, "loss": 0.0108, "step": 192140 }, { "epoch": 122.46653919694073, "grad_norm": 0.2515903115272522, "learning_rate": 3.136390759705512e-05, "loss": 0.0053, "step": 192150 }, { "epoch": 122.47291268323774, "grad_norm": 0.21310020983219147, "learning_rate": 3.135879329051335e-05, "loss": 0.0084, "step": 192160 }, { "epoch": 122.47928616953473, "grad_norm": 0.3019670248031616, "learning_rate": 3.135367921047985e-05, "loss": 0.0089, "step": 192170 }, { "epoch": 122.48565965583174, "grad_norm": 0.4517471492290497, "learning_rate": 3.134856535701675e-05, "loss": 0.0101, "step": 192180 }, { "epoch": 122.49203314212875, "grad_norm": 0.2234722226858139, "learning_rate": 3.1343451730186184e-05, "loss": 0.0067, "step": 192190 }, { "epoch": 122.49840662842576, "grad_norm": 0.1420173943042755, "learning_rate": 3.133833833005029e-05, "loss": 0.0102, "step": 192200 }, { "epoch": 122.50478011472275, "grad_norm": 0.42702534794807434, "learning_rate": 3.133322515667122e-05, "loss": 0.0099, "step": 192210 }, { "epoch": 122.51115360101976, "grad_norm": 0.3734264671802521, "learning_rate": 3.1328112210111085e-05, "loss": 0.01, "step": 192220 }, { "epoch": 122.51752708731676, "grad_norm": 0.19583597779273987, "learning_rate": 3.132299949043202e-05, "loss": 0.0093, "step": 192230 }, { "epoch": 122.52390057361377, "grad_norm": 0.17627353966236115, "learning_rate": 3.1317886997696136e-05, "loss": 0.0075, "step": 192240 }, { "epoch": 122.53027405991077, "grad_norm": 0.2630313038825989, "learning_rate": 3.131277473196555e-05, "loss": 0.006, "step": 192250 }, { "epoch": 122.53664754620777, "grad_norm": 0.5151763558387756, "learning_rate": 3.130766269330241e-05, "loss": 0.0067, "step": 192260 }, { "epoch": 122.54302103250478, "grad_norm": 0.19939100742340088, "learning_rate": 3.1302550881768824e-05, "loss": 0.0076, "step": 192270 }, { "epoch": 122.54939451880179, "grad_norm": 0.20961476862430573, "learning_rate": 3.12974392974269e-05, "loss": 0.0113, "step": 192280 }, { "epoch": 122.55576800509878, "grad_norm": 0.48451468348503113, "learning_rate": 3.129232794033873e-05, "loss": 0.0059, "step": 192290 }, { "epoch": 122.56214149139579, "grad_norm": 0.3034600615501404, "learning_rate": 3.128721681056644e-05, "loss": 0.0105, "step": 192300 }, { "epoch": 122.5685149776928, "grad_norm": 0.43503642082214355, "learning_rate": 3.128210590817214e-05, "loss": 0.0071, "step": 192310 }, { "epoch": 122.57488846398981, "grad_norm": 0.1837967336177826, "learning_rate": 3.127699523321794e-05, "loss": 0.0087, "step": 192320 }, { "epoch": 122.5812619502868, "grad_norm": 0.13591435551643372, "learning_rate": 3.1271884785765906e-05, "loss": 0.0056, "step": 192330 }, { "epoch": 122.58763543658381, "grad_norm": 0.23969879746437073, "learning_rate": 3.1266774565878154e-05, "loss": 0.0075, "step": 192340 }, { "epoch": 122.59400892288082, "grad_norm": 0.18110354244709015, "learning_rate": 3.12616645736168e-05, "loss": 0.0059, "step": 192350 }, { "epoch": 122.60038240917783, "grad_norm": 0.19977523386478424, "learning_rate": 3.125655480904389e-05, "loss": 0.0068, "step": 192360 }, { "epoch": 122.60675589547482, "grad_norm": 0.20604431629180908, "learning_rate": 3.1251445272221536e-05, "loss": 0.0077, "step": 192370 }, { "epoch": 122.61312938177183, "grad_norm": 0.17277485132217407, "learning_rate": 3.1246335963211825e-05, "loss": 0.0099, "step": 192380 }, { "epoch": 122.61950286806884, "grad_norm": 0.43369871377944946, "learning_rate": 3.1241226882076855e-05, "loss": 0.007, "step": 192390 }, { "epoch": 122.62587635436584, "grad_norm": 0.5706885457038879, "learning_rate": 3.1236118028878673e-05, "loss": 0.0175, "step": 192400 }, { "epoch": 122.63224984066284, "grad_norm": 0.2925262153148651, "learning_rate": 3.123100940367937e-05, "loss": 0.0075, "step": 192410 }, { "epoch": 122.63862332695984, "grad_norm": 0.1702062338590622, "learning_rate": 3.122590100654105e-05, "loss": 0.0066, "step": 192420 }, { "epoch": 122.64499681325685, "grad_norm": 0.1999903917312622, "learning_rate": 3.122079283752574e-05, "loss": 0.0057, "step": 192430 }, { "epoch": 122.65137029955386, "grad_norm": 0.3160587251186371, "learning_rate": 3.1215684896695515e-05, "loss": 0.0075, "step": 192440 }, { "epoch": 122.65774378585085, "grad_norm": 0.2055468112230301, "learning_rate": 3.121057718411246e-05, "loss": 0.0076, "step": 192450 }, { "epoch": 122.66411727214786, "grad_norm": 0.14877493679523468, "learning_rate": 3.120546969983866e-05, "loss": 0.0051, "step": 192460 }, { "epoch": 122.67049075844487, "grad_norm": 0.3934931755065918, "learning_rate": 3.120036244393612e-05, "loss": 0.0061, "step": 192470 }, { "epoch": 122.67686424474188, "grad_norm": 0.13346832990646362, "learning_rate": 3.119525541646693e-05, "loss": 0.0109, "step": 192480 }, { "epoch": 122.68323773103887, "grad_norm": 0.6143624782562256, "learning_rate": 3.1190148617493154e-05, "loss": 0.0157, "step": 192490 }, { "epoch": 122.68961121733588, "grad_norm": 0.2760361135005951, "learning_rate": 3.118504204707682e-05, "loss": 0.0083, "step": 192500 }, { "epoch": 122.69598470363289, "grad_norm": 0.3211762607097626, "learning_rate": 3.117993570527998e-05, "loss": 0.0097, "step": 192510 }, { "epoch": 122.7023581899299, "grad_norm": 0.315827339887619, "learning_rate": 3.11748295921647e-05, "loss": 0.0079, "step": 192520 }, { "epoch": 122.70873167622689, "grad_norm": 0.31762269139289856, "learning_rate": 3.116972370779303e-05, "loss": 0.02, "step": 192530 }, { "epoch": 122.7151051625239, "grad_norm": 0.2608756124973297, "learning_rate": 3.116461805222698e-05, "loss": 0.0145, "step": 192540 }, { "epoch": 122.7214786488209, "grad_norm": 0.321361243724823, "learning_rate": 3.11595126255286e-05, "loss": 0.0083, "step": 192550 }, { "epoch": 122.72785213511791, "grad_norm": 0.3235655128955841, "learning_rate": 3.115440742775993e-05, "loss": 0.0088, "step": 192560 }, { "epoch": 122.73422562141491, "grad_norm": 0.524125337600708, "learning_rate": 3.1149302458983026e-05, "loss": 0.01, "step": 192570 }, { "epoch": 122.74059910771192, "grad_norm": 0.6152228713035583, "learning_rate": 3.114419771925987e-05, "loss": 0.0088, "step": 192580 }, { "epoch": 122.74697259400892, "grad_norm": 0.2427012026309967, "learning_rate": 3.113909320865252e-05, "loss": 0.005, "step": 192590 }, { "epoch": 122.75334608030593, "grad_norm": 0.16364538669586182, "learning_rate": 3.113398892722301e-05, "loss": 0.0072, "step": 192600 }, { "epoch": 122.75971956660293, "grad_norm": 0.20695550739765167, "learning_rate": 3.112888487503333e-05, "loss": 0.0056, "step": 192610 }, { "epoch": 122.76609305289993, "grad_norm": 0.31452879309654236, "learning_rate": 3.112378105214552e-05, "loss": 0.0142, "step": 192620 }, { "epoch": 122.77246653919694, "grad_norm": 0.4529061019420624, "learning_rate": 3.1118677458621586e-05, "loss": 0.0074, "step": 192630 }, { "epoch": 122.77884002549395, "grad_norm": 0.40278327465057373, "learning_rate": 3.1113574094523566e-05, "loss": 0.0122, "step": 192640 }, { "epoch": 122.78521351179094, "grad_norm": 0.21653194725513458, "learning_rate": 3.110847095991344e-05, "loss": 0.0117, "step": 192650 }, { "epoch": 122.79158699808795, "grad_norm": 0.3082391023635864, "learning_rate": 3.110336805485322e-05, "loss": 0.0059, "step": 192660 }, { "epoch": 122.79796048438496, "grad_norm": 0.29961273074150085, "learning_rate": 3.1098265379404954e-05, "loss": 0.0177, "step": 192670 }, { "epoch": 122.80433397068197, "grad_norm": 0.32929688692092896, "learning_rate": 3.109316293363058e-05, "loss": 0.0133, "step": 192680 }, { "epoch": 122.81070745697897, "grad_norm": 0.22788172960281372, "learning_rate": 3.108806071759212e-05, "loss": 0.0098, "step": 192690 }, { "epoch": 122.81708094327597, "grad_norm": 0.25142911076545715, "learning_rate": 3.108295873135158e-05, "loss": 0.0069, "step": 192700 }, { "epoch": 122.82345442957298, "grad_norm": 0.2589869201183319, "learning_rate": 3.107785697497099e-05, "loss": 0.0112, "step": 192710 }, { "epoch": 122.82982791586998, "grad_norm": 0.16335755586624146, "learning_rate": 3.107275544851227e-05, "loss": 0.0089, "step": 192720 }, { "epoch": 122.83620140216699, "grad_norm": 0.1765221804380417, "learning_rate": 3.106765415203744e-05, "loss": 0.0149, "step": 192730 }, { "epoch": 122.84257488846399, "grad_norm": 0.30919453501701355, "learning_rate": 3.106255308560851e-05, "loss": 0.0083, "step": 192740 }, { "epoch": 122.848948374761, "grad_norm": 0.15959900617599487, "learning_rate": 3.105745224928742e-05, "loss": 0.0065, "step": 192750 }, { "epoch": 122.855321861058, "grad_norm": 0.23361870646476746, "learning_rate": 3.1052351643136166e-05, "loss": 0.0093, "step": 192760 }, { "epoch": 122.86169534735501, "grad_norm": 0.28216367959976196, "learning_rate": 3.104725126721674e-05, "loss": 0.0146, "step": 192770 }, { "epoch": 122.868068833652, "grad_norm": 0.2051176130771637, "learning_rate": 3.104215112159111e-05, "loss": 0.0122, "step": 192780 }, { "epoch": 122.87444231994901, "grad_norm": 0.40731021761894226, "learning_rate": 3.1037051206321235e-05, "loss": 0.0126, "step": 192790 }, { "epoch": 122.88081580624602, "grad_norm": 0.22044239938259125, "learning_rate": 3.103195152146908e-05, "loss": 0.0073, "step": 192800 }, { "epoch": 122.88718929254303, "grad_norm": 0.4919971227645874, "learning_rate": 3.1026852067096646e-05, "loss": 0.0084, "step": 192810 }, { "epoch": 122.89356277884002, "grad_norm": 0.19759079813957214, "learning_rate": 3.1021752843265863e-05, "loss": 0.0101, "step": 192820 }, { "epoch": 122.89993626513703, "grad_norm": 0.13890862464904785, "learning_rate": 3.1016653850038703e-05, "loss": 0.0108, "step": 192830 }, { "epoch": 122.90630975143404, "grad_norm": 0.4962965250015259, "learning_rate": 3.1011555087477114e-05, "loss": 0.0051, "step": 192840 }, { "epoch": 122.91268323773104, "grad_norm": 0.18363170325756073, "learning_rate": 3.100645655564307e-05, "loss": 0.0193, "step": 192850 }, { "epoch": 122.91905672402804, "grad_norm": 0.16887281835079193, "learning_rate": 3.100135825459851e-05, "loss": 0.0078, "step": 192860 }, { "epoch": 122.92543021032505, "grad_norm": 0.39558595418930054, "learning_rate": 3.099626018440538e-05, "loss": 0.0065, "step": 192870 }, { "epoch": 122.93180369662205, "grad_norm": 0.24186331033706665, "learning_rate": 3.0991162345125626e-05, "loss": 0.0084, "step": 192880 }, { "epoch": 122.93817718291906, "grad_norm": 0.11608146876096725, "learning_rate": 3.098606473682121e-05, "loss": 0.0148, "step": 192890 }, { "epoch": 122.94455066921606, "grad_norm": 0.39572426676750183, "learning_rate": 3.098096735955406e-05, "loss": 0.0104, "step": 192900 }, { "epoch": 122.95092415551306, "grad_norm": 0.14927202463150024, "learning_rate": 3.097587021338609e-05, "loss": 0.0238, "step": 192910 }, { "epoch": 122.95729764181007, "grad_norm": 0.21316328644752502, "learning_rate": 3.0970773298379286e-05, "loss": 0.0046, "step": 192920 }, { "epoch": 122.96367112810708, "grad_norm": 0.38775286078453064, "learning_rate": 3.096567661459554e-05, "loss": 0.0087, "step": 192930 }, { "epoch": 122.97004461440407, "grad_norm": 0.23423027992248535, "learning_rate": 3.09605801620968e-05, "loss": 0.0084, "step": 192940 }, { "epoch": 122.97641810070108, "grad_norm": 0.15848040580749512, "learning_rate": 3.095548394094497e-05, "loss": 0.0055, "step": 192950 }, { "epoch": 122.98279158699809, "grad_norm": 0.135626420378685, "learning_rate": 3.095038795120202e-05, "loss": 0.011, "step": 192960 }, { "epoch": 122.9891650732951, "grad_norm": 0.2128722071647644, "learning_rate": 3.094529219292982e-05, "loss": 0.0091, "step": 192970 }, { "epoch": 122.99553855959209, "grad_norm": 0.25398528575897217, "learning_rate": 3.0940196666190323e-05, "loss": 0.0081, "step": 192980 }, { "epoch": 123.0019120458891, "grad_norm": 0.16766732931137085, "learning_rate": 3.0935101371045436e-05, "loss": 0.0097, "step": 192990 }, { "epoch": 123.00828553218611, "grad_norm": 0.18847030401229858, "learning_rate": 3.0930006307557047e-05, "loss": 0.0075, "step": 193000 }, { "epoch": 123.01465901848312, "grad_norm": 0.31693515181541443, "learning_rate": 3.0924911475787096e-05, "loss": 0.0089, "step": 193010 }, { "epoch": 123.02103250478011, "grad_norm": 0.23979979753494263, "learning_rate": 3.091981687579749e-05, "loss": 0.0076, "step": 193020 }, { "epoch": 123.02740599107712, "grad_norm": 0.26390981674194336, "learning_rate": 3.091472250765012e-05, "loss": 0.0055, "step": 193030 }, { "epoch": 123.03377947737413, "grad_norm": 0.32940956950187683, "learning_rate": 3.090962837140688e-05, "loss": 0.0058, "step": 193040 }, { "epoch": 123.04015296367113, "grad_norm": 0.27856460213661194, "learning_rate": 3.090453446712967e-05, "loss": 0.01, "step": 193050 }, { "epoch": 123.04652644996813, "grad_norm": 0.5238555669784546, "learning_rate": 3.089944079488044e-05, "loss": 0.0092, "step": 193060 }, { "epoch": 123.05289993626513, "grad_norm": 0.18816393613815308, "learning_rate": 3.0894347354721e-05, "loss": 0.0155, "step": 193070 }, { "epoch": 123.05927342256214, "grad_norm": 0.2350817322731018, "learning_rate": 3.0889254146713286e-05, "loss": 0.0065, "step": 193080 }, { "epoch": 123.06564690885915, "grad_norm": 0.16594065725803375, "learning_rate": 3.088416117091917e-05, "loss": 0.0119, "step": 193090 }, { "epoch": 123.07202039515614, "grad_norm": 0.3026483356952667, "learning_rate": 3.087906842740057e-05, "loss": 0.0097, "step": 193100 }, { "epoch": 123.07839388145315, "grad_norm": 0.2790206968784332, "learning_rate": 3.0873975916219325e-05, "loss": 0.0137, "step": 193110 }, { "epoch": 123.08476736775016, "grad_norm": 0.3457379937171936, "learning_rate": 3.0868883637437327e-05, "loss": 0.0067, "step": 193120 }, { "epoch": 123.09114085404717, "grad_norm": 0.14758886396884918, "learning_rate": 3.0863791591116473e-05, "loss": 0.0072, "step": 193130 }, { "epoch": 123.09751434034416, "grad_norm": 0.21220912039279938, "learning_rate": 3.08586997773186e-05, "loss": 0.0091, "step": 193140 }, { "epoch": 123.10388782664117, "grad_norm": 0.2974901795387268, "learning_rate": 3.08536081961056e-05, "loss": 0.0064, "step": 193150 }, { "epoch": 123.11026131293818, "grad_norm": 0.1716255396604538, "learning_rate": 3.0848516847539336e-05, "loss": 0.0063, "step": 193160 }, { "epoch": 123.11663479923519, "grad_norm": 0.5280405879020691, "learning_rate": 3.084342573168169e-05, "loss": 0.0091, "step": 193170 }, { "epoch": 123.12300828553218, "grad_norm": 0.1558116376399994, "learning_rate": 3.0838334848594495e-05, "loss": 0.0103, "step": 193180 }, { "epoch": 123.12938177182919, "grad_norm": 0.3491559326648712, "learning_rate": 3.083324419833962e-05, "loss": 0.0048, "step": 193190 }, { "epoch": 123.1357552581262, "grad_norm": 0.352328896522522, "learning_rate": 3.082815378097892e-05, "loss": 0.0084, "step": 193200 }, { "epoch": 123.1421287444232, "grad_norm": 0.5559830665588379, "learning_rate": 3.082306359657428e-05, "loss": 0.0139, "step": 193210 }, { "epoch": 123.1485022307202, "grad_norm": 0.15496742725372314, "learning_rate": 3.0817973645187505e-05, "loss": 0.0053, "step": 193220 }, { "epoch": 123.1548757170172, "grad_norm": 0.22219574451446533, "learning_rate": 3.081288392688045e-05, "loss": 0.0085, "step": 193230 }, { "epoch": 123.16124920331421, "grad_norm": 0.23138543963432312, "learning_rate": 3.0807794441715e-05, "loss": 0.0109, "step": 193240 }, { "epoch": 123.16762268961122, "grad_norm": 0.4550197422504425, "learning_rate": 3.080270518975295e-05, "loss": 0.0061, "step": 193250 }, { "epoch": 123.17399617590821, "grad_norm": 0.6882684230804443, "learning_rate": 3.079761617105615e-05, "loss": 0.0158, "step": 193260 }, { "epoch": 123.18036966220522, "grad_norm": 0.177008256316185, "learning_rate": 3.0792527385686445e-05, "loss": 0.0223, "step": 193270 }, { "epoch": 123.18674314850223, "grad_norm": 0.163381889462471, "learning_rate": 3.078743883370569e-05, "loss": 0.0159, "step": 193280 }, { "epoch": 123.19311663479924, "grad_norm": 0.6858503222465515, "learning_rate": 3.078235051517567e-05, "loss": 0.0103, "step": 193290 }, { "epoch": 123.19949012109623, "grad_norm": 0.4511534869670868, "learning_rate": 3.077726243015824e-05, "loss": 0.0108, "step": 193300 }, { "epoch": 123.20586360739324, "grad_norm": 0.47110772132873535, "learning_rate": 3.0772174578715236e-05, "loss": 0.0123, "step": 193310 }, { "epoch": 123.21223709369025, "grad_norm": 0.20495618879795074, "learning_rate": 3.0767086960908444e-05, "loss": 0.0064, "step": 193320 }, { "epoch": 123.21861057998726, "grad_norm": 0.15175732970237732, "learning_rate": 3.076199957679971e-05, "loss": 0.0059, "step": 193330 }, { "epoch": 123.22498406628426, "grad_norm": 0.2631797194480896, "learning_rate": 3.075691242645084e-05, "loss": 0.0072, "step": 193340 }, { "epoch": 123.23135755258126, "grad_norm": 0.30508607625961304, "learning_rate": 3.075182550992367e-05, "loss": 0.0074, "step": 193350 }, { "epoch": 123.23773103887827, "grad_norm": 0.2644168436527252, "learning_rate": 3.074673882727999e-05, "loss": 0.0141, "step": 193360 }, { "epoch": 123.24410452517527, "grad_norm": 0.20039471983909607, "learning_rate": 3.0741652378581585e-05, "loss": 0.0199, "step": 193370 }, { "epoch": 123.25047801147228, "grad_norm": 0.491616815328598, "learning_rate": 3.0736566163890325e-05, "loss": 0.0142, "step": 193380 }, { "epoch": 123.25685149776928, "grad_norm": 0.22785735130310059, "learning_rate": 3.073148018326795e-05, "loss": 0.0098, "step": 193390 }, { "epoch": 123.26322498406628, "grad_norm": 0.47182556986808777, "learning_rate": 3.072639443677629e-05, "loss": 0.0089, "step": 193400 }, { "epoch": 123.26959847036329, "grad_norm": 0.2448681741952896, "learning_rate": 3.0721308924477124e-05, "loss": 0.0162, "step": 193410 }, { "epoch": 123.2759719566603, "grad_norm": 0.36747005581855774, "learning_rate": 3.0716223646432274e-05, "loss": 0.008, "step": 193420 }, { "epoch": 123.2823454429573, "grad_norm": 0.18667751550674438, "learning_rate": 3.0711138602703504e-05, "loss": 0.0077, "step": 193430 }, { "epoch": 123.2887189292543, "grad_norm": 0.19806817173957825, "learning_rate": 3.0706053793352606e-05, "loss": 0.0049, "step": 193440 }, { "epoch": 123.29509241555131, "grad_norm": 0.2224341183900833, "learning_rate": 3.0700969218441386e-05, "loss": 0.0064, "step": 193450 }, { "epoch": 123.30146590184832, "grad_norm": 0.3350595235824585, "learning_rate": 3.0695884878031604e-05, "loss": 0.0147, "step": 193460 }, { "epoch": 123.30783938814531, "grad_norm": 0.2077118456363678, "learning_rate": 3.069080077218504e-05, "loss": 0.0163, "step": 193470 }, { "epoch": 123.31421287444232, "grad_norm": 0.27197200059890747, "learning_rate": 3.068571690096348e-05, "loss": 0.0108, "step": 193480 }, { "epoch": 123.32058636073933, "grad_norm": 0.11452541500329971, "learning_rate": 3.068063326442871e-05, "loss": 0.0083, "step": 193490 }, { "epoch": 123.32695984703633, "grad_norm": 0.2217608243227005, "learning_rate": 3.067554986264247e-05, "loss": 0.0099, "step": 193500 }, { "epoch": 123.33333333333333, "grad_norm": 0.2224886566400528, "learning_rate": 3.0670466695666544e-05, "loss": 0.0049, "step": 193510 }, { "epoch": 123.33970681963034, "grad_norm": 0.4372996687889099, "learning_rate": 3.06653837635627e-05, "loss": 0.0118, "step": 193520 }, { "epoch": 123.34608030592734, "grad_norm": 0.3203122615814209, "learning_rate": 3.066030106639272e-05, "loss": 0.0077, "step": 193530 }, { "epoch": 123.35245379222435, "grad_norm": 0.2116926908493042, "learning_rate": 3.0655218604218315e-05, "loss": 0.0087, "step": 193540 }, { "epoch": 123.35882727852135, "grad_norm": 0.21601071953773499, "learning_rate": 3.065013637710127e-05, "loss": 0.0112, "step": 193550 }, { "epoch": 123.36520076481835, "grad_norm": 0.11225463449954987, "learning_rate": 3.064505438510335e-05, "loss": 0.0048, "step": 193560 }, { "epoch": 123.37157425111536, "grad_norm": 0.21525530517101288, "learning_rate": 3.063997262828629e-05, "loss": 0.0078, "step": 193570 }, { "epoch": 123.37794773741237, "grad_norm": 0.31269529461860657, "learning_rate": 3.0634891106711836e-05, "loss": 0.0166, "step": 193580 }, { "epoch": 123.38432122370936, "grad_norm": 0.3390798568725586, "learning_rate": 3.062980982044174e-05, "loss": 0.0112, "step": 193590 }, { "epoch": 123.39069471000637, "grad_norm": 0.5249605178833008, "learning_rate": 3.062472876953775e-05, "loss": 0.0117, "step": 193600 }, { "epoch": 123.39706819630338, "grad_norm": 0.21501420438289642, "learning_rate": 3.0619647954061594e-05, "loss": 0.0095, "step": 193610 }, { "epoch": 123.40344168260039, "grad_norm": 0.12762799859046936, "learning_rate": 3.0614567374075004e-05, "loss": 0.0194, "step": 193620 }, { "epoch": 123.40981516889738, "grad_norm": 0.15716956555843353, "learning_rate": 3.060948702963974e-05, "loss": 0.0095, "step": 193630 }, { "epoch": 123.41618865519439, "grad_norm": 0.3799980878829956, "learning_rate": 3.0604406920817504e-05, "loss": 0.0107, "step": 193640 }, { "epoch": 123.4225621414914, "grad_norm": 0.255869597196579, "learning_rate": 3.059932704767004e-05, "loss": 0.0051, "step": 193650 }, { "epoch": 123.4289356277884, "grad_norm": 0.45301467180252075, "learning_rate": 3.059424741025907e-05, "loss": 0.0068, "step": 193660 }, { "epoch": 123.4353091140854, "grad_norm": 0.13122892379760742, "learning_rate": 3.058916800864632e-05, "loss": 0.0144, "step": 193670 }, { "epoch": 123.4416826003824, "grad_norm": 0.25252214074134827, "learning_rate": 3.05840888428935e-05, "loss": 0.0091, "step": 193680 }, { "epoch": 123.44805608667941, "grad_norm": 0.31668245792388916, "learning_rate": 3.0579009913062334e-05, "loss": 0.0119, "step": 193690 }, { "epoch": 123.45442957297642, "grad_norm": 0.39272820949554443, "learning_rate": 3.057393121921453e-05, "loss": 0.0127, "step": 193700 }, { "epoch": 123.46080305927342, "grad_norm": 0.19693152606487274, "learning_rate": 3.0568852761411814e-05, "loss": 0.0062, "step": 193710 }, { "epoch": 123.46717654557042, "grad_norm": 0.2955210506916046, "learning_rate": 3.0563774539715864e-05, "loss": 0.0153, "step": 193720 }, { "epoch": 123.47355003186743, "grad_norm": 0.24506717920303345, "learning_rate": 3.055869655418843e-05, "loss": 0.0104, "step": 193730 }, { "epoch": 123.47992351816444, "grad_norm": 0.2107606679201126, "learning_rate": 3.055361880489117e-05, "loss": 0.0098, "step": 193740 }, { "epoch": 123.48629700446143, "grad_norm": 0.24309813976287842, "learning_rate": 3.0548541291885805e-05, "loss": 0.0079, "step": 193750 }, { "epoch": 123.49267049075844, "grad_norm": 0.3718718886375427, "learning_rate": 3.054346401523403e-05, "loss": 0.0108, "step": 193760 }, { "epoch": 123.49904397705545, "grad_norm": 0.11096420884132385, "learning_rate": 3.0538386974997536e-05, "loss": 0.0102, "step": 193770 }, { "epoch": 123.50541746335246, "grad_norm": 0.31085291504859924, "learning_rate": 3.053331017123802e-05, "loss": 0.0075, "step": 193780 }, { "epoch": 123.51179094964945, "grad_norm": 0.2206360548734665, "learning_rate": 3.052823360401716e-05, "loss": 0.017, "step": 193790 }, { "epoch": 123.51816443594646, "grad_norm": 0.28056588768959045, "learning_rate": 3.052315727339664e-05, "loss": 0.0142, "step": 193800 }, { "epoch": 123.52453792224347, "grad_norm": 0.19519668817520142, "learning_rate": 3.051808117943817e-05, "loss": 0.005, "step": 193810 }, { "epoch": 123.53091140854048, "grad_norm": 0.29346102476119995, "learning_rate": 3.0513005322203392e-05, "loss": 0.0068, "step": 193820 }, { "epoch": 123.53728489483747, "grad_norm": 0.37998542189598083, "learning_rate": 3.050792970175399e-05, "loss": 0.0163, "step": 193830 }, { "epoch": 123.54365838113448, "grad_norm": 0.2325873225927353, "learning_rate": 3.0502854318151653e-05, "loss": 0.0071, "step": 193840 }, { "epoch": 123.55003186743149, "grad_norm": 0.22250108420848846, "learning_rate": 3.0497779171458063e-05, "loss": 0.0119, "step": 193850 }, { "epoch": 123.5564053537285, "grad_norm": 0.29542219638824463, "learning_rate": 3.049270426173485e-05, "loss": 0.0069, "step": 193860 }, { "epoch": 123.56277884002549, "grad_norm": 0.2652048170566559, "learning_rate": 3.048762958904369e-05, "loss": 0.0054, "step": 193870 }, { "epoch": 123.5691523263225, "grad_norm": 0.20896486937999725, "learning_rate": 3.048255515344628e-05, "loss": 0.0053, "step": 193880 }, { "epoch": 123.5755258126195, "grad_norm": 0.19949908554553986, "learning_rate": 3.0477480955004233e-05, "loss": 0.0105, "step": 193890 }, { "epoch": 123.58189929891651, "grad_norm": 0.4890397787094116, "learning_rate": 3.0472406993779223e-05, "loss": 0.0115, "step": 193900 }, { "epoch": 123.5882727852135, "grad_norm": 0.42673739790916443, "learning_rate": 3.046733326983291e-05, "loss": 0.0106, "step": 193910 }, { "epoch": 123.59464627151051, "grad_norm": 0.2995816171169281, "learning_rate": 3.0462259783226954e-05, "loss": 0.0094, "step": 193920 }, { "epoch": 123.60101975780752, "grad_norm": 0.26002392172813416, "learning_rate": 3.0457186534022962e-05, "loss": 0.0088, "step": 193930 }, { "epoch": 123.60739324410453, "grad_norm": 0.18743771314620972, "learning_rate": 3.0452113522282618e-05, "loss": 0.0067, "step": 193940 }, { "epoch": 123.61376673040154, "grad_norm": 0.19961977005004883, "learning_rate": 3.0447040748067567e-05, "loss": 0.0076, "step": 193950 }, { "epoch": 123.62014021669853, "grad_norm": 0.3406122326850891, "learning_rate": 3.044196821143941e-05, "loss": 0.0072, "step": 193960 }, { "epoch": 123.62651370299554, "grad_norm": 0.3656028211116791, "learning_rate": 3.0436895912459808e-05, "loss": 0.0093, "step": 193970 }, { "epoch": 123.63288718929255, "grad_norm": 0.14921823143959045, "learning_rate": 3.0431823851190383e-05, "loss": 0.0086, "step": 193980 }, { "epoch": 123.63926067558955, "grad_norm": 0.26231762766838074, "learning_rate": 3.04267520276928e-05, "loss": 0.0087, "step": 193990 }, { "epoch": 123.64563416188655, "grad_norm": 0.4875856637954712, "learning_rate": 3.0421680442028645e-05, "loss": 0.0096, "step": 194000 }, { "epoch": 123.65200764818356, "grad_norm": 0.24778415262699127, "learning_rate": 3.0416609094259545e-05, "loss": 0.007, "step": 194010 }, { "epoch": 123.65838113448056, "grad_norm": 0.22273410856723785, "learning_rate": 3.0411537984447163e-05, "loss": 0.0173, "step": 194020 }, { "epoch": 123.66475462077757, "grad_norm": 0.21297365427017212, "learning_rate": 3.040646711265307e-05, "loss": 0.0099, "step": 194030 }, { "epoch": 123.67112810707457, "grad_norm": 0.3435242474079132, "learning_rate": 3.0401396478938892e-05, "loss": 0.0073, "step": 194040 }, { "epoch": 123.67750159337157, "grad_norm": 0.17552518844604492, "learning_rate": 3.039632608336626e-05, "loss": 0.0059, "step": 194050 }, { "epoch": 123.68387507966858, "grad_norm": 0.22267208993434906, "learning_rate": 3.03912559259968e-05, "loss": 0.0128, "step": 194060 }, { "epoch": 123.69024856596559, "grad_norm": 0.22082924842834473, "learning_rate": 3.038618600689206e-05, "loss": 0.0053, "step": 194070 }, { "epoch": 123.69662205226258, "grad_norm": 0.3329477906227112, "learning_rate": 3.038111632611369e-05, "loss": 0.0068, "step": 194080 }, { "epoch": 123.70299553855959, "grad_norm": 0.23524963855743408, "learning_rate": 3.0376046883723275e-05, "loss": 0.011, "step": 194090 }, { "epoch": 123.7093690248566, "grad_norm": 0.20997785031795502, "learning_rate": 3.037097767978243e-05, "loss": 0.0103, "step": 194100 }, { "epoch": 123.7157425111536, "grad_norm": 0.20670467615127563, "learning_rate": 3.0365908714352733e-05, "loss": 0.0054, "step": 194110 }, { "epoch": 123.7221159974506, "grad_norm": 0.198499858379364, "learning_rate": 3.0360839987495776e-05, "loss": 0.0068, "step": 194120 }, { "epoch": 123.72848948374761, "grad_norm": 0.20785042643547058, "learning_rate": 3.035577149927318e-05, "loss": 0.0143, "step": 194130 }, { "epoch": 123.73486297004462, "grad_norm": 0.22689852118492126, "learning_rate": 3.035070324974648e-05, "loss": 0.0085, "step": 194140 }, { "epoch": 123.74123645634162, "grad_norm": 0.3620717525482178, "learning_rate": 3.0345635238977294e-05, "loss": 0.0054, "step": 194150 }, { "epoch": 123.74760994263862, "grad_norm": 0.263260155916214, "learning_rate": 3.0340567467027193e-05, "loss": 0.0098, "step": 194160 }, { "epoch": 123.75398342893563, "grad_norm": 0.4023111164569855, "learning_rate": 3.0335499933957777e-05, "loss": 0.0079, "step": 194170 }, { "epoch": 123.76035691523263, "grad_norm": 0.4768292009830475, "learning_rate": 3.033043263983059e-05, "loss": 0.0132, "step": 194180 }, { "epoch": 123.76673040152964, "grad_norm": 0.23565667867660522, "learning_rate": 3.0325365584707216e-05, "loss": 0.0072, "step": 194190 }, { "epoch": 123.77310388782664, "grad_norm": 0.11988133937120438, "learning_rate": 3.0320298768649246e-05, "loss": 0.0168, "step": 194200 }, { "epoch": 123.77947737412364, "grad_norm": 0.2726680636405945, "learning_rate": 3.03152321917182e-05, "loss": 0.007, "step": 194210 }, { "epoch": 123.78585086042065, "grad_norm": 0.3902071714401245, "learning_rate": 3.0310165853975674e-05, "loss": 0.0145, "step": 194220 }, { "epoch": 123.79222434671766, "grad_norm": 0.16569702327251434, "learning_rate": 3.0305099755483224e-05, "loss": 0.0092, "step": 194230 }, { "epoch": 123.79859783301465, "grad_norm": 0.21275977790355682, "learning_rate": 3.0300033896302426e-05, "loss": 0.0089, "step": 194240 }, { "epoch": 123.80497131931166, "grad_norm": 0.5480740070343018, "learning_rate": 3.0294968276494796e-05, "loss": 0.0083, "step": 194250 }, { "epoch": 123.81134480560867, "grad_norm": 0.23318099975585938, "learning_rate": 3.0289902896121903e-05, "loss": 0.0121, "step": 194260 }, { "epoch": 123.81771829190568, "grad_norm": 0.19485698640346527, "learning_rate": 3.028483775524531e-05, "loss": 0.0065, "step": 194270 }, { "epoch": 123.82409177820267, "grad_norm": 0.27652695775032043, "learning_rate": 3.0279772853926556e-05, "loss": 0.0081, "step": 194280 }, { "epoch": 123.83046526449968, "grad_norm": 0.3735632002353668, "learning_rate": 3.0274708192227173e-05, "loss": 0.0106, "step": 194290 }, { "epoch": 123.83683875079669, "grad_norm": 0.38898158073425293, "learning_rate": 3.02696437702087e-05, "loss": 0.0134, "step": 194300 }, { "epoch": 123.8432122370937, "grad_norm": 0.2926764190196991, "learning_rate": 3.02645795879327e-05, "loss": 0.0084, "step": 194310 }, { "epoch": 123.84958572339069, "grad_norm": 0.1916087567806244, "learning_rate": 3.025951564546069e-05, "loss": 0.0114, "step": 194320 }, { "epoch": 123.8559592096877, "grad_norm": 0.299578994512558, "learning_rate": 3.0254451942854196e-05, "loss": 0.0086, "step": 194330 }, { "epoch": 123.8623326959847, "grad_norm": 0.2952232360839844, "learning_rate": 3.024938848017476e-05, "loss": 0.0097, "step": 194340 }, { "epoch": 123.86870618228171, "grad_norm": 0.22044870257377625, "learning_rate": 3.0244325257483897e-05, "loss": 0.0063, "step": 194350 }, { "epoch": 123.8750796685787, "grad_norm": 0.17639093101024628, "learning_rate": 3.023926227484314e-05, "loss": 0.0161, "step": 194360 }, { "epoch": 123.88145315487571, "grad_norm": 0.3567701578140259, "learning_rate": 3.023419953231399e-05, "loss": 0.0094, "step": 194370 }, { "epoch": 123.88782664117272, "grad_norm": 0.15139298141002655, "learning_rate": 3.0229137029957993e-05, "loss": 0.0178, "step": 194380 }, { "epoch": 123.89420012746973, "grad_norm": 0.3561333417892456, "learning_rate": 3.0224074767836637e-05, "loss": 0.0091, "step": 194390 }, { "epoch": 123.90057361376672, "grad_norm": 0.15362589061260223, "learning_rate": 3.0219012746011454e-05, "loss": 0.0101, "step": 194400 }, { "epoch": 123.90694710006373, "grad_norm": 0.5968348979949951, "learning_rate": 3.021395096454393e-05, "loss": 0.01, "step": 194410 }, { "epoch": 123.91332058636074, "grad_norm": 0.20135250687599182, "learning_rate": 3.0208889423495594e-05, "loss": 0.0118, "step": 194420 }, { "epoch": 123.91969407265775, "grad_norm": 0.24921338260173798, "learning_rate": 3.020382812292793e-05, "loss": 0.0119, "step": 194430 }, { "epoch": 123.92606755895474, "grad_norm": 0.30638304352760315, "learning_rate": 3.0198767062902455e-05, "loss": 0.0058, "step": 194440 }, { "epoch": 123.93244104525175, "grad_norm": 0.1478521227836609, "learning_rate": 3.0193706243480656e-05, "loss": 0.0121, "step": 194450 }, { "epoch": 123.93881453154876, "grad_norm": 0.4992946684360504, "learning_rate": 3.0188645664724014e-05, "loss": 0.0126, "step": 194460 }, { "epoch": 123.94518801784577, "grad_norm": 0.4132925570011139, "learning_rate": 3.018358532669403e-05, "loss": 0.0269, "step": 194470 }, { "epoch": 123.95156150414276, "grad_norm": 0.12741459906101227, "learning_rate": 3.0178525229452204e-05, "loss": 0.0061, "step": 194480 }, { "epoch": 123.95793499043977, "grad_norm": 0.18072228133678436, "learning_rate": 3.0173465373060018e-05, "loss": 0.0123, "step": 194490 }, { "epoch": 123.96430847673678, "grad_norm": 0.265250563621521, "learning_rate": 3.0168405757578934e-05, "loss": 0.0088, "step": 194500 }, { "epoch": 123.97068196303378, "grad_norm": 0.19822216033935547, "learning_rate": 3.016334638307044e-05, "loss": 0.0082, "step": 194510 }, { "epoch": 123.97705544933078, "grad_norm": 0.37239617109298706, "learning_rate": 3.015828724959604e-05, "loss": 0.0076, "step": 194520 }, { "epoch": 123.98342893562778, "grad_norm": 0.38056501746177673, "learning_rate": 3.0153228357217166e-05, "loss": 0.0098, "step": 194530 }, { "epoch": 123.9898024219248, "grad_norm": 0.4358058273792267, "learning_rate": 3.01481697059953e-05, "loss": 0.0091, "step": 194540 }, { "epoch": 123.9961759082218, "grad_norm": 0.12393330782651901, "learning_rate": 3.0143111295991922e-05, "loss": 0.014, "step": 194550 }, { "epoch": 124.00254939451881, "grad_norm": 0.3003818988800049, "learning_rate": 3.0138053127268506e-05, "loss": 0.0157, "step": 194560 }, { "epoch": 124.0089228808158, "grad_norm": 0.2756510078907013, "learning_rate": 3.013299519988647e-05, "loss": 0.0136, "step": 194570 }, { "epoch": 124.01529636711281, "grad_norm": 0.25998082756996155, "learning_rate": 3.012793751390731e-05, "loss": 0.0072, "step": 194580 }, { "epoch": 124.02166985340982, "grad_norm": 0.1702447533607483, "learning_rate": 3.012288006939249e-05, "loss": 0.0086, "step": 194590 }, { "epoch": 124.02804333970683, "grad_norm": 0.18394248187541962, "learning_rate": 3.0117822866403424e-05, "loss": 0.0049, "step": 194600 }, { "epoch": 124.03441682600382, "grad_norm": 0.20790612697601318, "learning_rate": 3.0112765905001583e-05, "loss": 0.0076, "step": 194610 }, { "epoch": 124.04079031230083, "grad_norm": 0.43269816040992737, "learning_rate": 3.0107709185248407e-05, "loss": 0.0072, "step": 194620 }, { "epoch": 124.04716379859784, "grad_norm": 0.21481989324092865, "learning_rate": 3.010265270720537e-05, "loss": 0.0155, "step": 194630 }, { "epoch": 124.05353728489484, "grad_norm": 0.5004616975784302, "learning_rate": 3.009759647093387e-05, "loss": 0.0056, "step": 194640 }, { "epoch": 124.05991077119184, "grad_norm": 0.27709218859672546, "learning_rate": 3.0092540476495362e-05, "loss": 0.0124, "step": 194650 }, { "epoch": 124.06628425748885, "grad_norm": 0.19980505108833313, "learning_rate": 3.0087484723951304e-05, "loss": 0.0122, "step": 194660 }, { "epoch": 124.07265774378585, "grad_norm": 0.4104327857494354, "learning_rate": 3.0082429213363084e-05, "loss": 0.0133, "step": 194670 }, { "epoch": 124.07903123008286, "grad_norm": 0.402022123336792, "learning_rate": 3.0077373944792158e-05, "loss": 0.0082, "step": 194680 }, { "epoch": 124.08540471637986, "grad_norm": 0.2894449234008789, "learning_rate": 3.0072318918299947e-05, "loss": 0.0092, "step": 194690 }, { "epoch": 124.09177820267686, "grad_norm": 0.6193724870681763, "learning_rate": 3.0067264133947897e-05, "loss": 0.0175, "step": 194700 }, { "epoch": 124.09815168897387, "grad_norm": 0.17684578895568848, "learning_rate": 3.0062209591797386e-05, "loss": 0.0118, "step": 194710 }, { "epoch": 124.10452517527088, "grad_norm": 0.3937324285507202, "learning_rate": 3.0057155291909855e-05, "loss": 0.0049, "step": 194720 }, { "epoch": 124.11089866156787, "grad_norm": 0.3949861526489258, "learning_rate": 3.0052101234346718e-05, "loss": 0.0103, "step": 194730 }, { "epoch": 124.11727214786488, "grad_norm": 0.27650776505470276, "learning_rate": 3.00470474191694e-05, "loss": 0.0091, "step": 194740 }, { "epoch": 124.12364563416189, "grad_norm": 0.11914093792438507, "learning_rate": 3.0041993846439274e-05, "loss": 0.0067, "step": 194750 }, { "epoch": 124.1300191204589, "grad_norm": 0.3424961268901825, "learning_rate": 3.003694051621777e-05, "loss": 0.0145, "step": 194760 }, { "epoch": 124.13639260675589, "grad_norm": 0.5892871022224426, "learning_rate": 3.003188742856631e-05, "loss": 0.0144, "step": 194770 }, { "epoch": 124.1427660930529, "grad_norm": 0.2323741763830185, "learning_rate": 3.0026834583546244e-05, "loss": 0.0159, "step": 194780 }, { "epoch": 124.1491395793499, "grad_norm": 0.22123198211193085, "learning_rate": 3.0021781981218998e-05, "loss": 0.0095, "step": 194790 }, { "epoch": 124.15551306564691, "grad_norm": 0.24006396532058716, "learning_rate": 3.0016729621645967e-05, "loss": 0.009, "step": 194800 }, { "epoch": 124.16188655194391, "grad_norm": 0.42932242155075073, "learning_rate": 3.0011677504888554e-05, "loss": 0.0175, "step": 194810 }, { "epoch": 124.16826003824092, "grad_norm": 0.3413126766681671, "learning_rate": 3.000662563100811e-05, "loss": 0.0083, "step": 194820 }, { "epoch": 124.17463352453792, "grad_norm": 0.14406339824199677, "learning_rate": 3.0001574000066045e-05, "loss": 0.0056, "step": 194830 }, { "epoch": 124.18100701083493, "grad_norm": 0.3312388062477112, "learning_rate": 2.9996522612123755e-05, "loss": 0.0157, "step": 194840 }, { "epoch": 124.18738049713193, "grad_norm": 0.23184509575366974, "learning_rate": 2.999147146724258e-05, "loss": 0.0131, "step": 194850 }, { "epoch": 124.19375398342893, "grad_norm": 0.17721779644489288, "learning_rate": 2.9986420565483925e-05, "loss": 0.0092, "step": 194860 }, { "epoch": 124.20012746972594, "grad_norm": 0.22155708074569702, "learning_rate": 2.998136990690915e-05, "loss": 0.0084, "step": 194870 }, { "epoch": 124.20650095602295, "grad_norm": 0.27825823426246643, "learning_rate": 2.997631949157966e-05, "loss": 0.0125, "step": 194880 }, { "epoch": 124.21287444231994, "grad_norm": 0.1488322913646698, "learning_rate": 2.997126931955676e-05, "loss": 0.0042, "step": 194890 }, { "epoch": 124.21924792861695, "grad_norm": 0.2583620548248291, "learning_rate": 2.9966219390901863e-05, "loss": 0.0102, "step": 194900 }, { "epoch": 124.22562141491396, "grad_norm": 0.5099391937255859, "learning_rate": 2.9961169705676324e-05, "loss": 0.0143, "step": 194910 }, { "epoch": 124.23199490121097, "grad_norm": 0.34421345591545105, "learning_rate": 2.9956120263941478e-05, "loss": 0.0132, "step": 194920 }, { "epoch": 124.23836838750796, "grad_norm": 0.1473533660173416, "learning_rate": 2.9951071065758697e-05, "loss": 0.0077, "step": 194930 }, { "epoch": 124.24474187380497, "grad_norm": 0.1545647829771042, "learning_rate": 2.994602211118932e-05, "loss": 0.0087, "step": 194940 }, { "epoch": 124.25111536010198, "grad_norm": 0.21019339561462402, "learning_rate": 2.9940973400294743e-05, "loss": 0.0059, "step": 194950 }, { "epoch": 124.25748884639899, "grad_norm": 0.21434855461120605, "learning_rate": 2.993592493313625e-05, "loss": 0.0119, "step": 194960 }, { "epoch": 124.26386233269598, "grad_norm": 0.13276071846485138, "learning_rate": 2.993087670977521e-05, "loss": 0.0101, "step": 194970 }, { "epoch": 124.27023581899299, "grad_norm": 0.10364490747451782, "learning_rate": 2.9925828730272976e-05, "loss": 0.0084, "step": 194980 }, { "epoch": 124.27660930529, "grad_norm": 0.3415667414665222, "learning_rate": 2.992078099469089e-05, "loss": 0.0072, "step": 194990 }, { "epoch": 124.282982791587, "grad_norm": 0.31141987442970276, "learning_rate": 2.991573350309026e-05, "loss": 0.0091, "step": 195000 }, { "epoch": 124.289356277884, "grad_norm": 0.2065601348876953, "learning_rate": 2.991068625553242e-05, "loss": 0.01, "step": 195010 }, { "epoch": 124.295729764181, "grad_norm": 0.2878758907318115, "learning_rate": 2.9905639252078727e-05, "loss": 0.0081, "step": 195020 }, { "epoch": 124.30210325047801, "grad_norm": 0.12330178916454315, "learning_rate": 2.9900592492790485e-05, "loss": 0.0057, "step": 195030 }, { "epoch": 124.30847673677502, "grad_norm": 0.411286324262619, "learning_rate": 2.989554597772901e-05, "loss": 0.0095, "step": 195040 }, { "epoch": 124.31485022307201, "grad_norm": 0.5992331504821777, "learning_rate": 2.9890499706955638e-05, "loss": 0.0086, "step": 195050 }, { "epoch": 124.32122370936902, "grad_norm": 0.34840303659439087, "learning_rate": 2.988545368053169e-05, "loss": 0.0071, "step": 195060 }, { "epoch": 124.32759719566603, "grad_norm": 0.3583376705646515, "learning_rate": 2.988040789851847e-05, "loss": 0.019, "step": 195070 }, { "epoch": 124.33397068196304, "grad_norm": 0.606151819229126, "learning_rate": 2.9875362360977278e-05, "loss": 0.0074, "step": 195080 }, { "epoch": 124.34034416826003, "grad_norm": 0.2980544865131378, "learning_rate": 2.9870317067969445e-05, "loss": 0.008, "step": 195090 }, { "epoch": 124.34671765455704, "grad_norm": 0.2723540663719177, "learning_rate": 2.9865272019556256e-05, "loss": 0.0089, "step": 195100 }, { "epoch": 124.35309114085405, "grad_norm": 0.25209295749664307, "learning_rate": 2.9860227215799034e-05, "loss": 0.0064, "step": 195110 }, { "epoch": 124.35946462715106, "grad_norm": 0.19697128236293793, "learning_rate": 2.9855182656759057e-05, "loss": 0.0062, "step": 195120 }, { "epoch": 124.36583811344805, "grad_norm": 0.23616041243076324, "learning_rate": 2.9850138342497635e-05, "loss": 0.0116, "step": 195130 }, { "epoch": 124.37221159974506, "grad_norm": 0.16120995581150055, "learning_rate": 2.9845094273076057e-05, "loss": 0.0094, "step": 195140 }, { "epoch": 124.37858508604207, "grad_norm": 0.32665741443634033, "learning_rate": 2.9840050448555614e-05, "loss": 0.0085, "step": 195150 }, { "epoch": 124.38495857233907, "grad_norm": 0.0946430116891861, "learning_rate": 2.9835006868997594e-05, "loss": 0.0074, "step": 195160 }, { "epoch": 124.39133205863607, "grad_norm": 0.27223920822143555, "learning_rate": 2.9829963534463268e-05, "loss": 0.0218, "step": 195170 }, { "epoch": 124.39770554493307, "grad_norm": 0.3153654634952545, "learning_rate": 2.9824920445013937e-05, "loss": 0.009, "step": 195180 }, { "epoch": 124.40407903123008, "grad_norm": 0.09322527796030045, "learning_rate": 2.9819877600710873e-05, "loss": 0.0092, "step": 195190 }, { "epoch": 124.41045251752709, "grad_norm": 0.3795059323310852, "learning_rate": 2.981483500161535e-05, "loss": 0.0101, "step": 195200 }, { "epoch": 124.4168260038241, "grad_norm": 0.12803059816360474, "learning_rate": 2.980979264778863e-05, "loss": 0.0077, "step": 195210 }, { "epoch": 124.42319949012109, "grad_norm": 0.19239139556884766, "learning_rate": 2.9804750539291992e-05, "loss": 0.0057, "step": 195220 }, { "epoch": 124.4295729764181, "grad_norm": 0.3768424093723297, "learning_rate": 2.979970867618672e-05, "loss": 0.0111, "step": 195230 }, { "epoch": 124.43594646271511, "grad_norm": 0.6562380790710449, "learning_rate": 2.9794667058534047e-05, "loss": 0.0154, "step": 195240 }, { "epoch": 124.44231994901212, "grad_norm": 0.15717823803424835, "learning_rate": 2.9789625686395245e-05, "loss": 0.0252, "step": 195250 }, { "epoch": 124.44869343530911, "grad_norm": 0.21724513173103333, "learning_rate": 2.978458455983157e-05, "loss": 0.0224, "step": 195260 }, { "epoch": 124.45506692160612, "grad_norm": 0.18350377678871155, "learning_rate": 2.97795436789043e-05, "loss": 0.0099, "step": 195270 }, { "epoch": 124.46144040790313, "grad_norm": 0.16167892515659332, "learning_rate": 2.977450304367464e-05, "loss": 0.0077, "step": 195280 }, { "epoch": 124.46781389420013, "grad_norm": 0.2502082884311676, "learning_rate": 2.9769462654203874e-05, "loss": 0.0158, "step": 195290 }, { "epoch": 124.47418738049713, "grad_norm": 0.3745588958263397, "learning_rate": 2.976442251055323e-05, "loss": 0.0081, "step": 195300 }, { "epoch": 124.48056086679414, "grad_norm": 0.24032437801361084, "learning_rate": 2.9759382612783982e-05, "loss": 0.008, "step": 195310 }, { "epoch": 124.48693435309114, "grad_norm": 0.1810990273952484, "learning_rate": 2.9754342960957326e-05, "loss": 0.0131, "step": 195320 }, { "epoch": 124.49330783938815, "grad_norm": 0.18262255191802979, "learning_rate": 2.9749303555134512e-05, "loss": 0.0048, "step": 195330 }, { "epoch": 124.49968132568515, "grad_norm": 0.15219296514987946, "learning_rate": 2.974426439537681e-05, "loss": 0.0095, "step": 195340 }, { "epoch": 124.50605481198215, "grad_norm": 0.2906431257724762, "learning_rate": 2.973922548174539e-05, "loss": 0.0074, "step": 195350 }, { "epoch": 124.51242829827916, "grad_norm": 0.2543080747127533, "learning_rate": 2.9734186814301513e-05, "loss": 0.0072, "step": 195360 }, { "epoch": 124.51880178457617, "grad_norm": 0.27557772397994995, "learning_rate": 2.97291483931064e-05, "loss": 0.0043, "step": 195370 }, { "epoch": 124.52517527087316, "grad_norm": 0.17684492468833923, "learning_rate": 2.9724110218221298e-05, "loss": 0.0126, "step": 195380 }, { "epoch": 124.53154875717017, "grad_norm": 0.2387172132730484, "learning_rate": 2.9719072289707374e-05, "loss": 0.0071, "step": 195390 }, { "epoch": 124.53792224346718, "grad_norm": 0.27833616733551025, "learning_rate": 2.9714034607625875e-05, "loss": 0.0116, "step": 195400 }, { "epoch": 124.54429572976419, "grad_norm": 0.4206775724887848, "learning_rate": 2.970899717203802e-05, "loss": 0.0081, "step": 195410 }, { "epoch": 124.55066921606118, "grad_norm": 0.36260709166526794, "learning_rate": 2.970395998300499e-05, "loss": 0.0116, "step": 195420 }, { "epoch": 124.55704270235819, "grad_norm": 0.16343200206756592, "learning_rate": 2.9698923040588017e-05, "loss": 0.0062, "step": 195430 }, { "epoch": 124.5634161886552, "grad_norm": 0.4385316073894501, "learning_rate": 2.9693886344848294e-05, "loss": 0.0094, "step": 195440 }, { "epoch": 124.5697896749522, "grad_norm": 0.12429248541593552, "learning_rate": 2.9688849895847036e-05, "loss": 0.0112, "step": 195450 }, { "epoch": 124.5761631612492, "grad_norm": 0.3054317533969879, "learning_rate": 2.968381369364542e-05, "loss": 0.0129, "step": 195460 }, { "epoch": 124.5825366475462, "grad_norm": 0.3460492193698883, "learning_rate": 2.9678777738304642e-05, "loss": 0.008, "step": 195470 }, { "epoch": 124.58891013384321, "grad_norm": 0.18363817036151886, "learning_rate": 2.967374202988592e-05, "loss": 0.011, "step": 195480 }, { "epoch": 124.59528362014022, "grad_norm": 0.47019293904304504, "learning_rate": 2.966870656845041e-05, "loss": 0.0168, "step": 195490 }, { "epoch": 124.60165710643722, "grad_norm": 0.2745073139667511, "learning_rate": 2.9663671354059307e-05, "loss": 0.01, "step": 195500 }, { "epoch": 124.60803059273422, "grad_norm": 0.17695190012454987, "learning_rate": 2.9658636386773797e-05, "loss": 0.0095, "step": 195510 }, { "epoch": 124.61440407903123, "grad_norm": 0.1320931315422058, "learning_rate": 2.965360166665508e-05, "loss": 0.0155, "step": 195520 }, { "epoch": 124.62077756532824, "grad_norm": 0.2745317220687866, "learning_rate": 2.964856719376429e-05, "loss": 0.0125, "step": 195530 }, { "epoch": 124.62715105162523, "grad_norm": 0.273525595664978, "learning_rate": 2.9643532968162636e-05, "loss": 0.009, "step": 195540 }, { "epoch": 124.63352453792224, "grad_norm": 0.21105709671974182, "learning_rate": 2.9638498989911283e-05, "loss": 0.0091, "step": 195550 }, { "epoch": 124.63989802421925, "grad_norm": 0.43757233023643494, "learning_rate": 2.963346525907138e-05, "loss": 0.0092, "step": 195560 }, { "epoch": 124.64627151051626, "grad_norm": 0.8090785145759583, "learning_rate": 2.9628431775704097e-05, "loss": 0.0253, "step": 195570 }, { "epoch": 124.65264499681325, "grad_norm": 0.3130844235420227, "learning_rate": 2.9623398539870607e-05, "loss": 0.0106, "step": 195580 }, { "epoch": 124.65901848311026, "grad_norm": 0.47694408893585205, "learning_rate": 2.9618365551632076e-05, "loss": 0.0061, "step": 195590 }, { "epoch": 124.66539196940727, "grad_norm": 0.1901947259902954, "learning_rate": 2.9613332811049633e-05, "loss": 0.0172, "step": 195600 }, { "epoch": 124.67176545570427, "grad_norm": 0.2707735598087311, "learning_rate": 2.9608300318184446e-05, "loss": 0.0165, "step": 195610 }, { "epoch": 124.67813894200127, "grad_norm": 0.21613809466362, "learning_rate": 2.9603268073097657e-05, "loss": 0.0048, "step": 195620 }, { "epoch": 124.68451242829828, "grad_norm": 0.6139352321624756, "learning_rate": 2.9598236075850444e-05, "loss": 0.0181, "step": 195630 }, { "epoch": 124.69088591459528, "grad_norm": 0.549331784248352, "learning_rate": 2.9593204326503904e-05, "loss": 0.0196, "step": 195640 }, { "epoch": 124.69725940089229, "grad_norm": 0.17842644453048706, "learning_rate": 2.9588172825119198e-05, "loss": 0.0064, "step": 195650 }, { "epoch": 124.70363288718929, "grad_norm": 0.14751750230789185, "learning_rate": 2.9583141571757482e-05, "loss": 0.0045, "step": 195660 }, { "epoch": 124.7100063734863, "grad_norm": 0.17491251230239868, "learning_rate": 2.9578110566479855e-05, "loss": 0.0106, "step": 195670 }, { "epoch": 124.7163798597833, "grad_norm": 0.27773991227149963, "learning_rate": 2.957307980934747e-05, "loss": 0.0049, "step": 195680 }, { "epoch": 124.72275334608031, "grad_norm": 0.20831206440925598, "learning_rate": 2.9568049300421447e-05, "loss": 0.0069, "step": 195690 }, { "epoch": 124.7291268323773, "grad_norm": 0.28060391545295715, "learning_rate": 2.9563019039762935e-05, "loss": 0.0093, "step": 195700 }, { "epoch": 124.73550031867431, "grad_norm": 0.24093586206436157, "learning_rate": 2.955798902743302e-05, "loss": 0.0046, "step": 195710 }, { "epoch": 124.74187380497132, "grad_norm": 0.19031812250614166, "learning_rate": 2.9552959263492835e-05, "loss": 0.0062, "step": 195720 }, { "epoch": 124.74824729126833, "grad_norm": 0.1629791408777237, "learning_rate": 2.954792974800351e-05, "loss": 0.0082, "step": 195730 }, { "epoch": 124.75462077756532, "grad_norm": 0.5118164420127869, "learning_rate": 2.954290048102616e-05, "loss": 0.0105, "step": 195740 }, { "epoch": 124.76099426386233, "grad_norm": 0.14322689175605774, "learning_rate": 2.9537871462621858e-05, "loss": 0.0044, "step": 195750 }, { "epoch": 124.76736775015934, "grad_norm": 0.4340987503528595, "learning_rate": 2.953284269285175e-05, "loss": 0.0071, "step": 195760 }, { "epoch": 124.77374123645635, "grad_norm": 0.2338060885667801, "learning_rate": 2.9527814171776923e-05, "loss": 0.0092, "step": 195770 }, { "epoch": 124.78011472275334, "grad_norm": 0.33366355299949646, "learning_rate": 2.9522785899458493e-05, "loss": 0.0068, "step": 195780 }, { "epoch": 124.78648820905035, "grad_norm": 0.19467169046401978, "learning_rate": 2.951775787595754e-05, "loss": 0.0098, "step": 195790 }, { "epoch": 124.79286169534736, "grad_norm": 0.1896657794713974, "learning_rate": 2.9512730101335175e-05, "loss": 0.0069, "step": 195800 }, { "epoch": 124.79923518164436, "grad_norm": 0.2471681386232376, "learning_rate": 2.9507702575652476e-05, "loss": 0.006, "step": 195810 }, { "epoch": 124.80560866794136, "grad_norm": 0.40041443705558777, "learning_rate": 2.9502675298970527e-05, "loss": 0.0125, "step": 195820 }, { "epoch": 124.81198215423836, "grad_norm": 0.4289628267288208, "learning_rate": 2.9497648271350442e-05, "loss": 0.0169, "step": 195830 }, { "epoch": 124.81835564053537, "grad_norm": 0.3060722053050995, "learning_rate": 2.949262149285329e-05, "loss": 0.0068, "step": 195840 }, { "epoch": 124.82472912683238, "grad_norm": 0.18447661399841309, "learning_rate": 2.948759496354014e-05, "loss": 0.0084, "step": 195850 }, { "epoch": 124.83110261312939, "grad_norm": 0.35352450609207153, "learning_rate": 2.948256868347207e-05, "loss": 0.0085, "step": 195860 }, { "epoch": 124.83747609942638, "grad_norm": 0.33736124634742737, "learning_rate": 2.9477542652710188e-05, "loss": 0.013, "step": 195870 }, { "epoch": 124.84384958572339, "grad_norm": 0.17685188353061676, "learning_rate": 2.9472516871315525e-05, "loss": 0.0071, "step": 195880 }, { "epoch": 124.8502230720204, "grad_norm": 0.24573083221912384, "learning_rate": 2.946749133934915e-05, "loss": 0.0088, "step": 195890 }, { "epoch": 124.8565965583174, "grad_norm": 0.37333130836486816, "learning_rate": 2.9462466056872152e-05, "loss": 0.0035, "step": 195900 }, { "epoch": 124.8629700446144, "grad_norm": 0.1525712013244629, "learning_rate": 2.9457441023945597e-05, "loss": 0.0059, "step": 195910 }, { "epoch": 124.86934353091141, "grad_norm": 0.17384737730026245, "learning_rate": 2.945241624063051e-05, "loss": 0.0053, "step": 195920 }, { "epoch": 124.87571701720842, "grad_norm": 0.10676485300064087, "learning_rate": 2.944739170698797e-05, "loss": 0.0058, "step": 195930 }, { "epoch": 124.88209050350542, "grad_norm": 0.20460061728954315, "learning_rate": 2.944236742307902e-05, "loss": 0.0093, "step": 195940 }, { "epoch": 124.88846398980242, "grad_norm": 0.27117428183555603, "learning_rate": 2.9437343388964737e-05, "loss": 0.0111, "step": 195950 }, { "epoch": 124.89483747609943, "grad_norm": 0.20363982021808624, "learning_rate": 2.9432319604706126e-05, "loss": 0.0101, "step": 195960 }, { "epoch": 124.90121096239643, "grad_norm": 0.2544470429420471, "learning_rate": 2.942729607036425e-05, "loss": 0.0112, "step": 195970 }, { "epoch": 124.90758444869344, "grad_norm": 0.33448684215545654, "learning_rate": 2.9422272786000172e-05, "loss": 0.0096, "step": 195980 }, { "epoch": 124.91395793499044, "grad_norm": 0.24747900664806366, "learning_rate": 2.941724975167489e-05, "loss": 0.011, "step": 195990 }, { "epoch": 124.92033142128744, "grad_norm": 0.2122986912727356, "learning_rate": 2.9412226967449452e-05, "loss": 0.0131, "step": 196000 }, { "epoch": 124.92670490758445, "grad_norm": 0.41425976157188416, "learning_rate": 2.9407204433384905e-05, "loss": 0.0186, "step": 196010 }, { "epoch": 124.93307839388146, "grad_norm": 0.32091447710990906, "learning_rate": 2.940218214954228e-05, "loss": 0.0113, "step": 196020 }, { "epoch": 124.93945188017845, "grad_norm": 0.13205525279045105, "learning_rate": 2.9397160115982562e-05, "loss": 0.0076, "step": 196030 }, { "epoch": 124.94582536647546, "grad_norm": 0.13531221449375153, "learning_rate": 2.939213833276681e-05, "loss": 0.006, "step": 196040 }, { "epoch": 124.95219885277247, "grad_norm": 0.24473237991333008, "learning_rate": 2.9387116799956048e-05, "loss": 0.0093, "step": 196050 }, { "epoch": 124.95857233906948, "grad_norm": 0.16550523042678833, "learning_rate": 2.9382095517611264e-05, "loss": 0.0069, "step": 196060 }, { "epoch": 124.96494582536647, "grad_norm": 0.30250614881515503, "learning_rate": 2.9377074485793486e-05, "loss": 0.0084, "step": 196070 }, { "epoch": 124.97131931166348, "grad_norm": 0.1277596354484558, "learning_rate": 2.9372053704563728e-05, "loss": 0.0096, "step": 196080 }, { "epoch": 124.97769279796049, "grad_norm": 0.26038870215415955, "learning_rate": 2.9367033173983006e-05, "loss": 0.0136, "step": 196090 }, { "epoch": 124.9840662842575, "grad_norm": 0.481698215007782, "learning_rate": 2.936201289411229e-05, "loss": 0.0147, "step": 196100 }, { "epoch": 124.99043977055449, "grad_norm": 0.11616504192352295, "learning_rate": 2.9356992865012612e-05, "loss": 0.0085, "step": 196110 }, { "epoch": 124.9968132568515, "grad_norm": 0.2974666953086853, "learning_rate": 2.935197308674498e-05, "loss": 0.0101, "step": 196120 }, { "epoch": 125.0031867431485, "grad_norm": 0.16829752922058105, "learning_rate": 2.9346953559370337e-05, "loss": 0.0077, "step": 196130 }, { "epoch": 125.00956022944551, "grad_norm": 0.19231849908828735, "learning_rate": 2.9341934282949724e-05, "loss": 0.0077, "step": 196140 }, { "epoch": 125.0159337157425, "grad_norm": 0.26085761189460754, "learning_rate": 2.9336915257544106e-05, "loss": 0.0068, "step": 196150 }, { "epoch": 125.02230720203951, "grad_norm": 0.27187037467956543, "learning_rate": 2.9331896483214494e-05, "loss": 0.0163, "step": 196160 }, { "epoch": 125.02868068833652, "grad_norm": 0.2064451277256012, "learning_rate": 2.9326877960021836e-05, "loss": 0.0051, "step": 196170 }, { "epoch": 125.03505417463353, "grad_norm": 0.2781285345554352, "learning_rate": 2.9321859688027132e-05, "loss": 0.0047, "step": 196180 }, { "epoch": 125.04142766093052, "grad_norm": 0.17986169457435608, "learning_rate": 2.9316841667291356e-05, "loss": 0.0063, "step": 196190 }, { "epoch": 125.04780114722753, "grad_norm": 0.3693870007991791, "learning_rate": 2.9311823897875503e-05, "loss": 0.0118, "step": 196200 }, { "epoch": 125.05417463352454, "grad_norm": 0.11500784009695053, "learning_rate": 2.9306806379840502e-05, "loss": 0.0109, "step": 196210 }, { "epoch": 125.06054811982155, "grad_norm": 0.21102988719940186, "learning_rate": 2.9301789113247334e-05, "loss": 0.003, "step": 196220 }, { "epoch": 125.06692160611854, "grad_norm": 0.2638266980648041, "learning_rate": 2.9296772098157e-05, "loss": 0.0099, "step": 196230 }, { "epoch": 125.07329509241555, "grad_norm": 0.1605229377746582, "learning_rate": 2.9291755334630405e-05, "loss": 0.0105, "step": 196240 }, { "epoch": 125.07966857871256, "grad_norm": 0.24191419780254364, "learning_rate": 2.928673882272854e-05, "loss": 0.0101, "step": 196250 }, { "epoch": 125.08604206500956, "grad_norm": 0.20784182846546173, "learning_rate": 2.928172256251235e-05, "loss": 0.0094, "step": 196260 }, { "epoch": 125.09241555130656, "grad_norm": 0.8278200626373291, "learning_rate": 2.9276706554042814e-05, "loss": 0.0108, "step": 196270 }, { "epoch": 125.09878903760357, "grad_norm": 0.32401642203330994, "learning_rate": 2.927169079738084e-05, "loss": 0.0072, "step": 196280 }, { "epoch": 125.10516252390057, "grad_norm": 0.4524100720882416, "learning_rate": 2.9266675292587396e-05, "loss": 0.0076, "step": 196290 }, { "epoch": 125.11153601019758, "grad_norm": 0.5503368973731995, "learning_rate": 2.9261660039723438e-05, "loss": 0.0073, "step": 196300 }, { "epoch": 125.11790949649458, "grad_norm": 0.2043905109167099, "learning_rate": 2.925664503884987e-05, "loss": 0.0059, "step": 196310 }, { "epoch": 125.12428298279158, "grad_norm": 0.3287336826324463, "learning_rate": 2.9251630290027654e-05, "loss": 0.0107, "step": 196320 }, { "epoch": 125.13065646908859, "grad_norm": 0.686756432056427, "learning_rate": 2.9246615793317723e-05, "loss": 0.014, "step": 196330 }, { "epoch": 125.1370299553856, "grad_norm": 0.3008434772491455, "learning_rate": 2.9241601548781016e-05, "loss": 0.0081, "step": 196340 }, { "epoch": 125.1434034416826, "grad_norm": 0.22148941457271576, "learning_rate": 2.9236587556478435e-05, "loss": 0.0066, "step": 196350 }, { "epoch": 125.1497769279796, "grad_norm": 0.38935595750808716, "learning_rate": 2.9231573816470914e-05, "loss": 0.0102, "step": 196360 }, { "epoch": 125.15615041427661, "grad_norm": 0.1647345870733261, "learning_rate": 2.9226560328819396e-05, "loss": 0.0069, "step": 196370 }, { "epoch": 125.16252390057362, "grad_norm": 0.23153837025165558, "learning_rate": 2.9221547093584777e-05, "loss": 0.0081, "step": 196380 }, { "epoch": 125.16889738687061, "grad_norm": 0.20046988129615784, "learning_rate": 2.9216534110827976e-05, "loss": 0.0085, "step": 196390 }, { "epoch": 125.17527087316762, "grad_norm": 0.20695732533931732, "learning_rate": 2.921152138060991e-05, "loss": 0.0111, "step": 196400 }, { "epoch": 125.18164435946463, "grad_norm": 0.26992568373680115, "learning_rate": 2.9206508902991485e-05, "loss": 0.0101, "step": 196410 }, { "epoch": 125.18801784576164, "grad_norm": 0.21153752505779266, "learning_rate": 2.920149667803362e-05, "loss": 0.0167, "step": 196420 }, { "epoch": 125.19439133205863, "grad_norm": 0.17092803120613098, "learning_rate": 2.919648470579719e-05, "loss": 0.0057, "step": 196430 }, { "epoch": 125.20076481835564, "grad_norm": 0.1919468492269516, "learning_rate": 2.9191472986343138e-05, "loss": 0.0072, "step": 196440 }, { "epoch": 125.20713830465264, "grad_norm": 0.20765632390975952, "learning_rate": 2.91864615197323e-05, "loss": 0.0052, "step": 196450 }, { "epoch": 125.21351179094965, "grad_norm": 0.12564438581466675, "learning_rate": 2.9181450306025633e-05, "loss": 0.0087, "step": 196460 }, { "epoch": 125.21988527724666, "grad_norm": 0.19719967246055603, "learning_rate": 2.917643934528398e-05, "loss": 0.0056, "step": 196470 }, { "epoch": 125.22625876354365, "grad_norm": 0.23955143988132477, "learning_rate": 2.9171428637568282e-05, "loss": 0.0106, "step": 196480 }, { "epoch": 125.23263224984066, "grad_norm": 0.3045954704284668, "learning_rate": 2.916641818293936e-05, "loss": 0.0049, "step": 196490 }, { "epoch": 125.23900573613767, "grad_norm": 0.17071834206581116, "learning_rate": 2.9161407981458144e-05, "loss": 0.0093, "step": 196500 }, { "epoch": 125.24537922243468, "grad_norm": 0.30231741070747375, "learning_rate": 2.9156398033185472e-05, "loss": 0.0137, "step": 196510 }, { "epoch": 125.25175270873167, "grad_norm": 0.2472502887248993, "learning_rate": 2.9151388338182282e-05, "loss": 0.0165, "step": 196520 }, { "epoch": 125.25812619502868, "grad_norm": 0.1514027714729309, "learning_rate": 2.9146378896509374e-05, "loss": 0.0065, "step": 196530 }, { "epoch": 125.26449968132569, "grad_norm": 0.226918563246727, "learning_rate": 2.9141369708227673e-05, "loss": 0.0074, "step": 196540 }, { "epoch": 125.2708731676227, "grad_norm": 0.4695672392845154, "learning_rate": 2.913636077339802e-05, "loss": 0.0112, "step": 196550 }, { "epoch": 125.27724665391969, "grad_norm": 0.244410902261734, "learning_rate": 2.9131352092081286e-05, "loss": 0.007, "step": 196560 }, { "epoch": 125.2836201402167, "grad_norm": 0.4626779854297638, "learning_rate": 2.912634366433831e-05, "loss": 0.0066, "step": 196570 }, { "epoch": 125.2899936265137, "grad_norm": 0.20521649718284607, "learning_rate": 2.9121335490229972e-05, "loss": 0.0117, "step": 196580 }, { "epoch": 125.29636711281071, "grad_norm": 0.20352604985237122, "learning_rate": 2.9116327569817138e-05, "loss": 0.0149, "step": 196590 }, { "epoch": 125.30274059910771, "grad_norm": 0.13538797199726105, "learning_rate": 2.911131990316063e-05, "loss": 0.0062, "step": 196600 }, { "epoch": 125.30911408540472, "grad_norm": 0.17445415258407593, "learning_rate": 2.910631249032129e-05, "loss": 0.0056, "step": 196610 }, { "epoch": 125.31548757170172, "grad_norm": 0.36246395111083984, "learning_rate": 2.9101305331360016e-05, "loss": 0.0103, "step": 196620 }, { "epoch": 125.32186105799873, "grad_norm": 0.2716695964336395, "learning_rate": 2.9096298426337587e-05, "loss": 0.0061, "step": 196630 }, { "epoch": 125.32823454429573, "grad_norm": 0.16645899415016174, "learning_rate": 2.9091291775314877e-05, "loss": 0.0143, "step": 196640 }, { "epoch": 125.33460803059273, "grad_norm": 0.1416938155889511, "learning_rate": 2.9086285378352696e-05, "loss": 0.0068, "step": 196650 }, { "epoch": 125.34098151688974, "grad_norm": 0.46119752526283264, "learning_rate": 2.908127923551194e-05, "loss": 0.0072, "step": 196660 }, { "epoch": 125.34735500318675, "grad_norm": 0.313097208738327, "learning_rate": 2.907627334685334e-05, "loss": 0.0056, "step": 196670 }, { "epoch": 125.35372848948374, "grad_norm": 0.12267223745584488, "learning_rate": 2.90712677124378e-05, "loss": 0.0145, "step": 196680 }, { "epoch": 125.36010197578075, "grad_norm": 0.47033122181892395, "learning_rate": 2.9066262332326115e-05, "loss": 0.0119, "step": 196690 }, { "epoch": 125.36647546207776, "grad_norm": 0.2758324444293976, "learning_rate": 2.9061257206579106e-05, "loss": 0.0072, "step": 196700 }, { "epoch": 125.37284894837477, "grad_norm": 0.1466747224330902, "learning_rate": 2.905625233525758e-05, "loss": 0.0089, "step": 196710 }, { "epoch": 125.37922243467176, "grad_norm": 0.31110793352127075, "learning_rate": 2.9051247718422374e-05, "loss": 0.0066, "step": 196720 }, { "epoch": 125.38559592096877, "grad_norm": 0.621121883392334, "learning_rate": 2.9046243356134295e-05, "loss": 0.009, "step": 196730 }, { "epoch": 125.39196940726578, "grad_norm": 0.2956647276878357, "learning_rate": 2.9041239248454132e-05, "loss": 0.0112, "step": 196740 }, { "epoch": 125.39834289356278, "grad_norm": 0.13216300308704376, "learning_rate": 2.903623539544269e-05, "loss": 0.0064, "step": 196750 }, { "epoch": 125.40471637985978, "grad_norm": 0.16943196952342987, "learning_rate": 2.9031231797160795e-05, "loss": 0.0055, "step": 196760 }, { "epoch": 125.41108986615679, "grad_norm": 0.32361194491386414, "learning_rate": 2.9026228453669234e-05, "loss": 0.0086, "step": 196770 }, { "epoch": 125.4174633524538, "grad_norm": 0.15293119847774506, "learning_rate": 2.9021225365028802e-05, "loss": 0.0072, "step": 196780 }, { "epoch": 125.4238368387508, "grad_norm": 0.17287875711917877, "learning_rate": 2.9016222531300263e-05, "loss": 0.0076, "step": 196790 }, { "epoch": 125.4302103250478, "grad_norm": 0.36638689041137695, "learning_rate": 2.901121995254446e-05, "loss": 0.0086, "step": 196800 }, { "epoch": 125.4365838113448, "grad_norm": 0.32797807455062866, "learning_rate": 2.9006217628822146e-05, "loss": 0.0067, "step": 196810 }, { "epoch": 125.44295729764181, "grad_norm": 0.39272716641426086, "learning_rate": 2.900121556019411e-05, "loss": 0.014, "step": 196820 }, { "epoch": 125.44933078393882, "grad_norm": 0.24111925065517426, "learning_rate": 2.8996213746721113e-05, "loss": 0.0063, "step": 196830 }, { "epoch": 125.45570427023581, "grad_norm": 0.34341537952423096, "learning_rate": 2.8991212188463967e-05, "loss": 0.0111, "step": 196840 }, { "epoch": 125.46207775653282, "grad_norm": 0.26800042390823364, "learning_rate": 2.8986210885483433e-05, "loss": 0.0064, "step": 196850 }, { "epoch": 125.46845124282983, "grad_norm": 0.3929288685321808, "learning_rate": 2.8981209837840272e-05, "loss": 0.0072, "step": 196860 }, { "epoch": 125.47482472912684, "grad_norm": 0.14261744916439056, "learning_rate": 2.897620904559526e-05, "loss": 0.0049, "step": 196870 }, { "epoch": 125.48119821542383, "grad_norm": 0.37415969371795654, "learning_rate": 2.8971208508809134e-05, "loss": 0.0092, "step": 196880 }, { "epoch": 125.48757170172084, "grad_norm": 0.17268061637878418, "learning_rate": 2.89662082275427e-05, "loss": 0.0098, "step": 196890 }, { "epoch": 125.49394518801785, "grad_norm": 0.2813112735748291, "learning_rate": 2.8961208201856698e-05, "loss": 0.0066, "step": 196900 }, { "epoch": 125.50031867431485, "grad_norm": 0.2182254195213318, "learning_rate": 2.895620843181188e-05, "loss": 0.0097, "step": 196910 }, { "epoch": 125.50669216061185, "grad_norm": 0.32088884711265564, "learning_rate": 2.8951208917468975e-05, "loss": 0.0122, "step": 196920 }, { "epoch": 125.51306564690886, "grad_norm": 0.22960297763347626, "learning_rate": 2.8946209658888778e-05, "loss": 0.006, "step": 196930 }, { "epoch": 125.51943913320586, "grad_norm": 0.3567044734954834, "learning_rate": 2.8941210656132013e-05, "loss": 0.0109, "step": 196940 }, { "epoch": 125.52581261950287, "grad_norm": 0.19752779603004456, "learning_rate": 2.8936211909259416e-05, "loss": 0.0053, "step": 196950 }, { "epoch": 125.53218610579987, "grad_norm": 0.16517765820026398, "learning_rate": 2.8931213418331714e-05, "loss": 0.0057, "step": 196960 }, { "epoch": 125.53855959209687, "grad_norm": 0.44586023688316345, "learning_rate": 2.8926215183409683e-05, "loss": 0.0123, "step": 196970 }, { "epoch": 125.54493307839388, "grad_norm": 0.12886792421340942, "learning_rate": 2.8921217204554035e-05, "loss": 0.021, "step": 196980 }, { "epoch": 125.55130656469089, "grad_norm": 0.18880446255207062, "learning_rate": 2.89162194818255e-05, "loss": 0.0074, "step": 196990 }, { "epoch": 125.55768005098788, "grad_norm": 0.20462855696678162, "learning_rate": 2.891122201528478e-05, "loss": 0.0135, "step": 197000 }, { "epoch": 125.56405353728489, "grad_norm": 0.5157061219215393, "learning_rate": 2.8906224804992666e-05, "loss": 0.0129, "step": 197010 }, { "epoch": 125.5704270235819, "grad_norm": 0.22583110630512238, "learning_rate": 2.89012278510098e-05, "loss": 0.006, "step": 197020 }, { "epoch": 125.57680050987891, "grad_norm": 0.3290700614452362, "learning_rate": 2.8896231153396953e-05, "loss": 0.0149, "step": 197030 }, { "epoch": 125.5831739961759, "grad_norm": 0.16598835587501526, "learning_rate": 2.8891234712214798e-05, "loss": 0.0077, "step": 197040 }, { "epoch": 125.58954748247291, "grad_norm": 0.20109929144382477, "learning_rate": 2.8886238527524122e-05, "loss": 0.0103, "step": 197050 }, { "epoch": 125.59592096876992, "grad_norm": 0.2176167219877243, "learning_rate": 2.8881242599385534e-05, "loss": 0.0057, "step": 197060 }, { "epoch": 125.60229445506693, "grad_norm": 0.1446818858385086, "learning_rate": 2.8876246927859808e-05, "loss": 0.0078, "step": 197070 }, { "epoch": 125.60866794136393, "grad_norm": 0.5060468912124634, "learning_rate": 2.887125151300763e-05, "loss": 0.0085, "step": 197080 }, { "epoch": 125.61504142766093, "grad_norm": 0.11048492789268494, "learning_rate": 2.88662563548897e-05, "loss": 0.0075, "step": 197090 }, { "epoch": 125.62141491395793, "grad_norm": 0.7820450067520142, "learning_rate": 2.886126145356668e-05, "loss": 0.0076, "step": 197100 }, { "epoch": 125.62778840025494, "grad_norm": 0.2526898682117462, "learning_rate": 2.885626680909931e-05, "loss": 0.0058, "step": 197110 }, { "epoch": 125.63416188655195, "grad_norm": 0.18632888793945312, "learning_rate": 2.8851272421548263e-05, "loss": 0.0079, "step": 197120 }, { "epoch": 125.64053537284894, "grad_norm": 0.16216059029102325, "learning_rate": 2.8846278290974227e-05, "loss": 0.0143, "step": 197130 }, { "epoch": 125.64690885914595, "grad_norm": 0.15594497323036194, "learning_rate": 2.8841284417437864e-05, "loss": 0.0039, "step": 197140 }, { "epoch": 125.65328234544296, "grad_norm": 0.6816205382347107, "learning_rate": 2.8836290800999888e-05, "loss": 0.0154, "step": 197150 }, { "epoch": 125.65965583173997, "grad_norm": 0.1644778549671173, "learning_rate": 2.883129744172096e-05, "loss": 0.0088, "step": 197160 }, { "epoch": 125.66602931803696, "grad_norm": 0.0899663046002388, "learning_rate": 2.8826304339661757e-05, "loss": 0.0042, "step": 197170 }, { "epoch": 125.67240280433397, "grad_norm": 0.13184207677841187, "learning_rate": 2.882131149488293e-05, "loss": 0.0079, "step": 197180 }, { "epoch": 125.67877629063098, "grad_norm": 0.33414265513420105, "learning_rate": 2.88163189074452e-05, "loss": 0.0063, "step": 197190 }, { "epoch": 125.68514977692799, "grad_norm": 0.24380983412265778, "learning_rate": 2.8811326577409157e-05, "loss": 0.0045, "step": 197200 }, { "epoch": 125.69152326322498, "grad_norm": 0.23243363201618195, "learning_rate": 2.8806334504835525e-05, "loss": 0.0077, "step": 197210 }, { "epoch": 125.69789674952199, "grad_norm": 0.17076542973518372, "learning_rate": 2.8801342689784916e-05, "loss": 0.0063, "step": 197220 }, { "epoch": 125.704270235819, "grad_norm": 0.3009606897830963, "learning_rate": 2.8796351132318046e-05, "loss": 0.0101, "step": 197230 }, { "epoch": 125.710643722116, "grad_norm": 0.4789252281188965, "learning_rate": 2.879135983249549e-05, "loss": 0.0128, "step": 197240 }, { "epoch": 125.717017208413, "grad_norm": 0.6188280582427979, "learning_rate": 2.878636879037796e-05, "loss": 0.0075, "step": 197250 }, { "epoch": 125.72339069471, "grad_norm": 0.3010062873363495, "learning_rate": 2.878137800602608e-05, "loss": 0.0097, "step": 197260 }, { "epoch": 125.72976418100701, "grad_norm": 0.1720682829618454, "learning_rate": 2.877638747950049e-05, "loss": 0.0062, "step": 197270 }, { "epoch": 125.73613766730402, "grad_norm": 0.21856006979942322, "learning_rate": 2.8771397210861807e-05, "loss": 0.02, "step": 197280 }, { "epoch": 125.74251115360101, "grad_norm": 0.23255573213100433, "learning_rate": 2.876640720017071e-05, "loss": 0.0071, "step": 197290 }, { "epoch": 125.74888463989802, "grad_norm": 0.3347095251083374, "learning_rate": 2.8761417447487816e-05, "loss": 0.0113, "step": 197300 }, { "epoch": 125.75525812619503, "grad_norm": 0.3669491708278656, "learning_rate": 2.8756427952873744e-05, "loss": 0.0096, "step": 197310 }, { "epoch": 125.76163161249204, "grad_norm": 0.19584420323371887, "learning_rate": 2.8751438716389124e-05, "loss": 0.0072, "step": 197320 }, { "epoch": 125.76800509878903, "grad_norm": 0.2670430541038513, "learning_rate": 2.8746449738094617e-05, "loss": 0.005, "step": 197330 }, { "epoch": 125.77437858508604, "grad_norm": 0.15251244604587555, "learning_rate": 2.8741461018050776e-05, "loss": 0.0174, "step": 197340 }, { "epoch": 125.78075207138305, "grad_norm": 0.19011330604553223, "learning_rate": 2.8736472556318272e-05, "loss": 0.0055, "step": 197350 }, { "epoch": 125.78712555768006, "grad_norm": 0.44587957859039307, "learning_rate": 2.8731484352957684e-05, "loss": 0.0062, "step": 197360 }, { "epoch": 125.79349904397705, "grad_norm": 0.25222015380859375, "learning_rate": 2.8726496408029678e-05, "loss": 0.0111, "step": 197370 }, { "epoch": 125.79987253027406, "grad_norm": 0.315752238035202, "learning_rate": 2.8721508721594792e-05, "loss": 0.0065, "step": 197380 }, { "epoch": 125.80624601657107, "grad_norm": 0.12070008367300034, "learning_rate": 2.871652129371368e-05, "loss": 0.0058, "step": 197390 }, { "epoch": 125.81261950286807, "grad_norm": 0.26489463448524475, "learning_rate": 2.8711534124446915e-05, "loss": 0.0072, "step": 197400 }, { "epoch": 125.81899298916507, "grad_norm": 0.4508625566959381, "learning_rate": 2.8706547213855144e-05, "loss": 0.0109, "step": 197410 }, { "epoch": 125.82536647546208, "grad_norm": 0.36500680446624756, "learning_rate": 2.870156056199889e-05, "loss": 0.0108, "step": 197420 }, { "epoch": 125.83173996175908, "grad_norm": 0.2514684796333313, "learning_rate": 2.8696574168938805e-05, "loss": 0.0082, "step": 197430 }, { "epoch": 125.83811344805609, "grad_norm": 0.15965072810649872, "learning_rate": 2.869158803473546e-05, "loss": 0.0076, "step": 197440 }, { "epoch": 125.84448693435309, "grad_norm": 0.18650934100151062, "learning_rate": 2.868660215944944e-05, "loss": 0.0075, "step": 197450 }, { "epoch": 125.8508604206501, "grad_norm": 0.2209753394126892, "learning_rate": 2.8681616543141304e-05, "loss": 0.0038, "step": 197460 }, { "epoch": 125.8572339069471, "grad_norm": 0.43495649099349976, "learning_rate": 2.867663118587167e-05, "loss": 0.0065, "step": 197470 }, { "epoch": 125.86360739324411, "grad_norm": 0.4003203809261322, "learning_rate": 2.8671646087701108e-05, "loss": 0.0073, "step": 197480 }, { "epoch": 125.8699808795411, "grad_norm": 0.21371005475521088, "learning_rate": 2.8666661248690186e-05, "loss": 0.0102, "step": 197490 }, { "epoch": 125.87635436583811, "grad_norm": 0.2036741077899933, "learning_rate": 2.866167666889944e-05, "loss": 0.0064, "step": 197500 }, { "epoch": 125.88272785213512, "grad_norm": 0.17573480308055878, "learning_rate": 2.86566923483895e-05, "loss": 0.0064, "step": 197510 }, { "epoch": 125.88910133843213, "grad_norm": 0.16813971102237701, "learning_rate": 2.865170828722089e-05, "loss": 0.0111, "step": 197520 }, { "epoch": 125.89547482472912, "grad_norm": 0.20874767005443573, "learning_rate": 2.864672448545419e-05, "loss": 0.008, "step": 197530 }, { "epoch": 125.90184831102613, "grad_norm": 0.18716560304164886, "learning_rate": 2.864174094314992e-05, "loss": 0.0073, "step": 197540 }, { "epoch": 125.90822179732314, "grad_norm": 0.33911046385765076, "learning_rate": 2.863675766036869e-05, "loss": 0.0066, "step": 197550 }, { "epoch": 125.91459528362014, "grad_norm": 0.3855462968349457, "learning_rate": 2.8631774637171022e-05, "loss": 0.0082, "step": 197560 }, { "epoch": 125.92096876991714, "grad_norm": 0.22067157924175262, "learning_rate": 2.8626791873617463e-05, "loss": 0.0063, "step": 197570 }, { "epoch": 125.92734225621415, "grad_norm": 0.26538386940956116, "learning_rate": 2.862180936976857e-05, "loss": 0.0072, "step": 197580 }, { "epoch": 125.93371574251115, "grad_norm": 0.5340567827224731, "learning_rate": 2.8616827125684853e-05, "loss": 0.0131, "step": 197590 }, { "epoch": 125.94008922880816, "grad_norm": 0.4617607593536377, "learning_rate": 2.8611845141426895e-05, "loss": 0.0111, "step": 197600 }, { "epoch": 125.94646271510516, "grad_norm": 0.19652675092220306, "learning_rate": 2.8606863417055212e-05, "loss": 0.0073, "step": 197610 }, { "epoch": 125.95283620140216, "grad_norm": 0.20543380081653595, "learning_rate": 2.860188195263034e-05, "loss": 0.0087, "step": 197620 }, { "epoch": 125.95920968769917, "grad_norm": 0.20030157268047333, "learning_rate": 2.8596900748212785e-05, "loss": 0.0081, "step": 197630 }, { "epoch": 125.96558317399618, "grad_norm": 0.3677026331424713, "learning_rate": 2.859191980386311e-05, "loss": 0.0077, "step": 197640 }, { "epoch": 125.97195666029317, "grad_norm": 0.6349195837974548, "learning_rate": 2.858693911964182e-05, "loss": 0.0191, "step": 197650 }, { "epoch": 125.97833014659018, "grad_norm": 0.2655300796031952, "learning_rate": 2.8581958695609447e-05, "loss": 0.009, "step": 197660 }, { "epoch": 125.98470363288719, "grad_norm": 0.23030579090118408, "learning_rate": 2.857697853182647e-05, "loss": 0.0074, "step": 197670 }, { "epoch": 125.9910771191842, "grad_norm": 0.8053467273712158, "learning_rate": 2.8571998628353453e-05, "loss": 0.0248, "step": 197680 }, { "epoch": 125.99745060548119, "grad_norm": 0.2174738198518753, "learning_rate": 2.856701898525088e-05, "loss": 0.015, "step": 197690 }, { "epoch": 126.0038240917782, "grad_norm": 0.20798933506011963, "learning_rate": 2.856203960257927e-05, "loss": 0.0104, "step": 197700 }, { "epoch": 126.0101975780752, "grad_norm": 0.24492913484573364, "learning_rate": 2.8557060480399095e-05, "loss": 0.0108, "step": 197710 }, { "epoch": 126.01657106437222, "grad_norm": 0.5725833177566528, "learning_rate": 2.85520816187709e-05, "loss": 0.0106, "step": 197720 }, { "epoch": 126.02294455066922, "grad_norm": 0.4352588355541229, "learning_rate": 2.8547103017755167e-05, "loss": 0.0091, "step": 197730 }, { "epoch": 126.02931803696622, "grad_norm": 0.17685511708259583, "learning_rate": 2.8542124677412385e-05, "loss": 0.0081, "step": 197740 }, { "epoch": 126.03569152326322, "grad_norm": 0.3888845145702362, "learning_rate": 2.8537146597803027e-05, "loss": 0.0084, "step": 197750 }, { "epoch": 126.04206500956023, "grad_norm": 0.14820259809494019, "learning_rate": 2.8532168778987643e-05, "loss": 0.0043, "step": 197760 }, { "epoch": 126.04843849585724, "grad_norm": 0.3113950788974762, "learning_rate": 2.852719122102664e-05, "loss": 0.0162, "step": 197770 }, { "epoch": 126.05481198215423, "grad_norm": 0.14589384198188782, "learning_rate": 2.8522213923980556e-05, "loss": 0.0049, "step": 197780 }, { "epoch": 126.06118546845124, "grad_norm": 0.2854650616645813, "learning_rate": 2.8517236887909826e-05, "loss": 0.0093, "step": 197790 }, { "epoch": 126.06755895474825, "grad_norm": 0.19038431346416473, "learning_rate": 2.8512260112874996e-05, "loss": 0.0052, "step": 197800 }, { "epoch": 126.07393244104526, "grad_norm": 0.29014313220977783, "learning_rate": 2.8507283598936452e-05, "loss": 0.0054, "step": 197810 }, { "epoch": 126.08030592734225, "grad_norm": 0.09691594541072845, "learning_rate": 2.850230734615472e-05, "loss": 0.0053, "step": 197820 }, { "epoch": 126.08667941363926, "grad_norm": 0.14593985676765442, "learning_rate": 2.8497331354590262e-05, "loss": 0.0071, "step": 197830 }, { "epoch": 126.09305289993627, "grad_norm": 0.4179464280605316, "learning_rate": 2.8492355624303525e-05, "loss": 0.0074, "step": 197840 }, { "epoch": 126.09942638623328, "grad_norm": 0.23104937374591827, "learning_rate": 2.848738015535495e-05, "loss": 0.0049, "step": 197850 }, { "epoch": 126.10579987253027, "grad_norm": 0.16739416122436523, "learning_rate": 2.848240494780504e-05, "loss": 0.0075, "step": 197860 }, { "epoch": 126.11217335882728, "grad_norm": 0.18643039464950562, "learning_rate": 2.8477430001714223e-05, "loss": 0.0121, "step": 197870 }, { "epoch": 126.11854684512429, "grad_norm": 0.14592353999614716, "learning_rate": 2.8472455317142955e-05, "loss": 0.0053, "step": 197880 }, { "epoch": 126.1249203314213, "grad_norm": 0.1915207952260971, "learning_rate": 2.8467480894151666e-05, "loss": 0.0104, "step": 197890 }, { "epoch": 126.13129381771829, "grad_norm": 0.19105112552642822, "learning_rate": 2.846250673280084e-05, "loss": 0.0053, "step": 197900 }, { "epoch": 126.1376673040153, "grad_norm": 0.1462220698595047, "learning_rate": 2.8457532833150867e-05, "loss": 0.0064, "step": 197910 }, { "epoch": 126.1440407903123, "grad_norm": 0.414176881313324, "learning_rate": 2.845255919526222e-05, "loss": 0.0086, "step": 197920 }, { "epoch": 126.15041427660931, "grad_norm": 0.5208079218864441, "learning_rate": 2.8447585819195312e-05, "loss": 0.0097, "step": 197930 }, { "epoch": 126.1567877629063, "grad_norm": 0.5385300517082214, "learning_rate": 2.844261270501062e-05, "loss": 0.014, "step": 197940 }, { "epoch": 126.16316124920331, "grad_norm": 0.2180018275976181, "learning_rate": 2.8437639852768496e-05, "loss": 0.0115, "step": 197950 }, { "epoch": 126.16953473550032, "grad_norm": 0.2894710600376129, "learning_rate": 2.843266726252942e-05, "loss": 0.0107, "step": 197960 }, { "epoch": 126.17590822179733, "grad_norm": 0.4887109696865082, "learning_rate": 2.842769493435381e-05, "loss": 0.0103, "step": 197970 }, { "epoch": 126.18228170809432, "grad_norm": 0.2988354563713074, "learning_rate": 2.8422722868302065e-05, "loss": 0.0074, "step": 197980 }, { "epoch": 126.18865519439133, "grad_norm": 0.21350686252117157, "learning_rate": 2.8417751064434596e-05, "loss": 0.0062, "step": 197990 }, { "epoch": 126.19502868068834, "grad_norm": 0.38789546489715576, "learning_rate": 2.8412779522811848e-05, "loss": 0.0057, "step": 198000 }, { "epoch": 126.20140216698535, "grad_norm": 0.17803208529949188, "learning_rate": 2.840780824349421e-05, "loss": 0.018, "step": 198010 }, { "epoch": 126.20777565328234, "grad_norm": 0.286005437374115, "learning_rate": 2.840283722654209e-05, "loss": 0.0125, "step": 198020 }, { "epoch": 126.21414913957935, "grad_norm": 0.40575096011161804, "learning_rate": 2.8397866472015867e-05, "loss": 0.0082, "step": 198030 }, { "epoch": 126.22052262587636, "grad_norm": 0.44766706228256226, "learning_rate": 2.8392895979975975e-05, "loss": 0.008, "step": 198040 }, { "epoch": 126.22689611217336, "grad_norm": 0.22247818112373352, "learning_rate": 2.8387925750482807e-05, "loss": 0.0078, "step": 198050 }, { "epoch": 126.23326959847036, "grad_norm": 0.3506118357181549, "learning_rate": 2.838295578359674e-05, "loss": 0.0064, "step": 198060 }, { "epoch": 126.23964308476737, "grad_norm": 0.14720302820205688, "learning_rate": 2.837798607937816e-05, "loss": 0.005, "step": 198070 }, { "epoch": 126.24601657106437, "grad_norm": 0.16680817306041718, "learning_rate": 2.8373016637887497e-05, "loss": 0.0061, "step": 198080 }, { "epoch": 126.25239005736138, "grad_norm": 0.240431547164917, "learning_rate": 2.836804745918506e-05, "loss": 0.0061, "step": 198090 }, { "epoch": 126.25876354365838, "grad_norm": 0.1293327659368515, "learning_rate": 2.836307854333129e-05, "loss": 0.0136, "step": 198100 }, { "epoch": 126.26513702995538, "grad_norm": 0.20851664245128632, "learning_rate": 2.835810989038653e-05, "loss": 0.0066, "step": 198110 }, { "epoch": 126.27151051625239, "grad_norm": 0.6637753248214722, "learning_rate": 2.8353141500411202e-05, "loss": 0.0155, "step": 198120 }, { "epoch": 126.2778840025494, "grad_norm": 0.151699960231781, "learning_rate": 2.8348173373465602e-05, "loss": 0.0055, "step": 198130 }, { "epoch": 126.2842574888464, "grad_norm": 0.31775903701782227, "learning_rate": 2.8343205509610166e-05, "loss": 0.0119, "step": 198140 }, { "epoch": 126.2906309751434, "grad_norm": 0.23005913197994232, "learning_rate": 2.8338237908905225e-05, "loss": 0.0088, "step": 198150 }, { "epoch": 126.29700446144041, "grad_norm": 0.1427742838859558, "learning_rate": 2.8333270571411147e-05, "loss": 0.0125, "step": 198160 }, { "epoch": 126.30337794773742, "grad_norm": 0.6330360174179077, "learning_rate": 2.8328303497188268e-05, "loss": 0.0077, "step": 198170 }, { "epoch": 126.30975143403441, "grad_norm": 0.2638310492038727, "learning_rate": 2.8323336686296982e-05, "loss": 0.0136, "step": 198180 }, { "epoch": 126.31612492033142, "grad_norm": 0.5466791987419128, "learning_rate": 2.8318370138797622e-05, "loss": 0.0089, "step": 198190 }, { "epoch": 126.32249840662843, "grad_norm": 0.1482134312391281, "learning_rate": 2.831340385475053e-05, "loss": 0.0224, "step": 198200 }, { "epoch": 126.32887189292543, "grad_norm": 0.388662725687027, "learning_rate": 2.830843783421605e-05, "loss": 0.0092, "step": 198210 }, { "epoch": 126.33524537922243, "grad_norm": 0.34755179286003113, "learning_rate": 2.830347207725454e-05, "loss": 0.0086, "step": 198220 }, { "epoch": 126.34161886551944, "grad_norm": 0.22883322834968567, "learning_rate": 2.8298506583926333e-05, "loss": 0.0085, "step": 198230 }, { "epoch": 126.34799235181644, "grad_norm": 0.1829129159450531, "learning_rate": 2.829354135429176e-05, "loss": 0.0099, "step": 198240 }, { "epoch": 126.35436583811345, "grad_norm": 0.12744157016277313, "learning_rate": 2.8288576388411132e-05, "loss": 0.0085, "step": 198250 }, { "epoch": 126.36073932441045, "grad_norm": 0.3372187316417694, "learning_rate": 2.8283611686344825e-05, "loss": 0.0096, "step": 198260 }, { "epoch": 126.36711281070745, "grad_norm": 0.3092075288295746, "learning_rate": 2.8278647248153134e-05, "loss": 0.0142, "step": 198270 }, { "epoch": 126.37348629700446, "grad_norm": 0.4255455732345581, "learning_rate": 2.8273683073896384e-05, "loss": 0.0087, "step": 198280 }, { "epoch": 126.37985978330147, "grad_norm": 0.13824491202831268, "learning_rate": 2.8268719163634888e-05, "loss": 0.0033, "step": 198290 }, { "epoch": 126.38623326959846, "grad_norm": 0.23851940035820007, "learning_rate": 2.8263755517428987e-05, "loss": 0.0176, "step": 198300 }, { "epoch": 126.39260675589547, "grad_norm": 0.2156737595796585, "learning_rate": 2.8258792135338974e-05, "loss": 0.0069, "step": 198310 }, { "epoch": 126.39898024219248, "grad_norm": 0.34475618600845337, "learning_rate": 2.8253829017425172e-05, "loss": 0.011, "step": 198320 }, { "epoch": 126.40535372848949, "grad_norm": 0.30467942357063293, "learning_rate": 2.8248866163747878e-05, "loss": 0.0137, "step": 198330 }, { "epoch": 126.4117272147865, "grad_norm": 0.1600867062807083, "learning_rate": 2.824390357436737e-05, "loss": 0.013, "step": 198340 }, { "epoch": 126.41810070108349, "grad_norm": 0.3943819999694824, "learning_rate": 2.8238941249343996e-05, "loss": 0.0083, "step": 198350 }, { "epoch": 126.4244741873805, "grad_norm": 0.11975240707397461, "learning_rate": 2.823397918873804e-05, "loss": 0.0078, "step": 198360 }, { "epoch": 126.4308476736775, "grad_norm": 0.16965749859809875, "learning_rate": 2.822901739260978e-05, "loss": 0.0115, "step": 198370 }, { "epoch": 126.43722115997451, "grad_norm": 0.4310756325721741, "learning_rate": 2.82240558610195e-05, "loss": 0.0083, "step": 198380 }, { "epoch": 126.4435946462715, "grad_norm": 0.6035828590393066, "learning_rate": 2.821909459402752e-05, "loss": 0.0091, "step": 198390 }, { "epoch": 126.44996813256851, "grad_norm": 0.2692941129207611, "learning_rate": 2.8214133591694098e-05, "loss": 0.0134, "step": 198400 }, { "epoch": 126.45634161886552, "grad_norm": 0.26975223422050476, "learning_rate": 2.8209172854079524e-05, "loss": 0.0072, "step": 198410 }, { "epoch": 126.46271510516253, "grad_norm": 0.298441618680954, "learning_rate": 2.8204212381244054e-05, "loss": 0.0064, "step": 198420 }, { "epoch": 126.46908859145952, "grad_norm": 0.4304404854774475, "learning_rate": 2.8199252173248007e-05, "loss": 0.015, "step": 198430 }, { "epoch": 126.47546207775653, "grad_norm": 0.10676310956478119, "learning_rate": 2.8194292230151632e-05, "loss": 0.0143, "step": 198440 }, { "epoch": 126.48183556405354, "grad_norm": 0.40520498156547546, "learning_rate": 2.8189332552015192e-05, "loss": 0.0062, "step": 198450 }, { "epoch": 126.48820905035055, "grad_norm": 0.19742025434970856, "learning_rate": 2.8184373138898923e-05, "loss": 0.0116, "step": 198460 }, { "epoch": 126.49458253664754, "grad_norm": 0.44220492243766785, "learning_rate": 2.817941399086317e-05, "loss": 0.0099, "step": 198470 }, { "epoch": 126.50095602294455, "grad_norm": 0.1653134822845459, "learning_rate": 2.8174455107968094e-05, "loss": 0.0094, "step": 198480 }, { "epoch": 126.50732950924156, "grad_norm": 0.3326775133609772, "learning_rate": 2.8169496490274018e-05, "loss": 0.0062, "step": 198490 }, { "epoch": 126.51370299553857, "grad_norm": 0.20542016625404358, "learning_rate": 2.8164538137841146e-05, "loss": 0.0092, "step": 198500 }, { "epoch": 126.52007648183556, "grad_norm": 0.45563623309135437, "learning_rate": 2.8159580050729795e-05, "loss": 0.0129, "step": 198510 }, { "epoch": 126.52644996813257, "grad_norm": 0.19810527563095093, "learning_rate": 2.8154622229000128e-05, "loss": 0.02, "step": 198520 }, { "epoch": 126.53282345442958, "grad_norm": 0.31982913613319397, "learning_rate": 2.814966467271245e-05, "loss": 0.0158, "step": 198530 }, { "epoch": 126.53919694072658, "grad_norm": 0.1911347508430481, "learning_rate": 2.814470738192697e-05, "loss": 0.0177, "step": 198540 }, { "epoch": 126.54557042702358, "grad_norm": 0.2938292920589447, "learning_rate": 2.8139750356703932e-05, "loss": 0.006, "step": 198550 }, { "epoch": 126.55194391332059, "grad_norm": 0.189845010638237, "learning_rate": 2.8134793597103548e-05, "loss": 0.0047, "step": 198560 }, { "epoch": 126.5583173996176, "grad_norm": 0.26161471009254456, "learning_rate": 2.8129837103186086e-05, "loss": 0.0059, "step": 198570 }, { "epoch": 126.5646908859146, "grad_norm": 0.42105570435523987, "learning_rate": 2.812488087501175e-05, "loss": 0.0084, "step": 198580 }, { "epoch": 126.5710643722116, "grad_norm": 0.2764506936073303, "learning_rate": 2.8119924912640766e-05, "loss": 0.0172, "step": 198590 }, { "epoch": 126.5774378585086, "grad_norm": 0.23773936927318573, "learning_rate": 2.8114969216133334e-05, "loss": 0.0111, "step": 198600 }, { "epoch": 126.58381134480561, "grad_norm": 0.29746660590171814, "learning_rate": 2.8110013785549705e-05, "loss": 0.0145, "step": 198610 }, { "epoch": 126.59018483110262, "grad_norm": 0.14938341081142426, "learning_rate": 2.8105058620950077e-05, "loss": 0.0098, "step": 198620 }, { "epoch": 126.59655831739961, "grad_norm": 0.3326406478881836, "learning_rate": 2.8100103722394662e-05, "loss": 0.0087, "step": 198630 }, { "epoch": 126.60293180369662, "grad_norm": 0.37615835666656494, "learning_rate": 2.809514908994364e-05, "loss": 0.011, "step": 198640 }, { "epoch": 126.60930528999363, "grad_norm": 0.2907828986644745, "learning_rate": 2.809019472365728e-05, "loss": 0.0082, "step": 198650 }, { "epoch": 126.61567877629064, "grad_norm": 0.4388159513473511, "learning_rate": 2.8085240623595698e-05, "loss": 0.009, "step": 198660 }, { "epoch": 126.62205226258763, "grad_norm": 0.424688458442688, "learning_rate": 2.808028678981916e-05, "loss": 0.0075, "step": 198670 }, { "epoch": 126.62842574888464, "grad_norm": 0.5923977494239807, "learning_rate": 2.807533322238781e-05, "loss": 0.0068, "step": 198680 }, { "epoch": 126.63479923518165, "grad_norm": 0.18348939716815948, "learning_rate": 2.8070379921361902e-05, "loss": 0.0077, "step": 198690 }, { "epoch": 126.64117272147865, "grad_norm": 0.1712603121995926, "learning_rate": 2.806542688680154e-05, "loss": 0.0117, "step": 198700 }, { "epoch": 126.64754620777565, "grad_norm": 0.37787821888923645, "learning_rate": 2.8060474118766978e-05, "loss": 0.0057, "step": 198710 }, { "epoch": 126.65391969407266, "grad_norm": 0.34639501571655273, "learning_rate": 2.8055521617318357e-05, "loss": 0.0125, "step": 198720 }, { "epoch": 126.66029318036966, "grad_norm": 0.2640224099159241, "learning_rate": 2.805056938251588e-05, "loss": 0.0099, "step": 198730 }, { "epoch": 126.66666666666667, "grad_norm": 0.2971501052379608, "learning_rate": 2.804561741441968e-05, "loss": 0.0052, "step": 198740 }, { "epoch": 126.67304015296367, "grad_norm": 0.15603354573249817, "learning_rate": 2.8040665713089982e-05, "loss": 0.0107, "step": 198750 }, { "epoch": 126.67941363926067, "grad_norm": 0.262604683637619, "learning_rate": 2.803571427858693e-05, "loss": 0.0098, "step": 198760 }, { "epoch": 126.68578712555768, "grad_norm": 0.17156529426574707, "learning_rate": 2.8030763110970682e-05, "loss": 0.0109, "step": 198770 }, { "epoch": 126.69216061185469, "grad_norm": 0.18227963149547577, "learning_rate": 2.8025812210301382e-05, "loss": 0.0155, "step": 198780 }, { "epoch": 126.69853409815168, "grad_norm": 0.15834851562976837, "learning_rate": 2.8020861576639256e-05, "loss": 0.0095, "step": 198790 }, { "epoch": 126.70490758444869, "grad_norm": 0.24925892055034637, "learning_rate": 2.8015911210044378e-05, "loss": 0.0074, "step": 198800 }, { "epoch": 126.7112810707457, "grad_norm": 0.3078984022140503, "learning_rate": 2.801096111057695e-05, "loss": 0.0055, "step": 198810 }, { "epoch": 126.7176545570427, "grad_norm": 0.1954808235168457, "learning_rate": 2.8006011278297085e-05, "loss": 0.0071, "step": 198820 }, { "epoch": 126.7240280433397, "grad_norm": 0.6801706552505493, "learning_rate": 2.8001061713264986e-05, "loss": 0.0084, "step": 198830 }, { "epoch": 126.73040152963671, "grad_norm": 0.26151278614997864, "learning_rate": 2.7996112415540727e-05, "loss": 0.0099, "step": 198840 }, { "epoch": 126.73677501593372, "grad_norm": 0.5048986673355103, "learning_rate": 2.7991163385184495e-05, "loss": 0.0094, "step": 198850 }, { "epoch": 126.74314850223072, "grad_norm": 0.6186419129371643, "learning_rate": 2.79862146222564e-05, "loss": 0.013, "step": 198860 }, { "epoch": 126.74952198852772, "grad_norm": 0.422387957572937, "learning_rate": 2.7981266126816587e-05, "loss": 0.0143, "step": 198870 }, { "epoch": 126.75589547482473, "grad_norm": 0.49659648537635803, "learning_rate": 2.797631789892516e-05, "loss": 0.0184, "step": 198880 }, { "epoch": 126.76226896112173, "grad_norm": 0.5345154404640198, "learning_rate": 2.7971369938642278e-05, "loss": 0.0157, "step": 198890 }, { "epoch": 126.76864244741874, "grad_norm": 0.17307379841804504, "learning_rate": 2.7966422246028055e-05, "loss": 0.0071, "step": 198900 }, { "epoch": 126.77501593371574, "grad_norm": 0.20822319388389587, "learning_rate": 2.7961474821142608e-05, "loss": 0.0128, "step": 198910 }, { "epoch": 126.78138942001274, "grad_norm": 0.5677980780601501, "learning_rate": 2.7956527664046027e-05, "loss": 0.0098, "step": 198920 }, { "epoch": 126.78776290630975, "grad_norm": 0.28547775745391846, "learning_rate": 2.795158077479847e-05, "loss": 0.0168, "step": 198930 }, { "epoch": 126.79413639260676, "grad_norm": 0.33065417408943176, "learning_rate": 2.7946634153460016e-05, "loss": 0.0083, "step": 198940 }, { "epoch": 126.80050987890377, "grad_norm": 0.2738056778907776, "learning_rate": 2.7941687800090783e-05, "loss": 0.0071, "step": 198950 }, { "epoch": 126.80688336520076, "grad_norm": 0.19329331815242767, "learning_rate": 2.7936741714750848e-05, "loss": 0.0053, "step": 198960 }, { "epoch": 126.81325685149777, "grad_norm": 0.2924012541770935, "learning_rate": 2.7931795897500358e-05, "loss": 0.0072, "step": 198970 }, { "epoch": 126.81963033779478, "grad_norm": 0.19470210373401642, "learning_rate": 2.792685034839938e-05, "loss": 0.0097, "step": 198980 }, { "epoch": 126.82600382409179, "grad_norm": 0.34322813153266907, "learning_rate": 2.7921905067508012e-05, "loss": 0.0073, "step": 198990 }, { "epoch": 126.83237731038878, "grad_norm": 0.18539214134216309, "learning_rate": 2.791696005488632e-05, "loss": 0.008, "step": 199000 }, { "epoch": 126.83875079668579, "grad_norm": 0.12626871466636658, "learning_rate": 2.7912015310594436e-05, "loss": 0.012, "step": 199010 }, { "epoch": 126.8451242829828, "grad_norm": 0.183848038315773, "learning_rate": 2.7907070834692416e-05, "loss": 0.0173, "step": 199020 }, { "epoch": 126.8514977692798, "grad_norm": 0.3664945065975189, "learning_rate": 2.790212662724035e-05, "loss": 0.0036, "step": 199030 }, { "epoch": 126.8578712555768, "grad_norm": 0.1435866355895996, "learning_rate": 2.7897182688298307e-05, "loss": 0.0063, "step": 199040 }, { "epoch": 126.8642447418738, "grad_norm": 0.513117253780365, "learning_rate": 2.789223901792634e-05, "loss": 0.0154, "step": 199050 }, { "epoch": 126.87061822817081, "grad_norm": 0.26633331179618835, "learning_rate": 2.7887295616184566e-05, "loss": 0.0101, "step": 199060 }, { "epoch": 126.87699171446782, "grad_norm": 0.2305534929037094, "learning_rate": 2.788235248313303e-05, "loss": 0.0064, "step": 199070 }, { "epoch": 126.88336520076481, "grad_norm": 0.42612719535827637, "learning_rate": 2.787740961883179e-05, "loss": 0.0098, "step": 199080 }, { "epoch": 126.88973868706182, "grad_norm": 0.13581795990467072, "learning_rate": 2.787246702334089e-05, "loss": 0.0124, "step": 199090 }, { "epoch": 126.89611217335883, "grad_norm": 0.3455899655818939, "learning_rate": 2.7867524696720425e-05, "loss": 0.0081, "step": 199100 }, { "epoch": 126.90248565965584, "grad_norm": 0.2599731683731079, "learning_rate": 2.7862582639030445e-05, "loss": 0.0198, "step": 199110 }, { "epoch": 126.90885914595283, "grad_norm": 0.13413748145103455, "learning_rate": 2.7857640850330974e-05, "loss": 0.0052, "step": 199120 }, { "epoch": 126.91523263224984, "grad_norm": 0.43669337034225464, "learning_rate": 2.785269933068205e-05, "loss": 0.0143, "step": 199130 }, { "epoch": 126.92160611854685, "grad_norm": 0.21402224898338318, "learning_rate": 2.784775808014377e-05, "loss": 0.0097, "step": 199140 }, { "epoch": 126.92797960484386, "grad_norm": 0.2747464179992676, "learning_rate": 2.784281709877614e-05, "loss": 0.0076, "step": 199150 }, { "epoch": 126.93435309114085, "grad_norm": 0.16309727728366852, "learning_rate": 2.7837876386639205e-05, "loss": 0.0048, "step": 199160 }, { "epoch": 126.94072657743786, "grad_norm": 0.42993178963661194, "learning_rate": 2.783293594379297e-05, "loss": 0.0136, "step": 199170 }, { "epoch": 126.94710006373487, "grad_norm": 0.3854418694972992, "learning_rate": 2.7827995770297537e-05, "loss": 0.0072, "step": 199180 }, { "epoch": 126.95347355003187, "grad_norm": 0.11821999400854111, "learning_rate": 2.7823055866212848e-05, "loss": 0.0098, "step": 199190 }, { "epoch": 126.95984703632887, "grad_norm": 0.19786791503429413, "learning_rate": 2.781811623159899e-05, "loss": 0.0086, "step": 199200 }, { "epoch": 126.96622052262587, "grad_norm": 0.25949153304100037, "learning_rate": 2.7813176866515943e-05, "loss": 0.0096, "step": 199210 }, { "epoch": 126.97259400892288, "grad_norm": 0.44777950644493103, "learning_rate": 2.7808237771023782e-05, "loss": 0.0112, "step": 199220 }, { "epoch": 126.97896749521989, "grad_norm": 0.3157767653465271, "learning_rate": 2.7803298945182443e-05, "loss": 0.0091, "step": 199230 }, { "epoch": 126.98534098151688, "grad_norm": 0.33726799488067627, "learning_rate": 2.7798360389052003e-05, "loss": 0.0072, "step": 199240 }, { "epoch": 126.99171446781389, "grad_norm": 0.4377405345439911, "learning_rate": 2.779342210269243e-05, "loss": 0.0103, "step": 199250 }, { "epoch": 126.9980879541109, "grad_norm": 0.6322977542877197, "learning_rate": 2.778848408616378e-05, "loss": 0.0121, "step": 199260 }, { "epoch": 127.00446144040791, "grad_norm": 0.48156124353408813, "learning_rate": 2.778354633952598e-05, "loss": 0.0157, "step": 199270 }, { "epoch": 127.0108349267049, "grad_norm": 0.31823232769966125, "learning_rate": 2.7778608862839084e-05, "loss": 0.0253, "step": 199280 }, { "epoch": 127.01720841300191, "grad_norm": 0.4866085648536682, "learning_rate": 2.777367165616308e-05, "loss": 0.0133, "step": 199290 }, { "epoch": 127.02358189929892, "grad_norm": 0.4049694538116455, "learning_rate": 2.776873471955794e-05, "loss": 0.0059, "step": 199300 }, { "epoch": 127.02995538559593, "grad_norm": 0.12478294223546982, "learning_rate": 2.776379805308365e-05, "loss": 0.0086, "step": 199310 }, { "epoch": 127.03632887189292, "grad_norm": 0.17719706892967224, "learning_rate": 2.775886165680023e-05, "loss": 0.0074, "step": 199320 }, { "epoch": 127.04270235818993, "grad_norm": 0.1743023693561554, "learning_rate": 2.7753925530767633e-05, "loss": 0.0086, "step": 199330 }, { "epoch": 127.04907584448694, "grad_norm": 0.17824669182300568, "learning_rate": 2.7748989675045843e-05, "loss": 0.0116, "step": 199340 }, { "epoch": 127.05544933078394, "grad_norm": 0.16326531767845154, "learning_rate": 2.7744054089694828e-05, "loss": 0.016, "step": 199350 }, { "epoch": 127.06182281708094, "grad_norm": 0.3225385546684265, "learning_rate": 2.77391187747746e-05, "loss": 0.0083, "step": 199360 }, { "epoch": 127.06819630337795, "grad_norm": 0.17103545367717743, "learning_rate": 2.7734183730345058e-05, "loss": 0.0073, "step": 199370 }, { "epoch": 127.07456978967495, "grad_norm": 0.22132916748523712, "learning_rate": 2.772924895646623e-05, "loss": 0.0095, "step": 199380 }, { "epoch": 127.08094327597196, "grad_norm": 0.6603155136108398, "learning_rate": 2.772431445319803e-05, "loss": 0.0143, "step": 199390 }, { "epoch": 127.08731676226896, "grad_norm": 0.20807558298110962, "learning_rate": 2.771938022060048e-05, "loss": 0.0108, "step": 199400 }, { "epoch": 127.09369024856596, "grad_norm": 0.10031045973300934, "learning_rate": 2.771444625873346e-05, "loss": 0.0047, "step": 199410 }, { "epoch": 127.10006373486297, "grad_norm": 0.3120976984500885, "learning_rate": 2.770951256765698e-05, "loss": 0.0093, "step": 199420 }, { "epoch": 127.10643722115998, "grad_norm": 0.28135383129119873, "learning_rate": 2.7704579147430965e-05, "loss": 0.0108, "step": 199430 }, { "epoch": 127.11281070745697, "grad_norm": 0.226432204246521, "learning_rate": 2.7699645998115366e-05, "loss": 0.0086, "step": 199440 }, { "epoch": 127.11918419375398, "grad_norm": 0.3047170639038086, "learning_rate": 2.7694713119770106e-05, "loss": 0.0115, "step": 199450 }, { "epoch": 127.12555768005099, "grad_norm": 0.2596748471260071, "learning_rate": 2.7689780512455156e-05, "loss": 0.0124, "step": 199460 }, { "epoch": 127.131931166348, "grad_norm": 0.49248626828193665, "learning_rate": 2.7684848176230437e-05, "loss": 0.0084, "step": 199470 }, { "epoch": 127.13830465264499, "grad_norm": 0.3361526131629944, "learning_rate": 2.7679916111155886e-05, "loss": 0.0126, "step": 199480 }, { "epoch": 127.144678138942, "grad_norm": 0.6345490217208862, "learning_rate": 2.76749843172914e-05, "loss": 0.0152, "step": 199490 }, { "epoch": 127.151051625239, "grad_norm": 0.24118992686271667, "learning_rate": 2.767005279469696e-05, "loss": 0.0072, "step": 199500 }, { "epoch": 127.15742511153601, "grad_norm": 0.4203263819217682, "learning_rate": 2.766512154343246e-05, "loss": 0.009, "step": 199510 }, { "epoch": 127.16379859783301, "grad_norm": 0.16252964735031128, "learning_rate": 2.766019056355782e-05, "loss": 0.0062, "step": 199520 }, { "epoch": 127.17017208413002, "grad_norm": 0.3377378284931183, "learning_rate": 2.765525985513294e-05, "loss": 0.0129, "step": 199530 }, { "epoch": 127.17654557042702, "grad_norm": 0.29249268770217896, "learning_rate": 2.7650329418217786e-05, "loss": 0.0064, "step": 199540 }, { "epoch": 127.18291905672403, "grad_norm": 0.2156192511320114, "learning_rate": 2.7645399252872195e-05, "loss": 0.0072, "step": 199550 }, { "epoch": 127.18929254302103, "grad_norm": 0.2746632397174835, "learning_rate": 2.7640469359156124e-05, "loss": 0.018, "step": 199560 }, { "epoch": 127.19566602931803, "grad_norm": 3.730924129486084, "learning_rate": 2.7635539737129447e-05, "loss": 0.0292, "step": 199570 }, { "epoch": 127.20203951561504, "grad_norm": 0.1469079554080963, "learning_rate": 2.7630610386852107e-05, "loss": 0.0138, "step": 199580 }, { "epoch": 127.20841300191205, "grad_norm": 0.23291261494159698, "learning_rate": 2.7625681308383944e-05, "loss": 0.0069, "step": 199590 }, { "epoch": 127.21478648820906, "grad_norm": 0.46576833724975586, "learning_rate": 2.762075250178489e-05, "loss": 0.0098, "step": 199600 }, { "epoch": 127.22115997450605, "grad_norm": 0.21424056589603424, "learning_rate": 2.7615823967114823e-05, "loss": 0.0046, "step": 199610 }, { "epoch": 127.22753346080306, "grad_norm": 0.22806721925735474, "learning_rate": 2.7610895704433632e-05, "loss": 0.0067, "step": 199620 }, { "epoch": 127.23390694710007, "grad_norm": 0.23299694061279297, "learning_rate": 2.7605967713801173e-05, "loss": 0.0063, "step": 199630 }, { "epoch": 127.24028043339707, "grad_norm": 0.3848362863063812, "learning_rate": 2.760103999527738e-05, "loss": 0.0075, "step": 199640 }, { "epoch": 127.24665391969407, "grad_norm": 0.5437709093093872, "learning_rate": 2.7596112548922088e-05, "loss": 0.0076, "step": 199650 }, { "epoch": 127.25302740599108, "grad_norm": 0.4205312132835388, "learning_rate": 2.7591185374795186e-05, "loss": 0.0112, "step": 199660 }, { "epoch": 127.25940089228808, "grad_norm": 0.21332435309886932, "learning_rate": 2.7586258472956515e-05, "loss": 0.011, "step": 199670 }, { "epoch": 127.26577437858509, "grad_norm": 0.41189083456993103, "learning_rate": 2.758133184346599e-05, "loss": 0.0106, "step": 199680 }, { "epoch": 127.27214786488209, "grad_norm": 0.4762002229690552, "learning_rate": 2.7576405486383448e-05, "loss": 0.0088, "step": 199690 }, { "epoch": 127.2785213511791, "grad_norm": 0.3577072322368622, "learning_rate": 2.757147940176875e-05, "loss": 0.0074, "step": 199700 }, { "epoch": 127.2848948374761, "grad_norm": 0.3204646706581116, "learning_rate": 2.756655358968174e-05, "loss": 0.0203, "step": 199710 }, { "epoch": 127.29126832377311, "grad_norm": 0.2125365138053894, "learning_rate": 2.7561628050182298e-05, "loss": 0.0085, "step": 199720 }, { "epoch": 127.2976418100701, "grad_norm": 0.2725948393344879, "learning_rate": 2.755670278333026e-05, "loss": 0.0102, "step": 199730 }, { "epoch": 127.30401529636711, "grad_norm": 0.3569667339324951, "learning_rate": 2.7551777789185472e-05, "loss": 0.0169, "step": 199740 }, { "epoch": 127.31038878266412, "grad_norm": 0.2869185209274292, "learning_rate": 2.7546853067807786e-05, "loss": 0.0079, "step": 199750 }, { "epoch": 127.31676226896113, "grad_norm": 0.20556366443634033, "learning_rate": 2.7541928619257007e-05, "loss": 0.0098, "step": 199760 }, { "epoch": 127.32313575525812, "grad_norm": 0.19095194339752197, "learning_rate": 2.753700444359303e-05, "loss": 0.0073, "step": 199770 }, { "epoch": 127.32950924155513, "grad_norm": 0.20733429491519928, "learning_rate": 2.753208054087565e-05, "loss": 0.0157, "step": 199780 }, { "epoch": 127.33588272785214, "grad_norm": 0.20680662989616394, "learning_rate": 2.7527156911164698e-05, "loss": 0.0043, "step": 199790 }, { "epoch": 127.34225621414915, "grad_norm": 0.18338492512702942, "learning_rate": 2.7522233554519994e-05, "loss": 0.0089, "step": 199800 }, { "epoch": 127.34862970044614, "grad_norm": 0.48213234543800354, "learning_rate": 2.75173104710014e-05, "loss": 0.0113, "step": 199810 }, { "epoch": 127.35500318674315, "grad_norm": 0.11713185161352158, "learning_rate": 2.7512387660668703e-05, "loss": 0.0051, "step": 199820 }, { "epoch": 127.36137667304016, "grad_norm": 0.5753323435783386, "learning_rate": 2.7507465123581732e-05, "loss": 0.0146, "step": 199830 }, { "epoch": 127.36775015933716, "grad_norm": 0.4802349805831909, "learning_rate": 2.7502542859800273e-05, "loss": 0.0073, "step": 199840 }, { "epoch": 127.37412364563416, "grad_norm": 0.26197919249534607, "learning_rate": 2.749762086938418e-05, "loss": 0.0091, "step": 199850 }, { "epoch": 127.38049713193116, "grad_norm": 0.4542310833930969, "learning_rate": 2.749269915239324e-05, "loss": 0.016, "step": 199860 }, { "epoch": 127.38687061822817, "grad_norm": 0.23014982044696808, "learning_rate": 2.748777770888726e-05, "loss": 0.0089, "step": 199870 }, { "epoch": 127.39324410452518, "grad_norm": 0.27310824394226074, "learning_rate": 2.7482856538926004e-05, "loss": 0.0099, "step": 199880 }, { "epoch": 127.39961759082217, "grad_norm": 0.2587435841560364, "learning_rate": 2.747793564256933e-05, "loss": 0.0066, "step": 199890 }, { "epoch": 127.40599107711918, "grad_norm": 0.13177193701267242, "learning_rate": 2.7473015019877e-05, "loss": 0.0112, "step": 199900 }, { "epoch": 127.41236456341619, "grad_norm": 0.17646090686321259, "learning_rate": 2.7468094670908813e-05, "loss": 0.013, "step": 199910 }, { "epoch": 127.4187380497132, "grad_norm": 0.16768956184387207, "learning_rate": 2.7463174595724523e-05, "loss": 0.0057, "step": 199920 }, { "epoch": 127.42511153601019, "grad_norm": 0.13569262623786926, "learning_rate": 2.745825479438398e-05, "loss": 0.0136, "step": 199930 }, { "epoch": 127.4314850223072, "grad_norm": 0.22334648668766022, "learning_rate": 2.7453335266946888e-05, "loss": 0.0059, "step": 199940 }, { "epoch": 127.43785850860421, "grad_norm": 0.3281746506690979, "learning_rate": 2.7448416013473076e-05, "loss": 0.0097, "step": 199950 }, { "epoch": 127.44423199490122, "grad_norm": 0.10700041055679321, "learning_rate": 2.7443497034022282e-05, "loss": 0.0041, "step": 199960 }, { "epoch": 127.45060548119821, "grad_norm": 0.4646055996417999, "learning_rate": 2.7438578328654335e-05, "loss": 0.0161, "step": 199970 }, { "epoch": 127.45697896749522, "grad_norm": 0.264865517616272, "learning_rate": 2.7433659897428927e-05, "loss": 0.0082, "step": 199980 }, { "epoch": 127.46335245379223, "grad_norm": 0.1436857283115387, "learning_rate": 2.742874174040587e-05, "loss": 0.0063, "step": 199990 }, { "epoch": 127.46972594008923, "grad_norm": 0.19324944913387299, "learning_rate": 2.742382385764492e-05, "loss": 0.0197, "step": 200000 }, { "epoch": 127.47609942638623, "grad_norm": 0.17750951647758484, "learning_rate": 2.7418906249205823e-05, "loss": 0.0089, "step": 200010 }, { "epoch": 127.48247291268324, "grad_norm": 0.2844674587249756, "learning_rate": 2.7413988915148314e-05, "loss": 0.0144, "step": 200020 }, { "epoch": 127.48884639898024, "grad_norm": 0.3282513916492462, "learning_rate": 2.7409071855532187e-05, "loss": 0.0103, "step": 200030 }, { "epoch": 127.49521988527725, "grad_norm": 0.1513226479291916, "learning_rate": 2.740415507041716e-05, "loss": 0.0125, "step": 200040 }, { "epoch": 127.50159337157424, "grad_norm": 0.39882776141166687, "learning_rate": 2.739923855986299e-05, "loss": 0.0059, "step": 200050 }, { "epoch": 127.50796685787125, "grad_norm": 0.11472208797931671, "learning_rate": 2.7394322323929383e-05, "loss": 0.0116, "step": 200060 }, { "epoch": 127.51434034416826, "grad_norm": 0.29740431904792786, "learning_rate": 2.7389406362676152e-05, "loss": 0.0048, "step": 200070 }, { "epoch": 127.52071383046527, "grad_norm": 0.19246573746204376, "learning_rate": 2.7384490676162933e-05, "loss": 0.0086, "step": 200080 }, { "epoch": 127.52708731676226, "grad_norm": 0.200848788022995, "learning_rate": 2.737957526444953e-05, "loss": 0.0117, "step": 200090 }, { "epoch": 127.53346080305927, "grad_norm": 0.18686717748641968, "learning_rate": 2.7374660127595618e-05, "loss": 0.0069, "step": 200100 }, { "epoch": 127.53983428935628, "grad_norm": 0.1579282134771347, "learning_rate": 2.7369745265660994e-05, "loss": 0.0058, "step": 200110 }, { "epoch": 127.54620777565329, "grad_norm": 0.2118227481842041, "learning_rate": 2.736483067870529e-05, "loss": 0.0042, "step": 200120 }, { "epoch": 127.55258126195028, "grad_norm": 0.12798602879047394, "learning_rate": 2.7359916366788284e-05, "loss": 0.0091, "step": 200130 }, { "epoch": 127.55895474824729, "grad_norm": 0.5178753733634949, "learning_rate": 2.7355002329969654e-05, "loss": 0.0117, "step": 200140 }, { "epoch": 127.5653282345443, "grad_norm": 0.4923287034034729, "learning_rate": 2.735008856830916e-05, "loss": 0.0116, "step": 200150 }, { "epoch": 127.5717017208413, "grad_norm": 0.17540161311626434, "learning_rate": 2.7345175081866438e-05, "loss": 0.0074, "step": 200160 }, { "epoch": 127.5780752071383, "grad_norm": 0.1839684695005417, "learning_rate": 2.734026187070125e-05, "loss": 0.0088, "step": 200170 }, { "epoch": 127.5844486934353, "grad_norm": 0.20221124589443207, "learning_rate": 2.7335348934873273e-05, "loss": 0.0116, "step": 200180 }, { "epoch": 127.59082217973231, "grad_norm": 0.21611729264259338, "learning_rate": 2.7330436274442206e-05, "loss": 0.0083, "step": 200190 }, { "epoch": 127.59719566602932, "grad_norm": 0.2278890758752823, "learning_rate": 2.732552388946772e-05, "loss": 0.0039, "step": 200200 }, { "epoch": 127.60356915232632, "grad_norm": 0.3366321325302124, "learning_rate": 2.732061178000955e-05, "loss": 0.0118, "step": 200210 }, { "epoch": 127.60994263862332, "grad_norm": 0.35906782746315, "learning_rate": 2.7315699946127353e-05, "loss": 0.0117, "step": 200220 }, { "epoch": 127.61631612492033, "grad_norm": 0.1594858169555664, "learning_rate": 2.7310788387880825e-05, "loss": 0.0091, "step": 200230 }, { "epoch": 127.62268961121734, "grad_norm": 0.11879772692918777, "learning_rate": 2.730587710532962e-05, "loss": 0.0051, "step": 200240 }, { "epoch": 127.62906309751435, "grad_norm": 0.3912765681743622, "learning_rate": 2.7300966098533475e-05, "loss": 0.0131, "step": 200250 }, { "epoch": 127.63543658381134, "grad_norm": 0.1706879884004593, "learning_rate": 2.729605536755198e-05, "loss": 0.0122, "step": 200260 }, { "epoch": 127.64181007010835, "grad_norm": 0.31269025802612305, "learning_rate": 2.729114491244487e-05, "loss": 0.0073, "step": 200270 }, { "epoch": 127.64818355640536, "grad_norm": 0.3658025562763214, "learning_rate": 2.7286234733271777e-05, "loss": 0.0174, "step": 200280 }, { "epoch": 127.65455704270236, "grad_norm": 0.21759721636772156, "learning_rate": 2.728132483009241e-05, "loss": 0.0043, "step": 200290 }, { "epoch": 127.66093052899936, "grad_norm": 0.29243898391723633, "learning_rate": 2.7276415202966353e-05, "loss": 0.0119, "step": 200300 }, { "epoch": 127.66730401529637, "grad_norm": 0.11951319128274918, "learning_rate": 2.7271505851953333e-05, "loss": 0.0122, "step": 200310 }, { "epoch": 127.67367750159337, "grad_norm": 0.4138694703578949, "learning_rate": 2.7266596777112973e-05, "loss": 0.0068, "step": 200320 }, { "epoch": 127.68005098789038, "grad_norm": 0.21406427025794983, "learning_rate": 2.7261687978504924e-05, "loss": 0.0156, "step": 200330 }, { "epoch": 127.68642447418738, "grad_norm": 0.34734028577804565, "learning_rate": 2.725677945618882e-05, "loss": 0.0168, "step": 200340 }, { "epoch": 127.69279796048438, "grad_norm": 0.2705456018447876, "learning_rate": 2.7251871210224334e-05, "loss": 0.0155, "step": 200350 }, { "epoch": 127.69917144678139, "grad_norm": 0.3272731602191925, "learning_rate": 2.7246963240671097e-05, "loss": 0.0118, "step": 200360 }, { "epoch": 127.7055449330784, "grad_norm": 0.11797334998846054, "learning_rate": 2.724205554758873e-05, "loss": 0.0084, "step": 200370 }, { "epoch": 127.7119184193754, "grad_norm": 0.4042567312717438, "learning_rate": 2.723714813103686e-05, "loss": 0.0084, "step": 200380 }, { "epoch": 127.7182919056724, "grad_norm": 0.15402689576148987, "learning_rate": 2.723224099107515e-05, "loss": 0.0127, "step": 200390 }, { "epoch": 127.72466539196941, "grad_norm": 0.20309236645698547, "learning_rate": 2.7227334127763216e-05, "loss": 0.0047, "step": 200400 }, { "epoch": 127.73103887826642, "grad_norm": 0.2533794939517975, "learning_rate": 2.7222427541160666e-05, "loss": 0.0108, "step": 200410 }, { "epoch": 127.73741236456341, "grad_norm": 0.11371153593063354, "learning_rate": 2.7217521231327102e-05, "loss": 0.0038, "step": 200420 }, { "epoch": 127.74378585086042, "grad_norm": 0.36699748039245605, "learning_rate": 2.7212615198322196e-05, "loss": 0.0071, "step": 200430 }, { "epoch": 127.75015933715743, "grad_norm": 0.26567962765693665, "learning_rate": 2.7207709442205525e-05, "loss": 0.01, "step": 200440 }, { "epoch": 127.75653282345444, "grad_norm": 0.13068482279777527, "learning_rate": 2.7202803963036706e-05, "loss": 0.0096, "step": 200450 }, { "epoch": 127.76290630975143, "grad_norm": 0.24388867616653442, "learning_rate": 2.7197898760875318e-05, "loss": 0.0068, "step": 200460 }, { "epoch": 127.76927979604844, "grad_norm": 0.23288846015930176, "learning_rate": 2.7192993835781012e-05, "loss": 0.0148, "step": 200470 }, { "epoch": 127.77565328234544, "grad_norm": 0.39870479702949524, "learning_rate": 2.718808918781337e-05, "loss": 0.0109, "step": 200480 }, { "epoch": 127.78202676864245, "grad_norm": 0.23000825941562653, "learning_rate": 2.718318481703197e-05, "loss": 0.0106, "step": 200490 }, { "epoch": 127.78840025493945, "grad_norm": 0.29756593704223633, "learning_rate": 2.717828072349643e-05, "loss": 0.006, "step": 200500 }, { "epoch": 127.79477374123645, "grad_norm": 0.10081999748945236, "learning_rate": 2.71733769072663e-05, "loss": 0.0108, "step": 200510 }, { "epoch": 127.80114722753346, "grad_norm": 0.3133156895637512, "learning_rate": 2.7168473368401214e-05, "loss": 0.0119, "step": 200520 }, { "epoch": 127.80752071383047, "grad_norm": 0.2802475690841675, "learning_rate": 2.716357010696073e-05, "loss": 0.0044, "step": 200530 }, { "epoch": 127.81389420012746, "grad_norm": 0.390235036611557, "learning_rate": 2.715866712300443e-05, "loss": 0.007, "step": 200540 }, { "epoch": 127.82026768642447, "grad_norm": 0.16827784478664398, "learning_rate": 2.7153764416591877e-05, "loss": 0.0049, "step": 200550 }, { "epoch": 127.82664117272148, "grad_norm": 0.33225077390670776, "learning_rate": 2.714886198778267e-05, "loss": 0.0059, "step": 200560 }, { "epoch": 127.83301465901849, "grad_norm": 0.2228832244873047, "learning_rate": 2.7143959836636367e-05, "loss": 0.008, "step": 200570 }, { "epoch": 127.83938814531548, "grad_norm": 0.15034542977809906, "learning_rate": 2.7139057963212533e-05, "loss": 0.0069, "step": 200580 }, { "epoch": 127.84576163161249, "grad_norm": 0.4365427792072296, "learning_rate": 2.7134156367570708e-05, "loss": 0.0097, "step": 200590 }, { "epoch": 127.8521351179095, "grad_norm": 0.212889164686203, "learning_rate": 2.7129255049770492e-05, "loss": 0.0048, "step": 200600 }, { "epoch": 127.8585086042065, "grad_norm": 0.2169175148010254, "learning_rate": 2.7124354009871422e-05, "loss": 0.0151, "step": 200610 }, { "epoch": 127.8648820905035, "grad_norm": 0.09389228373765945, "learning_rate": 2.7119453247933047e-05, "loss": 0.0079, "step": 200620 }, { "epoch": 127.87125557680051, "grad_norm": 0.1914530098438263, "learning_rate": 2.71145527640149e-05, "loss": 0.0083, "step": 200630 }, { "epoch": 127.87762906309752, "grad_norm": 0.2625027298927307, "learning_rate": 2.710965255817658e-05, "loss": 0.0167, "step": 200640 }, { "epoch": 127.88400254939452, "grad_norm": 0.4761947989463806, "learning_rate": 2.710475263047756e-05, "loss": 0.0157, "step": 200650 }, { "epoch": 127.89037603569152, "grad_norm": 0.21925418078899384, "learning_rate": 2.7099852980977425e-05, "loss": 0.0039, "step": 200660 }, { "epoch": 127.89674952198853, "grad_norm": 0.14250962436199188, "learning_rate": 2.7094953609735683e-05, "loss": 0.0138, "step": 200670 }, { "epoch": 127.90312300828553, "grad_norm": 0.24509933590888977, "learning_rate": 2.7090054516811912e-05, "loss": 0.0166, "step": 200680 }, { "epoch": 127.90949649458254, "grad_norm": 0.11367535591125488, "learning_rate": 2.708515570226558e-05, "loss": 0.0124, "step": 200690 }, { "epoch": 127.91586998087953, "grad_norm": 0.604600191116333, "learning_rate": 2.7080257166156258e-05, "loss": 0.0112, "step": 200700 }, { "epoch": 127.92224346717654, "grad_norm": 0.2618337869644165, "learning_rate": 2.7075358908543436e-05, "loss": 0.0079, "step": 200710 }, { "epoch": 127.92861695347355, "grad_norm": 0.20634189248085022, "learning_rate": 2.7070460929486673e-05, "loss": 0.0108, "step": 200720 }, { "epoch": 127.93499043977056, "grad_norm": 0.5139747262001038, "learning_rate": 2.7065563229045433e-05, "loss": 0.0133, "step": 200730 }, { "epoch": 127.94136392606755, "grad_norm": 0.21058739721775055, "learning_rate": 2.7060665807279273e-05, "loss": 0.0069, "step": 200740 }, { "epoch": 127.94773741236456, "grad_norm": 0.15359072387218475, "learning_rate": 2.705576866424767e-05, "loss": 0.0078, "step": 200750 }, { "epoch": 127.95411089866157, "grad_norm": 0.15210184454917908, "learning_rate": 2.7050871800010153e-05, "loss": 0.0076, "step": 200760 }, { "epoch": 127.96048438495858, "grad_norm": 0.4296680688858032, "learning_rate": 2.7045975214626185e-05, "loss": 0.0117, "step": 200770 }, { "epoch": 127.96685787125557, "grad_norm": 0.17622649669647217, "learning_rate": 2.7041078908155317e-05, "loss": 0.0081, "step": 200780 }, { "epoch": 127.97323135755258, "grad_norm": 0.24343577027320862, "learning_rate": 2.703618288065701e-05, "loss": 0.0061, "step": 200790 }, { "epoch": 127.97960484384959, "grad_norm": 0.2050786316394806, "learning_rate": 2.7031287132190765e-05, "loss": 0.0065, "step": 200800 }, { "epoch": 127.9859783301466, "grad_norm": 0.14434196054935455, "learning_rate": 2.7026391662816042e-05, "loss": 0.007, "step": 200810 }, { "epoch": 127.99235181644359, "grad_norm": 0.23856617510318756, "learning_rate": 2.7021496472592395e-05, "loss": 0.0111, "step": 200820 }, { "epoch": 127.9987253027406, "grad_norm": 0.23431099951267242, "learning_rate": 2.7016601561579223e-05, "loss": 0.0135, "step": 200830 }, { "epoch": 128.00509878903762, "grad_norm": 0.17556962370872498, "learning_rate": 2.7011706929836056e-05, "loss": 0.0096, "step": 200840 }, { "epoch": 128.0114722753346, "grad_norm": 0.17222192883491516, "learning_rate": 2.7006812577422346e-05, "loss": 0.0131, "step": 200850 }, { "epoch": 128.0178457616316, "grad_norm": 0.2893871068954468, "learning_rate": 2.7001918504397606e-05, "loss": 0.0096, "step": 200860 }, { "epoch": 128.0242192479286, "grad_norm": 0.19670899212360382, "learning_rate": 2.6997024710821228e-05, "loss": 0.0111, "step": 200870 }, { "epoch": 128.03059273422562, "grad_norm": 0.11805979162454605, "learning_rate": 2.6992131196752733e-05, "loss": 0.0164, "step": 200880 }, { "epoch": 128.03696622052263, "grad_norm": 0.18099980056285858, "learning_rate": 2.6987237962251576e-05, "loss": 0.0145, "step": 200890 }, { "epoch": 128.04333970681964, "grad_norm": 0.4033251404762268, "learning_rate": 2.6982345007377207e-05, "loss": 0.0108, "step": 200900 }, { "epoch": 128.04971319311665, "grad_norm": 0.37093913555145264, "learning_rate": 2.6977452332189052e-05, "loss": 0.0105, "step": 200910 }, { "epoch": 128.05608667941365, "grad_norm": 0.31927454471588135, "learning_rate": 2.6972559936746612e-05, "loss": 0.0052, "step": 200920 }, { "epoch": 128.06246016571063, "grad_norm": 0.22643785178661346, "learning_rate": 2.696766782110931e-05, "loss": 0.008, "step": 200930 }, { "epoch": 128.06883365200764, "grad_norm": 0.22454096376895905, "learning_rate": 2.6962775985336586e-05, "loss": 0.0121, "step": 200940 }, { "epoch": 128.07520713830465, "grad_norm": 0.16636742651462555, "learning_rate": 2.6957884429487867e-05, "loss": 0.0075, "step": 200950 }, { "epoch": 128.08158062460166, "grad_norm": 0.1990029662847519, "learning_rate": 2.6952993153622647e-05, "loss": 0.0064, "step": 200960 }, { "epoch": 128.08795411089866, "grad_norm": 0.22347576916217804, "learning_rate": 2.6948102157800282e-05, "loss": 0.0066, "step": 200970 }, { "epoch": 128.09432759719567, "grad_norm": 0.5067558884620667, "learning_rate": 2.6943211442080264e-05, "loss": 0.0087, "step": 200980 }, { "epoch": 128.10070108349268, "grad_norm": 0.2367233783006668, "learning_rate": 2.6938321006521967e-05, "loss": 0.005, "step": 200990 }, { "epoch": 128.1070745697897, "grad_norm": 0.35985928773880005, "learning_rate": 2.6933430851184887e-05, "loss": 0.0095, "step": 201000 }, { "epoch": 128.11344805608667, "grad_norm": 0.365728497505188, "learning_rate": 2.6928540976128362e-05, "loss": 0.0044, "step": 201010 }, { "epoch": 128.11982154238368, "grad_norm": 0.4418471157550812, "learning_rate": 2.692365138141187e-05, "loss": 0.009, "step": 201020 }, { "epoch": 128.12619502868068, "grad_norm": 0.21126984059810638, "learning_rate": 2.6918762067094777e-05, "loss": 0.0077, "step": 201030 }, { "epoch": 128.1325685149777, "grad_norm": 0.15385712683200836, "learning_rate": 2.691387303323656e-05, "loss": 0.0127, "step": 201040 }, { "epoch": 128.1389420012747, "grad_norm": 0.2718387842178345, "learning_rate": 2.6908984279896537e-05, "loss": 0.0076, "step": 201050 }, { "epoch": 128.1453154875717, "grad_norm": 0.5965019464492798, "learning_rate": 2.6904095807134184e-05, "loss": 0.0164, "step": 201060 }, { "epoch": 128.15168897386872, "grad_norm": 0.22329246997833252, "learning_rate": 2.689920761500887e-05, "loss": 0.0117, "step": 201070 }, { "epoch": 128.15806246016572, "grad_norm": 0.13188399374485016, "learning_rate": 2.6894319703579996e-05, "loss": 0.0047, "step": 201080 }, { "epoch": 128.1644359464627, "grad_norm": 0.09121834486722946, "learning_rate": 2.688943207290693e-05, "loss": 0.0069, "step": 201090 }, { "epoch": 128.1708094327597, "grad_norm": 0.4453027546405792, "learning_rate": 2.68845447230491e-05, "loss": 0.0188, "step": 201100 }, { "epoch": 128.17718291905672, "grad_norm": 0.18615788221359253, "learning_rate": 2.687965765406588e-05, "loss": 0.016, "step": 201110 }, { "epoch": 128.18355640535373, "grad_norm": 0.22961190342903137, "learning_rate": 2.6874770866016653e-05, "loss": 0.0104, "step": 201120 }, { "epoch": 128.18992989165073, "grad_norm": 0.10465698689222336, "learning_rate": 2.6869884358960774e-05, "loss": 0.008, "step": 201130 }, { "epoch": 128.19630337794774, "grad_norm": 0.19774428009986877, "learning_rate": 2.6864998132957653e-05, "loss": 0.0039, "step": 201140 }, { "epoch": 128.20267686424475, "grad_norm": 0.2653988301753998, "learning_rate": 2.6860112188066654e-05, "loss": 0.0093, "step": 201150 }, { "epoch": 128.20905035054176, "grad_norm": 0.3510421812534332, "learning_rate": 2.685522652434713e-05, "loss": 0.0073, "step": 201160 }, { "epoch": 128.21542383683874, "grad_norm": 0.20161433517932892, "learning_rate": 2.6850341141858447e-05, "loss": 0.0078, "step": 201170 }, { "epoch": 128.22179732313575, "grad_norm": 0.13608260452747345, "learning_rate": 2.684545604065999e-05, "loss": 0.0125, "step": 201180 }, { "epoch": 128.22817080943275, "grad_norm": 0.15047970414161682, "learning_rate": 2.6840571220811107e-05, "loss": 0.0082, "step": 201190 }, { "epoch": 128.23454429572976, "grad_norm": 0.5855533480644226, "learning_rate": 2.683568668237114e-05, "loss": 0.0215, "step": 201200 }, { "epoch": 128.24091778202677, "grad_norm": 0.18668928742408752, "learning_rate": 2.6830802425399458e-05, "loss": 0.0175, "step": 201210 }, { "epoch": 128.24729126832378, "grad_norm": 0.193975031375885, "learning_rate": 2.6825918449955383e-05, "loss": 0.0054, "step": 201220 }, { "epoch": 128.2536647546208, "grad_norm": 0.12121458351612091, "learning_rate": 2.68210347560983e-05, "loss": 0.0074, "step": 201230 }, { "epoch": 128.2600382409178, "grad_norm": 0.20039980113506317, "learning_rate": 2.681615134388753e-05, "loss": 0.0058, "step": 201240 }, { "epoch": 128.26641172721477, "grad_norm": 0.18997140228748322, "learning_rate": 2.68112682133824e-05, "loss": 0.004, "step": 201250 }, { "epoch": 128.27278521351178, "grad_norm": 0.5600760579109192, "learning_rate": 2.6806385364642246e-05, "loss": 0.0155, "step": 201260 }, { "epoch": 128.2791586998088, "grad_norm": 0.40120694041252136, "learning_rate": 2.6801502797726424e-05, "loss": 0.0069, "step": 201270 }, { "epoch": 128.2855321861058, "grad_norm": 0.3968052566051483, "learning_rate": 2.679662051269425e-05, "loss": 0.0075, "step": 201280 }, { "epoch": 128.2919056724028, "grad_norm": 0.32792818546295166, "learning_rate": 2.6791738509605037e-05, "loss": 0.0063, "step": 201290 }, { "epoch": 128.2982791586998, "grad_norm": 0.3480658531188965, "learning_rate": 2.67868567885181e-05, "loss": 0.0085, "step": 201300 }, { "epoch": 128.30465264499682, "grad_norm": 0.29507365822792053, "learning_rate": 2.6781975349492784e-05, "loss": 0.0141, "step": 201310 }, { "epoch": 128.31102613129383, "grad_norm": 0.5968568921089172, "learning_rate": 2.6777094192588392e-05, "loss": 0.0089, "step": 201320 }, { "epoch": 128.3173996175908, "grad_norm": 0.11710251122713089, "learning_rate": 2.677221331786423e-05, "loss": 0.0051, "step": 201330 }, { "epoch": 128.32377310388782, "grad_norm": 0.16336210072040558, "learning_rate": 2.676733272537959e-05, "loss": 0.0098, "step": 201340 }, { "epoch": 128.33014659018482, "grad_norm": 0.19184382259845734, "learning_rate": 2.676245241519381e-05, "loss": 0.0111, "step": 201350 }, { "epoch": 128.33652007648183, "grad_norm": 0.17981837689876556, "learning_rate": 2.6757572387366174e-05, "loss": 0.0092, "step": 201360 }, { "epoch": 128.34289356277884, "grad_norm": 0.16607683897018433, "learning_rate": 2.6752692641955978e-05, "loss": 0.006, "step": 201370 }, { "epoch": 128.34926704907585, "grad_norm": 0.4631597697734833, "learning_rate": 2.6747813179022498e-05, "loss": 0.0099, "step": 201380 }, { "epoch": 128.35564053537286, "grad_norm": 0.47018057107925415, "learning_rate": 2.6742933998625077e-05, "loss": 0.0083, "step": 201390 }, { "epoch": 128.36201402166986, "grad_norm": 0.4272370934486389, "learning_rate": 2.6738055100822936e-05, "loss": 0.0084, "step": 201400 }, { "epoch": 128.36838750796684, "grad_norm": 0.3102502226829529, "learning_rate": 2.6733176485675395e-05, "loss": 0.0056, "step": 201410 }, { "epoch": 128.37476099426385, "grad_norm": 0.23946969211101532, "learning_rate": 2.6728298153241715e-05, "loss": 0.006, "step": 201420 }, { "epoch": 128.38113448056086, "grad_norm": 0.33482825756073, "learning_rate": 2.672342010358122e-05, "loss": 0.0063, "step": 201430 }, { "epoch": 128.38750796685787, "grad_norm": 0.23752012848854065, "learning_rate": 2.671854233675311e-05, "loss": 0.0047, "step": 201440 }, { "epoch": 128.39388145315488, "grad_norm": 0.22271791100502014, "learning_rate": 2.671366485281671e-05, "loss": 0.0068, "step": 201450 }, { "epoch": 128.40025493945188, "grad_norm": 0.43115323781967163, "learning_rate": 2.670878765183127e-05, "loss": 0.0114, "step": 201460 }, { "epoch": 128.4066284257489, "grad_norm": 0.16715078055858612, "learning_rate": 2.670391073385605e-05, "loss": 0.0087, "step": 201470 }, { "epoch": 128.4130019120459, "grad_norm": 0.41718506813049316, "learning_rate": 2.6699034098950293e-05, "loss": 0.0091, "step": 201480 }, { "epoch": 128.4193753983429, "grad_norm": 0.3674412965774536, "learning_rate": 2.6694157747173287e-05, "loss": 0.0102, "step": 201490 }, { "epoch": 128.4257488846399, "grad_norm": 0.428159236907959, "learning_rate": 2.6689281678584278e-05, "loss": 0.007, "step": 201500 }, { "epoch": 128.4321223709369, "grad_norm": 0.13151143491268158, "learning_rate": 2.6684405893242493e-05, "loss": 0.0128, "step": 201510 }, { "epoch": 128.4384958572339, "grad_norm": 0.1838671714067459, "learning_rate": 2.6679530391207174e-05, "loss": 0.0107, "step": 201520 }, { "epoch": 128.4448693435309, "grad_norm": 0.32290318608283997, "learning_rate": 2.667465517253762e-05, "loss": 0.0118, "step": 201530 }, { "epoch": 128.45124282982792, "grad_norm": 0.15457265079021454, "learning_rate": 2.666978023729299e-05, "loss": 0.0075, "step": 201540 }, { "epoch": 128.45761631612493, "grad_norm": 0.2038186490535736, "learning_rate": 2.6664905585532575e-05, "loss": 0.0127, "step": 201550 }, { "epoch": 128.46398980242193, "grad_norm": 0.2714236080646515, "learning_rate": 2.6660031217315573e-05, "loss": 0.0085, "step": 201560 }, { "epoch": 128.47036328871894, "grad_norm": 0.3566865622997284, "learning_rate": 2.6655157132701265e-05, "loss": 0.0078, "step": 201570 }, { "epoch": 128.47673677501592, "grad_norm": 0.2617378234863281, "learning_rate": 2.6650283331748793e-05, "loss": 0.0115, "step": 201580 }, { "epoch": 128.48311026131293, "grad_norm": 0.258346289396286, "learning_rate": 2.6645409814517452e-05, "loss": 0.0108, "step": 201590 }, { "epoch": 128.48948374760994, "grad_norm": 0.3217862546443939, "learning_rate": 2.6640536581066433e-05, "loss": 0.0076, "step": 201600 }, { "epoch": 128.49585723390695, "grad_norm": 0.21698512136936188, "learning_rate": 2.663566363145495e-05, "loss": 0.0086, "step": 201610 }, { "epoch": 128.50223072020395, "grad_norm": 0.29522010684013367, "learning_rate": 2.663079096574219e-05, "loss": 0.0065, "step": 201620 }, { "epoch": 128.50860420650096, "grad_norm": 0.18469898402690887, "learning_rate": 2.6625918583987402e-05, "loss": 0.0065, "step": 201630 }, { "epoch": 128.51497769279797, "grad_norm": 0.5895656943321228, "learning_rate": 2.6621046486249777e-05, "loss": 0.0096, "step": 201640 }, { "epoch": 128.52135117909498, "grad_norm": 0.27485907077789307, "learning_rate": 2.6616174672588507e-05, "loss": 0.012, "step": 201650 }, { "epoch": 128.52772466539196, "grad_norm": 0.18595381081104279, "learning_rate": 2.6611303143062772e-05, "loss": 0.0082, "step": 201660 }, { "epoch": 128.53409815168897, "grad_norm": 0.18804024159908295, "learning_rate": 2.660643189773181e-05, "loss": 0.0061, "step": 201670 }, { "epoch": 128.54047163798597, "grad_norm": 0.14784978330135345, "learning_rate": 2.660156093665479e-05, "loss": 0.0089, "step": 201680 }, { "epoch": 128.54684512428298, "grad_norm": 0.13248702883720398, "learning_rate": 2.659669025989089e-05, "loss": 0.0243, "step": 201690 }, { "epoch": 128.55321861058, "grad_norm": 0.20314982533454895, "learning_rate": 2.659181986749928e-05, "loss": 0.0121, "step": 201700 }, { "epoch": 128.559592096877, "grad_norm": 0.47495874762535095, "learning_rate": 2.6586949759539203e-05, "loss": 0.0098, "step": 201710 }, { "epoch": 128.565965583174, "grad_norm": 0.3356379270553589, "learning_rate": 2.6582079936069747e-05, "loss": 0.0111, "step": 201720 }, { "epoch": 128.572339069471, "grad_norm": 0.30108532309532166, "learning_rate": 2.6577210397150155e-05, "loss": 0.0056, "step": 201730 }, { "epoch": 128.578712555768, "grad_norm": 0.2091444879770279, "learning_rate": 2.6572341142839552e-05, "loss": 0.0117, "step": 201740 }, { "epoch": 128.585086042065, "grad_norm": 0.18537040054798126, "learning_rate": 2.656747217319716e-05, "loss": 0.0103, "step": 201750 }, { "epoch": 128.591459528362, "grad_norm": 0.14974622428417206, "learning_rate": 2.6562603488282067e-05, "loss": 0.0097, "step": 201760 }, { "epoch": 128.59783301465902, "grad_norm": 0.34468311071395874, "learning_rate": 2.655773508815349e-05, "loss": 0.0054, "step": 201770 }, { "epoch": 128.60420650095602, "grad_norm": 0.5328456163406372, "learning_rate": 2.6552866972870572e-05, "loss": 0.0176, "step": 201780 }, { "epoch": 128.61057998725303, "grad_norm": 0.40615132451057434, "learning_rate": 2.6547999142492447e-05, "loss": 0.0071, "step": 201790 }, { "epoch": 128.61695347355004, "grad_norm": 0.1888032704591751, "learning_rate": 2.654313159707826e-05, "loss": 0.0114, "step": 201800 }, { "epoch": 128.62332695984705, "grad_norm": 0.568735659122467, "learning_rate": 2.6538264336687192e-05, "loss": 0.0136, "step": 201810 }, { "epoch": 128.62970044614403, "grad_norm": 0.37821081280708313, "learning_rate": 2.6533397361378354e-05, "loss": 0.0077, "step": 201820 }, { "epoch": 128.63607393244104, "grad_norm": 0.20578810572624207, "learning_rate": 2.6528530671210904e-05, "loss": 0.0065, "step": 201830 }, { "epoch": 128.64244741873804, "grad_norm": 0.5283336639404297, "learning_rate": 2.6523664266243946e-05, "loss": 0.0083, "step": 201840 }, { "epoch": 128.64882090503505, "grad_norm": 0.4756122827529907, "learning_rate": 2.651879814653665e-05, "loss": 0.0051, "step": 201850 }, { "epoch": 128.65519439133206, "grad_norm": 0.5701627731323242, "learning_rate": 2.6513932312148126e-05, "loss": 0.0142, "step": 201860 }, { "epoch": 128.66156787762907, "grad_norm": 0.3024049997329712, "learning_rate": 2.6509066763137502e-05, "loss": 0.007, "step": 201870 }, { "epoch": 128.66794136392608, "grad_norm": 0.36942753195762634, "learning_rate": 2.6504201499563873e-05, "loss": 0.0106, "step": 201880 }, { "epoch": 128.67431485022308, "grad_norm": 0.2773287296295166, "learning_rate": 2.6499336521486406e-05, "loss": 0.0073, "step": 201890 }, { "epoch": 128.68068833652006, "grad_norm": 0.32265010476112366, "learning_rate": 2.6494471828964185e-05, "loss": 0.0067, "step": 201900 }, { "epoch": 128.68706182281707, "grad_norm": 0.2764343023300171, "learning_rate": 2.6489607422056318e-05, "loss": 0.0185, "step": 201910 }, { "epoch": 128.69343530911408, "grad_norm": 0.7233279347419739, "learning_rate": 2.6484743300821903e-05, "loss": 0.0107, "step": 201920 }, { "epoch": 128.6998087954111, "grad_norm": 0.32044875621795654, "learning_rate": 2.647987946532008e-05, "loss": 0.0055, "step": 201930 }, { "epoch": 128.7061822817081, "grad_norm": 0.18795166909694672, "learning_rate": 2.6475015915609928e-05, "loss": 0.0091, "step": 201940 }, { "epoch": 128.7125557680051, "grad_norm": 0.23943138122558594, "learning_rate": 2.6470152651750546e-05, "loss": 0.019, "step": 201950 }, { "epoch": 128.7189292543021, "grad_norm": 0.17294742166996002, "learning_rate": 2.646528967380102e-05, "loss": 0.0042, "step": 201960 }, { "epoch": 128.72530274059912, "grad_norm": 0.1914478987455368, "learning_rate": 2.646042698182043e-05, "loss": 0.0046, "step": 201970 }, { "epoch": 128.7316762268961, "grad_norm": 0.25807562470436096, "learning_rate": 2.645556457586789e-05, "loss": 0.0071, "step": 201980 }, { "epoch": 128.7380497131931, "grad_norm": 0.13331399857997894, "learning_rate": 2.6450702456002474e-05, "loss": 0.0121, "step": 201990 }, { "epoch": 128.74442319949011, "grad_norm": 0.2777356505393982, "learning_rate": 2.6445840622283257e-05, "loss": 0.0238, "step": 202000 }, { "epoch": 128.75079668578712, "grad_norm": 0.21796062588691711, "learning_rate": 2.64409790747693e-05, "loss": 0.008, "step": 202010 }, { "epoch": 128.75717017208413, "grad_norm": 0.2600943446159363, "learning_rate": 2.6436117813519705e-05, "loss": 0.0069, "step": 202020 }, { "epoch": 128.76354365838114, "grad_norm": 0.3485572338104248, "learning_rate": 2.6431256838593533e-05, "loss": 0.0145, "step": 202030 }, { "epoch": 128.76991714467815, "grad_norm": 0.10060109943151474, "learning_rate": 2.6426396150049837e-05, "loss": 0.0109, "step": 202040 }, { "epoch": 128.77629063097515, "grad_norm": 0.10025067627429962, "learning_rate": 2.6421535747947667e-05, "loss": 0.0115, "step": 202050 }, { "epoch": 128.78266411727213, "grad_norm": 0.40822428464889526, "learning_rate": 2.6416675632346117e-05, "loss": 0.0109, "step": 202060 }, { "epoch": 128.78903760356914, "grad_norm": 0.17337612807750702, "learning_rate": 2.641181580330423e-05, "loss": 0.006, "step": 202070 }, { "epoch": 128.79541108986615, "grad_norm": 0.23642848432064056, "learning_rate": 2.6406956260881045e-05, "loss": 0.0108, "step": 202080 }, { "epoch": 128.80178457616316, "grad_norm": 0.2946702241897583, "learning_rate": 2.6402097005135596e-05, "loss": 0.0052, "step": 202090 }, { "epoch": 128.80815806246017, "grad_norm": 0.30230385065078735, "learning_rate": 2.639723803612699e-05, "loss": 0.0103, "step": 202100 }, { "epoch": 128.81453154875717, "grad_norm": 0.24974043667316437, "learning_rate": 2.63923793539142e-05, "loss": 0.0076, "step": 202110 }, { "epoch": 128.82090503505418, "grad_norm": 0.21464158594608307, "learning_rate": 2.6387520958556296e-05, "loss": 0.0104, "step": 202120 }, { "epoch": 128.8272785213512, "grad_norm": 0.18596114218235016, "learning_rate": 2.6382662850112288e-05, "loss": 0.0056, "step": 202130 }, { "epoch": 128.8336520076482, "grad_norm": 0.355320006608963, "learning_rate": 2.6377805028641268e-05, "loss": 0.0147, "step": 202140 }, { "epoch": 128.84002549394518, "grad_norm": 0.3553478419780731, "learning_rate": 2.637294749420217e-05, "loss": 0.0142, "step": 202150 }, { "epoch": 128.84639898024218, "grad_norm": 0.21570025384426117, "learning_rate": 2.6368090246854104e-05, "loss": 0.0107, "step": 202160 }, { "epoch": 128.8527724665392, "grad_norm": 0.41694822907447815, "learning_rate": 2.6363233286656043e-05, "loss": 0.0108, "step": 202170 }, { "epoch": 128.8591459528362, "grad_norm": 0.2748437225818634, "learning_rate": 2.6358376613667014e-05, "loss": 0.0059, "step": 202180 }, { "epoch": 128.8655194391332, "grad_norm": 0.5416483283042908, "learning_rate": 2.6353520227946015e-05, "loss": 0.0078, "step": 202190 }, { "epoch": 128.87189292543022, "grad_norm": 0.20469613373279572, "learning_rate": 2.6348664129552092e-05, "loss": 0.0105, "step": 202200 }, { "epoch": 128.87826641172722, "grad_norm": 0.23214127123355865, "learning_rate": 2.6343808318544228e-05, "loss": 0.0074, "step": 202210 }, { "epoch": 128.88463989802423, "grad_norm": 0.26488614082336426, "learning_rate": 2.633895279498143e-05, "loss": 0.0054, "step": 202220 }, { "epoch": 128.8910133843212, "grad_norm": 0.13827252388000488, "learning_rate": 2.633409755892267e-05, "loss": 0.0055, "step": 202230 }, { "epoch": 128.89738687061822, "grad_norm": 0.13809968531131744, "learning_rate": 2.6329242610426996e-05, "loss": 0.0057, "step": 202240 }, { "epoch": 128.90376035691523, "grad_norm": 0.1700212061405182, "learning_rate": 2.6324387949553363e-05, "loss": 0.006, "step": 202250 }, { "epoch": 128.91013384321224, "grad_norm": 0.2074444741010666, "learning_rate": 2.631953357636078e-05, "loss": 0.005, "step": 202260 }, { "epoch": 128.91650732950924, "grad_norm": 0.34349799156188965, "learning_rate": 2.63146794909082e-05, "loss": 0.0075, "step": 202270 }, { "epoch": 128.92288081580625, "grad_norm": 0.6581512093544006, "learning_rate": 2.6309825693254663e-05, "loss": 0.0206, "step": 202280 }, { "epoch": 128.92925430210326, "grad_norm": 0.12933507561683655, "learning_rate": 2.6304972183459076e-05, "loss": 0.006, "step": 202290 }, { "epoch": 128.93562778840027, "grad_norm": 0.28715604543685913, "learning_rate": 2.6300118961580467e-05, "loss": 0.017, "step": 202300 }, { "epoch": 128.94200127469725, "grad_norm": 0.19940119981765747, "learning_rate": 2.6295266027677766e-05, "loss": 0.0055, "step": 202310 }, { "epoch": 128.94837476099426, "grad_norm": 0.14642265439033508, "learning_rate": 2.629041338181001e-05, "loss": 0.0065, "step": 202320 }, { "epoch": 128.95474824729126, "grad_norm": 0.23565346002578735, "learning_rate": 2.6285561024036065e-05, "loss": 0.0061, "step": 202330 }, { "epoch": 128.96112173358827, "grad_norm": 0.19423213601112366, "learning_rate": 2.628070895441498e-05, "loss": 0.0145, "step": 202340 }, { "epoch": 128.96749521988528, "grad_norm": 0.27094408869743347, "learning_rate": 2.6275857173005662e-05, "loss": 0.0142, "step": 202350 }, { "epoch": 128.9738687061823, "grad_norm": 0.22188986837863922, "learning_rate": 2.627100567986709e-05, "loss": 0.0099, "step": 202360 }, { "epoch": 128.9802421924793, "grad_norm": 0.3882010579109192, "learning_rate": 2.6266154475058168e-05, "loss": 0.0052, "step": 202370 }, { "epoch": 128.9866156787763, "grad_norm": 0.10922522097826004, "learning_rate": 2.626130355863791e-05, "loss": 0.0041, "step": 202380 }, { "epoch": 128.99298916507328, "grad_norm": 0.4033176004886627, "learning_rate": 2.625645293066522e-05, "loss": 0.009, "step": 202390 }, { "epoch": 128.9993626513703, "grad_norm": 0.15650835633277893, "learning_rate": 2.6251602591199053e-05, "loss": 0.0069, "step": 202400 }, { "epoch": 129.0057361376673, "grad_norm": 0.172312930226326, "learning_rate": 2.6246752540298315e-05, "loss": 0.0068, "step": 202410 }, { "epoch": 129.0121096239643, "grad_norm": 0.5073386430740356, "learning_rate": 2.6241902778021994e-05, "loss": 0.0073, "step": 202420 }, { "epoch": 129.01848311026131, "grad_norm": 0.26125961542129517, "learning_rate": 2.623705330442895e-05, "loss": 0.0162, "step": 202430 }, { "epoch": 129.02485659655832, "grad_norm": 0.1797134280204773, "learning_rate": 2.6232204119578162e-05, "loss": 0.0051, "step": 202440 }, { "epoch": 129.03123008285533, "grad_norm": 0.11675703525543213, "learning_rate": 2.6227355223528514e-05, "loss": 0.007, "step": 202450 }, { "epoch": 129.03760356915234, "grad_norm": 0.28571367263793945, "learning_rate": 2.622250661633899e-05, "loss": 0.0059, "step": 202460 }, { "epoch": 129.04397705544932, "grad_norm": 0.2734951376914978, "learning_rate": 2.6217658298068414e-05, "loss": 0.0071, "step": 202470 }, { "epoch": 129.05035054174633, "grad_norm": 0.49292874336242676, "learning_rate": 2.6212810268775767e-05, "loss": 0.0098, "step": 202480 }, { "epoch": 129.05672402804333, "grad_norm": 0.18214693665504456, "learning_rate": 2.6207962528519936e-05, "loss": 0.0062, "step": 202490 }, { "epoch": 129.06309751434034, "grad_norm": 0.09160583466291428, "learning_rate": 2.6203115077359818e-05, "loss": 0.0127, "step": 202500 }, { "epoch": 129.06947100063735, "grad_norm": 0.2528248131275177, "learning_rate": 2.6198267915354302e-05, "loss": 0.0049, "step": 202510 }, { "epoch": 129.07584448693436, "grad_norm": 0.16120734810829163, "learning_rate": 2.6193421042562328e-05, "loss": 0.0066, "step": 202520 }, { "epoch": 129.08221797323137, "grad_norm": 0.35117223858833313, "learning_rate": 2.6188574459042757e-05, "loss": 0.0169, "step": 202530 }, { "epoch": 129.08859145952837, "grad_norm": 0.1304357945919037, "learning_rate": 2.61837281648545e-05, "loss": 0.0074, "step": 202540 }, { "epoch": 129.09496494582535, "grad_norm": 0.18364275991916656, "learning_rate": 2.6178882160056405e-05, "loss": 0.0079, "step": 202550 }, { "epoch": 129.10133843212236, "grad_norm": 0.21299563348293304, "learning_rate": 2.61740364447074e-05, "loss": 0.0087, "step": 202560 }, { "epoch": 129.10771191841937, "grad_norm": 0.40032705664634705, "learning_rate": 2.616919101886635e-05, "loss": 0.0075, "step": 202570 }, { "epoch": 129.11408540471638, "grad_norm": 0.32902464270591736, "learning_rate": 2.6164345882592135e-05, "loss": 0.0088, "step": 202580 }, { "epoch": 129.12045889101339, "grad_norm": 0.3389374613761902, "learning_rate": 2.6159501035943594e-05, "loss": 0.0105, "step": 202590 }, { "epoch": 129.1268323773104, "grad_norm": 0.21612390875816345, "learning_rate": 2.6154656478979644e-05, "loss": 0.0047, "step": 202600 }, { "epoch": 129.1332058636074, "grad_norm": 0.38838982582092285, "learning_rate": 2.614981221175914e-05, "loss": 0.0093, "step": 202610 }, { "epoch": 129.1395793499044, "grad_norm": 0.15586718916893005, "learning_rate": 2.614496823434093e-05, "loss": 0.0063, "step": 202620 }, { "epoch": 129.1459528362014, "grad_norm": 0.11626745760440826, "learning_rate": 2.6140124546783863e-05, "loss": 0.0095, "step": 202630 }, { "epoch": 129.1523263224984, "grad_norm": 0.33422550559043884, "learning_rate": 2.6135281149146823e-05, "loss": 0.0099, "step": 202640 }, { "epoch": 129.1586998087954, "grad_norm": 0.4161488711833954, "learning_rate": 2.6130438041488653e-05, "loss": 0.0088, "step": 202650 }, { "epoch": 129.1650732950924, "grad_norm": 0.17237727344036102, "learning_rate": 2.6125595223868195e-05, "loss": 0.0064, "step": 202660 }, { "epoch": 129.17144678138942, "grad_norm": 0.1467214822769165, "learning_rate": 2.6120752696344296e-05, "loss": 0.0069, "step": 202670 }, { "epoch": 129.17782026768643, "grad_norm": 0.1396697461605072, "learning_rate": 2.6115910458975783e-05, "loss": 0.0043, "step": 202680 }, { "epoch": 129.18419375398344, "grad_norm": 0.27544263005256653, "learning_rate": 2.6111068511821522e-05, "loss": 0.0131, "step": 202690 }, { "epoch": 129.19056724028044, "grad_norm": 0.13144370913505554, "learning_rate": 2.610622685494034e-05, "loss": 0.0061, "step": 202700 }, { "epoch": 129.19694072657742, "grad_norm": 0.15196441113948822, "learning_rate": 2.6101385488391062e-05, "loss": 0.01, "step": 202710 }, { "epoch": 129.20331421287443, "grad_norm": 0.5668461322784424, "learning_rate": 2.6096544412232494e-05, "loss": 0.0097, "step": 202720 }, { "epoch": 129.20968769917144, "grad_norm": 0.14445194602012634, "learning_rate": 2.609170362652349e-05, "loss": 0.008, "step": 202730 }, { "epoch": 129.21606118546845, "grad_norm": 0.217608243227005, "learning_rate": 2.6086863131322868e-05, "loss": 0.0129, "step": 202740 }, { "epoch": 129.22243467176546, "grad_norm": 0.35155317187309265, "learning_rate": 2.6082022926689432e-05, "loss": 0.0065, "step": 202750 }, { "epoch": 129.22880815806246, "grad_norm": 0.1446237862110138, "learning_rate": 2.6077183012681982e-05, "loss": 0.0078, "step": 202760 }, { "epoch": 129.23518164435947, "grad_norm": 0.22685512900352478, "learning_rate": 2.6072343389359367e-05, "loss": 0.0133, "step": 202770 }, { "epoch": 129.24155513065648, "grad_norm": 0.10368503630161285, "learning_rate": 2.6067504056780374e-05, "loss": 0.0085, "step": 202780 }, { "epoch": 129.2479286169535, "grad_norm": 0.4845616817474365, "learning_rate": 2.6062665015003794e-05, "loss": 0.009, "step": 202790 }, { "epoch": 129.25430210325047, "grad_norm": 0.5047348141670227, "learning_rate": 2.6057826264088424e-05, "loss": 0.0118, "step": 202800 }, { "epoch": 129.26067558954747, "grad_norm": 0.4574735760688782, "learning_rate": 2.605298780409311e-05, "loss": 0.0047, "step": 202810 }, { "epoch": 129.26704907584448, "grad_norm": 0.11115851998329163, "learning_rate": 2.604814963507656e-05, "loss": 0.0186, "step": 202820 }, { "epoch": 129.2734225621415, "grad_norm": 0.23267658054828644, "learning_rate": 2.604331175709762e-05, "loss": 0.0146, "step": 202830 }, { "epoch": 129.2797960484385, "grad_norm": 0.4211127460002899, "learning_rate": 2.603847417021505e-05, "loss": 0.007, "step": 202840 }, { "epoch": 129.2861695347355, "grad_norm": 0.3651857376098633, "learning_rate": 2.6033636874487676e-05, "loss": 0.0075, "step": 202850 }, { "epoch": 129.29254302103251, "grad_norm": 0.19807501137256622, "learning_rate": 2.602879986997421e-05, "loss": 0.0082, "step": 202860 }, { "epoch": 129.29891650732952, "grad_norm": 0.3074679672718048, "learning_rate": 2.602396315673347e-05, "loss": 0.0088, "step": 202870 }, { "epoch": 129.3052899936265, "grad_norm": 0.24016742408275604, "learning_rate": 2.60191267348242e-05, "loss": 0.013, "step": 202880 }, { "epoch": 129.3116634799235, "grad_norm": 0.16443519294261932, "learning_rate": 2.601429060430522e-05, "loss": 0.0064, "step": 202890 }, { "epoch": 129.31803696622052, "grad_norm": 0.17981985211372375, "learning_rate": 2.6009454765235212e-05, "loss": 0.0109, "step": 202900 }, { "epoch": 129.32441045251753, "grad_norm": 0.0979493260383606, "learning_rate": 2.600461921767301e-05, "loss": 0.0088, "step": 202910 }, { "epoch": 129.33078393881453, "grad_norm": 0.22251436114311218, "learning_rate": 2.5999783961677327e-05, "loss": 0.0056, "step": 202920 }, { "epoch": 129.33715742511154, "grad_norm": 0.6743789315223694, "learning_rate": 2.5994948997306935e-05, "loss": 0.008, "step": 202930 }, { "epoch": 129.34353091140855, "grad_norm": 0.1749187856912613, "learning_rate": 2.5990114324620558e-05, "loss": 0.0085, "step": 202940 }, { "epoch": 129.34990439770556, "grad_norm": 0.31134966015815735, "learning_rate": 2.5985279943676977e-05, "loss": 0.0038, "step": 202950 }, { "epoch": 129.35627788400254, "grad_norm": 0.18267089128494263, "learning_rate": 2.5980445854534924e-05, "loss": 0.0058, "step": 202960 }, { "epoch": 129.36265137029955, "grad_norm": 0.2632170617580414, "learning_rate": 2.597561205725313e-05, "loss": 0.0068, "step": 202970 }, { "epoch": 129.36902485659655, "grad_norm": 0.4263994097709656, "learning_rate": 2.5970778551890312e-05, "loss": 0.013, "step": 202980 }, { "epoch": 129.37539834289356, "grad_norm": 0.14906151592731476, "learning_rate": 2.596594533850526e-05, "loss": 0.0073, "step": 202990 }, { "epoch": 129.38177182919057, "grad_norm": 0.30064916610717773, "learning_rate": 2.596111241715663e-05, "loss": 0.0089, "step": 203000 }, { "epoch": 129.38814531548758, "grad_norm": 0.18335579335689545, "learning_rate": 2.5956279787903193e-05, "loss": 0.0214, "step": 203010 }, { "epoch": 129.39451880178459, "grad_norm": 0.18681636452674866, "learning_rate": 2.595144745080364e-05, "loss": 0.0169, "step": 203020 }, { "epoch": 129.4008922880816, "grad_norm": 0.28161710500717163, "learning_rate": 2.5946615405916746e-05, "loss": 0.0114, "step": 203030 }, { "epoch": 129.40726577437857, "grad_norm": 0.234013631939888, "learning_rate": 2.5941783653301143e-05, "loss": 0.0091, "step": 203040 }, { "epoch": 129.41363926067558, "grad_norm": 0.324063777923584, "learning_rate": 2.5936952193015608e-05, "loss": 0.0131, "step": 203050 }, { "epoch": 129.4200127469726, "grad_norm": 0.18940018117427826, "learning_rate": 2.5932121025118822e-05, "loss": 0.0064, "step": 203060 }, { "epoch": 129.4263862332696, "grad_norm": 0.24792657792568207, "learning_rate": 2.5927290149669493e-05, "loss": 0.0127, "step": 203070 }, { "epoch": 129.4327597195666, "grad_norm": 0.12790410220623016, "learning_rate": 2.5922459566726288e-05, "loss": 0.005, "step": 203080 }, { "epoch": 129.4391332058636, "grad_norm": 0.17799130082130432, "learning_rate": 2.591762927634796e-05, "loss": 0.0086, "step": 203090 }, { "epoch": 129.44550669216062, "grad_norm": 0.2548878490924835, "learning_rate": 2.591279927859317e-05, "loss": 0.0114, "step": 203100 }, { "epoch": 129.45188017845763, "grad_norm": 0.2024044543504715, "learning_rate": 2.5907969573520606e-05, "loss": 0.0041, "step": 203110 }, { "epoch": 129.4582536647546, "grad_norm": 0.3045634925365448, "learning_rate": 2.590314016118895e-05, "loss": 0.0124, "step": 203120 }, { "epoch": 129.46462715105162, "grad_norm": 0.33561813831329346, "learning_rate": 2.589831104165691e-05, "loss": 0.009, "step": 203130 }, { "epoch": 129.47100063734862, "grad_norm": 0.1336328387260437, "learning_rate": 2.5893482214983138e-05, "loss": 0.0069, "step": 203140 }, { "epoch": 129.47737412364563, "grad_norm": 0.18706083297729492, "learning_rate": 2.5888653681226322e-05, "loss": 0.0171, "step": 203150 }, { "epoch": 129.48374760994264, "grad_norm": 0.3050249516963959, "learning_rate": 2.5883825440445108e-05, "loss": 0.0087, "step": 203160 }, { "epoch": 129.49012109623965, "grad_norm": 0.1969534456729889, "learning_rate": 2.5878997492698228e-05, "loss": 0.0064, "step": 203170 }, { "epoch": 129.49649458253666, "grad_norm": 0.20880259573459625, "learning_rate": 2.587416983804426e-05, "loss": 0.0077, "step": 203180 }, { "epoch": 129.50286806883366, "grad_norm": 0.13278162479400635, "learning_rate": 2.5869342476541923e-05, "loss": 0.0081, "step": 203190 }, { "epoch": 129.50924155513064, "grad_norm": 0.36129990220069885, "learning_rate": 2.5864515408249844e-05, "loss": 0.0063, "step": 203200 }, { "epoch": 129.51561504142765, "grad_norm": 0.7779558300971985, "learning_rate": 2.585968863322673e-05, "loss": 0.0062, "step": 203210 }, { "epoch": 129.52198852772466, "grad_norm": 0.12443215399980545, "learning_rate": 2.5854862151531146e-05, "loss": 0.0087, "step": 203220 }, { "epoch": 129.52836201402167, "grad_norm": 0.39480623602867126, "learning_rate": 2.585003596322181e-05, "loss": 0.0064, "step": 203230 }, { "epoch": 129.53473550031867, "grad_norm": 0.18563735485076904, "learning_rate": 2.5845210068357333e-05, "loss": 0.0044, "step": 203240 }, { "epoch": 129.54110898661568, "grad_norm": 0.3223111629486084, "learning_rate": 2.5840384466996366e-05, "loss": 0.0061, "step": 203250 }, { "epoch": 129.5474824729127, "grad_norm": 0.18024712800979614, "learning_rate": 2.583555915919751e-05, "loss": 0.011, "step": 203260 }, { "epoch": 129.5538559592097, "grad_norm": 0.2438982129096985, "learning_rate": 2.5830734145019453e-05, "loss": 0.0098, "step": 203270 }, { "epoch": 129.56022944550668, "grad_norm": 0.3518161177635193, "learning_rate": 2.5825909424520788e-05, "loss": 0.0136, "step": 203280 }, { "epoch": 129.5666029318037, "grad_norm": 0.19394667446613312, "learning_rate": 2.582108499776016e-05, "loss": 0.0082, "step": 203290 }, { "epoch": 129.5729764181007, "grad_norm": 0.26625239849090576, "learning_rate": 2.581626086479615e-05, "loss": 0.0138, "step": 203300 }, { "epoch": 129.5793499043977, "grad_norm": 0.27189064025878906, "learning_rate": 2.581143702568742e-05, "loss": 0.0048, "step": 203310 }, { "epoch": 129.5857233906947, "grad_norm": 0.24016514420509338, "learning_rate": 2.580661348049257e-05, "loss": 0.0075, "step": 203320 }, { "epoch": 129.59209687699172, "grad_norm": 0.1307680606842041, "learning_rate": 2.5801790229270213e-05, "loss": 0.0124, "step": 203330 }, { "epoch": 129.59847036328873, "grad_norm": 0.19258537888526917, "learning_rate": 2.5796967272078926e-05, "loss": 0.007, "step": 203340 }, { "epoch": 129.60484384958573, "grad_norm": 0.15844811499118805, "learning_rate": 2.5792144608977363e-05, "loss": 0.0123, "step": 203350 }, { "epoch": 129.61121733588274, "grad_norm": 0.19302867352962494, "learning_rate": 2.5787322240024092e-05, "loss": 0.01, "step": 203360 }, { "epoch": 129.61759082217972, "grad_norm": 0.5176437497138977, "learning_rate": 2.5782500165277723e-05, "loss": 0.0106, "step": 203370 }, { "epoch": 129.62396430847673, "grad_norm": 0.3900178372859955, "learning_rate": 2.5777678384796845e-05, "loss": 0.0102, "step": 203380 }, { "epoch": 129.63033779477374, "grad_norm": 0.25941208004951477, "learning_rate": 2.577285689864002e-05, "loss": 0.0106, "step": 203390 }, { "epoch": 129.63671128107075, "grad_norm": 0.5293005704879761, "learning_rate": 2.5768035706865875e-05, "loss": 0.0132, "step": 203400 }, { "epoch": 129.64308476736775, "grad_norm": 0.3466799259185791, "learning_rate": 2.5763214809532975e-05, "loss": 0.0114, "step": 203410 }, { "epoch": 129.64945825366476, "grad_norm": 0.16785390675067902, "learning_rate": 2.57583942066999e-05, "loss": 0.008, "step": 203420 }, { "epoch": 129.65583173996177, "grad_norm": 0.25047382712364197, "learning_rate": 2.575357389842521e-05, "loss": 0.0132, "step": 203430 }, { "epoch": 129.66220522625878, "grad_norm": 0.3747299611568451, "learning_rate": 2.5748753884767497e-05, "loss": 0.0124, "step": 203440 }, { "epoch": 129.66857871255576, "grad_norm": 0.7488142848014832, "learning_rate": 2.5743934165785333e-05, "loss": 0.0122, "step": 203450 }, { "epoch": 129.67495219885276, "grad_norm": 0.312488853931427, "learning_rate": 2.573911474153726e-05, "loss": 0.0085, "step": 203460 }, { "epoch": 129.68132568514977, "grad_norm": 0.45418694615364075, "learning_rate": 2.573429561208184e-05, "loss": 0.0065, "step": 203470 }, { "epoch": 129.68769917144678, "grad_norm": 0.12168841809034348, "learning_rate": 2.5729476777477647e-05, "loss": 0.0066, "step": 203480 }, { "epoch": 129.6940726577438, "grad_norm": 0.2186029851436615, "learning_rate": 2.572465823778324e-05, "loss": 0.0076, "step": 203490 }, { "epoch": 129.7004461440408, "grad_norm": 0.26509636640548706, "learning_rate": 2.5719839993057142e-05, "loss": 0.01, "step": 203500 }, { "epoch": 129.7068196303378, "grad_norm": 0.16011284291744232, "learning_rate": 2.5715022043357896e-05, "loss": 0.0034, "step": 203510 }, { "epoch": 129.7131931166348, "grad_norm": 0.22467052936553955, "learning_rate": 2.5710204388744085e-05, "loss": 0.0086, "step": 203520 }, { "epoch": 129.7195666029318, "grad_norm": 0.2755078673362732, "learning_rate": 2.5705387029274218e-05, "loss": 0.007, "step": 203530 }, { "epoch": 129.7259400892288, "grad_norm": 0.2272060662508011, "learning_rate": 2.570056996500685e-05, "loss": 0.0144, "step": 203540 }, { "epoch": 129.7323135755258, "grad_norm": 0.21427418291568756, "learning_rate": 2.5695753196000466e-05, "loss": 0.0075, "step": 203550 }, { "epoch": 129.73868706182282, "grad_norm": 0.39880743622779846, "learning_rate": 2.5690936722313668e-05, "loss": 0.0114, "step": 203560 }, { "epoch": 129.74506054811982, "grad_norm": 0.6059857606887817, "learning_rate": 2.568612054400491e-05, "loss": 0.012, "step": 203570 }, { "epoch": 129.75143403441683, "grad_norm": 0.2822572588920593, "learning_rate": 2.5681304661132755e-05, "loss": 0.0123, "step": 203580 }, { "epoch": 129.75780752071384, "grad_norm": 0.4574579894542694, "learning_rate": 2.5676489073755693e-05, "loss": 0.011, "step": 203590 }, { "epoch": 129.76418100701085, "grad_norm": 0.226291224360466, "learning_rate": 2.5671673781932292e-05, "loss": 0.0101, "step": 203600 }, { "epoch": 129.77055449330783, "grad_norm": 0.1992907077074051, "learning_rate": 2.566685878572098e-05, "loss": 0.0073, "step": 203610 }, { "epoch": 129.77692797960484, "grad_norm": 0.1332230418920517, "learning_rate": 2.5662044085180337e-05, "loss": 0.008, "step": 203620 }, { "epoch": 129.78330146590184, "grad_norm": 0.1762542724609375, "learning_rate": 2.5657229680368832e-05, "loss": 0.0138, "step": 203630 }, { "epoch": 129.78967495219885, "grad_norm": 0.3905273675918579, "learning_rate": 2.5652415571344966e-05, "loss": 0.006, "step": 203640 }, { "epoch": 129.79604843849586, "grad_norm": 0.23811188340187073, "learning_rate": 2.564760175816723e-05, "loss": 0.0087, "step": 203650 }, { "epoch": 129.80242192479287, "grad_norm": 0.3286784589290619, "learning_rate": 2.5642788240894132e-05, "loss": 0.0078, "step": 203660 }, { "epoch": 129.80879541108987, "grad_norm": 0.34452104568481445, "learning_rate": 2.5637975019584167e-05, "loss": 0.0078, "step": 203670 }, { "epoch": 129.81516889738688, "grad_norm": 0.07334115356206894, "learning_rate": 2.5633162094295797e-05, "loss": 0.0066, "step": 203680 }, { "epoch": 129.82154238368386, "grad_norm": 0.1720007359981537, "learning_rate": 2.56283494650875e-05, "loss": 0.0095, "step": 203690 }, { "epoch": 129.82791586998087, "grad_norm": 0.3404781222343445, "learning_rate": 2.5623537132017804e-05, "loss": 0.0081, "step": 203700 }, { "epoch": 129.83428935627788, "grad_norm": 0.10320843756198883, "learning_rate": 2.5618725095145108e-05, "loss": 0.0055, "step": 203710 }, { "epoch": 129.8406628425749, "grad_norm": 0.12091800570487976, "learning_rate": 2.5613913354527942e-05, "loss": 0.0066, "step": 203720 }, { "epoch": 129.8470363288719, "grad_norm": 0.5115314722061157, "learning_rate": 2.560910191022473e-05, "loss": 0.0078, "step": 203730 }, { "epoch": 129.8534098151689, "grad_norm": 0.09364311397075653, "learning_rate": 2.5604290762294003e-05, "loss": 0.0079, "step": 203740 }, { "epoch": 129.8597833014659, "grad_norm": 0.19010484218597412, "learning_rate": 2.5599479910794132e-05, "loss": 0.0071, "step": 203750 }, { "epoch": 129.86615678776292, "grad_norm": 0.21015828847885132, "learning_rate": 2.559466935578364e-05, "loss": 0.0234, "step": 203760 }, { "epoch": 129.8725302740599, "grad_norm": 0.17504602670669556, "learning_rate": 2.5589859097320946e-05, "loss": 0.0076, "step": 203770 }, { "epoch": 129.8789037603569, "grad_norm": 0.40908533334732056, "learning_rate": 2.558504913546454e-05, "loss": 0.0068, "step": 203780 }, { "epoch": 129.8852772466539, "grad_norm": 0.26003724336624146, "learning_rate": 2.5580239470272804e-05, "loss": 0.0061, "step": 203790 }, { "epoch": 129.89165073295092, "grad_norm": 0.2799641489982605, "learning_rate": 2.557543010180423e-05, "loss": 0.0081, "step": 203800 }, { "epoch": 129.89802421924793, "grad_norm": 0.22890987992286682, "learning_rate": 2.557062103011725e-05, "loss": 0.0067, "step": 203810 }, { "epoch": 129.90439770554494, "grad_norm": 0.16260161995887756, "learning_rate": 2.5565812255270277e-05, "loss": 0.0086, "step": 203820 }, { "epoch": 129.91077119184195, "grad_norm": 0.2116127610206604, "learning_rate": 2.5561003777321746e-05, "loss": 0.0126, "step": 203830 }, { "epoch": 129.91714467813895, "grad_norm": 0.3136760890483856, "learning_rate": 2.55561955963301e-05, "loss": 0.0074, "step": 203840 }, { "epoch": 129.92351816443593, "grad_norm": 0.45059698820114136, "learning_rate": 2.555138771235377e-05, "loss": 0.0109, "step": 203850 }, { "epoch": 129.92989165073294, "grad_norm": 0.1941019594669342, "learning_rate": 2.554658012545116e-05, "loss": 0.006, "step": 203860 }, { "epoch": 129.93626513702995, "grad_norm": 0.5583682060241699, "learning_rate": 2.5541772835680666e-05, "loss": 0.014, "step": 203870 }, { "epoch": 129.94263862332696, "grad_norm": 0.18063558638095856, "learning_rate": 2.5536965843100763e-05, "loss": 0.0128, "step": 203880 }, { "epoch": 129.94901210962396, "grad_norm": 0.5083077549934387, "learning_rate": 2.5532159147769784e-05, "loss": 0.0135, "step": 203890 }, { "epoch": 129.95538559592097, "grad_norm": 0.22553980350494385, "learning_rate": 2.5527352749746193e-05, "loss": 0.006, "step": 203900 }, { "epoch": 129.96175908221798, "grad_norm": 0.33638080954551697, "learning_rate": 2.552254664908835e-05, "loss": 0.0107, "step": 203910 }, { "epoch": 129.968132568515, "grad_norm": 0.11249227821826935, "learning_rate": 2.5517740845854722e-05, "loss": 0.0062, "step": 203920 }, { "epoch": 129.97450605481197, "grad_norm": 0.30092769861221313, "learning_rate": 2.551293534010361e-05, "loss": 0.0201, "step": 203930 }, { "epoch": 129.98087954110898, "grad_norm": 0.4219214618206024, "learning_rate": 2.5508130131893475e-05, "loss": 0.0107, "step": 203940 }, { "epoch": 129.98725302740598, "grad_norm": 0.41899439692497253, "learning_rate": 2.5503325221282687e-05, "loss": 0.0072, "step": 203950 }, { "epoch": 129.993626513703, "grad_norm": 0.23885054886341095, "learning_rate": 2.5498520608329625e-05, "loss": 0.0088, "step": 203960 }, { "epoch": 130.0, "grad_norm": 0.4144451916217804, "learning_rate": 2.549371629309265e-05, "loss": 0.0055, "step": 203970 }, { "epoch": 130.006373486297, "grad_norm": 0.5424145460128784, "learning_rate": 2.5488912275630178e-05, "loss": 0.0123, "step": 203980 }, { "epoch": 130.01274697259402, "grad_norm": 0.4223598539829254, "learning_rate": 2.5484108556000562e-05, "loss": 0.0145, "step": 203990 }, { "epoch": 130.01912045889102, "grad_norm": 0.27775782346725464, "learning_rate": 2.547930513426218e-05, "loss": 0.0065, "step": 204000 }, { "epoch": 130.02549394518803, "grad_norm": 0.6879072785377502, "learning_rate": 2.5474502010473373e-05, "loss": 0.0164, "step": 204010 }, { "epoch": 130.031867431485, "grad_norm": 0.4655729830265045, "learning_rate": 2.546969918469254e-05, "loss": 0.0078, "step": 204020 }, { "epoch": 130.03824091778202, "grad_norm": 0.3302982449531555, "learning_rate": 2.5464896656978022e-05, "loss": 0.0132, "step": 204030 }, { "epoch": 130.04461440407903, "grad_norm": 0.21374432742595673, "learning_rate": 2.546009442738817e-05, "loss": 0.0094, "step": 204040 }, { "epoch": 130.05098789037604, "grad_norm": 0.2773694694042206, "learning_rate": 2.5455292495981326e-05, "loss": 0.0161, "step": 204050 }, { "epoch": 130.05736137667304, "grad_norm": 0.48528310656547546, "learning_rate": 2.5450490862815867e-05, "loss": 0.0065, "step": 204060 }, { "epoch": 130.06373486297005, "grad_norm": 0.2164522260427475, "learning_rate": 2.5445689527950133e-05, "loss": 0.0062, "step": 204070 }, { "epoch": 130.07010834926706, "grad_norm": 0.13984619081020355, "learning_rate": 2.5440888491442444e-05, "loss": 0.0046, "step": 204080 }, { "epoch": 130.07648183556407, "grad_norm": 0.19094206392765045, "learning_rate": 2.5436087753351134e-05, "loss": 0.0091, "step": 204090 }, { "epoch": 130.08285532186105, "grad_norm": 0.19528043270111084, "learning_rate": 2.5431287313734566e-05, "loss": 0.0115, "step": 204100 }, { "epoch": 130.08922880815805, "grad_norm": 0.4596606492996216, "learning_rate": 2.5426487172651053e-05, "loss": 0.0097, "step": 204110 }, { "epoch": 130.09560229445506, "grad_norm": 0.3454800546169281, "learning_rate": 2.5421687330158927e-05, "loss": 0.0048, "step": 204120 }, { "epoch": 130.10197578075207, "grad_norm": 0.27776193618774414, "learning_rate": 2.5416887786316502e-05, "loss": 0.0112, "step": 204130 }, { "epoch": 130.10834926704908, "grad_norm": 0.4476628303527832, "learning_rate": 2.5412088541182077e-05, "loss": 0.0102, "step": 204140 }, { "epoch": 130.1147227533461, "grad_norm": 0.2306569367647171, "learning_rate": 2.540728959481402e-05, "loss": 0.0074, "step": 204150 }, { "epoch": 130.1210962396431, "grad_norm": 0.20499762892723083, "learning_rate": 2.5402490947270608e-05, "loss": 0.0157, "step": 204160 }, { "epoch": 130.1274697259401, "grad_norm": 0.21188080310821533, "learning_rate": 2.5397692598610158e-05, "loss": 0.0093, "step": 204170 }, { "epoch": 130.13384321223708, "grad_norm": 0.2868412137031555, "learning_rate": 2.5392894548890945e-05, "loss": 0.0094, "step": 204180 }, { "epoch": 130.1402166985341, "grad_norm": 0.43743157386779785, "learning_rate": 2.5388096798171324e-05, "loss": 0.0092, "step": 204190 }, { "epoch": 130.1465901848311, "grad_norm": 0.24163392186164856, "learning_rate": 2.538329934650956e-05, "loss": 0.0109, "step": 204200 }, { "epoch": 130.1529636711281, "grad_norm": 0.16932232677936554, "learning_rate": 2.5378502193963948e-05, "loss": 0.0062, "step": 204210 }, { "epoch": 130.1593371574251, "grad_norm": 0.3387359380722046, "learning_rate": 2.537370534059277e-05, "loss": 0.0154, "step": 204220 }, { "epoch": 130.16571064372212, "grad_norm": 0.29826265573501587, "learning_rate": 2.536890878645434e-05, "loss": 0.0093, "step": 204230 }, { "epoch": 130.17208413001913, "grad_norm": 0.20850063860416412, "learning_rate": 2.536411253160692e-05, "loss": 0.011, "step": 204240 }, { "epoch": 130.17845761631614, "grad_norm": 0.27400171756744385, "learning_rate": 2.5359316576108795e-05, "loss": 0.0064, "step": 204250 }, { "epoch": 130.18483110261312, "grad_norm": 0.1968202441930771, "learning_rate": 2.5354520920018216e-05, "loss": 0.0126, "step": 204260 }, { "epoch": 130.19120458891013, "grad_norm": 0.40038472414016724, "learning_rate": 2.5349725563393516e-05, "loss": 0.0083, "step": 204270 }, { "epoch": 130.19757807520713, "grad_norm": 0.22628332674503326, "learning_rate": 2.534493050629289e-05, "loss": 0.0105, "step": 204280 }, { "epoch": 130.20395156150414, "grad_norm": 0.264456182718277, "learning_rate": 2.5340135748774653e-05, "loss": 0.0071, "step": 204290 }, { "epoch": 130.21032504780115, "grad_norm": 0.14140188694000244, "learning_rate": 2.533534129089703e-05, "loss": 0.0079, "step": 204300 }, { "epoch": 130.21669853409816, "grad_norm": 0.21525242924690247, "learning_rate": 2.533054713271834e-05, "loss": 0.0154, "step": 204310 }, { "epoch": 130.22307202039516, "grad_norm": 0.14802193641662598, "learning_rate": 2.5325753274296748e-05, "loss": 0.0117, "step": 204320 }, { "epoch": 130.22944550669217, "grad_norm": 0.16991712152957916, "learning_rate": 2.5320959715690573e-05, "loss": 0.005, "step": 204330 }, { "epoch": 130.23581899298915, "grad_norm": 0.2962770164012909, "learning_rate": 2.5316166456958022e-05, "loss": 0.006, "step": 204340 }, { "epoch": 130.24219247928616, "grad_norm": 0.5064720511436462, "learning_rate": 2.5311373498157392e-05, "loss": 0.0055, "step": 204350 }, { "epoch": 130.24856596558317, "grad_norm": 0.3200368881225586, "learning_rate": 2.5306580839346843e-05, "loss": 0.0088, "step": 204360 }, { "epoch": 130.25493945188018, "grad_norm": 0.30870670080184937, "learning_rate": 2.530178848058467e-05, "loss": 0.0085, "step": 204370 }, { "epoch": 130.26131293817718, "grad_norm": 0.386036217212677, "learning_rate": 2.5296996421929087e-05, "loss": 0.0177, "step": 204380 }, { "epoch": 130.2676864244742, "grad_norm": 0.13145886361598969, "learning_rate": 2.5292204663438325e-05, "loss": 0.0052, "step": 204390 }, { "epoch": 130.2740599107712, "grad_norm": 0.2739108204841614, "learning_rate": 2.528741320517058e-05, "loss": 0.0096, "step": 204400 }, { "epoch": 130.2804333970682, "grad_norm": 0.15848767757415771, "learning_rate": 2.5282622047184117e-05, "loss": 0.0088, "step": 204410 }, { "epoch": 130.2868068833652, "grad_norm": 0.13098376989364624, "learning_rate": 2.527783118953714e-05, "loss": 0.0055, "step": 204420 }, { "epoch": 130.2931803696622, "grad_norm": 0.20952025055885315, "learning_rate": 2.5273040632287848e-05, "loss": 0.0072, "step": 204430 }, { "epoch": 130.2995538559592, "grad_norm": 0.1418273001909256, "learning_rate": 2.526825037549444e-05, "loss": 0.0069, "step": 204440 }, { "epoch": 130.3059273422562, "grad_norm": 0.44514045119285583, "learning_rate": 2.526346041921518e-05, "loss": 0.0081, "step": 204450 }, { "epoch": 130.31230082855322, "grad_norm": 0.14574477076530457, "learning_rate": 2.5258670763508184e-05, "loss": 0.0112, "step": 204460 }, { "epoch": 130.31867431485023, "grad_norm": 0.2248559147119522, "learning_rate": 2.525388140843173e-05, "loss": 0.0127, "step": 204470 }, { "epoch": 130.32504780114724, "grad_norm": 0.16867229342460632, "learning_rate": 2.5249092354043956e-05, "loss": 0.0066, "step": 204480 }, { "epoch": 130.33142128744424, "grad_norm": 0.23356327414512634, "learning_rate": 2.524430360040312e-05, "loss": 0.01, "step": 204490 }, { "epoch": 130.33779477374122, "grad_norm": 0.15978217124938965, "learning_rate": 2.5239515147567328e-05, "loss": 0.004, "step": 204500 }, { "epoch": 130.34416826003823, "grad_norm": 0.3093019723892212, "learning_rate": 2.5234726995594827e-05, "loss": 0.0111, "step": 204510 }, { "epoch": 130.35054174633524, "grad_norm": 0.23174501955509186, "learning_rate": 2.5229939144543767e-05, "loss": 0.0045, "step": 204520 }, { "epoch": 130.35691523263225, "grad_norm": 0.4314015507698059, "learning_rate": 2.522515159447234e-05, "loss": 0.0121, "step": 204530 }, { "epoch": 130.36328871892925, "grad_norm": 0.16118159890174866, "learning_rate": 2.5220364345438696e-05, "loss": 0.0073, "step": 204540 }, { "epoch": 130.36966220522626, "grad_norm": 0.24550795555114746, "learning_rate": 2.5215577397501032e-05, "loss": 0.0095, "step": 204550 }, { "epoch": 130.37603569152327, "grad_norm": 0.1948462724685669, "learning_rate": 2.521079075071751e-05, "loss": 0.0106, "step": 204560 }, { "epoch": 130.38240917782028, "grad_norm": 0.17258599400520325, "learning_rate": 2.5206004405146284e-05, "loss": 0.0145, "step": 204570 }, { "epoch": 130.38878266411726, "grad_norm": 0.29702651500701904, "learning_rate": 2.520121836084549e-05, "loss": 0.0095, "step": 204580 }, { "epoch": 130.39515615041427, "grad_norm": 0.21919219195842743, "learning_rate": 2.5196432617873356e-05, "loss": 0.0081, "step": 204590 }, { "epoch": 130.40152963671127, "grad_norm": 0.346382200717926, "learning_rate": 2.5191647176287936e-05, "loss": 0.0084, "step": 204600 }, { "epoch": 130.40790312300828, "grad_norm": 0.18166732788085938, "learning_rate": 2.5186862036147447e-05, "loss": 0.0074, "step": 204610 }, { "epoch": 130.4142766093053, "grad_norm": 0.15286493301391602, "learning_rate": 2.518207719750999e-05, "loss": 0.0066, "step": 204620 }, { "epoch": 130.4206500956023, "grad_norm": 0.10362542420625687, "learning_rate": 2.5177292660433773e-05, "loss": 0.0083, "step": 204630 }, { "epoch": 130.4270235818993, "grad_norm": 0.37373635172843933, "learning_rate": 2.5172508424976838e-05, "loss": 0.0119, "step": 204640 }, { "epoch": 130.4333970681963, "grad_norm": 0.20126494765281677, "learning_rate": 2.5167724491197396e-05, "loss": 0.0055, "step": 204650 }, { "epoch": 130.43977055449332, "grad_norm": 0.3069716989994049, "learning_rate": 2.5162940859153523e-05, "loss": 0.0051, "step": 204660 }, { "epoch": 130.4461440407903, "grad_norm": 0.21693997085094452, "learning_rate": 2.5158157528903403e-05, "loss": 0.0078, "step": 204670 }, { "epoch": 130.4525175270873, "grad_norm": 0.1993100643157959, "learning_rate": 2.5153374500505088e-05, "loss": 0.0124, "step": 204680 }, { "epoch": 130.45889101338432, "grad_norm": 0.22253312170505524, "learning_rate": 2.514859177401676e-05, "loss": 0.0096, "step": 204690 }, { "epoch": 130.46526449968133, "grad_norm": 0.43263453245162964, "learning_rate": 2.5143809349496493e-05, "loss": 0.0154, "step": 204700 }, { "epoch": 130.47163798597833, "grad_norm": 0.3017336428165436, "learning_rate": 2.513902722700242e-05, "loss": 0.0156, "step": 204710 }, { "epoch": 130.47801147227534, "grad_norm": 0.12881934642791748, "learning_rate": 2.5134245406592617e-05, "loss": 0.0112, "step": 204720 }, { "epoch": 130.48438495857235, "grad_norm": 0.23425479233264923, "learning_rate": 2.5129463888325223e-05, "loss": 0.0144, "step": 204730 }, { "epoch": 130.49075844486936, "grad_norm": 0.17108944058418274, "learning_rate": 2.5124682672258325e-05, "loss": 0.0063, "step": 204740 }, { "epoch": 130.49713193116634, "grad_norm": 0.3847046494483948, "learning_rate": 2.5119901758450027e-05, "loss": 0.0139, "step": 204750 }, { "epoch": 130.50350541746334, "grad_norm": 0.5529868006706238, "learning_rate": 2.5115121146958387e-05, "loss": 0.0081, "step": 204760 }, { "epoch": 130.50987890376035, "grad_norm": 0.4789715111255646, "learning_rate": 2.5110340837841544e-05, "loss": 0.0118, "step": 204770 }, { "epoch": 130.51625239005736, "grad_norm": 0.14101357758045197, "learning_rate": 2.510556083115756e-05, "loss": 0.0084, "step": 204780 }, { "epoch": 130.52262587635437, "grad_norm": 0.11282239854335785, "learning_rate": 2.5100781126964513e-05, "loss": 0.0146, "step": 204790 }, { "epoch": 130.52899936265138, "grad_norm": 0.22952084243297577, "learning_rate": 2.5096001725320463e-05, "loss": 0.0076, "step": 204800 }, { "epoch": 130.53537284894838, "grad_norm": 0.4081629812717438, "learning_rate": 2.509122262628353e-05, "loss": 0.0052, "step": 204810 }, { "epoch": 130.5417463352454, "grad_norm": 0.32308846712112427, "learning_rate": 2.5086443829911754e-05, "loss": 0.0059, "step": 204820 }, { "epoch": 130.54811982154237, "grad_norm": 0.20988717675209045, "learning_rate": 2.508166533626321e-05, "loss": 0.0149, "step": 204830 }, { "epoch": 130.55449330783938, "grad_norm": 0.22845757007598877, "learning_rate": 2.507688714539596e-05, "loss": 0.0063, "step": 204840 }, { "epoch": 130.5608667941364, "grad_norm": 0.23069381713867188, "learning_rate": 2.5072109257368042e-05, "loss": 0.0066, "step": 204850 }, { "epoch": 130.5672402804334, "grad_norm": 0.08751989156007767, "learning_rate": 2.5067331672237552e-05, "loss": 0.0061, "step": 204860 }, { "epoch": 130.5736137667304, "grad_norm": 0.10620729625225067, "learning_rate": 2.5062554390062525e-05, "loss": 0.0062, "step": 204870 }, { "epoch": 130.5799872530274, "grad_norm": 0.31744667887687683, "learning_rate": 2.5057777410901005e-05, "loss": 0.0065, "step": 204880 }, { "epoch": 130.58636073932442, "grad_norm": 0.4427950978279114, "learning_rate": 2.5053000734811018e-05, "loss": 0.0116, "step": 204890 }, { "epoch": 130.59273422562143, "grad_norm": 0.13335059583187103, "learning_rate": 2.5048224361850646e-05, "loss": 0.0112, "step": 204900 }, { "epoch": 130.5991077119184, "grad_norm": 0.15718965232372284, "learning_rate": 2.504344829207791e-05, "loss": 0.0126, "step": 204910 }, { "epoch": 130.60548119821541, "grad_norm": 0.1721978783607483, "learning_rate": 2.503867252555084e-05, "loss": 0.003, "step": 204920 }, { "epoch": 130.61185468451242, "grad_norm": 0.38555750250816345, "learning_rate": 2.5033897062327438e-05, "loss": 0.0119, "step": 204930 }, { "epoch": 130.61822817080943, "grad_norm": 0.3778753876686096, "learning_rate": 2.502912190246579e-05, "loss": 0.0079, "step": 204940 }, { "epoch": 130.62460165710644, "grad_norm": 0.11952663213014603, "learning_rate": 2.5024347046023876e-05, "loss": 0.0111, "step": 204950 }, { "epoch": 130.63097514340345, "grad_norm": 0.2837376892566681, "learning_rate": 2.5019572493059735e-05, "loss": 0.0096, "step": 204960 }, { "epoch": 130.63734862970045, "grad_norm": 0.1795782595872879, "learning_rate": 2.5014798243631356e-05, "loss": 0.0088, "step": 204970 }, { "epoch": 130.64372211599746, "grad_norm": 0.22531481087207794, "learning_rate": 2.5010024297796776e-05, "loss": 0.0083, "step": 204980 }, { "epoch": 130.65009560229444, "grad_norm": 0.32225263118743896, "learning_rate": 2.5005250655614e-05, "loss": 0.0099, "step": 204990 }, { "epoch": 130.65646908859145, "grad_norm": 0.323777437210083, "learning_rate": 2.5000477317141034e-05, "loss": 0.0044, "step": 205000 }, { "epoch": 130.66284257488846, "grad_norm": 0.20042437314987183, "learning_rate": 2.4995704282435845e-05, "loss": 0.0087, "step": 205010 }, { "epoch": 130.66921606118547, "grad_norm": 0.12271622568368912, "learning_rate": 2.49909315515565e-05, "loss": 0.0118, "step": 205020 }, { "epoch": 130.67558954748247, "grad_norm": 0.10890910029411316, "learning_rate": 2.498615912456091e-05, "loss": 0.0106, "step": 205030 }, { "epoch": 130.68196303377948, "grad_norm": 0.4239848256111145, "learning_rate": 2.498138700150712e-05, "loss": 0.0095, "step": 205040 }, { "epoch": 130.6883365200765, "grad_norm": 0.2537029981613159, "learning_rate": 2.4976615182453084e-05, "loss": 0.0056, "step": 205050 }, { "epoch": 130.6947100063735, "grad_norm": 0.1294068545103073, "learning_rate": 2.4971843667456834e-05, "loss": 0.0065, "step": 205060 }, { "epoch": 130.70108349267048, "grad_norm": 0.25873249769210815, "learning_rate": 2.4967072456576278e-05, "loss": 0.0083, "step": 205070 }, { "epoch": 130.70745697896749, "grad_norm": 0.19523939490318298, "learning_rate": 2.4962301549869448e-05, "loss": 0.0052, "step": 205080 }, { "epoch": 130.7138304652645, "grad_norm": 0.14750796556472778, "learning_rate": 2.495753094739429e-05, "loss": 0.0167, "step": 205090 }, { "epoch": 130.7202039515615, "grad_norm": 0.11027756333351135, "learning_rate": 2.495276064920878e-05, "loss": 0.0045, "step": 205100 }, { "epoch": 130.7265774378585, "grad_norm": 0.23904204368591309, "learning_rate": 2.494799065537085e-05, "loss": 0.0059, "step": 205110 }, { "epoch": 130.73295092415552, "grad_norm": 0.17435112595558167, "learning_rate": 2.4943220965938507e-05, "loss": 0.0116, "step": 205120 }, { "epoch": 130.73932441045253, "grad_norm": 0.29373982548713684, "learning_rate": 2.493845158096969e-05, "loss": 0.0066, "step": 205130 }, { "epoch": 130.74569789674953, "grad_norm": 0.23561398684978485, "learning_rate": 2.493368250052235e-05, "loss": 0.0073, "step": 205140 }, { "epoch": 130.7520713830465, "grad_norm": 0.1624774932861328, "learning_rate": 2.4928913724654413e-05, "loss": 0.0064, "step": 205150 }, { "epoch": 130.75844486934352, "grad_norm": 0.11124633997678757, "learning_rate": 2.492414525342388e-05, "loss": 0.0077, "step": 205160 }, { "epoch": 130.76481835564053, "grad_norm": 0.21933723986148834, "learning_rate": 2.4919377086888624e-05, "loss": 0.0051, "step": 205170 }, { "epoch": 130.77119184193754, "grad_norm": 0.40369653701782227, "learning_rate": 2.4914609225106634e-05, "loss": 0.011, "step": 205180 }, { "epoch": 130.77756532823454, "grad_norm": 0.28697699308395386, "learning_rate": 2.490984166813581e-05, "loss": 0.0086, "step": 205190 }, { "epoch": 130.78393881453155, "grad_norm": 0.343045711517334, "learning_rate": 2.4905074416034136e-05, "loss": 0.0041, "step": 205200 }, { "epoch": 130.79031230082856, "grad_norm": 0.1814769059419632, "learning_rate": 2.4900307468859457e-05, "loss": 0.0064, "step": 205210 }, { "epoch": 130.79668578712557, "grad_norm": 0.23198506236076355, "learning_rate": 2.4895540826669765e-05, "loss": 0.0075, "step": 205220 }, { "epoch": 130.80305927342255, "grad_norm": 0.25850948691368103, "learning_rate": 2.4890774489522957e-05, "loss": 0.0075, "step": 205230 }, { "epoch": 130.80943275971956, "grad_norm": 0.1450684666633606, "learning_rate": 2.488600845747694e-05, "loss": 0.0065, "step": 205240 }, { "epoch": 130.81580624601656, "grad_norm": 0.41491881012916565, "learning_rate": 2.4881242730589617e-05, "loss": 0.0111, "step": 205250 }, { "epoch": 130.82217973231357, "grad_norm": 0.1883465051651001, "learning_rate": 2.4876477308918928e-05, "loss": 0.0064, "step": 205260 }, { "epoch": 130.82855321861058, "grad_norm": 0.20829954743385315, "learning_rate": 2.4871712192522756e-05, "loss": 0.0072, "step": 205270 }, { "epoch": 130.8349267049076, "grad_norm": 0.12111140787601471, "learning_rate": 2.486694738145901e-05, "loss": 0.0064, "step": 205280 }, { "epoch": 130.8413001912046, "grad_norm": 0.20877590775489807, "learning_rate": 2.4862182875785567e-05, "loss": 0.0049, "step": 205290 }, { "epoch": 130.8476736775016, "grad_norm": 0.43158772587776184, "learning_rate": 2.4857418675560352e-05, "loss": 0.0075, "step": 205300 }, { "epoch": 130.8540471637986, "grad_norm": 0.3143948018550873, "learning_rate": 2.485265478084124e-05, "loss": 0.0093, "step": 205310 }, { "epoch": 130.8604206500956, "grad_norm": 0.47183483839035034, "learning_rate": 2.4847891191686112e-05, "loss": 0.0151, "step": 205320 }, { "epoch": 130.8667941363926, "grad_norm": 0.3349553942680359, "learning_rate": 2.4843127908152835e-05, "loss": 0.0088, "step": 205330 }, { "epoch": 130.8731676226896, "grad_norm": 0.4241543412208557, "learning_rate": 2.4838364930299346e-05, "loss": 0.0071, "step": 205340 }, { "epoch": 130.87954110898661, "grad_norm": 0.21545414626598358, "learning_rate": 2.4833602258183443e-05, "loss": 0.0086, "step": 205350 }, { "epoch": 130.88591459528362, "grad_norm": 0.24071644246578217, "learning_rate": 2.482883989186305e-05, "loss": 0.0062, "step": 205360 }, { "epoch": 130.89228808158063, "grad_norm": 0.1948540210723877, "learning_rate": 2.4824077831395996e-05, "loss": 0.0063, "step": 205370 }, { "epoch": 130.89866156787764, "grad_norm": 0.4545469880104065, "learning_rate": 2.481931607684021e-05, "loss": 0.0091, "step": 205380 }, { "epoch": 130.90503505417465, "grad_norm": 0.3421158194541931, "learning_rate": 2.4814554628253473e-05, "loss": 0.0108, "step": 205390 }, { "epoch": 130.91140854047163, "grad_norm": 0.354618638753891, "learning_rate": 2.4809793485693688e-05, "loss": 0.0093, "step": 205400 }, { "epoch": 130.91778202676863, "grad_norm": 0.2194470316171646, "learning_rate": 2.4805032649218697e-05, "loss": 0.0102, "step": 205410 }, { "epoch": 130.92415551306564, "grad_norm": 0.3889772295951843, "learning_rate": 2.4800272118886347e-05, "loss": 0.0071, "step": 205420 }, { "epoch": 130.93052899936265, "grad_norm": 0.24751372635364532, "learning_rate": 2.479551189475447e-05, "loss": 0.0112, "step": 205430 }, { "epoch": 130.93690248565966, "grad_norm": 0.1915123164653778, "learning_rate": 2.4790751976880932e-05, "loss": 0.0085, "step": 205440 }, { "epoch": 130.94327597195667, "grad_norm": 0.16883181035518646, "learning_rate": 2.4785992365323563e-05, "loss": 0.0069, "step": 205450 }, { "epoch": 130.94964945825367, "grad_norm": 0.170342817902565, "learning_rate": 2.4781233060140192e-05, "loss": 0.0134, "step": 205460 }, { "epoch": 130.95602294455068, "grad_norm": 0.6638593673706055, "learning_rate": 2.4776474061388633e-05, "loss": 0.0081, "step": 205470 }, { "epoch": 130.96239643084766, "grad_norm": 0.5212392210960388, "learning_rate": 2.477171536912674e-05, "loss": 0.011, "step": 205480 }, { "epoch": 130.96876991714467, "grad_norm": 0.15780538320541382, "learning_rate": 2.476695698341233e-05, "loss": 0.0174, "step": 205490 }, { "epoch": 130.97514340344168, "grad_norm": 0.14549915492534637, "learning_rate": 2.476219890430322e-05, "loss": 0.0126, "step": 205500 }, { "epoch": 130.98151688973869, "grad_norm": 0.26643064618110657, "learning_rate": 2.4757441131857202e-05, "loss": 0.0071, "step": 205510 }, { "epoch": 130.9878903760357, "grad_norm": 0.2750958204269409, "learning_rate": 2.4752683666132122e-05, "loss": 0.0112, "step": 205520 }, { "epoch": 130.9942638623327, "grad_norm": 0.18083015084266663, "learning_rate": 2.4747926507185777e-05, "loss": 0.0088, "step": 205530 }, { "epoch": 131.0006373486297, "grad_norm": 0.17516420781612396, "learning_rate": 2.4743169655075965e-05, "loss": 0.0138, "step": 205540 }, { "epoch": 131.00701083492672, "grad_norm": 0.21830593049526215, "learning_rate": 2.473841310986047e-05, "loss": 0.0066, "step": 205550 }, { "epoch": 131.0133843212237, "grad_norm": 0.18747352063655853, "learning_rate": 2.4733656871597123e-05, "loss": 0.0061, "step": 205560 }, { "epoch": 131.0197578075207, "grad_norm": 0.5394179224967957, "learning_rate": 2.4728900940343702e-05, "loss": 0.0114, "step": 205570 }, { "epoch": 131.0261312938177, "grad_norm": 0.296905517578125, "learning_rate": 2.4724145316157994e-05, "loss": 0.0109, "step": 205580 }, { "epoch": 131.03250478011472, "grad_norm": 0.10902935266494751, "learning_rate": 2.4719389999097786e-05, "loss": 0.0067, "step": 205590 }, { "epoch": 131.03887826641173, "grad_norm": 0.29082468152046204, "learning_rate": 2.4714634989220836e-05, "loss": 0.0099, "step": 205600 }, { "epoch": 131.04525175270874, "grad_norm": 0.6398632526397705, "learning_rate": 2.4709880286584962e-05, "loss": 0.0169, "step": 205610 }, { "epoch": 131.05162523900574, "grad_norm": 0.4924894869327545, "learning_rate": 2.4705125891247928e-05, "loss": 0.0072, "step": 205620 }, { "epoch": 131.05799872530275, "grad_norm": 0.5675828456878662, "learning_rate": 2.4700371803267492e-05, "loss": 0.0123, "step": 205630 }, { "epoch": 131.06437221159973, "grad_norm": 0.5897438526153564, "learning_rate": 2.4695618022701405e-05, "loss": 0.0065, "step": 205640 }, { "epoch": 131.07074569789674, "grad_norm": 0.24331237375736237, "learning_rate": 2.4690864549607474e-05, "loss": 0.0033, "step": 205650 }, { "epoch": 131.07711918419375, "grad_norm": 0.5772973895072937, "learning_rate": 2.468611138404343e-05, "loss": 0.0087, "step": 205660 }, { "epoch": 131.08349267049076, "grad_norm": 0.32592612504959106, "learning_rate": 2.468135852606704e-05, "loss": 0.009, "step": 205670 }, { "epoch": 131.08986615678776, "grad_norm": 0.10574318468570709, "learning_rate": 2.4676605975736023e-05, "loss": 0.014, "step": 205680 }, { "epoch": 131.09623964308477, "grad_norm": 0.21793371438980103, "learning_rate": 2.467185373310817e-05, "loss": 0.0054, "step": 205690 }, { "epoch": 131.10261312938178, "grad_norm": 0.4275188148021698, "learning_rate": 2.466710179824121e-05, "loss": 0.0109, "step": 205700 }, { "epoch": 131.1089866156788, "grad_norm": 0.2141624540090561, "learning_rate": 2.4662350171192883e-05, "loss": 0.0088, "step": 205710 }, { "epoch": 131.11536010197577, "grad_norm": 0.4281606674194336, "learning_rate": 2.4657598852020908e-05, "loss": 0.0069, "step": 205720 }, { "epoch": 131.12173358827278, "grad_norm": 0.4291565418243408, "learning_rate": 2.4652847840783068e-05, "loss": 0.019, "step": 205730 }, { "epoch": 131.12810707456978, "grad_norm": 0.11615169048309326, "learning_rate": 2.464809713753702e-05, "loss": 0.0096, "step": 205740 }, { "epoch": 131.1344805608668, "grad_norm": 0.19146528840065002, "learning_rate": 2.464334674234055e-05, "loss": 0.0055, "step": 205750 }, { "epoch": 131.1408540471638, "grad_norm": 0.15006107091903687, "learning_rate": 2.4638596655251333e-05, "loss": 0.0061, "step": 205760 }, { "epoch": 131.1472275334608, "grad_norm": 0.5572911500930786, "learning_rate": 2.4633846876327155e-05, "loss": 0.0139, "step": 205770 }, { "epoch": 131.15360101975782, "grad_norm": 0.19073864817619324, "learning_rate": 2.4629097405625647e-05, "loss": 0.0073, "step": 205780 }, { "epoch": 131.15997450605482, "grad_norm": 0.43402501940727234, "learning_rate": 2.462434824320458e-05, "loss": 0.0148, "step": 205790 }, { "epoch": 131.1663479923518, "grad_norm": 0.20692744851112366, "learning_rate": 2.4619599389121638e-05, "loss": 0.0082, "step": 205800 }, { "epoch": 131.1727214786488, "grad_norm": 0.18872588872909546, "learning_rate": 2.4614850843434528e-05, "loss": 0.0051, "step": 205810 }, { "epoch": 131.17909496494582, "grad_norm": 0.2159077525138855, "learning_rate": 2.461010260620093e-05, "loss": 0.0058, "step": 205820 }, { "epoch": 131.18546845124283, "grad_norm": 0.18911594152450562, "learning_rate": 2.4605354677478575e-05, "loss": 0.0065, "step": 205830 }, { "epoch": 131.19184193753983, "grad_norm": 0.23971712589263916, "learning_rate": 2.4600607057325135e-05, "loss": 0.0096, "step": 205840 }, { "epoch": 131.19821542383684, "grad_norm": 0.19108088314533234, "learning_rate": 2.4595859745798304e-05, "loss": 0.0079, "step": 205850 }, { "epoch": 131.20458891013385, "grad_norm": 0.08424948155879974, "learning_rate": 2.4591112742955748e-05, "loss": 0.0051, "step": 205860 }, { "epoch": 131.21096239643086, "grad_norm": 0.3672334849834442, "learning_rate": 2.458636604885517e-05, "loss": 0.0064, "step": 205870 }, { "epoch": 131.21733588272787, "grad_norm": 0.1736650913953781, "learning_rate": 2.4581619663554244e-05, "loss": 0.0112, "step": 205880 }, { "epoch": 131.22370936902485, "grad_norm": 0.3354727625846863, "learning_rate": 2.4576873587110643e-05, "loss": 0.0081, "step": 205890 }, { "epoch": 131.23008285532185, "grad_norm": 0.28604698181152344, "learning_rate": 2.4572127819582007e-05, "loss": 0.0079, "step": 205900 }, { "epoch": 131.23645634161886, "grad_norm": 0.17971903085708618, "learning_rate": 2.4567382361026064e-05, "loss": 0.0098, "step": 205910 }, { "epoch": 131.24282982791587, "grad_norm": 0.580685555934906, "learning_rate": 2.45626372115004e-05, "loss": 0.0108, "step": 205920 }, { "epoch": 131.24920331421288, "grad_norm": 0.23209227621555328, "learning_rate": 2.455789237106273e-05, "loss": 0.0093, "step": 205930 }, { "epoch": 131.25557680050989, "grad_norm": 0.18523401021957397, "learning_rate": 2.455314783977067e-05, "loss": 0.0054, "step": 205940 }, { "epoch": 131.2619502868069, "grad_norm": 0.16882628202438354, "learning_rate": 2.4548403617681926e-05, "loss": 0.0056, "step": 205950 }, { "epoch": 131.2683237731039, "grad_norm": 0.330886572599411, "learning_rate": 2.454365970485407e-05, "loss": 0.0145, "step": 205960 }, { "epoch": 131.27469725940088, "grad_norm": 0.30368509888648987, "learning_rate": 2.4538916101344806e-05, "loss": 0.0117, "step": 205970 }, { "epoch": 131.2810707456979, "grad_norm": 0.21836785972118378, "learning_rate": 2.4534172807211753e-05, "loss": 0.0049, "step": 205980 }, { "epoch": 131.2874442319949, "grad_norm": 0.531607985496521, "learning_rate": 2.4529429822512538e-05, "loss": 0.0132, "step": 205990 }, { "epoch": 131.2938177182919, "grad_norm": 0.28655529022216797, "learning_rate": 2.4524687147304783e-05, "loss": 0.0075, "step": 206000 }, { "epoch": 131.3001912045889, "grad_norm": 0.227704256772995, "learning_rate": 2.4519944781646147e-05, "loss": 0.013, "step": 206010 }, { "epoch": 131.30656469088592, "grad_norm": 0.5265530943870544, "learning_rate": 2.4515202725594244e-05, "loss": 0.0057, "step": 206020 }, { "epoch": 131.31293817718293, "grad_norm": 0.4515119791030884, "learning_rate": 2.4510460979206685e-05, "loss": 0.0044, "step": 206030 }, { "epoch": 131.31931166347994, "grad_norm": 0.46073687076568604, "learning_rate": 2.450571954254108e-05, "loss": 0.0086, "step": 206040 }, { "epoch": 131.32568514977692, "grad_norm": 0.15698358416557312, "learning_rate": 2.450097841565508e-05, "loss": 0.0073, "step": 206050 }, { "epoch": 131.33205863607392, "grad_norm": 0.29000625014305115, "learning_rate": 2.4496237598606236e-05, "loss": 0.008, "step": 206060 }, { "epoch": 131.33843212237093, "grad_norm": 0.26535430550575256, "learning_rate": 2.4491497091452202e-05, "loss": 0.0065, "step": 206070 }, { "epoch": 131.34480560866794, "grad_norm": 0.15363667905330658, "learning_rate": 2.4486756894250545e-05, "loss": 0.0076, "step": 206080 }, { "epoch": 131.35117909496495, "grad_norm": 0.08641275018453598, "learning_rate": 2.4482017007058915e-05, "loss": 0.0077, "step": 206090 }, { "epoch": 131.35755258126196, "grad_norm": 0.5220246315002441, "learning_rate": 2.4477277429934836e-05, "loss": 0.0071, "step": 206100 }, { "epoch": 131.36392606755896, "grad_norm": 0.2640078663825989, "learning_rate": 2.4472538162935953e-05, "loss": 0.0143, "step": 206110 }, { "epoch": 131.37029955385597, "grad_norm": 0.1030304953455925, "learning_rate": 2.4467799206119834e-05, "loss": 0.0112, "step": 206120 }, { "epoch": 131.37667304015295, "grad_norm": 0.1499813050031662, "learning_rate": 2.446306055954406e-05, "loss": 0.0081, "step": 206130 }, { "epoch": 131.38304652644996, "grad_norm": 0.3317027986049652, "learning_rate": 2.4458322223266195e-05, "loss": 0.0066, "step": 206140 }, { "epoch": 131.38942001274697, "grad_norm": 0.3474537134170532, "learning_rate": 2.4453584197343842e-05, "loss": 0.0115, "step": 206150 }, { "epoch": 131.39579349904398, "grad_norm": 0.5639309883117676, "learning_rate": 2.4448846481834565e-05, "loss": 0.0108, "step": 206160 }, { "epoch": 131.40216698534098, "grad_norm": 0.2797049582004547, "learning_rate": 2.444410907679593e-05, "loss": 0.0062, "step": 206170 }, { "epoch": 131.408540471638, "grad_norm": 0.29822278022766113, "learning_rate": 2.4439371982285473e-05, "loss": 0.0045, "step": 206180 }, { "epoch": 131.414913957935, "grad_norm": 0.47615769505500793, "learning_rate": 2.4434635198360805e-05, "loss": 0.0126, "step": 206190 }, { "epoch": 131.421287444232, "grad_norm": 0.22557586431503296, "learning_rate": 2.4429898725079453e-05, "loss": 0.0055, "step": 206200 }, { "epoch": 131.427660930529, "grad_norm": 0.5891027450561523, "learning_rate": 2.4425162562498975e-05, "loss": 0.0198, "step": 206210 }, { "epoch": 131.434034416826, "grad_norm": 0.326540470123291, "learning_rate": 2.44204267106769e-05, "loss": 0.0073, "step": 206220 }, { "epoch": 131.440407903123, "grad_norm": 0.24969254434108734, "learning_rate": 2.4415691169670812e-05, "loss": 0.0075, "step": 206230 }, { "epoch": 131.44678138942, "grad_norm": 0.3137982189655304, "learning_rate": 2.4410955939538226e-05, "loss": 0.0088, "step": 206240 }, { "epoch": 131.45315487571702, "grad_norm": 0.5127069354057312, "learning_rate": 2.4406221020336696e-05, "loss": 0.0085, "step": 206250 }, { "epoch": 131.45952836201403, "grad_norm": 0.2369498908519745, "learning_rate": 2.4401486412123713e-05, "loss": 0.0064, "step": 206260 }, { "epoch": 131.46590184831103, "grad_norm": 0.41332313418388367, "learning_rate": 2.4396752114956867e-05, "loss": 0.0114, "step": 206270 }, { "epoch": 131.47227533460804, "grad_norm": 0.40945568680763245, "learning_rate": 2.4392018128893656e-05, "loss": 0.0079, "step": 206280 }, { "epoch": 131.47864882090502, "grad_norm": 0.40940797328948975, "learning_rate": 2.43872844539916e-05, "loss": 0.0088, "step": 206290 }, { "epoch": 131.48502230720203, "grad_norm": 0.27717167139053345, "learning_rate": 2.438255109030822e-05, "loss": 0.0059, "step": 206300 }, { "epoch": 131.49139579349904, "grad_norm": 0.161263570189476, "learning_rate": 2.4377818037901013e-05, "loss": 0.0075, "step": 206310 }, { "epoch": 131.49776927979605, "grad_norm": 0.5426060557365417, "learning_rate": 2.437308529682753e-05, "loss": 0.0094, "step": 206320 }, { "epoch": 131.50414276609305, "grad_norm": 0.18327857553958893, "learning_rate": 2.436835286714525e-05, "loss": 0.015, "step": 206330 }, { "epoch": 131.51051625239006, "grad_norm": 0.5656985640525818, "learning_rate": 2.436362074891169e-05, "loss": 0.0116, "step": 206340 }, { "epoch": 131.51688973868707, "grad_norm": 0.4857151508331299, "learning_rate": 2.435888894218432e-05, "loss": 0.0232, "step": 206350 }, { "epoch": 131.52326322498408, "grad_norm": 0.5164250135421753, "learning_rate": 2.4354157447020676e-05, "loss": 0.0139, "step": 206360 }, { "epoch": 131.52963671128106, "grad_norm": 0.35452786087989807, "learning_rate": 2.4349426263478237e-05, "loss": 0.0096, "step": 206370 }, { "epoch": 131.53601019757807, "grad_norm": 0.14995358884334564, "learning_rate": 2.4344695391614486e-05, "loss": 0.0046, "step": 206380 }, { "epoch": 131.54238368387507, "grad_norm": 0.23027701675891876, "learning_rate": 2.433996483148689e-05, "loss": 0.0154, "step": 206390 }, { "epoch": 131.54875717017208, "grad_norm": 0.09216442704200745, "learning_rate": 2.433523458315296e-05, "loss": 0.0096, "step": 206400 }, { "epoch": 131.5551306564691, "grad_norm": 0.2610524296760559, "learning_rate": 2.433050464667017e-05, "loss": 0.0072, "step": 206410 }, { "epoch": 131.5615041427661, "grad_norm": 0.4036546051502228, "learning_rate": 2.432577502209598e-05, "loss": 0.0094, "step": 206420 }, { "epoch": 131.5678776290631, "grad_norm": 0.24170011281967163, "learning_rate": 2.432104570948785e-05, "loss": 0.0089, "step": 206430 }, { "epoch": 131.5742511153601, "grad_norm": 0.33132192492485046, "learning_rate": 2.4316316708903293e-05, "loss": 0.0076, "step": 206440 }, { "epoch": 131.5806246016571, "grad_norm": 0.2783168852329254, "learning_rate": 2.43115880203997e-05, "loss": 0.0079, "step": 206450 }, { "epoch": 131.5869980879541, "grad_norm": 0.42520779371261597, "learning_rate": 2.4306859644034587e-05, "loss": 0.0083, "step": 206460 }, { "epoch": 131.5933715742511, "grad_norm": 0.6975628733634949, "learning_rate": 2.4302131579865366e-05, "loss": 0.0167, "step": 206470 }, { "epoch": 131.59974506054812, "grad_norm": 0.6585759520530701, "learning_rate": 2.429740382794955e-05, "loss": 0.013, "step": 206480 }, { "epoch": 131.60611854684512, "grad_norm": 0.20765124261379242, "learning_rate": 2.4292676388344504e-05, "loss": 0.0079, "step": 206490 }, { "epoch": 131.61249203314213, "grad_norm": 0.2686104476451874, "learning_rate": 2.4287949261107734e-05, "loss": 0.0039, "step": 206500 }, { "epoch": 131.61886551943914, "grad_norm": 0.3497500419616699, "learning_rate": 2.4283222446296634e-05, "loss": 0.0069, "step": 206510 }, { "epoch": 131.62523900573615, "grad_norm": 0.20553264021873474, "learning_rate": 2.4278495943968703e-05, "loss": 0.0083, "step": 206520 }, { "epoch": 131.63161249203316, "grad_norm": 0.27128714323043823, "learning_rate": 2.427376975418129e-05, "loss": 0.0046, "step": 206530 }, { "epoch": 131.63798597833014, "grad_norm": 0.1291462779045105, "learning_rate": 2.4269043876991888e-05, "loss": 0.0054, "step": 206540 }, { "epoch": 131.64435946462714, "grad_norm": 0.44430238008499146, "learning_rate": 2.4264318312457894e-05, "loss": 0.0145, "step": 206550 }, { "epoch": 131.65073295092415, "grad_norm": 0.23414826393127441, "learning_rate": 2.4259593060636732e-05, "loss": 0.0063, "step": 206560 }, { "epoch": 131.65710643722116, "grad_norm": 0.37287378311157227, "learning_rate": 2.4254868121585796e-05, "loss": 0.0078, "step": 206570 }, { "epoch": 131.66347992351817, "grad_norm": 0.47343963384628296, "learning_rate": 2.425014349536254e-05, "loss": 0.0107, "step": 206580 }, { "epoch": 131.66985340981518, "grad_norm": 0.341098815202713, "learning_rate": 2.4245419182024353e-05, "loss": 0.0128, "step": 206590 }, { "epoch": 131.67622689611218, "grad_norm": 0.1867770254611969, "learning_rate": 2.424069518162864e-05, "loss": 0.0065, "step": 206600 }, { "epoch": 131.6826003824092, "grad_norm": 0.22877442836761475, "learning_rate": 2.4235971494232778e-05, "loss": 0.0136, "step": 206610 }, { "epoch": 131.68897386870617, "grad_norm": 0.39753586053848267, "learning_rate": 2.4231248119894235e-05, "loss": 0.0075, "step": 206620 }, { "epoch": 131.69534735500318, "grad_norm": 0.18112720549106598, "learning_rate": 2.4226525058670314e-05, "loss": 0.0137, "step": 206630 }, { "epoch": 131.7017208413002, "grad_norm": 0.33424824476242065, "learning_rate": 2.4221802310618462e-05, "loss": 0.0085, "step": 206640 }, { "epoch": 131.7080943275972, "grad_norm": 0.09351005405187607, "learning_rate": 2.4217079875796033e-05, "loss": 0.0078, "step": 206650 }, { "epoch": 131.7144678138942, "grad_norm": 0.15434160828590393, "learning_rate": 2.421235775426047e-05, "loss": 0.0087, "step": 206660 }, { "epoch": 131.7208413001912, "grad_norm": 0.15922291576862335, "learning_rate": 2.4207635946069067e-05, "loss": 0.0046, "step": 206670 }, { "epoch": 131.72721478648822, "grad_norm": 0.20381693542003632, "learning_rate": 2.4202914451279256e-05, "loss": 0.0077, "step": 206680 }, { "epoch": 131.73358827278523, "grad_norm": 0.20925241708755493, "learning_rate": 2.4198193269948393e-05, "loss": 0.0059, "step": 206690 }, { "epoch": 131.7399617590822, "grad_norm": 0.346466064453125, "learning_rate": 2.4193472402133843e-05, "loss": 0.014, "step": 206700 }, { "epoch": 131.7463352453792, "grad_norm": 0.09483383595943451, "learning_rate": 2.4188751847892947e-05, "loss": 0.0082, "step": 206710 }, { "epoch": 131.75270873167622, "grad_norm": 0.21764998137950897, "learning_rate": 2.4184031607283113e-05, "loss": 0.0112, "step": 206720 }, { "epoch": 131.75908221797323, "grad_norm": 0.23262406885623932, "learning_rate": 2.4179311680361662e-05, "loss": 0.01, "step": 206730 }, { "epoch": 131.76545570427024, "grad_norm": 0.4581264555454254, "learning_rate": 2.4174592067185954e-05, "loss": 0.0092, "step": 206740 }, { "epoch": 131.77182919056725, "grad_norm": 0.36170992255210876, "learning_rate": 2.4169872767813317e-05, "loss": 0.0112, "step": 206750 }, { "epoch": 131.77820267686425, "grad_norm": 0.16303211450576782, "learning_rate": 2.4165153782301136e-05, "loss": 0.0098, "step": 206760 }, { "epoch": 131.78457616316126, "grad_norm": 0.2098337858915329, "learning_rate": 2.4160435110706724e-05, "loss": 0.0096, "step": 206770 }, { "epoch": 131.79094964945824, "grad_norm": 0.2524847388267517, "learning_rate": 2.4155716753087422e-05, "loss": 0.0062, "step": 206780 }, { "epoch": 131.79732313575525, "grad_norm": 0.2128397524356842, "learning_rate": 2.4150998709500544e-05, "loss": 0.0094, "step": 206790 }, { "epoch": 131.80369662205226, "grad_norm": 0.45467597246170044, "learning_rate": 2.4146280980003477e-05, "loss": 0.0149, "step": 206800 }, { "epoch": 131.81007010834927, "grad_norm": 0.26863136887550354, "learning_rate": 2.414156356465347e-05, "loss": 0.0059, "step": 206810 }, { "epoch": 131.81644359464627, "grad_norm": 0.106324203312397, "learning_rate": 2.4136846463507896e-05, "loss": 0.0076, "step": 206820 }, { "epoch": 131.82281708094328, "grad_norm": 0.3427966833114624, "learning_rate": 2.4132129676624038e-05, "loss": 0.0157, "step": 206830 }, { "epoch": 131.8291905672403, "grad_norm": 0.16947492957115173, "learning_rate": 2.4127413204059264e-05, "loss": 0.0051, "step": 206840 }, { "epoch": 131.8355640535373, "grad_norm": 0.1577671319246292, "learning_rate": 2.4122697045870805e-05, "loss": 0.0094, "step": 206850 }, { "epoch": 131.84193753983428, "grad_norm": 0.23673786222934723, "learning_rate": 2.4117981202116037e-05, "loss": 0.0119, "step": 206860 }, { "epoch": 131.84831102613128, "grad_norm": 0.12579423189163208, "learning_rate": 2.4113265672852232e-05, "loss": 0.0048, "step": 206870 }, { "epoch": 131.8546845124283, "grad_norm": 0.10091779381036758, "learning_rate": 2.410855045813669e-05, "loss": 0.0073, "step": 206880 }, { "epoch": 131.8610579987253, "grad_norm": 0.16133522987365723, "learning_rate": 2.410383555802668e-05, "loss": 0.0061, "step": 206890 }, { "epoch": 131.8674314850223, "grad_norm": 0.1579633355140686, "learning_rate": 2.4099120972579546e-05, "loss": 0.0142, "step": 206900 }, { "epoch": 131.87380497131932, "grad_norm": 0.3589632511138916, "learning_rate": 2.4094406701852544e-05, "loss": 0.0089, "step": 206910 }, { "epoch": 131.88017845761632, "grad_norm": 0.1615668386220932, "learning_rate": 2.408969274590296e-05, "loss": 0.0078, "step": 206920 }, { "epoch": 131.88655194391333, "grad_norm": 0.2257506102323532, "learning_rate": 2.4084979104788052e-05, "loss": 0.0073, "step": 206930 }, { "epoch": 131.8929254302103, "grad_norm": 0.1424528956413269, "learning_rate": 2.4080265778565138e-05, "loss": 0.0061, "step": 206940 }, { "epoch": 131.89929891650732, "grad_norm": 0.39749813079833984, "learning_rate": 2.407555276729146e-05, "loss": 0.0104, "step": 206950 }, { "epoch": 131.90567240280433, "grad_norm": 0.17432767152786255, "learning_rate": 2.4070840071024298e-05, "loss": 0.0049, "step": 206960 }, { "epoch": 131.91204588910134, "grad_norm": 0.11865808069705963, "learning_rate": 2.4066127689820888e-05, "loss": 0.0066, "step": 206970 }, { "epoch": 131.91841937539834, "grad_norm": 0.19276976585388184, "learning_rate": 2.4061415623738532e-05, "loss": 0.014, "step": 206980 }, { "epoch": 131.92479286169535, "grad_norm": 0.36953410506248474, "learning_rate": 2.405670387283447e-05, "loss": 0.0067, "step": 206990 }, { "epoch": 131.93116634799236, "grad_norm": 0.19844792783260345, "learning_rate": 2.405199243716595e-05, "loss": 0.0106, "step": 207000 }, { "epoch": 131.93753983428937, "grad_norm": 0.20311938226222992, "learning_rate": 2.4047281316790215e-05, "loss": 0.0104, "step": 207010 }, { "epoch": 131.94391332058635, "grad_norm": 0.14476488530635834, "learning_rate": 2.40425705117645e-05, "loss": 0.0083, "step": 207020 }, { "epoch": 131.95028680688335, "grad_norm": 0.18064384162425995, "learning_rate": 2.403786002214608e-05, "loss": 0.0077, "step": 207030 }, { "epoch": 131.95666029318036, "grad_norm": 0.5062242150306702, "learning_rate": 2.4033149847992174e-05, "loss": 0.0066, "step": 207040 }, { "epoch": 131.96303377947737, "grad_norm": 0.2981482446193695, "learning_rate": 2.4028439989360014e-05, "loss": 0.0052, "step": 207050 }, { "epoch": 131.96940726577438, "grad_norm": 0.957068920135498, "learning_rate": 2.4023730446306807e-05, "loss": 0.0157, "step": 207060 }, { "epoch": 131.9757807520714, "grad_norm": 0.19591619074344635, "learning_rate": 2.4019021218889826e-05, "loss": 0.006, "step": 207070 }, { "epoch": 131.9821542383684, "grad_norm": 0.11951790750026703, "learning_rate": 2.4014312307166264e-05, "loss": 0.006, "step": 207080 }, { "epoch": 131.9885277246654, "grad_norm": 0.17299455404281616, "learning_rate": 2.400960371119334e-05, "loss": 0.0158, "step": 207090 }, { "epoch": 131.99490121096238, "grad_norm": 0.2574929893016815, "learning_rate": 2.4004895431028257e-05, "loss": 0.0096, "step": 207100 }, { "epoch": 132.0012746972594, "grad_norm": 0.21659253537654877, "learning_rate": 2.400018746672825e-05, "loss": 0.0069, "step": 207110 }, { "epoch": 132.0076481835564, "grad_norm": 0.11378421634435654, "learning_rate": 2.3995479818350514e-05, "loss": 0.0081, "step": 207120 }, { "epoch": 132.0140216698534, "grad_norm": 0.13419456779956818, "learning_rate": 2.3990772485952256e-05, "loss": 0.0048, "step": 207130 }, { "epoch": 132.0203951561504, "grad_norm": 0.17366361618041992, "learning_rate": 2.3986065469590646e-05, "loss": 0.0193, "step": 207140 }, { "epoch": 132.02676864244742, "grad_norm": 0.20944753289222717, "learning_rate": 2.398135876932292e-05, "loss": 0.0044, "step": 207150 }, { "epoch": 132.03314212874443, "grad_norm": 0.12066478282213211, "learning_rate": 2.3976652385206257e-05, "loss": 0.0047, "step": 207160 }, { "epoch": 132.03951561504144, "grad_norm": 0.35059407353401184, "learning_rate": 2.3971946317297826e-05, "loss": 0.0126, "step": 207170 }, { "epoch": 132.04588910133845, "grad_norm": 0.29172593355178833, "learning_rate": 2.3967240565654815e-05, "loss": 0.0055, "step": 207180 }, { "epoch": 132.05226258763543, "grad_norm": 0.1553424596786499, "learning_rate": 2.396253513033444e-05, "loss": 0.0108, "step": 207190 }, { "epoch": 132.05863607393243, "grad_norm": 0.20036271214485168, "learning_rate": 2.395783001139381e-05, "loss": 0.0084, "step": 207200 }, { "epoch": 132.06500956022944, "grad_norm": 0.33523717522621155, "learning_rate": 2.3953125208890154e-05, "loss": 0.0046, "step": 207210 }, { "epoch": 132.07138304652645, "grad_norm": 0.07166878134012222, "learning_rate": 2.3948420722880598e-05, "loss": 0.0075, "step": 207220 }, { "epoch": 132.07775653282346, "grad_norm": 0.18013344705104828, "learning_rate": 2.394371655342237e-05, "loss": 0.0141, "step": 207230 }, { "epoch": 132.08413001912047, "grad_norm": 0.15018822252750397, "learning_rate": 2.3939012700572543e-05, "loss": 0.0088, "step": 207240 }, { "epoch": 132.09050350541747, "grad_norm": 0.26525023579597473, "learning_rate": 2.393430916438834e-05, "loss": 0.0091, "step": 207250 }, { "epoch": 132.09687699171448, "grad_norm": 0.1286991685628891, "learning_rate": 2.3929605944926885e-05, "loss": 0.0037, "step": 207260 }, { "epoch": 132.10325047801146, "grad_norm": 0.34881559014320374, "learning_rate": 2.3924903042245338e-05, "loss": 0.0114, "step": 207270 }, { "epoch": 132.10962396430847, "grad_norm": 0.2609415352344513, "learning_rate": 2.3920200456400822e-05, "loss": 0.0049, "step": 207280 }, { "epoch": 132.11599745060548, "grad_norm": 0.146498903632164, "learning_rate": 2.391549818745051e-05, "loss": 0.0112, "step": 207290 }, { "epoch": 132.12237093690248, "grad_norm": 0.3830110430717468, "learning_rate": 2.3910796235451514e-05, "loss": 0.0065, "step": 207300 }, { "epoch": 132.1287444231995, "grad_norm": 0.27069348096847534, "learning_rate": 2.3906094600460987e-05, "loss": 0.0099, "step": 207310 }, { "epoch": 132.1351179094965, "grad_norm": 0.18351000547409058, "learning_rate": 2.3901393282536027e-05, "loss": 0.0101, "step": 207320 }, { "epoch": 132.1414913957935, "grad_norm": 0.19313348829746246, "learning_rate": 2.3896692281733813e-05, "loss": 0.0104, "step": 207330 }, { "epoch": 132.14786488209052, "grad_norm": 0.09580203145742416, "learning_rate": 2.38919915981114e-05, "loss": 0.0064, "step": 207340 }, { "epoch": 132.1542383683875, "grad_norm": 0.17059122025966644, "learning_rate": 2.3887291231725957e-05, "loss": 0.0044, "step": 207350 }, { "epoch": 132.1606118546845, "grad_norm": 0.0986395999789238, "learning_rate": 2.388259118263456e-05, "loss": 0.006, "step": 207360 }, { "epoch": 132.1669853409815, "grad_norm": 0.341968297958374, "learning_rate": 2.3877891450894373e-05, "loss": 0.0059, "step": 207370 }, { "epoch": 132.17335882727852, "grad_norm": 0.3859848976135254, "learning_rate": 2.3873192036562437e-05, "loss": 0.0141, "step": 207380 }, { "epoch": 132.17973231357553, "grad_norm": 0.16214615106582642, "learning_rate": 2.3868492939695904e-05, "loss": 0.0057, "step": 207390 }, { "epoch": 132.18610579987254, "grad_norm": 0.189400777220726, "learning_rate": 2.386379416035183e-05, "loss": 0.0049, "step": 207400 }, { "epoch": 132.19247928616954, "grad_norm": 0.32515838742256165, "learning_rate": 2.3859095698587374e-05, "loss": 0.0141, "step": 207410 }, { "epoch": 132.19885277246655, "grad_norm": 0.1324828416109085, "learning_rate": 2.3854397554459553e-05, "loss": 0.0096, "step": 207420 }, { "epoch": 132.20522625876353, "grad_norm": 0.14252065122127533, "learning_rate": 2.3849699728025503e-05, "loss": 0.0047, "step": 207430 }, { "epoch": 132.21159974506054, "grad_norm": 0.3125499188899994, "learning_rate": 2.3845002219342287e-05, "loss": 0.0096, "step": 207440 }, { "epoch": 132.21797323135755, "grad_norm": 0.2859903573989868, "learning_rate": 2.3840305028467e-05, "loss": 0.0086, "step": 207450 }, { "epoch": 132.22434671765456, "grad_norm": 0.17509323358535767, "learning_rate": 2.3835608155456673e-05, "loss": 0.0092, "step": 207460 }, { "epoch": 132.23072020395156, "grad_norm": 0.2151523381471634, "learning_rate": 2.3830911600368434e-05, "loss": 0.0057, "step": 207470 }, { "epoch": 132.23709369024857, "grad_norm": 0.11621182411909103, "learning_rate": 2.382621536325933e-05, "loss": 0.0112, "step": 207480 }, { "epoch": 132.24346717654558, "grad_norm": 0.2138713300228119, "learning_rate": 2.382151944418642e-05, "loss": 0.0073, "step": 207490 }, { "epoch": 132.2498406628426, "grad_norm": 0.28453582525253296, "learning_rate": 2.3816823843206742e-05, "loss": 0.0067, "step": 207500 }, { "epoch": 132.25621414913957, "grad_norm": 0.09429771453142166, "learning_rate": 2.3812128560377416e-05, "loss": 0.0205, "step": 207510 }, { "epoch": 132.26258763543657, "grad_norm": 0.4035918116569519, "learning_rate": 2.3807433595755418e-05, "loss": 0.0113, "step": 207520 }, { "epoch": 132.26896112173358, "grad_norm": 0.27870121598243713, "learning_rate": 2.380273894939785e-05, "loss": 0.005, "step": 207530 }, { "epoch": 132.2753346080306, "grad_norm": 0.6407867670059204, "learning_rate": 2.3798044621361722e-05, "loss": 0.0136, "step": 207540 }, { "epoch": 132.2817080943276, "grad_norm": 0.14330652356147766, "learning_rate": 2.3793350611704123e-05, "loss": 0.0064, "step": 207550 }, { "epoch": 132.2880815806246, "grad_norm": 0.16099122166633606, "learning_rate": 2.378865692048202e-05, "loss": 0.0056, "step": 207560 }, { "epoch": 132.2944550669216, "grad_norm": 0.18897801637649536, "learning_rate": 2.3783963547752504e-05, "loss": 0.0051, "step": 207570 }, { "epoch": 132.30082855321862, "grad_norm": 0.4947758615016937, "learning_rate": 2.3779270493572582e-05, "loss": 0.0157, "step": 207580 }, { "epoch": 132.3072020395156, "grad_norm": 0.47310689091682434, "learning_rate": 2.377457775799928e-05, "loss": 0.0088, "step": 207590 }, { "epoch": 132.3135755258126, "grad_norm": 0.7535581588745117, "learning_rate": 2.3769885341089597e-05, "loss": 0.0081, "step": 207600 }, { "epoch": 132.31994901210962, "grad_norm": 0.1603311002254486, "learning_rate": 2.3765193242900597e-05, "loss": 0.0062, "step": 207610 }, { "epoch": 132.32632249840663, "grad_norm": 0.14210645854473114, "learning_rate": 2.3760501463489266e-05, "loss": 0.0084, "step": 207620 }, { "epoch": 132.33269598470363, "grad_norm": 0.2264302521944046, "learning_rate": 2.3755810002912615e-05, "loss": 0.0082, "step": 207630 }, { "epoch": 132.33906947100064, "grad_norm": 0.1599452793598175, "learning_rate": 2.3751118861227633e-05, "loss": 0.0054, "step": 207640 }, { "epoch": 132.34544295729765, "grad_norm": 0.44603487849235535, "learning_rate": 2.3746428038491363e-05, "loss": 0.0131, "step": 207650 }, { "epoch": 132.35181644359466, "grad_norm": 0.13056588172912598, "learning_rate": 2.3741737534760778e-05, "loss": 0.0219, "step": 207660 }, { "epoch": 132.35818992989164, "grad_norm": 0.1528797447681427, "learning_rate": 2.373704735009287e-05, "loss": 0.007, "step": 207670 }, { "epoch": 132.36456341618864, "grad_norm": 0.16512946784496307, "learning_rate": 2.3732357484544614e-05, "loss": 0.0041, "step": 207680 }, { "epoch": 132.37093690248565, "grad_norm": 0.17794589698314667, "learning_rate": 2.3727667938173038e-05, "loss": 0.0103, "step": 207690 }, { "epoch": 132.37731038878266, "grad_norm": 0.3872010111808777, "learning_rate": 2.3722978711035105e-05, "loss": 0.0088, "step": 207700 }, { "epoch": 132.38368387507967, "grad_norm": 0.21960937976837158, "learning_rate": 2.371828980318778e-05, "loss": 0.0051, "step": 207710 }, { "epoch": 132.39005736137668, "grad_norm": 0.3446994125843048, "learning_rate": 2.3713601214688034e-05, "loss": 0.0095, "step": 207720 }, { "epoch": 132.39643084767368, "grad_norm": 0.44875994324684143, "learning_rate": 2.3708912945592876e-05, "loss": 0.0128, "step": 207730 }, { "epoch": 132.4028043339707, "grad_norm": 0.38377678394317627, "learning_rate": 2.3704224995959247e-05, "loss": 0.0074, "step": 207740 }, { "epoch": 132.4091778202677, "grad_norm": 0.2330186516046524, "learning_rate": 2.3699537365844105e-05, "loss": 0.0103, "step": 207750 }, { "epoch": 132.41555130656468, "grad_norm": 0.47603967785835266, "learning_rate": 2.369485005530442e-05, "loss": 0.0109, "step": 207760 }, { "epoch": 132.4219247928617, "grad_norm": 0.4386747479438782, "learning_rate": 2.3690163064397124e-05, "loss": 0.0084, "step": 207770 }, { "epoch": 132.4282982791587, "grad_norm": 0.4101771116256714, "learning_rate": 2.3685476393179206e-05, "loss": 0.0135, "step": 207780 }, { "epoch": 132.4346717654557, "grad_norm": 0.0992397591471672, "learning_rate": 2.3680790041707595e-05, "loss": 0.0082, "step": 207790 }, { "epoch": 132.4410452517527, "grad_norm": 0.19728806614875793, "learning_rate": 2.3676104010039234e-05, "loss": 0.012, "step": 207800 }, { "epoch": 132.44741873804972, "grad_norm": 0.09261688590049744, "learning_rate": 2.3671418298231042e-05, "loss": 0.0034, "step": 207810 }, { "epoch": 132.45379222434673, "grad_norm": 0.3161605894565582, "learning_rate": 2.366673290634e-05, "loss": 0.0108, "step": 207820 }, { "epoch": 132.46016571064374, "grad_norm": 0.22057877480983734, "learning_rate": 2.3662047834423012e-05, "loss": 0.0237, "step": 207830 }, { "epoch": 132.46653919694072, "grad_norm": 0.10038279742002487, "learning_rate": 2.3657363082537008e-05, "loss": 0.0066, "step": 207840 }, { "epoch": 132.47291268323772, "grad_norm": 0.4873245656490326, "learning_rate": 2.36526786507389e-05, "loss": 0.0108, "step": 207850 }, { "epoch": 132.47928616953473, "grad_norm": 0.2103903591632843, "learning_rate": 2.3647994539085637e-05, "loss": 0.0157, "step": 207860 }, { "epoch": 132.48565965583174, "grad_norm": 0.2532047927379608, "learning_rate": 2.3643310747634127e-05, "loss": 0.0197, "step": 207870 }, { "epoch": 132.49203314212875, "grad_norm": 0.571959376335144, "learning_rate": 2.3638627276441272e-05, "loss": 0.0068, "step": 207880 }, { "epoch": 132.49840662842576, "grad_norm": 0.20890909433364868, "learning_rate": 2.3633944125563965e-05, "loss": 0.0067, "step": 207890 }, { "epoch": 132.50478011472276, "grad_norm": 0.13720613718032837, "learning_rate": 2.3629261295059175e-05, "loss": 0.0278, "step": 207900 }, { "epoch": 132.51115360101977, "grad_norm": 0.2171843945980072, "learning_rate": 2.3624578784983715e-05, "loss": 0.0054, "step": 207910 }, { "epoch": 132.51752708731675, "grad_norm": 0.5751456022262573, "learning_rate": 2.3619896595394542e-05, "loss": 0.0164, "step": 207920 }, { "epoch": 132.52390057361376, "grad_norm": 0.22876495122909546, "learning_rate": 2.361521472634852e-05, "loss": 0.0083, "step": 207930 }, { "epoch": 132.53027405991077, "grad_norm": 0.297566682100296, "learning_rate": 2.361053317790258e-05, "loss": 0.0043, "step": 207940 }, { "epoch": 132.53664754620777, "grad_norm": 0.47922930121421814, "learning_rate": 2.3605851950113543e-05, "loss": 0.0135, "step": 207950 }, { "epoch": 132.54302103250478, "grad_norm": 0.365920752286911, "learning_rate": 2.3601171043038345e-05, "loss": 0.0111, "step": 207960 }, { "epoch": 132.5493945188018, "grad_norm": 0.5293009281158447, "learning_rate": 2.359649045673383e-05, "loss": 0.0048, "step": 207970 }, { "epoch": 132.5557680050988, "grad_norm": 0.9994503855705261, "learning_rate": 2.3591810191256913e-05, "loss": 0.0068, "step": 207980 }, { "epoch": 132.5621414913958, "grad_norm": 0.4287843406200409, "learning_rate": 2.3587130246664402e-05, "loss": 0.0115, "step": 207990 }, { "epoch": 132.5685149776928, "grad_norm": 0.13526007533073425, "learning_rate": 2.358245062301322e-05, "loss": 0.0073, "step": 208000 }, { "epoch": 132.5748884639898, "grad_norm": 0.29134607315063477, "learning_rate": 2.3577771320360197e-05, "loss": 0.0097, "step": 208010 }, { "epoch": 132.5812619502868, "grad_norm": 0.09957001358270645, "learning_rate": 2.3573092338762202e-05, "loss": 0.0069, "step": 208020 }, { "epoch": 132.5876354365838, "grad_norm": 0.14430487155914307, "learning_rate": 2.3568413678276068e-05, "loss": 0.0087, "step": 208030 }, { "epoch": 132.59400892288082, "grad_norm": 0.1876784861087799, "learning_rate": 2.3563735338958687e-05, "loss": 0.0073, "step": 208040 }, { "epoch": 132.60038240917783, "grad_norm": 0.1714451164007187, "learning_rate": 2.3559057320866873e-05, "loss": 0.0105, "step": 208050 }, { "epoch": 132.60675589547483, "grad_norm": 0.21065591275691986, "learning_rate": 2.3554379624057482e-05, "loss": 0.0056, "step": 208060 }, { "epoch": 132.61312938177184, "grad_norm": 0.4196683168411255, "learning_rate": 2.354970224858733e-05, "loss": 0.0087, "step": 208070 }, { "epoch": 132.61950286806882, "grad_norm": 0.20988376438617706, "learning_rate": 2.3545025194513304e-05, "loss": 0.0075, "step": 208080 }, { "epoch": 132.62587635436583, "grad_norm": 0.4082799553871155, "learning_rate": 2.354034846189216e-05, "loss": 0.0058, "step": 208090 }, { "epoch": 132.63224984066284, "grad_norm": 0.3294207751750946, "learning_rate": 2.353567205078079e-05, "loss": 0.0076, "step": 208100 }, { "epoch": 132.63862332695984, "grad_norm": 0.10788829624652863, "learning_rate": 2.3530995961235962e-05, "loss": 0.0156, "step": 208110 }, { "epoch": 132.64499681325685, "grad_norm": 0.20633111894130707, "learning_rate": 2.3526320193314572e-05, "loss": 0.0131, "step": 208120 }, { "epoch": 132.65137029955386, "grad_norm": 0.31086769700050354, "learning_rate": 2.352164474707334e-05, "loss": 0.0061, "step": 208130 }, { "epoch": 132.65774378585087, "grad_norm": 0.2618934214115143, "learning_rate": 2.3516969622569146e-05, "loss": 0.0071, "step": 208140 }, { "epoch": 132.66411727214788, "grad_norm": 0.15089167654514313, "learning_rate": 2.3512294819858777e-05, "loss": 0.0084, "step": 208150 }, { "epoch": 132.67049075844486, "grad_norm": 0.2790548503398895, "learning_rate": 2.3507620338999038e-05, "loss": 0.0089, "step": 208160 }, { "epoch": 132.67686424474186, "grad_norm": 0.2508986294269562, "learning_rate": 2.3502946180046696e-05, "loss": 0.0075, "step": 208170 }, { "epoch": 132.68323773103887, "grad_norm": 0.7216712236404419, "learning_rate": 2.34982723430586e-05, "loss": 0.0074, "step": 208180 }, { "epoch": 132.68961121733588, "grad_norm": 0.3295466899871826, "learning_rate": 2.3493598828091518e-05, "loss": 0.0068, "step": 208190 }, { "epoch": 132.6959847036329, "grad_norm": 0.11823663860559464, "learning_rate": 2.3488925635202235e-05, "loss": 0.0063, "step": 208200 }, { "epoch": 132.7023581899299, "grad_norm": 0.15202485024929047, "learning_rate": 2.3484252764447516e-05, "loss": 0.0085, "step": 208210 }, { "epoch": 132.7087316762269, "grad_norm": 0.12270059436559677, "learning_rate": 2.3479580215884205e-05, "loss": 0.0068, "step": 208220 }, { "epoch": 132.7151051625239, "grad_norm": 0.25686901807785034, "learning_rate": 2.347490798956899e-05, "loss": 0.006, "step": 208230 }, { "epoch": 132.7214786488209, "grad_norm": 0.24630093574523926, "learning_rate": 2.3470236085558706e-05, "loss": 0.0112, "step": 208240 }, { "epoch": 132.7278521351179, "grad_norm": 0.10443603247404099, "learning_rate": 2.3465564503910088e-05, "loss": 0.0185, "step": 208250 }, { "epoch": 132.7342256214149, "grad_norm": 0.12331133335828781, "learning_rate": 2.3460893244679943e-05, "loss": 0.0039, "step": 208260 }, { "epoch": 132.74059910771192, "grad_norm": 0.20859533548355103, "learning_rate": 2.3456222307924968e-05, "loss": 0.0114, "step": 208270 }, { "epoch": 132.74697259400892, "grad_norm": 0.17044399678707123, "learning_rate": 2.3451551693701973e-05, "loss": 0.0068, "step": 208280 }, { "epoch": 132.75334608030593, "grad_norm": 0.2722562849521637, "learning_rate": 2.344688140206767e-05, "loss": 0.0057, "step": 208290 }, { "epoch": 132.75971956660294, "grad_norm": 0.1857479363679886, "learning_rate": 2.344221143307887e-05, "loss": 0.0112, "step": 208300 }, { "epoch": 132.76609305289995, "grad_norm": 0.16136100888252258, "learning_rate": 2.3437541786792238e-05, "loss": 0.014, "step": 208310 }, { "epoch": 132.77246653919693, "grad_norm": 0.20853658020496368, "learning_rate": 2.3432872463264573e-05, "loss": 0.0104, "step": 208320 }, { "epoch": 132.77884002549393, "grad_norm": 0.1788104921579361, "learning_rate": 2.3428203462552585e-05, "loss": 0.0079, "step": 208330 }, { "epoch": 132.78521351179094, "grad_norm": 0.2846098244190216, "learning_rate": 2.3423534784713013e-05, "loss": 0.0051, "step": 208340 }, { "epoch": 132.79158699808795, "grad_norm": 0.35715439915657043, "learning_rate": 2.341886642980257e-05, "loss": 0.0055, "step": 208350 }, { "epoch": 132.79796048438496, "grad_norm": 0.568095326423645, "learning_rate": 2.3414198397878018e-05, "loss": 0.0138, "step": 208360 }, { "epoch": 132.80433397068197, "grad_norm": 0.31306979060173035, "learning_rate": 2.340953068899605e-05, "loss": 0.0126, "step": 208370 }, { "epoch": 132.81070745697897, "grad_norm": 0.2277352511882782, "learning_rate": 2.3404863303213393e-05, "loss": 0.0075, "step": 208380 }, { "epoch": 132.81708094327598, "grad_norm": 0.1191881000995636, "learning_rate": 2.340019624058674e-05, "loss": 0.005, "step": 208390 }, { "epoch": 132.823454429573, "grad_norm": 0.31646883487701416, "learning_rate": 2.3395529501172836e-05, "loss": 0.01, "step": 208400 }, { "epoch": 132.82982791586997, "grad_norm": 0.09406021237373352, "learning_rate": 2.3390863085028364e-05, "loss": 0.0044, "step": 208410 }, { "epoch": 132.83620140216698, "grad_norm": 0.46196940541267395, "learning_rate": 2.338619699221003e-05, "loss": 0.0099, "step": 208420 }, { "epoch": 132.842574888464, "grad_norm": 0.5575750470161438, "learning_rate": 2.338153122277451e-05, "loss": 0.0079, "step": 208430 }, { "epoch": 132.848948374761, "grad_norm": 0.5282983779907227, "learning_rate": 2.3376865776778538e-05, "loss": 0.0098, "step": 208440 }, { "epoch": 132.855321861058, "grad_norm": 0.17658911645412445, "learning_rate": 2.3372200654278782e-05, "loss": 0.0057, "step": 208450 }, { "epoch": 132.861695347355, "grad_norm": 0.1923454999923706, "learning_rate": 2.336753585533193e-05, "loss": 0.0169, "step": 208460 }, { "epoch": 132.86806883365202, "grad_norm": 0.21284371614456177, "learning_rate": 2.336287137999466e-05, "loss": 0.0066, "step": 208470 }, { "epoch": 132.87444231994903, "grad_norm": 0.12414535880088806, "learning_rate": 2.3358207228323635e-05, "loss": 0.0164, "step": 208480 }, { "epoch": 132.880815806246, "grad_norm": 0.4316423535346985, "learning_rate": 2.335354340037557e-05, "loss": 0.0068, "step": 208490 }, { "epoch": 132.887189292543, "grad_norm": 0.5041974782943726, "learning_rate": 2.3348879896207105e-05, "loss": 0.0151, "step": 208500 }, { "epoch": 132.89356277884002, "grad_norm": 0.24005837738513947, "learning_rate": 2.3344216715874916e-05, "loss": 0.0101, "step": 208510 }, { "epoch": 132.89993626513703, "grad_norm": 0.422002911567688, "learning_rate": 2.333955385943564e-05, "loss": 0.0125, "step": 208520 }, { "epoch": 132.90630975143404, "grad_norm": 0.3373175263404846, "learning_rate": 2.3334891326945974e-05, "loss": 0.0116, "step": 208530 }, { "epoch": 132.91268323773104, "grad_norm": 0.11418671160936356, "learning_rate": 2.333022911846256e-05, "loss": 0.0055, "step": 208540 }, { "epoch": 132.91905672402805, "grad_norm": 0.11273584514856339, "learning_rate": 2.3325567234042033e-05, "loss": 0.0056, "step": 208550 }, { "epoch": 132.92543021032506, "grad_norm": 0.31412649154663086, "learning_rate": 2.332090567374104e-05, "loss": 0.0109, "step": 208560 }, { "epoch": 132.93180369662204, "grad_norm": 0.2153327465057373, "learning_rate": 2.3316244437616248e-05, "loss": 0.0068, "step": 208570 }, { "epoch": 132.93817718291905, "grad_norm": 0.18942472338676453, "learning_rate": 2.331158352572428e-05, "loss": 0.0059, "step": 208580 }, { "epoch": 132.94455066921606, "grad_norm": 0.18095986545085907, "learning_rate": 2.3306922938121768e-05, "loss": 0.0059, "step": 208590 }, { "epoch": 132.95092415551306, "grad_norm": 0.19290249049663544, "learning_rate": 2.3302262674865323e-05, "loss": 0.0121, "step": 208600 }, { "epoch": 132.95729764181007, "grad_norm": 0.21659448742866516, "learning_rate": 2.3297602736011615e-05, "loss": 0.005, "step": 208610 }, { "epoch": 132.96367112810708, "grad_norm": 0.10941784828901291, "learning_rate": 2.329294312161725e-05, "loss": 0.0065, "step": 208620 }, { "epoch": 132.9700446144041, "grad_norm": 0.4465150237083435, "learning_rate": 2.3288283831738834e-05, "loss": 0.0103, "step": 208630 }, { "epoch": 132.9764181007011, "grad_norm": 0.19225746393203735, "learning_rate": 2.3283624866432974e-05, "loss": 0.0052, "step": 208640 }, { "epoch": 132.98279158699808, "grad_norm": 0.4638186991214752, "learning_rate": 2.327896622575634e-05, "loss": 0.0082, "step": 208650 }, { "epoch": 132.98916507329508, "grad_norm": 0.4357037842273712, "learning_rate": 2.3274307909765448e-05, "loss": 0.0085, "step": 208660 }, { "epoch": 132.9955385595921, "grad_norm": 0.26878786087036133, "learning_rate": 2.3269649918516974e-05, "loss": 0.0155, "step": 208670 }, { "epoch": 133.0019120458891, "grad_norm": 0.13332504034042358, "learning_rate": 2.3264992252067464e-05, "loss": 0.0079, "step": 208680 }, { "epoch": 133.0082855321861, "grad_norm": 0.2609400451183319, "learning_rate": 2.3260334910473585e-05, "loss": 0.0074, "step": 208690 }, { "epoch": 133.01465901848312, "grad_norm": 0.2247215062379837, "learning_rate": 2.3255677893791845e-05, "loss": 0.0044, "step": 208700 }, { "epoch": 133.02103250478012, "grad_norm": 0.2998594045639038, "learning_rate": 2.3251021202078882e-05, "loss": 0.0077, "step": 208710 }, { "epoch": 133.02740599107713, "grad_norm": 0.24752506613731384, "learning_rate": 2.3246364835391272e-05, "loss": 0.003, "step": 208720 }, { "epoch": 133.0337794773741, "grad_norm": 0.2803555727005005, "learning_rate": 2.324170879378559e-05, "loss": 0.0085, "step": 208730 }, { "epoch": 133.04015296367112, "grad_norm": 0.2595332860946655, "learning_rate": 2.3237053077318387e-05, "loss": 0.0089, "step": 208740 }, { "epoch": 133.04652644996813, "grad_norm": 0.32761067152023315, "learning_rate": 2.3232397686046282e-05, "loss": 0.0074, "step": 208750 }, { "epoch": 133.05289993626513, "grad_norm": 0.14208076894283295, "learning_rate": 2.322774262002581e-05, "loss": 0.0078, "step": 208760 }, { "epoch": 133.05927342256214, "grad_norm": 0.32558414340019226, "learning_rate": 2.3223087879313548e-05, "loss": 0.0091, "step": 208770 }, { "epoch": 133.06564690885915, "grad_norm": 0.18155889213085175, "learning_rate": 2.3218433463966032e-05, "loss": 0.0052, "step": 208780 }, { "epoch": 133.07202039515616, "grad_norm": 0.11831308156251907, "learning_rate": 2.3213779374039874e-05, "loss": 0.0111, "step": 208790 }, { "epoch": 133.07839388145317, "grad_norm": 0.18850424885749817, "learning_rate": 2.3209125609591543e-05, "loss": 0.0041, "step": 208800 }, { "epoch": 133.08476736775015, "grad_norm": 0.28981122374534607, "learning_rate": 2.3204472170677655e-05, "loss": 0.0076, "step": 208810 }, { "epoch": 133.09114085404715, "grad_norm": 0.25061196088790894, "learning_rate": 2.319981905735471e-05, "loss": 0.0077, "step": 208820 }, { "epoch": 133.09751434034416, "grad_norm": 0.15318623185157776, "learning_rate": 2.3195166269679303e-05, "loss": 0.0156, "step": 208830 }, { "epoch": 133.10388782664117, "grad_norm": 0.2628040015697479, "learning_rate": 2.319051380770789e-05, "loss": 0.0054, "step": 208840 }, { "epoch": 133.11026131293818, "grad_norm": 0.1732509285211563, "learning_rate": 2.318586167149708e-05, "loss": 0.007, "step": 208850 }, { "epoch": 133.1166347992352, "grad_norm": 0.5343897342681885, "learning_rate": 2.3181209861103336e-05, "loss": 0.0152, "step": 208860 }, { "epoch": 133.1230082855322, "grad_norm": 0.11149527877569199, "learning_rate": 2.317655837658326e-05, "loss": 0.0085, "step": 208870 }, { "epoch": 133.1293817718292, "grad_norm": 0.4052315950393677, "learning_rate": 2.3171907217993277e-05, "loss": 0.009, "step": 208880 }, { "epoch": 133.13575525812618, "grad_norm": 0.3152958154678345, "learning_rate": 2.316725638538998e-05, "loss": 0.0097, "step": 208890 }, { "epoch": 133.1421287444232, "grad_norm": 0.23077090084552765, "learning_rate": 2.3162605878829847e-05, "loss": 0.0092, "step": 208900 }, { "epoch": 133.1485022307202, "grad_norm": 0.5838913321495056, "learning_rate": 2.3157955698369393e-05, "loss": 0.0116, "step": 208910 }, { "epoch": 133.1548757170172, "grad_norm": 0.37321364879608154, "learning_rate": 2.3153305844065097e-05, "loss": 0.0064, "step": 208920 }, { "epoch": 133.1612492033142, "grad_norm": 0.21337176859378815, "learning_rate": 2.3148656315973505e-05, "loss": 0.0068, "step": 208930 }, { "epoch": 133.16762268961122, "grad_norm": 0.3195952773094177, "learning_rate": 2.3144007114151096e-05, "loss": 0.0075, "step": 208940 }, { "epoch": 133.17399617590823, "grad_norm": 0.1713661104440689, "learning_rate": 2.3139358238654348e-05, "loss": 0.0069, "step": 208950 }, { "epoch": 133.18036966220524, "grad_norm": 0.30784472823143005, "learning_rate": 2.313470968953974e-05, "loss": 0.0075, "step": 208960 }, { "epoch": 133.18674314850222, "grad_norm": 0.4007425010204315, "learning_rate": 2.3130061466863816e-05, "loss": 0.0109, "step": 208970 }, { "epoch": 133.19311663479922, "grad_norm": 0.2947174906730652, "learning_rate": 2.312541357068298e-05, "loss": 0.0113, "step": 208980 }, { "epoch": 133.19949012109623, "grad_norm": 0.3589024841785431, "learning_rate": 2.3120766001053757e-05, "loss": 0.0108, "step": 208990 }, { "epoch": 133.20586360739324, "grad_norm": 0.1811601221561432, "learning_rate": 2.3116118758032583e-05, "loss": 0.0097, "step": 209000 }, { "epoch": 133.21223709369025, "grad_norm": 0.23793098330497742, "learning_rate": 2.311147184167599e-05, "loss": 0.0157, "step": 209010 }, { "epoch": 133.21861057998726, "grad_norm": 0.10718502104282379, "learning_rate": 2.3106825252040365e-05, "loss": 0.0123, "step": 209020 }, { "epoch": 133.22498406628426, "grad_norm": 0.3374631404876709, "learning_rate": 2.310217898918222e-05, "loss": 0.0061, "step": 209030 }, { "epoch": 133.23135755258127, "grad_norm": 0.31195932626724243, "learning_rate": 2.3097533053158006e-05, "loss": 0.0071, "step": 209040 }, { "epoch": 133.23773103887828, "grad_norm": 0.11818850785493851, "learning_rate": 2.3092887444024152e-05, "loss": 0.0059, "step": 209050 }, { "epoch": 133.24410452517526, "grad_norm": 0.3062325716018677, "learning_rate": 2.308824216183711e-05, "loss": 0.0082, "step": 209060 }, { "epoch": 133.25047801147227, "grad_norm": 0.22322800755500793, "learning_rate": 2.308359720665335e-05, "loss": 0.0054, "step": 209070 }, { "epoch": 133.25685149776928, "grad_norm": 0.11980710923671722, "learning_rate": 2.3078952578529296e-05, "loss": 0.0066, "step": 209080 }, { "epoch": 133.26322498406628, "grad_norm": 0.19181190431118011, "learning_rate": 2.307430827752139e-05, "loss": 0.0096, "step": 209090 }, { "epoch": 133.2695984703633, "grad_norm": 0.12450841814279556, "learning_rate": 2.306966430368604e-05, "loss": 0.0082, "step": 209100 }, { "epoch": 133.2759719566603, "grad_norm": 0.5066869854927063, "learning_rate": 2.3065020657079712e-05, "loss": 0.009, "step": 209110 }, { "epoch": 133.2823454429573, "grad_norm": 0.11152461171150208, "learning_rate": 2.3060377337758813e-05, "loss": 0.0049, "step": 209120 }, { "epoch": 133.28871892925432, "grad_norm": 0.2339470386505127, "learning_rate": 2.305573434577975e-05, "loss": 0.0087, "step": 209130 }, { "epoch": 133.2950924155513, "grad_norm": 0.1921219825744629, "learning_rate": 2.3051091681198972e-05, "loss": 0.0053, "step": 209140 }, { "epoch": 133.3014659018483, "grad_norm": 0.18182183802127838, "learning_rate": 2.3046449344072875e-05, "loss": 0.0076, "step": 209150 }, { "epoch": 133.3078393881453, "grad_norm": 0.2128463089466095, "learning_rate": 2.304180733445786e-05, "loss": 0.009, "step": 209160 }, { "epoch": 133.31421287444232, "grad_norm": 0.4023610055446625, "learning_rate": 2.3037165652410324e-05, "loss": 0.0085, "step": 209170 }, { "epoch": 133.32058636073933, "grad_norm": 0.13248057663440704, "learning_rate": 2.30325242979867e-05, "loss": 0.0046, "step": 209180 }, { "epoch": 133.32695984703633, "grad_norm": 0.3644568622112274, "learning_rate": 2.3027883271243372e-05, "loss": 0.0049, "step": 209190 }, { "epoch": 133.33333333333334, "grad_norm": 0.17187228798866272, "learning_rate": 2.3023242572236724e-05, "loss": 0.0094, "step": 209200 }, { "epoch": 133.33970681963035, "grad_norm": 0.19756832718849182, "learning_rate": 2.3018602201023133e-05, "loss": 0.0067, "step": 209210 }, { "epoch": 133.34608030592733, "grad_norm": 0.6667490005493164, "learning_rate": 2.3013962157659035e-05, "loss": 0.0116, "step": 209220 }, { "epoch": 133.35245379222434, "grad_norm": 0.08364143967628479, "learning_rate": 2.3009322442200743e-05, "loss": 0.0129, "step": 209230 }, { "epoch": 133.35882727852135, "grad_norm": 0.15380804240703583, "learning_rate": 2.3004683054704678e-05, "loss": 0.0096, "step": 209240 }, { "epoch": 133.36520076481835, "grad_norm": 0.18484210968017578, "learning_rate": 2.300004399522719e-05, "loss": 0.0075, "step": 209250 }, { "epoch": 133.37157425111536, "grad_norm": 0.3144412338733673, "learning_rate": 2.2995405263824692e-05, "loss": 0.0174, "step": 209260 }, { "epoch": 133.37794773741237, "grad_norm": 0.28573036193847656, "learning_rate": 2.299076686055348e-05, "loss": 0.0075, "step": 209270 }, { "epoch": 133.38432122370938, "grad_norm": 0.28654372692108154, "learning_rate": 2.2986128785469973e-05, "loss": 0.0074, "step": 209280 }, { "epoch": 133.3906947100064, "grad_norm": 0.28356608748435974, "learning_rate": 2.2981491038630505e-05, "loss": 0.0092, "step": 209290 }, { "epoch": 133.39706819630337, "grad_norm": 0.4689754545688629, "learning_rate": 2.2976853620091433e-05, "loss": 0.0166, "step": 209300 }, { "epoch": 133.40344168260037, "grad_norm": 0.148905411362648, "learning_rate": 2.297221652990908e-05, "loss": 0.0053, "step": 209310 }, { "epoch": 133.40981516889738, "grad_norm": 0.20044897496700287, "learning_rate": 2.2967579768139834e-05, "loss": 0.007, "step": 209320 }, { "epoch": 133.4161886551944, "grad_norm": 0.10816063731908798, "learning_rate": 2.296294333484002e-05, "loss": 0.0062, "step": 209330 }, { "epoch": 133.4225621414914, "grad_norm": 0.2579619288444519, "learning_rate": 2.2958307230065972e-05, "loss": 0.0077, "step": 209340 }, { "epoch": 133.4289356277884, "grad_norm": 0.18412673473358154, "learning_rate": 2.2953671453874003e-05, "loss": 0.0068, "step": 209350 }, { "epoch": 133.4353091140854, "grad_norm": 0.5633890628814697, "learning_rate": 2.2949036006320497e-05, "loss": 0.0124, "step": 209360 }, { "epoch": 133.44168260038242, "grad_norm": 0.1866784542798996, "learning_rate": 2.2944400887461708e-05, "loss": 0.0099, "step": 209370 }, { "epoch": 133.4480560866794, "grad_norm": 0.18230080604553223, "learning_rate": 2.293976609735401e-05, "loss": 0.0117, "step": 209380 }, { "epoch": 133.4544295729764, "grad_norm": 0.3487936854362488, "learning_rate": 2.2935131636053687e-05, "loss": 0.0051, "step": 209390 }, { "epoch": 133.46080305927342, "grad_norm": 0.1710105985403061, "learning_rate": 2.2930497503617105e-05, "loss": 0.0075, "step": 209400 }, { "epoch": 133.46717654557042, "grad_norm": 0.16034741699695587, "learning_rate": 2.2925863700100496e-05, "loss": 0.0053, "step": 209410 }, { "epoch": 133.47355003186743, "grad_norm": 0.20656411349773407, "learning_rate": 2.2921230225560224e-05, "loss": 0.0074, "step": 209420 }, { "epoch": 133.47992351816444, "grad_norm": 0.2137531042098999, "learning_rate": 2.2916597080052578e-05, "loss": 0.0088, "step": 209430 }, { "epoch": 133.48629700446145, "grad_norm": 0.19310688972473145, "learning_rate": 2.2911964263633846e-05, "loss": 0.0115, "step": 209440 }, { "epoch": 133.49267049075846, "grad_norm": 0.21999256312847137, "learning_rate": 2.2907331776360297e-05, "loss": 0.0068, "step": 209450 }, { "epoch": 133.49904397705544, "grad_norm": 0.21535074710845947, "learning_rate": 2.2902699618288275e-05, "loss": 0.0093, "step": 209460 }, { "epoch": 133.50541746335244, "grad_norm": 0.20467233657836914, "learning_rate": 2.2898067789474033e-05, "loss": 0.0049, "step": 209470 }, { "epoch": 133.51179094964945, "grad_norm": 0.19328881800174713, "learning_rate": 2.289343628997385e-05, "loss": 0.0043, "step": 209480 }, { "epoch": 133.51816443594646, "grad_norm": 0.2537834942340851, "learning_rate": 2.2888805119844e-05, "loss": 0.006, "step": 209490 }, { "epoch": 133.52453792224347, "grad_norm": 0.18327175080776215, "learning_rate": 2.288417427914078e-05, "loss": 0.0077, "step": 209500 }, { "epoch": 133.53091140854048, "grad_norm": 0.23857282102108002, "learning_rate": 2.287954376792044e-05, "loss": 0.0075, "step": 209510 }, { "epoch": 133.53728489483748, "grad_norm": 0.20872478187084198, "learning_rate": 2.2874913586239254e-05, "loss": 0.0066, "step": 209520 }, { "epoch": 133.5436583811345, "grad_norm": 0.23197956383228302, "learning_rate": 2.2870283734153458e-05, "loss": 0.0043, "step": 209530 }, { "epoch": 133.55003186743147, "grad_norm": 0.2325730174779892, "learning_rate": 2.2865654211719366e-05, "loss": 0.0075, "step": 209540 }, { "epoch": 133.55640535372848, "grad_norm": 0.15091219544410706, "learning_rate": 2.286102501899316e-05, "loss": 0.0125, "step": 209550 }, { "epoch": 133.5627788400255, "grad_norm": 0.10890689492225647, "learning_rate": 2.285639615603114e-05, "loss": 0.0045, "step": 209560 }, { "epoch": 133.5691523263225, "grad_norm": 0.2280227243900299, "learning_rate": 2.2851767622889513e-05, "loss": 0.0157, "step": 209570 }, { "epoch": 133.5755258126195, "grad_norm": 0.3846968710422516, "learning_rate": 2.2847139419624584e-05, "loss": 0.0082, "step": 209580 }, { "epoch": 133.5818992989165, "grad_norm": 0.2610900402069092, "learning_rate": 2.284251154629251e-05, "loss": 0.0083, "step": 209590 }, { "epoch": 133.58827278521352, "grad_norm": 0.4278898537158966, "learning_rate": 2.283788400294958e-05, "loss": 0.0099, "step": 209600 }, { "epoch": 133.59464627151053, "grad_norm": 0.1495106816291809, "learning_rate": 2.2833256789652e-05, "loss": 0.0078, "step": 209610 }, { "epoch": 133.60101975780753, "grad_norm": 0.2571331858634949, "learning_rate": 2.2828629906456005e-05, "loss": 0.0082, "step": 209620 }, { "epoch": 133.60739324410451, "grad_norm": 0.40233439207077026, "learning_rate": 2.2824003353417787e-05, "loss": 0.0069, "step": 209630 }, { "epoch": 133.61376673040152, "grad_norm": 0.21155031025409698, "learning_rate": 2.281937713059361e-05, "loss": 0.0097, "step": 209640 }, { "epoch": 133.62014021669853, "grad_norm": 0.22649331390857697, "learning_rate": 2.281475123803966e-05, "loss": 0.0155, "step": 209650 }, { "epoch": 133.62651370299554, "grad_norm": 0.10261798650026321, "learning_rate": 2.2810125675812154e-05, "loss": 0.0086, "step": 209660 }, { "epoch": 133.63288718929255, "grad_norm": 0.4476805627346039, "learning_rate": 2.280550044396727e-05, "loss": 0.0046, "step": 209670 }, { "epoch": 133.63926067558955, "grad_norm": 0.29650500416755676, "learning_rate": 2.2800875542561268e-05, "loss": 0.0095, "step": 209680 }, { "epoch": 133.64563416188656, "grad_norm": 0.2832626700401306, "learning_rate": 2.2796250971650275e-05, "loss": 0.0073, "step": 209690 }, { "epoch": 133.65200764818357, "grad_norm": 0.14608751237392426, "learning_rate": 2.279162673129053e-05, "loss": 0.0039, "step": 209700 }, { "epoch": 133.65838113448055, "grad_norm": 0.16947200894355774, "learning_rate": 2.27870028215382e-05, "loss": 0.0096, "step": 209710 }, { "epoch": 133.66475462077756, "grad_norm": 0.14382252097129822, "learning_rate": 2.2782379242449493e-05, "loss": 0.0083, "step": 209720 }, { "epoch": 133.67112810707457, "grad_norm": 0.2614913582801819, "learning_rate": 2.2777755994080575e-05, "loss": 0.0125, "step": 209730 }, { "epoch": 133.67750159337157, "grad_norm": 0.22043412923812866, "learning_rate": 2.2773133076487625e-05, "loss": 0.0063, "step": 209740 }, { "epoch": 133.68387507966858, "grad_norm": 0.12280425429344177, "learning_rate": 2.2768510489726808e-05, "loss": 0.0097, "step": 209750 }, { "epoch": 133.6902485659656, "grad_norm": 0.22298923134803772, "learning_rate": 2.2763888233854285e-05, "loss": 0.0055, "step": 209760 }, { "epoch": 133.6966220522626, "grad_norm": 0.24359771609306335, "learning_rate": 2.275926630892626e-05, "loss": 0.0048, "step": 209770 }, { "epoch": 133.7029955385596, "grad_norm": 0.09813204407691956, "learning_rate": 2.2754644714998862e-05, "loss": 0.0107, "step": 209780 }, { "epoch": 133.70936902485658, "grad_norm": 0.16631746292114258, "learning_rate": 2.2750023452128254e-05, "loss": 0.009, "step": 209790 }, { "epoch": 133.7157425111536, "grad_norm": 0.32303646206855774, "learning_rate": 2.2745402520370574e-05, "loss": 0.0106, "step": 209800 }, { "epoch": 133.7221159974506, "grad_norm": 0.28397852182388306, "learning_rate": 2.2740781919782007e-05, "loss": 0.0088, "step": 209810 }, { "epoch": 133.7284894837476, "grad_norm": 0.24999652802944183, "learning_rate": 2.2736161650418668e-05, "loss": 0.0075, "step": 209820 }, { "epoch": 133.73486297004462, "grad_norm": 0.2899072468280792, "learning_rate": 2.2731541712336714e-05, "loss": 0.0073, "step": 209830 }, { "epoch": 133.74123645634162, "grad_norm": 0.20497678220272064, "learning_rate": 2.2726922105592253e-05, "loss": 0.0118, "step": 209840 }, { "epoch": 133.74760994263863, "grad_norm": 0.5027830004692078, "learning_rate": 2.2722302830241454e-05, "loss": 0.0079, "step": 209850 }, { "epoch": 133.75398342893564, "grad_norm": 0.3720428943634033, "learning_rate": 2.2717683886340434e-05, "loss": 0.0057, "step": 209860 }, { "epoch": 133.76035691523262, "grad_norm": 0.17893846333026886, "learning_rate": 2.2713065273945318e-05, "loss": 0.0067, "step": 209870 }, { "epoch": 133.76673040152963, "grad_norm": 0.2457980066537857, "learning_rate": 2.2708446993112192e-05, "loss": 0.0263, "step": 209880 }, { "epoch": 133.77310388782664, "grad_norm": 0.1506902575492859, "learning_rate": 2.270382904389723e-05, "loss": 0.0059, "step": 209890 }, { "epoch": 133.77947737412364, "grad_norm": 0.24910348653793335, "learning_rate": 2.2699211426356515e-05, "loss": 0.0049, "step": 209900 }, { "epoch": 133.78585086042065, "grad_norm": 0.6322899460792542, "learning_rate": 2.2694594140546156e-05, "loss": 0.0098, "step": 209910 }, { "epoch": 133.79222434671766, "grad_norm": 0.2936779856681824, "learning_rate": 2.2689977186522237e-05, "loss": 0.0082, "step": 209920 }, { "epoch": 133.79859783301467, "grad_norm": 0.2799830138683319, "learning_rate": 2.2685360564340924e-05, "loss": 0.0112, "step": 209930 }, { "epoch": 133.80497131931168, "grad_norm": 0.37434443831443787, "learning_rate": 2.268074427405823e-05, "loss": 0.013, "step": 209940 }, { "epoch": 133.81134480560866, "grad_norm": 0.16197063028812408, "learning_rate": 2.26761283157303e-05, "loss": 0.0119, "step": 209950 }, { "epoch": 133.81771829190566, "grad_norm": 0.1695319265127182, "learning_rate": 2.2671512689413198e-05, "loss": 0.0059, "step": 209960 }, { "epoch": 133.82409177820267, "grad_norm": 0.1551632434129715, "learning_rate": 2.2666897395163052e-05, "loss": 0.0051, "step": 209970 }, { "epoch": 133.83046526449968, "grad_norm": 0.12212195992469788, "learning_rate": 2.2662282433035876e-05, "loss": 0.0091, "step": 209980 }, { "epoch": 133.8368387507967, "grad_norm": 0.3466648757457733, "learning_rate": 2.2657667803087795e-05, "loss": 0.0056, "step": 209990 }, { "epoch": 133.8432122370937, "grad_norm": 0.17436403036117554, "learning_rate": 2.2653053505374866e-05, "loss": 0.0062, "step": 210000 }, { "epoch": 133.8495857233907, "grad_norm": 0.42342859506607056, "learning_rate": 2.2648439539953155e-05, "loss": 0.0073, "step": 210010 }, { "epoch": 133.8559592096877, "grad_norm": 0.15706409513950348, "learning_rate": 2.264382590687871e-05, "loss": 0.0139, "step": 210020 }, { "epoch": 133.8623326959847, "grad_norm": 0.1505737453699112, "learning_rate": 2.2639212606207627e-05, "loss": 0.0072, "step": 210030 }, { "epoch": 133.8687061822817, "grad_norm": 0.1191784143447876, "learning_rate": 2.2634599637995945e-05, "loss": 0.0081, "step": 210040 }, { "epoch": 133.8750796685787, "grad_norm": 0.2672337293624878, "learning_rate": 2.2629987002299718e-05, "loss": 0.0073, "step": 210050 }, { "epoch": 133.88145315487571, "grad_norm": 0.2822096049785614, "learning_rate": 2.2625374699174968e-05, "loss": 0.0147, "step": 210060 }, { "epoch": 133.88782664117272, "grad_norm": 0.19563084840774536, "learning_rate": 2.262076272867778e-05, "loss": 0.0143, "step": 210070 }, { "epoch": 133.89420012746973, "grad_norm": 0.3436216115951538, "learning_rate": 2.2616151090864173e-05, "loss": 0.0086, "step": 210080 }, { "epoch": 133.90057361376674, "grad_norm": 0.1305565983057022, "learning_rate": 2.2611539785790187e-05, "loss": 0.0209, "step": 210090 }, { "epoch": 133.90694710006375, "grad_norm": 0.28106850385665894, "learning_rate": 2.2606928813511834e-05, "loss": 0.01, "step": 210100 }, { "epoch": 133.91332058636073, "grad_norm": 0.2698580026626587, "learning_rate": 2.2602318174085198e-05, "loss": 0.0115, "step": 210110 }, { "epoch": 133.91969407265773, "grad_norm": 0.22645150125026703, "learning_rate": 2.2597707867566227e-05, "loss": 0.0085, "step": 210120 }, { "epoch": 133.92606755895474, "grad_norm": 0.11963307112455368, "learning_rate": 2.259309789401099e-05, "loss": 0.0133, "step": 210130 }, { "epoch": 133.93244104525175, "grad_norm": 0.2699699401855469, "learning_rate": 2.2588488253475476e-05, "loss": 0.0053, "step": 210140 }, { "epoch": 133.93881453154876, "grad_norm": 0.2456587553024292, "learning_rate": 2.2583878946015746e-05, "loss": 0.0066, "step": 210150 }, { "epoch": 133.94518801784577, "grad_norm": 0.3160649538040161, "learning_rate": 2.2579269971687738e-05, "loss": 0.0057, "step": 210160 }, { "epoch": 133.95156150414277, "grad_norm": 0.10573146492242813, "learning_rate": 2.2574661330547502e-05, "loss": 0.0104, "step": 210170 }, { "epoch": 133.95793499043978, "grad_norm": 0.3590652644634247, "learning_rate": 2.2570053022651027e-05, "loss": 0.0068, "step": 210180 }, { "epoch": 133.96430847673676, "grad_norm": 0.2688010036945343, "learning_rate": 2.2565445048054313e-05, "loss": 0.0094, "step": 210190 }, { "epoch": 133.97068196303377, "grad_norm": 0.13986477255821228, "learning_rate": 2.2560837406813316e-05, "loss": 0.0124, "step": 210200 }, { "epoch": 133.97705544933078, "grad_norm": 0.19705986976623535, "learning_rate": 2.2556230098984072e-05, "loss": 0.0062, "step": 210210 }, { "epoch": 133.98342893562778, "grad_norm": 0.2982979714870453, "learning_rate": 2.2551623124622544e-05, "loss": 0.0123, "step": 210220 }, { "epoch": 133.9898024219248, "grad_norm": 0.13672609627246857, "learning_rate": 2.2547016483784718e-05, "loss": 0.012, "step": 210230 }, { "epoch": 133.9961759082218, "grad_norm": 0.453146755695343, "learning_rate": 2.2542410176526535e-05, "loss": 0.0101, "step": 210240 }, { "epoch": 134.0025493945188, "grad_norm": 0.12240495532751083, "learning_rate": 2.253780420290403e-05, "loss": 0.0041, "step": 210250 }, { "epoch": 134.00892288081582, "grad_norm": 0.43351635336875916, "learning_rate": 2.2533198562973097e-05, "loss": 0.004, "step": 210260 }, { "epoch": 134.01529636711282, "grad_norm": 0.13185563683509827, "learning_rate": 2.2528593256789755e-05, "loss": 0.0089, "step": 210270 }, { "epoch": 134.0216698534098, "grad_norm": 0.16643491387367249, "learning_rate": 2.252398828440992e-05, "loss": 0.0128, "step": 210280 }, { "epoch": 134.0280433397068, "grad_norm": 0.17559486627578735, "learning_rate": 2.251938364588961e-05, "loss": 0.0108, "step": 210290 }, { "epoch": 134.03441682600382, "grad_norm": 0.13750587403774261, "learning_rate": 2.251477934128469e-05, "loss": 0.0058, "step": 210300 }, { "epoch": 134.04079031230083, "grad_norm": 0.24808110296726227, "learning_rate": 2.2510175370651177e-05, "loss": 0.0105, "step": 210310 }, { "epoch": 134.04716379859784, "grad_norm": 0.24411945044994354, "learning_rate": 2.2505571734044984e-05, "loss": 0.0092, "step": 210320 }, { "epoch": 134.05353728489484, "grad_norm": 0.7124010324478149, "learning_rate": 2.2500968431522056e-05, "loss": 0.012, "step": 210330 }, { "epoch": 134.05991077119185, "grad_norm": 0.2935021221637726, "learning_rate": 2.2496365463138308e-05, "loss": 0.0051, "step": 210340 }, { "epoch": 134.06628425748886, "grad_norm": 0.1871609091758728, "learning_rate": 2.24917628289497e-05, "loss": 0.0085, "step": 210350 }, { "epoch": 134.07265774378584, "grad_norm": 0.147378072142601, "learning_rate": 2.248716052901215e-05, "loss": 0.0075, "step": 210360 }, { "epoch": 134.07903123008285, "grad_norm": 0.190736323595047, "learning_rate": 2.248255856338158e-05, "loss": 0.0094, "step": 210370 }, { "epoch": 134.08540471637986, "grad_norm": 0.1356031596660614, "learning_rate": 2.247795693211388e-05, "loss": 0.0059, "step": 210380 }, { "epoch": 134.09177820267686, "grad_norm": 0.43770015239715576, "learning_rate": 2.2473355635265008e-05, "loss": 0.0108, "step": 210390 }, { "epoch": 134.09815168897387, "grad_norm": 0.13097649812698364, "learning_rate": 2.2468754672890852e-05, "loss": 0.0063, "step": 210400 }, { "epoch": 134.10452517527088, "grad_norm": 0.1088082417845726, "learning_rate": 2.2464154045047325e-05, "loss": 0.0054, "step": 210410 }, { "epoch": 134.1108986615679, "grad_norm": 0.17975163459777832, "learning_rate": 2.2459553751790302e-05, "loss": 0.0043, "step": 210420 }, { "epoch": 134.1172721478649, "grad_norm": 0.1607961356639862, "learning_rate": 2.2454953793175725e-05, "loss": 0.0052, "step": 210430 }, { "epoch": 134.12364563416187, "grad_norm": 0.15270192921161652, "learning_rate": 2.2450354169259462e-05, "loss": 0.0077, "step": 210440 }, { "epoch": 134.13001912045888, "grad_norm": 0.20407156646251678, "learning_rate": 2.2445754880097412e-05, "loss": 0.006, "step": 210450 }, { "epoch": 134.1363926067559, "grad_norm": 0.09692882746458054, "learning_rate": 2.2441155925745437e-05, "loss": 0.0146, "step": 210460 }, { "epoch": 134.1427660930529, "grad_norm": 0.4712417423725128, "learning_rate": 2.2436557306259453e-05, "loss": 0.0111, "step": 210470 }, { "epoch": 134.1491395793499, "grad_norm": 0.1341615617275238, "learning_rate": 2.2431959021695325e-05, "loss": 0.0173, "step": 210480 }, { "epoch": 134.15551306564691, "grad_norm": 0.3448566794395447, "learning_rate": 2.2427361072108923e-05, "loss": 0.0107, "step": 210490 }, { "epoch": 134.16188655194392, "grad_norm": 0.30381837487220764, "learning_rate": 2.2422763457556112e-05, "loss": 0.0052, "step": 210500 }, { "epoch": 134.16826003824093, "grad_norm": 0.21424727141857147, "learning_rate": 2.2418166178092748e-05, "loss": 0.0192, "step": 210510 }, { "epoch": 134.1746335245379, "grad_norm": 0.15316540002822876, "learning_rate": 2.2413569233774727e-05, "loss": 0.0088, "step": 210520 }, { "epoch": 134.18100701083492, "grad_norm": 0.1952027678489685, "learning_rate": 2.2408972624657883e-05, "loss": 0.0127, "step": 210530 }, { "epoch": 134.18738049713193, "grad_norm": 0.40057408809661865, "learning_rate": 2.2404376350798075e-05, "loss": 0.0104, "step": 210540 }, { "epoch": 134.19375398342893, "grad_norm": 0.5672199130058289, "learning_rate": 2.2399780412251126e-05, "loss": 0.0151, "step": 210550 }, { "epoch": 134.20012746972594, "grad_norm": 0.21957193315029144, "learning_rate": 2.2395184809072923e-05, "loss": 0.0063, "step": 210560 }, { "epoch": 134.20650095602295, "grad_norm": 0.16344794631004333, "learning_rate": 2.2390589541319284e-05, "loss": 0.0073, "step": 210570 }, { "epoch": 134.21287444231996, "grad_norm": 0.15911363065242767, "learning_rate": 2.2385994609046056e-05, "loss": 0.0036, "step": 210580 }, { "epoch": 134.21924792861697, "grad_norm": 0.18829819560050964, "learning_rate": 2.2381400012309044e-05, "loss": 0.0073, "step": 210590 }, { "epoch": 134.22562141491395, "grad_norm": 0.25720247626304626, "learning_rate": 2.237680575116412e-05, "loss": 0.0097, "step": 210600 }, { "epoch": 134.23199490121095, "grad_norm": 0.24584287405014038, "learning_rate": 2.2372211825667082e-05, "loss": 0.0052, "step": 210610 }, { "epoch": 134.23836838750796, "grad_norm": 0.13489261269569397, "learning_rate": 2.236761823587376e-05, "loss": 0.0073, "step": 210620 }, { "epoch": 134.24474187380497, "grad_norm": 0.10012228786945343, "learning_rate": 2.2363024981839947e-05, "loss": 0.0077, "step": 210630 }, { "epoch": 134.25111536010198, "grad_norm": 0.3901284635066986, "learning_rate": 2.235843206362151e-05, "loss": 0.0113, "step": 210640 }, { "epoch": 134.25748884639899, "grad_norm": 0.33367785811424255, "learning_rate": 2.235383948127418e-05, "loss": 0.0104, "step": 210650 }, { "epoch": 134.263862332696, "grad_norm": 0.20280121266841888, "learning_rate": 2.2349247234853825e-05, "loss": 0.012, "step": 210660 }, { "epoch": 134.270235818993, "grad_norm": 0.2010260969400406, "learning_rate": 2.2344655324416204e-05, "loss": 0.0065, "step": 210670 }, { "epoch": 134.27660930528998, "grad_norm": 0.20480984449386597, "learning_rate": 2.234006375001717e-05, "loss": 0.0052, "step": 210680 }, { "epoch": 134.282982791587, "grad_norm": 0.254938006401062, "learning_rate": 2.2335472511712434e-05, "loss": 0.0082, "step": 210690 }, { "epoch": 134.289356277884, "grad_norm": 0.17401942610740662, "learning_rate": 2.2330881609557853e-05, "loss": 0.0135, "step": 210700 }, { "epoch": 134.295729764181, "grad_norm": 0.3407895565032959, "learning_rate": 2.232629104360916e-05, "loss": 0.0188, "step": 210710 }, { "epoch": 134.302103250478, "grad_norm": 0.20335036516189575, "learning_rate": 2.2321700813922202e-05, "loss": 0.0107, "step": 210720 }, { "epoch": 134.30847673677502, "grad_norm": 0.24819743633270264, "learning_rate": 2.231711092055267e-05, "loss": 0.0093, "step": 210730 }, { "epoch": 134.31485022307203, "grad_norm": 0.39702072739601135, "learning_rate": 2.2312521363556398e-05, "loss": 0.0069, "step": 210740 }, { "epoch": 134.32122370936904, "grad_norm": 0.36989593505859375, "learning_rate": 2.230793214298913e-05, "loss": 0.0124, "step": 210750 }, { "epoch": 134.32759719566602, "grad_norm": 0.2982294261455536, "learning_rate": 2.2303343258906633e-05, "loss": 0.0074, "step": 210760 }, { "epoch": 134.33397068196302, "grad_norm": 0.30951061844825745, "learning_rate": 2.229875471136465e-05, "loss": 0.0129, "step": 210770 }, { "epoch": 134.34034416826003, "grad_norm": 0.15971703827381134, "learning_rate": 2.229416650041897e-05, "loss": 0.0081, "step": 210780 }, { "epoch": 134.34671765455704, "grad_norm": 0.2453572303056717, "learning_rate": 2.2289578626125322e-05, "loss": 0.0093, "step": 210790 }, { "epoch": 134.35309114085405, "grad_norm": 0.3462895154953003, "learning_rate": 2.228499108853946e-05, "loss": 0.005, "step": 210800 }, { "epoch": 134.35946462715106, "grad_norm": 0.39686232805252075, "learning_rate": 2.22804038877171e-05, "loss": 0.0102, "step": 210810 }, { "epoch": 134.36583811344806, "grad_norm": 0.34944719076156616, "learning_rate": 2.2275817023714047e-05, "loss": 0.0074, "step": 210820 }, { "epoch": 134.37221159974507, "grad_norm": 0.6007533669471741, "learning_rate": 2.2271230496585954e-05, "loss": 0.006, "step": 210830 }, { "epoch": 134.37858508604205, "grad_norm": 0.12282685190439224, "learning_rate": 2.2266644306388607e-05, "loss": 0.0082, "step": 210840 }, { "epoch": 134.38495857233906, "grad_norm": 0.07289889454841614, "learning_rate": 2.2262058453177704e-05, "loss": 0.0117, "step": 210850 }, { "epoch": 134.39133205863607, "grad_norm": 0.209162637591362, "learning_rate": 2.225747293700901e-05, "loss": 0.0042, "step": 210860 }, { "epoch": 134.39770554493307, "grad_norm": 0.27758467197418213, "learning_rate": 2.225288775793818e-05, "loss": 0.0065, "step": 210870 }, { "epoch": 134.40407903123008, "grad_norm": 0.16966426372528076, "learning_rate": 2.224830291602098e-05, "loss": 0.0092, "step": 210880 }, { "epoch": 134.4104525175271, "grad_norm": 0.26019057631492615, "learning_rate": 2.2243718411313098e-05, "loss": 0.0053, "step": 210890 }, { "epoch": 134.4168260038241, "grad_norm": 0.20146602392196655, "learning_rate": 2.2239134243870247e-05, "loss": 0.0054, "step": 210900 }, { "epoch": 134.4231994901211, "grad_norm": 0.3462006449699402, "learning_rate": 2.2234550413748106e-05, "loss": 0.0045, "step": 210910 }, { "epoch": 134.42957297641811, "grad_norm": 0.29210877418518066, "learning_rate": 2.222996692100241e-05, "loss": 0.0061, "step": 210920 }, { "epoch": 134.4359464627151, "grad_norm": 0.1909114271402359, "learning_rate": 2.2225383765688835e-05, "loss": 0.0063, "step": 210930 }, { "epoch": 134.4423199490121, "grad_norm": 0.4576992988586426, "learning_rate": 2.2220800947863074e-05, "loss": 0.0127, "step": 210940 }, { "epoch": 134.4486934353091, "grad_norm": 0.3411645293235779, "learning_rate": 2.2216218467580786e-05, "loss": 0.0087, "step": 210950 }, { "epoch": 134.45506692160612, "grad_norm": 0.35631993412971497, "learning_rate": 2.2211636324897716e-05, "loss": 0.0054, "step": 210960 }, { "epoch": 134.46144040790313, "grad_norm": 0.15816296637058258, "learning_rate": 2.2207054519869463e-05, "loss": 0.0049, "step": 210970 }, { "epoch": 134.46781389420013, "grad_norm": 0.28029578924179077, "learning_rate": 2.220247305255176e-05, "loss": 0.0057, "step": 210980 }, { "epoch": 134.47418738049714, "grad_norm": 0.19956247508525848, "learning_rate": 2.219789192300023e-05, "loss": 0.0099, "step": 210990 }, { "epoch": 134.48056086679415, "grad_norm": 0.22313429415225983, "learning_rate": 2.2193311131270605e-05, "loss": 0.01, "step": 211000 }, { "epoch": 134.48693435309113, "grad_norm": 0.5808376669883728, "learning_rate": 2.2188730677418467e-05, "loss": 0.0104, "step": 211010 }, { "epoch": 134.49330783938814, "grad_norm": 0.14204061031341553, "learning_rate": 2.2184150561499523e-05, "loss": 0.0058, "step": 211020 }, { "epoch": 134.49968132568515, "grad_norm": 0.1267332285642624, "learning_rate": 2.2179570783569398e-05, "loss": 0.0085, "step": 211030 }, { "epoch": 134.50605481198215, "grad_norm": 0.5718514323234558, "learning_rate": 2.2174991343683794e-05, "loss": 0.0097, "step": 211040 }, { "epoch": 134.51242829827916, "grad_norm": 0.26839250326156616, "learning_rate": 2.217041224189827e-05, "loss": 0.0061, "step": 211050 }, { "epoch": 134.51880178457617, "grad_norm": 0.17503976821899414, "learning_rate": 2.216583347826854e-05, "loss": 0.0044, "step": 211060 }, { "epoch": 134.52517527087318, "grad_norm": 0.08636821806430817, "learning_rate": 2.2161255052850212e-05, "loss": 0.0144, "step": 211070 }, { "epoch": 134.53154875717019, "grad_norm": 0.17307239770889282, "learning_rate": 2.2156676965698914e-05, "loss": 0.0131, "step": 211080 }, { "epoch": 134.53792224346716, "grad_norm": 0.4154033660888672, "learning_rate": 2.2152099216870265e-05, "loss": 0.0147, "step": 211090 }, { "epoch": 134.54429572976417, "grad_norm": 0.44530752301216125, "learning_rate": 2.2147521806419918e-05, "loss": 0.0111, "step": 211100 }, { "epoch": 134.55066921606118, "grad_norm": 0.33803001046180725, "learning_rate": 2.2142944734403486e-05, "loss": 0.0183, "step": 211110 }, { "epoch": 134.5570427023582, "grad_norm": 0.284896582365036, "learning_rate": 2.2138368000876568e-05, "loss": 0.0048, "step": 211120 }, { "epoch": 134.5634161886552, "grad_norm": 0.3404543101787567, "learning_rate": 2.2133791605894773e-05, "loss": 0.0053, "step": 211130 }, { "epoch": 134.5697896749522, "grad_norm": 0.5210334658622742, "learning_rate": 2.212921554951374e-05, "loss": 0.0182, "step": 211140 }, { "epoch": 134.5761631612492, "grad_norm": 0.30585920810699463, "learning_rate": 2.212463983178904e-05, "loss": 0.0073, "step": 211150 }, { "epoch": 134.58253664754622, "grad_norm": 0.4875738322734833, "learning_rate": 2.2120064452776295e-05, "loss": 0.0115, "step": 211160 }, { "epoch": 134.5889101338432, "grad_norm": 0.227563738822937, "learning_rate": 2.211548941253107e-05, "loss": 0.0142, "step": 211170 }, { "epoch": 134.5952836201402, "grad_norm": 0.2789423167705536, "learning_rate": 2.2110914711108995e-05, "loss": 0.0064, "step": 211180 }, { "epoch": 134.60165710643722, "grad_norm": 0.13202789425849915, "learning_rate": 2.2106340348565637e-05, "loss": 0.0062, "step": 211190 }, { "epoch": 134.60803059273422, "grad_norm": 0.3562984764575958, "learning_rate": 2.210176632495658e-05, "loss": 0.0055, "step": 211200 }, { "epoch": 134.61440407903123, "grad_norm": 0.17401845753192902, "learning_rate": 2.20971926403374e-05, "loss": 0.011, "step": 211210 }, { "epoch": 134.62077756532824, "grad_norm": 0.3664025664329529, "learning_rate": 2.2092619294763655e-05, "loss": 0.0069, "step": 211220 }, { "epoch": 134.62715105162525, "grad_norm": 0.2732543647289276, "learning_rate": 2.208804628829096e-05, "loss": 0.0037, "step": 211230 }, { "epoch": 134.63352453792226, "grad_norm": 0.4632740318775177, "learning_rate": 2.208347362097485e-05, "loss": 0.0196, "step": 211240 }, { "epoch": 134.63989802421924, "grad_norm": 0.2158307433128357, "learning_rate": 2.207890129287089e-05, "loss": 0.0232, "step": 211250 }, { "epoch": 134.64627151051624, "grad_norm": 0.5104135274887085, "learning_rate": 2.207432930403463e-05, "loss": 0.0111, "step": 211260 }, { "epoch": 134.65264499681325, "grad_norm": 0.2780402898788452, "learning_rate": 2.206975765452165e-05, "loss": 0.005, "step": 211270 }, { "epoch": 134.65901848311026, "grad_norm": 0.37218034267425537, "learning_rate": 2.2065186344387484e-05, "loss": 0.0076, "step": 211280 }, { "epoch": 134.66539196940727, "grad_norm": 0.1586906760931015, "learning_rate": 2.206061537368768e-05, "loss": 0.006, "step": 211290 }, { "epoch": 134.67176545570427, "grad_norm": 0.47443345189094543, "learning_rate": 2.205604474247776e-05, "loss": 0.0111, "step": 211300 }, { "epoch": 134.67813894200128, "grad_norm": 0.23930025100708008, "learning_rate": 2.2051474450813302e-05, "loss": 0.0051, "step": 211310 }, { "epoch": 134.6845124282983, "grad_norm": 0.1945635974407196, "learning_rate": 2.2046904498749816e-05, "loss": 0.005, "step": 211320 }, { "epoch": 134.69088591459527, "grad_norm": 0.4199976921081543, "learning_rate": 2.2042334886342832e-05, "loss": 0.0129, "step": 211330 }, { "epoch": 134.69725940089228, "grad_norm": 0.21287508308887482, "learning_rate": 2.2037765613647847e-05, "loss": 0.007, "step": 211340 }, { "epoch": 134.7036328871893, "grad_norm": 0.4822179079055786, "learning_rate": 2.203319668072044e-05, "loss": 0.0129, "step": 211350 }, { "epoch": 134.7100063734863, "grad_norm": 0.2833027243614197, "learning_rate": 2.20286280876161e-05, "loss": 0.0066, "step": 211360 }, { "epoch": 134.7163798597833, "grad_norm": 0.13288819789886475, "learning_rate": 2.2024059834390333e-05, "loss": 0.0073, "step": 211370 }, { "epoch": 134.7227533460803, "grad_norm": 0.22996635735034943, "learning_rate": 2.2019491921098633e-05, "loss": 0.0078, "step": 211380 }, { "epoch": 134.72912683237732, "grad_norm": 0.14441357553005219, "learning_rate": 2.201492434779656e-05, "loss": 0.0054, "step": 211390 }, { "epoch": 134.73550031867433, "grad_norm": 0.1747668832540512, "learning_rate": 2.2010357114539547e-05, "loss": 0.006, "step": 211400 }, { "epoch": 134.7418738049713, "grad_norm": 0.21042746305465698, "learning_rate": 2.2005790221383138e-05, "loss": 0.016, "step": 211410 }, { "epoch": 134.7482472912683, "grad_norm": 0.23353299498558044, "learning_rate": 2.200122366838279e-05, "loss": 0.0097, "step": 211420 }, { "epoch": 134.75462077756532, "grad_norm": 0.18179288506507874, "learning_rate": 2.1996657455594043e-05, "loss": 0.0084, "step": 211430 }, { "epoch": 134.76099426386233, "grad_norm": 0.1281246542930603, "learning_rate": 2.1992091583072315e-05, "loss": 0.0103, "step": 211440 }, { "epoch": 134.76736775015934, "grad_norm": 0.24861060082912445, "learning_rate": 2.1987526050873138e-05, "loss": 0.008, "step": 211450 }, { "epoch": 134.77374123645635, "grad_norm": 0.3975648880004883, "learning_rate": 2.1982960859051965e-05, "loss": 0.0098, "step": 211460 }, { "epoch": 134.78011472275335, "grad_norm": 0.1934133768081665, "learning_rate": 2.197839600766427e-05, "loss": 0.0093, "step": 211470 }, { "epoch": 134.78648820905036, "grad_norm": 0.20758500695228577, "learning_rate": 2.1973831496765502e-05, "loss": 0.0141, "step": 211480 }, { "epoch": 134.79286169534734, "grad_norm": 0.2142581194639206, "learning_rate": 2.1969267326411163e-05, "loss": 0.0104, "step": 211490 }, { "epoch": 134.79923518164435, "grad_norm": 0.3010588586330414, "learning_rate": 2.1964703496656686e-05, "loss": 0.0069, "step": 211500 }, { "epoch": 134.80560866794136, "grad_norm": 0.11897353827953339, "learning_rate": 2.196014000755754e-05, "loss": 0.0095, "step": 211510 }, { "epoch": 134.81198215423836, "grad_norm": 0.20221944153308868, "learning_rate": 2.1955576859169142e-05, "loss": 0.0055, "step": 211520 }, { "epoch": 134.81835564053537, "grad_norm": 0.497948557138443, "learning_rate": 2.1951014051547e-05, "loss": 0.0107, "step": 211530 }, { "epoch": 134.82472912683238, "grad_norm": 0.289578914642334, "learning_rate": 2.1946451584746485e-05, "loss": 0.0062, "step": 211540 }, { "epoch": 134.8311026131294, "grad_norm": 0.13647302985191345, "learning_rate": 2.1941889458823088e-05, "loss": 0.0067, "step": 211550 }, { "epoch": 134.8374760994264, "grad_norm": 0.14777643978595734, "learning_rate": 2.1937327673832207e-05, "loss": 0.0189, "step": 211560 }, { "epoch": 134.8438495857234, "grad_norm": 0.20364797115325928, "learning_rate": 2.1932766229829327e-05, "loss": 0.0079, "step": 211570 }, { "epoch": 134.85022307202038, "grad_norm": 0.220905140042305, "learning_rate": 2.19282051268698e-05, "loss": 0.0106, "step": 211580 }, { "epoch": 134.8565965583174, "grad_norm": 0.12299200147390366, "learning_rate": 2.1923644365009104e-05, "loss": 0.007, "step": 211590 }, { "epoch": 134.8629700446144, "grad_norm": 0.39609310030937195, "learning_rate": 2.191908394430262e-05, "loss": 0.0099, "step": 211600 }, { "epoch": 134.8693435309114, "grad_norm": 0.223884716629982, "learning_rate": 2.1914523864805814e-05, "loss": 0.0047, "step": 211610 }, { "epoch": 134.87571701720842, "grad_norm": 0.41045689582824707, "learning_rate": 2.1909964126574024e-05, "loss": 0.0063, "step": 211620 }, { "epoch": 134.88209050350542, "grad_norm": 0.12326137721538544, "learning_rate": 2.190540472966271e-05, "loss": 0.0046, "step": 211630 }, { "epoch": 134.88846398980243, "grad_norm": 0.1739259660243988, "learning_rate": 2.1900845674127253e-05, "loss": 0.0076, "step": 211640 }, { "epoch": 134.89483747609944, "grad_norm": 0.147774800658226, "learning_rate": 2.189628696002305e-05, "loss": 0.0091, "step": 211650 }, { "epoch": 134.90121096239642, "grad_norm": 0.42832204699516296, "learning_rate": 2.189172858740548e-05, "loss": 0.0144, "step": 211660 }, { "epoch": 134.90758444869343, "grad_norm": 0.22030484676361084, "learning_rate": 2.188717055632996e-05, "loss": 0.0105, "step": 211670 }, { "epoch": 134.91395793499044, "grad_norm": 0.25332707166671753, "learning_rate": 2.1882612866851864e-05, "loss": 0.0095, "step": 211680 }, { "epoch": 134.92033142128744, "grad_norm": 0.489096462726593, "learning_rate": 2.187805551902657e-05, "loss": 0.0056, "step": 211690 }, { "epoch": 134.92670490758445, "grad_norm": 0.37678277492523193, "learning_rate": 2.187349851290943e-05, "loss": 0.0065, "step": 211700 }, { "epoch": 134.93307839388146, "grad_norm": 0.14802716672420502, "learning_rate": 2.186894184855588e-05, "loss": 0.0071, "step": 211710 }, { "epoch": 134.93945188017847, "grad_norm": 0.571172833442688, "learning_rate": 2.186438552602121e-05, "loss": 0.0061, "step": 211720 }, { "epoch": 134.94582536647547, "grad_norm": 0.12006988376379013, "learning_rate": 2.1859829545360838e-05, "loss": 0.008, "step": 211730 }, { "epoch": 134.95219885277245, "grad_norm": 0.41379931569099426, "learning_rate": 2.1855273906630087e-05, "loss": 0.0051, "step": 211740 }, { "epoch": 134.95857233906946, "grad_norm": 0.15348254144191742, "learning_rate": 2.1850718609884362e-05, "loss": 0.0064, "step": 211750 }, { "epoch": 134.96494582536647, "grad_norm": 0.14955303072929382, "learning_rate": 2.1846163655178954e-05, "loss": 0.0057, "step": 211760 }, { "epoch": 134.97131931166348, "grad_norm": 0.32789069414138794, "learning_rate": 2.1841609042569256e-05, "loss": 0.0118, "step": 211770 }, { "epoch": 134.9776927979605, "grad_norm": 0.2096279263496399, "learning_rate": 2.1837054772110595e-05, "loss": 0.0127, "step": 211780 }, { "epoch": 134.9840662842575, "grad_norm": 0.741711437702179, "learning_rate": 2.1832500843858306e-05, "loss": 0.017, "step": 211790 }, { "epoch": 134.9904397705545, "grad_norm": 0.4261437654495239, "learning_rate": 2.182794725786771e-05, "loss": 0.007, "step": 211800 }, { "epoch": 134.9968132568515, "grad_norm": 0.5088883638381958, "learning_rate": 2.182339401419417e-05, "loss": 0.0138, "step": 211810 }, { "epoch": 135.0031867431485, "grad_norm": 0.343952476978302, "learning_rate": 2.1818841112892996e-05, "loss": 0.0096, "step": 211820 }, { "epoch": 135.0095602294455, "grad_norm": 0.3865063190460205, "learning_rate": 2.1814288554019513e-05, "loss": 0.0063, "step": 211830 }, { "epoch": 135.0159337157425, "grad_norm": 0.17202940583229065, "learning_rate": 2.180973633762901e-05, "loss": 0.0122, "step": 211840 }, { "epoch": 135.0223072020395, "grad_norm": 0.21199725568294525, "learning_rate": 2.180518446377685e-05, "loss": 0.0049, "step": 211850 }, { "epoch": 135.02868068833652, "grad_norm": 0.44800689816474915, "learning_rate": 2.180063293251832e-05, "loss": 0.0123, "step": 211860 }, { "epoch": 135.03505417463353, "grad_norm": 0.1354338377714157, "learning_rate": 2.1796081743908727e-05, "loss": 0.0107, "step": 211870 }, { "epoch": 135.04142766093054, "grad_norm": 0.27365365624427795, "learning_rate": 2.179153089800335e-05, "loss": 0.012, "step": 211880 }, { "epoch": 135.04780114722755, "grad_norm": 0.1217835322022438, "learning_rate": 2.178698039485752e-05, "loss": 0.0069, "step": 211890 }, { "epoch": 135.05417463352453, "grad_norm": 0.3476071059703827, "learning_rate": 2.178243023452652e-05, "loss": 0.0093, "step": 211900 }, { "epoch": 135.06054811982153, "grad_norm": 0.26551002264022827, "learning_rate": 2.1777880417065625e-05, "loss": 0.0095, "step": 211910 }, { "epoch": 135.06692160611854, "grad_norm": 0.4020671546459198, "learning_rate": 2.1773330942530124e-05, "loss": 0.016, "step": 211920 }, { "epoch": 135.07329509241555, "grad_norm": 0.2801743745803833, "learning_rate": 2.1768781810975312e-05, "loss": 0.0068, "step": 211930 }, { "epoch": 135.07966857871256, "grad_norm": 0.18695494532585144, "learning_rate": 2.176423302245646e-05, "loss": 0.006, "step": 211940 }, { "epoch": 135.08604206500956, "grad_norm": 0.2884143888950348, "learning_rate": 2.1759684577028833e-05, "loss": 0.0064, "step": 211950 }, { "epoch": 135.09241555130657, "grad_norm": 0.22526617348194122, "learning_rate": 2.17551364747477e-05, "loss": 0.0117, "step": 211960 }, { "epoch": 135.09878903760358, "grad_norm": 0.38338503241539, "learning_rate": 2.1750588715668314e-05, "loss": 0.0068, "step": 211970 }, { "epoch": 135.10516252390056, "grad_norm": 0.28445637226104736, "learning_rate": 2.1746041299845964e-05, "loss": 0.007, "step": 211980 }, { "epoch": 135.11153601019757, "grad_norm": 0.25995105504989624, "learning_rate": 2.1741494227335883e-05, "loss": 0.0064, "step": 211990 }, { "epoch": 135.11790949649458, "grad_norm": 0.12934431433677673, "learning_rate": 2.173694749819333e-05, "loss": 0.0081, "step": 212000 }, { "epoch": 135.12428298279158, "grad_norm": 0.26911118626594543, "learning_rate": 2.173240111247354e-05, "loss": 0.0052, "step": 212010 }, { "epoch": 135.1306564690886, "grad_norm": 0.31595638394355774, "learning_rate": 2.1727855070231773e-05, "loss": 0.0056, "step": 212020 }, { "epoch": 135.1370299553856, "grad_norm": 0.18661841750144958, "learning_rate": 2.1723309371523264e-05, "loss": 0.0177, "step": 212030 }, { "epoch": 135.1434034416826, "grad_norm": 0.20691196620464325, "learning_rate": 2.171876401640324e-05, "loss": 0.0069, "step": 212040 }, { "epoch": 135.14977692797962, "grad_norm": 0.5235318541526794, "learning_rate": 2.171421900492692e-05, "loss": 0.0081, "step": 212050 }, { "epoch": 135.1561504142766, "grad_norm": 0.3175920844078064, "learning_rate": 2.1709674337149564e-05, "loss": 0.0186, "step": 212060 }, { "epoch": 135.1625239005736, "grad_norm": 0.46879518032073975, "learning_rate": 2.170513001312638e-05, "loss": 0.0104, "step": 212070 }, { "epoch": 135.1688973868706, "grad_norm": 0.21721284091472626, "learning_rate": 2.170058603291257e-05, "loss": 0.0079, "step": 212080 }, { "epoch": 135.17527087316762, "grad_norm": 0.1568175107240677, "learning_rate": 2.169604239656335e-05, "loss": 0.0085, "step": 212090 }, { "epoch": 135.18164435946463, "grad_norm": 0.22517365217208862, "learning_rate": 2.169149910413397e-05, "loss": 0.0039, "step": 212100 }, { "epoch": 135.18801784576164, "grad_norm": 0.26193705201148987, "learning_rate": 2.168695615567957e-05, "loss": 0.0183, "step": 212110 }, { "epoch": 135.19439133205864, "grad_norm": 0.2578570246696472, "learning_rate": 2.1682413551255405e-05, "loss": 0.0069, "step": 212120 }, { "epoch": 135.20076481835565, "grad_norm": 0.16327111423015594, "learning_rate": 2.1677871290916622e-05, "loss": 0.008, "step": 212130 }, { "epoch": 135.20713830465266, "grad_norm": 0.4605794847011566, "learning_rate": 2.1673329374718493e-05, "loss": 0.0122, "step": 212140 }, { "epoch": 135.21351179094964, "grad_norm": 0.34715983271598816, "learning_rate": 2.166878780271611e-05, "loss": 0.0069, "step": 212150 }, { "epoch": 135.21988527724665, "grad_norm": 0.07942170649766922, "learning_rate": 2.1664246574964732e-05, "loss": 0.0043, "step": 212160 }, { "epoch": 135.22625876354365, "grad_norm": 0.17364414036273956, "learning_rate": 2.1659705691519506e-05, "loss": 0.0104, "step": 212170 }, { "epoch": 135.23263224984066, "grad_norm": 0.5600479245185852, "learning_rate": 2.1655165152435613e-05, "loss": 0.0094, "step": 212180 }, { "epoch": 135.23900573613767, "grad_norm": 0.16968543827533722, "learning_rate": 2.1650624957768212e-05, "loss": 0.0057, "step": 212190 }, { "epoch": 135.24537922243468, "grad_norm": 0.42828094959259033, "learning_rate": 2.1646085107572495e-05, "loss": 0.0044, "step": 212200 }, { "epoch": 135.2517527087317, "grad_norm": 0.3854599595069885, "learning_rate": 2.1641545601903624e-05, "loss": 0.0076, "step": 212210 }, { "epoch": 135.2581261950287, "grad_norm": 0.18513348698616028, "learning_rate": 2.1637006440816738e-05, "loss": 0.0079, "step": 212220 }, { "epoch": 135.26449968132567, "grad_norm": 0.3479118049144745, "learning_rate": 2.1632467624366993e-05, "loss": 0.0215, "step": 212230 }, { "epoch": 135.27087316762268, "grad_norm": 0.42604342103004456, "learning_rate": 2.162792915260956e-05, "loss": 0.0065, "step": 212240 }, { "epoch": 135.2772466539197, "grad_norm": 0.3940887749195099, "learning_rate": 2.162339102559958e-05, "loss": 0.0155, "step": 212250 }, { "epoch": 135.2836201402167, "grad_norm": 0.19024859368801117, "learning_rate": 2.1618853243392195e-05, "loss": 0.0052, "step": 212260 }, { "epoch": 135.2899936265137, "grad_norm": 0.3000456690788269, "learning_rate": 2.1614315806042518e-05, "loss": 0.0158, "step": 212270 }, { "epoch": 135.2963671128107, "grad_norm": 0.14191168546676636, "learning_rate": 2.160977871360574e-05, "loss": 0.0057, "step": 212280 }, { "epoch": 135.30274059910772, "grad_norm": 0.17346419394016266, "learning_rate": 2.160524196613692e-05, "loss": 0.0064, "step": 212290 }, { "epoch": 135.30911408540473, "grad_norm": 0.15001586079597473, "learning_rate": 2.1600705563691236e-05, "loss": 0.0049, "step": 212300 }, { "epoch": 135.3154875717017, "grad_norm": 0.1571478545665741, "learning_rate": 2.1596169506323777e-05, "loss": 0.0048, "step": 212310 }, { "epoch": 135.32186105799872, "grad_norm": 0.12339052557945251, "learning_rate": 2.159163379408971e-05, "loss": 0.007, "step": 212320 }, { "epoch": 135.32823454429573, "grad_norm": 0.2581508755683899, "learning_rate": 2.1587098427044078e-05, "loss": 0.0059, "step": 212330 }, { "epoch": 135.33460803059273, "grad_norm": 0.48464351892471313, "learning_rate": 2.1582563405242047e-05, "loss": 0.0078, "step": 212340 }, { "epoch": 135.34098151688974, "grad_norm": 0.12201835215091705, "learning_rate": 2.1578028728738697e-05, "loss": 0.0159, "step": 212350 }, { "epoch": 135.34735500318675, "grad_norm": 0.1260623037815094, "learning_rate": 2.1573494397589134e-05, "loss": 0.007, "step": 212360 }, { "epoch": 135.35372848948376, "grad_norm": 0.25506848096847534, "learning_rate": 2.156896041184844e-05, "loss": 0.0059, "step": 212370 }, { "epoch": 135.36010197578076, "grad_norm": 0.21629399061203003, "learning_rate": 2.156442677157173e-05, "loss": 0.0043, "step": 212380 }, { "epoch": 135.36647546207774, "grad_norm": 0.0760716050863266, "learning_rate": 2.1559893476814085e-05, "loss": 0.0076, "step": 212390 }, { "epoch": 135.37284894837475, "grad_norm": 0.3333495557308197, "learning_rate": 2.1555360527630593e-05, "loss": 0.007, "step": 212400 }, { "epoch": 135.37922243467176, "grad_norm": 0.12179385870695114, "learning_rate": 2.15508279240763e-05, "loss": 0.0068, "step": 212410 }, { "epoch": 135.38559592096877, "grad_norm": 0.3156780004501343, "learning_rate": 2.154629566620635e-05, "loss": 0.0082, "step": 212420 }, { "epoch": 135.39196940726578, "grad_norm": 0.19014763832092285, "learning_rate": 2.1541763754075733e-05, "loss": 0.0133, "step": 212430 }, { "epoch": 135.39834289356278, "grad_norm": 0.3375912010669708, "learning_rate": 2.1537232187739576e-05, "loss": 0.0087, "step": 212440 }, { "epoch": 135.4047163798598, "grad_norm": 0.19370487332344055, "learning_rate": 2.15327009672529e-05, "loss": 0.01, "step": 212450 }, { "epoch": 135.4110898661568, "grad_norm": 0.1493513137102127, "learning_rate": 2.1528170092670825e-05, "loss": 0.0078, "step": 212460 }, { "epoch": 135.41746335245378, "grad_norm": 0.6569372415542603, "learning_rate": 2.1523639564048325e-05, "loss": 0.012, "step": 212470 }, { "epoch": 135.4238368387508, "grad_norm": 0.23086106777191162, "learning_rate": 2.1519109381440512e-05, "loss": 0.006, "step": 212480 }, { "epoch": 135.4302103250478, "grad_norm": 0.20073993504047394, "learning_rate": 2.1514579544902392e-05, "loss": 0.005, "step": 212490 }, { "epoch": 135.4365838113448, "grad_norm": 0.22186900675296783, "learning_rate": 2.1510050054489063e-05, "loss": 0.0048, "step": 212500 }, { "epoch": 135.4429572976418, "grad_norm": 0.2886353135108948, "learning_rate": 2.1505520910255484e-05, "loss": 0.0046, "step": 212510 }, { "epoch": 135.44933078393882, "grad_norm": 0.12588785588741302, "learning_rate": 2.1500992112256752e-05, "loss": 0.0097, "step": 212520 }, { "epoch": 135.45570427023583, "grad_norm": 0.3593915104866028, "learning_rate": 2.1496463660547865e-05, "loss": 0.0083, "step": 212530 }, { "epoch": 135.46207775653284, "grad_norm": 0.18087786436080933, "learning_rate": 2.149193555518387e-05, "loss": 0.0098, "step": 212540 }, { "epoch": 135.46845124282981, "grad_norm": 0.33163660764694214, "learning_rate": 2.1487407796219745e-05, "loss": 0.007, "step": 212550 }, { "epoch": 135.47482472912682, "grad_norm": 0.15079915523529053, "learning_rate": 2.1482880383710552e-05, "loss": 0.015, "step": 212560 }, { "epoch": 135.48119821542383, "grad_norm": 0.5074668526649475, "learning_rate": 2.147835331771129e-05, "loss": 0.0137, "step": 212570 }, { "epoch": 135.48757170172084, "grad_norm": 0.27219390869140625, "learning_rate": 2.147382659827697e-05, "loss": 0.0135, "step": 212580 }, { "epoch": 135.49394518801785, "grad_norm": 0.2886810004711151, "learning_rate": 2.1469300225462567e-05, "loss": 0.0109, "step": 212590 }, { "epoch": 135.50031867431485, "grad_norm": 0.2376013696193695, "learning_rate": 2.1464774199323123e-05, "loss": 0.0104, "step": 212600 }, { "epoch": 135.50669216061186, "grad_norm": 0.34352415800094604, "learning_rate": 2.146024851991361e-05, "loss": 0.0064, "step": 212610 }, { "epoch": 135.51306564690887, "grad_norm": 0.4872688949108124, "learning_rate": 2.145572318728903e-05, "loss": 0.0119, "step": 212620 }, { "epoch": 135.51943913320585, "grad_norm": 0.3728056252002716, "learning_rate": 2.145119820150434e-05, "loss": 0.0057, "step": 212630 }, { "epoch": 135.52581261950286, "grad_norm": 0.2764419615268707, "learning_rate": 2.1446673562614573e-05, "loss": 0.0077, "step": 212640 }, { "epoch": 135.53218610579987, "grad_norm": 0.3749164044857025, "learning_rate": 2.144214927067467e-05, "loss": 0.0259, "step": 212650 }, { "epoch": 135.53855959209687, "grad_norm": 0.20033389329910278, "learning_rate": 2.1437625325739623e-05, "loss": 0.0072, "step": 212660 }, { "epoch": 135.54493307839388, "grad_norm": 0.17414845526218414, "learning_rate": 2.1433101727864396e-05, "loss": 0.0045, "step": 212670 }, { "epoch": 135.5513065646909, "grad_norm": 0.1688651293516159, "learning_rate": 2.142857847710393e-05, "loss": 0.0085, "step": 212680 }, { "epoch": 135.5576800509879, "grad_norm": 0.19054532051086426, "learning_rate": 2.1424055573513236e-05, "loss": 0.0116, "step": 212690 }, { "epoch": 135.5640535372849, "grad_norm": 0.14326991140842438, "learning_rate": 2.1419533017147248e-05, "loss": 0.0094, "step": 212700 }, { "epoch": 135.57042702358189, "grad_norm": 0.4654884934425354, "learning_rate": 2.1415010808060915e-05, "loss": 0.0176, "step": 212710 }, { "epoch": 135.5768005098789, "grad_norm": 0.2108447551727295, "learning_rate": 2.141048894630917e-05, "loss": 0.0064, "step": 212720 }, { "epoch": 135.5831739961759, "grad_norm": 0.37011808156967163, "learning_rate": 2.1405967431947e-05, "loss": 0.0051, "step": 212730 }, { "epoch": 135.5895474824729, "grad_norm": 0.34583258628845215, "learning_rate": 2.140144626502932e-05, "loss": 0.0086, "step": 212740 }, { "epoch": 135.59592096876992, "grad_norm": 0.4000834822654724, "learning_rate": 2.1396925445611067e-05, "loss": 0.0061, "step": 212750 }, { "epoch": 135.60229445506693, "grad_norm": 0.13910187780857086, "learning_rate": 2.139240497374716e-05, "loss": 0.0037, "step": 212760 }, { "epoch": 135.60866794136393, "grad_norm": 0.1321016401052475, "learning_rate": 2.1387884849492563e-05, "loss": 0.004, "step": 212770 }, { "epoch": 135.61504142766094, "grad_norm": 0.31886717677116394, "learning_rate": 2.1383365072902174e-05, "loss": 0.0083, "step": 212780 }, { "epoch": 135.62141491395795, "grad_norm": 0.14390034973621368, "learning_rate": 2.137884564403092e-05, "loss": 0.0255, "step": 212790 }, { "epoch": 135.62778840025493, "grad_norm": 0.24213743209838867, "learning_rate": 2.13743265629337e-05, "loss": 0.0085, "step": 212800 }, { "epoch": 135.63416188655194, "grad_norm": 0.13267230987548828, "learning_rate": 2.1369807829665457e-05, "loss": 0.0085, "step": 212810 }, { "epoch": 135.64053537284894, "grad_norm": 0.09220178425312042, "learning_rate": 2.1365289444281073e-05, "loss": 0.0084, "step": 212820 }, { "epoch": 135.64690885914595, "grad_norm": 0.2293420433998108, "learning_rate": 2.1360771406835465e-05, "loss": 0.011, "step": 212830 }, { "epoch": 135.65328234544296, "grad_norm": 0.18800315260887146, "learning_rate": 2.13562537173835e-05, "loss": 0.009, "step": 212840 }, { "epoch": 135.65965583173997, "grad_norm": 0.3201417326927185, "learning_rate": 2.135173637598014e-05, "loss": 0.0155, "step": 212850 }, { "epoch": 135.66602931803698, "grad_norm": 0.1581268012523651, "learning_rate": 2.134721938268019e-05, "loss": 0.0089, "step": 212860 }, { "epoch": 135.67240280433398, "grad_norm": 0.2719762921333313, "learning_rate": 2.13427027375386e-05, "loss": 0.0073, "step": 212870 }, { "epoch": 135.67877629063096, "grad_norm": 0.20226597785949707, "learning_rate": 2.133818644061021e-05, "loss": 0.0052, "step": 212880 }, { "epoch": 135.68514977692797, "grad_norm": 0.10212904214859009, "learning_rate": 2.1333670491949946e-05, "loss": 0.0157, "step": 212890 }, { "epoch": 135.69152326322498, "grad_norm": 0.14631590247154236, "learning_rate": 2.132915489161262e-05, "loss": 0.0074, "step": 212900 }, { "epoch": 135.697896749522, "grad_norm": 0.16071784496307373, "learning_rate": 2.1324639639653148e-05, "loss": 0.0063, "step": 212910 }, { "epoch": 135.704270235819, "grad_norm": 0.4736170470714569, "learning_rate": 2.132012473612638e-05, "loss": 0.0085, "step": 212920 }, { "epoch": 135.710643722116, "grad_norm": 0.40461069345474243, "learning_rate": 2.1315610181087174e-05, "loss": 0.0049, "step": 212930 }, { "epoch": 135.717017208413, "grad_norm": 0.20723459124565125, "learning_rate": 2.1311095974590366e-05, "loss": 0.0076, "step": 212940 }, { "epoch": 135.72339069471002, "grad_norm": 0.19738644361495972, "learning_rate": 2.1306582116690855e-05, "loss": 0.0051, "step": 212950 }, { "epoch": 135.729764181007, "grad_norm": 0.16554057598114014, "learning_rate": 2.1302068607443454e-05, "loss": 0.0074, "step": 212960 }, { "epoch": 135.736137667304, "grad_norm": 0.1698579639196396, "learning_rate": 2.1297555446903017e-05, "loss": 0.0068, "step": 212970 }, { "epoch": 135.74251115360101, "grad_norm": 0.12326567620038986, "learning_rate": 2.1293042635124365e-05, "loss": 0.0039, "step": 212980 }, { "epoch": 135.74888463989802, "grad_norm": 0.3507702648639679, "learning_rate": 2.1288530172162384e-05, "loss": 0.0072, "step": 212990 }, { "epoch": 135.75525812619503, "grad_norm": 0.3645098805427551, "learning_rate": 2.128401805807183e-05, "loss": 0.018, "step": 213000 }, { "epoch": 135.76163161249204, "grad_norm": 0.1429995745420456, "learning_rate": 2.1279506292907593e-05, "loss": 0.0135, "step": 213010 }, { "epoch": 135.76800509878905, "grad_norm": 0.14482153952121735, "learning_rate": 2.1274994876724446e-05, "loss": 0.0067, "step": 213020 }, { "epoch": 135.77437858508605, "grad_norm": 0.14793351292610168, "learning_rate": 2.1270483809577275e-05, "loss": 0.0036, "step": 213030 }, { "epoch": 135.78075207138303, "grad_norm": 0.2649751603603363, "learning_rate": 2.1265973091520814e-05, "loss": 0.0083, "step": 213040 }, { "epoch": 135.78712555768004, "grad_norm": 0.20104902982711792, "learning_rate": 2.1261462722609922e-05, "loss": 0.0055, "step": 213050 }, { "epoch": 135.79349904397705, "grad_norm": 0.16276291012763977, "learning_rate": 2.1256952702899397e-05, "loss": 0.01, "step": 213060 }, { "epoch": 135.79987253027406, "grad_norm": 0.24855856597423553, "learning_rate": 2.125244303244403e-05, "loss": 0.0249, "step": 213070 }, { "epoch": 135.80624601657107, "grad_norm": 0.2934115529060364, "learning_rate": 2.1247933711298608e-05, "loss": 0.0083, "step": 213080 }, { "epoch": 135.81261950286807, "grad_norm": 0.32084575295448303, "learning_rate": 2.1243424739517954e-05, "loss": 0.0115, "step": 213090 }, { "epoch": 135.81899298916508, "grad_norm": 0.4747924208641052, "learning_rate": 2.123891611715684e-05, "loss": 0.0111, "step": 213100 }, { "epoch": 135.8253664754621, "grad_norm": 0.24367476999759674, "learning_rate": 2.1234407844270055e-05, "loss": 0.0075, "step": 213110 }, { "epoch": 135.83173996175907, "grad_norm": 0.27824899554252625, "learning_rate": 2.122989992091235e-05, "loss": 0.0057, "step": 213120 }, { "epoch": 135.83811344805608, "grad_norm": 0.63909512758255, "learning_rate": 2.1225392347138544e-05, "loss": 0.0179, "step": 213130 }, { "epoch": 135.84448693435309, "grad_norm": 0.3811812996864319, "learning_rate": 2.122088512300339e-05, "loss": 0.0125, "step": 213140 }, { "epoch": 135.8508604206501, "grad_norm": 0.20173919200897217, "learning_rate": 2.121637824856165e-05, "loss": 0.011, "step": 213150 }, { "epoch": 135.8572339069471, "grad_norm": 0.15838345885276794, "learning_rate": 2.1211871723868075e-05, "loss": 0.0057, "step": 213160 }, { "epoch": 135.8636073932441, "grad_norm": 0.2711038887500763, "learning_rate": 2.120736554897748e-05, "loss": 0.0043, "step": 213170 }, { "epoch": 135.86998087954112, "grad_norm": 0.29591137170791626, "learning_rate": 2.1202859723944536e-05, "loss": 0.0052, "step": 213180 }, { "epoch": 135.87635436583813, "grad_norm": 0.13398239016532898, "learning_rate": 2.1198354248824054e-05, "loss": 0.0104, "step": 213190 }, { "epoch": 135.8827278521351, "grad_norm": 0.5928769111633301, "learning_rate": 2.1193849123670745e-05, "loss": 0.0084, "step": 213200 }, { "epoch": 135.8891013384321, "grad_norm": 0.19763579964637756, "learning_rate": 2.11893443485394e-05, "loss": 0.0056, "step": 213210 }, { "epoch": 135.89547482472912, "grad_norm": 0.13865207135677338, "learning_rate": 2.1184839923484695e-05, "loss": 0.0067, "step": 213220 }, { "epoch": 135.90184831102613, "grad_norm": 0.4520626366138458, "learning_rate": 2.11803358485614e-05, "loss": 0.008, "step": 213230 }, { "epoch": 135.90822179732314, "grad_norm": 0.26298758387565613, "learning_rate": 2.1175832123824242e-05, "loss": 0.0064, "step": 213240 }, { "epoch": 135.91459528362014, "grad_norm": 0.16152207553386688, "learning_rate": 2.117132874932793e-05, "loss": 0.0096, "step": 213250 }, { "epoch": 135.92096876991715, "grad_norm": 0.2570919692516327, "learning_rate": 2.1166825725127176e-05, "loss": 0.0113, "step": 213260 }, { "epoch": 135.92734225621416, "grad_norm": 0.2801651656627655, "learning_rate": 2.1162323051276734e-05, "loss": 0.0108, "step": 213270 }, { "epoch": 135.93371574251114, "grad_norm": 0.33410516381263733, "learning_rate": 2.1157820727831296e-05, "loss": 0.0123, "step": 213280 }, { "epoch": 135.94008922880815, "grad_norm": 0.22175276279449463, "learning_rate": 2.1153318754845563e-05, "loss": 0.0116, "step": 213290 }, { "epoch": 135.94646271510516, "grad_norm": 0.30354589223861694, "learning_rate": 2.1148817132374227e-05, "loss": 0.0053, "step": 213300 }, { "epoch": 135.95283620140216, "grad_norm": 0.24541723728179932, "learning_rate": 2.114431586047202e-05, "loss": 0.0113, "step": 213310 }, { "epoch": 135.95920968769917, "grad_norm": 0.3029096722602844, "learning_rate": 2.113981493919362e-05, "loss": 0.0101, "step": 213320 }, { "epoch": 135.96558317399618, "grad_norm": 0.5495606660842896, "learning_rate": 2.1135314368593718e-05, "loss": 0.0069, "step": 213330 }, { "epoch": 135.9719566602932, "grad_norm": 0.16166569292545319, "learning_rate": 2.113081414872698e-05, "loss": 0.006, "step": 213340 }, { "epoch": 135.9783301465902, "grad_norm": 0.2461986541748047, "learning_rate": 2.1126314279648124e-05, "loss": 0.0082, "step": 213350 }, { "epoch": 135.98470363288718, "grad_norm": 0.0883156806230545, "learning_rate": 2.112181476141181e-05, "loss": 0.0074, "step": 213360 }, { "epoch": 135.99107711918418, "grad_norm": 0.18657638132572174, "learning_rate": 2.1117315594072717e-05, "loss": 0.0152, "step": 213370 }, { "epoch": 135.9974506054812, "grad_norm": 0.4201924204826355, "learning_rate": 2.1112816777685505e-05, "loss": 0.007, "step": 213380 }, { "epoch": 136.0038240917782, "grad_norm": 0.1302538812160492, "learning_rate": 2.1108318312304827e-05, "loss": 0.0063, "step": 213390 }, { "epoch": 136.0101975780752, "grad_norm": 0.10396663099527359, "learning_rate": 2.110382019798538e-05, "loss": 0.0081, "step": 213400 }, { "epoch": 136.01657106437222, "grad_norm": 0.17184054851531982, "learning_rate": 2.109932243478179e-05, "loss": 0.0053, "step": 213410 }, { "epoch": 136.02294455066922, "grad_norm": 0.11072632670402527, "learning_rate": 2.1094825022748732e-05, "loss": 0.0049, "step": 213420 }, { "epoch": 136.02931803696623, "grad_norm": 0.6501540541648865, "learning_rate": 2.109032796194082e-05, "loss": 0.0132, "step": 213430 }, { "epoch": 136.03569152326324, "grad_norm": 0.5307726860046387, "learning_rate": 2.108583125241273e-05, "loss": 0.0065, "step": 213440 }, { "epoch": 136.04206500956022, "grad_norm": 0.49301254749298096, "learning_rate": 2.1081334894219097e-05, "loss": 0.01, "step": 213450 }, { "epoch": 136.04843849585723, "grad_norm": 0.24696093797683716, "learning_rate": 2.1076838887414547e-05, "loss": 0.0095, "step": 213460 }, { "epoch": 136.05481198215423, "grad_norm": 0.48420456051826477, "learning_rate": 2.1072343232053694e-05, "loss": 0.0074, "step": 213470 }, { "epoch": 136.06118546845124, "grad_norm": 0.2565116286277771, "learning_rate": 2.10678479281912e-05, "loss": 0.0079, "step": 213480 }, { "epoch": 136.06755895474825, "grad_norm": 0.26978838443756104, "learning_rate": 2.1063352975881666e-05, "loss": 0.0102, "step": 213490 }, { "epoch": 136.07393244104526, "grad_norm": 0.11817779392004013, "learning_rate": 2.1058858375179723e-05, "loss": 0.0057, "step": 213500 }, { "epoch": 136.08030592734227, "grad_norm": 0.18701745569705963, "learning_rate": 2.105436412613995e-05, "loss": 0.0074, "step": 213510 }, { "epoch": 136.08667941363927, "grad_norm": 0.26516616344451904, "learning_rate": 2.1049870228817004e-05, "loss": 0.006, "step": 213520 }, { "epoch": 136.09305289993625, "grad_norm": 0.1392369419336319, "learning_rate": 2.1045376683265467e-05, "loss": 0.0062, "step": 213530 }, { "epoch": 136.09942638623326, "grad_norm": 0.15188799798488617, "learning_rate": 2.104088348953994e-05, "loss": 0.0085, "step": 213540 }, { "epoch": 136.10579987253027, "grad_norm": 0.3183937966823578, "learning_rate": 2.103639064769501e-05, "loss": 0.0097, "step": 213550 }, { "epoch": 136.11217335882728, "grad_norm": 0.23545844852924347, "learning_rate": 2.1031898157785306e-05, "loss": 0.0038, "step": 213560 }, { "epoch": 136.11854684512429, "grad_norm": 0.471622496843338, "learning_rate": 2.102740601986536e-05, "loss": 0.009, "step": 213570 }, { "epoch": 136.1249203314213, "grad_norm": 0.332684189081192, "learning_rate": 2.10229142339898e-05, "loss": 0.005, "step": 213580 }, { "epoch": 136.1312938177183, "grad_norm": 0.4104095995426178, "learning_rate": 2.1018422800213173e-05, "loss": 0.0098, "step": 213590 }, { "epoch": 136.1376673040153, "grad_norm": 0.7375832200050354, "learning_rate": 2.101393171859011e-05, "loss": 0.008, "step": 213600 }, { "epoch": 136.1440407903123, "grad_norm": 0.2254675030708313, "learning_rate": 2.1009440989175107e-05, "loss": 0.0048, "step": 213610 }, { "epoch": 136.1504142766093, "grad_norm": 0.2113407999277115, "learning_rate": 2.1004950612022788e-05, "loss": 0.0074, "step": 213620 }, { "epoch": 136.1567877629063, "grad_norm": 0.13531656563282013, "learning_rate": 2.1000460587187692e-05, "loss": 0.0147, "step": 213630 }, { "epoch": 136.1631612492033, "grad_norm": 0.2048831433057785, "learning_rate": 2.0995970914724376e-05, "loss": 0.0148, "step": 213640 }, { "epoch": 136.16953473550032, "grad_norm": 0.10962924361228943, "learning_rate": 2.099148159468739e-05, "loss": 0.0062, "step": 213650 }, { "epoch": 136.17590822179733, "grad_norm": 0.40644192695617676, "learning_rate": 2.09869926271313e-05, "loss": 0.0101, "step": 213660 }, { "epoch": 136.18228170809434, "grad_norm": 0.19095663726329803, "learning_rate": 2.098250401211065e-05, "loss": 0.0058, "step": 213670 }, { "epoch": 136.18865519439134, "grad_norm": 0.11047486960887909, "learning_rate": 2.097801574967997e-05, "loss": 0.0048, "step": 213680 }, { "epoch": 136.19502868068832, "grad_norm": 0.36260491609573364, "learning_rate": 2.097352783989378e-05, "loss": 0.0091, "step": 213690 }, { "epoch": 136.20140216698533, "grad_norm": 0.22654445469379425, "learning_rate": 2.096904028280665e-05, "loss": 0.0048, "step": 213700 }, { "epoch": 136.20777565328234, "grad_norm": 0.2336762696504593, "learning_rate": 2.0964553078473094e-05, "loss": 0.0083, "step": 213710 }, { "epoch": 136.21414913957935, "grad_norm": 0.3674794137477875, "learning_rate": 2.0960066226947628e-05, "loss": 0.0101, "step": 213720 }, { "epoch": 136.22052262587636, "grad_norm": 0.5140051245689392, "learning_rate": 2.0955579728284765e-05, "loss": 0.0119, "step": 213730 }, { "epoch": 136.22689611217336, "grad_norm": 0.13995486497879028, "learning_rate": 2.0951093582539057e-05, "loss": 0.0068, "step": 213740 }, { "epoch": 136.23326959847037, "grad_norm": 0.2652316689491272, "learning_rate": 2.0946607789764954e-05, "loss": 0.0125, "step": 213750 }, { "epoch": 136.23964308476738, "grad_norm": 0.21143247187137604, "learning_rate": 2.0942122350017023e-05, "loss": 0.0067, "step": 213760 }, { "epoch": 136.24601657106436, "grad_norm": 0.10480085760354996, "learning_rate": 2.0937637263349718e-05, "loss": 0.0075, "step": 213770 }, { "epoch": 136.25239005736137, "grad_norm": 0.23682402074337006, "learning_rate": 2.0933152529817594e-05, "loss": 0.0049, "step": 213780 }, { "epoch": 136.25876354365838, "grad_norm": 0.14742425084114075, "learning_rate": 2.092866814947508e-05, "loss": 0.0077, "step": 213790 }, { "epoch": 136.26513702995538, "grad_norm": 0.31505948305130005, "learning_rate": 2.0924184122376706e-05, "loss": 0.0127, "step": 213800 }, { "epoch": 136.2715105162524, "grad_norm": 0.3149089813232422, "learning_rate": 2.0919700448576952e-05, "loss": 0.0062, "step": 213810 }, { "epoch": 136.2778840025494, "grad_norm": 0.1662469357252121, "learning_rate": 2.0915217128130294e-05, "loss": 0.0088, "step": 213820 }, { "epoch": 136.2842574888464, "grad_norm": 0.305391788482666, "learning_rate": 2.0910734161091188e-05, "loss": 0.0201, "step": 213830 }, { "epoch": 136.29063097514342, "grad_norm": 0.32638218998908997, "learning_rate": 2.0906251547514144e-05, "loss": 0.0059, "step": 213840 }, { "epoch": 136.2970044614404, "grad_norm": 0.15127144753932953, "learning_rate": 2.0901769287453616e-05, "loss": 0.0053, "step": 213850 }, { "epoch": 136.3033779477374, "grad_norm": 0.1368200033903122, "learning_rate": 2.0897287380964064e-05, "loss": 0.0176, "step": 213860 }, { "epoch": 136.3097514340344, "grad_norm": 0.2882726788520813, "learning_rate": 2.0892805828099933e-05, "loss": 0.0087, "step": 213870 }, { "epoch": 136.31612492033142, "grad_norm": 0.27897438406944275, "learning_rate": 2.0888324628915723e-05, "loss": 0.0079, "step": 213880 }, { "epoch": 136.32249840662843, "grad_norm": 0.30435091257095337, "learning_rate": 2.0883843783465817e-05, "loss": 0.0193, "step": 213890 }, { "epoch": 136.32887189292543, "grad_norm": 0.08660493046045303, "learning_rate": 2.0879363291804728e-05, "loss": 0.0122, "step": 213900 }, { "epoch": 136.33524537922244, "grad_norm": 0.16569320857524872, "learning_rate": 2.087488315398684e-05, "loss": 0.006, "step": 213910 }, { "epoch": 136.34161886551945, "grad_norm": 0.1915445327758789, "learning_rate": 2.087040337006666e-05, "loss": 0.0058, "step": 213920 }, { "epoch": 136.34799235181643, "grad_norm": 0.08381646871566772, "learning_rate": 2.0865923940098547e-05, "loss": 0.0049, "step": 213930 }, { "epoch": 136.35436583811344, "grad_norm": 0.48380205035209656, "learning_rate": 2.0861444864136986e-05, "loss": 0.011, "step": 213940 }, { "epoch": 136.36073932441045, "grad_norm": 0.38144272565841675, "learning_rate": 2.085696614223638e-05, "loss": 0.0055, "step": 213950 }, { "epoch": 136.36711281070745, "grad_norm": 0.34462448954582214, "learning_rate": 2.0852487774451146e-05, "loss": 0.0116, "step": 213960 }, { "epoch": 136.37348629700446, "grad_norm": 0.44125086069107056, "learning_rate": 2.0848009760835692e-05, "loss": 0.0101, "step": 213970 }, { "epoch": 136.37985978330147, "grad_norm": 0.3154473900794983, "learning_rate": 2.0843532101444456e-05, "loss": 0.0071, "step": 213980 }, { "epoch": 136.38623326959848, "grad_norm": 0.3568072021007538, "learning_rate": 2.083905479633184e-05, "loss": 0.0067, "step": 213990 }, { "epoch": 136.39260675589549, "grad_norm": 0.29297301173210144, "learning_rate": 2.0834577845552238e-05, "loss": 0.0069, "step": 214000 }, { "epoch": 136.39898024219247, "grad_norm": 0.18737384676933289, "learning_rate": 2.0830101249160032e-05, "loss": 0.0086, "step": 214010 }, { "epoch": 136.40535372848947, "grad_norm": 0.24415071308612823, "learning_rate": 2.082562500720966e-05, "loss": 0.0086, "step": 214020 }, { "epoch": 136.41172721478648, "grad_norm": 0.14238430559635162, "learning_rate": 2.0821149119755484e-05, "loss": 0.0044, "step": 214030 }, { "epoch": 136.4181007010835, "grad_norm": 0.15700280666351318, "learning_rate": 2.0816673586851896e-05, "loss": 0.0087, "step": 214040 }, { "epoch": 136.4244741873805, "grad_norm": 0.17414222657680511, "learning_rate": 2.0812198408553262e-05, "loss": 0.0065, "step": 214050 }, { "epoch": 136.4308476736775, "grad_norm": 0.29874080419540405, "learning_rate": 2.0807723584913998e-05, "loss": 0.0106, "step": 214060 }, { "epoch": 136.4372211599745, "grad_norm": 0.09836408495903015, "learning_rate": 2.0803249115988445e-05, "loss": 0.007, "step": 214070 }, { "epoch": 136.44359464627152, "grad_norm": 0.34556227922439575, "learning_rate": 2.0798775001830983e-05, "loss": 0.0066, "step": 214080 }, { "epoch": 136.44996813256853, "grad_norm": 0.18025055527687073, "learning_rate": 2.0794301242495963e-05, "loss": 0.0155, "step": 214090 }, { "epoch": 136.4563416188655, "grad_norm": 0.17312857508659363, "learning_rate": 2.0789827838037772e-05, "loss": 0.0067, "step": 214100 }, { "epoch": 136.46271510516252, "grad_norm": 0.1143730953335762, "learning_rate": 2.0785354788510753e-05, "loss": 0.0086, "step": 214110 }, { "epoch": 136.46908859145952, "grad_norm": 0.2479863464832306, "learning_rate": 2.0780882093969256e-05, "loss": 0.0065, "step": 214120 }, { "epoch": 136.47546207775653, "grad_norm": 0.2708296477794647, "learning_rate": 2.077640975446763e-05, "loss": 0.0049, "step": 214130 }, { "epoch": 136.48183556405354, "grad_norm": 0.3996363878250122, "learning_rate": 2.0771937770060195e-05, "loss": 0.0081, "step": 214140 }, { "epoch": 136.48820905035055, "grad_norm": 0.1458505094051361, "learning_rate": 2.0767466140801332e-05, "loss": 0.0087, "step": 214150 }, { "epoch": 136.49458253664756, "grad_norm": 0.37913718819618225, "learning_rate": 2.0762994866745357e-05, "loss": 0.019, "step": 214160 }, { "epoch": 136.50095602294456, "grad_norm": 0.18216155469417572, "learning_rate": 2.0758523947946597e-05, "loss": 0.0085, "step": 214170 }, { "epoch": 136.50732950924154, "grad_norm": 0.22557097673416138, "learning_rate": 2.0754053384459364e-05, "loss": 0.0057, "step": 214180 }, { "epoch": 136.51370299553855, "grad_norm": 0.3659454882144928, "learning_rate": 2.074958317633801e-05, "loss": 0.014, "step": 214190 }, { "epoch": 136.52007648183556, "grad_norm": 0.2237318605184555, "learning_rate": 2.074511332363684e-05, "loss": 0.0078, "step": 214200 }, { "epoch": 136.52644996813257, "grad_norm": 0.12356472760438919, "learning_rate": 2.0740643826410162e-05, "loss": 0.0039, "step": 214210 }, { "epoch": 136.53282345442958, "grad_norm": 0.15497785806655884, "learning_rate": 2.073617468471227e-05, "loss": 0.0101, "step": 214220 }, { "epoch": 136.53919694072658, "grad_norm": 0.12978628277778625, "learning_rate": 2.0731705898597505e-05, "loss": 0.011, "step": 214230 }, { "epoch": 136.5455704270236, "grad_norm": 0.2079639583826065, "learning_rate": 2.0727237468120146e-05, "loss": 0.0124, "step": 214240 }, { "epoch": 136.5519439133206, "grad_norm": 0.26640909910202026, "learning_rate": 2.072276939333449e-05, "loss": 0.0085, "step": 214250 }, { "epoch": 136.55831739961758, "grad_norm": 0.28812164068222046, "learning_rate": 2.071830167429481e-05, "loss": 0.0066, "step": 214260 }, { "epoch": 136.5646908859146, "grad_norm": 0.18223504722118378, "learning_rate": 2.0713834311055446e-05, "loss": 0.0066, "step": 214270 }, { "epoch": 136.5710643722116, "grad_norm": 0.23256836831569672, "learning_rate": 2.0709367303670607e-05, "loss": 0.0048, "step": 214280 }, { "epoch": 136.5774378585086, "grad_norm": 0.14232970774173737, "learning_rate": 2.0704900652194636e-05, "loss": 0.0053, "step": 214290 }, { "epoch": 136.5838113448056, "grad_norm": 0.19669640064239502, "learning_rate": 2.0700434356681758e-05, "loss": 0.0094, "step": 214300 }, { "epoch": 136.59018483110262, "grad_norm": 0.18829213082790375, "learning_rate": 2.06959684171863e-05, "loss": 0.0083, "step": 214310 }, { "epoch": 136.59655831739963, "grad_norm": 0.14604702591896057, "learning_rate": 2.069150283376246e-05, "loss": 0.0337, "step": 214320 }, { "epoch": 136.60293180369663, "grad_norm": 0.3785502314567566, "learning_rate": 2.0687037606464553e-05, "loss": 0.0107, "step": 214330 }, { "epoch": 136.6093052899936, "grad_norm": 0.14911101758480072, "learning_rate": 2.0682572735346795e-05, "loss": 0.0061, "step": 214340 }, { "epoch": 136.61567877629062, "grad_norm": 0.3832978904247284, "learning_rate": 2.0678108220463492e-05, "loss": 0.0111, "step": 214350 }, { "epoch": 136.62205226258763, "grad_norm": 0.28972962498664856, "learning_rate": 2.0673644061868823e-05, "loss": 0.0057, "step": 214360 }, { "epoch": 136.62842574888464, "grad_norm": 0.3557979464530945, "learning_rate": 2.066918025961709e-05, "loss": 0.0101, "step": 214370 }, { "epoch": 136.63479923518165, "grad_norm": 0.2938990592956543, "learning_rate": 2.066471681376251e-05, "loss": 0.0151, "step": 214380 }, { "epoch": 136.64117272147865, "grad_norm": 0.5680206418037415, "learning_rate": 2.0660253724359314e-05, "loss": 0.0154, "step": 214390 }, { "epoch": 136.64754620777566, "grad_norm": 0.8622773289680481, "learning_rate": 2.0655790991461717e-05, "loss": 0.0055, "step": 214400 }, { "epoch": 136.65391969407267, "grad_norm": 0.35901886224746704, "learning_rate": 2.0651328615123983e-05, "loss": 0.0069, "step": 214410 }, { "epoch": 136.66029318036965, "grad_norm": 0.4831865429878235, "learning_rate": 2.064686659540032e-05, "loss": 0.0155, "step": 214420 }, { "epoch": 136.66666666666666, "grad_norm": 0.1644202023744583, "learning_rate": 2.064240493234494e-05, "loss": 0.004, "step": 214430 }, { "epoch": 136.67304015296367, "grad_norm": 0.3256523609161377, "learning_rate": 2.063794362601204e-05, "loss": 0.0118, "step": 214440 }, { "epoch": 136.67941363926067, "grad_norm": 0.6449172496795654, "learning_rate": 2.0633482676455883e-05, "loss": 0.0137, "step": 214450 }, { "epoch": 136.68578712555768, "grad_norm": 0.14071586728096008, "learning_rate": 2.0629022083730605e-05, "loss": 0.0029, "step": 214460 }, { "epoch": 136.6921606118547, "grad_norm": 0.14959993958473206, "learning_rate": 2.0624561847890455e-05, "loss": 0.0041, "step": 214470 }, { "epoch": 136.6985340981517, "grad_norm": 0.21583238244056702, "learning_rate": 2.0620101968989597e-05, "loss": 0.0093, "step": 214480 }, { "epoch": 136.7049075844487, "grad_norm": 0.3415376842021942, "learning_rate": 2.0615642447082273e-05, "loss": 0.0123, "step": 214490 }, { "epoch": 136.71128107074568, "grad_norm": 0.3900931179523468, "learning_rate": 2.0611183282222603e-05, "loss": 0.0081, "step": 214500 }, { "epoch": 136.7176545570427, "grad_norm": 0.24180026352405548, "learning_rate": 2.0606724474464817e-05, "loss": 0.0065, "step": 214510 }, { "epoch": 136.7240280433397, "grad_norm": 0.1273355484008789, "learning_rate": 2.0602266023863088e-05, "loss": 0.0052, "step": 214520 }, { "epoch": 136.7304015296367, "grad_norm": 0.15706554055213928, "learning_rate": 2.0597807930471575e-05, "loss": 0.0108, "step": 214530 }, { "epoch": 136.73677501593372, "grad_norm": 0.17727452516555786, "learning_rate": 2.059335019434444e-05, "loss": 0.0081, "step": 214540 }, { "epoch": 136.74314850223072, "grad_norm": 0.16430386900901794, "learning_rate": 2.058889281553589e-05, "loss": 0.0084, "step": 214550 }, { "epoch": 136.74952198852773, "grad_norm": 0.21821972727775574, "learning_rate": 2.0584435794100054e-05, "loss": 0.0075, "step": 214560 }, { "epoch": 136.75589547482474, "grad_norm": 0.32589447498321533, "learning_rate": 2.05799791300911e-05, "loss": 0.0062, "step": 214570 }, { "epoch": 136.76226896112172, "grad_norm": 0.1501391977071762, "learning_rate": 2.0575522823563155e-05, "loss": 0.0049, "step": 214580 }, { "epoch": 136.76864244741873, "grad_norm": 0.19366590678691864, "learning_rate": 2.0571066874570423e-05, "loss": 0.0072, "step": 214590 }, { "epoch": 136.77501593371574, "grad_norm": 0.1863098293542862, "learning_rate": 2.056661128316698e-05, "loss": 0.0063, "step": 214600 }, { "epoch": 136.78138942001274, "grad_norm": 0.23261062800884247, "learning_rate": 2.0562156049407023e-05, "loss": 0.0063, "step": 214610 }, { "epoch": 136.78776290630975, "grad_norm": 0.10359074175357819, "learning_rate": 2.055770117334464e-05, "loss": 0.0105, "step": 214620 }, { "epoch": 136.79413639260676, "grad_norm": 0.11596986651420593, "learning_rate": 2.0553246655034018e-05, "loss": 0.0093, "step": 214630 }, { "epoch": 136.80050987890377, "grad_norm": 0.5954976677894592, "learning_rate": 2.0548792494529223e-05, "loss": 0.0128, "step": 214640 }, { "epoch": 136.80688336520078, "grad_norm": 0.14024420082569122, "learning_rate": 2.0544338691884422e-05, "loss": 0.0161, "step": 214650 }, { "epoch": 136.81325685149778, "grad_norm": 0.3221265971660614, "learning_rate": 2.05398852471537e-05, "loss": 0.0073, "step": 214660 }, { "epoch": 136.81963033779476, "grad_norm": 0.22511564195156097, "learning_rate": 2.0535432160391222e-05, "loss": 0.0081, "step": 214670 }, { "epoch": 136.82600382409177, "grad_norm": 0.13945631682872772, "learning_rate": 2.0530979431651025e-05, "loss": 0.0125, "step": 214680 }, { "epoch": 136.83237731038878, "grad_norm": 0.12314986437559128, "learning_rate": 2.052652706098727e-05, "loss": 0.0045, "step": 214690 }, { "epoch": 136.8387507966858, "grad_norm": 0.39638280868530273, "learning_rate": 2.0522075048454038e-05, "loss": 0.0041, "step": 214700 }, { "epoch": 136.8451242829828, "grad_norm": 0.507981538772583, "learning_rate": 2.0517623394105427e-05, "loss": 0.0076, "step": 214710 }, { "epoch": 136.8514977692798, "grad_norm": 0.2415122240781784, "learning_rate": 2.051317209799551e-05, "loss": 0.0109, "step": 214720 }, { "epoch": 136.8578712555768, "grad_norm": 0.17225311696529388, "learning_rate": 2.0508721160178413e-05, "loss": 0.006, "step": 214730 }, { "epoch": 136.86424474187382, "grad_norm": 0.1127123311161995, "learning_rate": 2.0504270580708195e-05, "loss": 0.0081, "step": 214740 }, { "epoch": 136.8706182281708, "grad_norm": 0.2488560825586319, "learning_rate": 2.049982035963894e-05, "loss": 0.0112, "step": 214750 }, { "epoch": 136.8769917144678, "grad_norm": 0.1299217939376831, "learning_rate": 2.0495370497024695e-05, "loss": 0.0077, "step": 214760 }, { "epoch": 136.8833652007648, "grad_norm": 0.45304012298583984, "learning_rate": 2.0490920992919582e-05, "loss": 0.0067, "step": 214770 }, { "epoch": 136.88973868706182, "grad_norm": 0.16393931210041046, "learning_rate": 2.0486471847377637e-05, "loss": 0.0043, "step": 214780 }, { "epoch": 136.89611217335883, "grad_norm": 0.20457911491394043, "learning_rate": 2.0482023060452925e-05, "loss": 0.0063, "step": 214790 }, { "epoch": 136.90248565965584, "grad_norm": 0.23626191914081573, "learning_rate": 2.047757463219948e-05, "loss": 0.0074, "step": 214800 }, { "epoch": 136.90885914595285, "grad_norm": 0.461063414812088, "learning_rate": 2.0473126562671403e-05, "loss": 0.0099, "step": 214810 }, { "epoch": 136.91523263224985, "grad_norm": 0.33914920687675476, "learning_rate": 2.0468678851922712e-05, "loss": 0.0083, "step": 214820 }, { "epoch": 136.92160611854683, "grad_norm": 0.13905245065689087, "learning_rate": 2.0464231500007453e-05, "loss": 0.006, "step": 214830 }, { "epoch": 136.92797960484384, "grad_norm": 0.28175315260887146, "learning_rate": 2.0459784506979675e-05, "loss": 0.0069, "step": 214840 }, { "epoch": 136.93435309114085, "grad_norm": 0.42558473348617554, "learning_rate": 2.0455337872893383e-05, "loss": 0.0145, "step": 214850 }, { "epoch": 136.94072657743786, "grad_norm": 0.22245673835277557, "learning_rate": 2.0450891597802647e-05, "loss": 0.0084, "step": 214860 }, { "epoch": 136.94710006373487, "grad_norm": 0.6886193752288818, "learning_rate": 2.044644568176148e-05, "loss": 0.0166, "step": 214870 }, { "epoch": 136.95347355003187, "grad_norm": 0.2331005185842514, "learning_rate": 2.0442000124823896e-05, "loss": 0.0047, "step": 214880 }, { "epoch": 136.95984703632888, "grad_norm": 0.16700595617294312, "learning_rate": 2.0437554927043906e-05, "loss": 0.0118, "step": 214890 }, { "epoch": 136.9662205226259, "grad_norm": 0.31448015570640564, "learning_rate": 2.043311008847555e-05, "loss": 0.0084, "step": 214900 }, { "epoch": 136.97259400892287, "grad_norm": 0.3295111060142517, "learning_rate": 2.042866560917282e-05, "loss": 0.0106, "step": 214910 }, { "epoch": 136.97896749521988, "grad_norm": 0.15931065380573273, "learning_rate": 2.0424221489189727e-05, "loss": 0.0088, "step": 214920 }, { "epoch": 136.98534098151688, "grad_norm": 0.23004789650440216, "learning_rate": 2.0419777728580246e-05, "loss": 0.0035, "step": 214930 }, { "epoch": 136.9917144678139, "grad_norm": 0.0768420472741127, "learning_rate": 2.0415334327398418e-05, "loss": 0.0088, "step": 214940 }, { "epoch": 136.9980879541109, "grad_norm": 0.22034122049808502, "learning_rate": 2.0410891285698203e-05, "loss": 0.0056, "step": 214950 }, { "epoch": 137.0044614404079, "grad_norm": 0.10780317336320877, "learning_rate": 2.04064486035336e-05, "loss": 0.0053, "step": 214960 }, { "epoch": 137.01083492670492, "grad_norm": 0.1692921668291092, "learning_rate": 2.0402006280958567e-05, "loss": 0.0072, "step": 214970 }, { "epoch": 137.01720841300192, "grad_norm": 0.20485062897205353, "learning_rate": 2.0397564318027123e-05, "loss": 0.0042, "step": 214980 }, { "epoch": 137.0235818992989, "grad_norm": 0.5315746068954468, "learning_rate": 2.039312271479322e-05, "loss": 0.0133, "step": 214990 }, { "epoch": 137.0299553855959, "grad_norm": 0.19246551394462585, "learning_rate": 2.0388681471310834e-05, "loss": 0.0069, "step": 215000 }, { "epoch": 137.03632887189292, "grad_norm": 0.4767412543296814, "learning_rate": 2.03842405876339e-05, "loss": 0.0098, "step": 215010 }, { "epoch": 137.04270235818993, "grad_norm": 0.14428064227104187, "learning_rate": 2.0379800063816453e-05, "loss": 0.0045, "step": 215020 }, { "epoch": 137.04907584448694, "grad_norm": 0.15024906396865845, "learning_rate": 2.0375359899912356e-05, "loss": 0.0047, "step": 215030 }, { "epoch": 137.05544933078394, "grad_norm": 0.2372472882270813, "learning_rate": 2.037092009597562e-05, "loss": 0.0054, "step": 215040 }, { "epoch": 137.06182281708095, "grad_norm": 0.16086696088314056, "learning_rate": 2.036648065206017e-05, "loss": 0.0037, "step": 215050 }, { "epoch": 137.06819630337796, "grad_norm": 0.23183639347553253, "learning_rate": 2.0362041568219992e-05, "loss": 0.0093, "step": 215060 }, { "epoch": 137.07456978967494, "grad_norm": 0.4025565981864929, "learning_rate": 2.0357602844508954e-05, "loss": 0.0058, "step": 215070 }, { "epoch": 137.08094327597195, "grad_norm": 0.18822360038757324, "learning_rate": 2.035316448098105e-05, "loss": 0.013, "step": 215080 }, { "epoch": 137.08731676226896, "grad_norm": 0.38169190287590027, "learning_rate": 2.034872647769019e-05, "loss": 0.0133, "step": 215090 }, { "epoch": 137.09369024856596, "grad_norm": 0.21098847687244415, "learning_rate": 2.0344288834690296e-05, "loss": 0.0092, "step": 215100 }, { "epoch": 137.10006373486297, "grad_norm": 0.2817094624042511, "learning_rate": 2.0339851552035272e-05, "loss": 0.0115, "step": 215110 }, { "epoch": 137.10643722115998, "grad_norm": 0.21323652565479279, "learning_rate": 2.0335414629779076e-05, "loss": 0.0066, "step": 215120 }, { "epoch": 137.112810707457, "grad_norm": 0.09612549096345901, "learning_rate": 2.03309780679756e-05, "loss": 0.0042, "step": 215130 }, { "epoch": 137.119184193754, "grad_norm": 0.3293696939945221, "learning_rate": 2.0326541866678756e-05, "loss": 0.0054, "step": 215140 }, { "epoch": 137.12555768005097, "grad_norm": 0.2206740826368332, "learning_rate": 2.0322106025942423e-05, "loss": 0.0044, "step": 215150 }, { "epoch": 137.13193116634798, "grad_norm": 0.223127081990242, "learning_rate": 2.0317670545820555e-05, "loss": 0.007, "step": 215160 }, { "epoch": 137.138304652645, "grad_norm": 0.3096785843372345, "learning_rate": 2.0313235426366983e-05, "loss": 0.0061, "step": 215170 }, { "epoch": 137.144678138942, "grad_norm": 0.11705898493528366, "learning_rate": 2.0308800667635646e-05, "loss": 0.005, "step": 215180 }, { "epoch": 137.151051625239, "grad_norm": 0.6569610834121704, "learning_rate": 2.0304366269680398e-05, "loss": 0.0069, "step": 215190 }, { "epoch": 137.157425111536, "grad_norm": 0.08012557029724121, "learning_rate": 2.0299932232555175e-05, "loss": 0.005, "step": 215200 }, { "epoch": 137.16379859783302, "grad_norm": 0.10249363631010056, "learning_rate": 2.0295498556313774e-05, "loss": 0.003, "step": 215210 }, { "epoch": 137.17017208413003, "grad_norm": 0.17512385547161102, "learning_rate": 2.0291065241010137e-05, "loss": 0.0101, "step": 215220 }, { "epoch": 137.176545570427, "grad_norm": 0.16405276954174042, "learning_rate": 2.028663228669808e-05, "loss": 0.0111, "step": 215230 }, { "epoch": 137.18291905672402, "grad_norm": 0.1404777318239212, "learning_rate": 2.0282199693431543e-05, "loss": 0.0049, "step": 215240 }, { "epoch": 137.18929254302103, "grad_norm": 0.11998290568590164, "learning_rate": 2.027776746126429e-05, "loss": 0.0118, "step": 215250 }, { "epoch": 137.19566602931803, "grad_norm": 0.1581176370382309, "learning_rate": 2.0273335590250247e-05, "loss": 0.0189, "step": 215260 }, { "epoch": 137.20203951561504, "grad_norm": 0.18453535437583923, "learning_rate": 2.0268904080443247e-05, "loss": 0.005, "step": 215270 }, { "epoch": 137.20841300191205, "grad_norm": 0.4634920656681061, "learning_rate": 2.0264472931897133e-05, "loss": 0.0086, "step": 215280 }, { "epoch": 137.21478648820906, "grad_norm": 0.3214248716831207, "learning_rate": 2.0260042144665727e-05, "loss": 0.0047, "step": 215290 }, { "epoch": 137.22115997450607, "grad_norm": 0.42361024022102356, "learning_rate": 2.0255611718802903e-05, "loss": 0.007, "step": 215300 }, { "epoch": 137.22753346080307, "grad_norm": 0.18484309315681458, "learning_rate": 2.0251181654362484e-05, "loss": 0.0056, "step": 215310 }, { "epoch": 137.23390694710005, "grad_norm": 0.14186127483844757, "learning_rate": 2.0246751951398296e-05, "loss": 0.0059, "step": 215320 }, { "epoch": 137.24028043339706, "grad_norm": 0.22717846930027008, "learning_rate": 2.0242322609964144e-05, "loss": 0.0071, "step": 215330 }, { "epoch": 137.24665391969407, "grad_norm": 0.13992241024971008, "learning_rate": 2.0237893630113904e-05, "loss": 0.0047, "step": 215340 }, { "epoch": 137.25302740599108, "grad_norm": 0.16880561411380768, "learning_rate": 2.023346501190132e-05, "loss": 0.0056, "step": 215350 }, { "epoch": 137.25940089228808, "grad_norm": 0.2879900336265564, "learning_rate": 2.022903675538026e-05, "loss": 0.0068, "step": 215360 }, { "epoch": 137.2657743785851, "grad_norm": 0.20925010740756989, "learning_rate": 2.0224608860604493e-05, "loss": 0.0081, "step": 215370 }, { "epoch": 137.2721478648821, "grad_norm": 0.2062181532382965, "learning_rate": 2.022018132762788e-05, "loss": 0.0035, "step": 215380 }, { "epoch": 137.2785213511791, "grad_norm": 0.16855232417583466, "learning_rate": 2.021575415650414e-05, "loss": 0.0063, "step": 215390 }, { "epoch": 137.2848948374761, "grad_norm": 0.13148003816604614, "learning_rate": 2.0211327347287128e-05, "loss": 0.0082, "step": 215400 }, { "epoch": 137.2912683237731, "grad_norm": 0.22454619407653809, "learning_rate": 2.020690090003061e-05, "loss": 0.0077, "step": 215410 }, { "epoch": 137.2976418100701, "grad_norm": 0.1821751743555069, "learning_rate": 2.020247481478838e-05, "loss": 0.0043, "step": 215420 }, { "epoch": 137.3040152963671, "grad_norm": 0.4660310745239258, "learning_rate": 2.0198049091614195e-05, "loss": 0.0052, "step": 215430 }, { "epoch": 137.31038878266412, "grad_norm": 0.2703034281730652, "learning_rate": 2.0193623730561874e-05, "loss": 0.0072, "step": 215440 }, { "epoch": 137.31676226896113, "grad_norm": 0.24420884251594543, "learning_rate": 2.018919873168516e-05, "loss": 0.0055, "step": 215450 }, { "epoch": 137.32313575525814, "grad_norm": 0.29462483525276184, "learning_rate": 2.0184774095037828e-05, "loss": 0.0132, "step": 215460 }, { "epoch": 137.32950924155514, "grad_norm": 0.301673948764801, "learning_rate": 2.018034982067363e-05, "loss": 0.0128, "step": 215470 }, { "epoch": 137.33588272785212, "grad_norm": 0.2570488750934601, "learning_rate": 2.0175925908646347e-05, "loss": 0.005, "step": 215480 }, { "epoch": 137.34225621414913, "grad_norm": 0.1159176155924797, "learning_rate": 2.0171502359009724e-05, "loss": 0.0054, "step": 215490 }, { "epoch": 137.34862970044614, "grad_norm": 0.23873300850391388, "learning_rate": 2.016707917181751e-05, "loss": 0.0062, "step": 215500 }, { "epoch": 137.35500318674315, "grad_norm": 0.42516404390335083, "learning_rate": 2.016265634712344e-05, "loss": 0.0134, "step": 215510 }, { "epoch": 137.36137667304016, "grad_norm": 0.16932326555252075, "learning_rate": 2.0158233884981277e-05, "loss": 0.0036, "step": 215520 }, { "epoch": 137.36775015933716, "grad_norm": 0.1374453902244568, "learning_rate": 2.0153811785444748e-05, "loss": 0.0055, "step": 215530 }, { "epoch": 137.37412364563417, "grad_norm": 0.27588120102882385, "learning_rate": 2.0149390048567585e-05, "loss": 0.0041, "step": 215540 }, { "epoch": 137.38049713193118, "grad_norm": 0.09452113509178162, "learning_rate": 2.0144968674403496e-05, "loss": 0.0051, "step": 215550 }, { "epoch": 137.38687061822816, "grad_norm": 0.15802104771137238, "learning_rate": 2.0140547663006247e-05, "loss": 0.0098, "step": 215560 }, { "epoch": 137.39324410452517, "grad_norm": 0.2564275860786438, "learning_rate": 2.013612701442954e-05, "loss": 0.0062, "step": 215570 }, { "epoch": 137.39961759082217, "grad_norm": 0.11630161851644516, "learning_rate": 2.0131706728727085e-05, "loss": 0.0062, "step": 215580 }, { "epoch": 137.40599107711918, "grad_norm": 0.08030468225479126, "learning_rate": 2.012728680595259e-05, "loss": 0.0039, "step": 215590 }, { "epoch": 137.4123645634162, "grad_norm": 0.2308715134859085, "learning_rate": 2.0122867246159745e-05, "loss": 0.0134, "step": 215600 }, { "epoch": 137.4187380497132, "grad_norm": 0.16734735667705536, "learning_rate": 2.0118448049402294e-05, "loss": 0.005, "step": 215610 }, { "epoch": 137.4251115360102, "grad_norm": 0.3642149567604065, "learning_rate": 2.0114029215733908e-05, "loss": 0.004, "step": 215620 }, { "epoch": 137.4314850223072, "grad_norm": 0.20153072476387024, "learning_rate": 2.0109610745208286e-05, "loss": 0.0055, "step": 215630 }, { "epoch": 137.4378585086042, "grad_norm": 0.13427330553531647, "learning_rate": 2.01051926378791e-05, "loss": 0.0047, "step": 215640 }, { "epoch": 137.4442319949012, "grad_norm": 0.10932818800210953, "learning_rate": 2.010077489380006e-05, "loss": 0.0061, "step": 215650 }, { "epoch": 137.4506054811982, "grad_norm": 0.13909660279750824, "learning_rate": 2.009635751302484e-05, "loss": 0.0056, "step": 215660 }, { "epoch": 137.45697896749522, "grad_norm": 0.11812387406826019, "learning_rate": 2.0091940495607108e-05, "loss": 0.0079, "step": 215670 }, { "epoch": 137.46335245379223, "grad_norm": 0.35648247599601746, "learning_rate": 2.008752384160052e-05, "loss": 0.0074, "step": 215680 }, { "epoch": 137.46972594008923, "grad_norm": 0.31298592686653137, "learning_rate": 2.0083107551058778e-05, "loss": 0.0049, "step": 215690 }, { "epoch": 137.47609942638624, "grad_norm": 0.13687099516391754, "learning_rate": 2.0078691624035522e-05, "loss": 0.0043, "step": 215700 }, { "epoch": 137.48247291268325, "grad_norm": 0.29463210701942444, "learning_rate": 2.0074276060584417e-05, "loss": 0.008, "step": 215710 }, { "epoch": 137.48884639898023, "grad_norm": 0.4544181227684021, "learning_rate": 2.0069860860759092e-05, "loss": 0.0106, "step": 215720 }, { "epoch": 137.49521988527724, "grad_norm": 0.17018002271652222, "learning_rate": 2.0065446024613255e-05, "loss": 0.0048, "step": 215730 }, { "epoch": 137.50159337157424, "grad_norm": 0.13080976903438568, "learning_rate": 2.006103155220047e-05, "loss": 0.0086, "step": 215740 }, { "epoch": 137.50796685787125, "grad_norm": 0.1831139326095581, "learning_rate": 2.0056617443574444e-05, "loss": 0.0045, "step": 215750 }, { "epoch": 137.51434034416826, "grad_norm": 0.23537980020046234, "learning_rate": 2.0052203698788764e-05, "loss": 0.0051, "step": 215760 }, { "epoch": 137.52071383046527, "grad_norm": 0.3794563114643097, "learning_rate": 2.004779031789712e-05, "loss": 0.0066, "step": 215770 }, { "epoch": 137.52708731676228, "grad_norm": 0.1977662593126297, "learning_rate": 2.004337730095307e-05, "loss": 0.0114, "step": 215780 }, { "epoch": 137.53346080305928, "grad_norm": 0.15584811568260193, "learning_rate": 2.003896464801029e-05, "loss": 0.006, "step": 215790 }, { "epoch": 137.53983428935626, "grad_norm": 0.2632865607738495, "learning_rate": 2.0034552359122354e-05, "loss": 0.0069, "step": 215800 }, { "epoch": 137.54620777565327, "grad_norm": 0.2375778704881668, "learning_rate": 2.0030140434342936e-05, "loss": 0.0056, "step": 215810 }, { "epoch": 137.55258126195028, "grad_norm": 0.340436190366745, "learning_rate": 2.0025728873725575e-05, "loss": 0.0078, "step": 215820 }, { "epoch": 137.5589547482473, "grad_norm": 0.18441683053970337, "learning_rate": 2.0021317677323924e-05, "loss": 0.0041, "step": 215830 }, { "epoch": 137.5653282345443, "grad_norm": 0.21237874031066895, "learning_rate": 2.001690684519157e-05, "loss": 0.0087, "step": 215840 }, { "epoch": 137.5717017208413, "grad_norm": 0.32374295592308044, "learning_rate": 2.00124963773821e-05, "loss": 0.0129, "step": 215850 }, { "epoch": 137.5780752071383, "grad_norm": 0.37981003522872925, "learning_rate": 2.00080862739491e-05, "loss": 0.0103, "step": 215860 }, { "epoch": 137.58444869343532, "grad_norm": 0.1968267858028412, "learning_rate": 2.0003676534946186e-05, "loss": 0.0102, "step": 215870 }, { "epoch": 137.5908221797323, "grad_norm": 0.13946771621704102, "learning_rate": 1.9999267160426925e-05, "loss": 0.009, "step": 215880 }, { "epoch": 137.5971956660293, "grad_norm": 0.14451716840267181, "learning_rate": 1.999485815044489e-05, "loss": 0.0123, "step": 215890 }, { "epoch": 137.60356915232632, "grad_norm": 0.13509872555732727, "learning_rate": 1.9990449505053642e-05, "loss": 0.0047, "step": 215900 }, { "epoch": 137.60994263862332, "grad_norm": 0.21363605558872223, "learning_rate": 1.99860412243068e-05, "loss": 0.0057, "step": 215910 }, { "epoch": 137.61631612492033, "grad_norm": 0.18379932641983032, "learning_rate": 1.9981633308257863e-05, "loss": 0.0087, "step": 215920 }, { "epoch": 137.62268961121734, "grad_norm": 0.3258916437625885, "learning_rate": 1.9977225756960437e-05, "loss": 0.0059, "step": 215930 }, { "epoch": 137.62906309751435, "grad_norm": 0.16606158018112183, "learning_rate": 1.997281857046805e-05, "loss": 0.0143, "step": 215940 }, { "epoch": 137.63543658381136, "grad_norm": 0.20259936153888702, "learning_rate": 1.9968411748834304e-05, "loss": 0.0049, "step": 215950 }, { "epoch": 137.64181007010836, "grad_norm": 0.30687448382377625, "learning_rate": 1.996400529211267e-05, "loss": 0.0103, "step": 215960 }, { "epoch": 137.64818355640534, "grad_norm": 0.364359050989151, "learning_rate": 1.9959599200356748e-05, "loss": 0.0106, "step": 215970 }, { "epoch": 137.65455704270235, "grad_norm": 0.19447696208953857, "learning_rate": 1.9955193473620064e-05, "loss": 0.0058, "step": 215980 }, { "epoch": 137.66093052899936, "grad_norm": 0.36738014221191406, "learning_rate": 1.9950788111956142e-05, "loss": 0.0067, "step": 215990 }, { "epoch": 137.66730401529637, "grad_norm": 0.4452177882194519, "learning_rate": 1.9946383115418498e-05, "loss": 0.0083, "step": 216000 }, { "epoch": 137.67367750159337, "grad_norm": 0.13779163360595703, "learning_rate": 1.9941978484060698e-05, "loss": 0.0046, "step": 216010 }, { "epoch": 137.68005098789038, "grad_norm": 0.163221076130867, "learning_rate": 1.9937574217936233e-05, "loss": 0.0096, "step": 216020 }, { "epoch": 137.6864244741874, "grad_norm": 0.278967946767807, "learning_rate": 1.9933170317098627e-05, "loss": 0.0082, "step": 216030 }, { "epoch": 137.6927979604844, "grad_norm": 0.3218470811843872, "learning_rate": 1.9928766781601365e-05, "loss": 0.0136, "step": 216040 }, { "epoch": 137.69917144678138, "grad_norm": 0.10997866839170456, "learning_rate": 1.9924363611498022e-05, "loss": 0.0057, "step": 216050 }, { "epoch": 137.7055449330784, "grad_norm": 0.21389636397361755, "learning_rate": 1.991996080684202e-05, "loss": 0.0127, "step": 216060 }, { "epoch": 137.7119184193754, "grad_norm": 0.25612780451774597, "learning_rate": 1.991555836768691e-05, "loss": 0.0049, "step": 216070 }, { "epoch": 137.7182919056724, "grad_norm": 0.13881614804267883, "learning_rate": 1.9911156294086153e-05, "loss": 0.0074, "step": 216080 }, { "epoch": 137.7246653919694, "grad_norm": 0.14689861238002777, "learning_rate": 1.9906754586093296e-05, "loss": 0.0047, "step": 216090 }, { "epoch": 137.73103887826642, "grad_norm": 0.11211752146482468, "learning_rate": 1.990235324376174e-05, "loss": 0.0044, "step": 216100 }, { "epoch": 137.73741236456343, "grad_norm": 0.3128451108932495, "learning_rate": 1.989795226714503e-05, "loss": 0.0144, "step": 216110 }, { "epoch": 137.74378585086043, "grad_norm": 0.13104771077632904, "learning_rate": 1.9893551656296606e-05, "loss": 0.005, "step": 216120 }, { "epoch": 137.7501593371574, "grad_norm": 0.4680614471435547, "learning_rate": 1.9889151411269992e-05, "loss": 0.0064, "step": 216130 }, { "epoch": 137.75653282345442, "grad_norm": 0.11157342046499252, "learning_rate": 1.988475153211858e-05, "loss": 0.0078, "step": 216140 }, { "epoch": 137.76290630975143, "grad_norm": 0.2098930925130844, "learning_rate": 1.988035201889588e-05, "loss": 0.0068, "step": 216150 }, { "epoch": 137.76927979604844, "grad_norm": 0.38682860136032104, "learning_rate": 1.987595287165535e-05, "loss": 0.0091, "step": 216160 }, { "epoch": 137.77565328234544, "grad_norm": 0.16524016857147217, "learning_rate": 1.9871554090450432e-05, "loss": 0.0098, "step": 216170 }, { "epoch": 137.78202676864245, "grad_norm": 0.13451556861400604, "learning_rate": 1.986715567533456e-05, "loss": 0.0055, "step": 216180 }, { "epoch": 137.78840025493946, "grad_norm": 0.26001229882240295, "learning_rate": 1.986275762636121e-05, "loss": 0.0141, "step": 216190 }, { "epoch": 137.79477374123647, "grad_norm": 0.24435845017433167, "learning_rate": 1.9858359943583814e-05, "loss": 0.0075, "step": 216200 }, { "epoch": 137.80114722753345, "grad_norm": 0.23263774812221527, "learning_rate": 1.9853962627055807e-05, "loss": 0.0096, "step": 216210 }, { "epoch": 137.80752071383046, "grad_norm": 0.4045054614543915, "learning_rate": 1.984956567683059e-05, "loss": 0.0034, "step": 216220 }, { "epoch": 137.81389420012746, "grad_norm": 0.439545214176178, "learning_rate": 1.9845169092961642e-05, "loss": 0.0284, "step": 216230 }, { "epoch": 137.82026768642447, "grad_norm": 0.23075170814990997, "learning_rate": 1.9840772875502356e-05, "loss": 0.0062, "step": 216240 }, { "epoch": 137.82664117272148, "grad_norm": 0.3378181755542755, "learning_rate": 1.9836377024506153e-05, "loss": 0.0066, "step": 216250 }, { "epoch": 137.8330146590185, "grad_norm": 0.4385509490966797, "learning_rate": 1.9831981540026435e-05, "loss": 0.0084, "step": 216260 }, { "epoch": 137.8393881453155, "grad_norm": 0.43561646342277527, "learning_rate": 1.982758642211664e-05, "loss": 0.0197, "step": 216270 }, { "epoch": 137.8457616316125, "grad_norm": 0.12385480850934982, "learning_rate": 1.982319167083016e-05, "loss": 0.005, "step": 216280 }, { "epoch": 137.85213511790948, "grad_norm": 0.11731193959712982, "learning_rate": 1.981879728622039e-05, "loss": 0.0033, "step": 216290 }, { "epoch": 137.8585086042065, "grad_norm": 0.4248856008052826, "learning_rate": 1.981440326834073e-05, "loss": 0.0115, "step": 216300 }, { "epoch": 137.8648820905035, "grad_norm": 0.386565625667572, "learning_rate": 1.981000961724455e-05, "loss": 0.0071, "step": 216310 }, { "epoch": 137.8712555768005, "grad_norm": 0.24144095182418823, "learning_rate": 1.9805616332985276e-05, "loss": 0.0112, "step": 216320 }, { "epoch": 137.87762906309752, "grad_norm": 0.11142895370721817, "learning_rate": 1.9801223415616278e-05, "loss": 0.0055, "step": 216330 }, { "epoch": 137.88400254939452, "grad_norm": 0.11268014460802078, "learning_rate": 1.979683086519092e-05, "loss": 0.0047, "step": 216340 }, { "epoch": 137.89037603569153, "grad_norm": 0.13851545751094818, "learning_rate": 1.979243868176257e-05, "loss": 0.0072, "step": 216350 }, { "epoch": 137.89674952198854, "grad_norm": 0.17895470559597015, "learning_rate": 1.9788046865384624e-05, "loss": 0.0195, "step": 216360 }, { "epoch": 137.90312300828552, "grad_norm": 0.16493888199329376, "learning_rate": 1.9783655416110436e-05, "loss": 0.0072, "step": 216370 }, { "epoch": 137.90949649458253, "grad_norm": 0.4220949709415436, "learning_rate": 1.977926433399337e-05, "loss": 0.0096, "step": 216380 }, { "epoch": 137.91586998087953, "grad_norm": 0.15920333564281464, "learning_rate": 1.9774873619086747e-05, "loss": 0.0087, "step": 216390 }, { "epoch": 137.92224346717654, "grad_norm": 0.1126646101474762, "learning_rate": 1.977048327144397e-05, "loss": 0.0056, "step": 216400 }, { "epoch": 137.92861695347355, "grad_norm": 0.10267665237188339, "learning_rate": 1.9766093291118364e-05, "loss": 0.0065, "step": 216410 }, { "epoch": 137.93499043977056, "grad_norm": 0.16757726669311523, "learning_rate": 1.9761703678163267e-05, "loss": 0.01, "step": 216420 }, { "epoch": 137.94136392606757, "grad_norm": 0.20163755118846893, "learning_rate": 1.9757314432632e-05, "loss": 0.0092, "step": 216430 }, { "epoch": 137.94773741236457, "grad_norm": 0.11756237596273422, "learning_rate": 1.975292555457794e-05, "loss": 0.0081, "step": 216440 }, { "epoch": 137.95411089866155, "grad_norm": 0.2323741912841797, "learning_rate": 1.9748537044054387e-05, "loss": 0.0123, "step": 216450 }, { "epoch": 137.96048438495856, "grad_norm": 0.162825807929039, "learning_rate": 1.9744148901114677e-05, "loss": 0.0074, "step": 216460 }, { "epoch": 137.96685787125557, "grad_norm": 0.22952409088611603, "learning_rate": 1.97397611258121e-05, "loss": 0.0063, "step": 216470 }, { "epoch": 137.97323135755258, "grad_norm": 0.2733537554740906, "learning_rate": 1.973537371820004e-05, "loss": 0.0059, "step": 216480 }, { "epoch": 137.9796048438496, "grad_norm": 0.1864379495382309, "learning_rate": 1.973098667833172e-05, "loss": 0.0129, "step": 216490 }, { "epoch": 137.9859783301466, "grad_norm": 0.2064846009016037, "learning_rate": 1.972660000626051e-05, "loss": 0.0078, "step": 216500 }, { "epoch": 137.9923518164436, "grad_norm": 0.11351625621318817, "learning_rate": 1.9722213702039678e-05, "loss": 0.0044, "step": 216510 }, { "epoch": 137.9987253027406, "grad_norm": 0.22929507493972778, "learning_rate": 1.971782776572257e-05, "loss": 0.0083, "step": 216520 }, { "epoch": 138.00509878903762, "grad_norm": 0.08664324134588242, "learning_rate": 1.971344219736241e-05, "loss": 0.0046, "step": 216530 }, { "epoch": 138.0114722753346, "grad_norm": 0.09552918374538422, "learning_rate": 1.970905699701254e-05, "loss": 0.0095, "step": 216540 }, { "epoch": 138.0178457616316, "grad_norm": 0.600584089756012, "learning_rate": 1.970467216472623e-05, "loss": 0.012, "step": 216550 }, { "epoch": 138.0242192479286, "grad_norm": 0.15261892974376678, "learning_rate": 1.9700287700556752e-05, "loss": 0.0072, "step": 216560 }, { "epoch": 138.03059273422562, "grad_norm": 0.2030014991760254, "learning_rate": 1.969590360455737e-05, "loss": 0.0069, "step": 216570 }, { "epoch": 138.03696622052263, "grad_norm": 0.8578542470932007, "learning_rate": 1.969151987678139e-05, "loss": 0.0144, "step": 216580 }, { "epoch": 138.04333970681964, "grad_norm": 0.10232788324356079, "learning_rate": 1.9687136517282057e-05, "loss": 0.0088, "step": 216590 }, { "epoch": 138.04971319311665, "grad_norm": 0.23699578642845154, "learning_rate": 1.968275352611264e-05, "loss": 0.0073, "step": 216600 }, { "epoch": 138.05608667941365, "grad_norm": 0.6361034512519836, "learning_rate": 1.967837090332637e-05, "loss": 0.0185, "step": 216610 }, { "epoch": 138.06246016571063, "grad_norm": 0.2789441645145416, "learning_rate": 1.967398864897656e-05, "loss": 0.0057, "step": 216620 }, { "epoch": 138.06883365200764, "grad_norm": 0.18753960728645325, "learning_rate": 1.9669606763116382e-05, "loss": 0.0051, "step": 216630 }, { "epoch": 138.07520713830465, "grad_norm": 0.3209294080734253, "learning_rate": 1.966522524579914e-05, "loss": 0.0075, "step": 216640 }, { "epoch": 138.08158062460166, "grad_norm": 0.24838177859783173, "learning_rate": 1.966084409707803e-05, "loss": 0.0186, "step": 216650 }, { "epoch": 138.08795411089866, "grad_norm": 0.21112890541553497, "learning_rate": 1.9656463317006346e-05, "loss": 0.0103, "step": 216660 }, { "epoch": 138.09432759719567, "grad_norm": 0.13986405730247498, "learning_rate": 1.9652082905637248e-05, "loss": 0.0107, "step": 216670 }, { "epoch": 138.10070108349268, "grad_norm": 0.10067927092313766, "learning_rate": 1.9647702863024015e-05, "loss": 0.0065, "step": 216680 }, { "epoch": 138.1070745697897, "grad_norm": 0.21283303201198578, "learning_rate": 1.9643323189219847e-05, "loss": 0.0088, "step": 216690 }, { "epoch": 138.11344805608667, "grad_norm": 0.1314735859632492, "learning_rate": 1.9638943884277962e-05, "loss": 0.0047, "step": 216700 }, { "epoch": 138.11982154238368, "grad_norm": 0.39588820934295654, "learning_rate": 1.963456494825156e-05, "loss": 0.0064, "step": 216710 }, { "epoch": 138.12619502868068, "grad_norm": 0.15886543691158295, "learning_rate": 1.963018638119388e-05, "loss": 0.0037, "step": 216720 }, { "epoch": 138.1325685149777, "grad_norm": 0.16288916766643524, "learning_rate": 1.9625808183158108e-05, "loss": 0.0049, "step": 216730 }, { "epoch": 138.1389420012747, "grad_norm": 0.1319841593503952, "learning_rate": 1.9621430354197446e-05, "loss": 0.0058, "step": 216740 }, { "epoch": 138.1453154875717, "grad_norm": 0.3118193745613098, "learning_rate": 1.9617052894365068e-05, "loss": 0.0043, "step": 216750 }, { "epoch": 138.15168897386872, "grad_norm": 0.2497536987066269, "learning_rate": 1.9612675803714203e-05, "loss": 0.0151, "step": 216760 }, { "epoch": 138.15806246016572, "grad_norm": 0.4935654103755951, "learning_rate": 1.960829908229802e-05, "loss": 0.0081, "step": 216770 }, { "epoch": 138.1644359464627, "grad_norm": 0.2018563449382782, "learning_rate": 1.9603922730169693e-05, "loss": 0.005, "step": 216780 }, { "epoch": 138.1708094327597, "grad_norm": 0.27714109420776367, "learning_rate": 1.9599546747382384e-05, "loss": 0.0082, "step": 216790 }, { "epoch": 138.17718291905672, "grad_norm": 0.25821447372436523, "learning_rate": 1.9595171133989325e-05, "loss": 0.0107, "step": 216800 }, { "epoch": 138.18355640535373, "grad_norm": 0.13369639217853546, "learning_rate": 1.9590795890043612e-05, "loss": 0.0049, "step": 216810 }, { "epoch": 138.18992989165073, "grad_norm": 0.20319180190563202, "learning_rate": 1.958642101559845e-05, "loss": 0.0053, "step": 216820 }, { "epoch": 138.19630337794774, "grad_norm": 0.4524848759174347, "learning_rate": 1.9582046510706975e-05, "loss": 0.006, "step": 216830 }, { "epoch": 138.20267686424475, "grad_norm": 0.21538785099983215, "learning_rate": 1.9577672375422384e-05, "loss": 0.0134, "step": 216840 }, { "epoch": 138.20905035054176, "grad_norm": 0.4451136291027069, "learning_rate": 1.9573298609797768e-05, "loss": 0.0078, "step": 216850 }, { "epoch": 138.21542383683874, "grad_norm": 0.2311607301235199, "learning_rate": 1.9568925213886312e-05, "loss": 0.0035, "step": 216860 }, { "epoch": 138.22179732313575, "grad_norm": 0.27848657965660095, "learning_rate": 1.9564552187741154e-05, "loss": 0.0061, "step": 216870 }, { "epoch": 138.22817080943275, "grad_norm": 0.17566482722759247, "learning_rate": 1.956017953141542e-05, "loss": 0.0094, "step": 216880 }, { "epoch": 138.23454429572976, "grad_norm": 0.2716692388057709, "learning_rate": 1.9555807244962226e-05, "loss": 0.011, "step": 216890 }, { "epoch": 138.24091778202677, "grad_norm": 0.25217172503471375, "learning_rate": 1.9551435328434738e-05, "loss": 0.005, "step": 216900 }, { "epoch": 138.24729126832378, "grad_norm": 0.2408551275730133, "learning_rate": 1.9547063781886054e-05, "loss": 0.0054, "step": 216910 }, { "epoch": 138.2536647546208, "grad_norm": 0.2152169942855835, "learning_rate": 1.95426926053693e-05, "loss": 0.0073, "step": 216920 }, { "epoch": 138.2600382409178, "grad_norm": 0.2156735062599182, "learning_rate": 1.953832179893757e-05, "loss": 0.0083, "step": 216930 }, { "epoch": 138.26641172721477, "grad_norm": 0.15170666575431824, "learning_rate": 1.953395136264401e-05, "loss": 0.0077, "step": 216940 }, { "epoch": 138.27278521351178, "grad_norm": 0.35297510027885437, "learning_rate": 1.9529581296541703e-05, "loss": 0.0074, "step": 216950 }, { "epoch": 138.2791586998088, "grad_norm": 0.29680559039115906, "learning_rate": 1.9525211600683753e-05, "loss": 0.0104, "step": 216960 }, { "epoch": 138.2855321861058, "grad_norm": 0.12522265315055847, "learning_rate": 1.9520842275123236e-05, "loss": 0.0049, "step": 216970 }, { "epoch": 138.2919056724028, "grad_norm": 0.2835368514060974, "learning_rate": 1.951647331991328e-05, "loss": 0.0074, "step": 216980 }, { "epoch": 138.2982791586998, "grad_norm": 0.17366357147693634, "learning_rate": 1.9512104735106956e-05, "loss": 0.0058, "step": 216990 }, { "epoch": 138.30465264499682, "grad_norm": 0.14256805181503296, "learning_rate": 1.9507736520757348e-05, "loss": 0.0041, "step": 217000 }, { "epoch": 138.31102613129383, "grad_norm": 0.07321905344724655, "learning_rate": 1.950336867691751e-05, "loss": 0.0047, "step": 217010 }, { "epoch": 138.3173996175908, "grad_norm": 0.23089925944805145, "learning_rate": 1.9499001203640554e-05, "loss": 0.0071, "step": 217020 }, { "epoch": 138.32377310388782, "grad_norm": 0.6149997711181641, "learning_rate": 1.9494634100979532e-05, "loss": 0.0218, "step": 217030 }, { "epoch": 138.33014659018482, "grad_norm": 0.4166319966316223, "learning_rate": 1.9490267368987503e-05, "loss": 0.0083, "step": 217040 }, { "epoch": 138.33652007648183, "grad_norm": 0.13370093703269958, "learning_rate": 1.948590100771754e-05, "loss": 0.0075, "step": 217050 }, { "epoch": 138.34289356277884, "grad_norm": 0.2979470193386078, "learning_rate": 1.9481535017222662e-05, "loss": 0.0077, "step": 217060 }, { "epoch": 138.34926704907585, "grad_norm": 0.16512531042099, "learning_rate": 1.9477169397555973e-05, "loss": 0.0043, "step": 217070 }, { "epoch": 138.35564053537286, "grad_norm": 0.19398194551467896, "learning_rate": 1.947280414877049e-05, "loss": 0.0067, "step": 217080 }, { "epoch": 138.36201402166986, "grad_norm": 0.17073993384838104, "learning_rate": 1.9468439270919258e-05, "loss": 0.0067, "step": 217090 }, { "epoch": 138.36838750796684, "grad_norm": 0.13601866364479065, "learning_rate": 1.94640747640553e-05, "loss": 0.0078, "step": 217100 }, { "epoch": 138.37476099426385, "grad_norm": 0.373274028301239, "learning_rate": 1.9459710628231686e-05, "loss": 0.0116, "step": 217110 }, { "epoch": 138.38113448056086, "grad_norm": 0.19002099335193634, "learning_rate": 1.9455346863501418e-05, "loss": 0.0057, "step": 217120 }, { "epoch": 138.38750796685787, "grad_norm": 0.24328990280628204, "learning_rate": 1.945098346991753e-05, "loss": 0.0089, "step": 217130 }, { "epoch": 138.39388145315488, "grad_norm": 0.16268181800842285, "learning_rate": 1.9446620447533008e-05, "loss": 0.0071, "step": 217140 }, { "epoch": 138.40025493945188, "grad_norm": 0.46984291076660156, "learning_rate": 1.944225779640092e-05, "loss": 0.0071, "step": 217150 }, { "epoch": 138.4066284257489, "grad_norm": 0.5643371939659119, "learning_rate": 1.943789551657425e-05, "loss": 0.0088, "step": 217160 }, { "epoch": 138.4130019120459, "grad_norm": 0.13881660997867584, "learning_rate": 1.9433533608106003e-05, "loss": 0.0084, "step": 217170 }, { "epoch": 138.4193753983429, "grad_norm": 0.23799097537994385, "learning_rate": 1.9429172071049168e-05, "loss": 0.0107, "step": 217180 }, { "epoch": 138.4257488846399, "grad_norm": 0.18116261065006256, "learning_rate": 1.9424810905456785e-05, "loss": 0.0075, "step": 217190 }, { "epoch": 138.4321223709369, "grad_norm": 0.2612605392932892, "learning_rate": 1.9420450111381788e-05, "loss": 0.0062, "step": 217200 }, { "epoch": 138.4384958572339, "grad_norm": 0.14925865828990936, "learning_rate": 1.941608968887721e-05, "loss": 0.0059, "step": 217210 }, { "epoch": 138.4448693435309, "grad_norm": 0.22032584249973297, "learning_rate": 1.9411729637996002e-05, "loss": 0.0072, "step": 217220 }, { "epoch": 138.45124282982792, "grad_norm": 0.18316610157489777, "learning_rate": 1.9407369958791188e-05, "loss": 0.0047, "step": 217230 }, { "epoch": 138.45761631612493, "grad_norm": 0.26096147298812866, "learning_rate": 1.9403010651315683e-05, "loss": 0.0069, "step": 217240 }, { "epoch": 138.46398980242193, "grad_norm": 0.22780154645442963, "learning_rate": 1.9398651715622508e-05, "loss": 0.0119, "step": 217250 }, { "epoch": 138.47036328871894, "grad_norm": 0.08070777356624603, "learning_rate": 1.9394293151764597e-05, "loss": 0.0058, "step": 217260 }, { "epoch": 138.47673677501592, "grad_norm": 0.17443892359733582, "learning_rate": 1.9389934959794932e-05, "loss": 0.0096, "step": 217270 }, { "epoch": 138.48311026131293, "grad_norm": 0.23521216213703156, "learning_rate": 1.938557713976643e-05, "loss": 0.0046, "step": 217280 }, { "epoch": 138.48948374760994, "grad_norm": 0.2960941791534424, "learning_rate": 1.9381219691732094e-05, "loss": 0.0089, "step": 217290 }, { "epoch": 138.49585723390695, "grad_norm": 0.5523257851600647, "learning_rate": 1.937686261574485e-05, "loss": 0.0082, "step": 217300 }, { "epoch": 138.50223072020395, "grad_norm": 0.18517239391803741, "learning_rate": 1.9372505911857637e-05, "loss": 0.0044, "step": 217310 }, { "epoch": 138.50860420650096, "grad_norm": 0.23026584088802338, "learning_rate": 1.9368149580123375e-05, "loss": 0.0033, "step": 217320 }, { "epoch": 138.51497769279797, "grad_norm": 0.3116680383682251, "learning_rate": 1.9363793620595033e-05, "loss": 0.0092, "step": 217330 }, { "epoch": 138.52135117909498, "grad_norm": 0.19293805956840515, "learning_rate": 1.9359438033325522e-05, "loss": 0.0073, "step": 217340 }, { "epoch": 138.52772466539196, "grad_norm": 0.12377513200044632, "learning_rate": 1.9355082818367775e-05, "loss": 0.0104, "step": 217350 }, { "epoch": 138.53409815168897, "grad_norm": 0.11693527549505234, "learning_rate": 1.9350727975774686e-05, "loss": 0.0051, "step": 217360 }, { "epoch": 138.54047163798597, "grad_norm": 0.18866145610809326, "learning_rate": 1.934637350559922e-05, "loss": 0.0052, "step": 217370 }, { "epoch": 138.54684512428298, "grad_norm": 0.4640321731567383, "learning_rate": 1.934201940789423e-05, "loss": 0.0105, "step": 217380 }, { "epoch": 138.55321861058, "grad_norm": 0.2934904098510742, "learning_rate": 1.9337665682712662e-05, "loss": 0.0076, "step": 217390 }, { "epoch": 138.559592096877, "grad_norm": 0.3265955448150635, "learning_rate": 1.933331233010739e-05, "loss": 0.0044, "step": 217400 }, { "epoch": 138.565965583174, "grad_norm": 0.26340439915657043, "learning_rate": 1.932895935013136e-05, "loss": 0.0101, "step": 217410 }, { "epoch": 138.572339069471, "grad_norm": 0.3804100453853607, "learning_rate": 1.9324606742837403e-05, "loss": 0.0098, "step": 217420 }, { "epoch": 138.578712555768, "grad_norm": 0.2450338751077652, "learning_rate": 1.932025450827845e-05, "loss": 0.0163, "step": 217430 }, { "epoch": 138.585086042065, "grad_norm": 0.20626382529735565, "learning_rate": 1.931590264650737e-05, "loss": 0.0054, "step": 217440 }, { "epoch": 138.591459528362, "grad_norm": 0.5379351377487183, "learning_rate": 1.9311551157577045e-05, "loss": 0.0103, "step": 217450 }, { "epoch": 138.59783301465902, "grad_norm": 0.21126289665699005, "learning_rate": 1.930720004154033e-05, "loss": 0.0095, "step": 217460 }, { "epoch": 138.60420650095602, "grad_norm": 0.25215640664100647, "learning_rate": 1.9302849298450133e-05, "loss": 0.009, "step": 217470 }, { "epoch": 138.61057998725303, "grad_norm": 0.17294710874557495, "learning_rate": 1.9298498928359297e-05, "loss": 0.0076, "step": 217480 }, { "epoch": 138.61695347355004, "grad_norm": 0.2178502380847931, "learning_rate": 1.9294148931320688e-05, "loss": 0.0083, "step": 217490 }, { "epoch": 138.62332695984705, "grad_norm": 0.19170378148555756, "learning_rate": 1.9289799307387145e-05, "loss": 0.01, "step": 217500 }, { "epoch": 138.62970044614403, "grad_norm": 0.12312039732933044, "learning_rate": 1.928545005661157e-05, "loss": 0.0078, "step": 217510 }, { "epoch": 138.63607393244104, "grad_norm": 0.2018243372440338, "learning_rate": 1.928110117904674e-05, "loss": 0.0111, "step": 217520 }, { "epoch": 138.64244741873804, "grad_norm": 0.32228222489356995, "learning_rate": 1.927675267474555e-05, "loss": 0.0058, "step": 217530 }, { "epoch": 138.64882090503505, "grad_norm": 0.3768638074398041, "learning_rate": 1.927240454376081e-05, "loss": 0.0076, "step": 217540 }, { "epoch": 138.65519439133206, "grad_norm": 0.11019770801067352, "learning_rate": 1.9268056786145396e-05, "loss": 0.0066, "step": 217550 }, { "epoch": 138.66156787762907, "grad_norm": 0.17740483582019806, "learning_rate": 1.9263709401952073e-05, "loss": 0.0065, "step": 217560 }, { "epoch": 138.66794136392608, "grad_norm": 0.4533747136592865, "learning_rate": 1.925936239123372e-05, "loss": 0.0076, "step": 217570 }, { "epoch": 138.67431485022308, "grad_norm": 0.20933827757835388, "learning_rate": 1.9255015754043138e-05, "loss": 0.0076, "step": 217580 }, { "epoch": 138.68068833652006, "grad_norm": 0.2551039457321167, "learning_rate": 1.9250669490433137e-05, "loss": 0.0049, "step": 217590 }, { "epoch": 138.68706182281707, "grad_norm": 0.5738994479179382, "learning_rate": 1.9246323600456522e-05, "loss": 0.012, "step": 217600 }, { "epoch": 138.69343530911408, "grad_norm": 0.16510044038295746, "learning_rate": 1.924197808416613e-05, "loss": 0.0109, "step": 217610 }, { "epoch": 138.6998087954111, "grad_norm": 0.2214079350233078, "learning_rate": 1.9237632941614748e-05, "loss": 0.0072, "step": 217620 }, { "epoch": 138.7061822817081, "grad_norm": 0.12332244217395782, "learning_rate": 1.9233288172855164e-05, "loss": 0.0063, "step": 217630 }, { "epoch": 138.7125557680051, "grad_norm": 0.20049341022968292, "learning_rate": 1.922894377794016e-05, "loss": 0.0113, "step": 217640 }, { "epoch": 138.7189292543021, "grad_norm": 0.25063055753707886, "learning_rate": 1.9224599756922563e-05, "loss": 0.009, "step": 217650 }, { "epoch": 138.72530274059912, "grad_norm": 0.21326684951782227, "learning_rate": 1.9220256109855128e-05, "loss": 0.0133, "step": 217660 }, { "epoch": 138.7316762268961, "grad_norm": 0.09025144577026367, "learning_rate": 1.921591283679065e-05, "loss": 0.0097, "step": 217670 }, { "epoch": 138.7380497131931, "grad_norm": 0.12948445975780487, "learning_rate": 1.9211569937781875e-05, "loss": 0.0082, "step": 217680 }, { "epoch": 138.74442319949011, "grad_norm": 0.627592921257019, "learning_rate": 1.920722741288161e-05, "loss": 0.0086, "step": 217690 }, { "epoch": 138.75079668578712, "grad_norm": 0.12666122615337372, "learning_rate": 1.920288526214261e-05, "loss": 0.0112, "step": 217700 }, { "epoch": 138.75717017208413, "grad_norm": 0.15319454669952393, "learning_rate": 1.9198543485617633e-05, "loss": 0.0076, "step": 217710 }, { "epoch": 138.76354365838114, "grad_norm": 0.5063431859016418, "learning_rate": 1.9194202083359415e-05, "loss": 0.0121, "step": 217720 }, { "epoch": 138.76991714467815, "grad_norm": 0.25424817204475403, "learning_rate": 1.918986105542074e-05, "loss": 0.0057, "step": 217730 }, { "epoch": 138.77629063097515, "grad_norm": 0.2298424392938614, "learning_rate": 1.9185520401854344e-05, "loss": 0.0093, "step": 217740 }, { "epoch": 138.78266411727213, "grad_norm": 0.2260652482509613, "learning_rate": 1.918118012271297e-05, "loss": 0.0059, "step": 217750 }, { "epoch": 138.78903760356914, "grad_norm": 0.3487110137939453, "learning_rate": 1.917684021804936e-05, "loss": 0.0059, "step": 217760 }, { "epoch": 138.79541108986615, "grad_norm": 0.1704239398241043, "learning_rate": 1.917250068791622e-05, "loss": 0.005, "step": 217770 }, { "epoch": 138.80178457616316, "grad_norm": 0.43441230058670044, "learning_rate": 1.916816153236632e-05, "loss": 0.015, "step": 217780 }, { "epoch": 138.80815806246017, "grad_norm": 0.6473061442375183, "learning_rate": 1.9163822751452372e-05, "loss": 0.0053, "step": 217790 }, { "epoch": 138.81453154875717, "grad_norm": 0.1532680094242096, "learning_rate": 1.9159484345227087e-05, "loss": 0.0107, "step": 217800 }, { "epoch": 138.82090503505418, "grad_norm": 0.1973857283592224, "learning_rate": 1.9155146313743167e-05, "loss": 0.0054, "step": 217810 }, { "epoch": 138.8272785213512, "grad_norm": 0.17762959003448486, "learning_rate": 1.915080865705336e-05, "loss": 0.0056, "step": 217820 }, { "epoch": 138.8336520076482, "grad_norm": 0.21083243191242218, "learning_rate": 1.9146471375210355e-05, "loss": 0.004, "step": 217830 }, { "epoch": 138.84002549394518, "grad_norm": 0.1812191903591156, "learning_rate": 1.9142134468266858e-05, "loss": 0.0105, "step": 217840 }, { "epoch": 138.84639898024218, "grad_norm": 0.1846897453069687, "learning_rate": 1.9137797936275542e-05, "loss": 0.0042, "step": 217850 }, { "epoch": 138.8527724665392, "grad_norm": 0.18683649599552155, "learning_rate": 1.9133461779289135e-05, "loss": 0.0086, "step": 217860 }, { "epoch": 138.8591459528362, "grad_norm": 0.20484687387943268, "learning_rate": 1.9129125997360307e-05, "loss": 0.0087, "step": 217870 }, { "epoch": 138.8655194391332, "grad_norm": 0.29858508706092834, "learning_rate": 1.912479059054175e-05, "loss": 0.0063, "step": 217880 }, { "epoch": 138.87189292543022, "grad_norm": 0.409422367811203, "learning_rate": 1.9120455558886126e-05, "loss": 0.0081, "step": 217890 }, { "epoch": 138.87826641172722, "grad_norm": 0.1844300776720047, "learning_rate": 1.911612090244615e-05, "loss": 0.0052, "step": 217900 }, { "epoch": 138.88463989802423, "grad_norm": 0.2832438051700592, "learning_rate": 1.911178662127443e-05, "loss": 0.0078, "step": 217910 }, { "epoch": 138.8910133843212, "grad_norm": 0.2232305109500885, "learning_rate": 1.9107452715423683e-05, "loss": 0.0095, "step": 217920 }, { "epoch": 138.89738687061822, "grad_norm": 0.11733000725507736, "learning_rate": 1.910311918494654e-05, "loss": 0.0101, "step": 217930 }, { "epoch": 138.90376035691523, "grad_norm": 0.2021319568157196, "learning_rate": 1.90987860298957e-05, "loss": 0.0055, "step": 217940 }, { "epoch": 138.91013384321224, "grad_norm": 0.10682687163352966, "learning_rate": 1.909445325032375e-05, "loss": 0.0089, "step": 217950 }, { "epoch": 138.91650732950924, "grad_norm": 0.1672673374414444, "learning_rate": 1.9090120846283394e-05, "loss": 0.0032, "step": 217960 }, { "epoch": 138.92288081580625, "grad_norm": 0.10852009803056717, "learning_rate": 1.9085788817827234e-05, "loss": 0.0073, "step": 217970 }, { "epoch": 138.92925430210326, "grad_norm": 0.16592496633529663, "learning_rate": 1.908145716500796e-05, "loss": 0.0103, "step": 217980 }, { "epoch": 138.93562778840027, "grad_norm": 0.22799095511436462, "learning_rate": 1.9077125887878146e-05, "loss": 0.0046, "step": 217990 }, { "epoch": 138.94200127469725, "grad_norm": 0.2313104122877121, "learning_rate": 1.9072794986490456e-05, "loss": 0.0056, "step": 218000 }, { "epoch": 138.94837476099426, "grad_norm": 0.2311987727880478, "learning_rate": 1.9068464460897512e-05, "loss": 0.0278, "step": 218010 }, { "epoch": 138.95474824729126, "grad_norm": 0.1301259845495224, "learning_rate": 1.906413431115193e-05, "loss": 0.0057, "step": 218020 }, { "epoch": 138.96112173358827, "grad_norm": 0.23620569705963135, "learning_rate": 1.9059804537306308e-05, "loss": 0.0058, "step": 218030 }, { "epoch": 138.96749521988528, "grad_norm": 0.12200727313756943, "learning_rate": 1.9055475139413287e-05, "loss": 0.0089, "step": 218040 }, { "epoch": 138.9738687061823, "grad_norm": 0.2264038622379303, "learning_rate": 1.9051146117525466e-05, "loss": 0.015, "step": 218050 }, { "epoch": 138.9802421924793, "grad_norm": 0.19326196610927582, "learning_rate": 1.9046817471695432e-05, "loss": 0.0086, "step": 218060 }, { "epoch": 138.9866156787763, "grad_norm": 0.2706233263015747, "learning_rate": 1.9042489201975778e-05, "loss": 0.0069, "step": 218070 }, { "epoch": 138.99298916507328, "grad_norm": 0.11414724588394165, "learning_rate": 1.9038161308419143e-05, "loss": 0.0073, "step": 218080 }, { "epoch": 138.9993626513703, "grad_norm": 0.11975690722465515, "learning_rate": 1.9033833791078043e-05, "loss": 0.0174, "step": 218090 }, { "epoch": 139.0057361376673, "grad_norm": 0.27074119448661804, "learning_rate": 1.902950665000512e-05, "loss": 0.0067, "step": 218100 }, { "epoch": 139.0121096239643, "grad_norm": 0.49011629819869995, "learning_rate": 1.9025179885252915e-05, "loss": 0.0088, "step": 218110 }, { "epoch": 139.01848311026131, "grad_norm": 0.10533132404088974, "learning_rate": 1.902085349687405e-05, "loss": 0.0065, "step": 218120 }, { "epoch": 139.02485659655832, "grad_norm": 0.2099779099225998, "learning_rate": 1.9016527484921037e-05, "loss": 0.0077, "step": 218130 }, { "epoch": 139.03123008285533, "grad_norm": 0.17016273736953735, "learning_rate": 1.9012201849446483e-05, "loss": 0.0145, "step": 218140 }, { "epoch": 139.03760356915234, "grad_norm": 0.21829332411289215, "learning_rate": 1.9007876590502933e-05, "loss": 0.0063, "step": 218150 }, { "epoch": 139.04397705544932, "grad_norm": 0.21048825979232788, "learning_rate": 1.900355170814294e-05, "loss": 0.0075, "step": 218160 }, { "epoch": 139.05035054174633, "grad_norm": 0.20325660705566406, "learning_rate": 1.8999227202419044e-05, "loss": 0.0042, "step": 218170 }, { "epoch": 139.05672402804333, "grad_norm": 0.1131449043750763, "learning_rate": 1.8994903073383822e-05, "loss": 0.0098, "step": 218180 }, { "epoch": 139.06309751434034, "grad_norm": 0.11882281303405762, "learning_rate": 1.8990579321089807e-05, "loss": 0.0075, "step": 218190 }, { "epoch": 139.06947100063735, "grad_norm": 0.14975681900978088, "learning_rate": 1.8986255945589532e-05, "loss": 0.0037, "step": 218200 }, { "epoch": 139.07584448693436, "grad_norm": 0.2736274003982544, "learning_rate": 1.8981932946935506e-05, "loss": 0.0072, "step": 218210 }, { "epoch": 139.08221797323137, "grad_norm": 0.08850269019603729, "learning_rate": 1.8977610325180302e-05, "loss": 0.0063, "step": 218220 }, { "epoch": 139.08859145952837, "grad_norm": 0.5580034852027893, "learning_rate": 1.8973288080376424e-05, "loss": 0.0138, "step": 218230 }, { "epoch": 139.09496494582535, "grad_norm": 0.27963021397590637, "learning_rate": 1.8968966212576384e-05, "loss": 0.0036, "step": 218240 }, { "epoch": 139.10133843212236, "grad_norm": 0.24302053451538086, "learning_rate": 1.8964644721832687e-05, "loss": 0.0054, "step": 218250 }, { "epoch": 139.10771191841937, "grad_norm": 0.17115093767642975, "learning_rate": 1.8960323608197894e-05, "loss": 0.0131, "step": 218260 }, { "epoch": 139.11408540471638, "grad_norm": 0.1921529620885849, "learning_rate": 1.8956002871724444e-05, "loss": 0.0049, "step": 218270 }, { "epoch": 139.12045889101339, "grad_norm": 0.2394205927848816, "learning_rate": 1.8951682512464885e-05, "loss": 0.0171, "step": 218280 }, { "epoch": 139.1268323773104, "grad_norm": 0.2466907799243927, "learning_rate": 1.894736253047168e-05, "loss": 0.0111, "step": 218290 }, { "epoch": 139.1332058636074, "grad_norm": 0.1804196685552597, "learning_rate": 1.8943042925797372e-05, "loss": 0.0057, "step": 218300 }, { "epoch": 139.1395793499044, "grad_norm": 0.07853078842163086, "learning_rate": 1.893872369849438e-05, "loss": 0.0122, "step": 218310 }, { "epoch": 139.1459528362014, "grad_norm": 0.18524841964244843, "learning_rate": 1.8934404848615246e-05, "loss": 0.0098, "step": 218320 }, { "epoch": 139.1523263224984, "grad_norm": 0.13718639314174652, "learning_rate": 1.8930086376212413e-05, "loss": 0.01, "step": 218330 }, { "epoch": 139.1586998087954, "grad_norm": 0.5309875011444092, "learning_rate": 1.8925768281338375e-05, "loss": 0.0077, "step": 218340 }, { "epoch": 139.1650732950924, "grad_norm": 0.12051234394311905, "learning_rate": 1.8921450564045567e-05, "loss": 0.0115, "step": 218350 }, { "epoch": 139.17144678138942, "grad_norm": 0.14545747637748718, "learning_rate": 1.89171332243865e-05, "loss": 0.0085, "step": 218360 }, { "epoch": 139.17782026768643, "grad_norm": 0.2368392050266266, "learning_rate": 1.891281626241361e-05, "loss": 0.0059, "step": 218370 }, { "epoch": 139.18419375398344, "grad_norm": 0.24410773813724518, "learning_rate": 1.8908499678179353e-05, "loss": 0.0135, "step": 218380 }, { "epoch": 139.19056724028044, "grad_norm": 0.26052534580230713, "learning_rate": 1.8904183471736163e-05, "loss": 0.0121, "step": 218390 }, { "epoch": 139.19694072657742, "grad_norm": 0.1728202998638153, "learning_rate": 1.889986764313652e-05, "loss": 0.004, "step": 218400 }, { "epoch": 139.20331421287443, "grad_norm": 0.4553849697113037, "learning_rate": 1.8895552192432846e-05, "loss": 0.0077, "step": 218410 }, { "epoch": 139.20968769917144, "grad_norm": 0.1822715550661087, "learning_rate": 1.8891237119677584e-05, "loss": 0.0051, "step": 218420 }, { "epoch": 139.21606118546845, "grad_norm": 0.1420987993478775, "learning_rate": 1.888692242492315e-05, "loss": 0.0122, "step": 218430 }, { "epoch": 139.22243467176546, "grad_norm": 0.1933155655860901, "learning_rate": 1.8882608108221994e-05, "loss": 0.0057, "step": 218440 }, { "epoch": 139.22880815806246, "grad_norm": 0.549008846282959, "learning_rate": 1.8878294169626532e-05, "loss": 0.0078, "step": 218450 }, { "epoch": 139.23518164435947, "grad_norm": 0.19976665079593658, "learning_rate": 1.887398060918918e-05, "loss": 0.0067, "step": 218460 }, { "epoch": 139.24155513065648, "grad_norm": 0.21710583567619324, "learning_rate": 1.8869667426962346e-05, "loss": 0.0045, "step": 218470 }, { "epoch": 139.2479286169535, "grad_norm": 0.1401902586221695, "learning_rate": 1.8865354622998433e-05, "loss": 0.004, "step": 218480 }, { "epoch": 139.25430210325047, "grad_norm": 0.5956876277923584, "learning_rate": 1.886104219734987e-05, "loss": 0.0086, "step": 218490 }, { "epoch": 139.26067558954747, "grad_norm": 0.18186087906360626, "learning_rate": 1.885673015006905e-05, "loss": 0.0086, "step": 218500 }, { "epoch": 139.26704907584448, "grad_norm": 0.2440667748451233, "learning_rate": 1.885241848120836e-05, "loss": 0.0077, "step": 218510 }, { "epoch": 139.2734225621415, "grad_norm": 0.14562219381332397, "learning_rate": 1.884810719082018e-05, "loss": 0.0073, "step": 218520 }, { "epoch": 139.2797960484385, "grad_norm": 0.8733640909194946, "learning_rate": 1.884379627895692e-05, "loss": 0.0087, "step": 218530 }, { "epoch": 139.2861695347355, "grad_norm": 0.3597100079059601, "learning_rate": 1.883948574567096e-05, "loss": 0.0091, "step": 218540 }, { "epoch": 139.29254302103251, "grad_norm": 0.3171570301055908, "learning_rate": 1.8835175591014658e-05, "loss": 0.0059, "step": 218550 }, { "epoch": 139.29891650732952, "grad_norm": 0.2664869725704193, "learning_rate": 1.8830865815040388e-05, "loss": 0.006, "step": 218560 }, { "epoch": 139.3052899936265, "grad_norm": 0.1019761860370636, "learning_rate": 1.8826556417800546e-05, "loss": 0.0109, "step": 218570 }, { "epoch": 139.3116634799235, "grad_norm": 0.46546390652656555, "learning_rate": 1.882224739934747e-05, "loss": 0.0128, "step": 218580 }, { "epoch": 139.31803696622052, "grad_norm": 0.8527642488479614, "learning_rate": 1.881793875973353e-05, "loss": 0.0102, "step": 218590 }, { "epoch": 139.32441045251753, "grad_norm": 0.23042652010917664, "learning_rate": 1.8813630499011047e-05, "loss": 0.007, "step": 218600 }, { "epoch": 139.33078393881453, "grad_norm": 0.37165820598602295, "learning_rate": 1.8809322617232427e-05, "loss": 0.0099, "step": 218610 }, { "epoch": 139.33715742511154, "grad_norm": 0.3688690662384033, "learning_rate": 1.8805015114449982e-05, "loss": 0.0129, "step": 218620 }, { "epoch": 139.34353091140855, "grad_norm": 0.28058117628097534, "learning_rate": 1.8800707990716054e-05, "loss": 0.0055, "step": 218630 }, { "epoch": 139.34990439770556, "grad_norm": 0.15723960101604462, "learning_rate": 1.8796401246082968e-05, "loss": 0.0062, "step": 218640 }, { "epoch": 139.35627788400254, "grad_norm": 0.1910787671804428, "learning_rate": 1.8792094880603102e-05, "loss": 0.0072, "step": 218650 }, { "epoch": 139.36265137029955, "grad_norm": 0.1690051406621933, "learning_rate": 1.8787788894328713e-05, "loss": 0.0067, "step": 218660 }, { "epoch": 139.36902485659655, "grad_norm": 0.4334966242313385, "learning_rate": 1.878348328731217e-05, "loss": 0.0079, "step": 218670 }, { "epoch": 139.37539834289356, "grad_norm": 0.0670698881149292, "learning_rate": 1.8779178059605767e-05, "loss": 0.0058, "step": 218680 }, { "epoch": 139.38177182919057, "grad_norm": 0.12287802249193192, "learning_rate": 1.8774873211261864e-05, "loss": 0.0053, "step": 218690 }, { "epoch": 139.38814531548758, "grad_norm": 0.23422783613204956, "learning_rate": 1.8770568742332694e-05, "loss": 0.0088, "step": 218700 }, { "epoch": 139.39451880178459, "grad_norm": 0.36214494705200195, "learning_rate": 1.876626465287062e-05, "loss": 0.0066, "step": 218710 }, { "epoch": 139.4008922880816, "grad_norm": 0.2337128072977066, "learning_rate": 1.8761960942927914e-05, "loss": 0.0067, "step": 218720 }, { "epoch": 139.40726577437857, "grad_norm": 0.18611645698547363, "learning_rate": 1.8757657612556884e-05, "loss": 0.0073, "step": 218730 }, { "epoch": 139.41363926067558, "grad_norm": 0.13999323546886444, "learning_rate": 1.8753354661809786e-05, "loss": 0.009, "step": 218740 }, { "epoch": 139.4200127469726, "grad_norm": 0.5177416205406189, "learning_rate": 1.8749052090738955e-05, "loss": 0.0134, "step": 218750 }, { "epoch": 139.4263862332696, "grad_norm": 0.1895226538181305, "learning_rate": 1.8744749899396647e-05, "loss": 0.0068, "step": 218760 }, { "epoch": 139.4327597195666, "grad_norm": 0.15582948923110962, "learning_rate": 1.8740448087835134e-05, "loss": 0.0108, "step": 218770 }, { "epoch": 139.4391332058636, "grad_norm": 0.20412197709083557, "learning_rate": 1.8736146656106674e-05, "loss": 0.0055, "step": 218780 }, { "epoch": 139.44550669216062, "grad_norm": 0.13851429522037506, "learning_rate": 1.8731845604263583e-05, "loss": 0.0064, "step": 218790 }, { "epoch": 139.45188017845763, "grad_norm": 0.18790236115455627, "learning_rate": 1.8727544932358056e-05, "loss": 0.0039, "step": 218800 }, { "epoch": 139.4582536647546, "grad_norm": 0.275541216135025, "learning_rate": 1.8723244640442405e-05, "loss": 0.007, "step": 218810 }, { "epoch": 139.46462715105162, "grad_norm": 0.6011103391647339, "learning_rate": 1.8718944728568842e-05, "loss": 0.0191, "step": 218820 }, { "epoch": 139.47100063734862, "grad_norm": 0.135948047041893, "learning_rate": 1.8714645196789672e-05, "loss": 0.0077, "step": 218830 }, { "epoch": 139.47737412364563, "grad_norm": 0.21325470507144928, "learning_rate": 1.8710346045157063e-05, "loss": 0.0076, "step": 218840 }, { "epoch": 139.48374760994264, "grad_norm": 0.10691758990287781, "learning_rate": 1.870604727372331e-05, "loss": 0.0061, "step": 218850 }, { "epoch": 139.49012109623965, "grad_norm": 0.12074794620275497, "learning_rate": 1.8701748882540608e-05, "loss": 0.0046, "step": 218860 }, { "epoch": 139.49649458253666, "grad_norm": 0.2639900743961334, "learning_rate": 1.869745087166125e-05, "loss": 0.0069, "step": 218870 }, { "epoch": 139.50286806883366, "grad_norm": 0.1159106194972992, "learning_rate": 1.869315324113738e-05, "loss": 0.0051, "step": 218880 }, { "epoch": 139.50924155513064, "grad_norm": 0.15131977200508118, "learning_rate": 1.868885599102127e-05, "loss": 0.0056, "step": 218890 }, { "epoch": 139.51561504142765, "grad_norm": 0.176237091422081, "learning_rate": 1.868455912136513e-05, "loss": 0.0053, "step": 218900 }, { "epoch": 139.52198852772466, "grad_norm": 0.25299641489982605, "learning_rate": 1.8680262632221157e-05, "loss": 0.0068, "step": 218910 }, { "epoch": 139.52836201402167, "grad_norm": 0.5523607730865479, "learning_rate": 1.8675966523641547e-05, "loss": 0.0095, "step": 218920 }, { "epoch": 139.53473550031867, "grad_norm": 0.13358797132968903, "learning_rate": 1.8671670795678543e-05, "loss": 0.0122, "step": 218930 }, { "epoch": 139.54110898661568, "grad_norm": 0.22937746345996857, "learning_rate": 1.8667375448384312e-05, "loss": 0.0135, "step": 218940 }, { "epoch": 139.5474824729127, "grad_norm": 0.4625759422779083, "learning_rate": 1.8663080481811053e-05, "loss": 0.007, "step": 218950 }, { "epoch": 139.5538559592097, "grad_norm": 0.33126723766326904, "learning_rate": 1.8658785896010934e-05, "loss": 0.0053, "step": 218960 }, { "epoch": 139.56022944550668, "grad_norm": 0.18591174483299255, "learning_rate": 1.865449169103619e-05, "loss": 0.0084, "step": 218970 }, { "epoch": 139.5666029318037, "grad_norm": 0.20573800802230835, "learning_rate": 1.865019786693894e-05, "loss": 0.0056, "step": 218980 }, { "epoch": 139.5729764181007, "grad_norm": 0.20075666904449463, "learning_rate": 1.8645904423771398e-05, "loss": 0.0069, "step": 218990 }, { "epoch": 139.5793499043977, "grad_norm": 0.19623151421546936, "learning_rate": 1.8641611361585705e-05, "loss": 0.0079, "step": 219000 }, { "epoch": 139.5857233906947, "grad_norm": 0.17328153550624847, "learning_rate": 1.8637318680434073e-05, "loss": 0.0072, "step": 219010 }, { "epoch": 139.59209687699172, "grad_norm": 0.1487720012664795, "learning_rate": 1.8633026380368595e-05, "loss": 0.0039, "step": 219020 }, { "epoch": 139.59847036328873, "grad_norm": 0.16212336719036102, "learning_rate": 1.8628734461441482e-05, "loss": 0.006, "step": 219030 }, { "epoch": 139.60484384958573, "grad_norm": 0.22178006172180176, "learning_rate": 1.8624442923704864e-05, "loss": 0.0098, "step": 219040 }, { "epoch": 139.61121733588274, "grad_norm": 0.4180232584476471, "learning_rate": 1.8620151767210888e-05, "loss": 0.0116, "step": 219050 }, { "epoch": 139.61759082217972, "grad_norm": 0.08662152290344238, "learning_rate": 1.8615860992011675e-05, "loss": 0.0063, "step": 219060 }, { "epoch": 139.62396430847673, "grad_norm": 0.20701339840888977, "learning_rate": 1.8611570598159405e-05, "loss": 0.0071, "step": 219070 }, { "epoch": 139.63033779477374, "grad_norm": 0.10844285041093826, "learning_rate": 1.8607280585706182e-05, "loss": 0.0043, "step": 219080 }, { "epoch": 139.63671128107075, "grad_norm": 0.3546562194824219, "learning_rate": 1.8602990954704147e-05, "loss": 0.0113, "step": 219090 }, { "epoch": 139.64308476736775, "grad_norm": 0.21799764037132263, "learning_rate": 1.859870170520539e-05, "loss": 0.0047, "step": 219100 }, { "epoch": 139.64945825366476, "grad_norm": 0.30226820707321167, "learning_rate": 1.8594412837262082e-05, "loss": 0.006, "step": 219110 }, { "epoch": 139.65583173996177, "grad_norm": 0.19934721291065216, "learning_rate": 1.85901243509263e-05, "loss": 0.0041, "step": 219120 }, { "epoch": 139.66220522625878, "grad_norm": 0.20523478090763092, "learning_rate": 1.858583624625017e-05, "loss": 0.0059, "step": 219130 }, { "epoch": 139.66857871255576, "grad_norm": 0.3002677857875824, "learning_rate": 1.858154852328577e-05, "loss": 0.0071, "step": 219140 }, { "epoch": 139.67495219885276, "grad_norm": 0.1602572500705719, "learning_rate": 1.857726118208524e-05, "loss": 0.007, "step": 219150 }, { "epoch": 139.68132568514977, "grad_norm": 0.15535315871238708, "learning_rate": 1.8572974222700644e-05, "loss": 0.006, "step": 219160 }, { "epoch": 139.68769917144678, "grad_norm": 0.11012187600135803, "learning_rate": 1.856868764518409e-05, "loss": 0.0049, "step": 219170 }, { "epoch": 139.6940726577438, "grad_norm": 0.17553012073040009, "learning_rate": 1.8564401449587644e-05, "loss": 0.0037, "step": 219180 }, { "epoch": 139.7004461440408, "grad_norm": 0.19491469860076904, "learning_rate": 1.856011563596341e-05, "loss": 0.0084, "step": 219190 }, { "epoch": 139.7068196303378, "grad_norm": 0.13733184337615967, "learning_rate": 1.855583020436346e-05, "loss": 0.0046, "step": 219200 }, { "epoch": 139.7131931166348, "grad_norm": 0.11267005652189255, "learning_rate": 1.8551545154839862e-05, "loss": 0.0112, "step": 219210 }, { "epoch": 139.7195666029318, "grad_norm": 0.21132616698741913, "learning_rate": 1.8547260487444678e-05, "loss": 0.007, "step": 219220 }, { "epoch": 139.7259400892288, "grad_norm": 0.14365558326244354, "learning_rate": 1.8542976202229963e-05, "loss": 0.0049, "step": 219230 }, { "epoch": 139.7323135755258, "grad_norm": 0.21706852316856384, "learning_rate": 1.85386922992478e-05, "loss": 0.0149, "step": 219240 }, { "epoch": 139.73868706182282, "grad_norm": 0.13107416033744812, "learning_rate": 1.8534408778550234e-05, "loss": 0.009, "step": 219250 }, { "epoch": 139.74506054811982, "grad_norm": 0.23834921419620514, "learning_rate": 1.853012564018931e-05, "loss": 0.0073, "step": 219260 }, { "epoch": 139.75143403441683, "grad_norm": 0.1338113248348236, "learning_rate": 1.8525842884217053e-05, "loss": 0.0062, "step": 219270 }, { "epoch": 139.75780752071384, "grad_norm": 0.27806562185287476, "learning_rate": 1.852156051068554e-05, "loss": 0.0088, "step": 219280 }, { "epoch": 139.76418100701085, "grad_norm": 0.2482035756111145, "learning_rate": 1.8517278519646784e-05, "loss": 0.0042, "step": 219290 }, { "epoch": 139.77055449330783, "grad_norm": 0.3310551345348358, "learning_rate": 1.851299691115282e-05, "loss": 0.0104, "step": 219300 }, { "epoch": 139.77692797960484, "grad_norm": 0.2113998830318451, "learning_rate": 1.8508715685255657e-05, "loss": 0.0052, "step": 219310 }, { "epoch": 139.78330146590184, "grad_norm": 0.1343783587217331, "learning_rate": 1.8504434842007347e-05, "loss": 0.0063, "step": 219320 }, { "epoch": 139.78967495219885, "grad_norm": 0.1961672455072403, "learning_rate": 1.850015438145989e-05, "loss": 0.0065, "step": 219330 }, { "epoch": 139.79604843849586, "grad_norm": 0.26750287413597107, "learning_rate": 1.8495874303665302e-05, "loss": 0.0058, "step": 219340 }, { "epoch": 139.80242192479287, "grad_norm": 0.17070356011390686, "learning_rate": 1.849159460867556e-05, "loss": 0.0099, "step": 219350 }, { "epoch": 139.80879541108987, "grad_norm": 0.14049670100212097, "learning_rate": 1.8487315296542734e-05, "loss": 0.0059, "step": 219360 }, { "epoch": 139.81516889738688, "grad_norm": 0.1944490522146225, "learning_rate": 1.8483036367318745e-05, "loss": 0.0031, "step": 219370 }, { "epoch": 139.82154238368386, "grad_norm": 0.15455558896064758, "learning_rate": 1.8478757821055635e-05, "loss": 0.007, "step": 219380 }, { "epoch": 139.82791586998087, "grad_norm": 0.28136125206947327, "learning_rate": 1.8474479657805367e-05, "loss": 0.0106, "step": 219390 }, { "epoch": 139.83428935627788, "grad_norm": 0.1702367216348648, "learning_rate": 1.8470201877619965e-05, "loss": 0.0126, "step": 219400 }, { "epoch": 139.8406628425749, "grad_norm": 0.14116229116916656, "learning_rate": 1.8465924480551343e-05, "loss": 0.0032, "step": 219410 }, { "epoch": 139.8470363288719, "grad_norm": 0.34252244234085083, "learning_rate": 1.846164746665154e-05, "loss": 0.0058, "step": 219420 }, { "epoch": 139.8534098151689, "grad_norm": 0.19945086538791656, "learning_rate": 1.845737083597247e-05, "loss": 0.0044, "step": 219430 }, { "epoch": 139.8597833014659, "grad_norm": 0.1445310264825821, "learning_rate": 1.845309458856616e-05, "loss": 0.0099, "step": 219440 }, { "epoch": 139.86615678776292, "grad_norm": 0.2692975103855133, "learning_rate": 1.8448818724484506e-05, "loss": 0.006, "step": 219450 }, { "epoch": 139.8725302740599, "grad_norm": 0.13006673753261566, "learning_rate": 1.844454324377951e-05, "loss": 0.0037, "step": 219460 }, { "epoch": 139.8789037603569, "grad_norm": 0.09103123098611832, "learning_rate": 1.8440268146503103e-05, "loss": 0.0042, "step": 219470 }, { "epoch": 139.8852772466539, "grad_norm": 0.1021481305360794, "learning_rate": 1.8435993432707244e-05, "loss": 0.0179, "step": 219480 }, { "epoch": 139.89165073295092, "grad_norm": 0.3389931917190552, "learning_rate": 1.843171910244384e-05, "loss": 0.0066, "step": 219490 }, { "epoch": 139.89802421924793, "grad_norm": 0.2301708161830902, "learning_rate": 1.842744515576488e-05, "loss": 0.0142, "step": 219500 }, { "epoch": 139.90439770554494, "grad_norm": 0.136734738945961, "learning_rate": 1.842317159272226e-05, "loss": 0.012, "step": 219510 }, { "epoch": 139.91077119184195, "grad_norm": 0.3402259945869446, "learning_rate": 1.8418898413367923e-05, "loss": 0.0115, "step": 219520 }, { "epoch": 139.91714467813895, "grad_norm": 0.1586933434009552, "learning_rate": 1.8414625617753773e-05, "loss": 0.0056, "step": 219530 }, { "epoch": 139.92351816443593, "grad_norm": 0.36041581630706787, "learning_rate": 1.841035320593178e-05, "loss": 0.0181, "step": 219540 }, { "epoch": 139.92989165073294, "grad_norm": 0.1633589267730713, "learning_rate": 1.8406081177953783e-05, "loss": 0.0044, "step": 219550 }, { "epoch": 139.93626513702995, "grad_norm": 0.37941303849220276, "learning_rate": 1.8401809533871743e-05, "loss": 0.0079, "step": 219560 }, { "epoch": 139.94263862332696, "grad_norm": 0.14009816944599152, "learning_rate": 1.839753827373754e-05, "loss": 0.0043, "step": 219570 }, { "epoch": 139.94901210962396, "grad_norm": 0.08074986934661865, "learning_rate": 1.8393267397603115e-05, "loss": 0.0106, "step": 219580 }, { "epoch": 139.95538559592097, "grad_norm": 0.229388028383255, "learning_rate": 1.83889969055203e-05, "loss": 0.0133, "step": 219590 }, { "epoch": 139.96175908221798, "grad_norm": 0.2347557097673416, "learning_rate": 1.8384726797541034e-05, "loss": 0.009, "step": 219600 }, { "epoch": 139.968132568515, "grad_norm": 0.13059447705745697, "learning_rate": 1.8380457073717193e-05, "loss": 0.0158, "step": 219610 }, { "epoch": 139.97450605481197, "grad_norm": 0.23827268183231354, "learning_rate": 1.8376187734100646e-05, "loss": 0.0145, "step": 219620 }, { "epoch": 139.98087954110898, "grad_norm": 0.21870656311511993, "learning_rate": 1.8371918778743262e-05, "loss": 0.0056, "step": 219630 }, { "epoch": 139.98725302740598, "grad_norm": 0.2799355089664459, "learning_rate": 1.836765020769694e-05, "loss": 0.007, "step": 219640 }, { "epoch": 139.993626513703, "grad_norm": 0.3223740756511688, "learning_rate": 1.8363382021013537e-05, "loss": 0.0057, "step": 219650 }, { "epoch": 140.0, "grad_norm": 0.28990092873573303, "learning_rate": 1.8359114218744917e-05, "loss": 0.004, "step": 219660 }, { "epoch": 140.006373486297, "grad_norm": 0.15900322794914246, "learning_rate": 1.8354846800942908e-05, "loss": 0.0142, "step": 219670 }, { "epoch": 140.01274697259402, "grad_norm": 0.3820612132549286, "learning_rate": 1.8350579767659427e-05, "loss": 0.0102, "step": 219680 }, { "epoch": 140.01912045889102, "grad_norm": 0.08930078148841858, "learning_rate": 1.8346313118946245e-05, "loss": 0.0072, "step": 219690 }, { "epoch": 140.02549394518803, "grad_norm": 0.2837715744972229, "learning_rate": 1.8342046854855267e-05, "loss": 0.0095, "step": 219700 }, { "epoch": 140.031867431485, "grad_norm": 0.16587671637535095, "learning_rate": 1.833778097543829e-05, "loss": 0.0101, "step": 219710 }, { "epoch": 140.03824091778202, "grad_norm": 0.1795523762702942, "learning_rate": 1.8333515480747204e-05, "loss": 0.0103, "step": 219720 }, { "epoch": 140.04461440407903, "grad_norm": 0.4369167387485504, "learning_rate": 1.8329250370833766e-05, "loss": 0.0109, "step": 219730 }, { "epoch": 140.05098789037604, "grad_norm": 0.2244371473789215, "learning_rate": 1.8324985645749852e-05, "loss": 0.0098, "step": 219740 }, { "epoch": 140.05736137667304, "grad_norm": 0.08254300802946091, "learning_rate": 1.832072130554726e-05, "loss": 0.0093, "step": 219750 }, { "epoch": 140.06373486297005, "grad_norm": 0.12259356677532196, "learning_rate": 1.8316457350277843e-05, "loss": 0.0076, "step": 219760 }, { "epoch": 140.07010834926706, "grad_norm": 0.1291574239730835, "learning_rate": 1.8312193779993354e-05, "loss": 0.0059, "step": 219770 }, { "epoch": 140.07648183556407, "grad_norm": 0.3057321012020111, "learning_rate": 1.830793059474564e-05, "loss": 0.0053, "step": 219780 }, { "epoch": 140.08285532186105, "grad_norm": 0.4750802218914032, "learning_rate": 1.83036677945865e-05, "loss": 0.0117, "step": 219790 }, { "epoch": 140.08922880815805, "grad_norm": 0.4473167955875397, "learning_rate": 1.8299405379567712e-05, "loss": 0.0076, "step": 219800 }, { "epoch": 140.09560229445506, "grad_norm": 0.15109719336032867, "learning_rate": 1.8295143349741067e-05, "loss": 0.0097, "step": 219810 }, { "epoch": 140.10197578075207, "grad_norm": 0.08358108997344971, "learning_rate": 1.8290881705158387e-05, "loss": 0.0031, "step": 219820 }, { "epoch": 140.10834926704908, "grad_norm": 0.4532151520252228, "learning_rate": 1.8286620445871428e-05, "loss": 0.0085, "step": 219830 }, { "epoch": 140.1147227533461, "grad_norm": 0.15407487750053406, "learning_rate": 1.8282359571931968e-05, "loss": 0.0032, "step": 219840 }, { "epoch": 140.1210962396431, "grad_norm": 0.35642197728157043, "learning_rate": 1.8278099083391777e-05, "loss": 0.0073, "step": 219850 }, { "epoch": 140.1274697259401, "grad_norm": 0.3660007417201996, "learning_rate": 1.8273838980302643e-05, "loss": 0.0134, "step": 219860 }, { "epoch": 140.13384321223708, "grad_norm": 0.376760870218277, "learning_rate": 1.8269579262716323e-05, "loss": 0.0056, "step": 219870 }, { "epoch": 140.1402166985341, "grad_norm": 0.22132347524166107, "learning_rate": 1.826531993068457e-05, "loss": 0.0047, "step": 219880 }, { "epoch": 140.1465901848311, "grad_norm": 0.10501905530691147, "learning_rate": 1.8261060984259133e-05, "loss": 0.0066, "step": 219890 }, { "epoch": 140.1529636711281, "grad_norm": 0.11961213499307632, "learning_rate": 1.825680242349178e-05, "loss": 0.0058, "step": 219900 }, { "epoch": 140.1593371574251, "grad_norm": 0.08727774769067764, "learning_rate": 1.8252544248434254e-05, "loss": 0.0127, "step": 219910 }, { "epoch": 140.16571064372212, "grad_norm": 0.17957930266857147, "learning_rate": 1.8248286459138286e-05, "loss": 0.0089, "step": 219920 }, { "epoch": 140.17208413001913, "grad_norm": 0.10946113616228104, "learning_rate": 1.824402905565562e-05, "loss": 0.0105, "step": 219930 }, { "epoch": 140.17845761631614, "grad_norm": 0.21024806797504425, "learning_rate": 1.823977203803796e-05, "loss": 0.0065, "step": 219940 }, { "epoch": 140.18483110261312, "grad_norm": 0.18524643778800964, "learning_rate": 1.8235515406337083e-05, "loss": 0.0079, "step": 219950 }, { "epoch": 140.19120458891013, "grad_norm": 0.2649495601654053, "learning_rate": 1.823125916060468e-05, "loss": 0.0073, "step": 219960 }, { "epoch": 140.19757807520713, "grad_norm": 0.22991152107715607, "learning_rate": 1.822700330089247e-05, "loss": 0.0097, "step": 219970 }, { "epoch": 140.20395156150414, "grad_norm": 0.153776153922081, "learning_rate": 1.8222747827252156e-05, "loss": 0.0032, "step": 219980 }, { "epoch": 140.21032504780115, "grad_norm": 0.22328341007232666, "learning_rate": 1.8218492739735472e-05, "loss": 0.0064, "step": 219990 }, { "epoch": 140.21669853409816, "grad_norm": 0.21897631883621216, "learning_rate": 1.821423803839411e-05, "loss": 0.0053, "step": 220000 }, { "epoch": 140.22307202039516, "grad_norm": 0.13603763282299042, "learning_rate": 1.8209983723279767e-05, "loss": 0.0064, "step": 220010 }, { "epoch": 140.22944550669217, "grad_norm": 0.28398388624191284, "learning_rate": 1.8205729794444114e-05, "loss": 0.0037, "step": 220020 }, { "epoch": 140.23581899298915, "grad_norm": 0.17632527649402618, "learning_rate": 1.8201476251938887e-05, "loss": 0.0041, "step": 220030 }, { "epoch": 140.24219247928616, "grad_norm": 0.13595964014530182, "learning_rate": 1.8197223095815742e-05, "loss": 0.0043, "step": 220040 }, { "epoch": 140.24856596558317, "grad_norm": 0.21384021639823914, "learning_rate": 1.8192970326126368e-05, "loss": 0.0092, "step": 220050 }, { "epoch": 140.25493945188018, "grad_norm": 0.17454147338867188, "learning_rate": 1.818871794292241e-05, "loss": 0.0045, "step": 220060 }, { "epoch": 140.26131293817718, "grad_norm": 0.3004683256149292, "learning_rate": 1.8184465946255592e-05, "loss": 0.0134, "step": 220070 }, { "epoch": 140.2676864244742, "grad_norm": 0.2159430831670761, "learning_rate": 1.8180214336177542e-05, "loss": 0.0099, "step": 220080 }, { "epoch": 140.2740599107712, "grad_norm": 0.28665652871131897, "learning_rate": 1.8175963112739935e-05, "loss": 0.0077, "step": 220090 }, { "epoch": 140.2804333970682, "grad_norm": 0.3425051271915436, "learning_rate": 1.81717122759944e-05, "loss": 0.0093, "step": 220100 }, { "epoch": 140.2868068833652, "grad_norm": 0.3561354875564575, "learning_rate": 1.8167461825992654e-05, "loss": 0.0085, "step": 220110 }, { "epoch": 140.2931803696622, "grad_norm": 1.785398006439209, "learning_rate": 1.8163211762786264e-05, "loss": 0.011, "step": 220120 }, { "epoch": 140.2995538559592, "grad_norm": 0.21258361637592316, "learning_rate": 1.815896208642693e-05, "loss": 0.003, "step": 220130 }, { "epoch": 140.3059273422562, "grad_norm": 0.3343275487422943, "learning_rate": 1.815471279696625e-05, "loss": 0.0122, "step": 220140 }, { "epoch": 140.31230082855322, "grad_norm": 0.3418177366256714, "learning_rate": 1.815046389445591e-05, "loss": 0.0096, "step": 220150 }, { "epoch": 140.31867431485023, "grad_norm": 0.32580965757369995, "learning_rate": 1.8146215378947474e-05, "loss": 0.0079, "step": 220160 }, { "epoch": 140.32504780114724, "grad_norm": 0.21050234138965607, "learning_rate": 1.814196725049261e-05, "loss": 0.0118, "step": 220170 }, { "epoch": 140.33142128744424, "grad_norm": 0.4407993257045746, "learning_rate": 1.813771950914292e-05, "loss": 0.0119, "step": 220180 }, { "epoch": 140.33779477374122, "grad_norm": 0.370173841714859, "learning_rate": 1.8133472154950026e-05, "loss": 0.008, "step": 220190 }, { "epoch": 140.34416826003823, "grad_norm": 0.24275973439216614, "learning_rate": 1.8129225187965516e-05, "loss": 0.0079, "step": 220200 }, { "epoch": 140.35054174633524, "grad_norm": 0.2351212501525879, "learning_rate": 1.8124978608241024e-05, "loss": 0.0117, "step": 220210 }, { "epoch": 140.35691523263225, "grad_norm": 0.3334778845310211, "learning_rate": 1.8120732415828133e-05, "loss": 0.0081, "step": 220220 }, { "epoch": 140.36328871892925, "grad_norm": 0.2576521337032318, "learning_rate": 1.811648661077845e-05, "loss": 0.0041, "step": 220230 }, { "epoch": 140.36966220522626, "grad_norm": 0.18263818323612213, "learning_rate": 1.8112241193143537e-05, "loss": 0.0108, "step": 220240 }, { "epoch": 140.37603569152327, "grad_norm": 0.27221304178237915, "learning_rate": 1.8107996162975037e-05, "loss": 0.0056, "step": 220250 }, { "epoch": 140.38240917782028, "grad_norm": 0.19185245037078857, "learning_rate": 1.810375152032446e-05, "loss": 0.0043, "step": 220260 }, { "epoch": 140.38878266411726, "grad_norm": 0.268640398979187, "learning_rate": 1.8099507265243432e-05, "loss": 0.007, "step": 220270 }, { "epoch": 140.39515615041427, "grad_norm": 0.2392994612455368, "learning_rate": 1.8095263397783496e-05, "loss": 0.0083, "step": 220280 }, { "epoch": 140.40152963671127, "grad_norm": 0.41893985867500305, "learning_rate": 1.8091019917996265e-05, "loss": 0.0172, "step": 220290 }, { "epoch": 140.40790312300828, "grad_norm": 0.45640134811401367, "learning_rate": 1.8086776825933237e-05, "loss": 0.0084, "step": 220300 }, { "epoch": 140.4142766093053, "grad_norm": 0.13613304495811462, "learning_rate": 1.8082534121646018e-05, "loss": 0.0059, "step": 220310 }, { "epoch": 140.4206500956023, "grad_norm": 0.17509913444519043, "learning_rate": 1.8078291805186143e-05, "loss": 0.0074, "step": 220320 }, { "epoch": 140.4270235818993, "grad_norm": 0.22732633352279663, "learning_rate": 1.807404987660516e-05, "loss": 0.0077, "step": 220330 }, { "epoch": 140.4333970681963, "grad_norm": 0.1925801932811737, "learning_rate": 1.8069808335954598e-05, "loss": 0.0065, "step": 220340 }, { "epoch": 140.43977055449332, "grad_norm": 0.3293555974960327, "learning_rate": 1.8065567183286032e-05, "loss": 0.0077, "step": 220350 }, { "epoch": 140.4461440407903, "grad_norm": 0.3537825047969818, "learning_rate": 1.806132641865097e-05, "loss": 0.0052, "step": 220360 }, { "epoch": 140.4525175270873, "grad_norm": 0.22608786821365356, "learning_rate": 1.805708604210095e-05, "loss": 0.0151, "step": 220370 }, { "epoch": 140.45889101338432, "grad_norm": 0.2393665313720703, "learning_rate": 1.805284605368748e-05, "loss": 0.0071, "step": 220380 }, { "epoch": 140.46526449968133, "grad_norm": 0.18072739243507385, "learning_rate": 1.804860645346211e-05, "loss": 0.0044, "step": 220390 }, { "epoch": 140.47163798597833, "grad_norm": 0.23658569157123566, "learning_rate": 1.8044367241476333e-05, "loss": 0.0068, "step": 220400 }, { "epoch": 140.47801147227534, "grad_norm": 0.14497070014476776, "learning_rate": 1.804012841778167e-05, "loss": 0.0073, "step": 220410 }, { "epoch": 140.48438495857235, "grad_norm": 0.5404289960861206, "learning_rate": 1.8035889982429603e-05, "loss": 0.0067, "step": 220420 }, { "epoch": 140.49075844486936, "grad_norm": 0.1227465346455574, "learning_rate": 1.803165193547169e-05, "loss": 0.0066, "step": 220430 }, { "epoch": 140.49713193116634, "grad_norm": 0.5098342299461365, "learning_rate": 1.8027414276959348e-05, "loss": 0.0095, "step": 220440 }, { "epoch": 140.50350541746334, "grad_norm": 0.3183595538139343, "learning_rate": 1.802317700694413e-05, "loss": 0.0058, "step": 220450 }, { "epoch": 140.50987890376035, "grad_norm": 0.3361707329750061, "learning_rate": 1.801894012547749e-05, "loss": 0.0064, "step": 220460 }, { "epoch": 140.51625239005736, "grad_norm": 0.14095421135425568, "learning_rate": 1.801470363261095e-05, "loss": 0.0069, "step": 220470 }, { "epoch": 140.52262587635437, "grad_norm": 0.11626644432544708, "learning_rate": 1.801046752839593e-05, "loss": 0.0088, "step": 220480 }, { "epoch": 140.52899936265138, "grad_norm": 0.35819700360298157, "learning_rate": 1.8006231812883945e-05, "loss": 0.0066, "step": 220490 }, { "epoch": 140.53537284894838, "grad_norm": 0.3182438313961029, "learning_rate": 1.8001996486126456e-05, "loss": 0.0132, "step": 220500 }, { "epoch": 140.5417463352454, "grad_norm": 0.20108546316623688, "learning_rate": 1.7997761548174918e-05, "loss": 0.0155, "step": 220510 }, { "epoch": 140.54811982154237, "grad_norm": 0.3757680654525757, "learning_rate": 1.7993526999080773e-05, "loss": 0.0129, "step": 220520 }, { "epoch": 140.55449330783938, "grad_norm": 0.1651206910610199, "learning_rate": 1.7989292838895517e-05, "loss": 0.0079, "step": 220530 }, { "epoch": 140.5608667941364, "grad_norm": 0.644814670085907, "learning_rate": 1.7985059067670573e-05, "loss": 0.009, "step": 220540 }, { "epoch": 140.5672402804334, "grad_norm": 0.29670819640159607, "learning_rate": 1.7980825685457385e-05, "loss": 0.0044, "step": 220550 }, { "epoch": 140.5736137667304, "grad_norm": 0.19124442338943481, "learning_rate": 1.797659269230739e-05, "loss": 0.0077, "step": 220560 }, { "epoch": 140.5799872530274, "grad_norm": 0.19124968349933624, "learning_rate": 1.7972360088272032e-05, "loss": 0.0041, "step": 220570 }, { "epoch": 140.58636073932442, "grad_norm": 0.1938992142677307, "learning_rate": 1.7968127873402745e-05, "loss": 0.0105, "step": 220580 }, { "epoch": 140.59273422562143, "grad_norm": 0.25826480984687805, "learning_rate": 1.796389604775095e-05, "loss": 0.0186, "step": 220590 }, { "epoch": 140.5991077119184, "grad_norm": 0.10563621670007706, "learning_rate": 1.7959664611368044e-05, "loss": 0.0049, "step": 220600 }, { "epoch": 140.60548119821541, "grad_norm": 0.2666182816028595, "learning_rate": 1.795543356430548e-05, "loss": 0.0089, "step": 220610 }, { "epoch": 140.61185468451242, "grad_norm": 0.14325915277004242, "learning_rate": 1.7951202906614655e-05, "loss": 0.0064, "step": 220620 }, { "epoch": 140.61822817080943, "grad_norm": 0.21860522031784058, "learning_rate": 1.7946972638346976e-05, "loss": 0.0047, "step": 220630 }, { "epoch": 140.62460165710644, "grad_norm": 0.0965813398361206, "learning_rate": 1.7942742759553822e-05, "loss": 0.0046, "step": 220640 }, { "epoch": 140.63097514340345, "grad_norm": 0.1474827527999878, "learning_rate": 1.7938513270286628e-05, "loss": 0.0067, "step": 220650 }, { "epoch": 140.63734862970045, "grad_norm": 0.13083991408348083, "learning_rate": 1.793428417059677e-05, "loss": 0.0058, "step": 220660 }, { "epoch": 140.64372211599746, "grad_norm": 0.19891034066677094, "learning_rate": 1.793005546053564e-05, "loss": 0.0063, "step": 220670 }, { "epoch": 140.65009560229444, "grad_norm": 0.13821633160114288, "learning_rate": 1.7925827140154606e-05, "loss": 0.0079, "step": 220680 }, { "epoch": 140.65646908859145, "grad_norm": 0.36994603276252747, "learning_rate": 1.792159920950504e-05, "loss": 0.0066, "step": 220690 }, { "epoch": 140.66284257488846, "grad_norm": 0.33603814244270325, "learning_rate": 1.791737166863835e-05, "loss": 0.0087, "step": 220700 }, { "epoch": 140.66921606118547, "grad_norm": 0.21013738214969635, "learning_rate": 1.7913144517605883e-05, "loss": 0.0057, "step": 220710 }, { "epoch": 140.67558954748247, "grad_norm": 0.2536827027797699, "learning_rate": 1.790891775645901e-05, "loss": 0.0041, "step": 220720 }, { "epoch": 140.68196303377948, "grad_norm": 0.16587987542152405, "learning_rate": 1.7904691385249067e-05, "loss": 0.0073, "step": 220730 }, { "epoch": 140.6883365200765, "grad_norm": 0.1934317648410797, "learning_rate": 1.7900465404027438e-05, "loss": 0.0049, "step": 220740 }, { "epoch": 140.6947100063735, "grad_norm": 0.32700344920158386, "learning_rate": 1.7896239812845467e-05, "loss": 0.0121, "step": 220750 }, { "epoch": 140.70108349267048, "grad_norm": 0.4302576184272766, "learning_rate": 1.7892014611754492e-05, "loss": 0.0056, "step": 220760 }, { "epoch": 140.70745697896749, "grad_norm": 0.171657532453537, "learning_rate": 1.7887789800805838e-05, "loss": 0.0047, "step": 220770 }, { "epoch": 140.7138304652645, "grad_norm": 0.2033928483724594, "learning_rate": 1.788356538005087e-05, "loss": 0.0054, "step": 220780 }, { "epoch": 140.7202039515615, "grad_norm": 0.2042166143655777, "learning_rate": 1.7879341349540908e-05, "loss": 0.0068, "step": 220790 }, { "epoch": 140.7265774378585, "grad_norm": 0.42688652873039246, "learning_rate": 1.7875117709327278e-05, "loss": 0.0136, "step": 220800 }, { "epoch": 140.73295092415552, "grad_norm": 0.6247978210449219, "learning_rate": 1.7870894459461278e-05, "loss": 0.0133, "step": 220810 }, { "epoch": 140.73932441045253, "grad_norm": 0.27445662021636963, "learning_rate": 1.7866671599994277e-05, "loss": 0.0054, "step": 220820 }, { "epoch": 140.74569789674953, "grad_norm": 0.3651157319545746, "learning_rate": 1.7862449130977526e-05, "loss": 0.0081, "step": 220830 }, { "epoch": 140.7520713830465, "grad_norm": 0.1788710653781891, "learning_rate": 1.785822705246238e-05, "loss": 0.0084, "step": 220840 }, { "epoch": 140.75844486934352, "grad_norm": 0.3529968559741974, "learning_rate": 1.78540053645001e-05, "loss": 0.0082, "step": 220850 }, { "epoch": 140.76481835564053, "grad_norm": 0.29116302728652954, "learning_rate": 1.7849784067142038e-05, "loss": 0.0107, "step": 220860 }, { "epoch": 140.77119184193754, "grad_norm": 0.16405470669269562, "learning_rate": 1.7845563160439427e-05, "loss": 0.0066, "step": 220870 }, { "epoch": 140.77756532823454, "grad_norm": 0.16317057609558105, "learning_rate": 1.7841342644443598e-05, "loss": 0.0108, "step": 220880 }, { "epoch": 140.78393881453155, "grad_norm": 0.17679089307785034, "learning_rate": 1.783712251920582e-05, "loss": 0.0021, "step": 220890 }, { "epoch": 140.79031230082856, "grad_norm": 0.14910757541656494, "learning_rate": 1.7832902784777363e-05, "loss": 0.0081, "step": 220900 }, { "epoch": 140.79668578712557, "grad_norm": 0.24338221549987793, "learning_rate": 1.7828683441209492e-05, "loss": 0.0041, "step": 220910 }, { "epoch": 140.80305927342255, "grad_norm": 0.34123772382736206, "learning_rate": 1.7824464488553517e-05, "loss": 0.0052, "step": 220920 }, { "epoch": 140.80943275971956, "grad_norm": 0.1503811776638031, "learning_rate": 1.782024592686067e-05, "loss": 0.0082, "step": 220930 }, { "epoch": 140.81580624601656, "grad_norm": 0.11682815849781036, "learning_rate": 1.781602775618222e-05, "loss": 0.0138, "step": 220940 }, { "epoch": 140.82217973231357, "grad_norm": 0.21228380501270294, "learning_rate": 1.78118099765694e-05, "loss": 0.0049, "step": 220950 }, { "epoch": 140.82855321861058, "grad_norm": 0.4477880299091339, "learning_rate": 1.78075925880735e-05, "loss": 0.0061, "step": 220960 }, { "epoch": 140.8349267049076, "grad_norm": 0.19177992641925812, "learning_rate": 1.7803375590745746e-05, "loss": 0.0113, "step": 220970 }, { "epoch": 140.8413001912046, "grad_norm": 0.1381942629814148, "learning_rate": 1.7799158984637372e-05, "loss": 0.0056, "step": 220980 }, { "epoch": 140.8476736775016, "grad_norm": 0.32704535126686096, "learning_rate": 1.7794942769799605e-05, "loss": 0.0069, "step": 220990 }, { "epoch": 140.8540471637986, "grad_norm": 0.13372333347797394, "learning_rate": 1.7790726946283725e-05, "loss": 0.0147, "step": 221000 }, { "epoch": 140.8604206500956, "grad_norm": 0.14433011412620544, "learning_rate": 1.7786511514140884e-05, "loss": 0.0059, "step": 221010 }, { "epoch": 140.8667941363926, "grad_norm": 0.39552754163742065, "learning_rate": 1.7782296473422367e-05, "loss": 0.0172, "step": 221020 }, { "epoch": 140.8731676226896, "grad_norm": 0.1536208987236023, "learning_rate": 1.7778081824179344e-05, "loss": 0.0081, "step": 221030 }, { "epoch": 140.87954110898661, "grad_norm": 0.08347138017416, "learning_rate": 1.7773867566463088e-05, "loss": 0.0103, "step": 221040 }, { "epoch": 140.88591459528362, "grad_norm": 0.20116014778614044, "learning_rate": 1.776965370032473e-05, "loss": 0.0065, "step": 221050 }, { "epoch": 140.89228808158063, "grad_norm": 0.10449323803186417, "learning_rate": 1.7765440225815528e-05, "loss": 0.006, "step": 221060 }, { "epoch": 140.89866156787764, "grad_norm": 0.2492792159318924, "learning_rate": 1.7761227142986665e-05, "loss": 0.0104, "step": 221070 }, { "epoch": 140.90503505417465, "grad_norm": 0.18395520746707916, "learning_rate": 1.7757014451889324e-05, "loss": 0.0049, "step": 221080 }, { "epoch": 140.91140854047163, "grad_norm": 0.23973874747753143, "learning_rate": 1.775280215257469e-05, "loss": 0.0081, "step": 221090 }, { "epoch": 140.91778202676863, "grad_norm": 0.4157696068286896, "learning_rate": 1.7748590245093965e-05, "loss": 0.0094, "step": 221100 }, { "epoch": 140.92415551306564, "grad_norm": 0.2097785770893097, "learning_rate": 1.7744378729498322e-05, "loss": 0.0116, "step": 221110 }, { "epoch": 140.93052899936265, "grad_norm": 0.4149860143661499, "learning_rate": 1.7740167605838936e-05, "loss": 0.0075, "step": 221120 }, { "epoch": 140.93690248565966, "grad_norm": 0.265379399061203, "learning_rate": 1.7735956874166954e-05, "loss": 0.0095, "step": 221130 }, { "epoch": 140.94327597195667, "grad_norm": 0.21831855177879333, "learning_rate": 1.773174653453359e-05, "loss": 0.009, "step": 221140 }, { "epoch": 140.94964945825367, "grad_norm": 0.20626279711723328, "learning_rate": 1.7727536586989934e-05, "loss": 0.0062, "step": 221150 }, { "epoch": 140.95602294455068, "grad_norm": 0.2902730107307434, "learning_rate": 1.77233270315872e-05, "loss": 0.0119, "step": 221160 }, { "epoch": 140.96239643084766, "grad_norm": 0.10507030785083771, "learning_rate": 1.77191178683765e-05, "loss": 0.0076, "step": 221170 }, { "epoch": 140.96876991714467, "grad_norm": 0.21923404932022095, "learning_rate": 1.7714909097409033e-05, "loss": 0.0067, "step": 221180 }, { "epoch": 140.97514340344168, "grad_norm": 0.5707032680511475, "learning_rate": 1.771070071873586e-05, "loss": 0.0142, "step": 221190 }, { "epoch": 140.98151688973869, "grad_norm": 0.34689030051231384, "learning_rate": 1.7706492732408185e-05, "loss": 0.0065, "step": 221200 }, { "epoch": 140.9878903760357, "grad_norm": 0.3461723029613495, "learning_rate": 1.770228513847711e-05, "loss": 0.0075, "step": 221210 }, { "epoch": 140.9942638623327, "grad_norm": 0.143911674618721, "learning_rate": 1.7698077936993768e-05, "loss": 0.0059, "step": 221220 }, { "epoch": 141.0006373486297, "grad_norm": 0.2613365352153778, "learning_rate": 1.7693871128009254e-05, "loss": 0.0088, "step": 221230 }, { "epoch": 141.00701083492672, "grad_norm": 0.18823301792144775, "learning_rate": 1.7689664711574726e-05, "loss": 0.0135, "step": 221240 }, { "epoch": 141.0133843212237, "grad_norm": 0.1550752818584442, "learning_rate": 1.7685458687741287e-05, "loss": 0.0073, "step": 221250 }, { "epoch": 141.0197578075207, "grad_norm": 0.22006835043430328, "learning_rate": 1.768125305656002e-05, "loss": 0.0053, "step": 221260 }, { "epoch": 141.0261312938177, "grad_norm": 0.3705533444881439, "learning_rate": 1.7677047818082038e-05, "loss": 0.0069, "step": 221270 }, { "epoch": 141.03250478011472, "grad_norm": 0.15314915776252747, "learning_rate": 1.7672842972358454e-05, "loss": 0.0045, "step": 221280 }, { "epoch": 141.03887826641173, "grad_norm": 0.15448123216629028, "learning_rate": 1.766863851944035e-05, "loss": 0.0047, "step": 221290 }, { "epoch": 141.04525175270874, "grad_norm": 0.19619181752204895, "learning_rate": 1.7664434459378815e-05, "loss": 0.0063, "step": 221300 }, { "epoch": 141.05162523900574, "grad_norm": 0.19002564251422882, "learning_rate": 1.7660230792224914e-05, "loss": 0.0092, "step": 221310 }, { "epoch": 141.05799872530275, "grad_norm": 0.29226505756378174, "learning_rate": 1.765602751802976e-05, "loss": 0.0168, "step": 221320 }, { "epoch": 141.06437221159973, "grad_norm": 0.13016323745250702, "learning_rate": 1.7651824636844406e-05, "loss": 0.0038, "step": 221330 }, { "epoch": 141.07074569789674, "grad_norm": 0.23661595582962036, "learning_rate": 1.7647622148719928e-05, "loss": 0.0053, "step": 221340 }, { "epoch": 141.07711918419375, "grad_norm": 0.31940537691116333, "learning_rate": 1.764342005370737e-05, "loss": 0.0053, "step": 221350 }, { "epoch": 141.08349267049076, "grad_norm": 0.22778432071208954, "learning_rate": 1.7639218351857823e-05, "loss": 0.0065, "step": 221360 }, { "epoch": 141.08986615678776, "grad_norm": 0.23359635472297668, "learning_rate": 1.763501704322233e-05, "loss": 0.0075, "step": 221370 }, { "epoch": 141.09623964308477, "grad_norm": 0.31433185935020447, "learning_rate": 1.7630816127851933e-05, "loss": 0.0056, "step": 221380 }, { "epoch": 141.10261312938178, "grad_norm": 0.5687741637229919, "learning_rate": 1.7626615605797682e-05, "loss": 0.008, "step": 221390 }, { "epoch": 141.1089866156788, "grad_norm": 0.1416187584400177, "learning_rate": 1.7622415477110603e-05, "loss": 0.0074, "step": 221400 }, { "epoch": 141.11536010197577, "grad_norm": 0.3068394064903259, "learning_rate": 1.761821574184176e-05, "loss": 0.0054, "step": 221410 }, { "epoch": 141.12173358827278, "grad_norm": 0.22615231573581696, "learning_rate": 1.761401640004217e-05, "loss": 0.0094, "step": 221420 }, { "epoch": 141.12810707456978, "grad_norm": 0.3635083734989166, "learning_rate": 1.7609817451762863e-05, "loss": 0.0071, "step": 221430 }, { "epoch": 141.1344805608668, "grad_norm": 0.24417531490325928, "learning_rate": 1.7605618897054833e-05, "loss": 0.0046, "step": 221440 }, { "epoch": 141.1408540471638, "grad_norm": 0.46074816584587097, "learning_rate": 1.7601420735969132e-05, "loss": 0.0096, "step": 221450 }, { "epoch": 141.1472275334608, "grad_norm": 0.13238488137722015, "learning_rate": 1.7597222968556766e-05, "loss": 0.0068, "step": 221460 }, { "epoch": 141.15360101975782, "grad_norm": 0.21848319470882416, "learning_rate": 1.759302559486873e-05, "loss": 0.0111, "step": 221470 }, { "epoch": 141.15997450605482, "grad_norm": 0.3711802065372467, "learning_rate": 1.7588828614956016e-05, "loss": 0.0136, "step": 221480 }, { "epoch": 141.1663479923518, "grad_norm": 0.2637976109981537, "learning_rate": 1.7584632028869646e-05, "loss": 0.0044, "step": 221490 }, { "epoch": 141.1727214786488, "grad_norm": 0.327965646982193, "learning_rate": 1.758043583666061e-05, "loss": 0.0147, "step": 221500 }, { "epoch": 141.17909496494582, "grad_norm": 0.19220030307769775, "learning_rate": 1.757624003837989e-05, "loss": 0.0068, "step": 221510 }, { "epoch": 141.18546845124283, "grad_norm": 0.20436497032642365, "learning_rate": 1.757204463407844e-05, "loss": 0.0169, "step": 221520 }, { "epoch": 141.19184193753983, "grad_norm": 0.21801066398620605, "learning_rate": 1.7567849623807303e-05, "loss": 0.0083, "step": 221530 }, { "epoch": 141.19821542383684, "grad_norm": 0.2860826551914215, "learning_rate": 1.756365500761738e-05, "loss": 0.0059, "step": 221540 }, { "epoch": 141.20458891013385, "grad_norm": 0.11275922507047653, "learning_rate": 1.7559460785559688e-05, "loss": 0.0135, "step": 221550 }, { "epoch": 141.21096239643086, "grad_norm": 0.1271481066942215, "learning_rate": 1.7555266957685162e-05, "loss": 0.0054, "step": 221560 }, { "epoch": 141.21733588272787, "grad_norm": 0.3692796528339386, "learning_rate": 1.7551073524044815e-05, "loss": 0.0087, "step": 221570 }, { "epoch": 141.22370936902485, "grad_norm": 0.12893430888652802, "learning_rate": 1.754688048468952e-05, "loss": 0.0053, "step": 221580 }, { "epoch": 141.23008285532185, "grad_norm": 0.2826440632343292, "learning_rate": 1.7542687839670285e-05, "loss": 0.0068, "step": 221590 }, { "epoch": 141.23645634161886, "grad_norm": 0.19218450784683228, "learning_rate": 1.7538495589038022e-05, "loss": 0.0061, "step": 221600 }, { "epoch": 141.24282982791587, "grad_norm": 0.2343648076057434, "learning_rate": 1.753430373284372e-05, "loss": 0.0079, "step": 221610 }, { "epoch": 141.24920331421288, "grad_norm": 0.28832557797431946, "learning_rate": 1.7530112271138248e-05, "loss": 0.0051, "step": 221620 }, { "epoch": 141.25557680050989, "grad_norm": 0.2484668344259262, "learning_rate": 1.752592120397259e-05, "loss": 0.007, "step": 221630 }, { "epoch": 141.2619502868069, "grad_norm": 0.1306721717119217, "learning_rate": 1.752173053139764e-05, "loss": 0.0052, "step": 221640 }, { "epoch": 141.2683237731039, "grad_norm": 0.0700308233499527, "learning_rate": 1.751754025346434e-05, "loss": 0.0051, "step": 221650 }, { "epoch": 141.27469725940088, "grad_norm": 0.09825640916824341, "learning_rate": 1.751335037022357e-05, "loss": 0.0043, "step": 221660 }, { "epoch": 141.2810707456979, "grad_norm": 0.14058461785316467, "learning_rate": 1.7509160881726277e-05, "loss": 0.0055, "step": 221670 }, { "epoch": 141.2874442319949, "grad_norm": 0.12867912650108337, "learning_rate": 1.7504971788023367e-05, "loss": 0.0164, "step": 221680 }, { "epoch": 141.2938177182919, "grad_norm": 0.17467278242111206, "learning_rate": 1.7500783089165722e-05, "loss": 0.0086, "step": 221690 }, { "epoch": 141.3001912045889, "grad_norm": 0.21897061169147491, "learning_rate": 1.7496594785204236e-05, "loss": 0.0112, "step": 221700 }, { "epoch": 141.30656469088592, "grad_norm": 0.16487781703472137, "learning_rate": 1.7492406876189842e-05, "loss": 0.0106, "step": 221710 }, { "epoch": 141.31293817718293, "grad_norm": 0.3353691101074219, "learning_rate": 1.7488219362173363e-05, "loss": 0.005, "step": 221720 }, { "epoch": 141.31931166347994, "grad_norm": 0.1388065665960312, "learning_rate": 1.7484032243205732e-05, "loss": 0.0219, "step": 221730 }, { "epoch": 141.32568514977692, "grad_norm": 0.19596432149410248, "learning_rate": 1.7479845519337795e-05, "loss": 0.004, "step": 221740 }, { "epoch": 141.33205863607392, "grad_norm": 0.5944831967353821, "learning_rate": 1.747565919062047e-05, "loss": 0.0201, "step": 221750 }, { "epoch": 141.33843212237093, "grad_norm": 0.518428385257721, "learning_rate": 1.747147325710456e-05, "loss": 0.0139, "step": 221760 }, { "epoch": 141.34480560866794, "grad_norm": 0.14762887358665466, "learning_rate": 1.746728771884098e-05, "loss": 0.0082, "step": 221770 }, { "epoch": 141.35117909496495, "grad_norm": 0.3973854184150696, "learning_rate": 1.7463102575880568e-05, "loss": 0.0158, "step": 221780 }, { "epoch": 141.35755258126196, "grad_norm": 0.25409817695617676, "learning_rate": 1.7458917828274186e-05, "loss": 0.0056, "step": 221790 }, { "epoch": 141.36392606755896, "grad_norm": 0.4166826903820038, "learning_rate": 1.745473347607265e-05, "loss": 0.0109, "step": 221800 }, { "epoch": 141.37029955385597, "grad_norm": 0.4113784432411194, "learning_rate": 1.7450549519326853e-05, "loss": 0.0244, "step": 221810 }, { "epoch": 141.37667304015295, "grad_norm": 0.6147809028625488, "learning_rate": 1.7446365958087614e-05, "loss": 0.007, "step": 221820 }, { "epoch": 141.38304652644996, "grad_norm": 0.6864199042320251, "learning_rate": 1.7442182792405757e-05, "loss": 0.0103, "step": 221830 }, { "epoch": 141.38942001274697, "grad_norm": 0.17909379303455353, "learning_rate": 1.7438000022332106e-05, "loss": 0.0084, "step": 221840 }, { "epoch": 141.39579349904398, "grad_norm": 0.20839756727218628, "learning_rate": 1.743381764791751e-05, "loss": 0.0053, "step": 221850 }, { "epoch": 141.40216698534098, "grad_norm": 0.1796688288450241, "learning_rate": 1.7429635669212785e-05, "loss": 0.0051, "step": 221860 }, { "epoch": 141.408540471638, "grad_norm": 0.30249762535095215, "learning_rate": 1.7425454086268732e-05, "loss": 0.0137, "step": 221870 }, { "epoch": 141.414913957935, "grad_norm": 0.30034536123275757, "learning_rate": 1.7421272899136155e-05, "loss": 0.0115, "step": 221880 }, { "epoch": 141.421287444232, "grad_norm": 0.19269877672195435, "learning_rate": 1.7417092107865906e-05, "loss": 0.0044, "step": 221890 }, { "epoch": 141.427660930529, "grad_norm": 0.159627765417099, "learning_rate": 1.741291171250871e-05, "loss": 0.0053, "step": 221900 }, { "epoch": 141.434034416826, "grad_norm": 0.10359925776720047, "learning_rate": 1.740873171311543e-05, "loss": 0.0123, "step": 221910 }, { "epoch": 141.440407903123, "grad_norm": 0.14759159088134766, "learning_rate": 1.740455210973681e-05, "loss": 0.0052, "step": 221920 }, { "epoch": 141.44678138942, "grad_norm": 0.45801323652267456, "learning_rate": 1.7400372902423688e-05, "loss": 0.0064, "step": 221930 }, { "epoch": 141.45315487571702, "grad_norm": 0.43329721689224243, "learning_rate": 1.7396194091226785e-05, "loss": 0.0056, "step": 221940 }, { "epoch": 141.45952836201403, "grad_norm": 0.5555416941642761, "learning_rate": 1.7392015676196922e-05, "loss": 0.0055, "step": 221950 }, { "epoch": 141.46590184831103, "grad_norm": 0.10771285742521286, "learning_rate": 1.738783765738486e-05, "loss": 0.004, "step": 221960 }, { "epoch": 141.47227533460804, "grad_norm": 0.6864635944366455, "learning_rate": 1.7383660034841353e-05, "loss": 0.0094, "step": 221970 }, { "epoch": 141.47864882090502, "grad_norm": 0.1975284069776535, "learning_rate": 1.737948280861716e-05, "loss": 0.0058, "step": 221980 }, { "epoch": 141.48502230720203, "grad_norm": 0.48541972041130066, "learning_rate": 1.7375305978763067e-05, "loss": 0.0092, "step": 221990 }, { "epoch": 141.49139579349904, "grad_norm": 0.14134971797466278, "learning_rate": 1.7371129545329806e-05, "loss": 0.0127, "step": 222000 }, { "epoch": 141.49776927979605, "grad_norm": 0.2569088339805603, "learning_rate": 1.7366953508368128e-05, "loss": 0.0077, "step": 222010 }, { "epoch": 141.50414276609305, "grad_norm": 0.2546255588531494, "learning_rate": 1.736277786792876e-05, "loss": 0.0152, "step": 222020 }, { "epoch": 141.51051625239006, "grad_norm": 0.14736402034759521, "learning_rate": 1.7358602624062477e-05, "loss": 0.0074, "step": 222030 }, { "epoch": 141.51688973868707, "grad_norm": 0.230160653591156, "learning_rate": 1.7354427776819982e-05, "loss": 0.007, "step": 222040 }, { "epoch": 141.52326322498408, "grad_norm": 0.09785234928131104, "learning_rate": 1.735025332625202e-05, "loss": 0.0108, "step": 222050 }, { "epoch": 141.52963671128106, "grad_norm": 0.12944550812244415, "learning_rate": 1.734607927240928e-05, "loss": 0.0033, "step": 222060 }, { "epoch": 141.53601019757807, "grad_norm": 0.2078070193529129, "learning_rate": 1.7341905615342536e-05, "loss": 0.0094, "step": 222070 }, { "epoch": 141.54238368387507, "grad_norm": 0.354458212852478, "learning_rate": 1.7337732355102465e-05, "loss": 0.0145, "step": 222080 }, { "epoch": 141.54875717017208, "grad_norm": 0.1740775853395462, "learning_rate": 1.7333559491739788e-05, "loss": 0.01, "step": 222090 }, { "epoch": 141.5551306564691, "grad_norm": 0.21994371712207794, "learning_rate": 1.7329387025305206e-05, "loss": 0.0087, "step": 222100 }, { "epoch": 141.5615041427661, "grad_norm": 0.6948674321174622, "learning_rate": 1.7325214955849396e-05, "loss": 0.0129, "step": 222110 }, { "epoch": 141.5678776290631, "grad_norm": 0.20706094801425934, "learning_rate": 1.7321043283423092e-05, "loss": 0.0062, "step": 222120 }, { "epoch": 141.5742511153601, "grad_norm": 0.47677382826805115, "learning_rate": 1.7316872008076968e-05, "loss": 0.0071, "step": 222130 }, { "epoch": 141.5806246016571, "grad_norm": 0.3466886878013611, "learning_rate": 1.7312701129861708e-05, "loss": 0.0067, "step": 222140 }, { "epoch": 141.5869980879541, "grad_norm": 0.1289055347442627, "learning_rate": 1.7308530648827974e-05, "loss": 0.0055, "step": 222150 }, { "epoch": 141.5933715742511, "grad_norm": 0.05957920104265213, "learning_rate": 1.7304360565026477e-05, "loss": 0.0033, "step": 222160 }, { "epoch": 141.59974506054812, "grad_norm": 0.382127970457077, "learning_rate": 1.7300190878507865e-05, "loss": 0.0079, "step": 222170 }, { "epoch": 141.60611854684512, "grad_norm": 0.42512884736061096, "learning_rate": 1.729602158932281e-05, "loss": 0.0128, "step": 222180 }, { "epoch": 141.61249203314213, "grad_norm": 0.3287956416606903, "learning_rate": 1.7291852697521954e-05, "loss": 0.009, "step": 222190 }, { "epoch": 141.61886551943914, "grad_norm": 0.1864645928144455, "learning_rate": 1.7287684203155984e-05, "loss": 0.0036, "step": 222200 }, { "epoch": 141.62523900573615, "grad_norm": 0.31574419140815735, "learning_rate": 1.728351610627554e-05, "loss": 0.0044, "step": 222210 }, { "epoch": 141.63161249203316, "grad_norm": 0.13808618485927582, "learning_rate": 1.727934840693126e-05, "loss": 0.0049, "step": 222220 }, { "epoch": 141.63798597833014, "grad_norm": 0.44428083300590515, "learning_rate": 1.7275181105173775e-05, "loss": 0.0064, "step": 222230 }, { "epoch": 141.64435946462714, "grad_norm": 0.6149384379386902, "learning_rate": 1.7271014201053752e-05, "loss": 0.0077, "step": 222240 }, { "epoch": 141.65073295092415, "grad_norm": 0.2584187686443329, "learning_rate": 1.7266847694621814e-05, "loss": 0.0208, "step": 222250 }, { "epoch": 141.65710643722116, "grad_norm": 0.18182040750980377, "learning_rate": 1.726268158592858e-05, "loss": 0.0168, "step": 222260 }, { "epoch": 141.66347992351817, "grad_norm": 0.10008352249860764, "learning_rate": 1.7258515875024657e-05, "loss": 0.006, "step": 222270 }, { "epoch": 141.66985340981518, "grad_norm": 0.1574462503194809, "learning_rate": 1.7254350561960707e-05, "loss": 0.0087, "step": 222280 }, { "epoch": 141.67622689611218, "grad_norm": 0.08413814753293991, "learning_rate": 1.7250185646787286e-05, "loss": 0.0059, "step": 222290 }, { "epoch": 141.6826003824092, "grad_norm": 0.08020920306444168, "learning_rate": 1.7246021129555046e-05, "loss": 0.0152, "step": 222300 }, { "epoch": 141.68897386870617, "grad_norm": 0.3630838096141815, "learning_rate": 1.7241857010314555e-05, "loss": 0.0039, "step": 222310 }, { "epoch": 141.69534735500318, "grad_norm": 0.238372802734375, "learning_rate": 1.723769328911647e-05, "loss": 0.0096, "step": 222320 }, { "epoch": 141.7017208413002, "grad_norm": 0.18882954120635986, "learning_rate": 1.7233529966011308e-05, "loss": 0.0046, "step": 222330 }, { "epoch": 141.7080943275972, "grad_norm": 0.19463486969470978, "learning_rate": 1.7229367041049705e-05, "loss": 0.0111, "step": 222340 }, { "epoch": 141.7144678138942, "grad_norm": 0.29284659028053284, "learning_rate": 1.722520451428224e-05, "loss": 0.0073, "step": 222350 }, { "epoch": 141.7208413001912, "grad_norm": 0.289905846118927, "learning_rate": 1.7221042385759483e-05, "loss": 0.0043, "step": 222360 }, { "epoch": 141.72721478648822, "grad_norm": 0.13547484576702118, "learning_rate": 1.721688065553199e-05, "loss": 0.0098, "step": 222370 }, { "epoch": 141.73358827278523, "grad_norm": 0.6343623995780945, "learning_rate": 1.7212719323650368e-05, "loss": 0.0069, "step": 222380 }, { "epoch": 141.7399617590822, "grad_norm": 0.10618817061185837, "learning_rate": 1.7208558390165163e-05, "loss": 0.0189, "step": 222390 }, { "epoch": 141.7463352453792, "grad_norm": 0.2545408010482788, "learning_rate": 1.7204397855126936e-05, "loss": 0.0058, "step": 222400 }, { "epoch": 141.75270873167622, "grad_norm": 0.08432655036449432, "learning_rate": 1.7200237718586214e-05, "loss": 0.0079, "step": 222410 }, { "epoch": 141.75908221797323, "grad_norm": 0.2514033019542694, "learning_rate": 1.7196077980593617e-05, "loss": 0.0062, "step": 222420 }, { "epoch": 141.76545570427024, "grad_norm": 0.20969851315021515, "learning_rate": 1.71919186411996e-05, "loss": 0.0078, "step": 222430 }, { "epoch": 141.77182919056725, "grad_norm": 0.13189321756362915, "learning_rate": 1.718775970045477e-05, "loss": 0.0073, "step": 222440 }, { "epoch": 141.77820267686425, "grad_norm": 0.24360431730747223, "learning_rate": 1.7183601158409624e-05, "loss": 0.0075, "step": 222450 }, { "epoch": 141.78457616316126, "grad_norm": 0.24878472089767456, "learning_rate": 1.717944301511474e-05, "loss": 0.0096, "step": 222460 }, { "epoch": 141.79094964945824, "grad_norm": 0.44320568442344666, "learning_rate": 1.717528527062058e-05, "loss": 0.006, "step": 222470 }, { "epoch": 141.79732313575525, "grad_norm": 0.32331493496894836, "learning_rate": 1.7171127924977708e-05, "loss": 0.0047, "step": 222480 }, { "epoch": 141.80369662205226, "grad_norm": 0.18596979975700378, "learning_rate": 1.7166970978236614e-05, "loss": 0.006, "step": 222490 }, { "epoch": 141.81007010834927, "grad_norm": 0.08792056143283844, "learning_rate": 1.716281443044785e-05, "loss": 0.0071, "step": 222500 }, { "epoch": 141.81644359464627, "grad_norm": 0.17489083111286163, "learning_rate": 1.7158658281661864e-05, "loss": 0.0052, "step": 222510 }, { "epoch": 141.82281708094328, "grad_norm": 0.35323137044906616, "learning_rate": 1.7154502531929205e-05, "loss": 0.0087, "step": 222520 }, { "epoch": 141.8291905672403, "grad_norm": 0.15492358803749084, "learning_rate": 1.715034718130035e-05, "loss": 0.0059, "step": 222530 }, { "epoch": 141.8355640535373, "grad_norm": 0.13853055238723755, "learning_rate": 1.7146192229825792e-05, "loss": 0.0045, "step": 222540 }, { "epoch": 141.84193753983428, "grad_norm": 0.5773128271102905, "learning_rate": 1.7142037677556e-05, "loss": 0.011, "step": 222550 }, { "epoch": 141.84831102613128, "grad_norm": 0.300005704164505, "learning_rate": 1.713788352454149e-05, "loss": 0.0063, "step": 222560 }, { "epoch": 141.8546845124283, "grad_norm": 0.4641898274421692, "learning_rate": 1.713372977083273e-05, "loss": 0.0077, "step": 222570 }, { "epoch": 141.8610579987253, "grad_norm": 0.486402690410614, "learning_rate": 1.7129576416480175e-05, "loss": 0.0081, "step": 222580 }, { "epoch": 141.8674314850223, "grad_norm": 0.17339497804641724, "learning_rate": 1.712542346153429e-05, "loss": 0.0122, "step": 222590 }, { "epoch": 141.87380497131932, "grad_norm": 0.24792887270450592, "learning_rate": 1.7121270906045576e-05, "loss": 0.0057, "step": 222600 }, { "epoch": 141.88017845761632, "grad_norm": 0.10108092427253723, "learning_rate": 1.7117118750064438e-05, "loss": 0.0057, "step": 222610 }, { "epoch": 141.88655194391333, "grad_norm": 0.23539026081562042, "learning_rate": 1.7112966993641368e-05, "loss": 0.0061, "step": 222620 }, { "epoch": 141.8929254302103, "grad_norm": 0.2353391945362091, "learning_rate": 1.7108815636826775e-05, "loss": 0.0112, "step": 222630 }, { "epoch": 141.89929891650732, "grad_norm": 0.1363290548324585, "learning_rate": 1.7104664679671167e-05, "loss": 0.0103, "step": 222640 }, { "epoch": 141.90567240280433, "grad_norm": 0.09995895624160767, "learning_rate": 1.7100514122224908e-05, "loss": 0.006, "step": 222650 }, { "epoch": 141.91204588910134, "grad_norm": 0.18078768253326416, "learning_rate": 1.709636396453848e-05, "loss": 0.0101, "step": 222660 }, { "epoch": 141.91841937539834, "grad_norm": 0.11449802666902542, "learning_rate": 1.7092214206662294e-05, "loss": 0.0036, "step": 222670 }, { "epoch": 141.92479286169535, "grad_norm": 0.2188601940870285, "learning_rate": 1.708806484864678e-05, "loss": 0.0104, "step": 222680 }, { "epoch": 141.93116634799236, "grad_norm": 0.17458534240722656, "learning_rate": 1.7083915890542328e-05, "loss": 0.0055, "step": 222690 }, { "epoch": 141.93753983428937, "grad_norm": 0.11739835888147354, "learning_rate": 1.70797673323994e-05, "loss": 0.0039, "step": 222700 }, { "epoch": 141.94391332058635, "grad_norm": 0.12856076657772064, "learning_rate": 1.7075619174268377e-05, "loss": 0.0051, "step": 222710 }, { "epoch": 141.95028680688335, "grad_norm": 0.16349899768829346, "learning_rate": 1.707147141619967e-05, "loss": 0.003, "step": 222720 }, { "epoch": 141.95666029318036, "grad_norm": 0.11421487480401993, "learning_rate": 1.7067324058243657e-05, "loss": 0.006, "step": 222730 }, { "epoch": 141.96303377947737, "grad_norm": 0.07197015732526779, "learning_rate": 1.7063177100450765e-05, "loss": 0.0037, "step": 222740 }, { "epoch": 141.96940726577438, "grad_norm": 0.10414314270019531, "learning_rate": 1.7059030542871367e-05, "loss": 0.0036, "step": 222750 }, { "epoch": 141.9757807520714, "grad_norm": 0.22290970385074615, "learning_rate": 1.705488438555586e-05, "loss": 0.0114, "step": 222760 }, { "epoch": 141.9821542383684, "grad_norm": 0.10171223431825638, "learning_rate": 1.7050738628554585e-05, "loss": 0.0093, "step": 222770 }, { "epoch": 141.9885277246654, "grad_norm": 0.09284698218107224, "learning_rate": 1.7046593271917967e-05, "loss": 0.0076, "step": 222780 }, { "epoch": 141.99490121096238, "grad_norm": 0.14280208945274353, "learning_rate": 1.704244831569635e-05, "loss": 0.0065, "step": 222790 }, { "epoch": 142.0012746972594, "grad_norm": 0.3463272452354431, "learning_rate": 1.7038303759940106e-05, "loss": 0.0097, "step": 222800 }, { "epoch": 142.0076481835564, "grad_norm": 0.18160268664360046, "learning_rate": 1.7034159604699574e-05, "loss": 0.0064, "step": 222810 }, { "epoch": 142.0140216698534, "grad_norm": 0.17505916953086853, "learning_rate": 1.7030015850025144e-05, "loss": 0.007, "step": 222820 }, { "epoch": 142.0203951561504, "grad_norm": 0.2843824625015259, "learning_rate": 1.7025872495967155e-05, "loss": 0.0066, "step": 222830 }, { "epoch": 142.02676864244742, "grad_norm": 0.3279535472393036, "learning_rate": 1.7021729542575942e-05, "loss": 0.0059, "step": 222840 }, { "epoch": 142.03314212874443, "grad_norm": 0.0992545634508133, "learning_rate": 1.701758698990186e-05, "loss": 0.0044, "step": 222850 }, { "epoch": 142.03951561504144, "grad_norm": 0.34295231103897095, "learning_rate": 1.7013444837995206e-05, "loss": 0.0118, "step": 222860 }, { "epoch": 142.04588910133845, "grad_norm": 0.1480315774679184, "learning_rate": 1.700930308690637e-05, "loss": 0.0047, "step": 222870 }, { "epoch": 142.05226258763543, "grad_norm": 0.09314177930355072, "learning_rate": 1.700516173668565e-05, "loss": 0.0063, "step": 222880 }, { "epoch": 142.05863607393243, "grad_norm": 0.293195903301239, "learning_rate": 1.700102078738336e-05, "loss": 0.0042, "step": 222890 }, { "epoch": 142.06500956022944, "grad_norm": 0.3646708130836487, "learning_rate": 1.6996880239049813e-05, "loss": 0.0057, "step": 222900 }, { "epoch": 142.07138304652645, "grad_norm": 0.11736270040273666, "learning_rate": 1.6992740091735342e-05, "loss": 0.0107, "step": 222910 }, { "epoch": 142.07775653282346, "grad_norm": 0.09295795857906342, "learning_rate": 1.6988600345490247e-05, "loss": 0.006, "step": 222920 }, { "epoch": 142.08413001912047, "grad_norm": 0.37987443804740906, "learning_rate": 1.6984461000364827e-05, "loss": 0.005, "step": 222930 }, { "epoch": 142.09050350541747, "grad_norm": 0.20269158482551575, "learning_rate": 1.698032205640936e-05, "loss": 0.0061, "step": 222940 }, { "epoch": 142.09687699171448, "grad_norm": 0.2558862566947937, "learning_rate": 1.697618351367417e-05, "loss": 0.0105, "step": 222950 }, { "epoch": 142.10325047801146, "grad_norm": 0.15254458785057068, "learning_rate": 1.6972045372209532e-05, "loss": 0.0079, "step": 222960 }, { "epoch": 142.10962396430847, "grad_norm": 0.13688161969184875, "learning_rate": 1.6967907632065722e-05, "loss": 0.01, "step": 222970 }, { "epoch": 142.11599745060548, "grad_norm": 0.10799124836921692, "learning_rate": 1.6963770293293015e-05, "loss": 0.0046, "step": 222980 }, { "epoch": 142.12237093690248, "grad_norm": 0.28652235865592957, "learning_rate": 1.6959633355941713e-05, "loss": 0.01, "step": 222990 }, { "epoch": 142.1287444231995, "grad_norm": 0.20341689884662628, "learning_rate": 1.6955496820062032e-05, "loss": 0.0056, "step": 223000 }, { "epoch": 142.1351179094965, "grad_norm": 0.24782918393611908, "learning_rate": 1.6951360685704283e-05, "loss": 0.0079, "step": 223010 }, { "epoch": 142.1414913957935, "grad_norm": 0.2980453372001648, "learning_rate": 1.6947224952918684e-05, "loss": 0.0101, "step": 223020 }, { "epoch": 142.14786488209052, "grad_norm": 0.35206612944602966, "learning_rate": 1.6943089621755543e-05, "loss": 0.0132, "step": 223030 }, { "epoch": 142.1542383683875, "grad_norm": 0.41653314232826233, "learning_rate": 1.693895469226504e-05, "loss": 0.0092, "step": 223040 }, { "epoch": 142.1606118546845, "grad_norm": 0.07428012043237686, "learning_rate": 1.693482016449747e-05, "loss": 0.0047, "step": 223050 }, { "epoch": 142.1669853409815, "grad_norm": 0.11640966683626175, "learning_rate": 1.693068603850304e-05, "loss": 0.0046, "step": 223060 }, { "epoch": 142.17335882727852, "grad_norm": 0.1249624714255333, "learning_rate": 1.692655231433203e-05, "loss": 0.0098, "step": 223070 }, { "epoch": 142.17973231357553, "grad_norm": 0.6230377554893494, "learning_rate": 1.69224189920346e-05, "loss": 0.0156, "step": 223080 }, { "epoch": 142.18610579987254, "grad_norm": 0.23987340927124023, "learning_rate": 1.6918286071661026e-05, "loss": 0.0069, "step": 223090 }, { "epoch": 142.19247928616954, "grad_norm": 0.19283346831798553, "learning_rate": 1.6914153553261514e-05, "loss": 0.0069, "step": 223100 }, { "epoch": 142.19885277246655, "grad_norm": 0.13493730127811432, "learning_rate": 1.691002143688627e-05, "loss": 0.003, "step": 223110 }, { "epoch": 142.20522625876353, "grad_norm": 0.23026952147483826, "learning_rate": 1.6905889722585495e-05, "loss": 0.0073, "step": 223120 }, { "epoch": 142.21159974506054, "grad_norm": 0.09305721521377563, "learning_rate": 1.6901758410409418e-05, "loss": 0.007, "step": 223130 }, { "epoch": 142.21797323135755, "grad_norm": 0.27223172783851624, "learning_rate": 1.689762750040823e-05, "loss": 0.0188, "step": 223140 }, { "epoch": 142.22434671765456, "grad_norm": 0.17869116365909576, "learning_rate": 1.689349699263212e-05, "loss": 0.0089, "step": 223150 }, { "epoch": 142.23072020395156, "grad_norm": 0.09456615895032883, "learning_rate": 1.6889366887131265e-05, "loss": 0.0062, "step": 223160 }, { "epoch": 142.23709369024857, "grad_norm": 0.6450970768928528, "learning_rate": 1.688523718395589e-05, "loss": 0.0093, "step": 223170 }, { "epoch": 142.24346717654558, "grad_norm": 0.13820849359035492, "learning_rate": 1.6881107883156123e-05, "loss": 0.0105, "step": 223180 }, { "epoch": 142.2498406628426, "grad_norm": 0.25581905245780945, "learning_rate": 1.6876978984782183e-05, "loss": 0.0106, "step": 223190 }, { "epoch": 142.25621414913957, "grad_norm": 0.5324016213417053, "learning_rate": 1.6872850488884194e-05, "loss": 0.0105, "step": 223200 }, { "epoch": 142.26258763543657, "grad_norm": 0.2924562990665436, "learning_rate": 1.6868722395512392e-05, "loss": 0.0064, "step": 223210 }, { "epoch": 142.26896112173358, "grad_norm": 0.1492711752653122, "learning_rate": 1.6864594704716858e-05, "loss": 0.0064, "step": 223220 }, { "epoch": 142.2753346080306, "grad_norm": 0.3390948176383972, "learning_rate": 1.6860467416547797e-05, "loss": 0.0062, "step": 223230 }, { "epoch": 142.2817080943276, "grad_norm": 0.24670979380607605, "learning_rate": 1.6856340531055352e-05, "loss": 0.0141, "step": 223240 }, { "epoch": 142.2880815806246, "grad_norm": 0.09969007223844528, "learning_rate": 1.685221404828966e-05, "loss": 0.0151, "step": 223250 }, { "epoch": 142.2944550669216, "grad_norm": 0.502535879611969, "learning_rate": 1.6848087968300845e-05, "loss": 0.0114, "step": 223260 }, { "epoch": 142.30082855321862, "grad_norm": 0.17298352718353271, "learning_rate": 1.684396229113908e-05, "loss": 0.0096, "step": 223270 }, { "epoch": 142.3072020395156, "grad_norm": 0.27157846093177795, "learning_rate": 1.6839837016854477e-05, "loss": 0.0097, "step": 223280 }, { "epoch": 142.3135755258126, "grad_norm": 0.2435157597064972, "learning_rate": 1.683571214549716e-05, "loss": 0.0069, "step": 223290 }, { "epoch": 142.31994901210962, "grad_norm": 0.20911990106105804, "learning_rate": 1.6831587677117245e-05, "loss": 0.0088, "step": 223300 }, { "epoch": 142.32632249840663, "grad_norm": 0.43472909927368164, "learning_rate": 1.682746361176488e-05, "loss": 0.0255, "step": 223310 }, { "epoch": 142.33269598470363, "grad_norm": 0.28610959649086, "learning_rate": 1.6823339949490126e-05, "loss": 0.006, "step": 223320 }, { "epoch": 142.33906947100064, "grad_norm": 0.13839855790138245, "learning_rate": 1.6819216690343125e-05, "loss": 0.008, "step": 223330 }, { "epoch": 142.34544295729765, "grad_norm": 0.2291944921016693, "learning_rate": 1.6815093834373957e-05, "loss": 0.0073, "step": 223340 }, { "epoch": 142.35181644359466, "grad_norm": 0.2563689053058624, "learning_rate": 1.681097138163276e-05, "loss": 0.0098, "step": 223350 }, { "epoch": 142.35818992989164, "grad_norm": 0.10529392212629318, "learning_rate": 1.6806849332169568e-05, "loss": 0.0116, "step": 223360 }, { "epoch": 142.36456341618864, "grad_norm": 0.15075311064720154, "learning_rate": 1.6802727686034513e-05, "loss": 0.0064, "step": 223370 }, { "epoch": 142.37093690248565, "grad_norm": 0.13008180260658264, "learning_rate": 1.6798606443277636e-05, "loss": 0.0046, "step": 223380 }, { "epoch": 142.37731038878266, "grad_norm": 0.48646968603134155, "learning_rate": 1.679448560394908e-05, "loss": 0.0085, "step": 223390 }, { "epoch": 142.38368387507967, "grad_norm": 0.2572052478790283, "learning_rate": 1.6790365168098837e-05, "loss": 0.0189, "step": 223400 }, { "epoch": 142.39005736137668, "grad_norm": 0.5793009996414185, "learning_rate": 1.678624513577703e-05, "loss": 0.0122, "step": 223410 }, { "epoch": 142.39643084767368, "grad_norm": 0.4975297152996063, "learning_rate": 1.6782125507033704e-05, "loss": 0.0083, "step": 223420 }, { "epoch": 142.4028043339707, "grad_norm": 0.15746146440505981, "learning_rate": 1.677800628191892e-05, "loss": 0.0032, "step": 223430 }, { "epoch": 142.4091778202677, "grad_norm": 0.1744355410337448, "learning_rate": 1.6773887460482702e-05, "loss": 0.003, "step": 223440 }, { "epoch": 142.41555130656468, "grad_norm": 0.2221236675977707, "learning_rate": 1.6769769042775142e-05, "loss": 0.0073, "step": 223450 }, { "epoch": 142.4219247928617, "grad_norm": 0.3059946298599243, "learning_rate": 1.6765651028846263e-05, "loss": 0.0102, "step": 223460 }, { "epoch": 142.4282982791587, "grad_norm": 0.11842944473028183, "learning_rate": 1.67615334187461e-05, "loss": 0.0035, "step": 223470 }, { "epoch": 142.4346717654557, "grad_norm": 0.16297969222068787, "learning_rate": 1.6757416212524673e-05, "loss": 0.0058, "step": 223480 }, { "epoch": 142.4410452517527, "grad_norm": 0.5448606610298157, "learning_rate": 1.6753299410232044e-05, "loss": 0.0121, "step": 223490 }, { "epoch": 142.44741873804972, "grad_norm": 0.2545856237411499, "learning_rate": 1.6749183011918217e-05, "loss": 0.0063, "step": 223500 }, { "epoch": 142.45379222434673, "grad_norm": 0.29440224170684814, "learning_rate": 1.674506701763321e-05, "loss": 0.0063, "step": 223510 }, { "epoch": 142.46016571064374, "grad_norm": 0.07561805844306946, "learning_rate": 1.6740951427427022e-05, "loss": 0.0042, "step": 223520 }, { "epoch": 142.46653919694072, "grad_norm": 0.09724974632263184, "learning_rate": 1.6736836241349686e-05, "loss": 0.0063, "step": 223530 }, { "epoch": 142.47291268323772, "grad_norm": 0.20771656930446625, "learning_rate": 1.67327214594512e-05, "loss": 0.0154, "step": 223540 }, { "epoch": 142.47928616953473, "grad_norm": 0.2782699167728424, "learning_rate": 1.6728607081781556e-05, "loss": 0.0065, "step": 223550 }, { "epoch": 142.48565965583174, "grad_norm": 0.161574125289917, "learning_rate": 1.6724493108390747e-05, "loss": 0.0041, "step": 223560 }, { "epoch": 142.49203314212875, "grad_norm": 0.7838782668113708, "learning_rate": 1.6720379539328757e-05, "loss": 0.0091, "step": 223570 }, { "epoch": 142.49840662842576, "grad_norm": 0.2135651707649231, "learning_rate": 1.6716266374645585e-05, "loss": 0.0076, "step": 223580 }, { "epoch": 142.50478011472276, "grad_norm": 0.09519843757152557, "learning_rate": 1.67121536143912e-05, "loss": 0.0066, "step": 223590 }, { "epoch": 142.51115360101977, "grad_norm": 0.12873157858848572, "learning_rate": 1.6708041258615587e-05, "loss": 0.0129, "step": 223600 }, { "epoch": 142.51752708731675, "grad_norm": 0.1751462072134018, "learning_rate": 1.670392930736868e-05, "loss": 0.005, "step": 223610 }, { "epoch": 142.52390057361376, "grad_norm": 0.16708141565322876, "learning_rate": 1.669981776070049e-05, "loss": 0.0072, "step": 223620 }, { "epoch": 142.53027405991077, "grad_norm": 0.13569967448711395, "learning_rate": 1.669570661866095e-05, "loss": 0.0045, "step": 223630 }, { "epoch": 142.53664754620777, "grad_norm": 0.10741084814071655, "learning_rate": 1.6691595881300025e-05, "loss": 0.007, "step": 223640 }, { "epoch": 142.54302103250478, "grad_norm": 0.23126551508903503, "learning_rate": 1.6687485548667636e-05, "loss": 0.0061, "step": 223650 }, { "epoch": 142.5493945188018, "grad_norm": 0.13018648326396942, "learning_rate": 1.668337562081377e-05, "loss": 0.0061, "step": 223660 }, { "epoch": 142.5557680050988, "grad_norm": 0.49383875727653503, "learning_rate": 1.667926609778835e-05, "loss": 0.0106, "step": 223670 }, { "epoch": 142.5621414913958, "grad_norm": 0.21080762147903442, "learning_rate": 1.66751569796413e-05, "loss": 0.0132, "step": 223680 }, { "epoch": 142.5685149776928, "grad_norm": 0.3486484885215759, "learning_rate": 1.667104826642254e-05, "loss": 0.0077, "step": 223690 }, { "epoch": 142.5748884639898, "grad_norm": 0.1518961787223816, "learning_rate": 1.666693995818203e-05, "loss": 0.0073, "step": 223700 }, { "epoch": 142.5812619502868, "grad_norm": 0.47461792826652527, "learning_rate": 1.6662832054969674e-05, "loss": 0.0126, "step": 223710 }, { "epoch": 142.5876354365838, "grad_norm": 0.4194446802139282, "learning_rate": 1.665872455683538e-05, "loss": 0.0095, "step": 223720 }, { "epoch": 142.59400892288082, "grad_norm": 0.1736660748720169, "learning_rate": 1.6654617463829046e-05, "loss": 0.011, "step": 223730 }, { "epoch": 142.60038240917783, "grad_norm": 0.32411569356918335, "learning_rate": 1.665051077600063e-05, "loss": 0.009, "step": 223740 }, { "epoch": 142.60675589547483, "grad_norm": 0.19630643725395203, "learning_rate": 1.664640449339996e-05, "loss": 0.0071, "step": 223750 }, { "epoch": 142.61312938177184, "grad_norm": 0.2136065512895584, "learning_rate": 1.664229861607698e-05, "loss": 0.0059, "step": 223760 }, { "epoch": 142.61950286806882, "grad_norm": 0.2526579201221466, "learning_rate": 1.6638193144081555e-05, "loss": 0.0098, "step": 223770 }, { "epoch": 142.62587635436583, "grad_norm": 0.17659883201122284, "learning_rate": 1.663408807746361e-05, "loss": 0.0047, "step": 223780 }, { "epoch": 142.63224984066284, "grad_norm": 0.4329105615615845, "learning_rate": 1.662998341627297e-05, "loss": 0.0059, "step": 223790 }, { "epoch": 142.63862332695984, "grad_norm": 0.13348473608493805, "learning_rate": 1.662587916055955e-05, "loss": 0.0054, "step": 223800 }, { "epoch": 142.64499681325685, "grad_norm": 0.5647808909416199, "learning_rate": 1.662177531037321e-05, "loss": 0.0098, "step": 223810 }, { "epoch": 142.65137029955386, "grad_norm": 0.19389933347702026, "learning_rate": 1.6617671865763813e-05, "loss": 0.0065, "step": 223820 }, { "epoch": 142.65774378585087, "grad_norm": 0.26774150133132935, "learning_rate": 1.6613568826781205e-05, "loss": 0.0057, "step": 223830 }, { "epoch": 142.66411727214788, "grad_norm": 0.3297603726387024, "learning_rate": 1.6609466193475276e-05, "loss": 0.0148, "step": 223840 }, { "epoch": 142.67049075844486, "grad_norm": 0.14103105664253235, "learning_rate": 1.660536396589586e-05, "loss": 0.0084, "step": 223850 }, { "epoch": 142.67686424474186, "grad_norm": 0.20873518288135529, "learning_rate": 1.66012621440928e-05, "loss": 0.0088, "step": 223860 }, { "epoch": 142.68323773103887, "grad_norm": 0.27515900135040283, "learning_rate": 1.6597160728115916e-05, "loss": 0.0076, "step": 223870 }, { "epoch": 142.68961121733588, "grad_norm": 0.15990257263183594, "learning_rate": 1.6593059718015108e-05, "loss": 0.0036, "step": 223880 }, { "epoch": 142.6959847036329, "grad_norm": 0.15507255494594574, "learning_rate": 1.658895911384013e-05, "loss": 0.0061, "step": 223890 }, { "epoch": 142.7023581899299, "grad_norm": 0.15675707161426544, "learning_rate": 1.658485891564086e-05, "loss": 0.0035, "step": 223900 }, { "epoch": 142.7087316762269, "grad_norm": 0.07168799638748169, "learning_rate": 1.6580759123467087e-05, "loss": 0.009, "step": 223910 }, { "epoch": 142.7151051625239, "grad_norm": 0.15494346618652344, "learning_rate": 1.6576659737368677e-05, "loss": 0.0111, "step": 223920 }, { "epoch": 142.7214786488209, "grad_norm": 0.1237083151936531, "learning_rate": 1.657256075739538e-05, "loss": 0.0094, "step": 223930 }, { "epoch": 142.7278521351179, "grad_norm": 0.18058578670024872, "learning_rate": 1.656846218359704e-05, "loss": 0.004, "step": 223940 }, { "epoch": 142.7342256214149, "grad_norm": 0.31560254096984863, "learning_rate": 1.656436401602345e-05, "loss": 0.0096, "step": 223950 }, { "epoch": 142.74059910771192, "grad_norm": 0.12169413268566132, "learning_rate": 1.6560266254724405e-05, "loss": 0.0202, "step": 223960 }, { "epoch": 142.74697259400892, "grad_norm": 0.1686968207359314, "learning_rate": 1.6556168899749684e-05, "loss": 0.0075, "step": 223970 }, { "epoch": 142.75334608030593, "grad_norm": 0.4190077781677246, "learning_rate": 1.65520719511491e-05, "loss": 0.0124, "step": 223980 }, { "epoch": 142.75971956660294, "grad_norm": 0.1603570431470871, "learning_rate": 1.6547975408972426e-05, "loss": 0.0033, "step": 223990 }, { "epoch": 142.76609305289995, "grad_norm": 0.3188340961933136, "learning_rate": 1.654387927326943e-05, "loss": 0.0111, "step": 224000 }, { "epoch": 142.77246653919693, "grad_norm": 0.12722252309322357, "learning_rate": 1.6539783544089876e-05, "loss": 0.0085, "step": 224010 }, { "epoch": 142.77884002549393, "grad_norm": 0.5153641700744629, "learning_rate": 1.653568822148356e-05, "loss": 0.006, "step": 224020 }, { "epoch": 142.78521351179094, "grad_norm": 0.33786460757255554, "learning_rate": 1.6531593305500227e-05, "loss": 0.0149, "step": 224030 }, { "epoch": 142.79158699808795, "grad_norm": 0.13839225471019745, "learning_rate": 1.6527498796189634e-05, "loss": 0.006, "step": 224040 }, { "epoch": 142.79796048438496, "grad_norm": 0.6340293288230896, "learning_rate": 1.6523404693601517e-05, "loss": 0.0063, "step": 224050 }, { "epoch": 142.80433397068197, "grad_norm": 0.5718280673027039, "learning_rate": 1.6519310997785676e-05, "loss": 0.016, "step": 224060 }, { "epoch": 142.81070745697897, "grad_norm": 0.39942309260368347, "learning_rate": 1.651521770879178e-05, "loss": 0.0079, "step": 224070 }, { "epoch": 142.81708094327598, "grad_norm": 0.163779154419899, "learning_rate": 1.6511124826669622e-05, "loss": 0.0055, "step": 224080 }, { "epoch": 142.823454429573, "grad_norm": 0.12673279643058777, "learning_rate": 1.65070323514689e-05, "loss": 0.0119, "step": 224090 }, { "epoch": 142.82982791586997, "grad_norm": 0.17736245691776276, "learning_rate": 1.650294028323939e-05, "loss": 0.0059, "step": 224100 }, { "epoch": 142.83620140216698, "grad_norm": 0.4626528322696686, "learning_rate": 1.649884862203075e-05, "loss": 0.0096, "step": 224110 }, { "epoch": 142.842574888464, "grad_norm": 0.2328469455242157, "learning_rate": 1.649475736789275e-05, "loss": 0.0067, "step": 224120 }, { "epoch": 142.848948374761, "grad_norm": 0.19295397400856018, "learning_rate": 1.6490666520875076e-05, "loss": 0.0065, "step": 224130 }, { "epoch": 142.855321861058, "grad_norm": 0.13441258668899536, "learning_rate": 1.6486576081027443e-05, "loss": 0.0051, "step": 224140 }, { "epoch": 142.861695347355, "grad_norm": 0.08184042572975159, "learning_rate": 1.6482486048399536e-05, "loss": 0.0095, "step": 224150 }, { "epoch": 142.86806883365202, "grad_norm": 0.43711480498313904, "learning_rate": 1.647839642304109e-05, "loss": 0.006, "step": 224160 }, { "epoch": 142.87444231994903, "grad_norm": 0.32990381121635437, "learning_rate": 1.6474307205001772e-05, "loss": 0.0051, "step": 224170 }, { "epoch": 142.880815806246, "grad_norm": 0.2257634848356247, "learning_rate": 1.647021839433128e-05, "loss": 0.0051, "step": 224180 }, { "epoch": 142.887189292543, "grad_norm": 0.3547494113445282, "learning_rate": 1.6466129991079266e-05, "loss": 0.0113, "step": 224190 }, { "epoch": 142.89356277884002, "grad_norm": 0.2593889832496643, "learning_rate": 1.6462041995295458e-05, "loss": 0.0073, "step": 224200 }, { "epoch": 142.89993626513703, "grad_norm": 0.3130502998828888, "learning_rate": 1.64579544070295e-05, "loss": 0.0156, "step": 224210 }, { "epoch": 142.90630975143404, "grad_norm": 0.1463400423526764, "learning_rate": 1.645386722633107e-05, "loss": 0.0111, "step": 224220 }, { "epoch": 142.91268323773104, "grad_norm": 0.13244861364364624, "learning_rate": 1.644978045324981e-05, "loss": 0.0057, "step": 224230 }, { "epoch": 142.91905672402805, "grad_norm": 0.32308921217918396, "learning_rate": 1.6445694087835407e-05, "loss": 0.0101, "step": 224240 }, { "epoch": 142.92543021032506, "grad_norm": 0.21547842025756836, "learning_rate": 1.6441608130137502e-05, "loss": 0.0036, "step": 224250 }, { "epoch": 142.93180369662204, "grad_norm": 0.16276560723781586, "learning_rate": 1.6437522580205744e-05, "loss": 0.0087, "step": 224260 }, { "epoch": 142.93817718291905, "grad_norm": 0.409012109041214, "learning_rate": 1.6433437438089756e-05, "loss": 0.0041, "step": 224270 }, { "epoch": 142.94455066921606, "grad_norm": 0.16498953104019165, "learning_rate": 1.6429352703839213e-05, "loss": 0.0047, "step": 224280 }, { "epoch": 142.95092415551306, "grad_norm": 0.3706355392932892, "learning_rate": 1.6425268377503732e-05, "loss": 0.0081, "step": 224290 }, { "epoch": 142.95729764181007, "grad_norm": 0.3275659680366516, "learning_rate": 1.642118445913294e-05, "loss": 0.007, "step": 224300 }, { "epoch": 142.96367112810708, "grad_norm": 0.09132810682058334, "learning_rate": 1.641710094877646e-05, "loss": 0.0065, "step": 224310 }, { "epoch": 142.9700446144041, "grad_norm": 0.2875877618789673, "learning_rate": 1.641301784648389e-05, "loss": 0.0046, "step": 224320 }, { "epoch": 142.9764181007011, "grad_norm": 0.16997985541820526, "learning_rate": 1.640893515230489e-05, "loss": 0.0066, "step": 224330 }, { "epoch": 142.98279158699808, "grad_norm": 0.17990680038928986, "learning_rate": 1.6404852866289033e-05, "loss": 0.0134, "step": 224340 }, { "epoch": 142.98916507329508, "grad_norm": 0.19751402735710144, "learning_rate": 1.6400770988485937e-05, "loss": 0.0114, "step": 224350 }, { "epoch": 142.9955385595921, "grad_norm": 0.24292722344398499, "learning_rate": 1.639668951894518e-05, "loss": 0.0075, "step": 224360 }, { "epoch": 143.0019120458891, "grad_norm": 0.20462679862976074, "learning_rate": 1.639260845771639e-05, "loss": 0.0121, "step": 224370 }, { "epoch": 143.0082855321861, "grad_norm": 0.28396254777908325, "learning_rate": 1.6388527804849142e-05, "loss": 0.0034, "step": 224380 }, { "epoch": 143.01465901848312, "grad_norm": 0.15204665064811707, "learning_rate": 1.6384447560393007e-05, "loss": 0.0077, "step": 224390 }, { "epoch": 143.02103250478012, "grad_norm": 0.33738410472869873, "learning_rate": 1.6380367724397562e-05, "loss": 0.0048, "step": 224400 }, { "epoch": 143.02740599107713, "grad_norm": 0.15385234355926514, "learning_rate": 1.6376288296912402e-05, "loss": 0.0035, "step": 224410 }, { "epoch": 143.0337794773741, "grad_norm": 0.29666274785995483, "learning_rate": 1.637220927798709e-05, "loss": 0.0108, "step": 224420 }, { "epoch": 143.04015296367112, "grad_norm": 0.2879916727542877, "learning_rate": 1.6368130667671182e-05, "loss": 0.0104, "step": 224430 }, { "epoch": 143.04652644996813, "grad_norm": 0.3972392678260803, "learning_rate": 1.6364052466014228e-05, "loss": 0.0094, "step": 224440 }, { "epoch": 143.05289993626513, "grad_norm": 0.14366571605205536, "learning_rate": 1.6359974673065825e-05, "loss": 0.0141, "step": 224450 }, { "epoch": 143.05927342256214, "grad_norm": 0.5138731598854065, "learning_rate": 1.6355897288875456e-05, "loss": 0.0141, "step": 224460 }, { "epoch": 143.06564690885915, "grad_norm": 0.19632117450237274, "learning_rate": 1.635182031349272e-05, "loss": 0.0033, "step": 224470 }, { "epoch": 143.07202039515616, "grad_norm": 0.4396101236343384, "learning_rate": 1.634774374696712e-05, "loss": 0.0085, "step": 224480 }, { "epoch": 143.07839388145317, "grad_norm": 0.14552459120750427, "learning_rate": 1.634366758934824e-05, "loss": 0.0171, "step": 224490 }, { "epoch": 143.08476736775015, "grad_norm": 0.23912838101387024, "learning_rate": 1.6339591840685537e-05, "loss": 0.0103, "step": 224500 }, { "epoch": 143.09114085404715, "grad_norm": 0.17731693387031555, "learning_rate": 1.633551650102859e-05, "loss": 0.0079, "step": 224510 }, { "epoch": 143.09751434034416, "grad_norm": 0.21906428039073944, "learning_rate": 1.6331441570426908e-05, "loss": 0.0071, "step": 224520 }, { "epoch": 143.10388782664117, "grad_norm": 0.13123756647109985, "learning_rate": 1.632736704893e-05, "loss": 0.0053, "step": 224530 }, { "epoch": 143.11026131293818, "grad_norm": 0.16497020423412323, "learning_rate": 1.6323292936587354e-05, "loss": 0.004, "step": 224540 }, { "epoch": 143.1166347992352, "grad_norm": 0.26566457748413086, "learning_rate": 1.6319219233448514e-05, "loss": 0.0045, "step": 224550 }, { "epoch": 143.1230082855322, "grad_norm": 0.329656183719635, "learning_rate": 1.631514593956296e-05, "loss": 0.0072, "step": 224560 }, { "epoch": 143.1293817718292, "grad_norm": 0.0998678058385849, "learning_rate": 1.6311073054980187e-05, "loss": 0.0155, "step": 224570 }, { "epoch": 143.13575525812618, "grad_norm": 0.17964652180671692, "learning_rate": 1.6307000579749666e-05, "loss": 0.0057, "step": 224580 }, { "epoch": 143.1421287444232, "grad_norm": 0.14459532499313354, "learning_rate": 1.6302928513920913e-05, "loss": 0.0102, "step": 224590 }, { "epoch": 143.1485022307202, "grad_norm": 0.5662151575088501, "learning_rate": 1.62988568575434e-05, "loss": 0.0091, "step": 224600 }, { "epoch": 143.1548757170172, "grad_norm": 0.19715242087841034, "learning_rate": 1.629478561066659e-05, "loss": 0.007, "step": 224610 }, { "epoch": 143.1612492033142, "grad_norm": 0.3340986967086792, "learning_rate": 1.6290714773339944e-05, "loss": 0.0044, "step": 224620 }, { "epoch": 143.16762268961122, "grad_norm": 0.16687719523906708, "learning_rate": 1.628664434561297e-05, "loss": 0.0059, "step": 224630 }, { "epoch": 143.17399617590823, "grad_norm": 0.38039878010749817, "learning_rate": 1.6282574327535072e-05, "loss": 0.0193, "step": 224640 }, { "epoch": 143.18036966220524, "grad_norm": 0.15167196094989777, "learning_rate": 1.627850471915574e-05, "loss": 0.0146, "step": 224650 }, { "epoch": 143.18674314850222, "grad_norm": 0.36467984318733215, "learning_rate": 1.62744355205244e-05, "loss": 0.0073, "step": 224660 }, { "epoch": 143.19311663479922, "grad_norm": 0.2822469472885132, "learning_rate": 1.6270366731690546e-05, "loss": 0.0055, "step": 224670 }, { "epoch": 143.19949012109623, "grad_norm": 0.40562117099761963, "learning_rate": 1.6266298352703545e-05, "loss": 0.0119, "step": 224680 }, { "epoch": 143.20586360739324, "grad_norm": 0.21180079877376556, "learning_rate": 1.6262230383612887e-05, "loss": 0.0041, "step": 224690 }, { "epoch": 143.21223709369025, "grad_norm": 0.15101908147335052, "learning_rate": 1.625816282446799e-05, "loss": 0.0078, "step": 224700 }, { "epoch": 143.21861057998726, "grad_norm": 0.13209834694862366, "learning_rate": 1.6254095675318258e-05, "loss": 0.0058, "step": 224710 }, { "epoch": 143.22498406628426, "grad_norm": 0.1367049515247345, "learning_rate": 1.6250028936213118e-05, "loss": 0.0086, "step": 224720 }, { "epoch": 143.23135755258127, "grad_norm": 0.14438021183013916, "learning_rate": 1.624596260720201e-05, "loss": 0.0085, "step": 224730 }, { "epoch": 143.23773103887828, "grad_norm": 0.15149220824241638, "learning_rate": 1.6241896688334325e-05, "loss": 0.004, "step": 224740 }, { "epoch": 143.24410452517526, "grad_norm": 0.20297391712665558, "learning_rate": 1.6237831179659463e-05, "loss": 0.0117, "step": 224750 }, { "epoch": 143.25047801147227, "grad_norm": 0.337577223777771, "learning_rate": 1.6233766081226815e-05, "loss": 0.0085, "step": 224760 }, { "epoch": 143.25685149776928, "grad_norm": 0.8153294920921326, "learning_rate": 1.6229701393085823e-05, "loss": 0.013, "step": 224770 }, { "epoch": 143.26322498406628, "grad_norm": 0.1275549829006195, "learning_rate": 1.622563711528581e-05, "loss": 0.0168, "step": 224780 }, { "epoch": 143.2695984703633, "grad_norm": 0.13954618573188782, "learning_rate": 1.622157324787621e-05, "loss": 0.014, "step": 224790 }, { "epoch": 143.2759719566603, "grad_norm": 0.32289811968803406, "learning_rate": 1.6217509790906372e-05, "loss": 0.008, "step": 224800 }, { "epoch": 143.2823454429573, "grad_norm": 0.23852790892124176, "learning_rate": 1.621344674442572e-05, "loss": 0.0052, "step": 224810 }, { "epoch": 143.28871892925432, "grad_norm": 0.22710396349430084, "learning_rate": 1.6209384108483556e-05, "loss": 0.0052, "step": 224820 }, { "epoch": 143.2950924155513, "grad_norm": 0.24671529233455658, "learning_rate": 1.6205321883129293e-05, "loss": 0.0072, "step": 224830 }, { "epoch": 143.3014659018483, "grad_norm": 0.15161040425300598, "learning_rate": 1.6201260068412277e-05, "loss": 0.0056, "step": 224840 }, { "epoch": 143.3078393881453, "grad_norm": 0.2811957597732544, "learning_rate": 1.619719866438187e-05, "loss": 0.004, "step": 224850 }, { "epoch": 143.31421287444232, "grad_norm": 0.3880654275417328, "learning_rate": 1.619313767108739e-05, "loss": 0.0134, "step": 224860 }, { "epoch": 143.32058636073933, "grad_norm": 0.12099362909793854, "learning_rate": 1.6189077088578224e-05, "loss": 0.0115, "step": 224870 }, { "epoch": 143.32695984703633, "grad_norm": 0.1484343707561493, "learning_rate": 1.6185016916903694e-05, "loss": 0.0076, "step": 224880 }, { "epoch": 143.33333333333334, "grad_norm": 0.2620804011821747, "learning_rate": 1.6180957156113135e-05, "loss": 0.0053, "step": 224890 }, { "epoch": 143.33970681963035, "grad_norm": 0.20115764439105988, "learning_rate": 1.6176897806255866e-05, "loss": 0.01, "step": 224900 }, { "epoch": 143.34608030592733, "grad_norm": 0.19441337883472443, "learning_rate": 1.617283886738123e-05, "loss": 0.0058, "step": 224910 }, { "epoch": 143.35245379222434, "grad_norm": 0.5138246417045593, "learning_rate": 1.6168780339538546e-05, "loss": 0.0096, "step": 224920 }, { "epoch": 143.35882727852135, "grad_norm": 0.2560170888900757, "learning_rate": 1.6164722222777124e-05, "loss": 0.0053, "step": 224930 }, { "epoch": 143.36520076481835, "grad_norm": 0.17801013588905334, "learning_rate": 1.616066451714625e-05, "loss": 0.0111, "step": 224940 }, { "epoch": 143.37157425111536, "grad_norm": 0.2126264125108719, "learning_rate": 1.6156607222695275e-05, "loss": 0.0115, "step": 224950 }, { "epoch": 143.37794773741237, "grad_norm": 0.12381138652563095, "learning_rate": 1.615255033947347e-05, "loss": 0.0041, "step": 224960 }, { "epoch": 143.38432122370938, "grad_norm": 0.17490647733211517, "learning_rate": 1.614849386753014e-05, "loss": 0.0065, "step": 224970 }, { "epoch": 143.3906947100064, "grad_norm": 0.29355451464653015, "learning_rate": 1.6144437806914552e-05, "loss": 0.006, "step": 224980 }, { "epoch": 143.39706819630337, "grad_norm": 0.4571954607963562, "learning_rate": 1.6140382157676026e-05, "loss": 0.0176, "step": 224990 }, { "epoch": 143.40344168260037, "grad_norm": 0.13726352155208588, "learning_rate": 1.6136326919863826e-05, "loss": 0.006, "step": 225000 }, { "epoch": 143.40981516889738, "grad_norm": 0.17659443616867065, "learning_rate": 1.6132272093527228e-05, "loss": 0.0062, "step": 225010 }, { "epoch": 143.4161886551944, "grad_norm": 0.09846697747707367, "learning_rate": 1.6128217678715495e-05, "loss": 0.0039, "step": 225020 }, { "epoch": 143.4225621414914, "grad_norm": 0.11142260581254959, "learning_rate": 1.612416367547789e-05, "loss": 0.0056, "step": 225030 }, { "epoch": 143.4289356277884, "grad_norm": 0.12304805964231491, "learning_rate": 1.6120110083863694e-05, "loss": 0.0104, "step": 225040 }, { "epoch": 143.4353091140854, "grad_norm": 0.17697736620903015, "learning_rate": 1.6116056903922146e-05, "loss": 0.0207, "step": 225050 }, { "epoch": 143.44168260038242, "grad_norm": 0.1730627417564392, "learning_rate": 1.6112004135702496e-05, "loss": 0.0122, "step": 225060 }, { "epoch": 143.4480560866794, "grad_norm": 0.16922175884246826, "learning_rate": 1.6107951779253975e-05, "loss": 0.0056, "step": 225070 }, { "epoch": 143.4544295729764, "grad_norm": 0.1437993198633194, "learning_rate": 1.6103899834625863e-05, "loss": 0.0057, "step": 225080 }, { "epoch": 143.46080305927342, "grad_norm": 0.1714208722114563, "learning_rate": 1.6099848301867366e-05, "loss": 0.0066, "step": 225090 }, { "epoch": 143.46717654557042, "grad_norm": 0.3449191749095917, "learning_rate": 1.6095797181027717e-05, "loss": 0.0055, "step": 225100 }, { "epoch": 143.47355003186743, "grad_norm": 0.12146854400634766, "learning_rate": 1.6091746472156134e-05, "loss": 0.0061, "step": 225110 }, { "epoch": 143.47992351816444, "grad_norm": 0.4273850917816162, "learning_rate": 1.608769617530185e-05, "loss": 0.0154, "step": 225120 }, { "epoch": 143.48629700446145, "grad_norm": 0.24756619334220886, "learning_rate": 1.608364629051409e-05, "loss": 0.0076, "step": 225130 }, { "epoch": 143.49267049075846, "grad_norm": 0.2196638137102127, "learning_rate": 1.607959681784204e-05, "loss": 0.0119, "step": 225140 }, { "epoch": 143.49904397705544, "grad_norm": 0.2634517252445221, "learning_rate": 1.6075547757334907e-05, "loss": 0.0097, "step": 225150 }, { "epoch": 143.50541746335244, "grad_norm": 0.13953880965709686, "learning_rate": 1.607149910904191e-05, "loss": 0.0059, "step": 225160 }, { "epoch": 143.51179094964945, "grad_norm": 0.5551191568374634, "learning_rate": 1.6067450873012235e-05, "loss": 0.0078, "step": 225170 }, { "epoch": 143.51816443594646, "grad_norm": 0.1649959832429886, "learning_rate": 1.606340304929507e-05, "loss": 0.0046, "step": 225180 }, { "epoch": 143.52453792224347, "grad_norm": 0.22062504291534424, "learning_rate": 1.605935563793958e-05, "loss": 0.0074, "step": 225190 }, { "epoch": 143.53091140854048, "grad_norm": 0.8106417655944824, "learning_rate": 1.605530863899499e-05, "loss": 0.0113, "step": 225200 }, { "epoch": 143.53728489483748, "grad_norm": 0.10340183973312378, "learning_rate": 1.6051262052510428e-05, "loss": 0.0062, "step": 225210 }, { "epoch": 143.5436583811345, "grad_norm": 0.2961183488368988, "learning_rate": 1.6047215878535097e-05, "loss": 0.0114, "step": 225220 }, { "epoch": 143.55003186743147, "grad_norm": 0.1510486602783203, "learning_rate": 1.6043170117118135e-05, "loss": 0.0103, "step": 225230 }, { "epoch": 143.55640535372848, "grad_norm": 0.19390635192394257, "learning_rate": 1.603912476830875e-05, "loss": 0.0104, "step": 225240 }, { "epoch": 143.5627788400255, "grad_norm": 0.3772571384906769, "learning_rate": 1.6035079832156024e-05, "loss": 0.0108, "step": 225250 }, { "epoch": 143.5691523263225, "grad_norm": 0.200861394405365, "learning_rate": 1.603103530870917e-05, "loss": 0.0053, "step": 225260 }, { "epoch": 143.5755258126195, "grad_norm": 0.26273250579833984, "learning_rate": 1.6026991198017305e-05, "loss": 0.0056, "step": 225270 }, { "epoch": 143.5818992989165, "grad_norm": 0.1712527722120285, "learning_rate": 1.6022947500129574e-05, "loss": 0.0049, "step": 225280 }, { "epoch": 143.58827278521352, "grad_norm": 0.3439937233924866, "learning_rate": 1.6018904215095097e-05, "loss": 0.0046, "step": 225290 }, { "epoch": 143.59464627151053, "grad_norm": 0.20178638398647308, "learning_rate": 1.6014861342963032e-05, "loss": 0.0049, "step": 225300 }, { "epoch": 143.60101975780753, "grad_norm": 0.10588563233613968, "learning_rate": 1.6010818883782487e-05, "loss": 0.0067, "step": 225310 }, { "epoch": 143.60739324410451, "grad_norm": 0.2871628999710083, "learning_rate": 1.6006776837602582e-05, "loss": 0.0103, "step": 225320 }, { "epoch": 143.61376673040152, "grad_norm": 0.20240865647792816, "learning_rate": 1.6002735204472423e-05, "loss": 0.0061, "step": 225330 }, { "epoch": 143.62014021669853, "grad_norm": 0.7223148345947266, "learning_rate": 1.599869398444116e-05, "loss": 0.0085, "step": 225340 }, { "epoch": 143.62651370299554, "grad_norm": 0.35475578904151917, "learning_rate": 1.5994653177557834e-05, "loss": 0.0081, "step": 225350 }, { "epoch": 143.63288718929255, "grad_norm": 0.4761807322502136, "learning_rate": 1.5990612783871596e-05, "loss": 0.007, "step": 225360 }, { "epoch": 143.63926067558955, "grad_norm": 0.1462082415819168, "learning_rate": 1.5986572803431503e-05, "loss": 0.0054, "step": 225370 }, { "epoch": 143.64563416188656, "grad_norm": 0.5286439061164856, "learning_rate": 1.59825332362867e-05, "loss": 0.0129, "step": 225380 }, { "epoch": 143.65200764818357, "grad_norm": 0.14651764929294586, "learning_rate": 1.5978494082486206e-05, "loss": 0.0064, "step": 225390 }, { "epoch": 143.65838113448055, "grad_norm": 0.14293242990970612, "learning_rate": 1.5974455342079143e-05, "loss": 0.0049, "step": 225400 }, { "epoch": 143.66475462077756, "grad_norm": 0.09763351082801819, "learning_rate": 1.5970417015114576e-05, "loss": 0.0038, "step": 225410 }, { "epoch": 143.67112810707457, "grad_norm": 0.33910465240478516, "learning_rate": 1.5966379101641564e-05, "loss": 0.0151, "step": 225420 }, { "epoch": 143.67750159337157, "grad_norm": 0.09708832949399948, "learning_rate": 1.596234160170917e-05, "loss": 0.005, "step": 225430 }, { "epoch": 143.68387507966858, "grad_norm": 0.13284540176391602, "learning_rate": 1.595830451536648e-05, "loss": 0.0073, "step": 225440 }, { "epoch": 143.6902485659656, "grad_norm": 0.22393277287483215, "learning_rate": 1.5954267842662522e-05, "loss": 0.009, "step": 225450 }, { "epoch": 143.6966220522626, "grad_norm": 0.1836738884449005, "learning_rate": 1.5950231583646365e-05, "loss": 0.0168, "step": 225460 }, { "epoch": 143.7029955385596, "grad_norm": 0.17275460064411163, "learning_rate": 1.594619573836702e-05, "loss": 0.0036, "step": 225470 }, { "epoch": 143.70936902485658, "grad_norm": 0.21515274047851562, "learning_rate": 1.5942160306873567e-05, "loss": 0.0099, "step": 225480 }, { "epoch": 143.7157425111536, "grad_norm": 0.3610623776912689, "learning_rate": 1.5938125289215023e-05, "loss": 0.0056, "step": 225490 }, { "epoch": 143.7221159974506, "grad_norm": 0.15492260456085205, "learning_rate": 1.5934090685440413e-05, "loss": 0.0032, "step": 225500 }, { "epoch": 143.7284894837476, "grad_norm": 0.4117622673511505, "learning_rate": 1.593005649559875e-05, "loss": 0.015, "step": 225510 }, { "epoch": 143.73486297004462, "grad_norm": 0.1140320897102356, "learning_rate": 1.5926022719739104e-05, "loss": 0.0061, "step": 225520 }, { "epoch": 143.74123645634162, "grad_norm": 0.2544914186000824, "learning_rate": 1.5921989357910415e-05, "loss": 0.0113, "step": 225530 }, { "epoch": 143.74760994263863, "grad_norm": 0.1354125738143921, "learning_rate": 1.591795641016175e-05, "loss": 0.0062, "step": 225540 }, { "epoch": 143.75398342893564, "grad_norm": 0.24595403671264648, "learning_rate": 1.5913923876542074e-05, "loss": 0.0069, "step": 225550 }, { "epoch": 143.76035691523262, "grad_norm": 0.4165270924568176, "learning_rate": 1.5909891757100444e-05, "loss": 0.0071, "step": 225560 }, { "epoch": 143.76673040152963, "grad_norm": 0.3259887397289276, "learning_rate": 1.590586005188578e-05, "loss": 0.0087, "step": 225570 }, { "epoch": 143.77310388782664, "grad_norm": 0.24704866111278534, "learning_rate": 1.590182876094712e-05, "loss": 0.0145, "step": 225580 }, { "epoch": 143.77947737412364, "grad_norm": 0.42178696393966675, "learning_rate": 1.5897797884333436e-05, "loss": 0.005, "step": 225590 }, { "epoch": 143.78585086042065, "grad_norm": 0.24826645851135254, "learning_rate": 1.589376742209371e-05, "loss": 0.0102, "step": 225600 }, { "epoch": 143.79222434671766, "grad_norm": 0.36860865354537964, "learning_rate": 1.5889737374276886e-05, "loss": 0.0053, "step": 225610 }, { "epoch": 143.79859783301467, "grad_norm": 0.20645494759082794, "learning_rate": 1.5885707740931983e-05, "loss": 0.0046, "step": 225620 }, { "epoch": 143.80497131931168, "grad_norm": 0.16990475356578827, "learning_rate": 1.5881678522107935e-05, "loss": 0.0049, "step": 225630 }, { "epoch": 143.81134480560866, "grad_norm": 0.14551879465579987, "learning_rate": 1.587764971785371e-05, "loss": 0.0035, "step": 225640 }, { "epoch": 143.81771829190566, "grad_norm": 0.35754162073135376, "learning_rate": 1.5873621328218234e-05, "loss": 0.0071, "step": 225650 }, { "epoch": 143.82409177820267, "grad_norm": 0.2524584233760834, "learning_rate": 1.5869593353250505e-05, "loss": 0.0057, "step": 225660 }, { "epoch": 143.83046526449968, "grad_norm": 0.13398537039756775, "learning_rate": 1.586556579299944e-05, "loss": 0.0042, "step": 225670 }, { "epoch": 143.8368387507967, "grad_norm": 0.35300108790397644, "learning_rate": 1.5861538647513975e-05, "loss": 0.0071, "step": 225680 }, { "epoch": 143.8432122370937, "grad_norm": 0.30552247166633606, "learning_rate": 1.5857511916843033e-05, "loss": 0.0082, "step": 225690 }, { "epoch": 143.8495857233907, "grad_norm": 0.11105591058731079, "learning_rate": 1.5853485601035574e-05, "loss": 0.005, "step": 225700 }, { "epoch": 143.8559592096877, "grad_norm": 0.07956469804048538, "learning_rate": 1.58494597001405e-05, "loss": 0.0101, "step": 225710 }, { "epoch": 143.8623326959847, "grad_norm": 0.31186729669570923, "learning_rate": 1.584543421420674e-05, "loss": 0.0047, "step": 225720 }, { "epoch": 143.8687061822817, "grad_norm": 0.18373961746692657, "learning_rate": 1.5841409143283197e-05, "loss": 0.0054, "step": 225730 }, { "epoch": 143.8750796685787, "grad_norm": 0.19255299866199493, "learning_rate": 1.5837384487418772e-05, "loss": 0.0027, "step": 225740 }, { "epoch": 143.88145315487571, "grad_norm": 0.17850063741207123, "learning_rate": 1.583336024666239e-05, "loss": 0.0086, "step": 225750 }, { "epoch": 143.88782664117272, "grad_norm": 0.34518682956695557, "learning_rate": 1.5829336421062947e-05, "loss": 0.0055, "step": 225760 }, { "epoch": 143.89420012746973, "grad_norm": 0.137569859623909, "learning_rate": 1.582531301066933e-05, "loss": 0.007, "step": 225770 }, { "epoch": 143.90057361376674, "grad_norm": 0.1253690868616104, "learning_rate": 1.5821290015530403e-05, "loss": 0.0056, "step": 225780 }, { "epoch": 143.90694710006375, "grad_norm": 0.10396920889616013, "learning_rate": 1.5817267435695095e-05, "loss": 0.0055, "step": 225790 }, { "epoch": 143.91332058636073, "grad_norm": 0.2595668137073517, "learning_rate": 1.5813245271212258e-05, "loss": 0.0081, "step": 225800 }, { "epoch": 143.91969407265773, "grad_norm": 0.20857453346252441, "learning_rate": 1.5809223522130767e-05, "loss": 0.0062, "step": 225810 }, { "epoch": 143.92606755895474, "grad_norm": 0.18881870806217194, "learning_rate": 1.5805202188499472e-05, "loss": 0.0069, "step": 225820 }, { "epoch": 143.93244104525175, "grad_norm": 0.2678162157535553, "learning_rate": 1.580118127036728e-05, "loss": 0.0042, "step": 225830 }, { "epoch": 143.93881453154876, "grad_norm": 0.1871802657842636, "learning_rate": 1.579716076778302e-05, "loss": 0.007, "step": 225840 }, { "epoch": 143.94518801784577, "grad_norm": 0.22992292046546936, "learning_rate": 1.5793140680795554e-05, "loss": 0.0061, "step": 225850 }, { "epoch": 143.95156150414277, "grad_norm": 0.2679927349090576, "learning_rate": 1.5789121009453706e-05, "loss": 0.0086, "step": 225860 }, { "epoch": 143.95793499043978, "grad_norm": 0.3250144422054291, "learning_rate": 1.578510175380635e-05, "loss": 0.0051, "step": 225870 }, { "epoch": 143.96430847673676, "grad_norm": 0.21809595823287964, "learning_rate": 1.5781082913902324e-05, "loss": 0.0057, "step": 225880 }, { "epoch": 143.97068196303377, "grad_norm": 0.21071235835552216, "learning_rate": 1.577706448979044e-05, "loss": 0.0051, "step": 225890 }, { "epoch": 143.97705544933078, "grad_norm": 0.10393699258565903, "learning_rate": 1.5773046481519517e-05, "loss": 0.006, "step": 225900 }, { "epoch": 143.98342893562778, "grad_norm": 0.11725697666406631, "learning_rate": 1.576902888913843e-05, "loss": 0.0049, "step": 225910 }, { "epoch": 143.9898024219248, "grad_norm": 0.16064102947711945, "learning_rate": 1.5765011712695926e-05, "loss": 0.0177, "step": 225920 }, { "epoch": 143.9961759082218, "grad_norm": 0.2206147015094757, "learning_rate": 1.5760994952240877e-05, "loss": 0.0048, "step": 225930 }, { "epoch": 144.0025493945188, "grad_norm": 0.2967018187046051, "learning_rate": 1.5756978607822042e-05, "loss": 0.0047, "step": 225940 }, { "epoch": 144.00892288081582, "grad_norm": 0.45794758200645447, "learning_rate": 1.575296267948828e-05, "loss": 0.0083, "step": 225950 }, { "epoch": 144.01529636711282, "grad_norm": 0.5239172577857971, "learning_rate": 1.574894716728833e-05, "loss": 0.0151, "step": 225960 }, { "epoch": 144.0216698534098, "grad_norm": 0.4421536326408386, "learning_rate": 1.5744932071271022e-05, "loss": 0.009, "step": 225970 }, { "epoch": 144.0280433397068, "grad_norm": 0.4011935591697693, "learning_rate": 1.5740917391485133e-05, "loss": 0.0079, "step": 225980 }, { "epoch": 144.03441682600382, "grad_norm": 0.13928475975990295, "learning_rate": 1.573690312797944e-05, "loss": 0.0094, "step": 225990 }, { "epoch": 144.04079031230083, "grad_norm": 0.1675678789615631, "learning_rate": 1.5732889280802714e-05, "loss": 0.0128, "step": 226000 }, { "epoch": 144.04716379859784, "grad_norm": 0.18004503846168518, "learning_rate": 1.572887585000375e-05, "loss": 0.0058, "step": 226010 }, { "epoch": 144.05353728489484, "grad_norm": 0.4774920344352722, "learning_rate": 1.57248628356313e-05, "loss": 0.0076, "step": 226020 }, { "epoch": 144.05991077119185, "grad_norm": 0.2363836169242859, "learning_rate": 1.572085023773413e-05, "loss": 0.0062, "step": 226030 }, { "epoch": 144.06628425748886, "grad_norm": 0.25725817680358887, "learning_rate": 1.571683805636098e-05, "loss": 0.0111, "step": 226040 }, { "epoch": 144.07265774378584, "grad_norm": 0.18585015833377838, "learning_rate": 1.571282629156064e-05, "loss": 0.016, "step": 226050 }, { "epoch": 144.07903123008285, "grad_norm": 0.3751446008682251, "learning_rate": 1.5708814943381806e-05, "loss": 0.0087, "step": 226060 }, { "epoch": 144.08540471637986, "grad_norm": 0.19250191748142242, "learning_rate": 1.5704804011873264e-05, "loss": 0.0112, "step": 226070 }, { "epoch": 144.09177820267686, "grad_norm": 0.24147222936153412, "learning_rate": 1.5700793497083716e-05, "loss": 0.0049, "step": 226080 }, { "epoch": 144.09815168897387, "grad_norm": 0.29823848605155945, "learning_rate": 1.5696783399061943e-05, "loss": 0.005, "step": 226090 }, { "epoch": 144.10452517527088, "grad_norm": 0.446564257144928, "learning_rate": 1.5692773717856607e-05, "loss": 0.0199, "step": 226100 }, { "epoch": 144.1108986615679, "grad_norm": 0.12644363939762115, "learning_rate": 1.568876445351648e-05, "loss": 0.0064, "step": 226110 }, { "epoch": 144.1172721478649, "grad_norm": 0.28408557176589966, "learning_rate": 1.568475560609024e-05, "loss": 0.0124, "step": 226120 }, { "epoch": 144.12364563416187, "grad_norm": 0.17212222516536713, "learning_rate": 1.5680747175626653e-05, "loss": 0.0145, "step": 226130 }, { "epoch": 144.13001912045888, "grad_norm": 0.24081872403621674, "learning_rate": 1.567673916217436e-05, "loss": 0.0088, "step": 226140 }, { "epoch": 144.1363926067559, "grad_norm": 0.2013551890850067, "learning_rate": 1.5672731565782106e-05, "loss": 0.009, "step": 226150 }, { "epoch": 144.1427660930529, "grad_norm": 0.48688843846321106, "learning_rate": 1.566872438649858e-05, "loss": 0.0065, "step": 226160 }, { "epoch": 144.1491395793499, "grad_norm": 0.10167469829320908, "learning_rate": 1.566471762437246e-05, "loss": 0.0057, "step": 226170 }, { "epoch": 144.15551306564691, "grad_norm": 0.5825548768043518, "learning_rate": 1.5660711279452423e-05, "loss": 0.0165, "step": 226180 }, { "epoch": 144.16188655194392, "grad_norm": 0.10501135140657425, "learning_rate": 1.565670535178718e-05, "loss": 0.0066, "step": 226190 }, { "epoch": 144.16826003824093, "grad_norm": 0.3042411506175995, "learning_rate": 1.56526998414254e-05, "loss": 0.0054, "step": 226200 }, { "epoch": 144.1746335245379, "grad_norm": 0.3476267457008362, "learning_rate": 1.564869474841574e-05, "loss": 0.0105, "step": 226210 }, { "epoch": 144.18100701083492, "grad_norm": 0.2940092980861664, "learning_rate": 1.5644690072806855e-05, "loss": 0.0061, "step": 226220 }, { "epoch": 144.18738049713193, "grad_norm": 0.2107650190591812, "learning_rate": 1.564068581464746e-05, "loss": 0.0051, "step": 226230 }, { "epoch": 144.19375398342893, "grad_norm": 0.11873894929885864, "learning_rate": 1.5636681973986134e-05, "loss": 0.0054, "step": 226240 }, { "epoch": 144.20012746972594, "grad_norm": 0.1269804984331131, "learning_rate": 1.5632678550871587e-05, "loss": 0.0053, "step": 226250 }, { "epoch": 144.20650095602295, "grad_norm": 0.4865741431713104, "learning_rate": 1.562867554535243e-05, "loss": 0.0099, "step": 226260 }, { "epoch": 144.21287444231996, "grad_norm": 0.24571283161640167, "learning_rate": 1.5624672957477343e-05, "loss": 0.008, "step": 226270 }, { "epoch": 144.21924792861697, "grad_norm": 0.3333870470523834, "learning_rate": 1.5620670787294904e-05, "loss": 0.0038, "step": 226280 }, { "epoch": 144.22562141491395, "grad_norm": 0.1402440071105957, "learning_rate": 1.5616669034853794e-05, "loss": 0.0069, "step": 226290 }, { "epoch": 144.23199490121095, "grad_norm": 0.19813287258148193, "learning_rate": 1.5612667700202615e-05, "loss": 0.0095, "step": 226300 }, { "epoch": 144.23836838750796, "grad_norm": 0.19056899845600128, "learning_rate": 1.5608666783389997e-05, "loss": 0.0085, "step": 226310 }, { "epoch": 144.24474187380497, "grad_norm": 0.13832026720046997, "learning_rate": 1.5604666284464526e-05, "loss": 0.0104, "step": 226320 }, { "epoch": 144.25111536010198, "grad_norm": 0.12606970965862274, "learning_rate": 1.560066620347485e-05, "loss": 0.0049, "step": 226330 }, { "epoch": 144.25748884639899, "grad_norm": 0.12007533013820648, "learning_rate": 1.559666654046956e-05, "loss": 0.0064, "step": 226340 }, { "epoch": 144.263862332696, "grad_norm": 0.571018636226654, "learning_rate": 1.5592667295497253e-05, "loss": 0.0042, "step": 226350 }, { "epoch": 144.270235818993, "grad_norm": 0.2971462309360504, "learning_rate": 1.558866846860651e-05, "loss": 0.0058, "step": 226360 }, { "epoch": 144.27660930528998, "grad_norm": 0.1761477291584015, "learning_rate": 1.5584670059845952e-05, "loss": 0.0081, "step": 226370 }, { "epoch": 144.282982791587, "grad_norm": 0.20375579595565796, "learning_rate": 1.5580672069264142e-05, "loss": 0.0044, "step": 226380 }, { "epoch": 144.289356277884, "grad_norm": 0.1232871487736702, "learning_rate": 1.5576674496909664e-05, "loss": 0.007, "step": 226390 }, { "epoch": 144.295729764181, "grad_norm": 0.19285699725151062, "learning_rate": 1.557267734283108e-05, "loss": 0.006, "step": 226400 }, { "epoch": 144.302103250478, "grad_norm": 0.1792483627796173, "learning_rate": 1.5568680607076984e-05, "loss": 0.006, "step": 226410 }, { "epoch": 144.30847673677502, "grad_norm": 0.3803780972957611, "learning_rate": 1.556468428969593e-05, "loss": 0.0074, "step": 226420 }, { "epoch": 144.31485022307203, "grad_norm": 0.19736003875732422, "learning_rate": 1.5560688390736474e-05, "loss": 0.0086, "step": 226430 }, { "epoch": 144.32122370936904, "grad_norm": 0.2793513238430023, "learning_rate": 1.555669291024715e-05, "loss": 0.0066, "step": 226440 }, { "epoch": 144.32759719566602, "grad_norm": 0.14551664888858795, "learning_rate": 1.5552697848276544e-05, "loss": 0.006, "step": 226450 }, { "epoch": 144.33397068196302, "grad_norm": 0.17530031502246857, "learning_rate": 1.5548703204873184e-05, "loss": 0.007, "step": 226460 }, { "epoch": 144.34034416826003, "grad_norm": 0.1382693499326706, "learning_rate": 1.5544708980085603e-05, "loss": 0.0062, "step": 226470 }, { "epoch": 144.34671765455704, "grad_norm": 0.14815402030944824, "learning_rate": 1.5540715173962345e-05, "loss": 0.0109, "step": 226480 }, { "epoch": 144.35309114085405, "grad_norm": 0.26904764771461487, "learning_rate": 1.5536721786551916e-05, "loss": 0.0054, "step": 226490 }, { "epoch": 144.35946462715106, "grad_norm": 0.20341363549232483, "learning_rate": 1.553272881790287e-05, "loss": 0.009, "step": 226500 }, { "epoch": 144.36583811344806, "grad_norm": 0.346107542514801, "learning_rate": 1.55287362680637e-05, "loss": 0.0065, "step": 226510 }, { "epoch": 144.37221159974507, "grad_norm": 0.15285775065422058, "learning_rate": 1.552474413708296e-05, "loss": 0.0115, "step": 226520 }, { "epoch": 144.37858508604205, "grad_norm": 0.23759254813194275, "learning_rate": 1.5520752425009096e-05, "loss": 0.0072, "step": 226530 }, { "epoch": 144.38495857233906, "grad_norm": 0.13152790069580078, "learning_rate": 1.5516761131890657e-05, "loss": 0.0063, "step": 226540 }, { "epoch": 144.39133205863607, "grad_norm": 0.2234703004360199, "learning_rate": 1.5512770257776127e-05, "loss": 0.0046, "step": 226550 }, { "epoch": 144.39770554493307, "grad_norm": 0.17297591269016266, "learning_rate": 1.5508779802714007e-05, "loss": 0.0046, "step": 226560 }, { "epoch": 144.40407903123008, "grad_norm": 0.22704926133155823, "learning_rate": 1.5504789766752754e-05, "loss": 0.0115, "step": 226570 }, { "epoch": 144.4104525175271, "grad_norm": 0.1405377984046936, "learning_rate": 1.5500800149940897e-05, "loss": 0.0073, "step": 226580 }, { "epoch": 144.4168260038241, "grad_norm": 0.2320549637079239, "learning_rate": 1.5496810952326886e-05, "loss": 0.0092, "step": 226590 }, { "epoch": 144.4231994901211, "grad_norm": 0.23624710738658905, "learning_rate": 1.5492822173959197e-05, "loss": 0.0052, "step": 226600 }, { "epoch": 144.42957297641811, "grad_norm": 0.17679013311862946, "learning_rate": 1.548883381488629e-05, "loss": 0.0088, "step": 226610 }, { "epoch": 144.4359464627151, "grad_norm": 0.2825126647949219, "learning_rate": 1.548484587515666e-05, "loss": 0.0063, "step": 226620 }, { "epoch": 144.4423199490121, "grad_norm": 0.18597902357578278, "learning_rate": 1.5480858354818718e-05, "loss": 0.0066, "step": 226630 }, { "epoch": 144.4486934353091, "grad_norm": 0.14739449322223663, "learning_rate": 1.5476871253920943e-05, "loss": 0.0049, "step": 226640 }, { "epoch": 144.45506692160612, "grad_norm": 0.14840783178806305, "learning_rate": 1.547288457251177e-05, "loss": 0.0067, "step": 226650 }, { "epoch": 144.46144040790313, "grad_norm": 0.10507746785879135, "learning_rate": 1.546889831063968e-05, "loss": 0.0055, "step": 226660 }, { "epoch": 144.46781389420013, "grad_norm": 0.29006993770599365, "learning_rate": 1.5464912468353042e-05, "loss": 0.007, "step": 226670 }, { "epoch": 144.47418738049714, "grad_norm": 0.27605751156806946, "learning_rate": 1.546092704570034e-05, "loss": 0.0085, "step": 226680 }, { "epoch": 144.48056086679415, "grad_norm": 0.12746313214302063, "learning_rate": 1.5456942042729972e-05, "loss": 0.0071, "step": 226690 }, { "epoch": 144.48693435309113, "grad_norm": 0.39404764771461487, "learning_rate": 1.54529574594904e-05, "loss": 0.006, "step": 226700 }, { "epoch": 144.49330783938814, "grad_norm": 0.09981734305620193, "learning_rate": 1.5448973296029985e-05, "loss": 0.0048, "step": 226710 }, { "epoch": 144.49968132568515, "grad_norm": 0.4023379385471344, "learning_rate": 1.5444989552397184e-05, "loss": 0.0202, "step": 226720 }, { "epoch": 144.50605481198215, "grad_norm": 0.20769183337688446, "learning_rate": 1.5441006228640383e-05, "loss": 0.0077, "step": 226730 }, { "epoch": 144.51242829827916, "grad_norm": 0.22086171805858612, "learning_rate": 1.543702332480798e-05, "loss": 0.0097, "step": 226740 }, { "epoch": 144.51880178457617, "grad_norm": 0.5169291496276855, "learning_rate": 1.5433040840948366e-05, "loss": 0.0067, "step": 226750 }, { "epoch": 144.52517527087318, "grad_norm": 0.12141413986682892, "learning_rate": 1.5429058777109957e-05, "loss": 0.0071, "step": 226760 }, { "epoch": 144.53154875717019, "grad_norm": 0.2656896412372589, "learning_rate": 1.542507713334112e-05, "loss": 0.0086, "step": 226770 }, { "epoch": 144.53792224346716, "grad_norm": 0.3388252854347229, "learning_rate": 1.5421095909690248e-05, "loss": 0.0067, "step": 226780 }, { "epoch": 144.54429572976417, "grad_norm": 0.10282392054796219, "learning_rate": 1.5417115106205686e-05, "loss": 0.003, "step": 226790 }, { "epoch": 144.55066921606118, "grad_norm": 0.13434430956840515, "learning_rate": 1.5413134722935864e-05, "loss": 0.0069, "step": 226800 }, { "epoch": 144.5570427023582, "grad_norm": 0.4848576486110687, "learning_rate": 1.540915475992908e-05, "loss": 0.0097, "step": 226810 }, { "epoch": 144.5634161886552, "grad_norm": 0.18064625561237335, "learning_rate": 1.5405175217233737e-05, "loss": 0.0039, "step": 226820 }, { "epoch": 144.5697896749522, "grad_norm": 0.15147998929023743, "learning_rate": 1.5401196094898156e-05, "loss": 0.0036, "step": 226830 }, { "epoch": 144.5761631612492, "grad_norm": 0.0933188647031784, "learning_rate": 1.5397217392970746e-05, "loss": 0.0043, "step": 226840 }, { "epoch": 144.58253664754622, "grad_norm": 0.08971000462770462, "learning_rate": 1.5393239111499792e-05, "loss": 0.0097, "step": 226850 }, { "epoch": 144.5889101338432, "grad_norm": 0.25875234603881836, "learning_rate": 1.538926125053366e-05, "loss": 0.0095, "step": 226860 }, { "epoch": 144.5952836201402, "grad_norm": 0.4853735566139221, "learning_rate": 1.5385283810120688e-05, "loss": 0.0104, "step": 226870 }, { "epoch": 144.60165710643722, "grad_norm": 0.36051657795906067, "learning_rate": 1.53813067903092e-05, "loss": 0.0048, "step": 226880 }, { "epoch": 144.60803059273422, "grad_norm": 0.201873779296875, "learning_rate": 1.5377330191147504e-05, "loss": 0.0085, "step": 226890 }, { "epoch": 144.61440407903123, "grad_norm": 0.6413206458091736, "learning_rate": 1.537335401268395e-05, "loss": 0.0066, "step": 226900 }, { "epoch": 144.62077756532824, "grad_norm": 0.07918088138103485, "learning_rate": 1.5369378254966836e-05, "loss": 0.0079, "step": 226910 }, { "epoch": 144.62715105162525, "grad_norm": 0.3929083049297333, "learning_rate": 1.5365402918044475e-05, "loss": 0.0095, "step": 226920 }, { "epoch": 144.63352453792226, "grad_norm": 0.16377092897891998, "learning_rate": 1.5361428001965155e-05, "loss": 0.003, "step": 226930 }, { "epoch": 144.63989802421924, "grad_norm": 0.12919993698596954, "learning_rate": 1.535745350677722e-05, "loss": 0.0071, "step": 226940 }, { "epoch": 144.64627151051624, "grad_norm": 0.4291882812976837, "learning_rate": 1.5353479432528895e-05, "loss": 0.0052, "step": 226950 }, { "epoch": 144.65264499681325, "grad_norm": 0.18295274674892426, "learning_rate": 1.534950577926853e-05, "loss": 0.0072, "step": 226960 }, { "epoch": 144.65901848311026, "grad_norm": 0.33266887068748474, "learning_rate": 1.5345532547044365e-05, "loss": 0.0102, "step": 226970 }, { "epoch": 144.66539196940727, "grad_norm": 0.25147271156311035, "learning_rate": 1.5341559735904732e-05, "loss": 0.0122, "step": 226980 }, { "epoch": 144.67176545570427, "grad_norm": 0.2979556620121002, "learning_rate": 1.5337587345897842e-05, "loss": 0.0041, "step": 226990 }, { "epoch": 144.67813894200128, "grad_norm": 0.1354970932006836, "learning_rate": 1.5333615377072003e-05, "loss": 0.004, "step": 227000 }, { "epoch": 144.6845124282983, "grad_norm": 0.20539863407611847, "learning_rate": 1.5329643829475453e-05, "loss": 0.0049, "step": 227010 }, { "epoch": 144.69088591459527, "grad_norm": 0.22526027262210846, "learning_rate": 1.53256727031565e-05, "loss": 0.0042, "step": 227020 }, { "epoch": 144.69725940089228, "grad_norm": 0.510891318321228, "learning_rate": 1.532170199816333e-05, "loss": 0.0084, "step": 227030 }, { "epoch": 144.7036328871893, "grad_norm": 0.20799261331558228, "learning_rate": 1.5317731714544237e-05, "loss": 0.0044, "step": 227040 }, { "epoch": 144.7100063734863, "grad_norm": 0.193157359957695, "learning_rate": 1.5313761852347448e-05, "loss": 0.006, "step": 227050 }, { "epoch": 144.7163798597833, "grad_norm": 0.2458757609128952, "learning_rate": 1.5309792411621206e-05, "loss": 0.0074, "step": 227060 }, { "epoch": 144.7227533460803, "grad_norm": 0.37587881088256836, "learning_rate": 1.5305823392413717e-05, "loss": 0.0064, "step": 227070 }, { "epoch": 144.72912683237732, "grad_norm": 0.2452796995639801, "learning_rate": 1.5301854794773245e-05, "loss": 0.0077, "step": 227080 }, { "epoch": 144.73550031867433, "grad_norm": 0.18270200490951538, "learning_rate": 1.5297886618748003e-05, "loss": 0.0118, "step": 227090 }, { "epoch": 144.7418738049713, "grad_norm": 0.2584265172481537, "learning_rate": 1.5293918864386188e-05, "loss": 0.0121, "step": 227100 }, { "epoch": 144.7482472912683, "grad_norm": 0.18838201463222504, "learning_rate": 1.528995153173604e-05, "loss": 0.0077, "step": 227110 }, { "epoch": 144.75462077756532, "grad_norm": 0.3198460340499878, "learning_rate": 1.5285984620845757e-05, "loss": 0.0063, "step": 227120 }, { "epoch": 144.76099426386233, "grad_norm": 0.1506870985031128, "learning_rate": 1.528201813176353e-05, "loss": 0.0047, "step": 227130 }, { "epoch": 144.76736775015934, "grad_norm": 0.13989463448524475, "learning_rate": 1.5278052064537545e-05, "loss": 0.0097, "step": 227140 }, { "epoch": 144.77374123645635, "grad_norm": 0.29756104946136475, "learning_rate": 1.527408641921603e-05, "loss": 0.013, "step": 227150 }, { "epoch": 144.78011472275335, "grad_norm": 0.373996376991272, "learning_rate": 1.5270121195847152e-05, "loss": 0.0042, "step": 227160 }, { "epoch": 144.78648820905036, "grad_norm": 0.174320250749588, "learning_rate": 1.5266156394479092e-05, "loss": 0.0133, "step": 227170 }, { "epoch": 144.79286169534734, "grad_norm": 0.42530688643455505, "learning_rate": 1.526219201516001e-05, "loss": 0.0087, "step": 227180 }, { "epoch": 144.79923518164435, "grad_norm": 0.2248036116361618, "learning_rate": 1.5258228057938123e-05, "loss": 0.0048, "step": 227190 }, { "epoch": 144.80560866794136, "grad_norm": 0.20847749710083008, "learning_rate": 1.5254264522861538e-05, "loss": 0.0057, "step": 227200 }, { "epoch": 144.81198215423836, "grad_norm": 0.14450982213020325, "learning_rate": 1.5250301409978458e-05, "loss": 0.0041, "step": 227210 }, { "epoch": 144.81835564053537, "grad_norm": 0.41099387407302856, "learning_rate": 1.524633871933701e-05, "loss": 0.0049, "step": 227220 }, { "epoch": 144.82472912683238, "grad_norm": 0.1275016814470291, "learning_rate": 1.524237645098539e-05, "loss": 0.0029, "step": 227230 }, { "epoch": 144.8311026131294, "grad_norm": 0.17501626908779144, "learning_rate": 1.523841460497168e-05, "loss": 0.0056, "step": 227240 }, { "epoch": 144.8374760994264, "grad_norm": 0.12514974176883698, "learning_rate": 1.5234453181344071e-05, "loss": 0.0093, "step": 227250 }, { "epoch": 144.8438495857234, "grad_norm": 0.23536182940006256, "learning_rate": 1.523049218015068e-05, "loss": 0.0082, "step": 227260 }, { "epoch": 144.85022307202038, "grad_norm": 0.24760855734348297, "learning_rate": 1.5226531601439636e-05, "loss": 0.007, "step": 227270 }, { "epoch": 144.8565965583174, "grad_norm": 0.19204390048980713, "learning_rate": 1.5222571445259048e-05, "loss": 0.0052, "step": 227280 }, { "epoch": 144.8629700446144, "grad_norm": 0.2141609936952591, "learning_rate": 1.521861171165707e-05, "loss": 0.0068, "step": 227290 }, { "epoch": 144.8693435309114, "grad_norm": 0.37453508377075195, "learning_rate": 1.5214652400681801e-05, "loss": 0.0063, "step": 227300 }, { "epoch": 144.87571701720842, "grad_norm": 0.13136789202690125, "learning_rate": 1.521069351238134e-05, "loss": 0.0059, "step": 227310 }, { "epoch": 144.88209050350542, "grad_norm": 0.33715367317199707, "learning_rate": 1.5206735046803793e-05, "loss": 0.0058, "step": 227320 }, { "epoch": 144.88846398980243, "grad_norm": 0.5665774345397949, "learning_rate": 1.5202777003997275e-05, "loss": 0.0082, "step": 227330 }, { "epoch": 144.89483747609944, "grad_norm": 0.13421426713466644, "learning_rate": 1.519881938400987e-05, "loss": 0.0109, "step": 227340 }, { "epoch": 144.90121096239642, "grad_norm": 0.12596619129180908, "learning_rate": 1.5194862186889675e-05, "loss": 0.0029, "step": 227350 }, { "epoch": 144.90758444869343, "grad_norm": 0.2467360496520996, "learning_rate": 1.5190905412684741e-05, "loss": 0.0056, "step": 227360 }, { "epoch": 144.91395793499044, "grad_norm": 0.20946185290813446, "learning_rate": 1.5186949061443207e-05, "loss": 0.0054, "step": 227370 }, { "epoch": 144.92033142128744, "grad_norm": 0.10979580134153366, "learning_rate": 1.5182993133213081e-05, "loss": 0.0052, "step": 227380 }, { "epoch": 144.92670490758445, "grad_norm": 0.19245946407318115, "learning_rate": 1.5179037628042475e-05, "loss": 0.0093, "step": 227390 }, { "epoch": 144.93307839388146, "grad_norm": 0.32920244336128235, "learning_rate": 1.5175082545979424e-05, "loss": 0.0063, "step": 227400 }, { "epoch": 144.93945188017847, "grad_norm": 0.21090343594551086, "learning_rate": 1.5171127887072029e-05, "loss": 0.0048, "step": 227410 }, { "epoch": 144.94582536647547, "grad_norm": 0.1193012073636055, "learning_rate": 1.5167173651368283e-05, "loss": 0.0035, "step": 227420 }, { "epoch": 144.95219885277245, "grad_norm": 0.15372231602668762, "learning_rate": 1.5163219838916282e-05, "loss": 0.0067, "step": 227430 }, { "epoch": 144.95857233906946, "grad_norm": 0.27629056572914124, "learning_rate": 1.5159266449764049e-05, "loss": 0.0037, "step": 227440 }, { "epoch": 144.96494582536647, "grad_norm": 0.10327226668596268, "learning_rate": 1.5155313483959626e-05, "loss": 0.0055, "step": 227450 }, { "epoch": 144.97131931166348, "grad_norm": 0.11626359075307846, "learning_rate": 1.5151360941551029e-05, "loss": 0.0034, "step": 227460 }, { "epoch": 144.9776927979605, "grad_norm": 0.33307185769081116, "learning_rate": 1.5147408822586311e-05, "loss": 0.0041, "step": 227470 }, { "epoch": 144.9840662842575, "grad_norm": 0.4706706702709198, "learning_rate": 1.5143457127113486e-05, "loss": 0.0079, "step": 227480 }, { "epoch": 144.9904397705545, "grad_norm": 0.11135484278202057, "learning_rate": 1.5139505855180564e-05, "loss": 0.0114, "step": 227490 }, { "epoch": 144.9968132568515, "grad_norm": 0.18449585139751434, "learning_rate": 1.5135555006835544e-05, "loss": 0.0057, "step": 227500 }, { "epoch": 145.0031867431485, "grad_norm": 0.12175548076629639, "learning_rate": 1.513160458212648e-05, "loss": 0.0072, "step": 227510 }, { "epoch": 145.0095602294455, "grad_norm": 0.11966999620199203, "learning_rate": 1.5127654581101309e-05, "loss": 0.0069, "step": 227520 }, { "epoch": 145.0159337157425, "grad_norm": 0.21075889468193054, "learning_rate": 1.5123705003808075e-05, "loss": 0.0044, "step": 227530 }, { "epoch": 145.0223072020395, "grad_norm": 0.2459905445575714, "learning_rate": 1.5119755850294743e-05, "loss": 0.0123, "step": 227540 }, { "epoch": 145.02868068833652, "grad_norm": 0.3561978340148926, "learning_rate": 1.5115807120609337e-05, "loss": 0.0055, "step": 227550 }, { "epoch": 145.03505417463353, "grad_norm": 0.3527687191963196, "learning_rate": 1.5111858814799778e-05, "loss": 0.0066, "step": 227560 }, { "epoch": 145.04142766093054, "grad_norm": 0.2531128227710724, "learning_rate": 1.5107910932914093e-05, "loss": 0.0062, "step": 227570 }, { "epoch": 145.04780114722755, "grad_norm": 0.25589871406555176, "learning_rate": 1.510396347500022e-05, "loss": 0.0054, "step": 227580 }, { "epoch": 145.05417463352453, "grad_norm": 0.19705812633037567, "learning_rate": 1.5100016441106168e-05, "loss": 0.0088, "step": 227590 }, { "epoch": 145.06054811982153, "grad_norm": 0.3456308841705322, "learning_rate": 1.5096069831279835e-05, "loss": 0.009, "step": 227600 }, { "epoch": 145.06692160611854, "grad_norm": 0.10517881065607071, "learning_rate": 1.5092123645569222e-05, "loss": 0.007, "step": 227610 }, { "epoch": 145.07329509241555, "grad_norm": 0.20686915516853333, "learning_rate": 1.5088177884022264e-05, "loss": 0.007, "step": 227620 }, { "epoch": 145.07966857871256, "grad_norm": 0.2201867550611496, "learning_rate": 1.508423254668691e-05, "loss": 0.0075, "step": 227630 }, { "epoch": 145.08604206500956, "grad_norm": 0.19900812208652496, "learning_rate": 1.5080287633611085e-05, "loss": 0.0053, "step": 227640 }, { "epoch": 145.09241555130657, "grad_norm": 0.5026986002922058, "learning_rate": 1.5076343144842747e-05, "loss": 0.007, "step": 227650 }, { "epoch": 145.09878903760358, "grad_norm": 0.09794694185256958, "learning_rate": 1.507239908042981e-05, "loss": 0.0074, "step": 227660 }, { "epoch": 145.10516252390056, "grad_norm": 0.22241294384002686, "learning_rate": 1.5068455440420204e-05, "loss": 0.0062, "step": 227670 }, { "epoch": 145.11153601019757, "grad_norm": 0.21343955397605896, "learning_rate": 1.5064512224861832e-05, "loss": 0.0055, "step": 227680 }, { "epoch": 145.11790949649458, "grad_norm": 0.22179071605205536, "learning_rate": 1.506056943380263e-05, "loss": 0.007, "step": 227690 }, { "epoch": 145.12428298279158, "grad_norm": 0.13945268094539642, "learning_rate": 1.5056627067290503e-05, "loss": 0.0079, "step": 227700 }, { "epoch": 145.1306564690886, "grad_norm": 0.09631778299808502, "learning_rate": 1.5052685125373351e-05, "loss": 0.0051, "step": 227710 }, { "epoch": 145.1370299553856, "grad_norm": 0.15014615654945374, "learning_rate": 1.5048743608099047e-05, "loss": 0.0051, "step": 227720 }, { "epoch": 145.1434034416826, "grad_norm": 0.11132030189037323, "learning_rate": 1.5044802515515533e-05, "loss": 0.0034, "step": 227730 }, { "epoch": 145.14977692797962, "grad_norm": 0.5083252787590027, "learning_rate": 1.5040861847670662e-05, "loss": 0.0062, "step": 227740 }, { "epoch": 145.1561504142766, "grad_norm": 0.17011107504367828, "learning_rate": 1.503692160461233e-05, "loss": 0.0135, "step": 227750 }, { "epoch": 145.1625239005736, "grad_norm": 0.23517678678035736, "learning_rate": 1.5032981786388417e-05, "loss": 0.0066, "step": 227760 }, { "epoch": 145.1688973868706, "grad_norm": 0.3698781132698059, "learning_rate": 1.5029042393046767e-05, "loss": 0.0159, "step": 227770 }, { "epoch": 145.17527087316762, "grad_norm": 0.2571030259132385, "learning_rate": 1.502510342463529e-05, "loss": 0.0047, "step": 227780 }, { "epoch": 145.18164435946463, "grad_norm": 0.2895526587963104, "learning_rate": 1.5021164881201827e-05, "loss": 0.0116, "step": 227790 }, { "epoch": 145.18801784576164, "grad_norm": 0.15457205474376678, "learning_rate": 1.5017226762794233e-05, "loss": 0.0044, "step": 227800 }, { "epoch": 145.19439133205864, "grad_norm": 0.11645389348268509, "learning_rate": 1.5013289069460351e-05, "loss": 0.0071, "step": 227810 }, { "epoch": 145.20076481835565, "grad_norm": 0.09271256625652313, "learning_rate": 1.500935180124805e-05, "loss": 0.0069, "step": 227820 }, { "epoch": 145.20713830465266, "grad_norm": 0.27980831265449524, "learning_rate": 1.5005414958205166e-05, "loss": 0.0043, "step": 227830 }, { "epoch": 145.21351179094964, "grad_norm": 0.29680517315864563, "learning_rate": 1.5001478540379533e-05, "loss": 0.0078, "step": 227840 }, { "epoch": 145.21988527724665, "grad_norm": 0.4103247821331024, "learning_rate": 1.499754254781896e-05, "loss": 0.0132, "step": 227850 }, { "epoch": 145.22625876354365, "grad_norm": 0.20181937515735626, "learning_rate": 1.4993606980571306e-05, "loss": 0.0058, "step": 227860 }, { "epoch": 145.23263224984066, "grad_norm": 0.22445617616176605, "learning_rate": 1.4989671838684383e-05, "loss": 0.0039, "step": 227870 }, { "epoch": 145.23900573613767, "grad_norm": 0.18537554144859314, "learning_rate": 1.4985737122206001e-05, "loss": 0.0034, "step": 227880 }, { "epoch": 145.24537922243468, "grad_norm": 0.19707810878753662, "learning_rate": 1.4981802831183961e-05, "loss": 0.0121, "step": 227890 }, { "epoch": 145.2517527087317, "grad_norm": 0.15114180743694305, "learning_rate": 1.4977868965666087e-05, "loss": 0.0047, "step": 227900 }, { "epoch": 145.2581261950287, "grad_norm": 0.10661014169454575, "learning_rate": 1.4973935525700179e-05, "loss": 0.0064, "step": 227910 }, { "epoch": 145.26449968132567, "grad_norm": 0.11158113181591034, "learning_rate": 1.497000251133402e-05, "loss": 0.0056, "step": 227920 }, { "epoch": 145.27087316762268, "grad_norm": 0.11610811203718185, "learning_rate": 1.4966069922615389e-05, "loss": 0.0052, "step": 227930 }, { "epoch": 145.2772466539197, "grad_norm": 0.08212965726852417, "learning_rate": 1.4962137759592115e-05, "loss": 0.005, "step": 227940 }, { "epoch": 145.2836201402167, "grad_norm": 0.1767670065164566, "learning_rate": 1.4958206022311922e-05, "loss": 0.0032, "step": 227950 }, { "epoch": 145.2899936265137, "grad_norm": 0.15960727632045746, "learning_rate": 1.4954274710822625e-05, "loss": 0.0036, "step": 227960 }, { "epoch": 145.2963671128107, "grad_norm": 0.22515419125556946, "learning_rate": 1.4950343825171958e-05, "loss": 0.0048, "step": 227970 }, { "epoch": 145.30274059910772, "grad_norm": 0.30536845326423645, "learning_rate": 1.4946413365407747e-05, "loss": 0.0083, "step": 227980 }, { "epoch": 145.30911408540473, "grad_norm": 0.13144922256469727, "learning_rate": 1.4942483331577667e-05, "loss": 0.0051, "step": 227990 }, { "epoch": 145.3154875717017, "grad_norm": 0.31422317028045654, "learning_rate": 1.4938553723729537e-05, "loss": 0.0043, "step": 228000 }, { "epoch": 145.32186105799872, "grad_norm": 0.40613317489624023, "learning_rate": 1.4934624541911085e-05, "loss": 0.0162, "step": 228010 }, { "epoch": 145.32823454429573, "grad_norm": 0.16920389235019684, "learning_rate": 1.4930695786170046e-05, "loss": 0.0075, "step": 228020 }, { "epoch": 145.33460803059273, "grad_norm": 0.3333336114883423, "learning_rate": 1.4926767456554152e-05, "loss": 0.0119, "step": 228030 }, { "epoch": 145.34098151688974, "grad_norm": 0.24528813362121582, "learning_rate": 1.4922839553111162e-05, "loss": 0.0081, "step": 228040 }, { "epoch": 145.34735500318675, "grad_norm": 0.4606827199459076, "learning_rate": 1.4918912075888792e-05, "loss": 0.0134, "step": 228050 }, { "epoch": 145.35372848948376, "grad_norm": 0.4057120084762573, "learning_rate": 1.4914985024934757e-05, "loss": 0.0102, "step": 228060 }, { "epoch": 145.36010197578076, "grad_norm": 0.3306807279586792, "learning_rate": 1.4911058400296768e-05, "loss": 0.0054, "step": 228070 }, { "epoch": 145.36647546207774, "grad_norm": 0.1756065934896469, "learning_rate": 1.4907132202022583e-05, "loss": 0.0092, "step": 228080 }, { "epoch": 145.37284894837475, "grad_norm": 0.23395581543445587, "learning_rate": 1.4903206430159839e-05, "loss": 0.0075, "step": 228090 }, { "epoch": 145.37922243467176, "grad_norm": 0.07639873772859573, "learning_rate": 1.4899281084756295e-05, "loss": 0.0103, "step": 228100 }, { "epoch": 145.38559592096877, "grad_norm": 0.1367160826921463, "learning_rate": 1.4895356165859603e-05, "loss": 0.0047, "step": 228110 }, { "epoch": 145.39196940726578, "grad_norm": 0.3954692482948303, "learning_rate": 1.4891431673517514e-05, "loss": 0.0078, "step": 228120 }, { "epoch": 145.39834289356278, "grad_norm": 0.19256088137626648, "learning_rate": 1.4887507607777646e-05, "loss": 0.0059, "step": 228130 }, { "epoch": 145.4047163798598, "grad_norm": 0.27945202589035034, "learning_rate": 1.4883583968687725e-05, "loss": 0.0054, "step": 228140 }, { "epoch": 145.4110898661568, "grad_norm": 0.3534424602985382, "learning_rate": 1.4879660756295422e-05, "loss": 0.0053, "step": 228150 }, { "epoch": 145.41746335245378, "grad_norm": 0.42237770557403564, "learning_rate": 1.487573797064839e-05, "loss": 0.0092, "step": 228160 }, { "epoch": 145.4238368387508, "grad_norm": 0.1444535255432129, "learning_rate": 1.4871815611794293e-05, "loss": 0.0064, "step": 228170 }, { "epoch": 145.4302103250478, "grad_norm": 0.1489943116903305, "learning_rate": 1.486789367978082e-05, "loss": 0.0056, "step": 228180 }, { "epoch": 145.4365838113448, "grad_norm": 0.35448843240737915, "learning_rate": 1.486397217465561e-05, "loss": 0.0067, "step": 228190 }, { "epoch": 145.4429572976418, "grad_norm": 0.161909282207489, "learning_rate": 1.486005109646631e-05, "loss": 0.0061, "step": 228200 }, { "epoch": 145.44933078393882, "grad_norm": 0.2328248918056488, "learning_rate": 1.4856130445260552e-05, "loss": 0.0027, "step": 228210 }, { "epoch": 145.45570427023583, "grad_norm": 0.15061675012111664, "learning_rate": 1.4852210221086005e-05, "loss": 0.0081, "step": 228220 }, { "epoch": 145.46207775653284, "grad_norm": 0.18382276594638824, "learning_rate": 1.4848290423990291e-05, "loss": 0.0064, "step": 228230 }, { "epoch": 145.46845124282981, "grad_norm": 0.24096079170703888, "learning_rate": 1.4844371054021033e-05, "loss": 0.0046, "step": 228240 }, { "epoch": 145.47482472912682, "grad_norm": 0.1333199441432953, "learning_rate": 1.4840452111225844e-05, "loss": 0.003, "step": 228250 }, { "epoch": 145.48119821542383, "grad_norm": 0.6774123311042786, "learning_rate": 1.4836533595652391e-05, "loss": 0.0107, "step": 228260 }, { "epoch": 145.48757170172084, "grad_norm": 0.11709422618150711, "learning_rate": 1.4832615507348224e-05, "loss": 0.0071, "step": 228270 }, { "epoch": 145.49394518801785, "grad_norm": 0.37946441769599915, "learning_rate": 1.4828697846360995e-05, "loss": 0.0071, "step": 228280 }, { "epoch": 145.50031867431485, "grad_norm": 0.141106516122818, "learning_rate": 1.4824780612738282e-05, "loss": 0.0057, "step": 228290 }, { "epoch": 145.50669216061186, "grad_norm": 0.4044273793697357, "learning_rate": 1.4820863806527724e-05, "loss": 0.0068, "step": 228300 }, { "epoch": 145.51306564690887, "grad_norm": 0.28323909640312195, "learning_rate": 1.4816947427776857e-05, "loss": 0.0051, "step": 228310 }, { "epoch": 145.51943913320585, "grad_norm": 0.12003398686647415, "learning_rate": 1.4813031476533313e-05, "loss": 0.0041, "step": 228320 }, { "epoch": 145.52581261950286, "grad_norm": 0.35429275035858154, "learning_rate": 1.4809115952844655e-05, "loss": 0.0077, "step": 228330 }, { "epoch": 145.53218610579987, "grad_norm": 0.1340225785970688, "learning_rate": 1.4805200856758466e-05, "loss": 0.0072, "step": 228340 }, { "epoch": 145.53855959209687, "grad_norm": 0.6338981986045837, "learning_rate": 1.4801286188322299e-05, "loss": 0.0101, "step": 228350 }, { "epoch": 145.54493307839388, "grad_norm": 0.25278061628341675, "learning_rate": 1.4797371947583755e-05, "loss": 0.0106, "step": 228360 }, { "epoch": 145.5513065646909, "grad_norm": 0.22144877910614014, "learning_rate": 1.4793458134590376e-05, "loss": 0.0057, "step": 228370 }, { "epoch": 145.5576800509879, "grad_norm": 0.46135491132736206, "learning_rate": 1.4789544749389728e-05, "loss": 0.0079, "step": 228380 }, { "epoch": 145.5640535372849, "grad_norm": 0.19717702269554138, "learning_rate": 1.478563179202933e-05, "loss": 0.0062, "step": 228390 }, { "epoch": 145.57042702358189, "grad_norm": 0.16003091633319855, "learning_rate": 1.4781719262556776e-05, "loss": 0.0067, "step": 228400 }, { "epoch": 145.5768005098789, "grad_norm": 0.24014656245708466, "learning_rate": 1.4777807161019586e-05, "loss": 0.0039, "step": 228410 }, { "epoch": 145.5831739961759, "grad_norm": 0.49954599142074585, "learning_rate": 1.4773895487465288e-05, "loss": 0.0062, "step": 228420 }, { "epoch": 145.5895474824729, "grad_norm": 0.16124196350574493, "learning_rate": 1.476998424194141e-05, "loss": 0.0052, "step": 228430 }, { "epoch": 145.59592096876992, "grad_norm": 0.2714800238609314, "learning_rate": 1.4766073424495497e-05, "loss": 0.0067, "step": 228440 }, { "epoch": 145.60229445506693, "grad_norm": 0.11671970039606094, "learning_rate": 1.4762163035175059e-05, "loss": 0.0036, "step": 228450 }, { "epoch": 145.60866794136393, "grad_norm": 0.11825868487358093, "learning_rate": 1.4758253074027612e-05, "loss": 0.0113, "step": 228460 }, { "epoch": 145.61504142766094, "grad_norm": 0.17619585990905762, "learning_rate": 1.4754343541100657e-05, "loss": 0.0042, "step": 228470 }, { "epoch": 145.62141491395795, "grad_norm": 0.2436034232378006, "learning_rate": 1.4750434436441697e-05, "loss": 0.0074, "step": 228480 }, { "epoch": 145.62778840025493, "grad_norm": 0.23132699728012085, "learning_rate": 1.4746525760098256e-05, "loss": 0.0088, "step": 228490 }, { "epoch": 145.63416188655194, "grad_norm": 0.1469157189130783, "learning_rate": 1.4742617512117807e-05, "loss": 0.0041, "step": 228500 }, { "epoch": 145.64053537284894, "grad_norm": 0.25196486711502075, "learning_rate": 1.4738709692547847e-05, "loss": 0.0058, "step": 228510 }, { "epoch": 145.64690885914595, "grad_norm": 0.1670246720314026, "learning_rate": 1.473480230143584e-05, "loss": 0.0088, "step": 228520 }, { "epoch": 145.65328234544296, "grad_norm": 0.4039517045021057, "learning_rate": 1.4730895338829292e-05, "loss": 0.0092, "step": 228530 }, { "epoch": 145.65965583173997, "grad_norm": 0.10780546814203262, "learning_rate": 1.4726988804775672e-05, "loss": 0.0056, "step": 228540 }, { "epoch": 145.66602931803698, "grad_norm": 0.34122276306152344, "learning_rate": 1.4723082699322438e-05, "loss": 0.006, "step": 228550 }, { "epoch": 145.67240280433398, "grad_norm": 0.11637257039546967, "learning_rate": 1.4719177022517044e-05, "loss": 0.0132, "step": 228560 }, { "epoch": 145.67877629063096, "grad_norm": 0.14015494287014008, "learning_rate": 1.4715271774406969e-05, "loss": 0.0135, "step": 228570 }, { "epoch": 145.68514977692797, "grad_norm": 0.42805030941963196, "learning_rate": 1.4711366955039663e-05, "loss": 0.0108, "step": 228580 }, { "epoch": 145.69152326322498, "grad_norm": 0.1416362076997757, "learning_rate": 1.470746256446256e-05, "loss": 0.0053, "step": 228590 }, { "epoch": 145.697896749522, "grad_norm": 0.24157129228115082, "learning_rate": 1.47035586027231e-05, "loss": 0.0099, "step": 228600 }, { "epoch": 145.704270235819, "grad_norm": 0.15091082453727722, "learning_rate": 1.469965506986874e-05, "loss": 0.0103, "step": 228610 }, { "epoch": 145.710643722116, "grad_norm": 0.09477147459983826, "learning_rate": 1.4695751965946897e-05, "loss": 0.0073, "step": 228620 }, { "epoch": 145.717017208413, "grad_norm": 0.44003233313560486, "learning_rate": 1.4691849291005006e-05, "loss": 0.0199, "step": 228630 }, { "epoch": 145.72339069471002, "grad_norm": 0.13582241535186768, "learning_rate": 1.4687947045090461e-05, "loss": 0.0073, "step": 228640 }, { "epoch": 145.729764181007, "grad_norm": 0.11023824661970139, "learning_rate": 1.4684045228250738e-05, "loss": 0.0046, "step": 228650 }, { "epoch": 145.736137667304, "grad_norm": 0.29448580741882324, "learning_rate": 1.4680143840533173e-05, "loss": 0.013, "step": 228660 }, { "epoch": 145.74251115360101, "grad_norm": 0.07959394156932831, "learning_rate": 1.4676242881985225e-05, "loss": 0.0027, "step": 228670 }, { "epoch": 145.74888463989802, "grad_norm": 0.6436347961425781, "learning_rate": 1.4672342352654261e-05, "loss": 0.0183, "step": 228680 }, { "epoch": 145.75525812619503, "grad_norm": 0.16580788791179657, "learning_rate": 1.4668442252587722e-05, "loss": 0.0082, "step": 228690 }, { "epoch": 145.76163161249204, "grad_norm": 0.3812027871608734, "learning_rate": 1.466454258183294e-05, "loss": 0.0056, "step": 228700 }, { "epoch": 145.76800509878905, "grad_norm": 0.10444636642932892, "learning_rate": 1.4660643340437342e-05, "loss": 0.0052, "step": 228710 }, { "epoch": 145.77437858508605, "grad_norm": 0.16096851229667664, "learning_rate": 1.4656744528448291e-05, "loss": 0.0067, "step": 228720 }, { "epoch": 145.78075207138303, "grad_norm": 0.15501444041728973, "learning_rate": 1.4652846145913169e-05, "loss": 0.0046, "step": 228730 }, { "epoch": 145.78712555768004, "grad_norm": 0.25014162063598633, "learning_rate": 1.464894819287932e-05, "loss": 0.0099, "step": 228740 }, { "epoch": 145.79349904397705, "grad_norm": 0.109885074198246, "learning_rate": 1.4645050669394144e-05, "loss": 0.0036, "step": 228750 }, { "epoch": 145.79987253027406, "grad_norm": 0.23974356055259705, "learning_rate": 1.4641153575504979e-05, "loss": 0.0063, "step": 228760 }, { "epoch": 145.80624601657107, "grad_norm": 0.11871036142110825, "learning_rate": 1.463725691125919e-05, "loss": 0.0062, "step": 228770 }, { "epoch": 145.81261950286807, "grad_norm": 0.501052975654602, "learning_rate": 1.4633360676704095e-05, "loss": 0.0094, "step": 228780 }, { "epoch": 145.81899298916508, "grad_norm": 0.19800443947315216, "learning_rate": 1.4629464871887077e-05, "loss": 0.0107, "step": 228790 }, { "epoch": 145.8253664754621, "grad_norm": 0.13129322230815887, "learning_rate": 1.4625569496855457e-05, "loss": 0.0055, "step": 228800 }, { "epoch": 145.83173996175907, "grad_norm": 0.1197645291686058, "learning_rate": 1.4621674551656566e-05, "loss": 0.0035, "step": 228810 }, { "epoch": 145.83811344805608, "grad_norm": 0.21273265779018402, "learning_rate": 1.461778003633772e-05, "loss": 0.0063, "step": 228820 }, { "epoch": 145.84448693435309, "grad_norm": 0.5302979350090027, "learning_rate": 1.4613885950946282e-05, "loss": 0.0172, "step": 228830 }, { "epoch": 145.8508604206501, "grad_norm": 0.1933857500553131, "learning_rate": 1.4609992295529506e-05, "loss": 0.0078, "step": 228840 }, { "epoch": 145.8572339069471, "grad_norm": 0.29662972688674927, "learning_rate": 1.4606099070134754e-05, "loss": 0.0057, "step": 228850 }, { "epoch": 145.8636073932441, "grad_norm": 0.09551730006933212, "learning_rate": 1.4602206274809305e-05, "loss": 0.0052, "step": 228860 }, { "epoch": 145.86998087954112, "grad_norm": 0.22108913958072662, "learning_rate": 1.4598313909600497e-05, "loss": 0.0057, "step": 228870 }, { "epoch": 145.87635436583813, "grad_norm": 0.2201194018125534, "learning_rate": 1.4594421974555566e-05, "loss": 0.0058, "step": 228880 }, { "epoch": 145.8827278521351, "grad_norm": 0.1781148612499237, "learning_rate": 1.4590530469721857e-05, "loss": 0.0037, "step": 228890 }, { "epoch": 145.8891013384321, "grad_norm": 0.16160006821155548, "learning_rate": 1.4586639395146633e-05, "loss": 0.0052, "step": 228900 }, { "epoch": 145.89547482472912, "grad_norm": 0.28174635767936707, "learning_rate": 1.4582748750877174e-05, "loss": 0.0062, "step": 228910 }, { "epoch": 145.90184831102613, "grad_norm": 0.23766647279262543, "learning_rate": 1.4578858536960744e-05, "loss": 0.0072, "step": 228920 }, { "epoch": 145.90822179732314, "grad_norm": 0.4224328100681305, "learning_rate": 1.457496875344464e-05, "loss": 0.0053, "step": 228930 }, { "epoch": 145.91459528362014, "grad_norm": 0.1559581160545349, "learning_rate": 1.4571079400376114e-05, "loss": 0.0031, "step": 228940 }, { "epoch": 145.92096876991715, "grad_norm": 0.45385313034057617, "learning_rate": 1.4567190477802423e-05, "loss": 0.0089, "step": 228950 }, { "epoch": 145.92734225621416, "grad_norm": 0.28406861424446106, "learning_rate": 1.4563301985770811e-05, "loss": 0.0046, "step": 228960 }, { "epoch": 145.93371574251114, "grad_norm": 0.17357593774795532, "learning_rate": 1.4559413924328563e-05, "loss": 0.01, "step": 228970 }, { "epoch": 145.94008922880815, "grad_norm": 0.09440873563289642, "learning_rate": 1.4555526293522863e-05, "loss": 0.0054, "step": 228980 }, { "epoch": 145.94646271510516, "grad_norm": 0.2369723618030548, "learning_rate": 1.455163909340101e-05, "loss": 0.0123, "step": 228990 }, { "epoch": 145.95283620140216, "grad_norm": 0.14450988173484802, "learning_rate": 1.4547752324010184e-05, "loss": 0.0064, "step": 229000 }, { "epoch": 145.95920968769917, "grad_norm": 0.11576925963163376, "learning_rate": 1.4543865985397676e-05, "loss": 0.0041, "step": 229010 }, { "epoch": 145.96558317399618, "grad_norm": 0.26243042945861816, "learning_rate": 1.4539980077610644e-05, "loss": 0.0091, "step": 229020 }, { "epoch": 145.9719566602932, "grad_norm": 0.2236713469028473, "learning_rate": 1.4536094600696344e-05, "loss": 0.0156, "step": 229030 }, { "epoch": 145.9783301465902, "grad_norm": 0.22654026746749878, "learning_rate": 1.4532209554701981e-05, "loss": 0.0042, "step": 229040 }, { "epoch": 145.98470363288718, "grad_norm": 0.172376349568367, "learning_rate": 1.4528324939674759e-05, "loss": 0.0083, "step": 229050 }, { "epoch": 145.99107711918418, "grad_norm": 0.151407852768898, "learning_rate": 1.4524440755661867e-05, "loss": 0.0048, "step": 229060 }, { "epoch": 145.9974506054812, "grad_norm": 0.16473519802093506, "learning_rate": 1.4520557002710523e-05, "loss": 0.0085, "step": 229070 }, { "epoch": 146.0038240917782, "grad_norm": 0.08213087916374207, "learning_rate": 1.4516673680867915e-05, "loss": 0.0095, "step": 229080 }, { "epoch": 146.0101975780752, "grad_norm": 0.09170263260602951, "learning_rate": 1.451279079018122e-05, "loss": 0.0044, "step": 229090 }, { "epoch": 146.01657106437222, "grad_norm": 0.09691045433282852, "learning_rate": 1.4508908330697607e-05, "loss": 0.0038, "step": 229100 }, { "epoch": 146.02294455066922, "grad_norm": 0.47342944145202637, "learning_rate": 1.4505026302464287e-05, "loss": 0.0058, "step": 229110 }, { "epoch": 146.02931803696623, "grad_norm": 0.14520218968391418, "learning_rate": 1.4501144705528407e-05, "loss": 0.0072, "step": 229120 }, { "epoch": 146.03569152326324, "grad_norm": 0.05859525874257088, "learning_rate": 1.4497263539937134e-05, "loss": 0.0046, "step": 229130 }, { "epoch": 146.04206500956022, "grad_norm": 0.192449688911438, "learning_rate": 1.4493382805737616e-05, "loss": 0.0068, "step": 229140 }, { "epoch": 146.04843849585723, "grad_norm": 0.11506391316652298, "learning_rate": 1.4489502502977037e-05, "loss": 0.0078, "step": 229150 }, { "epoch": 146.05481198215423, "grad_norm": 0.14291131496429443, "learning_rate": 1.4485622631702528e-05, "loss": 0.005, "step": 229160 }, { "epoch": 146.06118546845124, "grad_norm": 0.28070947527885437, "learning_rate": 1.4481743191961239e-05, "loss": 0.0053, "step": 229170 }, { "epoch": 146.06755895474825, "grad_norm": 1.1232478618621826, "learning_rate": 1.4477864183800282e-05, "loss": 0.0141, "step": 229180 }, { "epoch": 146.07393244104526, "grad_norm": 0.24636362493038177, "learning_rate": 1.4473985607266832e-05, "loss": 0.0054, "step": 229190 }, { "epoch": 146.08030592734227, "grad_norm": 0.27431124448776245, "learning_rate": 1.4470107462408006e-05, "loss": 0.0089, "step": 229200 }, { "epoch": 146.08667941363927, "grad_norm": 0.14979800581932068, "learning_rate": 1.446622974927091e-05, "loss": 0.0062, "step": 229210 }, { "epoch": 146.09305289993625, "grad_norm": 0.16710713505744934, "learning_rate": 1.4462352467902678e-05, "loss": 0.0043, "step": 229220 }, { "epoch": 146.09942638623326, "grad_norm": 0.12424956262111664, "learning_rate": 1.4458475618350398e-05, "loss": 0.0036, "step": 229230 }, { "epoch": 146.10579987253027, "grad_norm": 0.15998202562332153, "learning_rate": 1.4454599200661217e-05, "loss": 0.012, "step": 229240 }, { "epoch": 146.11217335882728, "grad_norm": 0.20244650542736053, "learning_rate": 1.4450723214882211e-05, "loss": 0.0057, "step": 229250 }, { "epoch": 146.11854684512429, "grad_norm": 0.1104392260313034, "learning_rate": 1.4446847661060481e-05, "loss": 0.0054, "step": 229260 }, { "epoch": 146.1249203314213, "grad_norm": 0.23346073925495148, "learning_rate": 1.444297253924311e-05, "loss": 0.0079, "step": 229270 }, { "epoch": 146.1312938177183, "grad_norm": 0.15168584883213043, "learning_rate": 1.4439097849477202e-05, "loss": 0.0048, "step": 229280 }, { "epoch": 146.1376673040153, "grad_norm": 0.3850065767765045, "learning_rate": 1.4435223591809838e-05, "loss": 0.0115, "step": 229290 }, { "epoch": 146.1440407903123, "grad_norm": 0.11552002280950546, "learning_rate": 1.4431349766288082e-05, "loss": 0.0058, "step": 229300 }, { "epoch": 146.1504142766093, "grad_norm": 0.19613653421401978, "learning_rate": 1.4427476372958993e-05, "loss": 0.0049, "step": 229310 }, { "epoch": 146.1567877629063, "grad_norm": 0.16833806037902832, "learning_rate": 1.4423603411869669e-05, "loss": 0.0052, "step": 229320 }, { "epoch": 146.1631612492033, "grad_norm": 0.14095905423164368, "learning_rate": 1.441973088306715e-05, "loss": 0.0041, "step": 229330 }, { "epoch": 146.16953473550032, "grad_norm": 0.16995474696159363, "learning_rate": 1.4415858786598495e-05, "loss": 0.0067, "step": 229340 }, { "epoch": 146.17590822179733, "grad_norm": 0.19026309251785278, "learning_rate": 1.4411987122510739e-05, "loss": 0.0059, "step": 229350 }, { "epoch": 146.18228170809434, "grad_norm": 0.4292844533920288, "learning_rate": 1.440811589085097e-05, "loss": 0.0148, "step": 229360 }, { "epoch": 146.18865519439134, "grad_norm": 0.5145989060401917, "learning_rate": 1.4404245091666168e-05, "loss": 0.0077, "step": 229370 }, { "epoch": 146.19502868068832, "grad_norm": 0.2548743784427643, "learning_rate": 1.4400374725003413e-05, "loss": 0.0089, "step": 229380 }, { "epoch": 146.20140216698533, "grad_norm": 0.12681807577610016, "learning_rate": 1.4396504790909699e-05, "loss": 0.0079, "step": 229390 }, { "epoch": 146.20777565328234, "grad_norm": 0.376625657081604, "learning_rate": 1.4392635289432094e-05, "loss": 0.0128, "step": 229400 }, { "epoch": 146.21414913957935, "grad_norm": 0.23386399447917938, "learning_rate": 1.4388766220617562e-05, "loss": 0.0052, "step": 229410 }, { "epoch": 146.22052262587636, "grad_norm": 0.19036853313446045, "learning_rate": 1.4384897584513158e-05, "loss": 0.0058, "step": 229420 }, { "epoch": 146.22689611217336, "grad_norm": 0.21427351236343384, "learning_rate": 1.438102938116586e-05, "loss": 0.0051, "step": 229430 }, { "epoch": 146.23326959847037, "grad_norm": 0.2941793203353882, "learning_rate": 1.437716161062272e-05, "loss": 0.0039, "step": 229440 }, { "epoch": 146.23964308476738, "grad_norm": 0.16183796525001526, "learning_rate": 1.437329427293066e-05, "loss": 0.0097, "step": 229450 }, { "epoch": 146.24601657106436, "grad_norm": 0.15116186439990997, "learning_rate": 1.4369427368136735e-05, "loss": 0.0034, "step": 229460 }, { "epoch": 146.25239005736137, "grad_norm": 0.07427627593278885, "learning_rate": 1.4365560896287905e-05, "loss": 0.0117, "step": 229470 }, { "epoch": 146.25876354365838, "grad_norm": 0.16363295912742615, "learning_rate": 1.4361694857431151e-05, "loss": 0.0035, "step": 229480 }, { "epoch": 146.26513702995538, "grad_norm": 0.268706351518631, "learning_rate": 1.4357829251613442e-05, "loss": 0.0047, "step": 229490 }, { "epoch": 146.2715105162524, "grad_norm": 0.4635080397129059, "learning_rate": 1.4353964078881771e-05, "loss": 0.0092, "step": 229500 }, { "epoch": 146.2778840025494, "grad_norm": 0.15024632215499878, "learning_rate": 1.4350099339283096e-05, "loss": 0.0102, "step": 229510 }, { "epoch": 146.2842574888464, "grad_norm": 0.20931799709796906, "learning_rate": 1.4346235032864374e-05, "loss": 0.0088, "step": 229520 }, { "epoch": 146.29063097514342, "grad_norm": 0.23227857053279877, "learning_rate": 1.434237115967254e-05, "loss": 0.0057, "step": 229530 }, { "epoch": 146.2970044614404, "grad_norm": 0.09001592546701431, "learning_rate": 1.4338507719754596e-05, "loss": 0.0058, "step": 229540 }, { "epoch": 146.3033779477374, "grad_norm": 0.47839462757110596, "learning_rate": 1.4334644713157418e-05, "loss": 0.0053, "step": 229550 }, { "epoch": 146.3097514340344, "grad_norm": 0.21979102492332458, "learning_rate": 1.4330782139928e-05, "loss": 0.0063, "step": 229560 }, { "epoch": 146.31612492033142, "grad_norm": 0.7053107023239136, "learning_rate": 1.4326920000113241e-05, "loss": 0.0089, "step": 229570 }, { "epoch": 146.32249840662843, "grad_norm": 0.2154509425163269, "learning_rate": 1.4323058293760117e-05, "loss": 0.0053, "step": 229580 }, { "epoch": 146.32887189292543, "grad_norm": 0.21582239866256714, "learning_rate": 1.4319197020915482e-05, "loss": 0.0059, "step": 229590 }, { "epoch": 146.33524537922244, "grad_norm": 0.4056919813156128, "learning_rate": 1.4315336181626315e-05, "loss": 0.0074, "step": 229600 }, { "epoch": 146.34161886551945, "grad_norm": 0.16535748541355133, "learning_rate": 1.4311475775939498e-05, "loss": 0.007, "step": 229610 }, { "epoch": 146.34799235181643, "grad_norm": 0.14858786761760712, "learning_rate": 1.4307615803901953e-05, "loss": 0.0042, "step": 229620 }, { "epoch": 146.35436583811344, "grad_norm": 0.11891040951013565, "learning_rate": 1.4303756265560553e-05, "loss": 0.0098, "step": 229630 }, { "epoch": 146.36073932441045, "grad_norm": 0.12216638773679733, "learning_rate": 1.4299897160962239e-05, "loss": 0.0068, "step": 229640 }, { "epoch": 146.36711281070745, "grad_norm": 0.13997729122638702, "learning_rate": 1.4296038490153874e-05, "loss": 0.0085, "step": 229650 }, { "epoch": 146.37348629700446, "grad_norm": 0.2532150149345398, "learning_rate": 1.4292180253182358e-05, "loss": 0.0077, "step": 229660 }, { "epoch": 146.37985978330147, "grad_norm": 0.2429455816745758, "learning_rate": 1.4288322450094543e-05, "loss": 0.0047, "step": 229670 }, { "epoch": 146.38623326959848, "grad_norm": 0.3707597255706787, "learning_rate": 1.4284465080937365e-05, "loss": 0.0098, "step": 229680 }, { "epoch": 146.39260675589549, "grad_norm": 0.17996057868003845, "learning_rate": 1.4280608145757623e-05, "loss": 0.005, "step": 229690 }, { "epoch": 146.39898024219247, "grad_norm": 0.11039119958877563, "learning_rate": 1.4276751644602237e-05, "loss": 0.0058, "step": 229700 }, { "epoch": 146.40535372848947, "grad_norm": 0.40957286953926086, "learning_rate": 1.4272895577518026e-05, "loss": 0.0099, "step": 229710 }, { "epoch": 146.41172721478648, "grad_norm": 0.1614452749490738, "learning_rate": 1.4269039944551898e-05, "loss": 0.0075, "step": 229720 }, { "epoch": 146.4181007010835, "grad_norm": 0.17613622546195984, "learning_rate": 1.4265184745750643e-05, "loss": 0.0054, "step": 229730 }, { "epoch": 146.4244741873805, "grad_norm": 0.17340636253356934, "learning_rate": 1.4261329981161142e-05, "loss": 0.0076, "step": 229740 }, { "epoch": 146.4308476736775, "grad_norm": 0.3763548731803894, "learning_rate": 1.4257475650830215e-05, "loss": 0.0069, "step": 229750 }, { "epoch": 146.4372211599745, "grad_norm": 0.12967796623706818, "learning_rate": 1.425362175480473e-05, "loss": 0.0032, "step": 229760 }, { "epoch": 146.44359464627152, "grad_norm": 0.2462339699268341, "learning_rate": 1.4249768293131465e-05, "loss": 0.0063, "step": 229770 }, { "epoch": 146.44996813256853, "grad_norm": 0.23264603316783905, "learning_rate": 1.4245915265857285e-05, "loss": 0.0055, "step": 229780 }, { "epoch": 146.4563416188655, "grad_norm": 0.10804267227649689, "learning_rate": 1.4242062673028988e-05, "loss": 0.012, "step": 229790 }, { "epoch": 146.46271510516252, "grad_norm": 0.19699342548847198, "learning_rate": 1.4238210514693395e-05, "loss": 0.0099, "step": 229800 }, { "epoch": 146.46908859145952, "grad_norm": 0.406103253364563, "learning_rate": 1.423435879089729e-05, "loss": 0.006, "step": 229810 }, { "epoch": 146.47546207775653, "grad_norm": 0.24304145574569702, "learning_rate": 1.4230507501687507e-05, "loss": 0.0063, "step": 229820 }, { "epoch": 146.48183556405354, "grad_norm": 0.19330839812755585, "learning_rate": 1.4226656647110837e-05, "loss": 0.006, "step": 229830 }, { "epoch": 146.48820905035055, "grad_norm": 0.34584107995033264, "learning_rate": 1.4222806227214059e-05, "loss": 0.0078, "step": 229840 }, { "epoch": 146.49458253664756, "grad_norm": 0.16232502460479736, "learning_rate": 1.4218956242043951e-05, "loss": 0.0044, "step": 229850 }, { "epoch": 146.50095602294456, "grad_norm": 0.2700670659542084, "learning_rate": 1.4215106691647323e-05, "loss": 0.0047, "step": 229860 }, { "epoch": 146.50732950924154, "grad_norm": 0.24864323437213898, "learning_rate": 1.4211257576070936e-05, "loss": 0.0056, "step": 229870 }, { "epoch": 146.51370299553855, "grad_norm": 0.23524950444698334, "learning_rate": 1.4207408895361556e-05, "loss": 0.0062, "step": 229880 }, { "epoch": 146.52007648183556, "grad_norm": 0.2737785875797272, "learning_rate": 1.420356064956594e-05, "loss": 0.009, "step": 229890 }, { "epoch": 146.52644996813257, "grad_norm": 0.19469556212425232, "learning_rate": 1.4199712838730878e-05, "loss": 0.0181, "step": 229900 }, { "epoch": 146.53282345442958, "grad_norm": 0.5883970856666565, "learning_rate": 1.4195865462903102e-05, "loss": 0.0082, "step": 229910 }, { "epoch": 146.53919694072658, "grad_norm": 0.15376649796962738, "learning_rate": 1.4192018522129364e-05, "loss": 0.0063, "step": 229920 }, { "epoch": 146.5455704270236, "grad_norm": 0.5990418195724487, "learning_rate": 1.4188172016456408e-05, "loss": 0.0082, "step": 229930 }, { "epoch": 146.5519439133206, "grad_norm": 0.19673782587051392, "learning_rate": 1.4184325945930965e-05, "loss": 0.0074, "step": 229940 }, { "epoch": 146.55831739961758, "grad_norm": 0.14075760543346405, "learning_rate": 1.4180480310599787e-05, "loss": 0.0048, "step": 229950 }, { "epoch": 146.5646908859146, "grad_norm": 0.20699933171272278, "learning_rate": 1.4176635110509595e-05, "loss": 0.0041, "step": 229960 }, { "epoch": 146.5710643722116, "grad_norm": 0.1954485923051834, "learning_rate": 1.4172790345707109e-05, "loss": 0.0068, "step": 229970 }, { "epoch": 146.5774378585086, "grad_norm": 0.5888501405715942, "learning_rate": 1.4168946016239033e-05, "loss": 0.0085, "step": 229980 }, { "epoch": 146.5838113448056, "grad_norm": 0.4887137711048126, "learning_rate": 1.4165102122152102e-05, "loss": 0.0128, "step": 229990 }, { "epoch": 146.59018483110262, "grad_norm": 0.0972149446606636, "learning_rate": 1.4161258663493021e-05, "loss": 0.007, "step": 230000 }, { "epoch": 146.59655831739963, "grad_norm": 0.16588035225868225, "learning_rate": 1.415741564030848e-05, "loss": 0.0056, "step": 230010 }, { "epoch": 146.60293180369663, "grad_norm": 0.15226854383945465, "learning_rate": 1.415357305264517e-05, "loss": 0.0155, "step": 230020 }, { "epoch": 146.6093052899936, "grad_norm": 0.2906543016433716, "learning_rate": 1.4149730900549802e-05, "loss": 0.0091, "step": 230030 }, { "epoch": 146.61567877629062, "grad_norm": 0.3884744346141815, "learning_rate": 1.4145889184069056e-05, "loss": 0.0114, "step": 230040 }, { "epoch": 146.62205226258763, "grad_norm": 0.4443701505661011, "learning_rate": 1.4142047903249605e-05, "loss": 0.0061, "step": 230050 }, { "epoch": 146.62842574888464, "grad_norm": 0.26957422494888306, "learning_rate": 1.4138207058138114e-05, "loss": 0.0108, "step": 230060 }, { "epoch": 146.63479923518165, "grad_norm": 0.11714526265859604, "learning_rate": 1.4134366648781283e-05, "loss": 0.0153, "step": 230070 }, { "epoch": 146.64117272147865, "grad_norm": 0.4254423677921295, "learning_rate": 1.4130526675225759e-05, "loss": 0.0099, "step": 230080 }, { "epoch": 146.64754620777566, "grad_norm": 0.2455018162727356, "learning_rate": 1.4126687137518197e-05, "loss": 0.0165, "step": 230090 }, { "epoch": 146.65391969407267, "grad_norm": 0.2683461010456085, "learning_rate": 1.412284803570525e-05, "loss": 0.0055, "step": 230100 }, { "epoch": 146.66029318036965, "grad_norm": 0.15093660354614258, "learning_rate": 1.4119009369833597e-05, "loss": 0.0048, "step": 230110 }, { "epoch": 146.66666666666666, "grad_norm": 0.16625837981700897, "learning_rate": 1.4115171139949828e-05, "loss": 0.0115, "step": 230120 }, { "epoch": 146.67304015296367, "grad_norm": 0.26042813062667847, "learning_rate": 1.4111333346100625e-05, "loss": 0.0088, "step": 230130 }, { "epoch": 146.67941363926067, "grad_norm": 0.4327771067619324, "learning_rate": 1.4107495988332598e-05, "loss": 0.0075, "step": 230140 }, { "epoch": 146.68578712555768, "grad_norm": 0.1564124971628189, "learning_rate": 1.4103659066692404e-05, "loss": 0.0044, "step": 230150 }, { "epoch": 146.6921606118547, "grad_norm": 0.1392391175031662, "learning_rate": 1.4099822581226617e-05, "loss": 0.0051, "step": 230160 }, { "epoch": 146.6985340981517, "grad_norm": 0.10497743636369705, "learning_rate": 1.4095986531981897e-05, "loss": 0.0045, "step": 230170 }, { "epoch": 146.7049075844487, "grad_norm": 0.18490326404571533, "learning_rate": 1.4092150919004842e-05, "loss": 0.01, "step": 230180 }, { "epoch": 146.71128107074568, "grad_norm": 0.28222087025642395, "learning_rate": 1.4088315742342056e-05, "loss": 0.0087, "step": 230190 }, { "epoch": 146.7176545570427, "grad_norm": 0.22529542446136475, "learning_rate": 1.4084481002040118e-05, "loss": 0.0077, "step": 230200 }, { "epoch": 146.7240280433397, "grad_norm": 0.14904595911502838, "learning_rate": 1.4080646698145666e-05, "loss": 0.0045, "step": 230210 }, { "epoch": 146.7304015296367, "grad_norm": 0.19025710225105286, "learning_rate": 1.407681283070527e-05, "loss": 0.0113, "step": 230220 }, { "epoch": 146.73677501593372, "grad_norm": 0.3802643418312073, "learning_rate": 1.4072979399765517e-05, "loss": 0.0081, "step": 230230 }, { "epoch": 146.74314850223072, "grad_norm": 0.06289494037628174, "learning_rate": 1.4069146405372968e-05, "loss": 0.0073, "step": 230240 }, { "epoch": 146.74952198852773, "grad_norm": 0.1236472800374031, "learning_rate": 1.4065313847574242e-05, "loss": 0.0044, "step": 230250 }, { "epoch": 146.75589547482474, "grad_norm": 0.14459240436553955, "learning_rate": 1.4061481726415854e-05, "loss": 0.0074, "step": 230260 }, { "epoch": 146.76226896112172, "grad_norm": 0.4187425971031189, "learning_rate": 1.4057650041944408e-05, "loss": 0.0102, "step": 230270 }, { "epoch": 146.76864244741873, "grad_norm": 0.10355677455663681, "learning_rate": 1.4053818794206437e-05, "loss": 0.0065, "step": 230280 }, { "epoch": 146.77501593371574, "grad_norm": 0.19129635393619537, "learning_rate": 1.404998798324853e-05, "loss": 0.0072, "step": 230290 }, { "epoch": 146.78138942001274, "grad_norm": 0.2322043627500534, "learning_rate": 1.4046157609117188e-05, "loss": 0.0053, "step": 230300 }, { "epoch": 146.78776290630975, "grad_norm": 0.18258507549762726, "learning_rate": 1.404232767185899e-05, "loss": 0.0095, "step": 230310 }, { "epoch": 146.79413639260676, "grad_norm": 0.12532195448875427, "learning_rate": 1.4038498171520442e-05, "loss": 0.0042, "step": 230320 }, { "epoch": 146.80050987890377, "grad_norm": 0.32474809885025024, "learning_rate": 1.403466910814813e-05, "loss": 0.006, "step": 230330 }, { "epoch": 146.80688336520078, "grad_norm": 0.3265034258365631, "learning_rate": 1.4030840481788516e-05, "loss": 0.0168, "step": 230340 }, { "epoch": 146.81325685149778, "grad_norm": 0.3278328776359558, "learning_rate": 1.4027012292488162e-05, "loss": 0.0046, "step": 230350 }, { "epoch": 146.81963033779476, "grad_norm": 0.7820243239402771, "learning_rate": 1.4023184540293572e-05, "loss": 0.0059, "step": 230360 }, { "epoch": 146.82600382409177, "grad_norm": 0.13403350114822388, "learning_rate": 1.4019357225251256e-05, "loss": 0.0044, "step": 230370 }, { "epoch": 146.83237731038878, "grad_norm": 0.2808988690376282, "learning_rate": 1.4015530347407713e-05, "loss": 0.0105, "step": 230380 }, { "epoch": 146.8387507966858, "grad_norm": 0.3914079964160919, "learning_rate": 1.401170390680946e-05, "loss": 0.0071, "step": 230390 }, { "epoch": 146.8451242829828, "grad_norm": 0.11227937787771225, "learning_rate": 1.4007877903502986e-05, "loss": 0.0101, "step": 230400 }, { "epoch": 146.8514977692798, "grad_norm": 0.3364669978618622, "learning_rate": 1.4004052337534778e-05, "loss": 0.009, "step": 230410 }, { "epoch": 146.8578712555768, "grad_norm": 0.13031305372714996, "learning_rate": 1.4000227208951306e-05, "loss": 0.0051, "step": 230420 }, { "epoch": 146.86424474187382, "grad_norm": 0.15974955260753632, "learning_rate": 1.399640251779909e-05, "loss": 0.009, "step": 230430 }, { "epoch": 146.8706182281708, "grad_norm": 0.6169048547744751, "learning_rate": 1.3992578264124545e-05, "loss": 0.0062, "step": 230440 }, { "epoch": 146.8769917144678, "grad_norm": 0.22873468697071075, "learning_rate": 1.3988754447974189e-05, "loss": 0.0082, "step": 230450 }, { "epoch": 146.8833652007648, "grad_norm": 0.4606171250343323, "learning_rate": 1.3984931069394447e-05, "loss": 0.0056, "step": 230460 }, { "epoch": 146.88973868706182, "grad_norm": 0.19214648008346558, "learning_rate": 1.3981108128431836e-05, "loss": 0.0046, "step": 230470 }, { "epoch": 146.89611217335883, "grad_norm": 0.1325155347585678, "learning_rate": 1.397728562513273e-05, "loss": 0.0061, "step": 230480 }, { "epoch": 146.90248565965584, "grad_norm": 0.2002006471157074, "learning_rate": 1.397346355954363e-05, "loss": 0.0172, "step": 230490 }, { "epoch": 146.90885914595285, "grad_norm": 0.09195851534605026, "learning_rate": 1.396964193171097e-05, "loss": 0.0092, "step": 230500 }, { "epoch": 146.91523263224985, "grad_norm": 0.1459778994321823, "learning_rate": 1.3965820741681169e-05, "loss": 0.0122, "step": 230510 }, { "epoch": 146.92160611854683, "grad_norm": 0.0678580030798912, "learning_rate": 1.3961999989500657e-05, "loss": 0.0077, "step": 230520 }, { "epoch": 146.92797960484384, "grad_norm": 0.22001485526561737, "learning_rate": 1.3958179675215887e-05, "loss": 0.004, "step": 230530 }, { "epoch": 146.93435309114085, "grad_norm": 0.5316879153251648, "learning_rate": 1.3954359798873262e-05, "loss": 0.0094, "step": 230540 }, { "epoch": 146.94072657743786, "grad_norm": 0.2560243010520935, "learning_rate": 1.3950540360519199e-05, "loss": 0.0105, "step": 230550 }, { "epoch": 146.94710006373487, "grad_norm": 0.1990157812833786, "learning_rate": 1.3946721360200094e-05, "loss": 0.0068, "step": 230560 }, { "epoch": 146.95347355003187, "grad_norm": 0.1150251030921936, "learning_rate": 1.3942902797962376e-05, "loss": 0.0066, "step": 230570 }, { "epoch": 146.95984703632888, "grad_norm": 0.18742910027503967, "learning_rate": 1.3939084673852431e-05, "loss": 0.0048, "step": 230580 }, { "epoch": 146.9662205226259, "grad_norm": 0.15594592690467834, "learning_rate": 1.3935266987916657e-05, "loss": 0.0088, "step": 230590 }, { "epoch": 146.97259400892287, "grad_norm": 0.4066248834133148, "learning_rate": 1.3931449740201424e-05, "loss": 0.0104, "step": 230600 }, { "epoch": 146.97896749521988, "grad_norm": 0.4051734209060669, "learning_rate": 1.3927632930753148e-05, "loss": 0.0076, "step": 230610 }, { "epoch": 146.98534098151688, "grad_norm": 0.062035515904426575, "learning_rate": 1.3923816559618191e-05, "loss": 0.0057, "step": 230620 }, { "epoch": 146.9917144678139, "grad_norm": 0.5718725919723511, "learning_rate": 1.3920000626842922e-05, "loss": 0.0075, "step": 230630 }, { "epoch": 146.9980879541109, "grad_norm": 0.13276082277297974, "learning_rate": 1.3916185132473691e-05, "loss": 0.0125, "step": 230640 }, { "epoch": 147.0044614404079, "grad_norm": 0.12106064707040787, "learning_rate": 1.3912370076556902e-05, "loss": 0.006, "step": 230650 }, { "epoch": 147.01083492670492, "grad_norm": 0.1556275486946106, "learning_rate": 1.3908555459138884e-05, "loss": 0.0079, "step": 230660 }, { "epoch": 147.01720841300192, "grad_norm": 0.16469427943229675, "learning_rate": 1.3904741280265999e-05, "loss": 0.007, "step": 230670 }, { "epoch": 147.0235818992989, "grad_norm": 0.14014211297035217, "learning_rate": 1.3900927539984581e-05, "loss": 0.0058, "step": 230680 }, { "epoch": 147.0299553855959, "grad_norm": 0.26843321323394775, "learning_rate": 1.3897114238340964e-05, "loss": 0.004, "step": 230690 }, { "epoch": 147.03632887189292, "grad_norm": 0.13289061188697815, "learning_rate": 1.3893301375381506e-05, "loss": 0.0078, "step": 230700 }, { "epoch": 147.04270235818993, "grad_norm": 0.14863981306552887, "learning_rate": 1.3889488951152535e-05, "loss": 0.0062, "step": 230710 }, { "epoch": 147.04907584448694, "grad_norm": 0.08731581270694733, "learning_rate": 1.3885676965700362e-05, "loss": 0.0051, "step": 230720 }, { "epoch": 147.05544933078394, "grad_norm": 0.20060239732265472, "learning_rate": 1.38818654190713e-05, "loss": 0.0058, "step": 230730 }, { "epoch": 147.06182281708095, "grad_norm": 0.20573590695858002, "learning_rate": 1.3878054311311684e-05, "loss": 0.0122, "step": 230740 }, { "epoch": 147.06819630337796, "grad_norm": 0.1303136646747589, "learning_rate": 1.3874243642467815e-05, "loss": 0.0039, "step": 230750 }, { "epoch": 147.07456978967494, "grad_norm": 0.13148541748523712, "learning_rate": 1.3870433412585998e-05, "loss": 0.005, "step": 230760 }, { "epoch": 147.08094327597195, "grad_norm": 0.5670254230499268, "learning_rate": 1.3866623621712505e-05, "loss": 0.0078, "step": 230770 }, { "epoch": 147.08731676226896, "grad_norm": 0.16587479412555695, "learning_rate": 1.3862814269893671e-05, "loss": 0.0065, "step": 230780 }, { "epoch": 147.09369024856596, "grad_norm": 0.18128807842731476, "learning_rate": 1.3859005357175763e-05, "loss": 0.0093, "step": 230790 }, { "epoch": 147.10006373486297, "grad_norm": 0.2686079740524292, "learning_rate": 1.385519688360506e-05, "loss": 0.0108, "step": 230800 }, { "epoch": 147.10643722115998, "grad_norm": 0.1845853626728058, "learning_rate": 1.3851388849227826e-05, "loss": 0.0157, "step": 230810 }, { "epoch": 147.112810707457, "grad_norm": 0.1257193386554718, "learning_rate": 1.3847581254090381e-05, "loss": 0.0041, "step": 230820 }, { "epoch": 147.119184193754, "grad_norm": 0.16785618662834167, "learning_rate": 1.384377409823892e-05, "loss": 0.0053, "step": 230830 }, { "epoch": 147.12555768005097, "grad_norm": 0.15733566880226135, "learning_rate": 1.3839967381719764e-05, "loss": 0.0102, "step": 230840 }, { "epoch": 147.13193116634798, "grad_norm": 0.1553560197353363, "learning_rate": 1.3836161104579126e-05, "loss": 0.0083, "step": 230850 }, { "epoch": 147.138304652645, "grad_norm": 0.09813115745782852, "learning_rate": 1.3832355266863305e-05, "loss": 0.0111, "step": 230860 }, { "epoch": 147.144678138942, "grad_norm": 0.12267216295003891, "learning_rate": 1.3828549868618485e-05, "loss": 0.0073, "step": 230870 }, { "epoch": 147.151051625239, "grad_norm": 0.06350033730268478, "learning_rate": 1.3824744909890953e-05, "loss": 0.0044, "step": 230880 }, { "epoch": 147.157425111536, "grad_norm": 0.13699950277805328, "learning_rate": 1.3820940390726928e-05, "loss": 0.0114, "step": 230890 }, { "epoch": 147.16379859783302, "grad_norm": 0.19963954389095306, "learning_rate": 1.3817136311172629e-05, "loss": 0.0088, "step": 230900 }, { "epoch": 147.17017208413003, "grad_norm": 0.3264200687408447, "learning_rate": 1.3813332671274276e-05, "loss": 0.0047, "step": 230910 }, { "epoch": 147.176545570427, "grad_norm": 0.15485531091690063, "learning_rate": 1.3809529471078109e-05, "loss": 0.0101, "step": 230920 }, { "epoch": 147.18291905672402, "grad_norm": 0.350297212600708, "learning_rate": 1.380572671063033e-05, "loss": 0.0055, "step": 230930 }, { "epoch": 147.18929254302103, "grad_norm": 0.260128378868103, "learning_rate": 1.3801924389977145e-05, "loss": 0.0036, "step": 230940 }, { "epoch": 147.19566602931803, "grad_norm": 0.3601965010166168, "learning_rate": 1.3798122509164735e-05, "loss": 0.0058, "step": 230950 }, { "epoch": 147.20203951561504, "grad_norm": 0.2827568054199219, "learning_rate": 1.3794321068239336e-05, "loss": 0.0096, "step": 230960 }, { "epoch": 147.20841300191205, "grad_norm": 0.22016441822052002, "learning_rate": 1.379052006724712e-05, "loss": 0.0114, "step": 230970 }, { "epoch": 147.21478648820906, "grad_norm": 0.2667921483516693, "learning_rate": 1.3786719506234274e-05, "loss": 0.0071, "step": 230980 }, { "epoch": 147.22115997450607, "grad_norm": 0.18096786737442017, "learning_rate": 1.3782919385246961e-05, "loss": 0.0085, "step": 230990 }, { "epoch": 147.22753346080307, "grad_norm": 0.34929051995277405, "learning_rate": 1.3779119704331401e-05, "loss": 0.0065, "step": 231000 }, { "epoch": 147.23390694710005, "grad_norm": 0.10103239864110947, "learning_rate": 1.3775320463533703e-05, "loss": 0.0072, "step": 231010 }, { "epoch": 147.24028043339706, "grad_norm": 0.35737091302871704, "learning_rate": 1.3771521662900083e-05, "loss": 0.025, "step": 231020 }, { "epoch": 147.24665391969407, "grad_norm": 0.17846372723579407, "learning_rate": 1.3767723302476665e-05, "loss": 0.0075, "step": 231030 }, { "epoch": 147.25302740599108, "grad_norm": 0.20076654851436615, "learning_rate": 1.3763925382309644e-05, "loss": 0.0038, "step": 231040 }, { "epoch": 147.25940089228808, "grad_norm": 0.20052886009216309, "learning_rate": 1.3760127902445114e-05, "loss": 0.0052, "step": 231050 }, { "epoch": 147.2657743785851, "grad_norm": 0.10064736008644104, "learning_rate": 1.3756330862929261e-05, "loss": 0.0035, "step": 231060 }, { "epoch": 147.2721478648821, "grad_norm": 0.7242658138275146, "learning_rate": 1.3752534263808204e-05, "loss": 0.0086, "step": 231070 }, { "epoch": 147.2785213511791, "grad_norm": 0.48999282717704773, "learning_rate": 1.3748738105128085e-05, "loss": 0.0158, "step": 231080 }, { "epoch": 147.2848948374761, "grad_norm": 0.24213802814483643, "learning_rate": 1.3744942386935005e-05, "loss": 0.013, "step": 231090 }, { "epoch": 147.2912683237731, "grad_norm": 0.16314677894115448, "learning_rate": 1.3741147109275126e-05, "loss": 0.0099, "step": 231100 }, { "epoch": 147.2976418100701, "grad_norm": 0.11622485518455505, "learning_rate": 1.3737352272194538e-05, "loss": 0.0045, "step": 231110 }, { "epoch": 147.3040152963671, "grad_norm": 0.10064952820539474, "learning_rate": 1.373355787573936e-05, "loss": 0.0033, "step": 231120 }, { "epoch": 147.31038878266412, "grad_norm": 0.10846184939146042, "learning_rate": 1.372976391995568e-05, "loss": 0.0046, "step": 231130 }, { "epoch": 147.31676226896113, "grad_norm": 0.43739619851112366, "learning_rate": 1.3725970404889643e-05, "loss": 0.006, "step": 231140 }, { "epoch": 147.32313575525814, "grad_norm": 0.07743804156780243, "learning_rate": 1.3722177330587282e-05, "loss": 0.0066, "step": 231150 }, { "epoch": 147.32950924155514, "grad_norm": 0.2034185230731964, "learning_rate": 1.3718384697094738e-05, "loss": 0.0061, "step": 231160 }, { "epoch": 147.33588272785212, "grad_norm": 0.19579452276229858, "learning_rate": 1.3714592504458057e-05, "loss": 0.0054, "step": 231170 }, { "epoch": 147.34225621414913, "grad_norm": 0.14814110100269318, "learning_rate": 1.3710800752723368e-05, "loss": 0.0051, "step": 231180 }, { "epoch": 147.34862970044614, "grad_norm": 0.4906712472438812, "learning_rate": 1.370700944193668e-05, "loss": 0.0069, "step": 231190 }, { "epoch": 147.35500318674315, "grad_norm": 0.25556260347366333, "learning_rate": 1.370321857214411e-05, "loss": 0.0084, "step": 231200 }, { "epoch": 147.36137667304016, "grad_norm": 0.4768398702144623, "learning_rate": 1.369942814339169e-05, "loss": 0.0051, "step": 231210 }, { "epoch": 147.36775015933716, "grad_norm": 0.343989759683609, "learning_rate": 1.369563815572552e-05, "loss": 0.0044, "step": 231220 }, { "epoch": 147.37412364563417, "grad_norm": 0.5031096339225769, "learning_rate": 1.3691848609191592e-05, "loss": 0.0111, "step": 231230 }, { "epoch": 147.38049713193118, "grad_norm": 0.15655165910720825, "learning_rate": 1.3688059503836003e-05, "loss": 0.0056, "step": 231240 }, { "epoch": 147.38687061822816, "grad_norm": 0.4207097589969635, "learning_rate": 1.368427083970477e-05, "loss": 0.0124, "step": 231250 }, { "epoch": 147.39324410452517, "grad_norm": 0.4240633249282837, "learning_rate": 1.3680482616843942e-05, "loss": 0.0096, "step": 231260 }, { "epoch": 147.39961759082217, "grad_norm": 0.15851199626922607, "learning_rate": 1.367669483529952e-05, "loss": 0.0138, "step": 231270 }, { "epoch": 147.40599107711918, "grad_norm": 0.4340016543865204, "learning_rate": 1.3672907495117565e-05, "loss": 0.005, "step": 231280 }, { "epoch": 147.4123645634162, "grad_norm": 1.1822664737701416, "learning_rate": 1.3669120596344088e-05, "loss": 0.0086, "step": 231290 }, { "epoch": 147.4187380497132, "grad_norm": 0.26524659991264343, "learning_rate": 1.3665334139025094e-05, "loss": 0.0072, "step": 231300 }, { "epoch": 147.4251115360102, "grad_norm": 0.23315855860710144, "learning_rate": 1.366154812320658e-05, "loss": 0.0212, "step": 231310 }, { "epoch": 147.4314850223072, "grad_norm": 0.44348856806755066, "learning_rate": 1.3657762548934583e-05, "loss": 0.0106, "step": 231320 }, { "epoch": 147.4378585086042, "grad_norm": 0.21090468764305115, "learning_rate": 1.3653977416255081e-05, "loss": 0.0042, "step": 231330 }, { "epoch": 147.4442319949012, "grad_norm": 0.4129892587661743, "learning_rate": 1.3650192725214073e-05, "loss": 0.0101, "step": 231340 }, { "epoch": 147.4506054811982, "grad_norm": 0.29510289430618286, "learning_rate": 1.3646408475857525e-05, "loss": 0.0069, "step": 231350 }, { "epoch": 147.45697896749522, "grad_norm": 0.17984406650066376, "learning_rate": 1.3642624668231457e-05, "loss": 0.0046, "step": 231360 }, { "epoch": 147.46335245379223, "grad_norm": 0.17009764909744263, "learning_rate": 1.3638841302381822e-05, "loss": 0.0077, "step": 231370 }, { "epoch": 147.46972594008923, "grad_norm": 0.11692831665277481, "learning_rate": 1.3635058378354599e-05, "loss": 0.0068, "step": 231380 }, { "epoch": 147.47609942638624, "grad_norm": 0.3617134988307953, "learning_rate": 1.363127589619575e-05, "loss": 0.0081, "step": 231390 }, { "epoch": 147.48247291268325, "grad_norm": 0.12677878141403198, "learning_rate": 1.3627493855951217e-05, "loss": 0.0101, "step": 231400 }, { "epoch": 147.48884639898023, "grad_norm": 0.11112502962350845, "learning_rate": 1.3623712257666998e-05, "loss": 0.0041, "step": 231410 }, { "epoch": 147.49521988527724, "grad_norm": 0.15369148552417755, "learning_rate": 1.3619931101389017e-05, "loss": 0.0045, "step": 231420 }, { "epoch": 147.50159337157424, "grad_norm": 0.203826442360878, "learning_rate": 1.3616150387163217e-05, "loss": 0.0072, "step": 231430 }, { "epoch": 147.50796685787125, "grad_norm": 0.23429155349731445, "learning_rate": 1.3612370115035533e-05, "loss": 0.0069, "step": 231440 }, { "epoch": 147.51434034416826, "grad_norm": 0.21577145159244537, "learning_rate": 1.360859028505192e-05, "loss": 0.0125, "step": 231450 }, { "epoch": 147.52071383046527, "grad_norm": 0.2572345435619354, "learning_rate": 1.360481089725829e-05, "loss": 0.0043, "step": 231460 }, { "epoch": 147.52708731676228, "grad_norm": 0.45092153549194336, "learning_rate": 1.3601031951700577e-05, "loss": 0.0155, "step": 231470 }, { "epoch": 147.53346080305928, "grad_norm": 0.1839090883731842, "learning_rate": 1.3597253448424674e-05, "loss": 0.0076, "step": 231480 }, { "epoch": 147.53983428935626, "grad_norm": 0.16502909362316132, "learning_rate": 1.3593475387476529e-05, "loss": 0.0051, "step": 231490 }, { "epoch": 147.54620777565327, "grad_norm": 0.47647660970687866, "learning_rate": 1.3589697768902027e-05, "loss": 0.0098, "step": 231500 }, { "epoch": 147.55258126195028, "grad_norm": 0.09755682945251465, "learning_rate": 1.3585920592747076e-05, "loss": 0.0059, "step": 231510 }, { "epoch": 147.5589547482473, "grad_norm": 0.12089536339044571, "learning_rate": 1.3582143859057555e-05, "loss": 0.0063, "step": 231520 }, { "epoch": 147.5653282345443, "grad_norm": 0.23056963086128235, "learning_rate": 1.357836756787939e-05, "loss": 0.0069, "step": 231530 }, { "epoch": 147.5717017208413, "grad_norm": 0.10073991119861603, "learning_rate": 1.3574591719258445e-05, "loss": 0.0055, "step": 231540 }, { "epoch": 147.5780752071383, "grad_norm": 0.08600801974534988, "learning_rate": 1.3570816313240598e-05, "loss": 0.01, "step": 231550 }, { "epoch": 147.58444869343532, "grad_norm": 0.29723936319351196, "learning_rate": 1.3567041349871712e-05, "loss": 0.0043, "step": 231560 }, { "epoch": 147.5908221797323, "grad_norm": 0.11986328661441803, "learning_rate": 1.3563266829197708e-05, "loss": 0.0078, "step": 231570 }, { "epoch": 147.5971956660293, "grad_norm": 0.43935543298721313, "learning_rate": 1.355949275126438e-05, "loss": 0.0139, "step": 231580 }, { "epoch": 147.60356915232632, "grad_norm": 0.15078893303871155, "learning_rate": 1.3555719116117638e-05, "loss": 0.0075, "step": 231590 }, { "epoch": 147.60994263862332, "grad_norm": 0.13202804327011108, "learning_rate": 1.35519459238033e-05, "loss": 0.0062, "step": 231600 }, { "epoch": 147.61631612492033, "grad_norm": 0.14231868088245392, "learning_rate": 1.3548173174367263e-05, "loss": 0.0074, "step": 231610 }, { "epoch": 147.62268961121734, "grad_norm": 0.21514803171157837, "learning_rate": 1.3544400867855306e-05, "loss": 0.005, "step": 231620 }, { "epoch": 147.62906309751435, "grad_norm": 0.4427289366722107, "learning_rate": 1.3540629004313315e-05, "loss": 0.0095, "step": 231630 }, { "epoch": 147.63543658381136, "grad_norm": 0.12804611027240753, "learning_rate": 1.3536857583787104e-05, "loss": 0.0068, "step": 231640 }, { "epoch": 147.64181007010836, "grad_norm": 0.14631308615207672, "learning_rate": 1.3533086606322504e-05, "loss": 0.0104, "step": 231650 }, { "epoch": 147.64818355640534, "grad_norm": 0.4031652808189392, "learning_rate": 1.3529316071965315e-05, "loss": 0.0077, "step": 231660 }, { "epoch": 147.65455704270235, "grad_norm": 0.13858748972415924, "learning_rate": 1.3525545980761383e-05, "loss": 0.0097, "step": 231670 }, { "epoch": 147.66093052899936, "grad_norm": 0.17067070305347443, "learning_rate": 1.3521776332756508e-05, "loss": 0.0118, "step": 231680 }, { "epoch": 147.66730401529637, "grad_norm": 0.23781611025333405, "learning_rate": 1.3518007127996491e-05, "loss": 0.0061, "step": 231690 }, { "epoch": 147.67367750159337, "grad_norm": 0.26112326979637146, "learning_rate": 1.3514238366527115e-05, "loss": 0.0054, "step": 231700 }, { "epoch": 147.68005098789038, "grad_norm": 0.28284817934036255, "learning_rate": 1.3510470048394225e-05, "loss": 0.013, "step": 231710 }, { "epoch": 147.6864244741874, "grad_norm": 0.4130135774612427, "learning_rate": 1.3506702173643543e-05, "loss": 0.0072, "step": 231720 }, { "epoch": 147.6927979604844, "grad_norm": 0.08164238929748535, "learning_rate": 1.35029347423209e-05, "loss": 0.0201, "step": 231730 }, { "epoch": 147.69917144678138, "grad_norm": 0.12314968556165695, "learning_rate": 1.3499167754472042e-05, "loss": 0.0067, "step": 231740 }, { "epoch": 147.7055449330784, "grad_norm": 0.1401493102312088, "learning_rate": 1.3495401210142788e-05, "loss": 0.0119, "step": 231750 }, { "epoch": 147.7119184193754, "grad_norm": 0.19445329904556274, "learning_rate": 1.3491635109378847e-05, "loss": 0.0083, "step": 231760 }, { "epoch": 147.7182919056724, "grad_norm": 0.17888762056827545, "learning_rate": 1.3487869452226025e-05, "loss": 0.0057, "step": 231770 }, { "epoch": 147.7246653919694, "grad_norm": 0.12920278310775757, "learning_rate": 1.3484104238730066e-05, "loss": 0.0074, "step": 231780 }, { "epoch": 147.73103887826642, "grad_norm": 0.11052410304546356, "learning_rate": 1.3480339468936708e-05, "loss": 0.0052, "step": 231790 }, { "epoch": 147.73741236456343, "grad_norm": 0.18832068145275116, "learning_rate": 1.3476575142891701e-05, "loss": 0.0045, "step": 231800 }, { "epoch": 147.74378585086043, "grad_norm": 0.16027134656906128, "learning_rate": 1.34728112606408e-05, "loss": 0.0123, "step": 231810 }, { "epoch": 147.7501593371574, "grad_norm": 0.13259199261665344, "learning_rate": 1.3469047822229725e-05, "loss": 0.0084, "step": 231820 }, { "epoch": 147.75653282345442, "grad_norm": 0.0860593169927597, "learning_rate": 1.3465284827704217e-05, "loss": 0.0089, "step": 231830 }, { "epoch": 147.76290630975143, "grad_norm": 0.25153884291648865, "learning_rate": 1.3461522277109973e-05, "loss": 0.0058, "step": 231840 }, { "epoch": 147.76927979604844, "grad_norm": 0.5641430020332336, "learning_rate": 1.345776017049275e-05, "loss": 0.0074, "step": 231850 }, { "epoch": 147.77565328234544, "grad_norm": 0.2326754629611969, "learning_rate": 1.3453998507898241e-05, "loss": 0.0072, "step": 231860 }, { "epoch": 147.78202676864245, "grad_norm": 0.3933635950088501, "learning_rate": 1.3450237289372154e-05, "loss": 0.0101, "step": 231870 }, { "epoch": 147.78840025493946, "grad_norm": 0.5358847379684448, "learning_rate": 1.3446476514960177e-05, "loss": 0.0155, "step": 231880 }, { "epoch": 147.79477374123647, "grad_norm": 0.466997891664505, "learning_rate": 1.3442716184708055e-05, "loss": 0.0083, "step": 231890 }, { "epoch": 147.80114722753345, "grad_norm": 0.2867925763130188, "learning_rate": 1.3438956298661414e-05, "loss": 0.0061, "step": 231900 }, { "epoch": 147.80752071383046, "grad_norm": 0.1589479297399521, "learning_rate": 1.3435196856865994e-05, "loss": 0.0164, "step": 231910 }, { "epoch": 147.81389420012746, "grad_norm": 0.30841875076293945, "learning_rate": 1.3431437859367435e-05, "loss": 0.007, "step": 231920 }, { "epoch": 147.82026768642447, "grad_norm": 0.31694185733795166, "learning_rate": 1.3427679306211466e-05, "loss": 0.0069, "step": 231930 }, { "epoch": 147.82664117272148, "grad_norm": 0.16745805740356445, "learning_rate": 1.342392119744369e-05, "loss": 0.0049, "step": 231940 }, { "epoch": 147.8330146590185, "grad_norm": 0.16027291119098663, "learning_rate": 1.3420163533109826e-05, "loss": 0.0035, "step": 231950 }, { "epoch": 147.8393881453155, "grad_norm": 0.1188591718673706, "learning_rate": 1.3416406313255509e-05, "loss": 0.0061, "step": 231960 }, { "epoch": 147.8457616316125, "grad_norm": 0.11228562146425247, "learning_rate": 1.3412649537926391e-05, "loss": 0.0066, "step": 231970 }, { "epoch": 147.85213511790948, "grad_norm": 0.14445006847381592, "learning_rate": 1.3408893207168117e-05, "loss": 0.0089, "step": 231980 }, { "epoch": 147.8585086042065, "grad_norm": 0.5604311227798462, "learning_rate": 1.340513732102635e-05, "loss": 0.0065, "step": 231990 }, { "epoch": 147.8648820905035, "grad_norm": 0.29856303334236145, "learning_rate": 1.3401381879546715e-05, "loss": 0.0069, "step": 232000 }, { "epoch": 147.8712555768005, "grad_norm": 0.4663613736629486, "learning_rate": 1.3397626882774844e-05, "loss": 0.0076, "step": 232010 }, { "epoch": 147.87762906309752, "grad_norm": 0.2202560007572174, "learning_rate": 1.3393872330756351e-05, "loss": 0.0068, "step": 232020 }, { "epoch": 147.88400254939452, "grad_norm": 0.10432467609643936, "learning_rate": 1.339011822353689e-05, "loss": 0.0034, "step": 232030 }, { "epoch": 147.89037603569153, "grad_norm": 0.42463141679763794, "learning_rate": 1.3386364561162058e-05, "loss": 0.0042, "step": 232040 }, { "epoch": 147.89674952198854, "grad_norm": 0.2179185301065445, "learning_rate": 1.3382611343677458e-05, "loss": 0.0121, "step": 232050 }, { "epoch": 147.90312300828552, "grad_norm": 0.20728446543216705, "learning_rate": 1.3378858571128694e-05, "loss": 0.004, "step": 232060 }, { "epoch": 147.90949649458253, "grad_norm": 0.15412725508213043, "learning_rate": 1.3375106243561386e-05, "loss": 0.0054, "step": 232070 }, { "epoch": 147.91586998087953, "grad_norm": 0.4244357943534851, "learning_rate": 1.3371354361021122e-05, "loss": 0.0125, "step": 232080 }, { "epoch": 147.92224346717654, "grad_norm": 0.21974675357341766, "learning_rate": 1.336760292355348e-05, "loss": 0.0084, "step": 232090 }, { "epoch": 147.92861695347355, "grad_norm": 0.1968841552734375, "learning_rate": 1.3363851931204051e-05, "loss": 0.0165, "step": 232100 }, { "epoch": 147.93499043977056, "grad_norm": 0.3265763521194458, "learning_rate": 1.3360101384018398e-05, "loss": 0.0035, "step": 232110 }, { "epoch": 147.94136392606757, "grad_norm": 0.15903055667877197, "learning_rate": 1.3356351282042123e-05, "loss": 0.0051, "step": 232120 }, { "epoch": 147.94773741236457, "grad_norm": 0.18278174102306366, "learning_rate": 1.3352601625320772e-05, "loss": 0.0051, "step": 232130 }, { "epoch": 147.95411089866155, "grad_norm": 0.16593188047409058, "learning_rate": 1.3348852413899914e-05, "loss": 0.0071, "step": 232140 }, { "epoch": 147.96048438495856, "grad_norm": 0.28658682107925415, "learning_rate": 1.3345103647825091e-05, "loss": 0.0077, "step": 232150 }, { "epoch": 147.96685787125557, "grad_norm": 0.21765802800655365, "learning_rate": 1.334135532714188e-05, "loss": 0.0068, "step": 232160 }, { "epoch": 147.97323135755258, "grad_norm": 0.23874154686927795, "learning_rate": 1.3337607451895818e-05, "loss": 0.0054, "step": 232170 }, { "epoch": 147.9796048438496, "grad_norm": 0.3335578739643097, "learning_rate": 1.3333860022132432e-05, "loss": 0.0106, "step": 232180 }, { "epoch": 147.9859783301466, "grad_norm": 0.6209019422531128, "learning_rate": 1.3330113037897257e-05, "loss": 0.0077, "step": 232190 }, { "epoch": 147.9923518164436, "grad_norm": 0.3286232650279999, "learning_rate": 1.3326366499235842e-05, "loss": 0.0061, "step": 232200 }, { "epoch": 147.9987253027406, "grad_norm": 0.19875946640968323, "learning_rate": 1.3322620406193698e-05, "loss": 0.0045, "step": 232210 }, { "epoch": 148.00509878903762, "grad_norm": 0.3401239514350891, "learning_rate": 1.3318874758816352e-05, "loss": 0.0064, "step": 232220 }, { "epoch": 148.0114722753346, "grad_norm": 0.2540838122367859, "learning_rate": 1.3315129557149292e-05, "loss": 0.0061, "step": 232230 }, { "epoch": 148.0178457616316, "grad_norm": 0.13751426339149475, "learning_rate": 1.3311384801238059e-05, "loss": 0.0048, "step": 232240 }, { "epoch": 148.0242192479286, "grad_norm": 0.2280208170413971, "learning_rate": 1.3307640491128136e-05, "loss": 0.0066, "step": 232250 }, { "epoch": 148.03059273422562, "grad_norm": 0.11532506346702576, "learning_rate": 1.330389662686503e-05, "loss": 0.0041, "step": 232260 }, { "epoch": 148.03696622052263, "grad_norm": 0.5142546892166138, "learning_rate": 1.3300153208494209e-05, "loss": 0.0069, "step": 232270 }, { "epoch": 148.04333970681964, "grad_norm": 0.2282782942056656, "learning_rate": 1.3296410236061207e-05, "loss": 0.009, "step": 232280 }, { "epoch": 148.04971319311665, "grad_norm": 0.18216551840305328, "learning_rate": 1.329266770961145e-05, "loss": 0.009, "step": 232290 }, { "epoch": 148.05608667941365, "grad_norm": 0.13542209565639496, "learning_rate": 1.3288925629190447e-05, "loss": 0.0054, "step": 232300 }, { "epoch": 148.06246016571063, "grad_norm": 0.15006788074970245, "learning_rate": 1.3285183994843647e-05, "loss": 0.0037, "step": 232310 }, { "epoch": 148.06883365200764, "grad_norm": 0.08103980123996735, "learning_rate": 1.3281442806616551e-05, "loss": 0.0084, "step": 232320 }, { "epoch": 148.07520713830465, "grad_norm": 0.0726575180888176, "learning_rate": 1.3277702064554565e-05, "loss": 0.0066, "step": 232330 }, { "epoch": 148.08158062460166, "grad_norm": 0.23460239171981812, "learning_rate": 1.3273961768703191e-05, "loss": 0.0063, "step": 232340 }, { "epoch": 148.08795411089866, "grad_norm": 0.3625332713127136, "learning_rate": 1.3270221919107851e-05, "loss": 0.0059, "step": 232350 }, { "epoch": 148.09432759719567, "grad_norm": 0.10674367100000381, "learning_rate": 1.3266482515813995e-05, "loss": 0.0047, "step": 232360 }, { "epoch": 148.10070108349268, "grad_norm": 0.170624241232872, "learning_rate": 1.3262743558867041e-05, "loss": 0.0186, "step": 232370 }, { "epoch": 148.1070745697897, "grad_norm": 0.19475765526294708, "learning_rate": 1.3259005048312455e-05, "loss": 0.0106, "step": 232380 }, { "epoch": 148.11344805608667, "grad_norm": 0.5190044045448303, "learning_rate": 1.3255266984195647e-05, "loss": 0.0064, "step": 232390 }, { "epoch": 148.11982154238368, "grad_norm": 0.13973817229270935, "learning_rate": 1.3251529366562044e-05, "loss": 0.0099, "step": 232400 }, { "epoch": 148.12619502868068, "grad_norm": 0.11559075117111206, "learning_rate": 1.3247792195457032e-05, "loss": 0.0053, "step": 232410 }, { "epoch": 148.1325685149777, "grad_norm": 0.17755021154880524, "learning_rate": 1.3244055470926064e-05, "loss": 0.014, "step": 232420 }, { "epoch": 148.1389420012747, "grad_norm": 0.5896440148353577, "learning_rate": 1.3240319193014522e-05, "loss": 0.0099, "step": 232430 }, { "epoch": 148.1453154875717, "grad_norm": 0.14003683626651764, "learning_rate": 1.3236583361767812e-05, "loss": 0.0105, "step": 232440 }, { "epoch": 148.15168897386872, "grad_norm": 0.15270467102527618, "learning_rate": 1.3232847977231305e-05, "loss": 0.0069, "step": 232450 }, { "epoch": 148.15806246016572, "grad_norm": 0.46057790517807007, "learning_rate": 1.3229113039450447e-05, "loss": 0.0065, "step": 232460 }, { "epoch": 148.1644359464627, "grad_norm": 0.3423055410385132, "learning_rate": 1.3225378548470546e-05, "loss": 0.0054, "step": 232470 }, { "epoch": 148.1708094327597, "grad_norm": 0.45091527700424194, "learning_rate": 1.3221644504337033e-05, "loss": 0.0102, "step": 232480 }, { "epoch": 148.17718291905672, "grad_norm": 0.16299434006214142, "learning_rate": 1.3217910907095249e-05, "loss": 0.0073, "step": 232490 }, { "epoch": 148.18355640535373, "grad_norm": 0.5428645014762878, "learning_rate": 1.3214177756790608e-05, "loss": 0.0137, "step": 232500 }, { "epoch": 148.18992989165073, "grad_norm": 0.12421546131372452, "learning_rate": 1.3210445053468407e-05, "loss": 0.0061, "step": 232510 }, { "epoch": 148.19630337794774, "grad_norm": 0.1904636174440384, "learning_rate": 1.3206712797174048e-05, "loss": 0.0114, "step": 232520 }, { "epoch": 148.20267686424475, "grad_norm": 0.18979433178901672, "learning_rate": 1.3202980987952867e-05, "loss": 0.0063, "step": 232530 }, { "epoch": 148.20905035054176, "grad_norm": 0.12971900403499603, "learning_rate": 1.3199249625850213e-05, "loss": 0.0118, "step": 232540 }, { "epoch": 148.21542383683874, "grad_norm": 0.35209280252456665, "learning_rate": 1.3195518710911404e-05, "loss": 0.0102, "step": 232550 }, { "epoch": 148.22179732313575, "grad_norm": 0.15650880336761475, "learning_rate": 1.3191788243181807e-05, "loss": 0.0115, "step": 232560 }, { "epoch": 148.22817080943275, "grad_norm": 0.18687325716018677, "learning_rate": 1.3188058222706734e-05, "loss": 0.0069, "step": 232570 }, { "epoch": 148.23454429572976, "grad_norm": 0.40891870856285095, "learning_rate": 1.3184328649531513e-05, "loss": 0.0056, "step": 232580 }, { "epoch": 148.24091778202677, "grad_norm": 0.3174339234828949, "learning_rate": 1.3180599523701443e-05, "loss": 0.0062, "step": 232590 }, { "epoch": 148.24729126832378, "grad_norm": 0.06915030628442764, "learning_rate": 1.317687084526188e-05, "loss": 0.0062, "step": 232600 }, { "epoch": 148.2536647546208, "grad_norm": 0.3812072277069092, "learning_rate": 1.3173142614258072e-05, "loss": 0.0058, "step": 232610 }, { "epoch": 148.2600382409178, "grad_norm": 0.13262461125850677, "learning_rate": 1.3169414830735372e-05, "loss": 0.0062, "step": 232620 }, { "epoch": 148.26641172721477, "grad_norm": 0.20724186301231384, "learning_rate": 1.3165687494739032e-05, "loss": 0.0049, "step": 232630 }, { "epoch": 148.27278521351178, "grad_norm": 0.06304415315389633, "learning_rate": 1.3161960606314406e-05, "loss": 0.0036, "step": 232640 }, { "epoch": 148.2791586998088, "grad_norm": 0.12902003526687622, "learning_rate": 1.31582341655067e-05, "loss": 0.0081, "step": 232650 }, { "epoch": 148.2855321861058, "grad_norm": 0.07150037586688995, "learning_rate": 1.3154508172361257e-05, "loss": 0.0023, "step": 232660 }, { "epoch": 148.2919056724028, "grad_norm": 0.1340705156326294, "learning_rate": 1.3150782626923325e-05, "loss": 0.0088, "step": 232670 }, { "epoch": 148.2982791586998, "grad_norm": 0.1056206226348877, "learning_rate": 1.3147057529238172e-05, "loss": 0.0137, "step": 232680 }, { "epoch": 148.30465264499682, "grad_norm": 0.11471828073263168, "learning_rate": 1.314333287935105e-05, "loss": 0.0077, "step": 232690 }, { "epoch": 148.31102613129383, "grad_norm": 0.21185848116874695, "learning_rate": 1.3139608677307252e-05, "loss": 0.0057, "step": 232700 }, { "epoch": 148.3173996175908, "grad_norm": 0.12353729456663132, "learning_rate": 1.3135884923152009e-05, "loss": 0.0078, "step": 232710 }, { "epoch": 148.32377310388782, "grad_norm": 0.30304497480392456, "learning_rate": 1.3132161616930566e-05, "loss": 0.009, "step": 232720 }, { "epoch": 148.33014659018482, "grad_norm": 0.23670955002307892, "learning_rate": 1.3128438758688155e-05, "loss": 0.003, "step": 232730 }, { "epoch": 148.33652007648183, "grad_norm": 0.5640246272087097, "learning_rate": 1.3124716348470046e-05, "loss": 0.005, "step": 232740 }, { "epoch": 148.34289356277884, "grad_norm": 0.11031801253557205, "learning_rate": 1.312099438632144e-05, "loss": 0.0048, "step": 232750 }, { "epoch": 148.34926704907585, "grad_norm": 0.10414353758096695, "learning_rate": 1.3117272872287577e-05, "loss": 0.0048, "step": 232760 }, { "epoch": 148.35564053537286, "grad_norm": 0.1322323977947235, "learning_rate": 1.311355180641366e-05, "loss": 0.007, "step": 232770 }, { "epoch": 148.36201402166986, "grad_norm": 0.12791241705417633, "learning_rate": 1.3109831188744925e-05, "loss": 0.0065, "step": 232780 }, { "epoch": 148.36838750796684, "grad_norm": 0.1775730848312378, "learning_rate": 1.3106111019326573e-05, "loss": 0.0049, "step": 232790 }, { "epoch": 148.37476099426385, "grad_norm": 0.14202441275119781, "learning_rate": 1.3102391298203803e-05, "loss": 0.0082, "step": 232800 }, { "epoch": 148.38113448056086, "grad_norm": 0.1498994082212448, "learning_rate": 1.3098672025421804e-05, "loss": 0.0084, "step": 232810 }, { "epoch": 148.38750796685787, "grad_norm": 0.1790229231119156, "learning_rate": 1.3094953201025795e-05, "loss": 0.0124, "step": 232820 }, { "epoch": 148.39388145315488, "grad_norm": 0.22138449549674988, "learning_rate": 1.3091234825060949e-05, "loss": 0.0079, "step": 232830 }, { "epoch": 148.40025493945188, "grad_norm": 0.24517221748828888, "learning_rate": 1.3087516897572444e-05, "loss": 0.0037, "step": 232840 }, { "epoch": 148.4066284257489, "grad_norm": 0.15465041995048523, "learning_rate": 1.3083799418605464e-05, "loss": 0.0064, "step": 232850 }, { "epoch": 148.4130019120459, "grad_norm": 0.3331727683544159, "learning_rate": 1.3080082388205162e-05, "loss": 0.0041, "step": 232860 }, { "epoch": 148.4193753983429, "grad_norm": 0.12685626745224, "learning_rate": 1.307636580641673e-05, "loss": 0.0062, "step": 232870 }, { "epoch": 148.4257488846399, "grad_norm": 0.3257819414138794, "learning_rate": 1.3072649673285314e-05, "loss": 0.0048, "step": 232880 }, { "epoch": 148.4321223709369, "grad_norm": 0.11376507580280304, "learning_rate": 1.306893398885607e-05, "loss": 0.0054, "step": 232890 }, { "epoch": 148.4384958572339, "grad_norm": 0.14673596620559692, "learning_rate": 1.306521875317413e-05, "loss": 0.0063, "step": 232900 }, { "epoch": 148.4448693435309, "grad_norm": 0.10653230547904968, "learning_rate": 1.3061503966284671e-05, "loss": 0.0106, "step": 232910 }, { "epoch": 148.45124282982792, "grad_norm": 0.22247402369976044, "learning_rate": 1.3057789628232814e-05, "loss": 0.0078, "step": 232920 }, { "epoch": 148.45761631612493, "grad_norm": 0.25298336148262024, "learning_rate": 1.3054075739063698e-05, "loss": 0.0054, "step": 232930 }, { "epoch": 148.46398980242193, "grad_norm": 0.1964055895805359, "learning_rate": 1.3050362298822421e-05, "loss": 0.0091, "step": 232940 }, { "epoch": 148.47036328871894, "grad_norm": 0.14084932208061218, "learning_rate": 1.3046649307554148e-05, "loss": 0.0029, "step": 232950 }, { "epoch": 148.47673677501592, "grad_norm": 0.2994310259819031, "learning_rate": 1.3042936765303975e-05, "loss": 0.007, "step": 232960 }, { "epoch": 148.48311026131293, "grad_norm": 0.33933958411216736, "learning_rate": 1.3039224672117013e-05, "loss": 0.0172, "step": 232970 }, { "epoch": 148.48948374760994, "grad_norm": 0.128510981798172, "learning_rate": 1.3035513028038354e-05, "loss": 0.0089, "step": 232980 }, { "epoch": 148.49585723390695, "grad_norm": 0.12933802604675293, "learning_rate": 1.3031801833113139e-05, "loss": 0.0076, "step": 232990 }, { "epoch": 148.50223072020395, "grad_norm": 0.13759920001029968, "learning_rate": 1.3028091087386412e-05, "loss": 0.0062, "step": 233000 }, { "epoch": 148.50860420650096, "grad_norm": 0.49324068427085876, "learning_rate": 1.3024380790903295e-05, "loss": 0.0081, "step": 233010 }, { "epoch": 148.51497769279797, "grad_norm": 0.21729281544685364, "learning_rate": 1.3020670943708846e-05, "loss": 0.0061, "step": 233020 }, { "epoch": 148.52135117909498, "grad_norm": 0.14127928018569946, "learning_rate": 1.3016961545848195e-05, "loss": 0.0083, "step": 233030 }, { "epoch": 148.52772466539196, "grad_norm": 0.27154937386512756, "learning_rate": 1.3013252597366344e-05, "loss": 0.0072, "step": 233040 }, { "epoch": 148.53409815168897, "grad_norm": 0.12285394966602325, "learning_rate": 1.3009544098308413e-05, "loss": 0.0069, "step": 233050 }, { "epoch": 148.54047163798597, "grad_norm": 0.2829659581184387, "learning_rate": 1.3005836048719428e-05, "loss": 0.005, "step": 233060 }, { "epoch": 148.54684512428298, "grad_norm": 0.28437939286231995, "learning_rate": 1.3002128448644496e-05, "loss": 0.0084, "step": 233070 }, { "epoch": 148.55321861058, "grad_norm": 0.2734539210796356, "learning_rate": 1.2998421298128604e-05, "loss": 0.0086, "step": 233080 }, { "epoch": 148.559592096877, "grad_norm": 0.1262175738811493, "learning_rate": 1.2994714597216845e-05, "loss": 0.0041, "step": 233090 }, { "epoch": 148.565965583174, "grad_norm": 0.16775158047676086, "learning_rate": 1.2991008345954243e-05, "loss": 0.0037, "step": 233100 }, { "epoch": 148.572339069471, "grad_norm": 0.20074114203453064, "learning_rate": 1.2987302544385826e-05, "loss": 0.0078, "step": 233110 }, { "epoch": 148.578712555768, "grad_norm": 0.5076010823249817, "learning_rate": 1.2983597192556617e-05, "loss": 0.0099, "step": 233120 }, { "epoch": 148.585086042065, "grad_norm": 0.1800888478755951, "learning_rate": 1.297989229051167e-05, "loss": 0.0072, "step": 233130 }, { "epoch": 148.591459528362, "grad_norm": 0.17654410004615784, "learning_rate": 1.2976187838295983e-05, "loss": 0.0098, "step": 233140 }, { "epoch": 148.59783301465902, "grad_norm": 0.23805341124534607, "learning_rate": 1.2972483835954568e-05, "loss": 0.0123, "step": 233150 }, { "epoch": 148.60420650095602, "grad_norm": 0.11270268261432648, "learning_rate": 1.2968780283532423e-05, "loss": 0.005, "step": 233160 }, { "epoch": 148.61057998725303, "grad_norm": 0.13631890714168549, "learning_rate": 1.2965077181074586e-05, "loss": 0.0173, "step": 233170 }, { "epoch": 148.61695347355004, "grad_norm": 0.09446865320205688, "learning_rate": 1.2961374528626003e-05, "loss": 0.0082, "step": 233180 }, { "epoch": 148.62332695984705, "grad_norm": 0.1419980376958847, "learning_rate": 1.2957672326231702e-05, "loss": 0.009, "step": 233190 }, { "epoch": 148.62970044614403, "grad_norm": 0.12472062557935715, "learning_rate": 1.2953970573936636e-05, "loss": 0.0141, "step": 233200 }, { "epoch": 148.63607393244104, "grad_norm": 0.335350900888443, "learning_rate": 1.2950269271785843e-05, "loss": 0.016, "step": 233210 }, { "epoch": 148.64244741873804, "grad_norm": 0.22347284853458405, "learning_rate": 1.2946568419824223e-05, "loss": 0.0068, "step": 233220 }, { "epoch": 148.64882090503505, "grad_norm": 0.2511058449745178, "learning_rate": 1.2942868018096798e-05, "loss": 0.0064, "step": 233230 }, { "epoch": 148.65519439133206, "grad_norm": 0.15735067427158356, "learning_rate": 1.2939168066648515e-05, "loss": 0.0086, "step": 233240 }, { "epoch": 148.66156787762907, "grad_norm": 0.20694595575332642, "learning_rate": 1.2935468565524333e-05, "loss": 0.0086, "step": 233250 }, { "epoch": 148.66794136392608, "grad_norm": 0.24637439846992493, "learning_rate": 1.2931769514769182e-05, "loss": 0.0066, "step": 233260 }, { "epoch": 148.67431485022308, "grad_norm": 0.37594836950302124, "learning_rate": 1.2928070914428052e-05, "loss": 0.0061, "step": 233270 }, { "epoch": 148.68068833652006, "grad_norm": 0.09672008454799652, "learning_rate": 1.2924372764545856e-05, "loss": 0.0076, "step": 233280 }, { "epoch": 148.68706182281707, "grad_norm": 0.23345917463302612, "learning_rate": 1.2920675065167537e-05, "loss": 0.0106, "step": 233290 }, { "epoch": 148.69343530911408, "grad_norm": 0.3615751266479492, "learning_rate": 1.291697781633801e-05, "loss": 0.0072, "step": 233300 }, { "epoch": 148.6998087954111, "grad_norm": 0.09464187175035477, "learning_rate": 1.2913281018102236e-05, "loss": 0.0075, "step": 233310 }, { "epoch": 148.7061822817081, "grad_norm": 0.6298773288726807, "learning_rate": 1.2909584670505103e-05, "loss": 0.0128, "step": 233320 }, { "epoch": 148.7125557680051, "grad_norm": 0.19759316742420197, "learning_rate": 1.2905888773591545e-05, "loss": 0.0054, "step": 233330 }, { "epoch": 148.7189292543021, "grad_norm": 0.21844632923603058, "learning_rate": 1.2902193327406443e-05, "loss": 0.021, "step": 233340 }, { "epoch": 148.72530274059912, "grad_norm": 0.13136586546897888, "learning_rate": 1.2898498331994746e-05, "loss": 0.008, "step": 233350 }, { "epoch": 148.7316762268961, "grad_norm": 0.33112281560897827, "learning_rate": 1.2894803787401295e-05, "loss": 0.0101, "step": 233360 }, { "epoch": 148.7380497131931, "grad_norm": 0.10179667919874191, "learning_rate": 1.2891109693671022e-05, "loss": 0.0056, "step": 233370 }, { "epoch": 148.74442319949011, "grad_norm": 0.2741200923919678, "learning_rate": 1.2887416050848794e-05, "loss": 0.0047, "step": 233380 }, { "epoch": 148.75079668578712, "grad_norm": 0.2829958200454712, "learning_rate": 1.288372285897953e-05, "loss": 0.0079, "step": 233390 }, { "epoch": 148.75717017208413, "grad_norm": 0.23591956496238708, "learning_rate": 1.2880030118108044e-05, "loss": 0.0072, "step": 233400 }, { "epoch": 148.76354365838114, "grad_norm": 0.11868536472320557, "learning_rate": 1.2876337828279255e-05, "loss": 0.0074, "step": 233410 }, { "epoch": 148.76991714467815, "grad_norm": 0.5470077991485596, "learning_rate": 1.2872645989538012e-05, "loss": 0.0109, "step": 233420 }, { "epoch": 148.77629063097515, "grad_norm": 0.17777538299560547, "learning_rate": 1.2868954601929172e-05, "loss": 0.0118, "step": 233430 }, { "epoch": 148.78266411727213, "grad_norm": 0.3967754542827606, "learning_rate": 1.2865263665497584e-05, "loss": 0.0102, "step": 233440 }, { "epoch": 148.78903760356914, "grad_norm": 0.25147226452827454, "learning_rate": 1.2861573180288112e-05, "loss": 0.0055, "step": 233450 }, { "epoch": 148.79541108986615, "grad_norm": 0.14450353384017944, "learning_rate": 1.285788314634559e-05, "loss": 0.0086, "step": 233460 }, { "epoch": 148.80178457616316, "grad_norm": 0.18309634923934937, "learning_rate": 1.2854193563714855e-05, "loss": 0.0068, "step": 233470 }, { "epoch": 148.80815806246017, "grad_norm": 0.18427278101444244, "learning_rate": 1.2850504432440729e-05, "loss": 0.0077, "step": 233480 }, { "epoch": 148.81453154875717, "grad_norm": 0.22113947570323944, "learning_rate": 1.2846815752568065e-05, "loss": 0.0078, "step": 233490 }, { "epoch": 148.82090503505418, "grad_norm": 0.23150360584259033, "learning_rate": 1.2843127524141663e-05, "loss": 0.0039, "step": 233500 }, { "epoch": 148.8272785213512, "grad_norm": 0.16027235984802246, "learning_rate": 1.2839439747206345e-05, "loss": 0.0083, "step": 233510 }, { "epoch": 148.8336520076482, "grad_norm": 0.32608291506767273, "learning_rate": 1.2835752421806907e-05, "loss": 0.0062, "step": 233520 }, { "epoch": 148.84002549394518, "grad_norm": 0.1054784506559372, "learning_rate": 1.2832065547988182e-05, "loss": 0.0044, "step": 233530 }, { "epoch": 148.84639898024218, "grad_norm": 0.1759815514087677, "learning_rate": 1.2828379125794948e-05, "loss": 0.013, "step": 233540 }, { "epoch": 148.8527724665392, "grad_norm": 0.39203643798828125, "learning_rate": 1.2824693155272005e-05, "loss": 0.0106, "step": 233550 }, { "epoch": 148.8591459528362, "grad_norm": 0.11766216158866882, "learning_rate": 1.2821007636464138e-05, "loss": 0.0045, "step": 233560 }, { "epoch": 148.8655194391332, "grad_norm": 0.2620556652545929, "learning_rate": 1.2817322569416119e-05, "loss": 0.0055, "step": 233570 }, { "epoch": 148.87189292543022, "grad_norm": 0.2831522822380066, "learning_rate": 1.2813637954172752e-05, "loss": 0.0057, "step": 233580 }, { "epoch": 148.87826641172722, "grad_norm": 0.4666374623775482, "learning_rate": 1.280995379077879e-05, "loss": 0.0057, "step": 233590 }, { "epoch": 148.88463989802423, "grad_norm": 0.19181300699710846, "learning_rate": 1.2806270079279009e-05, "loss": 0.0073, "step": 233600 }, { "epoch": 148.8910133843212, "grad_norm": 0.22087514400482178, "learning_rate": 1.2802586819718144e-05, "loss": 0.0054, "step": 233610 }, { "epoch": 148.89738687061822, "grad_norm": 0.08910460770130157, "learning_rate": 1.2798904012140984e-05, "loss": 0.0063, "step": 233620 }, { "epoch": 148.90376035691523, "grad_norm": 0.08655358850955963, "learning_rate": 1.2795221656592266e-05, "loss": 0.0059, "step": 233630 }, { "epoch": 148.91013384321224, "grad_norm": 0.21635131537914276, "learning_rate": 1.2791539753116727e-05, "loss": 0.0059, "step": 233640 }, { "epoch": 148.91650732950924, "grad_norm": 0.2711957097053528, "learning_rate": 1.2787858301759104e-05, "loss": 0.0039, "step": 233650 }, { "epoch": 148.92288081580625, "grad_norm": 0.13388928771018982, "learning_rate": 1.2784177302564144e-05, "loss": 0.0075, "step": 233660 }, { "epoch": 148.92925430210326, "grad_norm": 0.16010725498199463, "learning_rate": 1.278049675557657e-05, "loss": 0.0044, "step": 233670 }, { "epoch": 148.93562778840027, "grad_norm": 0.14281818270683289, "learning_rate": 1.27768166608411e-05, "loss": 0.0059, "step": 233680 }, { "epoch": 148.94200127469725, "grad_norm": 0.1352490931749344, "learning_rate": 1.2773137018402442e-05, "loss": 0.0116, "step": 233690 }, { "epoch": 148.94837476099426, "grad_norm": 0.21411560475826263, "learning_rate": 1.2769457828305326e-05, "loss": 0.0035, "step": 233700 }, { "epoch": 148.95474824729126, "grad_norm": 0.13072001934051514, "learning_rate": 1.2765779090594454e-05, "loss": 0.0129, "step": 233710 }, { "epoch": 148.96112173358827, "grad_norm": 0.07905611395835876, "learning_rate": 1.2762100805314519e-05, "loss": 0.0051, "step": 233720 }, { "epoch": 148.96749521988528, "grad_norm": 0.17458899319171906, "learning_rate": 1.2758422972510204e-05, "loss": 0.0198, "step": 233730 }, { "epoch": 148.9738687061823, "grad_norm": 0.21020235121250153, "learning_rate": 1.2754745592226241e-05, "loss": 0.0042, "step": 233740 }, { "epoch": 148.9802421924793, "grad_norm": 0.18846341967582703, "learning_rate": 1.2751068664507254e-05, "loss": 0.0048, "step": 233750 }, { "epoch": 148.9866156787763, "grad_norm": 0.09811379760503769, "learning_rate": 1.2747392189397967e-05, "loss": 0.0085, "step": 233760 }, { "epoch": 148.99298916507328, "grad_norm": 0.1490439623594284, "learning_rate": 1.2743716166943021e-05, "loss": 0.0037, "step": 233770 }, { "epoch": 148.9993626513703, "grad_norm": 0.14308154582977295, "learning_rate": 1.2740040597187125e-05, "loss": 0.0074, "step": 233780 }, { "epoch": 149.0057361376673, "grad_norm": 0.13533087074756622, "learning_rate": 1.2736365480174895e-05, "loss": 0.0054, "step": 233790 }, { "epoch": 149.0121096239643, "grad_norm": 0.24483296275138855, "learning_rate": 1.2732690815951014e-05, "loss": 0.0074, "step": 233800 }, { "epoch": 149.01848311026131, "grad_norm": 0.387656033039093, "learning_rate": 1.2729016604560129e-05, "loss": 0.0056, "step": 233810 }, { "epoch": 149.02485659655832, "grad_norm": 0.15287582576274872, "learning_rate": 1.2725342846046884e-05, "loss": 0.0047, "step": 233820 }, { "epoch": 149.03123008285533, "grad_norm": 0.1243734285235405, "learning_rate": 1.2721669540455893e-05, "loss": 0.0027, "step": 233830 }, { "epoch": 149.03760356915234, "grad_norm": 0.14451630413532257, "learning_rate": 1.2717996687831841e-05, "loss": 0.0042, "step": 233840 }, { "epoch": 149.04397705544932, "grad_norm": 0.09370864182710648, "learning_rate": 1.271432428821932e-05, "loss": 0.0037, "step": 233850 }, { "epoch": 149.05035054174633, "grad_norm": 0.1270083487033844, "learning_rate": 1.2710652341662966e-05, "loss": 0.0077, "step": 233860 }, { "epoch": 149.05672402804333, "grad_norm": 0.15299318730831146, "learning_rate": 1.2706980848207378e-05, "loss": 0.0076, "step": 233870 }, { "epoch": 149.06309751434034, "grad_norm": 0.25964534282684326, "learning_rate": 1.2703309807897212e-05, "loss": 0.0091, "step": 233880 }, { "epoch": 149.06947100063735, "grad_norm": 0.4554256796836853, "learning_rate": 1.2699639220777015e-05, "loss": 0.0075, "step": 233890 }, { "epoch": 149.07584448693436, "grad_norm": 0.09146008640527725, "learning_rate": 1.2695969086891435e-05, "loss": 0.0059, "step": 233900 }, { "epoch": 149.08221797323137, "grad_norm": 0.2386639416217804, "learning_rate": 1.2692299406285036e-05, "loss": 0.0051, "step": 233910 }, { "epoch": 149.08859145952837, "grad_norm": 0.13722658157348633, "learning_rate": 1.268863017900246e-05, "loss": 0.0095, "step": 233920 }, { "epoch": 149.09496494582535, "grad_norm": 0.11712624877691269, "learning_rate": 1.2684961405088214e-05, "loss": 0.0033, "step": 233930 }, { "epoch": 149.10133843212236, "grad_norm": 0.4157274067401886, "learning_rate": 1.2681293084586937e-05, "loss": 0.0033, "step": 233940 }, { "epoch": 149.10771191841937, "grad_norm": 0.3428627550601959, "learning_rate": 1.267762521754317e-05, "loss": 0.0067, "step": 233950 }, { "epoch": 149.11408540471638, "grad_norm": 0.20579814910888672, "learning_rate": 1.2673957804001529e-05, "loss": 0.0166, "step": 233960 }, { "epoch": 149.12045889101339, "grad_norm": 0.1594051867723465, "learning_rate": 1.2670290844006504e-05, "loss": 0.0146, "step": 233970 }, { "epoch": 149.1268323773104, "grad_norm": 0.10205373167991638, "learning_rate": 1.2666624337602712e-05, "loss": 0.0049, "step": 233980 }, { "epoch": 149.1332058636074, "grad_norm": 0.190903902053833, "learning_rate": 1.266295828483468e-05, "loss": 0.0046, "step": 233990 }, { "epoch": 149.1395793499044, "grad_norm": 0.20151735842227936, "learning_rate": 1.265929268574696e-05, "loss": 0.0037, "step": 234000 }, { "epoch": 149.1459528362014, "grad_norm": 0.22080014646053314, "learning_rate": 1.265562754038407e-05, "loss": 0.0102, "step": 234010 }, { "epoch": 149.1523263224984, "grad_norm": 0.25175172090530396, "learning_rate": 1.2651962848790583e-05, "loss": 0.0052, "step": 234020 }, { "epoch": 149.1586998087954, "grad_norm": 0.1411723643541336, "learning_rate": 1.2648298611011006e-05, "loss": 0.01, "step": 234030 }, { "epoch": 149.1650732950924, "grad_norm": 0.11328615993261337, "learning_rate": 1.2644634827089869e-05, "loss": 0.0028, "step": 234040 }, { "epoch": 149.17144678138942, "grad_norm": 0.21432656049728394, "learning_rate": 1.2640971497071668e-05, "loss": 0.0044, "step": 234050 }, { "epoch": 149.17782026768643, "grad_norm": 0.3784433603286743, "learning_rate": 1.2637308621000976e-05, "loss": 0.0074, "step": 234060 }, { "epoch": 149.18419375398344, "grad_norm": 0.1752888262271881, "learning_rate": 1.2633646198922223e-05, "loss": 0.0127, "step": 234070 }, { "epoch": 149.19056724028044, "grad_norm": 0.15955489873886108, "learning_rate": 1.2629984230879966e-05, "loss": 0.0064, "step": 234080 }, { "epoch": 149.19694072657742, "grad_norm": 0.18475528061389923, "learning_rate": 1.2626322716918671e-05, "loss": 0.0112, "step": 234090 }, { "epoch": 149.20331421287443, "grad_norm": 0.09320053458213806, "learning_rate": 1.2622661657082863e-05, "loss": 0.0087, "step": 234100 }, { "epoch": 149.20968769917144, "grad_norm": 0.22905322909355164, "learning_rate": 1.2619001051416984e-05, "loss": 0.0043, "step": 234110 }, { "epoch": 149.21606118546845, "grad_norm": 0.16726231575012207, "learning_rate": 1.261534089996555e-05, "loss": 0.0077, "step": 234120 }, { "epoch": 149.22243467176546, "grad_norm": 0.10144130885601044, "learning_rate": 1.2611681202773018e-05, "loss": 0.0067, "step": 234130 }, { "epoch": 149.22880815806246, "grad_norm": 0.16374845802783966, "learning_rate": 1.260802195988386e-05, "loss": 0.0044, "step": 234140 }, { "epoch": 149.23518164435947, "grad_norm": 0.14380650222301483, "learning_rate": 1.2604363171342525e-05, "loss": 0.0073, "step": 234150 }, { "epoch": 149.24155513065648, "grad_norm": 0.09475073963403702, "learning_rate": 1.26007048371935e-05, "loss": 0.006, "step": 234160 }, { "epoch": 149.2479286169535, "grad_norm": 0.3592500686645508, "learning_rate": 1.2597046957481223e-05, "loss": 0.0085, "step": 234170 }, { "epoch": 149.25430210325047, "grad_norm": 0.08079465478658676, "learning_rate": 1.2593389532250139e-05, "loss": 0.0083, "step": 234180 }, { "epoch": 149.26067558954747, "grad_norm": 0.18330977857112885, "learning_rate": 1.2589732561544677e-05, "loss": 0.0084, "step": 234190 }, { "epoch": 149.26704907584448, "grad_norm": 0.3142426609992981, "learning_rate": 1.2586076045409295e-05, "loss": 0.0098, "step": 234200 }, { "epoch": 149.2734225621415, "grad_norm": 0.0897267609834671, "learning_rate": 1.2582419983888422e-05, "loss": 0.0048, "step": 234210 }, { "epoch": 149.2797960484385, "grad_norm": 0.1016467809677124, "learning_rate": 1.2578764377026464e-05, "loss": 0.0154, "step": 234220 }, { "epoch": 149.2861695347355, "grad_norm": 0.1672435998916626, "learning_rate": 1.2575109224867843e-05, "loss": 0.0046, "step": 234230 }, { "epoch": 149.29254302103251, "grad_norm": 0.2460492104291916, "learning_rate": 1.2571454527456993e-05, "loss": 0.0048, "step": 234240 }, { "epoch": 149.29891650732952, "grad_norm": 0.30469322204589844, "learning_rate": 1.2567800284838305e-05, "loss": 0.0048, "step": 234250 }, { "epoch": 149.3052899936265, "grad_norm": 0.17153117060661316, "learning_rate": 1.2564146497056189e-05, "loss": 0.0093, "step": 234260 }, { "epoch": 149.3116634799235, "grad_norm": 0.6332923769950867, "learning_rate": 1.2560493164155024e-05, "loss": 0.0087, "step": 234270 }, { "epoch": 149.31803696622052, "grad_norm": 0.09553353488445282, "learning_rate": 1.2556840286179233e-05, "loss": 0.0047, "step": 234280 }, { "epoch": 149.32441045251753, "grad_norm": 0.09133178740739822, "learning_rate": 1.2553187863173177e-05, "loss": 0.0078, "step": 234290 }, { "epoch": 149.33078393881453, "grad_norm": 0.31707969307899475, "learning_rate": 1.2549535895181253e-05, "loss": 0.0188, "step": 234300 }, { "epoch": 149.33715742511154, "grad_norm": 0.1187213733792305, "learning_rate": 1.254588438224782e-05, "loss": 0.0061, "step": 234310 }, { "epoch": 149.34353091140855, "grad_norm": 0.2839025557041168, "learning_rate": 1.2542233324417241e-05, "loss": 0.0069, "step": 234320 }, { "epoch": 149.34990439770556, "grad_norm": 0.167969211935997, "learning_rate": 1.2538582721733911e-05, "loss": 0.0065, "step": 234330 }, { "epoch": 149.35627788400254, "grad_norm": 0.118154376745224, "learning_rate": 1.253493257424217e-05, "loss": 0.0032, "step": 234340 }, { "epoch": 149.36265137029955, "grad_norm": 0.25641822814941406, "learning_rate": 1.2531282881986367e-05, "loss": 0.0036, "step": 234350 }, { "epoch": 149.36902485659655, "grad_norm": 0.5990939736366272, "learning_rate": 1.2527633645010845e-05, "loss": 0.0141, "step": 234360 }, { "epoch": 149.37539834289356, "grad_norm": 0.13706769049167633, "learning_rate": 1.2523984863359962e-05, "loss": 0.0084, "step": 234370 }, { "epoch": 149.38177182919057, "grad_norm": 0.5148871541023254, "learning_rate": 1.2520336537078048e-05, "loss": 0.0131, "step": 234380 }, { "epoch": 149.38814531548758, "grad_norm": 0.42778876423835754, "learning_rate": 1.251668866620943e-05, "loss": 0.008, "step": 234390 }, { "epoch": 149.39451880178459, "grad_norm": 0.10314369946718216, "learning_rate": 1.2513041250798423e-05, "loss": 0.0039, "step": 234400 }, { "epoch": 149.4008922880816, "grad_norm": 0.21825076639652252, "learning_rate": 1.2509394290889376e-05, "loss": 0.0078, "step": 234410 }, { "epoch": 149.40726577437857, "grad_norm": 0.19863875210285187, "learning_rate": 1.2505747786526583e-05, "loss": 0.0051, "step": 234420 }, { "epoch": 149.41363926067558, "grad_norm": 0.13831709325313568, "learning_rate": 1.2502101737754351e-05, "loss": 0.0061, "step": 234430 }, { "epoch": 149.4200127469726, "grad_norm": 0.20040389895439148, "learning_rate": 1.2498456144616976e-05, "loss": 0.0039, "step": 234440 }, { "epoch": 149.4263862332696, "grad_norm": 0.14563485980033875, "learning_rate": 1.2494811007158797e-05, "loss": 0.0103, "step": 234450 }, { "epoch": 149.4327597195666, "grad_norm": 0.16825658082962036, "learning_rate": 1.2491166325424041e-05, "loss": 0.0159, "step": 234460 }, { "epoch": 149.4391332058636, "grad_norm": 0.3055366277694702, "learning_rate": 1.2487522099457044e-05, "loss": 0.006, "step": 234470 }, { "epoch": 149.44550669216062, "grad_norm": 0.3046199083328247, "learning_rate": 1.2483878329302062e-05, "loss": 0.013, "step": 234480 }, { "epoch": 149.45188017845763, "grad_norm": 0.14802086353302002, "learning_rate": 1.2480235015003405e-05, "loss": 0.0055, "step": 234490 }, { "epoch": 149.4582536647546, "grad_norm": 0.1879676729440689, "learning_rate": 1.2476592156605282e-05, "loss": 0.0055, "step": 234500 }, { "epoch": 149.46462715105162, "grad_norm": 0.11979097872972488, "learning_rate": 1.2472949754152013e-05, "loss": 0.0038, "step": 234510 }, { "epoch": 149.47100063734862, "grad_norm": 0.09678881615400314, "learning_rate": 1.2469307807687825e-05, "loss": 0.0029, "step": 234520 }, { "epoch": 149.47737412364563, "grad_norm": 0.09595659375190735, "learning_rate": 1.2465666317257002e-05, "loss": 0.0072, "step": 234530 }, { "epoch": 149.48374760994264, "grad_norm": 0.259596049785614, "learning_rate": 1.246202528290375e-05, "loss": 0.0129, "step": 234540 }, { "epoch": 149.49012109623965, "grad_norm": 0.19345170259475708, "learning_rate": 1.2458384704672344e-05, "loss": 0.0095, "step": 234550 }, { "epoch": 149.49649458253666, "grad_norm": 0.20368358492851257, "learning_rate": 1.2454744582607008e-05, "loss": 0.008, "step": 234560 }, { "epoch": 149.50286806883366, "grad_norm": 0.36685600876808167, "learning_rate": 1.2451104916751976e-05, "loss": 0.0078, "step": 234570 }, { "epoch": 149.50924155513064, "grad_norm": 0.2722887694835663, "learning_rate": 1.2447465707151456e-05, "loss": 0.0104, "step": 234580 }, { "epoch": 149.51561504142765, "grad_norm": 0.3607155978679657, "learning_rate": 1.2443826953849696e-05, "loss": 0.005, "step": 234590 }, { "epoch": 149.52198852772466, "grad_norm": 0.22031743824481964, "learning_rate": 1.24401886568909e-05, "loss": 0.0088, "step": 234600 }, { "epoch": 149.52836201402167, "grad_norm": 0.1522902399301529, "learning_rate": 1.2436550816319275e-05, "loss": 0.0085, "step": 234610 }, { "epoch": 149.53473550031867, "grad_norm": 0.09942232817411423, "learning_rate": 1.2432913432179005e-05, "loss": 0.0061, "step": 234620 }, { "epoch": 149.54110898661568, "grad_norm": 0.1002417579293251, "learning_rate": 1.2429276504514337e-05, "loss": 0.0042, "step": 234630 }, { "epoch": 149.5474824729127, "grad_norm": 0.5143134593963623, "learning_rate": 1.2425640033369401e-05, "loss": 0.0091, "step": 234640 }, { "epoch": 149.5538559592097, "grad_norm": 0.15231755375862122, "learning_rate": 1.2422004018788436e-05, "loss": 0.0101, "step": 234650 }, { "epoch": 149.56022944550668, "grad_norm": 0.11713647097349167, "learning_rate": 1.2418368460815577e-05, "loss": 0.004, "step": 234660 }, { "epoch": 149.5666029318037, "grad_norm": 0.2575794458389282, "learning_rate": 1.2414733359495062e-05, "loss": 0.0062, "step": 234670 }, { "epoch": 149.5729764181007, "grad_norm": 0.1879737377166748, "learning_rate": 1.241109871487099e-05, "loss": 0.0044, "step": 234680 }, { "epoch": 149.5793499043977, "grad_norm": 0.23468631505966187, "learning_rate": 1.2407464526987571e-05, "loss": 0.0048, "step": 234690 }, { "epoch": 149.5857233906947, "grad_norm": 0.12593871355056763, "learning_rate": 1.2403830795888954e-05, "loss": 0.0036, "step": 234700 }, { "epoch": 149.59209687699172, "grad_norm": 0.14977331459522247, "learning_rate": 1.240019752161929e-05, "loss": 0.0039, "step": 234710 }, { "epoch": 149.59847036328873, "grad_norm": 0.2364838868379593, "learning_rate": 1.2396564704222714e-05, "loss": 0.0097, "step": 234720 }, { "epoch": 149.60484384958573, "grad_norm": 0.24193108081817627, "learning_rate": 1.239293234374339e-05, "loss": 0.0178, "step": 234730 }, { "epoch": 149.61121733588274, "grad_norm": 0.22975419461727142, "learning_rate": 1.2389300440225449e-05, "loss": 0.0095, "step": 234740 }, { "epoch": 149.61759082217972, "grad_norm": 0.11173935979604721, "learning_rate": 1.2385668993713012e-05, "loss": 0.0036, "step": 234750 }, { "epoch": 149.62396430847673, "grad_norm": 0.11936430633068085, "learning_rate": 1.2382038004250201e-05, "loss": 0.0034, "step": 234760 }, { "epoch": 149.63033779477374, "grad_norm": 0.29820409417152405, "learning_rate": 1.2378407471881176e-05, "loss": 0.0053, "step": 234770 }, { "epoch": 149.63671128107075, "grad_norm": 0.12079527229070663, "learning_rate": 1.2374777396649989e-05, "loss": 0.0191, "step": 234780 }, { "epoch": 149.64308476736775, "grad_norm": 0.5595112442970276, "learning_rate": 1.2371147778600794e-05, "loss": 0.0096, "step": 234790 }, { "epoch": 149.64945825366476, "grad_norm": 0.25093895196914673, "learning_rate": 1.2367518617777668e-05, "loss": 0.0065, "step": 234800 }, { "epoch": 149.65583173996177, "grad_norm": 0.4257009029388428, "learning_rate": 1.2363889914224753e-05, "loss": 0.0175, "step": 234810 }, { "epoch": 149.66220522625878, "grad_norm": 0.31472834944725037, "learning_rate": 1.2360261667986073e-05, "loss": 0.0049, "step": 234820 }, { "epoch": 149.66857871255576, "grad_norm": 0.613879919052124, "learning_rate": 1.2356633879105772e-05, "loss": 0.0063, "step": 234830 }, { "epoch": 149.67495219885276, "grad_norm": 0.1108255609869957, "learning_rate": 1.2353006547627898e-05, "loss": 0.0054, "step": 234840 }, { "epoch": 149.68132568514977, "grad_norm": 0.1081448495388031, "learning_rate": 1.2349379673596568e-05, "loss": 0.0087, "step": 234850 }, { "epoch": 149.68769917144678, "grad_norm": 0.44297870993614197, "learning_rate": 1.2345753257055786e-05, "loss": 0.009, "step": 234860 }, { "epoch": 149.6940726577438, "grad_norm": 0.20804305374622345, "learning_rate": 1.2342127298049676e-05, "loss": 0.0181, "step": 234870 }, { "epoch": 149.7004461440408, "grad_norm": 0.2931245267391205, "learning_rate": 1.2338501796622271e-05, "loss": 0.0066, "step": 234880 }, { "epoch": 149.7068196303378, "grad_norm": 0.22290216386318207, "learning_rate": 1.233487675281763e-05, "loss": 0.0045, "step": 234890 }, { "epoch": 149.7131931166348, "grad_norm": 0.3807258903980255, "learning_rate": 1.2331252166679786e-05, "loss": 0.0063, "step": 234900 }, { "epoch": 149.7195666029318, "grad_norm": 0.2820877730846405, "learning_rate": 1.2327628038252803e-05, "loss": 0.0054, "step": 234910 }, { "epoch": 149.7259400892288, "grad_norm": 0.1631147861480713, "learning_rate": 1.2324004367580705e-05, "loss": 0.0072, "step": 234920 }, { "epoch": 149.7323135755258, "grad_norm": 0.23666352033615112, "learning_rate": 1.2320381154707533e-05, "loss": 0.0051, "step": 234930 }, { "epoch": 149.73868706182282, "grad_norm": 0.21757416427135468, "learning_rate": 1.2316758399677286e-05, "loss": 0.005, "step": 234940 }, { "epoch": 149.74506054811982, "grad_norm": 0.17827455699443817, "learning_rate": 1.2313136102534024e-05, "loss": 0.0069, "step": 234950 }, { "epoch": 149.75143403441683, "grad_norm": 0.15116146206855774, "learning_rate": 1.2309514263321736e-05, "loss": 0.0059, "step": 234960 }, { "epoch": 149.75780752071384, "grad_norm": 0.1017964780330658, "learning_rate": 1.2305892882084436e-05, "loss": 0.0049, "step": 234970 }, { "epoch": 149.76418100701085, "grad_norm": 0.12447276711463928, "learning_rate": 1.2302271958866107e-05, "loss": 0.0059, "step": 234980 }, { "epoch": 149.77055449330783, "grad_norm": 0.09066111594438553, "learning_rate": 1.229865149371079e-05, "loss": 0.0068, "step": 234990 }, { "epoch": 149.77692797960484, "grad_norm": 0.21618957817554474, "learning_rate": 1.2295031486662451e-05, "loss": 0.0077, "step": 235000 }, { "epoch": 149.78330146590184, "grad_norm": 0.399005264043808, "learning_rate": 1.2291411937765074e-05, "loss": 0.0097, "step": 235010 }, { "epoch": 149.78967495219885, "grad_norm": 0.27522045373916626, "learning_rate": 1.2287792847062645e-05, "loss": 0.0101, "step": 235020 }, { "epoch": 149.79604843849586, "grad_norm": 0.895039439201355, "learning_rate": 1.2284174214599132e-05, "loss": 0.0089, "step": 235030 }, { "epoch": 149.80242192479287, "grad_norm": 0.17404618859291077, "learning_rate": 1.2280556040418517e-05, "loss": 0.0063, "step": 235040 }, { "epoch": 149.80879541108987, "grad_norm": 0.5419747829437256, "learning_rate": 1.2276938324564768e-05, "loss": 0.0081, "step": 235050 }, { "epoch": 149.81516889738688, "grad_norm": 0.1305268257856369, "learning_rate": 1.2273321067081827e-05, "loss": 0.0048, "step": 235060 }, { "epoch": 149.82154238368386, "grad_norm": 0.1913105696439743, "learning_rate": 1.2269704268013649e-05, "loss": 0.0036, "step": 235070 }, { "epoch": 149.82791586998087, "grad_norm": 0.533612847328186, "learning_rate": 1.2266087927404196e-05, "loss": 0.0076, "step": 235080 }, { "epoch": 149.83428935627788, "grad_norm": 0.10153429210186005, "learning_rate": 1.2262472045297403e-05, "loss": 0.0061, "step": 235090 }, { "epoch": 149.8406628425749, "grad_norm": 0.306753545999527, "learning_rate": 1.2258856621737203e-05, "loss": 0.0056, "step": 235100 }, { "epoch": 149.8470363288719, "grad_norm": 0.2295723855495453, "learning_rate": 1.2255241656767514e-05, "loss": 0.0035, "step": 235110 }, { "epoch": 149.8534098151689, "grad_norm": 0.4979839324951172, "learning_rate": 1.2251627150432293e-05, "loss": 0.0103, "step": 235120 }, { "epoch": 149.8597833014659, "grad_norm": 0.1703822910785675, "learning_rate": 1.2248013102775441e-05, "loss": 0.0051, "step": 235130 }, { "epoch": 149.86615678776292, "grad_norm": 0.7092081904411316, "learning_rate": 1.2244399513840877e-05, "loss": 0.0136, "step": 235140 }, { "epoch": 149.8725302740599, "grad_norm": 0.31269896030426025, "learning_rate": 1.2240786383672487e-05, "loss": 0.0045, "step": 235150 }, { "epoch": 149.8789037603569, "grad_norm": 0.1956835389137268, "learning_rate": 1.223717371231421e-05, "loss": 0.0143, "step": 235160 }, { "epoch": 149.8852772466539, "grad_norm": 0.2757374048233032, "learning_rate": 1.223356149980992e-05, "loss": 0.0047, "step": 235170 }, { "epoch": 149.89165073295092, "grad_norm": 0.4109291434288025, "learning_rate": 1.2229949746203523e-05, "loss": 0.0079, "step": 235180 }, { "epoch": 149.89802421924793, "grad_norm": 0.18587766587734222, "learning_rate": 1.2226338451538882e-05, "loss": 0.005, "step": 235190 }, { "epoch": 149.90439770554494, "grad_norm": 0.2630528509616852, "learning_rate": 1.2222727615859919e-05, "loss": 0.0052, "step": 235200 }, { "epoch": 149.91077119184195, "grad_norm": 0.1455637812614441, "learning_rate": 1.2219117239210454e-05, "loss": 0.0058, "step": 235210 }, { "epoch": 149.91714467813895, "grad_norm": 0.10366453230381012, "learning_rate": 1.2215507321634406e-05, "loss": 0.0077, "step": 235220 }, { "epoch": 149.92351816443593, "grad_norm": 0.22919270396232605, "learning_rate": 1.2211897863175597e-05, "loss": 0.0052, "step": 235230 }, { "epoch": 149.92989165073294, "grad_norm": 0.9232284426689148, "learning_rate": 1.2208288863877937e-05, "loss": 0.007, "step": 235240 }, { "epoch": 149.93626513702995, "grad_norm": 0.15251997113227844, "learning_rate": 1.2204680323785223e-05, "loss": 0.0073, "step": 235250 }, { "epoch": 149.94263862332696, "grad_norm": 0.19395169615745544, "learning_rate": 1.2201072242941342e-05, "loss": 0.0076, "step": 235260 }, { "epoch": 149.94901210962396, "grad_norm": 0.2142329216003418, "learning_rate": 1.219746462139012e-05, "loss": 0.0033, "step": 235270 }, { "epoch": 149.95538559592097, "grad_norm": 0.09699216485023499, "learning_rate": 1.2193857459175394e-05, "loss": 0.0044, "step": 235280 }, { "epoch": 149.96175908221798, "grad_norm": 0.49980416893959045, "learning_rate": 1.2190250756340981e-05, "loss": 0.0058, "step": 235290 }, { "epoch": 149.968132568515, "grad_norm": 0.3685292899608612, "learning_rate": 1.2186644512930733e-05, "loss": 0.0087, "step": 235300 }, { "epoch": 149.97450605481197, "grad_norm": 0.16016905009746552, "learning_rate": 1.2183038728988454e-05, "loss": 0.0044, "step": 235310 }, { "epoch": 149.98087954110898, "grad_norm": 0.24920368194580078, "learning_rate": 1.2179433404557966e-05, "loss": 0.0078, "step": 235320 }, { "epoch": 149.98725302740598, "grad_norm": 0.11483818292617798, "learning_rate": 1.2175828539683049e-05, "loss": 0.0101, "step": 235330 }, { "epoch": 149.993626513703, "grad_norm": 0.48779240250587463, "learning_rate": 1.2172224134407562e-05, "loss": 0.0091, "step": 235340 }, { "epoch": 150.0, "grad_norm": 0.638382077217102, "learning_rate": 1.2168620188775231e-05, "loss": 0.0091, "step": 235350 }, { "epoch": 150.006373486297, "grad_norm": 0.2441348433494568, "learning_rate": 1.2165016702829906e-05, "loss": 0.0076, "step": 235360 }, { "epoch": 150.01274697259402, "grad_norm": 0.6090556979179382, "learning_rate": 1.216141367661533e-05, "loss": 0.0082, "step": 235370 }, { "epoch": 150.01912045889102, "grad_norm": 0.8559185266494751, "learning_rate": 1.2157811110175338e-05, "loss": 0.008, "step": 235380 }, { "epoch": 150.02549394518803, "grad_norm": 0.22606950998306274, "learning_rate": 1.2154209003553634e-05, "loss": 0.0072, "step": 235390 }, { "epoch": 150.031867431485, "grad_norm": 0.22257943451404572, "learning_rate": 1.2150607356794041e-05, "loss": 0.0104, "step": 235400 }, { "epoch": 150.03824091778202, "grad_norm": 0.11405013501644135, "learning_rate": 1.2147006169940305e-05, "loss": 0.0041, "step": 235410 }, { "epoch": 150.04461440407903, "grad_norm": 0.11456470191478729, "learning_rate": 1.2143405443036182e-05, "loss": 0.0048, "step": 235420 }, { "epoch": 150.05098789037604, "grad_norm": 0.24218574166297913, "learning_rate": 1.2139805176125413e-05, "loss": 0.0052, "step": 235430 }, { "epoch": 150.05736137667304, "grad_norm": 0.1914362907409668, "learning_rate": 1.2136205369251768e-05, "loss": 0.0047, "step": 235440 }, { "epoch": 150.06373486297005, "grad_norm": 0.3381308615207672, "learning_rate": 1.2132606022458981e-05, "loss": 0.0082, "step": 235450 }, { "epoch": 150.07010834926706, "grad_norm": 0.300401896238327, "learning_rate": 1.2129007135790782e-05, "loss": 0.0102, "step": 235460 }, { "epoch": 150.07648183556407, "grad_norm": 0.20277172327041626, "learning_rate": 1.2125408709290886e-05, "loss": 0.0089, "step": 235470 }, { "epoch": 150.08285532186105, "grad_norm": 0.12556393444538116, "learning_rate": 1.2121810743003053e-05, "loss": 0.0046, "step": 235480 }, { "epoch": 150.08922880815805, "grad_norm": 0.15354801714420319, "learning_rate": 1.2118213236970976e-05, "loss": 0.0084, "step": 235490 }, { "epoch": 150.09560229445506, "grad_norm": 0.19296406209468842, "learning_rate": 1.2114616191238382e-05, "loss": 0.0223, "step": 235500 }, { "epoch": 150.10197578075207, "grad_norm": 0.06568257510662079, "learning_rate": 1.211101960584895e-05, "loss": 0.0035, "step": 235510 }, { "epoch": 150.10834926704908, "grad_norm": 0.16641020774841309, "learning_rate": 1.2107423480846437e-05, "loss": 0.005, "step": 235520 }, { "epoch": 150.1147227533461, "grad_norm": 0.13891515135765076, "learning_rate": 1.2103827816274472e-05, "loss": 0.004, "step": 235530 }, { "epoch": 150.1210962396431, "grad_norm": 0.12718673050403595, "learning_rate": 1.2100232612176793e-05, "loss": 0.006, "step": 235540 }, { "epoch": 150.1274697259401, "grad_norm": 0.1858956217765808, "learning_rate": 1.2096637868597061e-05, "loss": 0.0043, "step": 235550 }, { "epoch": 150.13384321223708, "grad_norm": 0.12023036926984787, "learning_rate": 1.209304358557899e-05, "loss": 0.0036, "step": 235560 }, { "epoch": 150.1402166985341, "grad_norm": 0.0798378586769104, "learning_rate": 1.20894497631662e-05, "loss": 0.0028, "step": 235570 }, { "epoch": 150.1465901848311, "grad_norm": 0.1121440976858139, "learning_rate": 1.2085856401402407e-05, "loss": 0.0055, "step": 235580 }, { "epoch": 150.1529636711281, "grad_norm": 0.5081676244735718, "learning_rate": 1.2082263500331254e-05, "loss": 0.008, "step": 235590 }, { "epoch": 150.1593371574251, "grad_norm": 0.28524309396743774, "learning_rate": 1.2078671059996405e-05, "loss": 0.0042, "step": 235600 }, { "epoch": 150.16571064372212, "grad_norm": 0.21537750959396362, "learning_rate": 1.2075079080441482e-05, "loss": 0.0036, "step": 235610 }, { "epoch": 150.17208413001913, "grad_norm": 0.07881610840559006, "learning_rate": 1.2071487561710176e-05, "loss": 0.0045, "step": 235620 }, { "epoch": 150.17845761631614, "grad_norm": 0.09913168102502823, "learning_rate": 1.2067896503846104e-05, "loss": 0.006, "step": 235630 }, { "epoch": 150.18483110261312, "grad_norm": 0.1884666532278061, "learning_rate": 1.2064305906892909e-05, "loss": 0.0046, "step": 235640 }, { "epoch": 150.19120458891013, "grad_norm": 0.1803775131702423, "learning_rate": 1.2060715770894193e-05, "loss": 0.006, "step": 235650 }, { "epoch": 150.19757807520713, "grad_norm": 0.08751586824655533, "learning_rate": 1.2057126095893618e-05, "loss": 0.0089, "step": 235660 }, { "epoch": 150.20395156150414, "grad_norm": 0.12290781736373901, "learning_rate": 1.2053536881934785e-05, "loss": 0.0052, "step": 235670 }, { "epoch": 150.21032504780115, "grad_norm": 0.15854735672473907, "learning_rate": 1.2049948129061307e-05, "loss": 0.0085, "step": 235680 }, { "epoch": 150.21669853409816, "grad_norm": 0.3418775200843811, "learning_rate": 1.2046359837316773e-05, "loss": 0.0049, "step": 235690 }, { "epoch": 150.22307202039516, "grad_norm": 0.18583443760871887, "learning_rate": 1.2042772006744818e-05, "loss": 0.0051, "step": 235700 }, { "epoch": 150.22944550669217, "grad_norm": 0.19189223647117615, "learning_rate": 1.2039184637389023e-05, "loss": 0.0036, "step": 235710 }, { "epoch": 150.23581899298915, "grad_norm": 0.27417081594467163, "learning_rate": 1.2035597729292974e-05, "loss": 0.009, "step": 235720 }, { "epoch": 150.24219247928616, "grad_norm": 0.18064627051353455, "learning_rate": 1.203201128250024e-05, "loss": 0.0046, "step": 235730 }, { "epoch": 150.24856596558317, "grad_norm": 0.22619940340518951, "learning_rate": 1.2028425297054429e-05, "loss": 0.0273, "step": 235740 }, { "epoch": 150.25493945188018, "grad_norm": 0.34876585006713867, "learning_rate": 1.2024839772999109e-05, "loss": 0.004, "step": 235750 }, { "epoch": 150.26131293817718, "grad_norm": 0.11266740411520004, "learning_rate": 1.2021254710377838e-05, "loss": 0.0082, "step": 235760 }, { "epoch": 150.2676864244742, "grad_norm": 0.3630247414112091, "learning_rate": 1.2017670109234175e-05, "loss": 0.0039, "step": 235770 }, { "epoch": 150.2740599107712, "grad_norm": 0.13329026103019714, "learning_rate": 1.2014085969611672e-05, "loss": 0.0146, "step": 235780 }, { "epoch": 150.2804333970682, "grad_norm": 0.07434424012899399, "learning_rate": 1.2010502291553904e-05, "loss": 0.0102, "step": 235790 }, { "epoch": 150.2868068833652, "grad_norm": 0.4945116937160492, "learning_rate": 1.2006919075104394e-05, "loss": 0.0082, "step": 235800 }, { "epoch": 150.2931803696622, "grad_norm": 0.17241373658180237, "learning_rate": 1.2003336320306696e-05, "loss": 0.0065, "step": 235810 }, { "epoch": 150.2995538559592, "grad_norm": 0.09497053921222687, "learning_rate": 1.1999754027204319e-05, "loss": 0.0051, "step": 235820 }, { "epoch": 150.3059273422562, "grad_norm": 0.12973612546920776, "learning_rate": 1.1996172195840826e-05, "loss": 0.0067, "step": 235830 }, { "epoch": 150.31230082855322, "grad_norm": 0.3372448682785034, "learning_rate": 1.199259082625972e-05, "loss": 0.0082, "step": 235840 }, { "epoch": 150.31867431485023, "grad_norm": 0.1288893073797226, "learning_rate": 1.1989009918504523e-05, "loss": 0.0042, "step": 235850 }, { "epoch": 150.32504780114724, "grad_norm": 0.10151948779821396, "learning_rate": 1.1985429472618726e-05, "loss": 0.0061, "step": 235860 }, { "epoch": 150.33142128744424, "grad_norm": 0.12099791318178177, "learning_rate": 1.1981849488645869e-05, "loss": 0.0042, "step": 235870 }, { "epoch": 150.33779477374122, "grad_norm": 0.06438552588224411, "learning_rate": 1.1978269966629436e-05, "loss": 0.004, "step": 235880 }, { "epoch": 150.34416826003823, "grad_norm": 0.5240074992179871, "learning_rate": 1.1974690906612918e-05, "loss": 0.0096, "step": 235890 }, { "epoch": 150.35054174633524, "grad_norm": 0.14994385838508606, "learning_rate": 1.1971112308639798e-05, "loss": 0.0061, "step": 235900 }, { "epoch": 150.35691523263225, "grad_norm": 0.3683505654335022, "learning_rate": 1.1967534172753591e-05, "loss": 0.0115, "step": 235910 }, { "epoch": 150.36328871892925, "grad_norm": 0.17379698157310486, "learning_rate": 1.196395649899773e-05, "loss": 0.006, "step": 235920 }, { "epoch": 150.36966220522626, "grad_norm": 0.10367763787508011, "learning_rate": 1.1960379287415724e-05, "loss": 0.0091, "step": 235930 }, { "epoch": 150.37603569152327, "grad_norm": 0.12135497480630875, "learning_rate": 1.1956802538051004e-05, "loss": 0.0074, "step": 235940 }, { "epoch": 150.38240917782028, "grad_norm": 0.17818927764892578, "learning_rate": 1.1953226250947086e-05, "loss": 0.0041, "step": 235950 }, { "epoch": 150.38878266411726, "grad_norm": 0.43479982018470764, "learning_rate": 1.1949650426147363e-05, "loss": 0.0064, "step": 235960 }, { "epoch": 150.39515615041427, "grad_norm": 0.17887380719184875, "learning_rate": 1.1946075063695321e-05, "loss": 0.0074, "step": 235970 }, { "epoch": 150.40152963671127, "grad_norm": 0.2646559178829193, "learning_rate": 1.1942500163634402e-05, "loss": 0.0074, "step": 235980 }, { "epoch": 150.40790312300828, "grad_norm": 0.14737741649150848, "learning_rate": 1.1938925726008038e-05, "loss": 0.0065, "step": 235990 }, { "epoch": 150.4142766093053, "grad_norm": 0.13834939897060394, "learning_rate": 1.1935351750859648e-05, "loss": 0.0185, "step": 236000 }, { "epoch": 150.4206500956023, "grad_norm": 0.18043173849582672, "learning_rate": 1.1931778238232683e-05, "loss": 0.0056, "step": 236010 }, { "epoch": 150.4270235818993, "grad_norm": 0.07299195975065231, "learning_rate": 1.192820518817056e-05, "loss": 0.0051, "step": 236020 }, { "epoch": 150.4333970681963, "grad_norm": 0.16213619709014893, "learning_rate": 1.1924632600716685e-05, "loss": 0.0112, "step": 236030 }, { "epoch": 150.43977055449332, "grad_norm": 0.29969093203544617, "learning_rate": 1.1921060475914464e-05, "loss": 0.0038, "step": 236040 }, { "epoch": 150.4461440407903, "grad_norm": 0.3936348855495453, "learning_rate": 1.191748881380732e-05, "loss": 0.0042, "step": 236050 }, { "epoch": 150.4525175270873, "grad_norm": 0.1318785846233368, "learning_rate": 1.1913917614438641e-05, "loss": 0.0083, "step": 236060 }, { "epoch": 150.45889101338432, "grad_norm": 0.12791626155376434, "learning_rate": 1.191034687785183e-05, "loss": 0.0064, "step": 236070 }, { "epoch": 150.46526449968133, "grad_norm": 0.11048905551433563, "learning_rate": 1.1906776604090247e-05, "loss": 0.0041, "step": 236080 }, { "epoch": 150.47163798597833, "grad_norm": 0.10832540690898895, "learning_rate": 1.1903206793197325e-05, "loss": 0.0064, "step": 236090 }, { "epoch": 150.47801147227534, "grad_norm": 0.29980218410491943, "learning_rate": 1.1899637445216377e-05, "loss": 0.0063, "step": 236100 }, { "epoch": 150.48438495857235, "grad_norm": 0.18110810220241547, "learning_rate": 1.1896068560190827e-05, "loss": 0.0101, "step": 236110 }, { "epoch": 150.49075844486936, "grad_norm": 0.4904848039150238, "learning_rate": 1.1892500138164003e-05, "loss": 0.0085, "step": 236120 }, { "epoch": 150.49713193116634, "grad_norm": 0.17653988301753998, "learning_rate": 1.1888932179179313e-05, "loss": 0.0088, "step": 236130 }, { "epoch": 150.50350541746334, "grad_norm": 0.12783211469650269, "learning_rate": 1.1885364683280048e-05, "loss": 0.0066, "step": 236140 }, { "epoch": 150.50987890376035, "grad_norm": 0.15844836831092834, "learning_rate": 1.1881797650509601e-05, "loss": 0.0084, "step": 236150 }, { "epoch": 150.51625239005736, "grad_norm": 0.45566150546073914, "learning_rate": 1.1878231080911307e-05, "loss": 0.0108, "step": 236160 }, { "epoch": 150.52262587635437, "grad_norm": 0.1041959896683693, "learning_rate": 1.1874664974528494e-05, "loss": 0.0112, "step": 236170 }, { "epoch": 150.52899936265138, "grad_norm": 0.12931789457798004, "learning_rate": 1.187109933140449e-05, "loss": 0.0046, "step": 236180 }, { "epoch": 150.53537284894838, "grad_norm": 0.22689352929592133, "learning_rate": 1.1867534151582637e-05, "loss": 0.0051, "step": 236190 }, { "epoch": 150.5417463352454, "grad_norm": 0.2117740958929062, "learning_rate": 1.1863969435106248e-05, "loss": 0.0052, "step": 236200 }, { "epoch": 150.54811982154237, "grad_norm": 0.11276483535766602, "learning_rate": 1.1860405182018635e-05, "loss": 0.0077, "step": 236210 }, { "epoch": 150.55449330783938, "grad_norm": 0.14775054156780243, "learning_rate": 1.18568413923631e-05, "loss": 0.0058, "step": 236220 }, { "epoch": 150.5608667941364, "grad_norm": 0.30365175008773804, "learning_rate": 1.1853278066182977e-05, "loss": 0.0063, "step": 236230 }, { "epoch": 150.5672402804334, "grad_norm": 0.42224234342575073, "learning_rate": 1.1849715203521516e-05, "loss": 0.007, "step": 236240 }, { "epoch": 150.5736137667304, "grad_norm": 0.13748694956302643, "learning_rate": 1.1846152804422056e-05, "loss": 0.0084, "step": 236250 }, { "epoch": 150.5799872530274, "grad_norm": 0.2531534433364868, "learning_rate": 1.184259086892784e-05, "loss": 0.0047, "step": 236260 }, { "epoch": 150.58636073932442, "grad_norm": 0.2857872247695923, "learning_rate": 1.1839029397082202e-05, "loss": 0.0051, "step": 236270 }, { "epoch": 150.59273422562143, "grad_norm": 0.10281901061534882, "learning_rate": 1.183546838892836e-05, "loss": 0.0075, "step": 236280 }, { "epoch": 150.5991077119184, "grad_norm": 0.1656620353460312, "learning_rate": 1.1831907844509621e-05, "loss": 0.0042, "step": 236290 }, { "epoch": 150.60548119821541, "grad_norm": 0.3610387444496155, "learning_rate": 1.182834776386924e-05, "loss": 0.0073, "step": 236300 }, { "epoch": 150.61185468451242, "grad_norm": 0.14559490978717804, "learning_rate": 1.182478814705048e-05, "loss": 0.0114, "step": 236310 }, { "epoch": 150.61822817080943, "grad_norm": 0.13993018865585327, "learning_rate": 1.1821228994096567e-05, "loss": 0.0041, "step": 236320 }, { "epoch": 150.62460165710644, "grad_norm": 0.48695507645606995, "learning_rate": 1.1817670305050783e-05, "loss": 0.0067, "step": 236330 }, { "epoch": 150.63097514340345, "grad_norm": 0.16241596639156342, "learning_rate": 1.1814112079956358e-05, "loss": 0.0057, "step": 236340 }, { "epoch": 150.63734862970045, "grad_norm": 0.18118096888065338, "learning_rate": 1.1810554318856521e-05, "loss": 0.0045, "step": 236350 }, { "epoch": 150.64372211599746, "grad_norm": 0.1556483805179596, "learning_rate": 1.1806997021794492e-05, "loss": 0.0064, "step": 236360 }, { "epoch": 150.65009560229444, "grad_norm": 0.23742584884166718, "learning_rate": 1.1803440188813524e-05, "loss": 0.007, "step": 236370 }, { "epoch": 150.65646908859145, "grad_norm": 0.5024580955505371, "learning_rate": 1.179988381995682e-05, "loss": 0.007, "step": 236380 }, { "epoch": 150.66284257488846, "grad_norm": 0.3111315071582794, "learning_rate": 1.1796327915267597e-05, "loss": 0.0113, "step": 236390 }, { "epoch": 150.66921606118547, "grad_norm": 0.14762452244758606, "learning_rate": 1.1792772474789043e-05, "loss": 0.0039, "step": 236400 }, { "epoch": 150.67558954748247, "grad_norm": 0.2676805555820465, "learning_rate": 1.1789217498564393e-05, "loss": 0.0047, "step": 236410 }, { "epoch": 150.68196303377948, "grad_norm": 0.28949517011642456, "learning_rate": 1.178566298663683e-05, "loss": 0.0048, "step": 236420 }, { "epoch": 150.6883365200765, "grad_norm": 0.16150279343128204, "learning_rate": 1.1782108939049535e-05, "loss": 0.004, "step": 236430 }, { "epoch": 150.6947100063735, "grad_norm": 0.14314092695713043, "learning_rate": 1.1778555355845695e-05, "loss": 0.0074, "step": 236440 }, { "epoch": 150.70108349267048, "grad_norm": 0.243682861328125, "learning_rate": 1.17750022370685e-05, "loss": 0.0109, "step": 236450 }, { "epoch": 150.70745697896749, "grad_norm": 0.18960748612880707, "learning_rate": 1.1771449582761124e-05, "loss": 0.0054, "step": 236460 }, { "epoch": 150.7138304652645, "grad_norm": 0.3697507977485657, "learning_rate": 1.1767897392966726e-05, "loss": 0.0058, "step": 236470 }, { "epoch": 150.7202039515615, "grad_norm": 0.2092149406671524, "learning_rate": 1.1764345667728476e-05, "loss": 0.0073, "step": 236480 }, { "epoch": 150.7265774378585, "grad_norm": 0.11561663448810577, "learning_rate": 1.176079440708951e-05, "loss": 0.0061, "step": 236490 }, { "epoch": 150.73295092415552, "grad_norm": 0.09377547353506088, "learning_rate": 1.175724361109301e-05, "loss": 0.0139, "step": 236500 }, { "epoch": 150.73932441045253, "grad_norm": 0.14164939522743225, "learning_rate": 1.1753693279782108e-05, "loss": 0.0042, "step": 236510 }, { "epoch": 150.74569789674953, "grad_norm": 0.12311585992574692, "learning_rate": 1.1750143413199938e-05, "loss": 0.0097, "step": 236520 }, { "epoch": 150.7520713830465, "grad_norm": 0.22552192211151123, "learning_rate": 1.1746594011389627e-05, "loss": 0.0052, "step": 236530 }, { "epoch": 150.75844486934352, "grad_norm": 0.23076364398002625, "learning_rate": 1.1743045074394337e-05, "loss": 0.0055, "step": 236540 }, { "epoch": 150.76481835564053, "grad_norm": 0.14563266932964325, "learning_rate": 1.1739496602257167e-05, "loss": 0.0092, "step": 236550 }, { "epoch": 150.77119184193754, "grad_norm": 0.4023200571537018, "learning_rate": 1.1735948595021235e-05, "loss": 0.0071, "step": 236560 }, { "epoch": 150.77756532823454, "grad_norm": 0.4778996407985687, "learning_rate": 1.1732401052729636e-05, "loss": 0.0105, "step": 236570 }, { "epoch": 150.78393881453155, "grad_norm": 0.1519021987915039, "learning_rate": 1.1728853975425519e-05, "loss": 0.007, "step": 236580 }, { "epoch": 150.79031230082856, "grad_norm": 0.22365067899227142, "learning_rate": 1.1725307363151954e-05, "loss": 0.0044, "step": 236590 }, { "epoch": 150.79668578712557, "grad_norm": 0.11142333596944809, "learning_rate": 1.1721761215952043e-05, "loss": 0.0079, "step": 236600 }, { "epoch": 150.80305927342255, "grad_norm": 0.3798063397407532, "learning_rate": 1.171821553386886e-05, "loss": 0.0042, "step": 236610 }, { "epoch": 150.80943275971956, "grad_norm": 0.46915313601493835, "learning_rate": 1.171467031694553e-05, "loss": 0.0054, "step": 236620 }, { "epoch": 150.81580624601656, "grad_norm": 0.44401276111602783, "learning_rate": 1.1711125565225073e-05, "loss": 0.0086, "step": 236630 }, { "epoch": 150.82217973231357, "grad_norm": 0.3458317518234253, "learning_rate": 1.1707581278750607e-05, "loss": 0.0168, "step": 236640 }, { "epoch": 150.82855321861058, "grad_norm": 0.12205207347869873, "learning_rate": 1.170403745756517e-05, "loss": 0.0119, "step": 236650 }, { "epoch": 150.8349267049076, "grad_norm": 0.4633494019508362, "learning_rate": 1.1700494101711861e-05, "loss": 0.0139, "step": 236660 }, { "epoch": 150.8413001912046, "grad_norm": 0.23499442636966705, "learning_rate": 1.169695121123368e-05, "loss": 0.0114, "step": 236670 }, { "epoch": 150.8476736775016, "grad_norm": 0.2817433774471283, "learning_rate": 1.1693408786173726e-05, "loss": 0.0066, "step": 236680 }, { "epoch": 150.8540471637986, "grad_norm": 0.10787025094032288, "learning_rate": 1.1689866826575003e-05, "loss": 0.0033, "step": 236690 }, { "epoch": 150.8604206500956, "grad_norm": 0.12301202118396759, "learning_rate": 1.1686325332480597e-05, "loss": 0.0053, "step": 236700 }, { "epoch": 150.8667941363926, "grad_norm": 0.22241850197315216, "learning_rate": 1.1682784303933486e-05, "loss": 0.0154, "step": 236710 }, { "epoch": 150.8731676226896, "grad_norm": 0.22483408451080322, "learning_rate": 1.1679243740976742e-05, "loss": 0.0074, "step": 236720 }, { "epoch": 150.87954110898661, "grad_norm": 0.26526129245758057, "learning_rate": 1.1675703643653358e-05, "loss": 0.0046, "step": 236730 }, { "epoch": 150.88591459528362, "grad_norm": 0.0849854126572609, "learning_rate": 1.1672164012006365e-05, "loss": 0.0045, "step": 236740 }, { "epoch": 150.89228808158063, "grad_norm": 0.28628477454185486, "learning_rate": 1.1668624846078752e-05, "loss": 0.0047, "step": 236750 }, { "epoch": 150.89866156787764, "grad_norm": 0.18216562271118164, "learning_rate": 1.166508614591355e-05, "loss": 0.0036, "step": 236760 }, { "epoch": 150.90503505417465, "grad_norm": 0.28208038210868835, "learning_rate": 1.1661547911553744e-05, "loss": 0.005, "step": 236770 }, { "epoch": 150.91140854047163, "grad_norm": 0.3695245087146759, "learning_rate": 1.1658010143042336e-05, "loss": 0.0064, "step": 236780 }, { "epoch": 150.91778202676863, "grad_norm": 0.1176077201962471, "learning_rate": 1.1654472840422282e-05, "loss": 0.0058, "step": 236790 }, { "epoch": 150.92415551306564, "grad_norm": 0.07304392009973526, "learning_rate": 1.1650936003736624e-05, "loss": 0.0043, "step": 236800 }, { "epoch": 150.93052899936265, "grad_norm": 0.20124207437038422, "learning_rate": 1.1647399633028266e-05, "loss": 0.0078, "step": 236810 }, { "epoch": 150.93690248565966, "grad_norm": 0.20539997518062592, "learning_rate": 1.1643863728340232e-05, "loss": 0.0089, "step": 236820 }, { "epoch": 150.94327597195667, "grad_norm": 0.12484586983919144, "learning_rate": 1.1640328289715452e-05, "loss": 0.0042, "step": 236830 }, { "epoch": 150.94964945825367, "grad_norm": 0.16127480566501617, "learning_rate": 1.1636793317196926e-05, "loss": 0.0088, "step": 236840 }, { "epoch": 150.95602294455068, "grad_norm": 0.21052499115467072, "learning_rate": 1.1633258810827558e-05, "loss": 0.0041, "step": 236850 }, { "epoch": 150.96239643084766, "grad_norm": 0.3261605203151703, "learning_rate": 1.162972477065033e-05, "loss": 0.0078, "step": 236860 }, { "epoch": 150.96876991714467, "grad_norm": 0.11079608649015427, "learning_rate": 1.162619119670818e-05, "loss": 0.0062, "step": 236870 }, { "epoch": 150.97514340344168, "grad_norm": 0.384586364030838, "learning_rate": 1.162265808904403e-05, "loss": 0.0065, "step": 236880 }, { "epoch": 150.98151688973869, "grad_norm": 0.20290836691856384, "learning_rate": 1.1619125447700813e-05, "loss": 0.0054, "step": 236890 }, { "epoch": 150.9878903760357, "grad_norm": 0.41740739345550537, "learning_rate": 1.1615593272721465e-05, "loss": 0.0066, "step": 236900 }, { "epoch": 150.9942638623327, "grad_norm": 0.17310117185115814, "learning_rate": 1.1612061564148902e-05, "loss": 0.007, "step": 236910 }, { "epoch": 151.0006373486297, "grad_norm": 0.18409860134124756, "learning_rate": 1.160853032202604e-05, "loss": 0.0046, "step": 236920 }, { "epoch": 151.00701083492672, "grad_norm": 0.23445867002010345, "learning_rate": 1.1604999546395762e-05, "loss": 0.0083, "step": 236930 }, { "epoch": 151.0133843212237, "grad_norm": 0.28854694962501526, "learning_rate": 1.1601469237301011e-05, "loss": 0.0049, "step": 236940 }, { "epoch": 151.0197578075207, "grad_norm": 0.18423497676849365, "learning_rate": 1.1597939394784657e-05, "loss": 0.0066, "step": 236950 }, { "epoch": 151.0261312938177, "grad_norm": 0.27017658948898315, "learning_rate": 1.1594410018889595e-05, "loss": 0.0067, "step": 236960 }, { "epoch": 151.03250478011472, "grad_norm": 0.08186306059360504, "learning_rate": 1.1590881109658702e-05, "loss": 0.0035, "step": 236970 }, { "epoch": 151.03887826641173, "grad_norm": 0.4054366946220398, "learning_rate": 1.1587352667134898e-05, "loss": 0.0092, "step": 236980 }, { "epoch": 151.04525175270874, "grad_norm": 0.6493861079216003, "learning_rate": 1.1583824691360995e-05, "loss": 0.0052, "step": 236990 }, { "epoch": 151.05162523900574, "grad_norm": 0.07330507785081863, "learning_rate": 1.1580297182379912e-05, "loss": 0.0113, "step": 237000 }, { "epoch": 151.05799872530275, "grad_norm": 0.5261068344116211, "learning_rate": 1.1576770140234472e-05, "loss": 0.0085, "step": 237010 }, { "epoch": 151.06437221159973, "grad_norm": 0.3437940180301666, "learning_rate": 1.1573243564967585e-05, "loss": 0.0052, "step": 237020 }, { "epoch": 151.07074569789674, "grad_norm": 0.10404927283525467, "learning_rate": 1.1569717456622037e-05, "loss": 0.0082, "step": 237030 }, { "epoch": 151.07711918419375, "grad_norm": 0.15611223876476288, "learning_rate": 1.1566191815240724e-05, "loss": 0.0062, "step": 237040 }, { "epoch": 151.08349267049076, "grad_norm": 0.1725742518901825, "learning_rate": 1.1562666640866465e-05, "loss": 0.0065, "step": 237050 }, { "epoch": 151.08986615678776, "grad_norm": 0.250772088766098, "learning_rate": 1.15591419335421e-05, "loss": 0.0215, "step": 237060 }, { "epoch": 151.09623964308477, "grad_norm": 0.1638173907995224, "learning_rate": 1.1555617693310438e-05, "loss": 0.0115, "step": 237070 }, { "epoch": 151.10261312938178, "grad_norm": 0.32563915848731995, "learning_rate": 1.1552093920214336e-05, "loss": 0.008, "step": 237080 }, { "epoch": 151.1089866156788, "grad_norm": 0.31375059485435486, "learning_rate": 1.1548570614296594e-05, "loss": 0.0111, "step": 237090 }, { "epoch": 151.11536010197577, "grad_norm": 0.11730597168207169, "learning_rate": 1.154504777560002e-05, "loss": 0.0161, "step": 237100 }, { "epoch": 151.12173358827278, "grad_norm": 0.2418045848608017, "learning_rate": 1.1541525404167408e-05, "loss": 0.0054, "step": 237110 }, { "epoch": 151.12810707456978, "grad_norm": 0.14176473021507263, "learning_rate": 1.1538003500041594e-05, "loss": 0.0144, "step": 237120 }, { "epoch": 151.1344805608668, "grad_norm": 0.2332388162612915, "learning_rate": 1.1534482063265345e-05, "loss": 0.0071, "step": 237130 }, { "epoch": 151.1408540471638, "grad_norm": 0.400095134973526, "learning_rate": 1.153096109388146e-05, "loss": 0.0139, "step": 237140 }, { "epoch": 151.1472275334608, "grad_norm": 0.3614988327026367, "learning_rate": 1.1527440591932697e-05, "loss": 0.0077, "step": 237150 }, { "epoch": 151.15360101975782, "grad_norm": 0.24638786911964417, "learning_rate": 1.1523920557461876e-05, "loss": 0.0041, "step": 237160 }, { "epoch": 151.15997450605482, "grad_norm": 0.07555806636810303, "learning_rate": 1.1520400990511749e-05, "loss": 0.0092, "step": 237170 }, { "epoch": 151.1663479923518, "grad_norm": 0.14322249591350555, "learning_rate": 1.1516881891125076e-05, "loss": 0.0084, "step": 237180 }, { "epoch": 151.1727214786488, "grad_norm": 0.08833607286214828, "learning_rate": 1.1513363259344628e-05, "loss": 0.0041, "step": 237190 }, { "epoch": 151.17909496494582, "grad_norm": 0.10736623406410217, "learning_rate": 1.1509845095213135e-05, "loss": 0.0032, "step": 237200 }, { "epoch": 151.18546845124283, "grad_norm": 0.09526318311691284, "learning_rate": 1.1506327398773382e-05, "loss": 0.0058, "step": 237210 }, { "epoch": 151.19184193753983, "grad_norm": 0.28359338641166687, "learning_rate": 1.150281017006809e-05, "loss": 0.0071, "step": 237220 }, { "epoch": 151.19821542383684, "grad_norm": 0.1082230731844902, "learning_rate": 1.1499293409140005e-05, "loss": 0.0109, "step": 237230 }, { "epoch": 151.20458891013385, "grad_norm": 0.09325654804706573, "learning_rate": 1.1495777116031841e-05, "loss": 0.003, "step": 237240 }, { "epoch": 151.21096239643086, "grad_norm": 0.31117525696754456, "learning_rate": 1.1492261290786355e-05, "loss": 0.0075, "step": 237250 }, { "epoch": 151.21733588272787, "grad_norm": 0.35126325488090515, "learning_rate": 1.148874593344625e-05, "loss": 0.0129, "step": 237260 }, { "epoch": 151.22370936902485, "grad_norm": 0.39034605026245117, "learning_rate": 1.148523104405424e-05, "loss": 0.0072, "step": 237270 }, { "epoch": 151.23008285532185, "grad_norm": 0.1640925258398056, "learning_rate": 1.1481716622653028e-05, "loss": 0.0032, "step": 237280 }, { "epoch": 151.23645634161886, "grad_norm": 0.16071783006191254, "learning_rate": 1.1478202669285337e-05, "loss": 0.0053, "step": 237290 }, { "epoch": 151.24282982791587, "grad_norm": 0.08718878775835037, "learning_rate": 1.1474689183993852e-05, "loss": 0.0074, "step": 237300 }, { "epoch": 151.24920331421288, "grad_norm": 0.2713512182235718, "learning_rate": 1.1471176166821274e-05, "loss": 0.0066, "step": 237310 }, { "epoch": 151.25557680050989, "grad_norm": 0.2789308726787567, "learning_rate": 1.1467663617810264e-05, "loss": 0.0035, "step": 237320 }, { "epoch": 151.2619502868069, "grad_norm": 0.11439848691225052, "learning_rate": 1.1464151537003537e-05, "loss": 0.0075, "step": 237330 }, { "epoch": 151.2683237731039, "grad_norm": 0.2413935661315918, "learning_rate": 1.1460639924443756e-05, "loss": 0.0055, "step": 237340 }, { "epoch": 151.27469725940088, "grad_norm": 0.4165845811367035, "learning_rate": 1.1457128780173581e-05, "loss": 0.006, "step": 237350 }, { "epoch": 151.2810707456979, "grad_norm": 0.2076205164194107, "learning_rate": 1.1453618104235669e-05, "loss": 0.0046, "step": 237360 }, { "epoch": 151.2874442319949, "grad_norm": 0.1321505755186081, "learning_rate": 1.1450107896672723e-05, "loss": 0.0042, "step": 237370 }, { "epoch": 151.2938177182919, "grad_norm": 0.25952520966529846, "learning_rate": 1.1446598157527333e-05, "loss": 0.0072, "step": 237380 }, { "epoch": 151.3001912045889, "grad_norm": 0.08443725109100342, "learning_rate": 1.1443088886842184e-05, "loss": 0.0031, "step": 237390 }, { "epoch": 151.30656469088592, "grad_norm": 0.17305836081504822, "learning_rate": 1.1439580084659901e-05, "loss": 0.01, "step": 237400 }, { "epoch": 151.31293817718293, "grad_norm": 0.6543185114860535, "learning_rate": 1.1436071751023158e-05, "loss": 0.0053, "step": 237410 }, { "epoch": 151.31931166347994, "grad_norm": 0.7884587645530701, "learning_rate": 1.1432563885974518e-05, "loss": 0.0075, "step": 237420 }, { "epoch": 151.32568514977692, "grad_norm": 0.09808345139026642, "learning_rate": 1.1429056489556661e-05, "loss": 0.0037, "step": 237430 }, { "epoch": 151.33205863607392, "grad_norm": 0.2920936346054077, "learning_rate": 1.1425549561812182e-05, "loss": 0.0033, "step": 237440 }, { "epoch": 151.33843212237093, "grad_norm": 0.09279225021600723, "learning_rate": 1.1422043102783697e-05, "loss": 0.0067, "step": 237450 }, { "epoch": 151.34480560866794, "grad_norm": 0.15559625625610352, "learning_rate": 1.1418537112513794e-05, "loss": 0.0049, "step": 237460 }, { "epoch": 151.35117909496495, "grad_norm": 0.202651247382164, "learning_rate": 1.1415031591045111e-05, "loss": 0.0087, "step": 237470 }, { "epoch": 151.35755258126196, "grad_norm": 0.09662780910730362, "learning_rate": 1.141152653842022e-05, "loss": 0.0054, "step": 237480 }, { "epoch": 151.36392606755896, "grad_norm": 0.10073503851890564, "learning_rate": 1.140802195468172e-05, "loss": 0.0077, "step": 237490 }, { "epoch": 151.37029955385597, "grad_norm": 0.1139586791396141, "learning_rate": 1.1404517839872175e-05, "loss": 0.0055, "step": 237500 }, { "epoch": 151.37667304015295, "grad_norm": 0.1834484040737152, "learning_rate": 1.1401014194034204e-05, "loss": 0.0063, "step": 237510 }, { "epoch": 151.38304652644996, "grad_norm": 0.19384871423244476, "learning_rate": 1.1397511017210328e-05, "loss": 0.0127, "step": 237520 }, { "epoch": 151.38942001274697, "grad_norm": 0.19138674437999725, "learning_rate": 1.139400830944315e-05, "loss": 0.0051, "step": 237530 }, { "epoch": 151.39579349904398, "grad_norm": 0.13412921130657196, "learning_rate": 1.1390506070775209e-05, "loss": 0.0055, "step": 237540 }, { "epoch": 151.40216698534098, "grad_norm": 0.28067389130592346, "learning_rate": 1.1387004301249105e-05, "loss": 0.0071, "step": 237550 }, { "epoch": 151.408540471638, "grad_norm": 0.11472296714782715, "learning_rate": 1.1383503000907325e-05, "loss": 0.0035, "step": 237560 }, { "epoch": 151.414913957935, "grad_norm": 0.28473126888275146, "learning_rate": 1.1380002169792454e-05, "loss": 0.0081, "step": 237570 }, { "epoch": 151.421287444232, "grad_norm": 0.13954687118530273, "learning_rate": 1.137650180794701e-05, "loss": 0.0033, "step": 237580 }, { "epoch": 151.427660930529, "grad_norm": 0.2636154294013977, "learning_rate": 1.1373001915413562e-05, "loss": 0.0204, "step": 237590 }, { "epoch": 151.434034416826, "grad_norm": 0.22443076968193054, "learning_rate": 1.1369502492234585e-05, "loss": 0.0038, "step": 237600 }, { "epoch": 151.440407903123, "grad_norm": 0.06897323578596115, "learning_rate": 1.1366003538452635e-05, "loss": 0.008, "step": 237610 }, { "epoch": 151.44678138942, "grad_norm": 0.20983876287937164, "learning_rate": 1.1362505054110224e-05, "loss": 0.0044, "step": 237620 }, { "epoch": 151.45315487571702, "grad_norm": 0.49170157313346863, "learning_rate": 1.1359007039249853e-05, "loss": 0.0067, "step": 237630 }, { "epoch": 151.45952836201403, "grad_norm": 0.07366879284381866, "learning_rate": 1.1355509493914018e-05, "loss": 0.0039, "step": 237640 }, { "epoch": 151.46590184831103, "grad_norm": 0.15822438895702362, "learning_rate": 1.1352012418145241e-05, "loss": 0.0053, "step": 237650 }, { "epoch": 151.47227533460804, "grad_norm": 0.5394644141197205, "learning_rate": 1.1348515811986005e-05, "loss": 0.0187, "step": 237660 }, { "epoch": 151.47864882090502, "grad_norm": 0.15793362259864807, "learning_rate": 1.1345019675478791e-05, "loss": 0.0044, "step": 237670 }, { "epoch": 151.48502230720203, "grad_norm": 0.2659788727760315, "learning_rate": 1.134152400866607e-05, "loss": 0.0048, "step": 237680 }, { "epoch": 151.49139579349904, "grad_norm": 0.15735593438148499, "learning_rate": 1.1338028811590367e-05, "loss": 0.0125, "step": 237690 }, { "epoch": 151.49776927979605, "grad_norm": 0.345195472240448, "learning_rate": 1.1334534084294086e-05, "loss": 0.0064, "step": 237700 }, { "epoch": 151.50414276609305, "grad_norm": 0.2027187943458557, "learning_rate": 1.1331039826819733e-05, "loss": 0.0038, "step": 237710 }, { "epoch": 151.51051625239006, "grad_norm": 0.17743629217147827, "learning_rate": 1.1327546039209742e-05, "loss": 0.0043, "step": 237720 }, { "epoch": 151.51688973868707, "grad_norm": 0.1860843300819397, "learning_rate": 1.1324052721506611e-05, "loss": 0.006, "step": 237730 }, { "epoch": 151.52326322498408, "grad_norm": 0.10929220914840698, "learning_rate": 1.1320559873752723e-05, "loss": 0.0052, "step": 237740 }, { "epoch": 151.52963671128106, "grad_norm": 0.15409216284751892, "learning_rate": 1.131706749599057e-05, "loss": 0.005, "step": 237750 }, { "epoch": 151.53601019757807, "grad_norm": 0.1766340285539627, "learning_rate": 1.1313575588262565e-05, "loss": 0.0052, "step": 237760 }, { "epoch": 151.54238368387507, "grad_norm": 0.48493921756744385, "learning_rate": 1.131008415061115e-05, "loss": 0.0064, "step": 237770 }, { "epoch": 151.54875717017208, "grad_norm": 0.2621578574180603, "learning_rate": 1.1306593183078723e-05, "loss": 0.006, "step": 237780 }, { "epoch": 151.5551306564691, "grad_norm": 0.16875682771205902, "learning_rate": 1.1303102685707734e-05, "loss": 0.01, "step": 237790 }, { "epoch": 151.5615041427661, "grad_norm": 0.22213903069496155, "learning_rate": 1.1299612658540587e-05, "loss": 0.0054, "step": 237800 }, { "epoch": 151.5678776290631, "grad_norm": 0.1519368439912796, "learning_rate": 1.1296123101619683e-05, "loss": 0.0067, "step": 237810 }, { "epoch": 151.5742511153601, "grad_norm": 0.09525824338197708, "learning_rate": 1.1292634014987413e-05, "loss": 0.0031, "step": 237820 }, { "epoch": 151.5806246016571, "grad_norm": 0.39388370513916016, "learning_rate": 1.1289145398686196e-05, "loss": 0.0063, "step": 237830 }, { "epoch": 151.5869980879541, "grad_norm": 0.723388135433197, "learning_rate": 1.128565725275842e-05, "loss": 0.0077, "step": 237840 }, { "epoch": 151.5933715742511, "grad_norm": 0.2672063112258911, "learning_rate": 1.1282169577246454e-05, "loss": 0.0028, "step": 237850 }, { "epoch": 151.59974506054812, "grad_norm": 0.3297936022281647, "learning_rate": 1.1278682372192667e-05, "loss": 0.0083, "step": 237860 }, { "epoch": 151.60611854684512, "grad_norm": 0.2811509668827057, "learning_rate": 1.1275195637639469e-05, "loss": 0.0117, "step": 237870 }, { "epoch": 151.61249203314213, "grad_norm": 0.35316169261932373, "learning_rate": 1.1271709373629202e-05, "loss": 0.0046, "step": 237880 }, { "epoch": 151.61886551943914, "grad_norm": 0.18266525864601135, "learning_rate": 1.1268223580204234e-05, "loss": 0.0045, "step": 237890 }, { "epoch": 151.62523900573615, "grad_norm": 0.26588135957717896, "learning_rate": 1.1264738257406904e-05, "loss": 0.0058, "step": 237900 }, { "epoch": 151.63161249203316, "grad_norm": 0.1801551729440689, "learning_rate": 1.1261253405279586e-05, "loss": 0.0111, "step": 237910 }, { "epoch": 151.63798597833014, "grad_norm": 0.11288680136203766, "learning_rate": 1.1257769023864622e-05, "loss": 0.0053, "step": 237920 }, { "epoch": 151.64435946462714, "grad_norm": 0.13578468561172485, "learning_rate": 1.1254285113204338e-05, "loss": 0.0044, "step": 237930 }, { "epoch": 151.65073295092415, "grad_norm": 0.3363684117794037, "learning_rate": 1.1250801673341071e-05, "loss": 0.0043, "step": 237940 }, { "epoch": 151.65710643722116, "grad_norm": 0.07922995090484619, "learning_rate": 1.124731870431714e-05, "loss": 0.0027, "step": 237950 }, { "epoch": 151.66347992351817, "grad_norm": 0.2759445607662201, "learning_rate": 1.1243836206174884e-05, "loss": 0.0074, "step": 237960 }, { "epoch": 151.66985340981518, "grad_norm": 0.14904655516147614, "learning_rate": 1.1240354178956614e-05, "loss": 0.0075, "step": 237970 }, { "epoch": 151.67622689611218, "grad_norm": 0.2219579815864563, "learning_rate": 1.1236872622704636e-05, "loss": 0.0052, "step": 237980 }, { "epoch": 151.6826003824092, "grad_norm": 0.3588109612464905, "learning_rate": 1.1233391537461235e-05, "loss": 0.0052, "step": 237990 }, { "epoch": 151.68897386870617, "grad_norm": 0.11617085337638855, "learning_rate": 1.122991092326875e-05, "loss": 0.0057, "step": 238000 }, { "epoch": 151.69534735500318, "grad_norm": 0.1799679696559906, "learning_rate": 1.1226430780169444e-05, "loss": 0.009, "step": 238010 }, { "epoch": 151.7017208413002, "grad_norm": 0.19937361776828766, "learning_rate": 1.122295110820562e-05, "loss": 0.0069, "step": 238020 }, { "epoch": 151.7080943275972, "grad_norm": 0.29403451085090637, "learning_rate": 1.1219471907419533e-05, "loss": 0.0043, "step": 238030 }, { "epoch": 151.7144678138942, "grad_norm": 0.08032389730215073, "learning_rate": 1.121599317785349e-05, "loss": 0.0072, "step": 238040 }, { "epoch": 151.7208413001912, "grad_norm": 0.2890909016132355, "learning_rate": 1.1212514919549755e-05, "loss": 0.0061, "step": 238050 }, { "epoch": 151.72721478648822, "grad_norm": 0.17063003778457642, "learning_rate": 1.1209037132550577e-05, "loss": 0.0078, "step": 238060 }, { "epoch": 151.73358827278523, "grad_norm": 0.1968267410993576, "learning_rate": 1.1205559816898215e-05, "loss": 0.0053, "step": 238070 }, { "epoch": 151.7399617590822, "grad_norm": 0.47364845871925354, "learning_rate": 1.1202082972634953e-05, "loss": 0.0073, "step": 238080 }, { "epoch": 151.7463352453792, "grad_norm": 0.21987837553024292, "learning_rate": 1.1198606599802986e-05, "loss": 0.0109, "step": 238090 }, { "epoch": 151.75270873167622, "grad_norm": 0.27803534269332886, "learning_rate": 1.11951306984446e-05, "loss": 0.0084, "step": 238100 }, { "epoch": 151.75908221797323, "grad_norm": 0.1946473866701126, "learning_rate": 1.1191655268602003e-05, "loss": 0.0108, "step": 238110 }, { "epoch": 151.76545570427024, "grad_norm": 0.17243921756744385, "learning_rate": 1.1188180310317458e-05, "loss": 0.0037, "step": 238120 }, { "epoch": 151.77182919056725, "grad_norm": 0.2625432312488556, "learning_rate": 1.118470582363314e-05, "loss": 0.0061, "step": 238130 }, { "epoch": 151.77820267686425, "grad_norm": 0.3290955722332001, "learning_rate": 1.1181231808591308e-05, "loss": 0.0044, "step": 238140 }, { "epoch": 151.78457616316126, "grad_norm": 0.16127069294452667, "learning_rate": 1.1177758265234146e-05, "loss": 0.006, "step": 238150 }, { "epoch": 151.79094964945824, "grad_norm": 0.3174325227737427, "learning_rate": 1.1174285193603906e-05, "loss": 0.0062, "step": 238160 }, { "epoch": 151.79732313575525, "grad_norm": 0.36680760979652405, "learning_rate": 1.1170812593742724e-05, "loss": 0.007, "step": 238170 }, { "epoch": 151.80369662205226, "grad_norm": 0.2588964104652405, "learning_rate": 1.116734046569285e-05, "loss": 0.0073, "step": 238180 }, { "epoch": 151.81007010834927, "grad_norm": 0.17771689593791962, "learning_rate": 1.1163868809496453e-05, "loss": 0.0064, "step": 238190 }, { "epoch": 151.81644359464627, "grad_norm": 0.2471923679113388, "learning_rate": 1.1160397625195717e-05, "loss": 0.0044, "step": 238200 }, { "epoch": 151.82281708094328, "grad_norm": 0.1557663530111313, "learning_rate": 1.1156926912832805e-05, "loss": 0.0059, "step": 238210 }, { "epoch": 151.8291905672403, "grad_norm": 0.3874107003211975, "learning_rate": 1.115345667244992e-05, "loss": 0.0063, "step": 238220 }, { "epoch": 151.8355640535373, "grad_norm": 0.17551377415657043, "learning_rate": 1.114998690408921e-05, "loss": 0.0052, "step": 238230 }, { "epoch": 151.84193753983428, "grad_norm": 0.12351218611001968, "learning_rate": 1.1146517607792844e-05, "loss": 0.0182, "step": 238240 }, { "epoch": 151.84831102613128, "grad_norm": 0.3795202672481537, "learning_rate": 1.114304878360296e-05, "loss": 0.0079, "step": 238250 }, { "epoch": 151.8546845124283, "grad_norm": 0.3907632529735565, "learning_rate": 1.1139580431561746e-05, "loss": 0.005, "step": 238260 }, { "epoch": 151.8610579987253, "grad_norm": 0.3264823257923126, "learning_rate": 1.1136112551711291e-05, "loss": 0.014, "step": 238270 }, { "epoch": 151.8674314850223, "grad_norm": 0.3757678270339966, "learning_rate": 1.1132645144093778e-05, "loss": 0.0069, "step": 238280 }, { "epoch": 151.87380497131932, "grad_norm": 0.1297888457775116, "learning_rate": 1.1129178208751307e-05, "loss": 0.0037, "step": 238290 }, { "epoch": 151.88017845761632, "grad_norm": 0.3975546956062317, "learning_rate": 1.1125711745726053e-05, "loss": 0.0116, "step": 238300 }, { "epoch": 151.88655194391333, "grad_norm": 0.09949151426553726, "learning_rate": 1.1122245755060073e-05, "loss": 0.0048, "step": 238310 }, { "epoch": 151.8929254302103, "grad_norm": 0.23479799926280975, "learning_rate": 1.111878023679553e-05, "loss": 0.0069, "step": 238320 }, { "epoch": 151.89929891650732, "grad_norm": 0.23076403141021729, "learning_rate": 1.1115315190974513e-05, "loss": 0.0065, "step": 238330 }, { "epoch": 151.90567240280433, "grad_norm": 0.16236372292041779, "learning_rate": 1.1111850617639137e-05, "loss": 0.0076, "step": 238340 }, { "epoch": 151.91204588910134, "grad_norm": 0.09478513896465302, "learning_rate": 1.1108386516831471e-05, "loss": 0.0052, "step": 238350 }, { "epoch": 151.91841937539834, "grad_norm": 0.3709292411804199, "learning_rate": 1.1104922888593643e-05, "loss": 0.0058, "step": 238360 }, { "epoch": 151.92479286169535, "grad_norm": 0.08321953564882278, "learning_rate": 1.1101459732967728e-05, "loss": 0.0062, "step": 238370 }, { "epoch": 151.93116634799236, "grad_norm": 0.43853774666786194, "learning_rate": 1.1097997049995806e-05, "loss": 0.004, "step": 238380 }, { "epoch": 151.93753983428937, "grad_norm": 0.24642184376716614, "learning_rate": 1.1094534839719934e-05, "loss": 0.0066, "step": 238390 }, { "epoch": 151.94391332058635, "grad_norm": 0.31854188442230225, "learning_rate": 1.109107310218222e-05, "loss": 0.0077, "step": 238400 }, { "epoch": 151.95028680688335, "grad_norm": 0.12839831411838531, "learning_rate": 1.1087611837424684e-05, "loss": 0.0064, "step": 238410 }, { "epoch": 151.95666029318036, "grad_norm": 0.22247222065925598, "learning_rate": 1.108415104548941e-05, "loss": 0.0067, "step": 238420 }, { "epoch": 151.96303377947737, "grad_norm": 0.2525670528411865, "learning_rate": 1.108069072641843e-05, "loss": 0.0038, "step": 238430 }, { "epoch": 151.96940726577438, "grad_norm": 0.17947496473789215, "learning_rate": 1.1077230880253835e-05, "loss": 0.0118, "step": 238440 }, { "epoch": 151.9757807520714, "grad_norm": 0.13216546177864075, "learning_rate": 1.1073771507037606e-05, "loss": 0.0064, "step": 238450 }, { "epoch": 151.9821542383684, "grad_norm": 0.1895257830619812, "learning_rate": 1.1070312606811817e-05, "loss": 0.0098, "step": 238460 }, { "epoch": 151.9885277246654, "grad_norm": 0.30190810561180115, "learning_rate": 1.1066854179618474e-05, "loss": 0.0042, "step": 238470 }, { "epoch": 151.99490121096238, "grad_norm": 0.18898946046829224, "learning_rate": 1.106339622549964e-05, "loss": 0.0148, "step": 238480 }, { "epoch": 152.0012746972594, "grad_norm": 0.4677102565765381, "learning_rate": 1.1059938744497272e-05, "loss": 0.0219, "step": 238490 }, { "epoch": 152.0076481835564, "grad_norm": 0.07938452810049057, "learning_rate": 1.1056481736653429e-05, "loss": 0.0037, "step": 238500 }, { "epoch": 152.0140216698534, "grad_norm": 0.3835509717464447, "learning_rate": 1.1053025202010103e-05, "loss": 0.0064, "step": 238510 }, { "epoch": 152.0203951561504, "grad_norm": 0.13830892741680145, "learning_rate": 1.1049569140609296e-05, "loss": 0.0045, "step": 238520 }, { "epoch": 152.02676864244742, "grad_norm": 0.0798812136054039, "learning_rate": 1.104611355249298e-05, "loss": 0.0035, "step": 238530 }, { "epoch": 152.03314212874443, "grad_norm": 0.12053446471691132, "learning_rate": 1.104265843770318e-05, "loss": 0.0094, "step": 238540 }, { "epoch": 152.03951561504144, "grad_norm": 0.09270553290843964, "learning_rate": 1.1039203796281861e-05, "loss": 0.0055, "step": 238550 }, { "epoch": 152.04588910133845, "grad_norm": 0.313518226146698, "learning_rate": 1.1035749628271002e-05, "loss": 0.0067, "step": 238560 }, { "epoch": 152.05226258763543, "grad_norm": 0.09681947529315948, "learning_rate": 1.103229593371256e-05, "loss": 0.0042, "step": 238570 }, { "epoch": 152.05863607393243, "grad_norm": 0.30112218856811523, "learning_rate": 1.1028842712648524e-05, "loss": 0.0066, "step": 238580 }, { "epoch": 152.06500956022944, "grad_norm": 0.24537093937397003, "learning_rate": 1.1025389965120847e-05, "loss": 0.004, "step": 238590 }, { "epoch": 152.07138304652645, "grad_norm": 0.4949844777584076, "learning_rate": 1.102193769117148e-05, "loss": 0.0043, "step": 238600 }, { "epoch": 152.07775653282346, "grad_norm": 0.08150278776884079, "learning_rate": 1.1018485890842356e-05, "loss": 0.0049, "step": 238610 }, { "epoch": 152.08413001912047, "grad_norm": 0.42641693353652954, "learning_rate": 1.1015034564175448e-05, "loss": 0.0191, "step": 238620 }, { "epoch": 152.09050350541747, "grad_norm": 0.2525986433029175, "learning_rate": 1.1011583711212674e-05, "loss": 0.0056, "step": 238630 }, { "epoch": 152.09687699171448, "grad_norm": 0.139072448015213, "learning_rate": 1.1008133331995974e-05, "loss": 0.0083, "step": 238640 }, { "epoch": 152.10325047801146, "grad_norm": 0.2118319422006607, "learning_rate": 1.100468342656727e-05, "loss": 0.0057, "step": 238650 }, { "epoch": 152.10962396430847, "grad_norm": 0.19130852818489075, "learning_rate": 1.1001233994968464e-05, "loss": 0.0035, "step": 238660 }, { "epoch": 152.11599745060548, "grad_norm": 0.5381157398223877, "learning_rate": 1.0997785037241498e-05, "loss": 0.0053, "step": 238670 }, { "epoch": 152.12237093690248, "grad_norm": 0.4782530963420868, "learning_rate": 1.0994336553428269e-05, "loss": 0.0075, "step": 238680 }, { "epoch": 152.1287444231995, "grad_norm": 0.1258234828710556, "learning_rate": 1.0990888543570677e-05, "loss": 0.0082, "step": 238690 }, { "epoch": 152.1351179094965, "grad_norm": 0.45518815517425537, "learning_rate": 1.0987441007710608e-05, "loss": 0.0066, "step": 238700 }, { "epoch": 152.1414913957935, "grad_norm": 0.4333152770996094, "learning_rate": 1.098399394588998e-05, "loss": 0.0042, "step": 238710 }, { "epoch": 152.14786488209052, "grad_norm": 0.08245629072189331, "learning_rate": 1.0980547358150661e-05, "loss": 0.011, "step": 238720 }, { "epoch": 152.1542383683875, "grad_norm": 0.552575409412384, "learning_rate": 1.0977101244534532e-05, "loss": 0.0117, "step": 238730 }, { "epoch": 152.1606118546845, "grad_norm": 0.34569039940834045, "learning_rate": 1.0973655605083449e-05, "loss": 0.005, "step": 238740 }, { "epoch": 152.1669853409815, "grad_norm": 0.11402948945760727, "learning_rate": 1.0970210439839318e-05, "loss": 0.0182, "step": 238750 }, { "epoch": 152.17335882727852, "grad_norm": 0.1053142100572586, "learning_rate": 1.0966765748843977e-05, "loss": 0.0023, "step": 238760 }, { "epoch": 152.17973231357553, "grad_norm": 0.18403871357440948, "learning_rate": 1.0963321532139287e-05, "loss": 0.0044, "step": 238770 }, { "epoch": 152.18610579987254, "grad_norm": 0.45705321431159973, "learning_rate": 1.0959877789767081e-05, "loss": 0.0095, "step": 238780 }, { "epoch": 152.19247928616954, "grad_norm": 0.1445131003856659, "learning_rate": 1.0956434521769231e-05, "loss": 0.0106, "step": 238790 }, { "epoch": 152.19885277246655, "grad_norm": 0.3490517735481262, "learning_rate": 1.0952991728187573e-05, "loss": 0.0052, "step": 238800 }, { "epoch": 152.20522625876353, "grad_norm": 0.23777265846729279, "learning_rate": 1.0949549409063925e-05, "loss": 0.0041, "step": 238810 }, { "epoch": 152.21159974506054, "grad_norm": 0.3912496864795685, "learning_rate": 1.094610756444011e-05, "loss": 0.0081, "step": 238820 }, { "epoch": 152.21797323135755, "grad_norm": 0.20541217923164368, "learning_rate": 1.0942666194357993e-05, "loss": 0.0109, "step": 238830 }, { "epoch": 152.22434671765456, "grad_norm": 0.32479628920555115, "learning_rate": 1.0939225298859323e-05, "loss": 0.0049, "step": 238840 }, { "epoch": 152.23072020395156, "grad_norm": 0.32706141471862793, "learning_rate": 1.0935784877985966e-05, "loss": 0.0076, "step": 238850 }, { "epoch": 152.23709369024857, "grad_norm": 0.1248515173792839, "learning_rate": 1.0932344931779687e-05, "loss": 0.0074, "step": 238860 }, { "epoch": 152.24346717654558, "grad_norm": 0.1821872889995575, "learning_rate": 1.092890546028233e-05, "loss": 0.0048, "step": 238870 }, { "epoch": 152.2498406628426, "grad_norm": 0.1834280639886856, "learning_rate": 1.0925466463535632e-05, "loss": 0.0081, "step": 238880 }, { "epoch": 152.25621414913957, "grad_norm": 0.27649715542793274, "learning_rate": 1.0922027941581432e-05, "loss": 0.0081, "step": 238890 }, { "epoch": 152.26258763543657, "grad_norm": 0.49059733748435974, "learning_rate": 1.0918589894461479e-05, "loss": 0.0074, "step": 238900 }, { "epoch": 152.26896112173358, "grad_norm": 0.291648268699646, "learning_rate": 1.091515232221756e-05, "loss": 0.0065, "step": 238910 }, { "epoch": 152.2753346080306, "grad_norm": 0.5350047945976257, "learning_rate": 1.0911715224891427e-05, "loss": 0.0055, "step": 238920 }, { "epoch": 152.2817080943276, "grad_norm": 0.16614589095115662, "learning_rate": 1.090827860252488e-05, "loss": 0.0087, "step": 238930 }, { "epoch": 152.2880815806246, "grad_norm": 0.47804561257362366, "learning_rate": 1.0904842455159647e-05, "loss": 0.0067, "step": 238940 }, { "epoch": 152.2944550669216, "grad_norm": 0.1987387090921402, "learning_rate": 1.0901406782837498e-05, "loss": 0.0052, "step": 238950 }, { "epoch": 152.30082855321862, "grad_norm": 0.4983787536621094, "learning_rate": 1.0897971585600154e-05, "loss": 0.0185, "step": 238960 }, { "epoch": 152.3072020395156, "grad_norm": 0.2504577040672302, "learning_rate": 1.0894536863489402e-05, "loss": 0.0083, "step": 238970 }, { "epoch": 152.3135755258126, "grad_norm": 0.22734545171260834, "learning_rate": 1.0891102616546922e-05, "loss": 0.0093, "step": 238980 }, { "epoch": 152.31994901210962, "grad_norm": 0.14679847657680511, "learning_rate": 1.0887668844814487e-05, "loss": 0.0044, "step": 238990 }, { "epoch": 152.32632249840663, "grad_norm": 0.158009335398674, "learning_rate": 1.0884235548333782e-05, "loss": 0.0056, "step": 239000 }, { "epoch": 152.33269598470363, "grad_norm": 0.09565307945013046, "learning_rate": 1.0880802727146584e-05, "loss": 0.0033, "step": 239010 }, { "epoch": 152.33906947100064, "grad_norm": 0.07512016594409943, "learning_rate": 1.0877370381294532e-05, "loss": 0.0028, "step": 239020 }, { "epoch": 152.34544295729765, "grad_norm": 0.16046959161758423, "learning_rate": 1.0873938510819382e-05, "loss": 0.0113, "step": 239030 }, { "epoch": 152.35181644359466, "grad_norm": 0.30236902832984924, "learning_rate": 1.0870507115762823e-05, "loss": 0.0053, "step": 239040 }, { "epoch": 152.35818992989164, "grad_norm": 0.20737190544605255, "learning_rate": 1.0867076196166548e-05, "loss": 0.0109, "step": 239050 }, { "epoch": 152.36456341618864, "grad_norm": 0.10032489150762558, "learning_rate": 1.0863645752072226e-05, "loss": 0.0071, "step": 239060 }, { "epoch": 152.37093690248565, "grad_norm": 0.25039300322532654, "learning_rate": 1.0860215783521577e-05, "loss": 0.0046, "step": 239070 }, { "epoch": 152.37731038878266, "grad_norm": 0.3127688765525818, "learning_rate": 1.085678629055626e-05, "loss": 0.0073, "step": 239080 }, { "epoch": 152.38368387507967, "grad_norm": 0.36033934354782104, "learning_rate": 1.0853357273217945e-05, "loss": 0.004, "step": 239090 }, { "epoch": 152.39005736137668, "grad_norm": 0.1274247020483017, "learning_rate": 1.0849928731548286e-05, "loss": 0.005, "step": 239100 }, { "epoch": 152.39643084767368, "grad_norm": 0.16821791231632233, "learning_rate": 1.0846500665588976e-05, "loss": 0.0044, "step": 239110 }, { "epoch": 152.4028043339707, "grad_norm": 0.12384303659200668, "learning_rate": 1.0843073075381644e-05, "loss": 0.0086, "step": 239120 }, { "epoch": 152.4091778202677, "grad_norm": 0.1714136004447937, "learning_rate": 1.0839645960967947e-05, "loss": 0.007, "step": 239130 }, { "epoch": 152.41555130656468, "grad_norm": 0.2269960641860962, "learning_rate": 1.0836219322389507e-05, "loss": 0.0124, "step": 239140 }, { "epoch": 152.4219247928617, "grad_norm": 0.45587271451950073, "learning_rate": 1.0832793159688015e-05, "loss": 0.0111, "step": 239150 }, { "epoch": 152.4282982791587, "grad_norm": 0.12278030067682266, "learning_rate": 1.0829367472905033e-05, "loss": 0.0045, "step": 239160 }, { "epoch": 152.4346717654557, "grad_norm": 0.11479590088129044, "learning_rate": 1.0825942262082234e-05, "loss": 0.0096, "step": 239170 }, { "epoch": 152.4410452517527, "grad_norm": 0.11814448982477188, "learning_rate": 1.0822517527261206e-05, "loss": 0.0056, "step": 239180 }, { "epoch": 152.44741873804972, "grad_norm": 0.21606558561325073, "learning_rate": 1.081909326848361e-05, "loss": 0.0088, "step": 239190 }, { "epoch": 152.45379222434673, "grad_norm": 0.09937078505754471, "learning_rate": 1.0815669485790997e-05, "loss": 0.0064, "step": 239200 }, { "epoch": 152.46016571064374, "grad_norm": 0.18784503638744354, "learning_rate": 1.0812246179225004e-05, "loss": 0.006, "step": 239210 }, { "epoch": 152.46653919694072, "grad_norm": 0.2463483214378357, "learning_rate": 1.080882334882723e-05, "loss": 0.0055, "step": 239220 }, { "epoch": 152.47291268323772, "grad_norm": 0.13640819489955902, "learning_rate": 1.0805400994639248e-05, "loss": 0.007, "step": 239230 }, { "epoch": 152.47928616953473, "grad_norm": 0.19041012227535248, "learning_rate": 1.0801979116702638e-05, "loss": 0.0046, "step": 239240 }, { "epoch": 152.48565965583174, "grad_norm": 0.568773627281189, "learning_rate": 1.0798557715059e-05, "loss": 0.011, "step": 239250 }, { "epoch": 152.49203314212875, "grad_norm": 0.11240726709365845, "learning_rate": 1.0795136789749904e-05, "loss": 0.0028, "step": 239260 }, { "epoch": 152.49840662842576, "grad_norm": 0.4726616144180298, "learning_rate": 1.0791716340816909e-05, "loss": 0.0098, "step": 239270 }, { "epoch": 152.50478011472276, "grad_norm": 0.3172299563884735, "learning_rate": 1.0788296368301564e-05, "loss": 0.0072, "step": 239280 }, { "epoch": 152.51115360101977, "grad_norm": 0.177865669131279, "learning_rate": 1.0784876872245452e-05, "loss": 0.0068, "step": 239290 }, { "epoch": 152.51752708731675, "grad_norm": 0.7384693622589111, "learning_rate": 1.0781457852690118e-05, "loss": 0.0118, "step": 239300 }, { "epoch": 152.52390057361376, "grad_norm": 0.2409648448228836, "learning_rate": 1.0778039309677096e-05, "loss": 0.0035, "step": 239310 }, { "epoch": 152.53027405991077, "grad_norm": 0.3215387165546417, "learning_rate": 1.0774621243247912e-05, "loss": 0.0044, "step": 239320 }, { "epoch": 152.53664754620777, "grad_norm": 0.07636915147304535, "learning_rate": 1.0771203653444134e-05, "loss": 0.0034, "step": 239330 }, { "epoch": 152.54302103250478, "grad_norm": 0.19294455647468567, "learning_rate": 1.0767786540307267e-05, "loss": 0.0062, "step": 239340 }, { "epoch": 152.5493945188018, "grad_norm": 0.16090402007102966, "learning_rate": 1.0764369903878834e-05, "loss": 0.0157, "step": 239350 }, { "epoch": 152.5557680050988, "grad_norm": 0.19415034353733063, "learning_rate": 1.0760953744200336e-05, "loss": 0.006, "step": 239360 }, { "epoch": 152.5621414913958, "grad_norm": 0.16888238489627838, "learning_rate": 1.0757538061313316e-05, "loss": 0.0033, "step": 239370 }, { "epoch": 152.5685149776928, "grad_norm": 0.47184422612190247, "learning_rate": 1.0754122855259258e-05, "loss": 0.0175, "step": 239380 }, { "epoch": 152.5748884639898, "grad_norm": 0.3765632212162018, "learning_rate": 1.0750708126079662e-05, "loss": 0.0107, "step": 239390 }, { "epoch": 152.5812619502868, "grad_norm": 0.2229687124490738, "learning_rate": 1.0747293873816016e-05, "loss": 0.0084, "step": 239400 }, { "epoch": 152.5876354365838, "grad_norm": 0.2053460031747818, "learning_rate": 1.07438800985098e-05, "loss": 0.006, "step": 239410 }, { "epoch": 152.59400892288082, "grad_norm": 0.6400021910667419, "learning_rate": 1.0740466800202515e-05, "loss": 0.0102, "step": 239420 }, { "epoch": 152.60038240917783, "grad_norm": 0.09022586792707443, "learning_rate": 1.0737053978935629e-05, "loss": 0.009, "step": 239430 }, { "epoch": 152.60675589547483, "grad_norm": 0.18781672418117523, "learning_rate": 1.0733641634750608e-05, "loss": 0.0038, "step": 239440 }, { "epoch": 152.61312938177184, "grad_norm": 0.24339288473129272, "learning_rate": 1.0730229767688898e-05, "loss": 0.0189, "step": 239450 }, { "epoch": 152.61950286806882, "grad_norm": 0.14460016787052155, "learning_rate": 1.0726818377791986e-05, "loss": 0.0048, "step": 239460 }, { "epoch": 152.62587635436583, "grad_norm": 0.14634519815444946, "learning_rate": 1.0723407465101314e-05, "loss": 0.0038, "step": 239470 }, { "epoch": 152.63224984066284, "grad_norm": 0.27079200744628906, "learning_rate": 1.0719997029658324e-05, "loss": 0.0087, "step": 239480 }, { "epoch": 152.63862332695984, "grad_norm": 0.20055142045021057, "learning_rate": 1.0716587071504442e-05, "loss": 0.0082, "step": 239490 }, { "epoch": 152.64499681325685, "grad_norm": 0.20050761103630066, "learning_rate": 1.0713177590681134e-05, "loss": 0.0081, "step": 239500 }, { "epoch": 152.65137029955386, "grad_norm": 0.14533106982707977, "learning_rate": 1.0709768587229807e-05, "loss": 0.003, "step": 239510 }, { "epoch": 152.65774378585087, "grad_norm": 0.14546574652194977, "learning_rate": 1.070636006119189e-05, "loss": 0.006, "step": 239520 }, { "epoch": 152.66411727214788, "grad_norm": 0.20141355693340302, "learning_rate": 1.070295201260878e-05, "loss": 0.0138, "step": 239530 }, { "epoch": 152.67049075844486, "grad_norm": 0.15076199173927307, "learning_rate": 1.0699544441521936e-05, "loss": 0.0049, "step": 239540 }, { "epoch": 152.67686424474186, "grad_norm": 0.2795514464378357, "learning_rate": 1.0696137347972707e-05, "loss": 0.0063, "step": 239550 }, { "epoch": 152.68323773103887, "grad_norm": 0.5130665898323059, "learning_rate": 1.0692730732002531e-05, "loss": 0.0112, "step": 239560 }, { "epoch": 152.68961121733588, "grad_norm": 0.17310839891433716, "learning_rate": 1.0689324593652772e-05, "loss": 0.0051, "step": 239570 }, { "epoch": 152.6959847036329, "grad_norm": 0.39640891551971436, "learning_rate": 1.0685918932964866e-05, "loss": 0.0074, "step": 239580 }, { "epoch": 152.7023581899299, "grad_norm": 0.23803569376468658, "learning_rate": 1.0682513749980133e-05, "loss": 0.0045, "step": 239590 }, { "epoch": 152.7087316762269, "grad_norm": 0.10886967927217484, "learning_rate": 1.0679109044739993e-05, "loss": 0.0044, "step": 239600 }, { "epoch": 152.7151051625239, "grad_norm": 0.3008319139480591, "learning_rate": 1.0675704817285809e-05, "loss": 0.0036, "step": 239610 }, { "epoch": 152.7214786488209, "grad_norm": 0.11812824010848999, "learning_rate": 1.0672301067658936e-05, "loss": 0.0076, "step": 239620 }, { "epoch": 152.7278521351179, "grad_norm": 0.14809662103652954, "learning_rate": 1.0668897795900723e-05, "loss": 0.0075, "step": 239630 }, { "epoch": 152.7342256214149, "grad_norm": 0.24317528307437897, "learning_rate": 1.0665495002052545e-05, "loss": 0.0156, "step": 239640 }, { "epoch": 152.74059910771192, "grad_norm": 0.421475887298584, "learning_rate": 1.066209268615575e-05, "loss": 0.0109, "step": 239650 }, { "epoch": 152.74697259400892, "grad_norm": 0.14228232204914093, "learning_rate": 1.0658690848251663e-05, "loss": 0.0088, "step": 239660 }, { "epoch": 152.75334608030593, "grad_norm": 0.09659285098314285, "learning_rate": 1.0655289488381614e-05, "loss": 0.0051, "step": 239670 }, { "epoch": 152.75971956660294, "grad_norm": 0.5493375658988953, "learning_rate": 1.0651888606586962e-05, "loss": 0.0066, "step": 239680 }, { "epoch": 152.76609305289995, "grad_norm": 0.12509386241436005, "learning_rate": 1.0648488202909013e-05, "loss": 0.0049, "step": 239690 }, { "epoch": 152.77246653919693, "grad_norm": 0.12238802015781403, "learning_rate": 1.0645088277389087e-05, "loss": 0.0046, "step": 239700 }, { "epoch": 152.77884002549393, "grad_norm": 0.1353289783000946, "learning_rate": 1.064168883006848e-05, "loss": 0.0042, "step": 239710 }, { "epoch": 152.78521351179094, "grad_norm": 0.19434022903442383, "learning_rate": 1.0638289860988543e-05, "loss": 0.0038, "step": 239720 }, { "epoch": 152.79158699808795, "grad_norm": 0.7702406048774719, "learning_rate": 1.063489137019052e-05, "loss": 0.0044, "step": 239730 }, { "epoch": 152.79796048438496, "grad_norm": 1.2415075302124023, "learning_rate": 1.0631493357715745e-05, "loss": 0.0082, "step": 239740 }, { "epoch": 152.80433397068197, "grad_norm": 0.11248290538787842, "learning_rate": 1.0628095823605483e-05, "loss": 0.0057, "step": 239750 }, { "epoch": 152.81070745697897, "grad_norm": 0.2688446044921875, "learning_rate": 1.0624698767901059e-05, "loss": 0.0035, "step": 239760 }, { "epoch": 152.81708094327598, "grad_norm": 0.3143787384033203, "learning_rate": 1.0621302190643694e-05, "loss": 0.0108, "step": 239770 }, { "epoch": 152.823454429573, "grad_norm": 0.13668130338191986, "learning_rate": 1.0617906091874703e-05, "loss": 0.0067, "step": 239780 }, { "epoch": 152.82982791586997, "grad_norm": 0.11751358211040497, "learning_rate": 1.061451047163533e-05, "loss": 0.0079, "step": 239790 }, { "epoch": 152.83620140216698, "grad_norm": 0.17384761571884155, "learning_rate": 1.0611115329966848e-05, "loss": 0.0037, "step": 239800 }, { "epoch": 152.842574888464, "grad_norm": 0.2499346286058426, "learning_rate": 1.0607720666910481e-05, "loss": 0.0031, "step": 239810 }, { "epoch": 152.848948374761, "grad_norm": 0.08709793537855148, "learning_rate": 1.0604326482507521e-05, "loss": 0.009, "step": 239820 }, { "epoch": 152.855321861058, "grad_norm": 0.3283195197582245, "learning_rate": 1.0600932776799189e-05, "loss": 0.0054, "step": 239830 }, { "epoch": 152.861695347355, "grad_norm": 0.08864174038171768, "learning_rate": 1.0597539549826724e-05, "loss": 0.0052, "step": 239840 }, { "epoch": 152.86806883365202, "grad_norm": 0.23591069877147675, "learning_rate": 1.059414680163134e-05, "loss": 0.005, "step": 239850 }, { "epoch": 152.87444231994903, "grad_norm": 0.17220915853977203, "learning_rate": 1.0590754532254304e-05, "loss": 0.0049, "step": 239860 }, { "epoch": 152.880815806246, "grad_norm": 0.16818006336688995, "learning_rate": 1.0587362741736783e-05, "loss": 0.0059, "step": 239870 }, { "epoch": 152.887189292543, "grad_norm": 0.302287757396698, "learning_rate": 1.0583971430120027e-05, "loss": 0.0059, "step": 239880 }, { "epoch": 152.89356277884002, "grad_norm": 0.10401882976293564, "learning_rate": 1.0580580597445221e-05, "loss": 0.0078, "step": 239890 }, { "epoch": 152.89993626513703, "grad_norm": 0.31304648518562317, "learning_rate": 1.0577190243753609e-05, "loss": 0.0165, "step": 239900 }, { "epoch": 152.90630975143404, "grad_norm": 0.26843589544296265, "learning_rate": 1.0573800369086328e-05, "loss": 0.0057, "step": 239910 }, { "epoch": 152.91268323773104, "grad_norm": 0.11181370168924332, "learning_rate": 1.0570410973484607e-05, "loss": 0.0032, "step": 239920 }, { "epoch": 152.91905672402805, "grad_norm": 0.2303232103586197, "learning_rate": 1.0567022056989623e-05, "loss": 0.0066, "step": 239930 }, { "epoch": 152.92543021032506, "grad_norm": 0.16739989817142487, "learning_rate": 1.0563633619642555e-05, "loss": 0.0052, "step": 239940 }, { "epoch": 152.93180369662204, "grad_norm": 0.11466255784034729, "learning_rate": 1.0560245661484558e-05, "loss": 0.0057, "step": 239950 }, { "epoch": 152.93817718291905, "grad_norm": 0.1784355491399765, "learning_rate": 1.0556858182556822e-05, "loss": 0.0068, "step": 239960 }, { "epoch": 152.94455066921606, "grad_norm": 0.20975004136562347, "learning_rate": 1.0553471182900504e-05, "loss": 0.0067, "step": 239970 }, { "epoch": 152.95092415551306, "grad_norm": 0.18366611003875732, "learning_rate": 1.0550084662556753e-05, "loss": 0.0051, "step": 239980 }, { "epoch": 152.95729764181007, "grad_norm": 0.08926151692867279, "learning_rate": 1.0546698621566708e-05, "loss": 0.0045, "step": 239990 }, { "epoch": 152.96367112810708, "grad_norm": 0.18475541472434998, "learning_rate": 1.0543313059971539e-05, "loss": 0.0043, "step": 240000 }, { "epoch": 152.9700446144041, "grad_norm": 0.34088414907455444, "learning_rate": 1.0539927977812369e-05, "loss": 0.0061, "step": 240010 }, { "epoch": 152.9764181007011, "grad_norm": 0.24515719711780548, "learning_rate": 1.053654337513033e-05, "loss": 0.0062, "step": 240020 }, { "epoch": 152.98279158699808, "grad_norm": 0.12750893831253052, "learning_rate": 1.0533159251966534e-05, "loss": 0.0047, "step": 240030 }, { "epoch": 152.98916507329508, "grad_norm": 0.25646457076072693, "learning_rate": 1.052977560836213e-05, "loss": 0.0048, "step": 240040 }, { "epoch": 152.9955385595921, "grad_norm": 0.11671438813209534, "learning_rate": 1.0526392444358224e-05, "loss": 0.0048, "step": 240050 }, { "epoch": 153.0019120458891, "grad_norm": 0.14454783499240875, "learning_rate": 1.0523009759995911e-05, "loss": 0.0051, "step": 240060 }, { "epoch": 153.0082855321861, "grad_norm": 0.06528137624263763, "learning_rate": 1.0519627555316292e-05, "loss": 0.0056, "step": 240070 }, { "epoch": 153.01465901848312, "grad_norm": 0.14195141196250916, "learning_rate": 1.0516245830360489e-05, "loss": 0.0064, "step": 240080 }, { "epoch": 153.02103250478012, "grad_norm": 0.32820719480514526, "learning_rate": 1.0512864585169575e-05, "loss": 0.0048, "step": 240090 }, { "epoch": 153.02740599107713, "grad_norm": 0.1828809678554535, "learning_rate": 1.0509483819784638e-05, "loss": 0.0081, "step": 240100 }, { "epoch": 153.0337794773741, "grad_norm": 0.12064191699028015, "learning_rate": 1.0506103534246759e-05, "loss": 0.0067, "step": 240110 }, { "epoch": 153.04015296367112, "grad_norm": 0.147539421916008, "learning_rate": 1.0502723728596997e-05, "loss": 0.0091, "step": 240120 }, { "epoch": 153.04652644996813, "grad_norm": 0.14696452021598816, "learning_rate": 1.0499344402876449e-05, "loss": 0.0139, "step": 240130 }, { "epoch": 153.05289993626513, "grad_norm": 0.23236083984375, "learning_rate": 1.0495965557126158e-05, "loss": 0.0087, "step": 240140 }, { "epoch": 153.05927342256214, "grad_norm": 0.15420351922512054, "learning_rate": 1.0492587191387187e-05, "loss": 0.0037, "step": 240150 }, { "epoch": 153.06564690885915, "grad_norm": 0.3660881817340851, "learning_rate": 1.0489209305700565e-05, "loss": 0.0113, "step": 240160 }, { "epoch": 153.07202039515616, "grad_norm": 0.5205349326133728, "learning_rate": 1.0485831900107368e-05, "loss": 0.0138, "step": 240170 }, { "epoch": 153.07839388145317, "grad_norm": 0.280001163482666, "learning_rate": 1.0482454974648626e-05, "loss": 0.0078, "step": 240180 }, { "epoch": 153.08476736775015, "grad_norm": 0.20147192478179932, "learning_rate": 1.047907852936536e-05, "loss": 0.0037, "step": 240190 }, { "epoch": 153.09114085404715, "grad_norm": 0.2087157517671585, "learning_rate": 1.0475702564298595e-05, "loss": 0.0085, "step": 240200 }, { "epoch": 153.09751434034416, "grad_norm": 0.34242263436317444, "learning_rate": 1.0472327079489375e-05, "loss": 0.0051, "step": 240210 }, { "epoch": 153.10388782664117, "grad_norm": 0.5774280428886414, "learning_rate": 1.0468952074978699e-05, "loss": 0.0085, "step": 240220 }, { "epoch": 153.11026131293818, "grad_norm": 0.1764027327299118, "learning_rate": 1.0465577550807581e-05, "loss": 0.0082, "step": 240230 }, { "epoch": 153.1166347992352, "grad_norm": 0.09546209126710892, "learning_rate": 1.0462203507017005e-05, "loss": 0.0023, "step": 240240 }, { "epoch": 153.1230082855322, "grad_norm": 0.11358706653118134, "learning_rate": 1.0458829943648019e-05, "loss": 0.0164, "step": 240250 }, { "epoch": 153.1293817718292, "grad_norm": 0.11143676936626434, "learning_rate": 1.0455456860741547e-05, "loss": 0.0055, "step": 240260 }, { "epoch": 153.13575525812618, "grad_norm": 0.2231888473033905, "learning_rate": 1.0452084258338629e-05, "loss": 0.0067, "step": 240270 }, { "epoch": 153.1421287444232, "grad_norm": 0.15816543996334076, "learning_rate": 1.0448712136480215e-05, "loss": 0.0042, "step": 240280 }, { "epoch": 153.1485022307202, "grad_norm": 0.1625976860523224, "learning_rate": 1.0445340495207317e-05, "loss": 0.0032, "step": 240290 }, { "epoch": 153.1548757170172, "grad_norm": 0.25558221340179443, "learning_rate": 1.0441969334560848e-05, "loss": 0.0144, "step": 240300 }, { "epoch": 153.1612492033142, "grad_norm": 0.5774677395820618, "learning_rate": 1.0438598654581817e-05, "loss": 0.0105, "step": 240310 }, { "epoch": 153.16762268961122, "grad_norm": 0.13606274127960205, "learning_rate": 1.0435228455311153e-05, "loss": 0.0092, "step": 240320 }, { "epoch": 153.17399617590823, "grad_norm": 0.4411952495574951, "learning_rate": 1.043185873678984e-05, "loss": 0.0081, "step": 240330 }, { "epoch": 153.18036966220524, "grad_norm": 0.2007512003183365, "learning_rate": 1.0428489499058775e-05, "loss": 0.0098, "step": 240340 }, { "epoch": 153.18674314850222, "grad_norm": 0.07616712898015976, "learning_rate": 1.0425120742158944e-05, "loss": 0.0076, "step": 240350 }, { "epoch": 153.19311663479922, "grad_norm": 0.3087804615497589, "learning_rate": 1.0421752466131256e-05, "loss": 0.0026, "step": 240360 }, { "epoch": 153.19949012109623, "grad_norm": 0.11723849922418594, "learning_rate": 1.0418384671016646e-05, "loss": 0.0047, "step": 240370 }, { "epoch": 153.20586360739324, "grad_norm": 0.1482083648443222, "learning_rate": 1.0415017356856017e-05, "loss": 0.0076, "step": 240380 }, { "epoch": 153.21223709369025, "grad_norm": 0.4511992633342743, "learning_rate": 1.0411650523690314e-05, "loss": 0.009, "step": 240390 }, { "epoch": 153.21861057998726, "grad_norm": 0.2744091749191284, "learning_rate": 1.040828417156044e-05, "loss": 0.0107, "step": 240400 }, { "epoch": 153.22498406628426, "grad_norm": 0.166437566280365, "learning_rate": 1.0404918300507288e-05, "loss": 0.0072, "step": 240410 }, { "epoch": 153.23135755258127, "grad_norm": 0.17098428308963776, "learning_rate": 1.0401552910571754e-05, "loss": 0.0028, "step": 240420 }, { "epoch": 153.23773103887828, "grad_norm": 0.2710188031196594, "learning_rate": 1.0398188001794768e-05, "loss": 0.0046, "step": 240430 }, { "epoch": 153.24410452517526, "grad_norm": 0.9622645378112793, "learning_rate": 1.0394823574217156e-05, "loss": 0.0081, "step": 240440 }, { "epoch": 153.25047801147227, "grad_norm": 0.5853045582771301, "learning_rate": 1.0391459627879847e-05, "loss": 0.0087, "step": 240450 }, { "epoch": 153.25685149776928, "grad_norm": 0.37261250615119934, "learning_rate": 1.0388096162823685e-05, "loss": 0.0072, "step": 240460 }, { "epoch": 153.26322498406628, "grad_norm": 0.3108242154121399, "learning_rate": 1.0384733179089585e-05, "loss": 0.0052, "step": 240470 }, { "epoch": 153.2695984703633, "grad_norm": 0.2694622576236725, "learning_rate": 1.038137067671835e-05, "loss": 0.0052, "step": 240480 }, { "epoch": 153.2759719566603, "grad_norm": 0.1434251070022583, "learning_rate": 1.0378008655750881e-05, "loss": 0.007, "step": 240490 }, { "epoch": 153.2823454429573, "grad_norm": 0.3930404782295227, "learning_rate": 1.0374647116228015e-05, "loss": 0.0071, "step": 240500 }, { "epoch": 153.28871892925432, "grad_norm": 0.19251401722431183, "learning_rate": 1.03712860581906e-05, "loss": 0.0046, "step": 240510 }, { "epoch": 153.2950924155513, "grad_norm": 0.17753320932388306, "learning_rate": 1.0367925481679464e-05, "loss": 0.0043, "step": 240520 }, { "epoch": 153.3014659018483, "grad_norm": 0.1329333782196045, "learning_rate": 1.0364565386735464e-05, "loss": 0.0084, "step": 240530 }, { "epoch": 153.3078393881453, "grad_norm": 0.5003127455711365, "learning_rate": 1.0361205773399418e-05, "loss": 0.0091, "step": 240540 }, { "epoch": 153.31421287444232, "grad_norm": 0.13218054175376892, "learning_rate": 1.0357846641712144e-05, "loss": 0.0085, "step": 240550 }, { "epoch": 153.32058636073933, "grad_norm": 0.12087587267160416, "learning_rate": 1.0354487991714452e-05, "loss": 0.005, "step": 240560 }, { "epoch": 153.32695984703633, "grad_norm": 0.1874856948852539, "learning_rate": 1.035112982344717e-05, "loss": 0.0068, "step": 240570 }, { "epoch": 153.33333333333334, "grad_norm": 0.34561944007873535, "learning_rate": 1.03477721369511e-05, "loss": 0.009, "step": 240580 }, { "epoch": 153.33970681963035, "grad_norm": 0.42082563042640686, "learning_rate": 1.0344414932267037e-05, "loss": 0.0117, "step": 240590 }, { "epoch": 153.34608030592733, "grad_norm": 0.369461327791214, "learning_rate": 1.034105820943576e-05, "loss": 0.0108, "step": 240600 }, { "epoch": 153.35245379222434, "grad_norm": 0.1669924557209015, "learning_rate": 1.0337701968498097e-05, "loss": 0.0136, "step": 240610 }, { "epoch": 153.35882727852135, "grad_norm": 0.6600402593612671, "learning_rate": 1.0334346209494771e-05, "loss": 0.0098, "step": 240620 }, { "epoch": 153.36520076481835, "grad_norm": 0.1831793338060379, "learning_rate": 1.0330990932466605e-05, "loss": 0.0038, "step": 240630 }, { "epoch": 153.37157425111536, "grad_norm": 0.24359260499477386, "learning_rate": 1.0327636137454344e-05, "loss": 0.0063, "step": 240640 }, { "epoch": 153.37794773741237, "grad_norm": 0.14939342439174652, "learning_rate": 1.0324281824498783e-05, "loss": 0.0141, "step": 240650 }, { "epoch": 153.38432122370938, "grad_norm": 0.12690356373786926, "learning_rate": 1.0320927993640634e-05, "loss": 0.0078, "step": 240660 }, { "epoch": 153.3906947100064, "grad_norm": 0.113094761967659, "learning_rate": 1.0317574644920686e-05, "loss": 0.0033, "step": 240670 }, { "epoch": 153.39706819630337, "grad_norm": 0.36972081661224365, "learning_rate": 1.0314221778379668e-05, "loss": 0.0153, "step": 240680 }, { "epoch": 153.40344168260037, "grad_norm": 0.1992025375366211, "learning_rate": 1.0310869394058331e-05, "loss": 0.0103, "step": 240690 }, { "epoch": 153.40981516889738, "grad_norm": 0.1382005214691162, "learning_rate": 1.0307517491997387e-05, "loss": 0.0047, "step": 240700 }, { "epoch": 153.4161886551944, "grad_norm": 0.15124450623989105, "learning_rate": 1.0304166072237593e-05, "loss": 0.0056, "step": 240710 }, { "epoch": 153.4225621414914, "grad_norm": 0.2732418179512024, "learning_rate": 1.0300815134819663e-05, "loss": 0.0083, "step": 240720 }, { "epoch": 153.4289356277884, "grad_norm": 0.5005336403846741, "learning_rate": 1.0297464679784314e-05, "loss": 0.0045, "step": 240730 }, { "epoch": 153.4353091140854, "grad_norm": 0.11522285640239716, "learning_rate": 1.0294114707172236e-05, "loss": 0.0068, "step": 240740 }, { "epoch": 153.44168260038242, "grad_norm": 0.1932888776063919, "learning_rate": 1.0290765217024167e-05, "loss": 0.0054, "step": 240750 }, { "epoch": 153.4480560866794, "grad_norm": 0.20597101747989655, "learning_rate": 1.0287416209380796e-05, "loss": 0.0219, "step": 240760 }, { "epoch": 153.4544295729764, "grad_norm": 0.30281391739845276, "learning_rate": 1.028406768428281e-05, "loss": 0.0051, "step": 240770 }, { "epoch": 153.46080305927342, "grad_norm": 0.1319674551486969, "learning_rate": 1.0280719641770886e-05, "loss": 0.0138, "step": 240780 }, { "epoch": 153.46717654557042, "grad_norm": 0.12069246917963028, "learning_rate": 1.0277372081885734e-05, "loss": 0.01, "step": 240790 }, { "epoch": 153.47355003186743, "grad_norm": 0.20646683871746063, "learning_rate": 1.0274025004668014e-05, "loss": 0.0061, "step": 240800 }, { "epoch": 153.47992351816444, "grad_norm": 0.1362047642469406, "learning_rate": 1.0270678410158401e-05, "loss": 0.0195, "step": 240810 }, { "epoch": 153.48629700446145, "grad_norm": 0.10900107026100159, "learning_rate": 1.0267332298397552e-05, "loss": 0.0089, "step": 240820 }, { "epoch": 153.49267049075846, "grad_norm": 0.1765272617340088, "learning_rate": 1.0263986669426124e-05, "loss": 0.0035, "step": 240830 }, { "epoch": 153.49904397705544, "grad_norm": 0.08345885574817657, "learning_rate": 1.0260641523284781e-05, "loss": 0.0055, "step": 240840 }, { "epoch": 153.50541746335244, "grad_norm": 0.19152872264385223, "learning_rate": 1.0257296860014166e-05, "loss": 0.0028, "step": 240850 }, { "epoch": 153.51179094964945, "grad_norm": 0.35748064517974854, "learning_rate": 1.0253952679654916e-05, "loss": 0.0201, "step": 240860 }, { "epoch": 153.51816443594646, "grad_norm": 0.1723119169473648, "learning_rate": 1.0250608982247656e-05, "loss": 0.0054, "step": 240870 }, { "epoch": 153.52453792224347, "grad_norm": 0.311603844165802, "learning_rate": 1.0247265767833036e-05, "loss": 0.008, "step": 240880 }, { "epoch": 153.53091140854048, "grad_norm": 0.09813393652439117, "learning_rate": 1.0243923036451675e-05, "loss": 0.0054, "step": 240890 }, { "epoch": 153.53728489483748, "grad_norm": 0.1547156274318695, "learning_rate": 1.0240580788144183e-05, "loss": 0.0057, "step": 240900 }, { "epoch": 153.5436583811345, "grad_norm": 0.3459078371524811, "learning_rate": 1.0237239022951156e-05, "loss": 0.005, "step": 240910 }, { "epoch": 153.55003186743147, "grad_norm": 0.11313871294260025, "learning_rate": 1.0233897740913235e-05, "loss": 0.0069, "step": 240920 }, { "epoch": 153.55640535372848, "grad_norm": 0.0772741511464119, "learning_rate": 1.0230556942071e-05, "loss": 0.0081, "step": 240930 }, { "epoch": 153.5627788400255, "grad_norm": 0.26498711109161377, "learning_rate": 1.0227216626465047e-05, "loss": 0.0042, "step": 240940 }, { "epoch": 153.5691523263225, "grad_norm": 0.17372864484786987, "learning_rate": 1.0223876794135956e-05, "loss": 0.0028, "step": 240950 }, { "epoch": 153.5755258126195, "grad_norm": 0.1560124009847641, "learning_rate": 1.0220537445124322e-05, "loss": 0.0055, "step": 240960 }, { "epoch": 153.5818992989165, "grad_norm": 0.3914065361022949, "learning_rate": 1.0217198579470722e-05, "loss": 0.0081, "step": 240970 }, { "epoch": 153.58827278521352, "grad_norm": 0.17217126488685608, "learning_rate": 1.0213860197215719e-05, "loss": 0.004, "step": 240980 }, { "epoch": 153.59464627151053, "grad_norm": 0.3448353111743927, "learning_rate": 1.021052229839986e-05, "loss": 0.0064, "step": 240990 }, { "epoch": 153.60101975780753, "grad_norm": 0.4798355996608734, "learning_rate": 1.0207184883063753e-05, "loss": 0.0102, "step": 241000 }, { "epoch": 153.60739324410451, "grad_norm": 0.0741119310259819, "learning_rate": 1.0203847951247897e-05, "loss": 0.0047, "step": 241010 }, { "epoch": 153.61376673040152, "grad_norm": 0.5032899975776672, "learning_rate": 1.020051150299287e-05, "loss": 0.0095, "step": 241020 }, { "epoch": 153.62014021669853, "grad_norm": 0.3894112706184387, "learning_rate": 1.0197175538339199e-05, "loss": 0.0055, "step": 241030 }, { "epoch": 153.62651370299554, "grad_norm": 0.10783681273460388, "learning_rate": 1.0193840057327447e-05, "loss": 0.0068, "step": 241040 }, { "epoch": 153.63288718929255, "grad_norm": 0.18386705219745636, "learning_rate": 1.0190505059998095e-05, "loss": 0.0106, "step": 241050 }, { "epoch": 153.63926067558955, "grad_norm": 0.15916840732097626, "learning_rate": 1.0187170546391705e-05, "loss": 0.0099, "step": 241060 }, { "epoch": 153.64563416188656, "grad_norm": 0.29758384823799133, "learning_rate": 1.0183836516548783e-05, "loss": 0.0048, "step": 241070 }, { "epoch": 153.65200764818357, "grad_norm": 0.39541128277778625, "learning_rate": 1.0180502970509842e-05, "loss": 0.013, "step": 241080 }, { "epoch": 153.65838113448055, "grad_norm": 0.14136402308940887, "learning_rate": 1.0177169908315376e-05, "loss": 0.0069, "step": 241090 }, { "epoch": 153.66475462077756, "grad_norm": 0.12444766610860825, "learning_rate": 1.0173837330005903e-05, "loss": 0.0058, "step": 241100 }, { "epoch": 153.67112810707457, "grad_norm": 0.07633435726165771, "learning_rate": 1.0170505235621908e-05, "loss": 0.0083, "step": 241110 }, { "epoch": 153.67750159337157, "grad_norm": 0.11175760626792908, "learning_rate": 1.016717362520388e-05, "loss": 0.0042, "step": 241120 }, { "epoch": 153.68387507966858, "grad_norm": 0.2120114415884018, "learning_rate": 1.0163842498792286e-05, "loss": 0.0062, "step": 241130 }, { "epoch": 153.6902485659656, "grad_norm": 0.3695414066314697, "learning_rate": 1.0160511856427651e-05, "loss": 0.0069, "step": 241140 }, { "epoch": 153.6966220522626, "grad_norm": 0.12056731432676315, "learning_rate": 1.015718169815038e-05, "loss": 0.0056, "step": 241150 }, { "epoch": 153.7029955385596, "grad_norm": 0.4663669466972351, "learning_rate": 1.0153852024000982e-05, "loss": 0.0081, "step": 241160 }, { "epoch": 153.70936902485658, "grad_norm": 0.224840447306633, "learning_rate": 1.0150522834019888e-05, "loss": 0.0055, "step": 241170 }, { "epoch": 153.7157425111536, "grad_norm": 0.3189881443977356, "learning_rate": 1.0147194128247595e-05, "loss": 0.0058, "step": 241180 }, { "epoch": 153.7221159974506, "grad_norm": 0.10657411813735962, "learning_rate": 1.0143865906724493e-05, "loss": 0.0107, "step": 241190 }, { "epoch": 153.7284894837476, "grad_norm": 0.2258238047361374, "learning_rate": 1.0140538169491065e-05, "loss": 0.0054, "step": 241200 }, { "epoch": 153.73486297004462, "grad_norm": 0.16274511814117432, "learning_rate": 1.013721091658772e-05, "loss": 0.0063, "step": 241210 }, { "epoch": 153.74123645634162, "grad_norm": 0.16322028636932373, "learning_rate": 1.0133884148054923e-05, "loss": 0.0056, "step": 241220 }, { "epoch": 153.74760994263863, "grad_norm": 0.35676756501197815, "learning_rate": 1.0130557863933048e-05, "loss": 0.0046, "step": 241230 }, { "epoch": 153.75398342893564, "grad_norm": 0.4929918944835663, "learning_rate": 1.0127232064262554e-05, "loss": 0.0102, "step": 241240 }, { "epoch": 153.76035691523262, "grad_norm": 0.3338485658168793, "learning_rate": 1.0123906749083834e-05, "loss": 0.0161, "step": 241250 }, { "epoch": 153.76673040152963, "grad_norm": 0.43881142139434814, "learning_rate": 1.0120581918437293e-05, "loss": 0.0114, "step": 241260 }, { "epoch": 153.77310388782664, "grad_norm": 0.21356521546840668, "learning_rate": 1.0117257572363325e-05, "loss": 0.006, "step": 241270 }, { "epoch": 153.77947737412364, "grad_norm": 0.19332140684127808, "learning_rate": 1.0113933710902345e-05, "loss": 0.0033, "step": 241280 }, { "epoch": 153.78585086042065, "grad_norm": 0.10612089931964874, "learning_rate": 1.0110610334094728e-05, "loss": 0.0093, "step": 241290 }, { "epoch": 153.79222434671766, "grad_norm": 0.1017225906252861, "learning_rate": 1.0107287441980857e-05, "loss": 0.0036, "step": 241300 }, { "epoch": 153.79859783301467, "grad_norm": 0.1539483666419983, "learning_rate": 1.0103965034601099e-05, "loss": 0.0053, "step": 241310 }, { "epoch": 153.80497131931168, "grad_norm": 0.10035558044910431, "learning_rate": 1.0100643111995855e-05, "loss": 0.0038, "step": 241320 }, { "epoch": 153.81134480560866, "grad_norm": 0.1915368139743805, "learning_rate": 1.0097321674205446e-05, "loss": 0.0052, "step": 241330 }, { "epoch": 153.81771829190566, "grad_norm": 0.16917827725410461, "learning_rate": 1.0094000721270264e-05, "loss": 0.0026, "step": 241340 }, { "epoch": 153.82409177820267, "grad_norm": 0.2218681126832962, "learning_rate": 1.0090680253230639e-05, "loss": 0.0024, "step": 241350 }, { "epoch": 153.83046526449968, "grad_norm": 0.4187619984149933, "learning_rate": 1.0087360270126955e-05, "loss": 0.0142, "step": 241360 }, { "epoch": 153.8368387507967, "grad_norm": 0.18404459953308105, "learning_rate": 1.0084040771999504e-05, "loss": 0.0038, "step": 241370 }, { "epoch": 153.8432122370937, "grad_norm": 0.1978590190410614, "learning_rate": 1.0080721758888656e-05, "loss": 0.0031, "step": 241380 }, { "epoch": 153.8495857233907, "grad_norm": 0.0931597352027893, "learning_rate": 1.0077403230834726e-05, "loss": 0.0042, "step": 241390 }, { "epoch": 153.8559592096877, "grad_norm": 0.1831062287092209, "learning_rate": 1.0074085187878047e-05, "loss": 0.0086, "step": 241400 }, { "epoch": 153.8623326959847, "grad_norm": 0.23328380286693573, "learning_rate": 1.0070767630058908e-05, "loss": 0.0063, "step": 241410 }, { "epoch": 153.8687061822817, "grad_norm": 0.3396683633327484, "learning_rate": 1.0067450557417657e-05, "loss": 0.0086, "step": 241420 }, { "epoch": 153.8750796685787, "grad_norm": 0.10786926746368408, "learning_rate": 1.0064133969994588e-05, "loss": 0.0027, "step": 241430 }, { "epoch": 153.88145315487571, "grad_norm": 0.1351919323205948, "learning_rate": 1.0060817867829991e-05, "loss": 0.0082, "step": 241440 }, { "epoch": 153.88782664117272, "grad_norm": 0.11853715032339096, "learning_rate": 1.0057502250964157e-05, "loss": 0.0068, "step": 241450 }, { "epoch": 153.89420012746973, "grad_norm": 0.26797258853912354, "learning_rate": 1.0054187119437397e-05, "loss": 0.0057, "step": 241460 }, { "epoch": 153.90057361376674, "grad_norm": 0.1729242354631424, "learning_rate": 1.0050872473289973e-05, "loss": 0.0096, "step": 241470 }, { "epoch": 153.90694710006375, "grad_norm": 0.08828655630350113, "learning_rate": 1.0047558312562171e-05, "loss": 0.0039, "step": 241480 }, { "epoch": 153.91332058636073, "grad_norm": 0.1528148651123047, "learning_rate": 1.0044244637294242e-05, "loss": 0.004, "step": 241490 }, { "epoch": 153.91969407265773, "grad_norm": 0.1189984455704689, "learning_rate": 1.004093144752648e-05, "loss": 0.0045, "step": 241500 }, { "epoch": 153.92606755895474, "grad_norm": 0.11748594790697098, "learning_rate": 1.0037618743299126e-05, "loss": 0.0054, "step": 241510 }, { "epoch": 153.93244104525175, "grad_norm": 0.08435408771038055, "learning_rate": 1.0034306524652437e-05, "loss": 0.0041, "step": 241520 }, { "epoch": 153.93881453154876, "grad_norm": 0.24117830395698547, "learning_rate": 1.0030994791626646e-05, "loss": 0.0078, "step": 241530 }, { "epoch": 153.94518801784577, "grad_norm": 0.29262417554855347, "learning_rate": 1.0027683544262017e-05, "loss": 0.0038, "step": 241540 }, { "epoch": 153.95156150414277, "grad_norm": 0.3208114802837372, "learning_rate": 1.0024372782598773e-05, "loss": 0.0058, "step": 241550 }, { "epoch": 153.95793499043978, "grad_norm": 0.2885794937610626, "learning_rate": 1.0021062506677143e-05, "loss": 0.0036, "step": 241560 }, { "epoch": 153.96430847673676, "grad_norm": 0.08062794804573059, "learning_rate": 1.0017752716537348e-05, "loss": 0.0036, "step": 241570 }, { "epoch": 153.97068196303377, "grad_norm": 0.11147583276033401, "learning_rate": 1.0014443412219593e-05, "loss": 0.0033, "step": 241580 }, { "epoch": 153.97705544933078, "grad_norm": 0.08459415286779404, "learning_rate": 1.001113459376412e-05, "loss": 0.0078, "step": 241590 }, { "epoch": 153.98342893562778, "grad_norm": 0.12759022414684296, "learning_rate": 1.0007826261211117e-05, "loss": 0.0127, "step": 241600 }, { "epoch": 153.9898024219248, "grad_norm": 0.08561825752258301, "learning_rate": 1.0004518414600782e-05, "loss": 0.0048, "step": 241610 }, { "epoch": 153.9961759082218, "grad_norm": 0.07710101455450058, "learning_rate": 1.0001211053973297e-05, "loss": 0.009, "step": 241620 }, { "epoch": 154.0025493945188, "grad_norm": 0.24412931501865387, "learning_rate": 9.997904179368878e-06, "loss": 0.005, "step": 241630 }, { "epoch": 154.00892288081582, "grad_norm": 0.0943923071026802, "learning_rate": 9.994597790827692e-06, "loss": 0.0057, "step": 241640 }, { "epoch": 154.01529636711282, "grad_norm": 0.0905103087425232, "learning_rate": 9.99129188838991e-06, "loss": 0.0032, "step": 241650 }, { "epoch": 154.0216698534098, "grad_norm": 0.2746722996234894, "learning_rate": 9.987986472095695e-06, "loss": 0.005, "step": 241660 }, { "epoch": 154.0280433397068, "grad_norm": 0.3655979335308075, "learning_rate": 9.984681541985235e-06, "loss": 0.0035, "step": 241670 }, { "epoch": 154.03441682600382, "grad_norm": 0.4572738707065582, "learning_rate": 9.981377098098676e-06, "loss": 0.0063, "step": 241680 }, { "epoch": 154.04079031230083, "grad_norm": 0.2224317193031311, "learning_rate": 9.978073140476169e-06, "loss": 0.0043, "step": 241690 }, { "epoch": 154.04716379859784, "grad_norm": 0.21383512020111084, "learning_rate": 9.974769669157846e-06, "loss": 0.0102, "step": 241700 }, { "epoch": 154.05353728489484, "grad_norm": 0.4679703116416931, "learning_rate": 9.971466684183888e-06, "loss": 0.0084, "step": 241710 }, { "epoch": 154.05991077119185, "grad_norm": 0.2616625130176544, "learning_rate": 9.96816418559438e-06, "loss": 0.0039, "step": 241720 }, { "epoch": 154.06628425748886, "grad_norm": 0.10306661576032639, "learning_rate": 9.964862173429484e-06, "loss": 0.0059, "step": 241730 }, { "epoch": 154.07265774378584, "grad_norm": 0.1973017305135727, "learning_rate": 9.961560647729302e-06, "loss": 0.0184, "step": 241740 }, { "epoch": 154.07903123008285, "grad_norm": 0.2946559488773346, "learning_rate": 9.958259608533982e-06, "loss": 0.0049, "step": 241750 }, { "epoch": 154.08540471637986, "grad_norm": 0.18592298030853271, "learning_rate": 9.95495905588359e-06, "loss": 0.0032, "step": 241760 }, { "epoch": 154.09177820267686, "grad_norm": 0.6475895643234253, "learning_rate": 9.951658989818264e-06, "loss": 0.0072, "step": 241770 }, { "epoch": 154.09815168897387, "grad_norm": 0.31212666630744934, "learning_rate": 9.948359410378084e-06, "loss": 0.0047, "step": 241780 }, { "epoch": 154.10452517527088, "grad_norm": 0.11561473459005356, "learning_rate": 9.945060317603177e-06, "loss": 0.0032, "step": 241790 }, { "epoch": 154.1108986615679, "grad_norm": 0.07584702968597412, "learning_rate": 9.941761711533577e-06, "loss": 0.0097, "step": 241800 }, { "epoch": 154.1172721478649, "grad_norm": 0.17127187550067902, "learning_rate": 9.938463592209408e-06, "loss": 0.0033, "step": 241810 }, { "epoch": 154.12364563416187, "grad_norm": 0.11616022139787674, "learning_rate": 9.935165959670733e-06, "loss": 0.0057, "step": 241820 }, { "epoch": 154.13001912045888, "grad_norm": 0.08590978384017944, "learning_rate": 9.931868813957618e-06, "loss": 0.0045, "step": 241830 }, { "epoch": 154.1363926067559, "grad_norm": 0.21696536242961884, "learning_rate": 9.92857215511011e-06, "loss": 0.003, "step": 241840 }, { "epoch": 154.1427660930529, "grad_norm": 0.08729539066553116, "learning_rate": 9.925275983168302e-06, "loss": 0.0095, "step": 241850 }, { "epoch": 154.1491395793499, "grad_norm": 0.10615651309490204, "learning_rate": 9.921980298172222e-06, "loss": 0.0041, "step": 241860 }, { "epoch": 154.15551306564691, "grad_norm": 0.2392064929008484, "learning_rate": 9.918685100161928e-06, "loss": 0.01, "step": 241870 }, { "epoch": 154.16188655194392, "grad_norm": 0.7082544565200806, "learning_rate": 9.915390389177437e-06, "loss": 0.0128, "step": 241880 }, { "epoch": 154.16826003824093, "grad_norm": 0.5741721391677856, "learning_rate": 9.912096165258827e-06, "loss": 0.0094, "step": 241890 }, { "epoch": 154.1746335245379, "grad_norm": 0.15962913632392883, "learning_rate": 9.908802428446073e-06, "loss": 0.0047, "step": 241900 }, { "epoch": 154.18100701083492, "grad_norm": 0.13236474990844727, "learning_rate": 9.905509178779233e-06, "loss": 0.0064, "step": 241910 }, { "epoch": 154.18738049713193, "grad_norm": 0.5458288788795471, "learning_rate": 9.902216416298293e-06, "loss": 0.0083, "step": 241920 }, { "epoch": 154.19375398342893, "grad_norm": 0.26745253801345825, "learning_rate": 9.898924141043313e-06, "loss": 0.0186, "step": 241930 }, { "epoch": 154.20012746972594, "grad_norm": 0.2943972051143646, "learning_rate": 9.895632353054245e-06, "loss": 0.0063, "step": 241940 }, { "epoch": 154.20650095602295, "grad_norm": 0.30074596405029297, "learning_rate": 9.892341052371118e-06, "loss": 0.0102, "step": 241950 }, { "epoch": 154.21287444231996, "grad_norm": 0.09977021813392639, "learning_rate": 9.889050239033914e-06, "loss": 0.0075, "step": 241960 }, { "epoch": 154.21924792861697, "grad_norm": 0.1770649403333664, "learning_rate": 9.88575991308262e-06, "loss": 0.0075, "step": 241970 }, { "epoch": 154.22562141491395, "grad_norm": 0.09287931025028229, "learning_rate": 9.882470074557204e-06, "loss": 0.0047, "step": 241980 }, { "epoch": 154.23199490121095, "grad_norm": 0.17502005398273468, "learning_rate": 9.879180723497666e-06, "loss": 0.0125, "step": 241990 }, { "epoch": 154.23836838750796, "grad_norm": 0.13929209113121033, "learning_rate": 9.875891859943969e-06, "loss": 0.0086, "step": 242000 }, { "epoch": 154.24474187380497, "grad_norm": 0.14298170804977417, "learning_rate": 9.872603483936066e-06, "loss": 0.0082, "step": 242010 }, { "epoch": 154.25111536010198, "grad_norm": 0.20309478044509888, "learning_rate": 9.869315595513901e-06, "loss": 0.0062, "step": 242020 }, { "epoch": 154.25748884639899, "grad_norm": 0.42509669065475464, "learning_rate": 9.866028194717469e-06, "loss": 0.0122, "step": 242030 }, { "epoch": 154.263862332696, "grad_norm": 0.12828698754310608, "learning_rate": 9.862741281586662e-06, "loss": 0.0088, "step": 242040 }, { "epoch": 154.270235818993, "grad_norm": 0.44338375329971313, "learning_rate": 9.859454856161454e-06, "loss": 0.0064, "step": 242050 }, { "epoch": 154.27660930528998, "grad_norm": 0.28331777453422546, "learning_rate": 9.856168918481762e-06, "loss": 0.0041, "step": 242060 }, { "epoch": 154.282982791587, "grad_norm": 0.14399749040603638, "learning_rate": 9.852883468587543e-06, "loss": 0.003, "step": 242070 }, { "epoch": 154.289356277884, "grad_norm": 0.12671075761318207, "learning_rate": 9.849598506518665e-06, "loss": 0.003, "step": 242080 }, { "epoch": 154.295729764181, "grad_norm": 0.11156364530324936, "learning_rate": 9.84631403231509e-06, "loss": 0.0054, "step": 242090 }, { "epoch": 154.302103250478, "grad_norm": 0.17718571424484253, "learning_rate": 9.843030046016699e-06, "loss": 0.0099, "step": 242100 }, { "epoch": 154.30847673677502, "grad_norm": 0.5053768754005432, "learning_rate": 9.83974654766343e-06, "loss": 0.0048, "step": 242110 }, { "epoch": 154.31485022307203, "grad_norm": 0.2751085162162781, "learning_rate": 9.836463537295132e-06, "loss": 0.006, "step": 242120 }, { "epoch": 154.32122370936904, "grad_norm": 0.16059571504592896, "learning_rate": 9.833181014951742e-06, "loss": 0.0059, "step": 242130 }, { "epoch": 154.32759719566602, "grad_norm": 0.15677402913570404, "learning_rate": 9.829898980673118e-06, "loss": 0.0059, "step": 242140 }, { "epoch": 154.33397068196302, "grad_norm": 0.167769655585289, "learning_rate": 9.826617434499152e-06, "loss": 0.0073, "step": 242150 }, { "epoch": 154.34034416826003, "grad_norm": 0.19041359424591064, "learning_rate": 9.823336376469705e-06, "loss": 0.0055, "step": 242160 }, { "epoch": 154.34671765455704, "grad_norm": 0.1510789394378662, "learning_rate": 9.820055806624662e-06, "loss": 0.0058, "step": 242170 }, { "epoch": 154.35309114085405, "grad_norm": 0.30743733048439026, "learning_rate": 9.816775725003879e-06, "loss": 0.0122, "step": 242180 }, { "epoch": 154.35946462715106, "grad_norm": 0.11961514502763748, "learning_rate": 9.81349613164721e-06, "loss": 0.0235, "step": 242190 }, { "epoch": 154.36583811344806, "grad_norm": 0.15657801926136017, "learning_rate": 9.810217026594493e-06, "loss": 0.0093, "step": 242200 }, { "epoch": 154.37221159974507, "grad_norm": 0.46439000964164734, "learning_rate": 9.80693840988559e-06, "loss": 0.0066, "step": 242210 }, { "epoch": 154.37858508604205, "grad_norm": 0.12229196727275848, "learning_rate": 9.803660281560345e-06, "loss": 0.0038, "step": 242220 }, { "epoch": 154.38495857233906, "grad_norm": 0.5458371043205261, "learning_rate": 9.800382641658573e-06, "loss": 0.0067, "step": 242230 }, { "epoch": 154.39133205863607, "grad_norm": 0.3249112069606781, "learning_rate": 9.797105490220093e-06, "loss": 0.0116, "step": 242240 }, { "epoch": 154.39770554493307, "grad_norm": 0.2549234628677368, "learning_rate": 9.793828827284756e-06, "loss": 0.0052, "step": 242250 }, { "epoch": 154.40407903123008, "grad_norm": 0.5056923031806946, "learning_rate": 9.79055265289236e-06, "loss": 0.0121, "step": 242260 }, { "epoch": 154.4104525175271, "grad_norm": 0.10784626752138138, "learning_rate": 9.78727696708271e-06, "loss": 0.0053, "step": 242270 }, { "epoch": 154.4168260038241, "grad_norm": 0.1326882541179657, "learning_rate": 9.784001769895613e-06, "loss": 0.0044, "step": 242280 }, { "epoch": 154.4231994901211, "grad_norm": 0.25860244035720825, "learning_rate": 9.780727061370854e-06, "loss": 0.0053, "step": 242290 }, { "epoch": 154.42957297641811, "grad_norm": 0.14098750054836273, "learning_rate": 9.777452841548246e-06, "loss": 0.0034, "step": 242300 }, { "epoch": 154.4359464627151, "grad_norm": 0.09809365123510361, "learning_rate": 9.774179110467558e-06, "loss": 0.0095, "step": 242310 }, { "epoch": 154.4423199490121, "grad_norm": 0.10327955335378647, "learning_rate": 9.770905868168578e-06, "loss": 0.0039, "step": 242320 }, { "epoch": 154.4486934353091, "grad_norm": 0.3629206717014313, "learning_rate": 9.767633114691061e-06, "loss": 0.0074, "step": 242330 }, { "epoch": 154.45506692160612, "grad_norm": 0.4611507058143616, "learning_rate": 9.7643608500748e-06, "loss": 0.0056, "step": 242340 }, { "epoch": 154.46144040790313, "grad_norm": 0.18422849476337433, "learning_rate": 9.761089074359542e-06, "loss": 0.0031, "step": 242350 }, { "epoch": 154.46781389420013, "grad_norm": 0.15624728798866272, "learning_rate": 9.757817787585044e-06, "loss": 0.0047, "step": 242360 }, { "epoch": 154.47418738049714, "grad_norm": 0.37981748580932617, "learning_rate": 9.754546989791042e-06, "loss": 0.0057, "step": 242370 }, { "epoch": 154.48056086679415, "grad_norm": 0.1838986873626709, "learning_rate": 9.751276681017307e-06, "loss": 0.0034, "step": 242380 }, { "epoch": 154.48693435309113, "grad_norm": 0.21307939291000366, "learning_rate": 9.74800686130356e-06, "loss": 0.0106, "step": 242390 }, { "epoch": 154.49330783938814, "grad_norm": 0.1023409441113472, "learning_rate": 9.744737530689534e-06, "loss": 0.0043, "step": 242400 }, { "epoch": 154.49968132568515, "grad_norm": 0.36730676889419556, "learning_rate": 9.741468689214938e-06, "loss": 0.0079, "step": 242410 }, { "epoch": 154.50605481198215, "grad_norm": 0.16347381472587585, "learning_rate": 9.738200336919528e-06, "loss": 0.0132, "step": 242420 }, { "epoch": 154.51242829827916, "grad_norm": 0.17229872941970825, "learning_rate": 9.734932473842988e-06, "loss": 0.0032, "step": 242430 }, { "epoch": 154.51880178457617, "grad_norm": 0.1920505166053772, "learning_rate": 9.731665100025045e-06, "loss": 0.0038, "step": 242440 }, { "epoch": 154.52517527087318, "grad_norm": 0.11388380080461502, "learning_rate": 9.72839821550537e-06, "loss": 0.0063, "step": 242450 }, { "epoch": 154.53154875717019, "grad_norm": 0.07977231591939926, "learning_rate": 9.725131820323707e-06, "loss": 0.0153, "step": 242460 }, { "epoch": 154.53792224346716, "grad_norm": 0.2018757164478302, "learning_rate": 9.721865914519696e-06, "loss": 0.0053, "step": 242470 }, { "epoch": 154.54429572976417, "grad_norm": 0.10775857418775558, "learning_rate": 9.718600498133051e-06, "loss": 0.0054, "step": 242480 }, { "epoch": 154.55066921606118, "grad_norm": 0.26215434074401855, "learning_rate": 9.715335571203431e-06, "loss": 0.0052, "step": 242490 }, { "epoch": 154.5570427023582, "grad_norm": 0.8017998337745667, "learning_rate": 9.712071133770545e-06, "loss": 0.0062, "step": 242500 }, { "epoch": 154.5634161886552, "grad_norm": 0.28315800428390503, "learning_rate": 9.708807185874008e-06, "loss": 0.0095, "step": 242510 }, { "epoch": 154.5697896749522, "grad_norm": 0.08532358705997467, "learning_rate": 9.705543727553517e-06, "loss": 0.0099, "step": 242520 }, { "epoch": 154.5761631612492, "grad_norm": 0.14261431992053986, "learning_rate": 9.702280758848714e-06, "loss": 0.0054, "step": 242530 }, { "epoch": 154.58253664754622, "grad_norm": 0.4081152081489563, "learning_rate": 9.699018279799249e-06, "loss": 0.0091, "step": 242540 }, { "epoch": 154.5889101338432, "grad_norm": 0.11121001839637756, "learning_rate": 9.695756290444746e-06, "loss": 0.0042, "step": 242550 }, { "epoch": 154.5952836201402, "grad_norm": 0.2566186487674713, "learning_rate": 9.692494790824869e-06, "loss": 0.0054, "step": 242560 }, { "epoch": 154.60165710643722, "grad_norm": 0.1360797882080078, "learning_rate": 9.68923378097924e-06, "loss": 0.0044, "step": 242570 }, { "epoch": 154.60803059273422, "grad_norm": 0.2238219678401947, "learning_rate": 9.68597326094748e-06, "loss": 0.0066, "step": 242580 }, { "epoch": 154.61440407903123, "grad_norm": 0.11817653477191925, "learning_rate": 9.682713230769197e-06, "loss": 0.0059, "step": 242590 }, { "epoch": 154.62077756532824, "grad_norm": 0.324648916721344, "learning_rate": 9.679453690484036e-06, "loss": 0.0073, "step": 242600 }, { "epoch": 154.62715105162525, "grad_norm": 0.44713419675827026, "learning_rate": 9.676194640131558e-06, "loss": 0.0059, "step": 242610 }, { "epoch": 154.63352453792226, "grad_norm": 0.19850006699562073, "learning_rate": 9.6729360797514e-06, "loss": 0.0039, "step": 242620 }, { "epoch": 154.63989802421924, "grad_norm": 0.42648351192474365, "learning_rate": 9.669678009383126e-06, "loss": 0.0065, "step": 242630 }, { "epoch": 154.64627151051624, "grad_norm": 0.1859533041715622, "learning_rate": 9.666420429066375e-06, "loss": 0.0174, "step": 242640 }, { "epoch": 154.65264499681325, "grad_norm": 0.17698591947555542, "learning_rate": 9.66316333884067e-06, "loss": 0.0066, "step": 242650 }, { "epoch": 154.65901848311026, "grad_norm": 0.2053345888853073, "learning_rate": 9.659906738745623e-06, "loss": 0.0042, "step": 242660 }, { "epoch": 154.66539196940727, "grad_norm": 0.2952617406845093, "learning_rate": 9.656650628820785e-06, "loss": 0.0075, "step": 242670 }, { "epoch": 154.67176545570427, "grad_norm": 0.1290670931339264, "learning_rate": 9.65339500910576e-06, "loss": 0.0038, "step": 242680 }, { "epoch": 154.67813894200128, "grad_norm": 0.40497326850891113, "learning_rate": 9.650139879640052e-06, "loss": 0.0168, "step": 242690 }, { "epoch": 154.6845124282983, "grad_norm": 0.16478300094604492, "learning_rate": 9.646885240463255e-06, "loss": 0.0083, "step": 242700 }, { "epoch": 154.69088591459527, "grad_norm": 0.18697090446949005, "learning_rate": 9.643631091614902e-06, "loss": 0.0067, "step": 242710 }, { "epoch": 154.69725940089228, "grad_norm": 0.2461615949869156, "learning_rate": 9.640377433134535e-06, "loss": 0.0153, "step": 242720 }, { "epoch": 154.7036328871893, "grad_norm": 0.10474171489477158, "learning_rate": 9.637124265061675e-06, "loss": 0.0043, "step": 242730 }, { "epoch": 154.7100063734863, "grad_norm": 0.19452454149723053, "learning_rate": 9.633871587435877e-06, "loss": 0.0035, "step": 242740 }, { "epoch": 154.7163798597833, "grad_norm": 0.07827474176883698, "learning_rate": 9.630619400296647e-06, "loss": 0.0037, "step": 242750 }, { "epoch": 154.7227533460803, "grad_norm": 0.14668941497802734, "learning_rate": 9.62736770368351e-06, "loss": 0.0036, "step": 242760 }, { "epoch": 154.72912683237732, "grad_norm": 0.19085222482681274, "learning_rate": 9.62411649763596e-06, "loss": 0.0164, "step": 242770 }, { "epoch": 154.73550031867433, "grad_norm": 0.12338367849588394, "learning_rate": 9.620865782193545e-06, "loss": 0.0026, "step": 242780 }, { "epoch": 154.7418738049713, "grad_norm": 0.5111455917358398, "learning_rate": 9.617615557395705e-06, "loss": 0.0166, "step": 242790 }, { "epoch": 154.7482472912683, "grad_norm": 0.14109760522842407, "learning_rate": 9.614365823281974e-06, "loss": 0.007, "step": 242800 }, { "epoch": 154.75462077756532, "grad_norm": 0.16530925035476685, "learning_rate": 9.611116579891821e-06, "loss": 0.0072, "step": 242810 }, { "epoch": 154.76099426386233, "grad_norm": 0.4507604241371155, "learning_rate": 9.607867827264755e-06, "loss": 0.0036, "step": 242820 }, { "epoch": 154.76736775015934, "grad_norm": 0.09511212259531021, "learning_rate": 9.60461956544021e-06, "loss": 0.005, "step": 242830 }, { "epoch": 154.77374123645635, "grad_norm": 0.06280311942100525, "learning_rate": 9.601371794457687e-06, "loss": 0.0039, "step": 242840 }, { "epoch": 154.78011472275335, "grad_norm": 0.20959234237670898, "learning_rate": 9.598124514356643e-06, "loss": 0.0049, "step": 242850 }, { "epoch": 154.78648820905036, "grad_norm": 0.20954900979995728, "learning_rate": 9.594877725176527e-06, "loss": 0.0047, "step": 242860 }, { "epoch": 154.79286169534734, "grad_norm": 0.3813554048538208, "learning_rate": 9.591631426956787e-06, "loss": 0.0064, "step": 242870 }, { "epoch": 154.79923518164435, "grad_norm": 0.14507296681404114, "learning_rate": 9.588385619736889e-06, "loss": 0.0038, "step": 242880 }, { "epoch": 154.80560866794136, "grad_norm": 0.28951409459114075, "learning_rate": 9.585140303556261e-06, "loss": 0.0114, "step": 242890 }, { "epoch": 154.81198215423836, "grad_norm": 0.33591973781585693, "learning_rate": 9.581895478454333e-06, "loss": 0.006, "step": 242900 }, { "epoch": 154.81835564053537, "grad_norm": 0.18983739614486694, "learning_rate": 9.578651144470524e-06, "loss": 0.0083, "step": 242910 }, { "epoch": 154.82472912683238, "grad_norm": 0.9829161167144775, "learning_rate": 9.575407301644285e-06, "loss": 0.0086, "step": 242920 }, { "epoch": 154.8311026131294, "grad_norm": 0.15292054414749146, "learning_rate": 9.572163950015006e-06, "loss": 0.0088, "step": 242930 }, { "epoch": 154.8374760994264, "grad_norm": 0.12873475253582, "learning_rate": 9.568921089622113e-06, "loss": 0.0051, "step": 242940 }, { "epoch": 154.8438495857234, "grad_norm": 0.4354668855667114, "learning_rate": 9.565678720504984e-06, "loss": 0.0064, "step": 242950 }, { "epoch": 154.85022307202038, "grad_norm": 0.42021775245666504, "learning_rate": 9.562436842703054e-06, "loss": 0.0047, "step": 242960 }, { "epoch": 154.8565965583174, "grad_norm": 0.09983206540346146, "learning_rate": 9.559195456255687e-06, "loss": 0.0076, "step": 242970 }, { "epoch": 154.8629700446144, "grad_norm": 0.20811142027378082, "learning_rate": 9.555954561202284e-06, "loss": 0.0054, "step": 242980 }, { "epoch": 154.8693435309114, "grad_norm": 0.19539637863636017, "learning_rate": 9.552714157582205e-06, "loss": 0.0059, "step": 242990 }, { "epoch": 154.87571701720842, "grad_norm": 0.3995899558067322, "learning_rate": 9.549474245434853e-06, "loss": 0.0061, "step": 243000 }, { "epoch": 154.88209050350542, "grad_norm": 0.1769992709159851, "learning_rate": 9.546234824799577e-06, "loss": 0.0049, "step": 243010 }, { "epoch": 154.88846398980243, "grad_norm": 0.2350369393825531, "learning_rate": 9.54299589571574e-06, "loss": 0.0053, "step": 243020 }, { "epoch": 154.89483747609944, "grad_norm": 0.1474510133266449, "learning_rate": 9.539757458222704e-06, "loss": 0.0043, "step": 243030 }, { "epoch": 154.90121096239642, "grad_norm": 0.15973564982414246, "learning_rate": 9.536519512359804e-06, "loss": 0.0042, "step": 243040 }, { "epoch": 154.90758444869343, "grad_norm": 0.270302951335907, "learning_rate": 9.533282058166404e-06, "loss": 0.0038, "step": 243050 }, { "epoch": 154.91395793499044, "grad_norm": 0.29310479760169983, "learning_rate": 9.53004509568184e-06, "loss": 0.0043, "step": 243060 }, { "epoch": 154.92033142128744, "grad_norm": 0.11072203516960144, "learning_rate": 9.526808624945432e-06, "loss": 0.0044, "step": 243070 }, { "epoch": 154.92670490758445, "grad_norm": 0.17872169613838196, "learning_rate": 9.523572645996504e-06, "loss": 0.0034, "step": 243080 }, { "epoch": 154.93307839388146, "grad_norm": 0.33901503682136536, "learning_rate": 9.520337158874393e-06, "loss": 0.0063, "step": 243090 }, { "epoch": 154.93945188017847, "grad_norm": 0.1906493604183197, "learning_rate": 9.517102163618408e-06, "loss": 0.0122, "step": 243100 }, { "epoch": 154.94582536647547, "grad_norm": 0.18293160200119019, "learning_rate": 9.513867660267856e-06, "loss": 0.0059, "step": 243110 }, { "epoch": 154.95219885277245, "grad_norm": 0.3226928114891052, "learning_rate": 9.510633648862015e-06, "loss": 0.0075, "step": 243120 }, { "epoch": 154.95857233906946, "grad_norm": 0.5704344511032104, "learning_rate": 9.507400129440226e-06, "loss": 0.0062, "step": 243130 }, { "epoch": 154.96494582536647, "grad_norm": 0.493643581867218, "learning_rate": 9.504167102041755e-06, "loss": 0.0078, "step": 243140 }, { "epoch": 154.97131931166348, "grad_norm": 0.4622494876384735, "learning_rate": 9.500934566705883e-06, "loss": 0.0081, "step": 243150 }, { "epoch": 154.9776927979605, "grad_norm": 0.18122877180576324, "learning_rate": 9.497702523471885e-06, "loss": 0.0051, "step": 243160 }, { "epoch": 154.9840662842575, "grad_norm": 0.558531641960144, "learning_rate": 9.494470972379066e-06, "loss": 0.0068, "step": 243170 }, { "epoch": 154.9904397705545, "grad_norm": 0.23913337290287018, "learning_rate": 9.491239913466642e-06, "loss": 0.0078, "step": 243180 }, { "epoch": 154.9968132568515, "grad_norm": 0.1835276335477829, "learning_rate": 9.488009346773919e-06, "loss": 0.0041, "step": 243190 }, { "epoch": 155.0031867431485, "grad_norm": 0.19274553656578064, "learning_rate": 9.48477927234011e-06, "loss": 0.0087, "step": 243200 }, { "epoch": 155.0095602294455, "grad_norm": 0.36473339796066284, "learning_rate": 9.481549690204517e-06, "loss": 0.0046, "step": 243210 }, { "epoch": 155.0159337157425, "grad_norm": 0.08644402027130127, "learning_rate": 9.478320600406327e-06, "loss": 0.0056, "step": 243220 }, { "epoch": 155.0223072020395, "grad_norm": 0.16964907944202423, "learning_rate": 9.475092002984808e-06, "loss": 0.0134, "step": 243230 }, { "epoch": 155.02868068833652, "grad_norm": 0.3152647912502289, "learning_rate": 9.471863897979182e-06, "loss": 0.005, "step": 243240 }, { "epoch": 155.03505417463353, "grad_norm": 0.4468255043029785, "learning_rate": 9.468636285428678e-06, "loss": 0.0238, "step": 243250 }, { "epoch": 155.04142766093054, "grad_norm": 0.1271553635597229, "learning_rate": 9.465409165372496e-06, "loss": 0.0094, "step": 243260 }, { "epoch": 155.04780114722755, "grad_norm": 0.0854572132229805, "learning_rate": 9.462182537849878e-06, "loss": 0.0165, "step": 243270 }, { "epoch": 155.05417463352453, "grad_norm": 0.09477397054433823, "learning_rate": 9.45895640290002e-06, "loss": 0.0026, "step": 243280 }, { "epoch": 155.06054811982153, "grad_norm": 0.10931822657585144, "learning_rate": 9.455730760562115e-06, "loss": 0.0144, "step": 243290 }, { "epoch": 155.06692160611854, "grad_norm": 0.13749998807907104, "learning_rate": 9.45250561087535e-06, "loss": 0.005, "step": 243300 }, { "epoch": 155.07329509241555, "grad_norm": 0.30055221915245056, "learning_rate": 9.449280953878937e-06, "loss": 0.0048, "step": 243310 }, { "epoch": 155.07966857871256, "grad_norm": 0.11119557917118073, "learning_rate": 9.446056789612046e-06, "loss": 0.0201, "step": 243320 }, { "epoch": 155.08604206500956, "grad_norm": 0.36265555024147034, "learning_rate": 9.442833118113853e-06, "loss": 0.0047, "step": 243330 }, { "epoch": 155.09241555130657, "grad_norm": 0.1375674158334732, "learning_rate": 9.439609939423516e-06, "loss": 0.0047, "step": 243340 }, { "epoch": 155.09878903760358, "grad_norm": 0.18713250756263733, "learning_rate": 9.436387253580242e-06, "loss": 0.0056, "step": 243350 }, { "epoch": 155.10516252390056, "grad_norm": 0.32762014865875244, "learning_rate": 9.43316506062314e-06, "loss": 0.004, "step": 243360 }, { "epoch": 155.11153601019757, "grad_norm": 0.1254536509513855, "learning_rate": 9.429943360591392e-06, "loss": 0.0133, "step": 243370 }, { "epoch": 155.11790949649458, "grad_norm": 0.18184436857700348, "learning_rate": 9.42672215352412e-06, "loss": 0.007, "step": 243380 }, { "epoch": 155.12428298279158, "grad_norm": 0.30995315313339233, "learning_rate": 9.423501439460513e-06, "loss": 0.0038, "step": 243390 }, { "epoch": 155.1306564690886, "grad_norm": 0.2547857463359833, "learning_rate": 9.420281218439648e-06, "loss": 0.0102, "step": 243400 }, { "epoch": 155.1370299553856, "grad_norm": 0.20533812046051025, "learning_rate": 9.41706149050069e-06, "loss": 0.0061, "step": 243410 }, { "epoch": 155.1434034416826, "grad_norm": 0.08458761125802994, "learning_rate": 9.413842255682753e-06, "loss": 0.0052, "step": 243420 }, { "epoch": 155.14977692797962, "grad_norm": 0.22753632068634033, "learning_rate": 9.410623514024952e-06, "loss": 0.0072, "step": 243430 }, { "epoch": 155.1561504142766, "grad_norm": 0.1655774563550949, "learning_rate": 9.407405265566382e-06, "loss": 0.0097, "step": 243440 }, { "epoch": 155.1625239005736, "grad_norm": 0.18072469532489777, "learning_rate": 9.40418751034618e-06, "loss": 0.0041, "step": 243450 }, { "epoch": 155.1688973868706, "grad_norm": 0.2747846245765686, "learning_rate": 9.400970248403428e-06, "loss": 0.0062, "step": 243460 }, { "epoch": 155.17527087316762, "grad_norm": 0.14877568185329437, "learning_rate": 9.397753479777222e-06, "loss": 0.0039, "step": 243470 }, { "epoch": 155.18164435946463, "grad_norm": 0.3607053756713867, "learning_rate": 9.394537204506631e-06, "loss": 0.0063, "step": 243480 }, { "epoch": 155.18801784576164, "grad_norm": 0.1308620721101761, "learning_rate": 9.391321422630777e-06, "loss": 0.0044, "step": 243490 }, { "epoch": 155.19439133205864, "grad_norm": 0.14316309988498688, "learning_rate": 9.388106134188685e-06, "loss": 0.009, "step": 243500 }, { "epoch": 155.20076481835565, "grad_norm": 0.1130523681640625, "learning_rate": 9.384891339219454e-06, "loss": 0.0053, "step": 243510 }, { "epoch": 155.20713830465266, "grad_norm": 0.07699710130691528, "learning_rate": 9.381677037762132e-06, "loss": 0.0076, "step": 243520 }, { "epoch": 155.21351179094964, "grad_norm": 0.16119468212127686, "learning_rate": 9.378463229855811e-06, "loss": 0.0049, "step": 243530 }, { "epoch": 155.21988527724665, "grad_norm": 0.153058260679245, "learning_rate": 9.375249915539485e-06, "loss": 0.0081, "step": 243540 }, { "epoch": 155.22625876354365, "grad_norm": 0.20625713467597961, "learning_rate": 9.372037094852243e-06, "loss": 0.0032, "step": 243550 }, { "epoch": 155.23263224984066, "grad_norm": 0.1123591959476471, "learning_rate": 9.368824767833112e-06, "loss": 0.0048, "step": 243560 }, { "epoch": 155.23900573613767, "grad_norm": 0.1966567188501358, "learning_rate": 9.365612934521117e-06, "loss": 0.0046, "step": 243570 }, { "epoch": 155.24537922243468, "grad_norm": 0.13199663162231445, "learning_rate": 9.362401594955283e-06, "loss": 0.006, "step": 243580 }, { "epoch": 155.2517527087317, "grad_norm": 0.10210982710123062, "learning_rate": 9.359190749174646e-06, "loss": 0.0049, "step": 243590 }, { "epoch": 155.2581261950287, "grad_norm": 0.15140534937381744, "learning_rate": 9.355980397218217e-06, "loss": 0.0052, "step": 243600 }, { "epoch": 155.26449968132567, "grad_norm": 0.2839451730251312, "learning_rate": 9.352770539125e-06, "loss": 0.0099, "step": 243610 }, { "epoch": 155.27087316762268, "grad_norm": 0.19480407238006592, "learning_rate": 9.34956117493398e-06, "loss": 0.0042, "step": 243620 }, { "epoch": 155.2772466539197, "grad_norm": 0.23654066026210785, "learning_rate": 9.346352304684191e-06, "loss": 0.0086, "step": 243630 }, { "epoch": 155.2836201402167, "grad_norm": 0.15060332417488098, "learning_rate": 9.3431439284146e-06, "loss": 0.0133, "step": 243640 }, { "epoch": 155.2899936265137, "grad_norm": 0.053042687475681305, "learning_rate": 9.339936046164199e-06, "loss": 0.0052, "step": 243650 }, { "epoch": 155.2963671128107, "grad_norm": 0.22590094804763794, "learning_rate": 9.336728657971954e-06, "loss": 0.0069, "step": 243660 }, { "epoch": 155.30274059910772, "grad_norm": 0.15809062123298645, "learning_rate": 9.333521763876857e-06, "loss": 0.0069, "step": 243670 }, { "epoch": 155.30911408540473, "grad_norm": 0.15088263154029846, "learning_rate": 9.330315363917869e-06, "loss": 0.0037, "step": 243680 }, { "epoch": 155.3154875717017, "grad_norm": 0.20888927578926086, "learning_rate": 9.327109458133948e-06, "loss": 0.0063, "step": 243690 }, { "epoch": 155.32186105799872, "grad_norm": 0.10383713245391846, "learning_rate": 9.323904046564035e-06, "loss": 0.0064, "step": 243700 }, { "epoch": 155.32823454429573, "grad_norm": 0.4224194884300232, "learning_rate": 9.320699129247107e-06, "loss": 0.0084, "step": 243710 }, { "epoch": 155.33460803059273, "grad_norm": 0.39350971579551697, "learning_rate": 9.31749470622209e-06, "loss": 0.0058, "step": 243720 }, { "epoch": 155.34098151688974, "grad_norm": 0.13168321549892426, "learning_rate": 9.314290777527923e-06, "loss": 0.0054, "step": 243730 }, { "epoch": 155.34735500318675, "grad_norm": 0.38547131419181824, "learning_rate": 9.311087343203534e-06, "loss": 0.0059, "step": 243740 }, { "epoch": 155.35372848948376, "grad_norm": 0.15253469347953796, "learning_rate": 9.307884403287842e-06, "loss": 0.003, "step": 243750 }, { "epoch": 155.36010197578076, "grad_norm": 0.3303816616535187, "learning_rate": 9.304681957819789e-06, "loss": 0.0044, "step": 243760 }, { "epoch": 155.36647546207774, "grad_norm": 0.2084241807460785, "learning_rate": 9.301480006838264e-06, "loss": 0.0057, "step": 243770 }, { "epoch": 155.37284894837475, "grad_norm": 0.30561086535453796, "learning_rate": 9.298278550382189e-06, "loss": 0.0084, "step": 243780 }, { "epoch": 155.37922243467176, "grad_norm": 0.09667328000068665, "learning_rate": 9.295077588490447e-06, "loss": 0.0051, "step": 243790 }, { "epoch": 155.38559592096877, "grad_norm": 0.5059058666229248, "learning_rate": 9.291877121201953e-06, "loss": 0.0101, "step": 243800 }, { "epoch": 155.39196940726578, "grad_norm": 0.11445721983909607, "learning_rate": 9.288677148555591e-06, "loss": 0.004, "step": 243810 }, { "epoch": 155.39834289356278, "grad_norm": 0.4681810140609741, "learning_rate": 9.285477670590236e-06, "loss": 0.0094, "step": 243820 }, { "epoch": 155.4047163798598, "grad_norm": 0.1900876760482788, "learning_rate": 9.282278687344759e-06, "loss": 0.0041, "step": 243830 }, { "epoch": 155.4110898661568, "grad_norm": 0.35082775354385376, "learning_rate": 9.279080198858048e-06, "loss": 0.0059, "step": 243840 }, { "epoch": 155.41746335245378, "grad_norm": 0.6234395503997803, "learning_rate": 9.275882205168962e-06, "loss": 0.0064, "step": 243850 }, { "epoch": 155.4238368387508, "grad_norm": 0.12593331933021545, "learning_rate": 9.27268470631636e-06, "loss": 0.005, "step": 243860 }, { "epoch": 155.4302103250478, "grad_norm": 0.0760035440325737, "learning_rate": 9.269487702339075e-06, "loss": 0.0106, "step": 243870 }, { "epoch": 155.4365838113448, "grad_norm": 0.08114614337682724, "learning_rate": 9.266291193275983e-06, "loss": 0.0047, "step": 243880 }, { "epoch": 155.4429572976418, "grad_norm": 0.08306657522916794, "learning_rate": 9.263095179165914e-06, "loss": 0.0046, "step": 243890 }, { "epoch": 155.44933078393882, "grad_norm": 0.1780644655227661, "learning_rate": 9.2598996600477e-06, "loss": 0.0036, "step": 243900 }, { "epoch": 155.45570427023583, "grad_norm": 0.38485926389694214, "learning_rate": 9.256704635960157e-06, "loss": 0.0114, "step": 243910 }, { "epoch": 155.46207775653284, "grad_norm": 0.782321035861969, "learning_rate": 9.253510106942142e-06, "loss": 0.0087, "step": 243920 }, { "epoch": 155.46845124282981, "grad_norm": 0.21115224063396454, "learning_rate": 9.250316073032433e-06, "loss": 0.0046, "step": 243930 }, { "epoch": 155.47482472912682, "grad_norm": 0.08891868591308594, "learning_rate": 9.247122534269864e-06, "loss": 0.0044, "step": 243940 }, { "epoch": 155.48119821542383, "grad_norm": 0.13955356180667877, "learning_rate": 9.243929490693226e-06, "loss": 0.0036, "step": 243950 }, { "epoch": 155.48757170172084, "grad_norm": 0.45552313327789307, "learning_rate": 9.24073694234135e-06, "loss": 0.0049, "step": 243960 }, { "epoch": 155.49394518801785, "grad_norm": 0.12675607204437256, "learning_rate": 9.237544889252969e-06, "loss": 0.0044, "step": 243970 }, { "epoch": 155.50031867431485, "grad_norm": 0.15341109037399292, "learning_rate": 9.234353331466927e-06, "loss": 0.0057, "step": 243980 }, { "epoch": 155.50669216061186, "grad_norm": 0.08130309730768204, "learning_rate": 9.231162269021975e-06, "loss": 0.0077, "step": 243990 }, { "epoch": 155.51306564690887, "grad_norm": 0.5816522240638733, "learning_rate": 9.227971701956894e-06, "loss": 0.0059, "step": 244000 }, { "epoch": 155.51943913320585, "grad_norm": 0.1288529932498932, "learning_rate": 9.224781630310442e-06, "loss": 0.0043, "step": 244010 }, { "epoch": 155.52581261950286, "grad_norm": 0.2327207624912262, "learning_rate": 9.221592054121404e-06, "loss": 0.0056, "step": 244020 }, { "epoch": 155.53218610579987, "grad_norm": 0.16735775768756866, "learning_rate": 9.21840297342852e-06, "loss": 0.0057, "step": 244030 }, { "epoch": 155.53855959209687, "grad_norm": 0.3005504608154297, "learning_rate": 9.21521438827055e-06, "loss": 0.0072, "step": 244040 }, { "epoch": 155.54493307839388, "grad_norm": 0.09811744838953018, "learning_rate": 9.212026298686216e-06, "loss": 0.0034, "step": 244050 }, { "epoch": 155.5513065646909, "grad_norm": 0.10105559229850769, "learning_rate": 9.208838704714295e-06, "loss": 0.0064, "step": 244060 }, { "epoch": 155.5576800509879, "grad_norm": 0.07322816550731659, "learning_rate": 9.205651606393479e-06, "loss": 0.0043, "step": 244070 }, { "epoch": 155.5640535372849, "grad_norm": 0.16656063497066498, "learning_rate": 9.202465003762517e-06, "loss": 0.005, "step": 244080 }, { "epoch": 155.57042702358189, "grad_norm": 0.20156756043434143, "learning_rate": 9.19927889686012e-06, "loss": 0.0059, "step": 244090 }, { "epoch": 155.5768005098789, "grad_norm": 0.15157294273376465, "learning_rate": 9.196093285725027e-06, "loss": 0.0068, "step": 244100 }, { "epoch": 155.5831739961759, "grad_norm": 0.22768676280975342, "learning_rate": 9.192908170395904e-06, "loss": 0.004, "step": 244110 }, { "epoch": 155.5895474824729, "grad_norm": 0.14211612939834595, "learning_rate": 9.189723550911483e-06, "loss": 0.0035, "step": 244120 }, { "epoch": 155.59592096876992, "grad_norm": 0.20951882004737854, "learning_rate": 9.186539427310454e-06, "loss": 0.0053, "step": 244130 }, { "epoch": 155.60229445506693, "grad_norm": 0.09940865635871887, "learning_rate": 9.183355799631504e-06, "loss": 0.0099, "step": 244140 }, { "epoch": 155.60866794136393, "grad_norm": 0.13684505224227905, "learning_rate": 9.180172667913306e-06, "loss": 0.0222, "step": 244150 }, { "epoch": 155.61504142766094, "grad_norm": 0.1711139976978302, "learning_rate": 9.176990032194565e-06, "loss": 0.0043, "step": 244160 }, { "epoch": 155.62141491395795, "grad_norm": 0.11415888369083405, "learning_rate": 9.173807892513935e-06, "loss": 0.0052, "step": 244170 }, { "epoch": 155.62778840025493, "grad_norm": 0.11032241582870483, "learning_rate": 9.17062624891009e-06, "loss": 0.0047, "step": 244180 }, { "epoch": 155.63416188655194, "grad_norm": 0.45404109358787537, "learning_rate": 9.167445101421673e-06, "loss": 0.0088, "step": 244190 }, { "epoch": 155.64053537284894, "grad_norm": 0.43854793906211853, "learning_rate": 9.164264450087356e-06, "loss": 0.0036, "step": 244200 }, { "epoch": 155.64690885914595, "grad_norm": 0.22503748536109924, "learning_rate": 9.161084294945787e-06, "loss": 0.0059, "step": 244210 }, { "epoch": 155.65328234544296, "grad_norm": 0.11611872911453247, "learning_rate": 9.157904636035602e-06, "loss": 0.0032, "step": 244220 }, { "epoch": 155.65965583173997, "grad_norm": 0.09952797740697861, "learning_rate": 9.154725473395426e-06, "loss": 0.0042, "step": 244230 }, { "epoch": 155.66602931803698, "grad_norm": 0.16069452464580536, "learning_rate": 9.151546807063922e-06, "loss": 0.0059, "step": 244240 }, { "epoch": 155.67240280433398, "grad_norm": 0.2007865309715271, "learning_rate": 9.148368637079663e-06, "loss": 0.0052, "step": 244250 }, { "epoch": 155.67877629063096, "grad_norm": 0.13745251297950745, "learning_rate": 9.145190963481315e-06, "loss": 0.0071, "step": 244260 }, { "epoch": 155.68514977692797, "grad_norm": 0.2964912950992584, "learning_rate": 9.142013786307451e-06, "loss": 0.0088, "step": 244270 }, { "epoch": 155.69152326322498, "grad_norm": 0.18545131385326385, "learning_rate": 9.138837105596725e-06, "loss": 0.0093, "step": 244280 }, { "epoch": 155.697896749522, "grad_norm": 0.17052920162677765, "learning_rate": 9.135660921387685e-06, "loss": 0.0051, "step": 244290 }, { "epoch": 155.704270235819, "grad_norm": 0.1775791198015213, "learning_rate": 9.132485233718957e-06, "loss": 0.0087, "step": 244300 }, { "epoch": 155.710643722116, "grad_norm": 0.10917146503925323, "learning_rate": 9.129310042629118e-06, "loss": 0.0049, "step": 244310 }, { "epoch": 155.717017208413, "grad_norm": 0.21426042914390564, "learning_rate": 9.12613534815675e-06, "loss": 0.0052, "step": 244320 }, { "epoch": 155.72339069471002, "grad_norm": 0.08221983909606934, "learning_rate": 9.122961150340414e-06, "loss": 0.0046, "step": 244330 }, { "epoch": 155.729764181007, "grad_norm": 0.5202838778495789, "learning_rate": 9.119787449218709e-06, "loss": 0.0049, "step": 244340 }, { "epoch": 155.736137667304, "grad_norm": 0.18295496702194214, "learning_rate": 9.116614244830184e-06, "loss": 0.0064, "step": 244350 }, { "epoch": 155.74251115360101, "grad_norm": 0.3205375373363495, "learning_rate": 9.113441537213402e-06, "loss": 0.0104, "step": 244360 }, { "epoch": 155.74888463989802, "grad_norm": 0.7098721265792847, "learning_rate": 9.110269326406889e-06, "loss": 0.0045, "step": 244370 }, { "epoch": 155.75525812619503, "grad_norm": 0.1180383712053299, "learning_rate": 9.107097612449222e-06, "loss": 0.0107, "step": 244380 }, { "epoch": 155.76163161249204, "grad_norm": 0.2574986219406128, "learning_rate": 9.10392639537893e-06, "loss": 0.0059, "step": 244390 }, { "epoch": 155.76800509878905, "grad_norm": 0.10383614152669907, "learning_rate": 9.100755675234546e-06, "loss": 0.0033, "step": 244400 }, { "epoch": 155.77437858508605, "grad_norm": 0.12747573852539062, "learning_rate": 9.097585452054585e-06, "loss": 0.0107, "step": 244410 }, { "epoch": 155.78075207138303, "grad_norm": 0.16129949688911438, "learning_rate": 9.094415725877586e-06, "loss": 0.0029, "step": 244420 }, { "epoch": 155.78712555768004, "grad_norm": 0.11507674306631088, "learning_rate": 9.091246496742061e-06, "loss": 0.0027, "step": 244430 }, { "epoch": 155.79349904397705, "grad_norm": 0.045433543622493744, "learning_rate": 9.088077764686514e-06, "loss": 0.005, "step": 244440 }, { "epoch": 155.79987253027406, "grad_norm": 0.36995330452919006, "learning_rate": 9.084909529749453e-06, "loss": 0.0094, "step": 244450 }, { "epoch": 155.80624601657107, "grad_norm": 0.18706971406936646, "learning_rate": 9.081741791969357e-06, "loss": 0.004, "step": 244460 }, { "epoch": 155.81261950286807, "grad_norm": 0.28293150663375854, "learning_rate": 9.078574551384739e-06, "loss": 0.0069, "step": 244470 }, { "epoch": 155.81899298916508, "grad_norm": 0.1369030624628067, "learning_rate": 9.07540780803407e-06, "loss": 0.0067, "step": 244480 }, { "epoch": 155.8253664754621, "grad_norm": 0.35261452198028564, "learning_rate": 9.072241561955857e-06, "loss": 0.0088, "step": 244490 }, { "epoch": 155.83173996175907, "grad_norm": 0.12511561810970306, "learning_rate": 9.069075813188527e-06, "loss": 0.0103, "step": 244500 }, { "epoch": 155.83811344805608, "grad_norm": 0.2506398856639862, "learning_rate": 9.065910561770585e-06, "loss": 0.0122, "step": 244510 }, { "epoch": 155.84448693435309, "grad_norm": 0.16417695581912994, "learning_rate": 9.062745807740463e-06, "loss": 0.0051, "step": 244520 }, { "epoch": 155.8508604206501, "grad_norm": 0.19370830059051514, "learning_rate": 9.05958155113666e-06, "loss": 0.0046, "step": 244530 }, { "epoch": 155.8572339069471, "grad_norm": 0.2065320909023285, "learning_rate": 9.056417791997568e-06, "loss": 0.003, "step": 244540 }, { "epoch": 155.8636073932441, "grad_norm": 0.116331547498703, "learning_rate": 9.053254530361666e-06, "loss": 0.0054, "step": 244550 }, { "epoch": 155.86998087954112, "grad_norm": 0.0910438671708107, "learning_rate": 9.050091766267387e-06, "loss": 0.0044, "step": 244560 }, { "epoch": 155.87635436583813, "grad_norm": 0.25420695543289185, "learning_rate": 9.046929499753155e-06, "loss": 0.0076, "step": 244570 }, { "epoch": 155.8827278521351, "grad_norm": 0.10622257739305496, "learning_rate": 9.043767730857377e-06, "loss": 0.0082, "step": 244580 }, { "epoch": 155.8891013384321, "grad_norm": 0.09603475034236908, "learning_rate": 9.040606459618511e-06, "loss": 0.0086, "step": 244590 }, { "epoch": 155.89547482472912, "grad_norm": 0.08635661751031876, "learning_rate": 9.037445686074941e-06, "loss": 0.0047, "step": 244600 }, { "epoch": 155.90184831102613, "grad_norm": 0.1881391406059265, "learning_rate": 9.034285410265087e-06, "loss": 0.0079, "step": 244610 }, { "epoch": 155.90822179732314, "grad_norm": 0.11894901096820831, "learning_rate": 9.031125632227327e-06, "loss": 0.0039, "step": 244620 }, { "epoch": 155.91459528362014, "grad_norm": 0.17168749868869781, "learning_rate": 9.0279663520001e-06, "loss": 0.0068, "step": 244630 }, { "epoch": 155.92096876991715, "grad_norm": 0.07049562782049179, "learning_rate": 9.02480756962174e-06, "loss": 0.0036, "step": 244640 }, { "epoch": 155.92734225621416, "grad_norm": 0.086650051176548, "learning_rate": 9.021649285130668e-06, "loss": 0.0032, "step": 244650 }, { "epoch": 155.93371574251114, "grad_norm": 0.45130181312561035, "learning_rate": 9.018491498565235e-06, "loss": 0.0084, "step": 244660 }, { "epoch": 155.94008922880815, "grad_norm": 0.30800339579582214, "learning_rate": 9.015334209963845e-06, "loss": 0.0046, "step": 244670 }, { "epoch": 155.94646271510516, "grad_norm": 0.1749805063009262, "learning_rate": 9.012177419364815e-06, "loss": 0.005, "step": 244680 }, { "epoch": 155.95283620140216, "grad_norm": 0.18762202560901642, "learning_rate": 9.009021126806543e-06, "loss": 0.0051, "step": 244690 }, { "epoch": 155.95920968769917, "grad_norm": 0.13522176444530487, "learning_rate": 9.005865332327369e-06, "loss": 0.0096, "step": 244700 }, { "epoch": 155.96558317399618, "grad_norm": 0.20997223258018494, "learning_rate": 9.002710035965627e-06, "loss": 0.0039, "step": 244710 }, { "epoch": 155.9719566602932, "grad_norm": 0.07524542510509491, "learning_rate": 8.999555237759661e-06, "loss": 0.0114, "step": 244720 }, { "epoch": 155.9783301465902, "grad_norm": 0.09874916076660156, "learning_rate": 8.996400937747824e-06, "loss": 0.0068, "step": 244730 }, { "epoch": 155.98470363288718, "grad_norm": 0.45509347319602966, "learning_rate": 8.993247135968425e-06, "loss": 0.0065, "step": 244740 }, { "epoch": 155.99107711918418, "grad_norm": 0.21857231855392456, "learning_rate": 8.99009383245979e-06, "loss": 0.0083, "step": 244750 }, { "epoch": 155.9974506054812, "grad_norm": 0.17306393384933472, "learning_rate": 8.986941027260226e-06, "loss": 0.0078, "step": 244760 }, { "epoch": 156.0038240917782, "grad_norm": 0.24290448427200317, "learning_rate": 8.98378872040807e-06, "loss": 0.003, "step": 244770 }, { "epoch": 156.0101975780752, "grad_norm": 0.05028976872563362, "learning_rate": 8.980636911941587e-06, "loss": 0.0057, "step": 244780 }, { "epoch": 156.01657106437222, "grad_norm": 0.4912436604499817, "learning_rate": 8.977485601899106e-06, "loss": 0.0063, "step": 244790 }, { "epoch": 156.02294455066922, "grad_norm": 0.16242550313472748, "learning_rate": 8.974334790318895e-06, "loss": 0.007, "step": 244800 }, { "epoch": 156.02931803696623, "grad_norm": 0.2679588496685028, "learning_rate": 8.97118447723927e-06, "loss": 0.0058, "step": 244810 }, { "epoch": 156.03569152326324, "grad_norm": 0.22199827432632446, "learning_rate": 8.968034662698472e-06, "loss": 0.0106, "step": 244820 }, { "epoch": 156.04206500956022, "grad_norm": 0.18536926805973053, "learning_rate": 8.964885346734802e-06, "loss": 0.0131, "step": 244830 }, { "epoch": 156.04843849585723, "grad_norm": 0.2475997805595398, "learning_rate": 8.961736529386511e-06, "loss": 0.0036, "step": 244840 }, { "epoch": 156.05481198215423, "grad_norm": 0.4272516667842865, "learning_rate": 8.958588210691886e-06, "loss": 0.0126, "step": 244850 }, { "epoch": 156.06118546845124, "grad_norm": 0.4382634162902832, "learning_rate": 8.955440390689146e-06, "loss": 0.0074, "step": 244860 }, { "epoch": 156.06755895474825, "grad_norm": 0.17858649790287018, "learning_rate": 8.952293069416567e-06, "loss": 0.0082, "step": 244870 }, { "epoch": 156.07393244104526, "grad_norm": 0.1358393430709839, "learning_rate": 8.949146246912382e-06, "loss": 0.0054, "step": 244880 }, { "epoch": 156.08030592734227, "grad_norm": 0.09886609017848969, "learning_rate": 8.945999923214832e-06, "loss": 0.008, "step": 244890 }, { "epoch": 156.08667941363927, "grad_norm": 0.39050349593162537, "learning_rate": 8.94285409836213e-06, "loss": 0.0101, "step": 244900 }, { "epoch": 156.09305289993625, "grad_norm": 0.1367151141166687, "learning_rate": 8.939708772392525e-06, "loss": 0.003, "step": 244910 }, { "epoch": 156.09942638623326, "grad_norm": 0.35479801893234253, "learning_rate": 8.936563945344228e-06, "loss": 0.0062, "step": 244920 }, { "epoch": 156.10579987253027, "grad_norm": 0.16092199087142944, "learning_rate": 8.93341961725545e-06, "loss": 0.0043, "step": 244930 }, { "epoch": 156.11217335882728, "grad_norm": 0.09768198430538177, "learning_rate": 8.930275788164383e-06, "loss": 0.0037, "step": 244940 }, { "epoch": 156.11854684512429, "grad_norm": 0.1401725709438324, "learning_rate": 8.927132458109266e-06, "loss": 0.0067, "step": 244950 }, { "epoch": 156.1249203314213, "grad_norm": 0.17204834520816803, "learning_rate": 8.923989627128244e-06, "loss": 0.0043, "step": 244960 }, { "epoch": 156.1312938177183, "grad_norm": 0.2004869282245636, "learning_rate": 8.920847295259544e-06, "loss": 0.0063, "step": 244970 }, { "epoch": 156.1376673040153, "grad_norm": 0.23878233134746552, "learning_rate": 8.917705462541321e-06, "loss": 0.0065, "step": 244980 }, { "epoch": 156.1440407903123, "grad_norm": 0.1562202274799347, "learning_rate": 8.914564129011787e-06, "loss": 0.0045, "step": 244990 }, { "epoch": 156.1504142766093, "grad_norm": 0.2741234600543976, "learning_rate": 8.91142329470907e-06, "loss": 0.0077, "step": 245000 }, { "epoch": 156.1567877629063, "grad_norm": 0.09296412020921707, "learning_rate": 8.908282959671366e-06, "loss": 0.0045, "step": 245010 }, { "epoch": 156.1631612492033, "grad_norm": 0.1184239313006401, "learning_rate": 8.905143123936815e-06, "loss": 0.0056, "step": 245020 }, { "epoch": 156.16953473550032, "grad_norm": 0.1140359565615654, "learning_rate": 8.902003787543572e-06, "loss": 0.0041, "step": 245030 }, { "epoch": 156.17590822179733, "grad_norm": 0.35513630509376526, "learning_rate": 8.89886495052979e-06, "loss": 0.0081, "step": 245040 }, { "epoch": 156.18228170809434, "grad_norm": 0.25426119565963745, "learning_rate": 8.895726612933608e-06, "loss": 0.0045, "step": 245050 }, { "epoch": 156.18865519439134, "grad_norm": 0.4569339156150818, "learning_rate": 8.892588774793159e-06, "loss": 0.008, "step": 245060 }, { "epoch": 156.19502868068832, "grad_norm": 0.11852329969406128, "learning_rate": 8.889451436146551e-06, "loss": 0.0043, "step": 245070 }, { "epoch": 156.20140216698533, "grad_norm": 0.0687749981880188, "learning_rate": 8.886314597031942e-06, "loss": 0.0034, "step": 245080 }, { "epoch": 156.20777565328234, "grad_norm": 0.30810868740081787, "learning_rate": 8.883178257487423e-06, "loss": 0.0036, "step": 245090 }, { "epoch": 156.21414913957935, "grad_norm": 0.12478208541870117, "learning_rate": 8.88004241755111e-06, "loss": 0.0097, "step": 245100 }, { "epoch": 156.22052262587636, "grad_norm": 0.11530270427465439, "learning_rate": 8.876907077261093e-06, "loss": 0.004, "step": 245110 }, { "epoch": 156.22689611217336, "grad_norm": 0.14133694767951965, "learning_rate": 8.873772236655493e-06, "loss": 0.0036, "step": 245120 }, { "epoch": 156.23326959847037, "grad_norm": 0.25446629524230957, "learning_rate": 8.870637895772387e-06, "loss": 0.005, "step": 245130 }, { "epoch": 156.23964308476738, "grad_norm": 0.07067503035068512, "learning_rate": 8.867504054649872e-06, "loss": 0.005, "step": 245140 }, { "epoch": 156.24601657106436, "grad_norm": 0.14535090327262878, "learning_rate": 8.864370713325999e-06, "loss": 0.0041, "step": 245150 }, { "epoch": 156.25239005736137, "grad_norm": 0.4369239807128906, "learning_rate": 8.861237871838873e-06, "loss": 0.0076, "step": 245160 }, { "epoch": 156.25876354365838, "grad_norm": 0.6973611116409302, "learning_rate": 8.858105530226551e-06, "loss": 0.0108, "step": 245170 }, { "epoch": 156.26513702995538, "grad_norm": 0.24474279582500458, "learning_rate": 8.854973688527091e-06, "loss": 0.0061, "step": 245180 }, { "epoch": 156.2715105162524, "grad_norm": 0.2485406994819641, "learning_rate": 8.851842346778533e-06, "loss": 0.0037, "step": 245190 }, { "epoch": 156.2778840025494, "grad_norm": 0.13072070479393005, "learning_rate": 8.848711505018969e-06, "loss": 0.0046, "step": 245200 }, { "epoch": 156.2842574888464, "grad_norm": 0.23046383261680603, "learning_rate": 8.845581163286387e-06, "loss": 0.0033, "step": 245210 }, { "epoch": 156.29063097514342, "grad_norm": 0.3048831820487976, "learning_rate": 8.84245132161886e-06, "loss": 0.0049, "step": 245220 }, { "epoch": 156.2970044614404, "grad_norm": 0.2937866151332855, "learning_rate": 8.839321980054399e-06, "loss": 0.0035, "step": 245230 }, { "epoch": 156.3033779477374, "grad_norm": 0.06964564323425293, "learning_rate": 8.836193138631066e-06, "loss": 0.0039, "step": 245240 }, { "epoch": 156.3097514340344, "grad_norm": 0.34270378947257996, "learning_rate": 8.83306479738682e-06, "loss": 0.0064, "step": 245250 }, { "epoch": 156.31612492033142, "grad_norm": 0.12945394217967987, "learning_rate": 8.829936956359719e-06, "loss": 0.0065, "step": 245260 }, { "epoch": 156.32249840662843, "grad_norm": 0.1661062091588974, "learning_rate": 8.826809615587762e-06, "loss": 0.0059, "step": 245270 }, { "epoch": 156.32887189292543, "grad_norm": 0.14703166484832764, "learning_rate": 8.823682775108932e-06, "loss": 0.0044, "step": 245280 }, { "epoch": 156.33524537922244, "grad_norm": 0.12768515944480896, "learning_rate": 8.820556434961225e-06, "loss": 0.0051, "step": 245290 }, { "epoch": 156.34161886551945, "grad_norm": 0.2581055760383606, "learning_rate": 8.817430595182652e-06, "loss": 0.0059, "step": 245300 }, { "epoch": 156.34799235181643, "grad_norm": 0.10233515501022339, "learning_rate": 8.814305255811179e-06, "loss": 0.0033, "step": 245310 }, { "epoch": 156.35436583811344, "grad_norm": 0.11321739107370377, "learning_rate": 8.811180416884778e-06, "loss": 0.005, "step": 245320 }, { "epoch": 156.36073932441045, "grad_norm": 0.31557947397232056, "learning_rate": 8.808056078441418e-06, "loss": 0.0057, "step": 245330 }, { "epoch": 156.36711281070745, "grad_norm": 0.14247146248817444, "learning_rate": 8.804932240519087e-06, "loss": 0.0135, "step": 245340 }, { "epoch": 156.37348629700446, "grad_norm": 0.09223005920648575, "learning_rate": 8.8018089031557e-06, "loss": 0.0124, "step": 245350 }, { "epoch": 156.37985978330147, "grad_norm": 0.4108101725578308, "learning_rate": 8.79868606638925e-06, "loss": 0.0066, "step": 245360 }, { "epoch": 156.38623326959848, "grad_norm": 0.26261040568351746, "learning_rate": 8.79556373025765e-06, "loss": 0.0026, "step": 245370 }, { "epoch": 156.39260675589549, "grad_norm": 0.21005186438560486, "learning_rate": 8.792441894798874e-06, "loss": 0.0095, "step": 245380 }, { "epoch": 156.39898024219247, "grad_norm": 0.062206193804740906, "learning_rate": 8.789320560050818e-06, "loss": 0.0058, "step": 245390 }, { "epoch": 156.40535372848947, "grad_norm": 0.2883543074131012, "learning_rate": 8.786199726051436e-06, "loss": 0.0039, "step": 245400 }, { "epoch": 156.41172721478648, "grad_norm": 0.3395735025405884, "learning_rate": 8.783079392838627e-06, "loss": 0.0103, "step": 245410 }, { "epoch": 156.4181007010835, "grad_norm": 0.17645268142223358, "learning_rate": 8.77995956045034e-06, "loss": 0.0027, "step": 245420 }, { "epoch": 156.4244741873805, "grad_norm": 0.2178686261177063, "learning_rate": 8.776840228924444e-06, "loss": 0.0103, "step": 245430 }, { "epoch": 156.4308476736775, "grad_norm": 0.2565377950668335, "learning_rate": 8.773721398298867e-06, "loss": 0.0074, "step": 245440 }, { "epoch": 156.4372211599745, "grad_norm": 0.1943625956773758, "learning_rate": 8.770603068611505e-06, "loss": 0.015, "step": 245450 }, { "epoch": 156.44359464627152, "grad_norm": 0.19277937710285187, "learning_rate": 8.767485239900237e-06, "loss": 0.0095, "step": 245460 }, { "epoch": 156.44996813256853, "grad_norm": 0.15980540215969086, "learning_rate": 8.764367912202942e-06, "loss": 0.0067, "step": 245470 }, { "epoch": 156.4563416188655, "grad_norm": 0.1401669830083847, "learning_rate": 8.761251085557525e-06, "loss": 0.0063, "step": 245480 }, { "epoch": 156.46271510516252, "grad_norm": 0.1475045084953308, "learning_rate": 8.75813476000184e-06, "loss": 0.0066, "step": 245490 }, { "epoch": 156.46908859145952, "grad_norm": 0.09757357835769653, "learning_rate": 8.755018935573761e-06, "loss": 0.0087, "step": 245500 }, { "epoch": 156.47546207775653, "grad_norm": 0.11805988848209381, "learning_rate": 8.751903612311124e-06, "loss": 0.0091, "step": 245510 }, { "epoch": 156.48183556405354, "grad_norm": 0.11991878598928452, "learning_rate": 8.748788790251833e-06, "loss": 0.0053, "step": 245520 }, { "epoch": 156.48820905035055, "grad_norm": 0.1247425451874733, "learning_rate": 8.74567446943368e-06, "loss": 0.0034, "step": 245530 }, { "epoch": 156.49458253664756, "grad_norm": 0.11779912561178207, "learning_rate": 8.742560649894544e-06, "loss": 0.0071, "step": 245540 }, { "epoch": 156.50095602294456, "grad_norm": 0.12567055225372314, "learning_rate": 8.73944733167224e-06, "loss": 0.003, "step": 245550 }, { "epoch": 156.50732950924154, "grad_norm": 0.20193694531917572, "learning_rate": 8.736334514804629e-06, "loss": 0.0077, "step": 245560 }, { "epoch": 156.51370299553855, "grad_norm": 0.07563360780477524, "learning_rate": 8.733222199329488e-06, "loss": 0.0047, "step": 245570 }, { "epoch": 156.52007648183556, "grad_norm": 0.16843779385089874, "learning_rate": 8.730110385284673e-06, "loss": 0.0073, "step": 245580 }, { "epoch": 156.52644996813257, "grad_norm": 0.2714974880218506, "learning_rate": 8.726999072707987e-06, "loss": 0.0046, "step": 245590 }, { "epoch": 156.53282345442958, "grad_norm": 0.08483027666807175, "learning_rate": 8.723888261637226e-06, "loss": 0.0087, "step": 245600 }, { "epoch": 156.53919694072658, "grad_norm": 0.1171664148569107, "learning_rate": 8.720777952110181e-06, "loss": 0.0076, "step": 245610 }, { "epoch": 156.5455704270236, "grad_norm": 0.06751760095357895, "learning_rate": 8.717668144164671e-06, "loss": 0.0036, "step": 245620 }, { "epoch": 156.5519439133206, "grad_norm": 0.10706749558448792, "learning_rate": 8.714558837838465e-06, "loss": 0.0069, "step": 245630 }, { "epoch": 156.55831739961758, "grad_norm": 0.30799272656440735, "learning_rate": 8.711450033169356e-06, "loss": 0.0088, "step": 245640 }, { "epoch": 156.5646908859146, "grad_norm": 0.1447061449289322, "learning_rate": 8.708341730195096e-06, "loss": 0.0079, "step": 245650 }, { "epoch": 156.5710643722116, "grad_norm": 0.37078920006752014, "learning_rate": 8.705233928953482e-06, "loss": 0.0062, "step": 245660 }, { "epoch": 156.5774378585086, "grad_norm": 0.30984458327293396, "learning_rate": 8.702126629482266e-06, "loss": 0.0124, "step": 245670 }, { "epoch": 156.5838113448056, "grad_norm": 0.14801068603992462, "learning_rate": 8.699019831819205e-06, "loss": 0.0043, "step": 245680 }, { "epoch": 156.59018483110262, "grad_norm": 0.2499924898147583, "learning_rate": 8.695913536002032e-06, "loss": 0.0049, "step": 245690 }, { "epoch": 156.59655831739963, "grad_norm": 0.4697401523590088, "learning_rate": 8.692807742068521e-06, "loss": 0.0059, "step": 245700 }, { "epoch": 156.60293180369663, "grad_norm": 0.40023934841156006, "learning_rate": 8.689702450056397e-06, "loss": 0.0047, "step": 245710 }, { "epoch": 156.6093052899936, "grad_norm": 0.1690974086523056, "learning_rate": 8.686597660003393e-06, "loss": 0.0094, "step": 245720 }, { "epoch": 156.61567877629062, "grad_norm": 0.12763629853725433, "learning_rate": 8.683493371947216e-06, "loss": 0.011, "step": 245730 }, { "epoch": 156.62205226258763, "grad_norm": 0.3960777521133423, "learning_rate": 8.68038958592562e-06, "loss": 0.0053, "step": 245740 }, { "epoch": 156.62842574888464, "grad_norm": 0.1754889041185379, "learning_rate": 8.677286301976301e-06, "loss": 0.0053, "step": 245750 }, { "epoch": 156.63479923518165, "grad_norm": 0.29509735107421875, "learning_rate": 8.674183520136969e-06, "loss": 0.0103, "step": 245760 }, { "epoch": 156.64117272147865, "grad_norm": 0.15855847299098969, "learning_rate": 8.671081240445327e-06, "loss": 0.0054, "step": 245770 }, { "epoch": 156.64754620777566, "grad_norm": 0.20231030881404877, "learning_rate": 8.667979462939052e-06, "loss": 0.0065, "step": 245780 }, { "epoch": 156.65391969407267, "grad_norm": 0.075351782143116, "learning_rate": 8.664878187655867e-06, "loss": 0.0057, "step": 245790 }, { "epoch": 156.66029318036965, "grad_norm": 0.18839547038078308, "learning_rate": 8.661777414633437e-06, "loss": 0.0081, "step": 245800 }, { "epoch": 156.66666666666666, "grad_norm": 0.10751429200172424, "learning_rate": 8.658677143909444e-06, "loss": 0.0059, "step": 245810 }, { "epoch": 156.67304015296367, "grad_norm": 0.5287175178527832, "learning_rate": 8.65557737552154e-06, "loss": 0.0091, "step": 245820 }, { "epoch": 156.67941363926067, "grad_norm": 0.23721922934055328, "learning_rate": 8.652478109507422e-06, "loss": 0.0158, "step": 245830 }, { "epoch": 156.68578712555768, "grad_norm": 0.21041728556156158, "learning_rate": 8.649379345904735e-06, "loss": 0.0037, "step": 245840 }, { "epoch": 156.6921606118547, "grad_norm": 0.3649423122406006, "learning_rate": 8.646281084751129e-06, "loss": 0.0063, "step": 245850 }, { "epoch": 156.6985340981517, "grad_norm": 0.3835855722427368, "learning_rate": 8.643183326084236e-06, "loss": 0.0074, "step": 245860 }, { "epoch": 156.7049075844487, "grad_norm": 0.5684453248977661, "learning_rate": 8.640086069941727e-06, "loss": 0.0053, "step": 245870 }, { "epoch": 156.71128107074568, "grad_norm": 0.41792386770248413, "learning_rate": 8.636989316361221e-06, "loss": 0.0238, "step": 245880 }, { "epoch": 156.7176545570427, "grad_norm": 0.1444835662841797, "learning_rate": 8.633893065380355e-06, "loss": 0.0068, "step": 245890 }, { "epoch": 156.7240280433397, "grad_norm": 0.1927569955587387, "learning_rate": 8.630797317036726e-06, "loss": 0.0036, "step": 245900 }, { "epoch": 156.7304015296367, "grad_norm": 0.18594269454479218, "learning_rate": 8.627702071367993e-06, "loss": 0.0022, "step": 245910 }, { "epoch": 156.73677501593372, "grad_norm": 0.09610047191381454, "learning_rate": 8.62460732841172e-06, "loss": 0.0054, "step": 245920 }, { "epoch": 156.74314850223072, "grad_norm": 0.13032053411006927, "learning_rate": 8.621513088205541e-06, "loss": 0.007, "step": 245930 }, { "epoch": 156.74952198852773, "grad_norm": 0.3036259114742279, "learning_rate": 8.618419350787039e-06, "loss": 0.0048, "step": 245940 }, { "epoch": 156.75589547482474, "grad_norm": 0.26232218742370605, "learning_rate": 8.615326116193833e-06, "loss": 0.0103, "step": 245950 }, { "epoch": 156.76226896112172, "grad_norm": 0.268343448638916, "learning_rate": 8.612233384463465e-06, "loss": 0.0156, "step": 245960 }, { "epoch": 156.76864244741873, "grad_norm": 0.11018366366624832, "learning_rate": 8.609141155633548e-06, "loss": 0.0025, "step": 245970 }, { "epoch": 156.77501593371574, "grad_norm": 0.1562703400850296, "learning_rate": 8.606049429741653e-06, "loss": 0.0048, "step": 245980 }, { "epoch": 156.78138942001274, "grad_norm": 0.5612740516662598, "learning_rate": 8.602958206825334e-06, "loss": 0.0137, "step": 245990 }, { "epoch": 156.78776290630975, "grad_norm": 0.5678074359893799, "learning_rate": 8.599867486922142e-06, "loss": 0.009, "step": 246000 }, { "epoch": 156.79413639260676, "grad_norm": 0.5891150236129761, "learning_rate": 8.596777270069666e-06, "loss": 0.0084, "step": 246010 }, { "epoch": 156.80050987890377, "grad_norm": 0.555412232875824, "learning_rate": 8.593687556305436e-06, "loss": 0.01, "step": 246020 }, { "epoch": 156.80688336520078, "grad_norm": 0.1644972711801529, "learning_rate": 8.590598345667e-06, "loss": 0.0033, "step": 246030 }, { "epoch": 156.81325685149778, "grad_norm": 0.36643755435943604, "learning_rate": 8.58750963819187e-06, "loss": 0.0045, "step": 246040 }, { "epoch": 156.81963033779476, "grad_norm": 0.19904690980911255, "learning_rate": 8.584421433917616e-06, "loss": 0.0078, "step": 246050 }, { "epoch": 156.82600382409177, "grad_norm": 0.13219456374645233, "learning_rate": 8.581333732881746e-06, "loss": 0.0052, "step": 246060 }, { "epoch": 156.83237731038878, "grad_norm": 0.08515002578496933, "learning_rate": 8.578246535121775e-06, "loss": 0.0069, "step": 246070 }, { "epoch": 156.8387507966858, "grad_norm": 0.09570056200027466, "learning_rate": 8.575159840675201e-06, "loss": 0.0028, "step": 246080 }, { "epoch": 156.8451242829828, "grad_norm": 0.06733711808919907, "learning_rate": 8.572073649579576e-06, "loss": 0.0088, "step": 246090 }, { "epoch": 156.8514977692798, "grad_norm": 0.21634441614151, "learning_rate": 8.568987961872343e-06, "loss": 0.0078, "step": 246100 }, { "epoch": 156.8578712555768, "grad_norm": 0.18247312307357788, "learning_rate": 8.565902777591039e-06, "loss": 0.0135, "step": 246110 }, { "epoch": 156.86424474187382, "grad_norm": 0.3333200216293335, "learning_rate": 8.562818096773123e-06, "loss": 0.0055, "step": 246120 }, { "epoch": 156.8706182281708, "grad_norm": 0.2520826458930969, "learning_rate": 8.559733919456115e-06, "loss": 0.0039, "step": 246130 }, { "epoch": 156.8769917144678, "grad_norm": 0.1581239551305771, "learning_rate": 8.55665024567744e-06, "loss": 0.0045, "step": 246140 }, { "epoch": 156.8833652007648, "grad_norm": 0.16351717710494995, "learning_rate": 8.553567075474605e-06, "loss": 0.0031, "step": 246150 }, { "epoch": 156.88973868706182, "grad_norm": 0.49117469787597656, "learning_rate": 8.550484408885062e-06, "loss": 0.007, "step": 246160 }, { "epoch": 156.89611217335883, "grad_norm": 0.19021213054656982, "learning_rate": 8.547402245946273e-06, "loss": 0.0081, "step": 246170 }, { "epoch": 156.90248565965584, "grad_norm": 0.13643525540828705, "learning_rate": 8.54432058669567e-06, "loss": 0.0069, "step": 246180 }, { "epoch": 156.90885914595285, "grad_norm": 0.3834240436553955, "learning_rate": 8.541239431170723e-06, "loss": 0.0085, "step": 246190 }, { "epoch": 156.91523263224985, "grad_norm": 0.1280355602502823, "learning_rate": 8.538158779408861e-06, "loss": 0.0049, "step": 246200 }, { "epoch": 156.92160611854683, "grad_norm": 0.1441369652748108, "learning_rate": 8.535078631447518e-06, "loss": 0.0071, "step": 246210 }, { "epoch": 156.92797960484384, "grad_norm": 0.6290425658226013, "learning_rate": 8.531998987324102e-06, "loss": 0.0072, "step": 246220 }, { "epoch": 156.93435309114085, "grad_norm": 0.4014973044395447, "learning_rate": 8.528919847076077e-06, "loss": 0.0084, "step": 246230 }, { "epoch": 156.94072657743786, "grad_norm": 0.07859345525503159, "learning_rate": 8.525841210740809e-06, "loss": 0.0036, "step": 246240 }, { "epoch": 156.94710006373487, "grad_norm": 0.2313801646232605, "learning_rate": 8.52276307835574e-06, "loss": 0.005, "step": 246250 }, { "epoch": 156.95347355003187, "grad_norm": 0.14955779910087585, "learning_rate": 8.519685449958248e-06, "loss": 0.0074, "step": 246260 }, { "epoch": 156.95984703632888, "grad_norm": 0.09999017417430878, "learning_rate": 8.516608325585768e-06, "loss": 0.0052, "step": 246270 }, { "epoch": 156.9662205226259, "grad_norm": 0.47228333353996277, "learning_rate": 8.513531705275635e-06, "loss": 0.0051, "step": 246280 }, { "epoch": 156.97259400892287, "grad_norm": 0.10737603902816772, "learning_rate": 8.510455589065275e-06, "loss": 0.0049, "step": 246290 }, { "epoch": 156.97896749521988, "grad_norm": 0.16509896516799927, "learning_rate": 8.507379976992042e-06, "loss": 0.0087, "step": 246300 }, { "epoch": 156.98534098151688, "grad_norm": 0.22981876134872437, "learning_rate": 8.504304869093343e-06, "loss": 0.0067, "step": 246310 }, { "epoch": 156.9917144678139, "grad_norm": 0.175724059343338, "learning_rate": 8.50123026540649e-06, "loss": 0.0097, "step": 246320 }, { "epoch": 156.9980879541109, "grad_norm": 0.2273656725883484, "learning_rate": 8.498156165968885e-06, "loss": 0.0043, "step": 246330 }, { "epoch": 157.0044614404079, "grad_norm": 0.20428447425365448, "learning_rate": 8.49508257081787e-06, "loss": 0.01, "step": 246340 }, { "epoch": 157.01083492670492, "grad_norm": 0.24522468447685242, "learning_rate": 8.492009479990782e-06, "loss": 0.0055, "step": 246350 }, { "epoch": 157.01720841300192, "grad_norm": 0.14402510225772858, "learning_rate": 8.488936893524962e-06, "loss": 0.004, "step": 246360 }, { "epoch": 157.0235818992989, "grad_norm": 0.09545891731977463, "learning_rate": 8.485864811457761e-06, "loss": 0.007, "step": 246370 }, { "epoch": 157.0299553855959, "grad_norm": 0.18048302829265594, "learning_rate": 8.482793233826503e-06, "loss": 0.0042, "step": 246380 }, { "epoch": 157.03632887189292, "grad_norm": 0.11721290647983551, "learning_rate": 8.479722160668502e-06, "loss": 0.0039, "step": 246390 }, { "epoch": 157.04270235818993, "grad_norm": 0.1944897621870041, "learning_rate": 8.476651592021062e-06, "loss": 0.0051, "step": 246400 }, { "epoch": 157.04907584448694, "grad_norm": 0.1394328474998474, "learning_rate": 8.473581527921525e-06, "loss": 0.0039, "step": 246410 }, { "epoch": 157.05544933078394, "grad_norm": 0.34833553433418274, "learning_rate": 8.470511968407185e-06, "loss": 0.0057, "step": 246420 }, { "epoch": 157.06182281708095, "grad_norm": 0.11988784372806549, "learning_rate": 8.467442913515333e-06, "loss": 0.008, "step": 246430 }, { "epoch": 157.06819630337796, "grad_norm": 0.2110205590724945, "learning_rate": 8.464374363283245e-06, "loss": 0.0082, "step": 246440 }, { "epoch": 157.07456978967494, "grad_norm": 0.08296080678701401, "learning_rate": 8.461306317748242e-06, "loss": 0.0053, "step": 246450 }, { "epoch": 157.08094327597195, "grad_norm": 0.2559543251991272, "learning_rate": 8.458238776947585e-06, "loss": 0.0054, "step": 246460 }, { "epoch": 157.08731676226896, "grad_norm": 0.125836580991745, "learning_rate": 8.455171740918549e-06, "loss": 0.0024, "step": 246470 }, { "epoch": 157.09369024856596, "grad_norm": 0.16120144724845886, "learning_rate": 8.452105209698402e-06, "loss": 0.0054, "step": 246480 }, { "epoch": 157.10006373486297, "grad_norm": 0.26803624629974365, "learning_rate": 8.449039183324393e-06, "loss": 0.008, "step": 246490 }, { "epoch": 157.10643722115998, "grad_norm": 0.3575916290283203, "learning_rate": 8.445973661833794e-06, "loss": 0.0075, "step": 246500 }, { "epoch": 157.112810707457, "grad_norm": 0.16366884112358093, "learning_rate": 8.442908645263858e-06, "loss": 0.0054, "step": 246510 }, { "epoch": 157.119184193754, "grad_norm": 0.1369766891002655, "learning_rate": 8.439844133651815e-06, "loss": 0.0044, "step": 246520 }, { "epoch": 157.12555768005097, "grad_norm": 0.15052731335163116, "learning_rate": 8.436780127034888e-06, "loss": 0.0067, "step": 246530 }, { "epoch": 157.13193116634798, "grad_norm": 0.250613272190094, "learning_rate": 8.433716625450339e-06, "loss": 0.0063, "step": 246540 }, { "epoch": 157.138304652645, "grad_norm": 0.15986096858978271, "learning_rate": 8.43065362893538e-06, "loss": 0.0095, "step": 246550 }, { "epoch": 157.144678138942, "grad_norm": 0.10279302299022675, "learning_rate": 8.42759113752723e-06, "loss": 0.0051, "step": 246560 }, { "epoch": 157.151051625239, "grad_norm": 0.1906105875968933, "learning_rate": 8.424529151263082e-06, "loss": 0.0058, "step": 246570 }, { "epoch": 157.157425111536, "grad_norm": 0.11767802387475967, "learning_rate": 8.421467670180173e-06, "loss": 0.0057, "step": 246580 }, { "epoch": 157.16379859783302, "grad_norm": 0.10285234451293945, "learning_rate": 8.418406694315689e-06, "loss": 0.0028, "step": 246590 }, { "epoch": 157.17017208413003, "grad_norm": 0.16644039750099182, "learning_rate": 8.415346223706822e-06, "loss": 0.0047, "step": 246600 }, { "epoch": 157.176545570427, "grad_norm": 0.15117670595645905, "learning_rate": 8.412286258390745e-06, "loss": 0.0053, "step": 246610 }, { "epoch": 157.18291905672402, "grad_norm": 0.08214733749628067, "learning_rate": 8.409226798404674e-06, "loss": 0.0053, "step": 246620 }, { "epoch": 157.18929254302103, "grad_norm": 0.17791637778282166, "learning_rate": 8.406167843785762e-06, "loss": 0.0055, "step": 246630 }, { "epoch": 157.19566602931803, "grad_norm": 0.12382975965738297, "learning_rate": 8.40310939457118e-06, "loss": 0.0042, "step": 246640 }, { "epoch": 157.20203951561504, "grad_norm": 0.49937117099761963, "learning_rate": 8.40005145079808e-06, "loss": 0.0091, "step": 246650 }, { "epoch": 157.20841300191205, "grad_norm": 0.28793957829475403, "learning_rate": 8.396994012503661e-06, "loss": 0.0153, "step": 246660 }, { "epoch": 157.21478648820906, "grad_norm": 0.3681637644767761, "learning_rate": 8.393937079725017e-06, "loss": 0.0067, "step": 246670 }, { "epoch": 157.22115997450607, "grad_norm": 0.16658098995685577, "learning_rate": 8.390880652499334e-06, "loss": 0.0093, "step": 246680 }, { "epoch": 157.22753346080307, "grad_norm": 0.06714574247598648, "learning_rate": 8.387824730863726e-06, "loss": 0.0037, "step": 246690 }, { "epoch": 157.23390694710005, "grad_norm": 0.2315053641796112, "learning_rate": 8.384769314855357e-06, "loss": 0.0084, "step": 246700 }, { "epoch": 157.24028043339706, "grad_norm": 0.39341574907302856, "learning_rate": 8.381714404511309e-06, "loss": 0.0054, "step": 246710 }, { "epoch": 157.24665391969407, "grad_norm": 0.17028139531612396, "learning_rate": 8.378659999868738e-06, "loss": 0.0048, "step": 246720 }, { "epoch": 157.25302740599108, "grad_norm": 0.1005636528134346, "learning_rate": 8.375606100964745e-06, "loss": 0.0041, "step": 246730 }, { "epoch": 157.25940089228808, "grad_norm": 0.4225819408893585, "learning_rate": 8.372552707836434e-06, "loss": 0.0061, "step": 246740 }, { "epoch": 157.2657743785851, "grad_norm": 0.15754415094852448, "learning_rate": 8.369499820520905e-06, "loss": 0.0054, "step": 246750 }, { "epoch": 157.2721478648821, "grad_norm": 0.08095015585422516, "learning_rate": 8.366447439055265e-06, "loss": 0.0049, "step": 246760 }, { "epoch": 157.2785213511791, "grad_norm": 0.12546943128108978, "learning_rate": 8.3633955634766e-06, "loss": 0.0034, "step": 246770 }, { "epoch": 157.2848948374761, "grad_norm": 0.25672274827957153, "learning_rate": 8.36034419382199e-06, "loss": 0.0058, "step": 246780 }, { "epoch": 157.2912683237731, "grad_norm": 0.08947079628705978, "learning_rate": 8.357293330128495e-06, "loss": 0.0032, "step": 246790 }, { "epoch": 157.2976418100701, "grad_norm": 0.12651129066944122, "learning_rate": 8.354242972433234e-06, "loss": 0.0057, "step": 246800 }, { "epoch": 157.3040152963671, "grad_norm": 0.10042692720890045, "learning_rate": 8.351193120773215e-06, "loss": 0.0062, "step": 246810 }, { "epoch": 157.31038878266412, "grad_norm": 0.1595425009727478, "learning_rate": 8.348143775185536e-06, "loss": 0.0024, "step": 246820 }, { "epoch": 157.31676226896113, "grad_norm": 0.34969210624694824, "learning_rate": 8.345094935707227e-06, "loss": 0.0056, "step": 246830 }, { "epoch": 157.32313575525814, "grad_norm": 0.07381212711334229, "learning_rate": 8.342046602375358e-06, "loss": 0.0091, "step": 246840 }, { "epoch": 157.32950924155514, "grad_norm": 0.22792819142341614, "learning_rate": 8.338998775226936e-06, "loss": 0.0068, "step": 246850 }, { "epoch": 157.33588272785212, "grad_norm": 0.26226356625556946, "learning_rate": 8.335951454299029e-06, "loss": 0.0098, "step": 246860 }, { "epoch": 157.34225621414913, "grad_norm": 0.26091504096984863, "learning_rate": 8.332904639628647e-06, "loss": 0.0062, "step": 246870 }, { "epoch": 157.34862970044614, "grad_norm": 0.28442952036857605, "learning_rate": 8.32985833125281e-06, "loss": 0.0061, "step": 246880 }, { "epoch": 157.35500318674315, "grad_norm": 0.0713464766740799, "learning_rate": 8.326812529208527e-06, "loss": 0.0047, "step": 246890 }, { "epoch": 157.36137667304016, "grad_norm": 0.05965150147676468, "learning_rate": 8.323767233532831e-06, "loss": 0.0045, "step": 246900 }, { "epoch": 157.36775015933716, "grad_norm": 0.2027149349451065, "learning_rate": 8.320722444262713e-06, "loss": 0.0048, "step": 246910 }, { "epoch": 157.37412364563417, "grad_norm": 0.22049428522586823, "learning_rate": 8.317678161435166e-06, "loss": 0.0039, "step": 246920 }, { "epoch": 157.38049713193118, "grad_norm": 0.44241124391555786, "learning_rate": 8.314634385087173e-06, "loss": 0.0094, "step": 246930 }, { "epoch": 157.38687061822816, "grad_norm": 0.0978999137878418, "learning_rate": 8.31159111525574e-06, "loss": 0.0103, "step": 246940 }, { "epoch": 157.39324410452517, "grad_norm": 0.20475277304649353, "learning_rate": 8.308548351977841e-06, "loss": 0.0036, "step": 246950 }, { "epoch": 157.39961759082217, "grad_norm": 0.09635647386312485, "learning_rate": 8.30550609529044e-06, "loss": 0.0034, "step": 246960 }, { "epoch": 157.40599107711918, "grad_norm": 0.21706771850585938, "learning_rate": 8.302464345230492e-06, "loss": 0.0083, "step": 246970 }, { "epoch": 157.4123645634162, "grad_norm": 0.5029513835906982, "learning_rate": 8.299423101834997e-06, "loss": 0.0079, "step": 246980 }, { "epoch": 157.4187380497132, "grad_norm": 0.08928275853395462, "learning_rate": 8.296382365140858e-06, "loss": 0.0036, "step": 246990 }, { "epoch": 157.4251115360102, "grad_norm": 0.29298070073127747, "learning_rate": 8.293342135185061e-06, "loss": 0.0063, "step": 247000 }, { "epoch": 157.4314850223072, "grad_norm": 0.14029574394226074, "learning_rate": 8.29030241200452e-06, "loss": 0.0059, "step": 247010 }, { "epoch": 157.4378585086042, "grad_norm": 0.2333037108182907, "learning_rate": 8.287263195636213e-06, "loss": 0.0037, "step": 247020 }, { "epoch": 157.4442319949012, "grad_norm": 0.15053924918174744, "learning_rate": 8.284224486117015e-06, "loss": 0.0051, "step": 247030 }, { "epoch": 157.4506054811982, "grad_norm": 0.08965332061052322, "learning_rate": 8.281186283483889e-06, "loss": 0.0086, "step": 247040 }, { "epoch": 157.45697896749522, "grad_norm": 0.14485067129135132, "learning_rate": 8.278148587773737e-06, "loss": 0.0124, "step": 247050 }, { "epoch": 157.46335245379223, "grad_norm": 0.2213018238544464, "learning_rate": 8.275111399023477e-06, "loss": 0.0036, "step": 247060 }, { "epoch": 157.46972594008923, "grad_norm": 0.1777385026216507, "learning_rate": 8.27207471726999e-06, "loss": 0.0066, "step": 247070 }, { "epoch": 157.47609942638624, "grad_norm": 0.09867897629737854, "learning_rate": 8.26903854255021e-06, "loss": 0.0063, "step": 247080 }, { "epoch": 157.48247291268325, "grad_norm": 0.06878365576267242, "learning_rate": 8.266002874901008e-06, "loss": 0.0057, "step": 247090 }, { "epoch": 157.48884639898023, "grad_norm": 0.06164753437042236, "learning_rate": 8.262967714359277e-06, "loss": 0.0041, "step": 247100 }, { "epoch": 157.49521988527724, "grad_norm": 0.14551801979541779, "learning_rate": 8.259933060961883e-06, "loss": 0.0052, "step": 247110 }, { "epoch": 157.50159337157424, "grad_norm": 0.13551150262355804, "learning_rate": 8.25689891474572e-06, "loss": 0.0062, "step": 247120 }, { "epoch": 157.50796685787125, "grad_norm": 0.15169855952262878, "learning_rate": 8.253865275747652e-06, "loss": 0.0071, "step": 247130 }, { "epoch": 157.51434034416826, "grad_norm": 0.25679096579551697, "learning_rate": 8.250832144004534e-06, "loss": 0.0112, "step": 247140 }, { "epoch": 157.52071383046527, "grad_norm": 0.34545019268989563, "learning_rate": 8.247799519553206e-06, "loss": 0.0045, "step": 247150 }, { "epoch": 157.52708731676228, "grad_norm": 0.09170223772525787, "learning_rate": 8.244767402430553e-06, "loss": 0.0189, "step": 247160 }, { "epoch": 157.53346080305928, "grad_norm": 0.09265198558568954, "learning_rate": 8.241735792673393e-06, "loss": 0.013, "step": 247170 }, { "epoch": 157.53983428935626, "grad_norm": 0.16836288571357727, "learning_rate": 8.238704690318578e-06, "loss": 0.0055, "step": 247180 }, { "epoch": 157.54620777565327, "grad_norm": 0.09581422060728073, "learning_rate": 8.235674095402924e-06, "loss": 0.0071, "step": 247190 }, { "epoch": 157.55258126195028, "grad_norm": 0.26170235872268677, "learning_rate": 8.232644007963253e-06, "loss": 0.0053, "step": 247200 }, { "epoch": 157.5589547482473, "grad_norm": 0.3238148093223572, "learning_rate": 8.229614428036402e-06, "loss": 0.0194, "step": 247210 }, { "epoch": 157.5653282345443, "grad_norm": 0.1911047250032425, "learning_rate": 8.226585355659178e-06, "loss": 0.0059, "step": 247220 }, { "epoch": 157.5717017208413, "grad_norm": 0.24100244045257568, "learning_rate": 8.223556790868374e-06, "loss": 0.0087, "step": 247230 }, { "epoch": 157.5780752071383, "grad_norm": 0.24493950605392456, "learning_rate": 8.220528733700795e-06, "loss": 0.0025, "step": 247240 }, { "epoch": 157.58444869343532, "grad_norm": 0.3814818859100342, "learning_rate": 8.217501184193244e-06, "loss": 0.0077, "step": 247250 }, { "epoch": 157.5908221797323, "grad_norm": 0.27398166060447693, "learning_rate": 8.214474142382506e-06, "loss": 0.0188, "step": 247260 }, { "epoch": 157.5971956660293, "grad_norm": 0.12433161586523056, "learning_rate": 8.211447608305362e-06, "loss": 0.0036, "step": 247270 }, { "epoch": 157.60356915232632, "grad_norm": 0.14989040791988373, "learning_rate": 8.208421581998566e-06, "loss": 0.004, "step": 247280 }, { "epoch": 157.60994263862332, "grad_norm": 0.2395733743906021, "learning_rate": 8.205396063498921e-06, "loss": 0.005, "step": 247290 }, { "epoch": 157.61631612492033, "grad_norm": 0.09656944870948792, "learning_rate": 8.202371052843177e-06, "loss": 0.0036, "step": 247300 }, { "epoch": 157.62268961121734, "grad_norm": 0.290469229221344, "learning_rate": 8.199346550068087e-06, "loss": 0.008, "step": 247310 }, { "epoch": 157.62906309751435, "grad_norm": 0.14345009624958038, "learning_rate": 8.19632255521039e-06, "loss": 0.009, "step": 247320 }, { "epoch": 157.63543658381136, "grad_norm": 0.2066667228937149, "learning_rate": 8.193299068306853e-06, "loss": 0.0037, "step": 247330 }, { "epoch": 157.64181007010836, "grad_norm": 0.27336013317108154, "learning_rate": 8.190276089394205e-06, "loss": 0.0047, "step": 247340 }, { "epoch": 157.64818355640534, "grad_norm": 0.07246526330709457, "learning_rate": 8.187253618509177e-06, "loss": 0.0033, "step": 247350 }, { "epoch": 157.65455704270235, "grad_norm": 0.3645360767841339, "learning_rate": 8.184231655688484e-06, "loss": 0.0032, "step": 247360 }, { "epoch": 157.66093052899936, "grad_norm": 0.155236154794693, "learning_rate": 8.181210200968887e-06, "loss": 0.0063, "step": 247370 }, { "epoch": 157.66730401529637, "grad_norm": 0.11469491571187973, "learning_rate": 8.178189254387036e-06, "loss": 0.0088, "step": 247380 }, { "epoch": 157.67367750159337, "grad_norm": 0.3662360906600952, "learning_rate": 8.17516881597969e-06, "loss": 0.006, "step": 247390 }, { "epoch": 157.68005098789038, "grad_norm": 0.31948429346084595, "learning_rate": 8.172148885783514e-06, "loss": 0.0041, "step": 247400 }, { "epoch": 157.6864244741874, "grad_norm": 0.11008699238300323, "learning_rate": 8.16912946383525e-06, "loss": 0.0074, "step": 247410 }, { "epoch": 157.6927979604844, "grad_norm": 0.32934099435806274, "learning_rate": 8.166110550171535e-06, "loss": 0.0025, "step": 247420 }, { "epoch": 157.69917144678138, "grad_norm": 0.3431851267814636, "learning_rate": 8.163092144829088e-06, "loss": 0.0045, "step": 247430 }, { "epoch": 157.7055449330784, "grad_norm": 0.1757257878780365, "learning_rate": 8.160074247844568e-06, "loss": 0.0048, "step": 247440 }, { "epoch": 157.7119184193754, "grad_norm": 0.14687000215053558, "learning_rate": 8.157056859254647e-06, "loss": 0.0105, "step": 247450 }, { "epoch": 157.7182919056724, "grad_norm": 0.2863249182701111, "learning_rate": 8.154039979095984e-06, "loss": 0.0071, "step": 247460 }, { "epoch": 157.7246653919694, "grad_norm": 0.10333648324012756, "learning_rate": 8.151023607405256e-06, "loss": 0.008, "step": 247470 }, { "epoch": 157.73103887826642, "grad_norm": 0.07823594659566879, "learning_rate": 8.1480077442191e-06, "loss": 0.0045, "step": 247480 }, { "epoch": 157.73741236456343, "grad_norm": 0.5380642414093018, "learning_rate": 8.144992389574168e-06, "loss": 0.007, "step": 247490 }, { "epoch": 157.74378585086043, "grad_norm": 0.4902053475379944, "learning_rate": 8.141977543507084e-06, "loss": 0.0143, "step": 247500 }, { "epoch": 157.7501593371574, "grad_norm": 0.4345225989818573, "learning_rate": 8.1389632060545e-06, "loss": 0.0101, "step": 247510 }, { "epoch": 157.75653282345442, "grad_norm": 0.3025950789451599, "learning_rate": 8.135949377253044e-06, "loss": 0.0049, "step": 247520 }, { "epoch": 157.76290630975143, "grad_norm": 0.26992887258529663, "learning_rate": 8.132936057139323e-06, "loss": 0.0076, "step": 247530 }, { "epoch": 157.76927979604844, "grad_norm": 0.3345283269882202, "learning_rate": 8.129923245749955e-06, "loss": 0.0072, "step": 247540 }, { "epoch": 157.77565328234544, "grad_norm": 0.13480901718139648, "learning_rate": 8.126910943121568e-06, "loss": 0.0048, "step": 247550 }, { "epoch": 157.78202676864245, "grad_norm": 0.18177825212478638, "learning_rate": 8.123899149290725e-06, "loss": 0.0056, "step": 247560 }, { "epoch": 157.78840025493946, "grad_norm": 0.16172954440116882, "learning_rate": 8.120887864294057e-06, "loss": 0.0077, "step": 247570 }, { "epoch": 157.79477374123647, "grad_norm": 0.24053344130516052, "learning_rate": 8.11787708816813e-06, "loss": 0.0038, "step": 247580 }, { "epoch": 157.80114722753345, "grad_norm": 0.13926604390144348, "learning_rate": 8.114866820949563e-06, "loss": 0.0053, "step": 247590 }, { "epoch": 157.80752071383046, "grad_norm": 0.22843950986862183, "learning_rate": 8.111857062674888e-06, "loss": 0.0073, "step": 247600 }, { "epoch": 157.81389420012746, "grad_norm": 0.1065191775560379, "learning_rate": 8.10884781338071e-06, "loss": 0.003, "step": 247610 }, { "epoch": 157.82026768642447, "grad_norm": 0.1265239715576172, "learning_rate": 8.105839073103583e-06, "loss": 0.007, "step": 247620 }, { "epoch": 157.82664117272148, "grad_norm": 0.16871027648448944, "learning_rate": 8.10283084188006e-06, "loss": 0.006, "step": 247630 }, { "epoch": 157.8330146590185, "grad_norm": 0.1453152894973755, "learning_rate": 8.099823119746697e-06, "loss": 0.0046, "step": 247640 }, { "epoch": 157.8393881453155, "grad_norm": 0.5710442066192627, "learning_rate": 8.096815906740052e-06, "loss": 0.0102, "step": 247650 }, { "epoch": 157.8457616316125, "grad_norm": 0.2301032841205597, "learning_rate": 8.093809202896658e-06, "loss": 0.0215, "step": 247660 }, { "epoch": 157.85213511790948, "grad_norm": 0.28420811891555786, "learning_rate": 8.090803008253046e-06, "loss": 0.0073, "step": 247670 }, { "epoch": 157.8585086042065, "grad_norm": 0.23472946882247925, "learning_rate": 8.087797322845731e-06, "loss": 0.0058, "step": 247680 }, { "epoch": 157.8648820905035, "grad_norm": 0.15643136203289032, "learning_rate": 8.08479214671128e-06, "loss": 0.0166, "step": 247690 }, { "epoch": 157.8712555768005, "grad_norm": 0.1126343309879303, "learning_rate": 8.081787479886154e-06, "loss": 0.0053, "step": 247700 }, { "epoch": 157.87762906309752, "grad_norm": 0.13131140172481537, "learning_rate": 8.078783322406896e-06, "loss": 0.0156, "step": 247710 }, { "epoch": 157.88400254939452, "grad_norm": 0.10864916443824768, "learning_rate": 8.07577967430999e-06, "loss": 0.0065, "step": 247720 }, { "epoch": 157.89037603569153, "grad_norm": 0.3276653587818146, "learning_rate": 8.072776535631971e-06, "loss": 0.0061, "step": 247730 }, { "epoch": 157.89674952198854, "grad_norm": 0.25712811946868896, "learning_rate": 8.069773906409279e-06, "loss": 0.0069, "step": 247740 }, { "epoch": 157.90312300828552, "grad_norm": 0.24003072082996368, "learning_rate": 8.066771786678434e-06, "loss": 0.0041, "step": 247750 }, { "epoch": 157.90949649458253, "grad_norm": 0.20890478789806366, "learning_rate": 8.0637701764759e-06, "loss": 0.0051, "step": 247760 }, { "epoch": 157.91586998087953, "grad_norm": 0.342383474111557, "learning_rate": 8.060769075838154e-06, "loss": 0.0063, "step": 247770 }, { "epoch": 157.92224346717654, "grad_norm": 0.8229806423187256, "learning_rate": 8.057768484801654e-06, "loss": 0.0069, "step": 247780 }, { "epoch": 157.92861695347355, "grad_norm": 1.0271625518798828, "learning_rate": 8.054768403402874e-06, "loss": 0.0059, "step": 247790 }, { "epoch": 157.93499043977056, "grad_norm": 0.16665276885032654, "learning_rate": 8.051768831678258e-06, "loss": 0.0029, "step": 247800 }, { "epoch": 157.94136392606757, "grad_norm": 0.17091815173625946, "learning_rate": 8.04876976966426e-06, "loss": 0.0029, "step": 247810 }, { "epoch": 157.94773741236457, "grad_norm": 0.07985825091600418, "learning_rate": 8.045771217397307e-06, "loss": 0.0037, "step": 247820 }, { "epoch": 157.95411089866155, "grad_norm": 0.1490887552499771, "learning_rate": 8.042773174913859e-06, "loss": 0.0031, "step": 247830 }, { "epoch": 157.96048438495856, "grad_norm": 0.190043643116951, "learning_rate": 8.039775642250325e-06, "loss": 0.0064, "step": 247840 }, { "epoch": 157.96685787125557, "grad_norm": 0.3319579064846039, "learning_rate": 8.036778619443137e-06, "loss": 0.009, "step": 247850 }, { "epoch": 157.97323135755258, "grad_norm": 0.210440993309021, "learning_rate": 8.033782106528698e-06, "loss": 0.0064, "step": 247860 }, { "epoch": 157.9796048438496, "grad_norm": 0.14524826407432556, "learning_rate": 8.030786103543441e-06, "loss": 0.0072, "step": 247870 }, { "epoch": 157.9859783301466, "grad_norm": 0.18050064146518707, "learning_rate": 8.027790610523756e-06, "loss": 0.0054, "step": 247880 }, { "epoch": 157.9923518164436, "grad_norm": 0.2862391769886017, "learning_rate": 8.024795627506044e-06, "loss": 0.0141, "step": 247890 }, { "epoch": 157.9987253027406, "grad_norm": 0.13313381373882294, "learning_rate": 8.021801154526687e-06, "loss": 0.0044, "step": 247900 }, { "epoch": 158.00509878903762, "grad_norm": 0.15967056155204773, "learning_rate": 8.018807191622084e-06, "loss": 0.0086, "step": 247910 }, { "epoch": 158.0114722753346, "grad_norm": 0.40269067883491516, "learning_rate": 8.015813738828621e-06, "loss": 0.0085, "step": 247920 }, { "epoch": 158.0178457616316, "grad_norm": 0.2882905602455139, "learning_rate": 8.012820796182658e-06, "loss": 0.011, "step": 247930 }, { "epoch": 158.0242192479286, "grad_norm": 0.21599845588207245, "learning_rate": 8.009828363720561e-06, "loss": 0.0058, "step": 247940 }, { "epoch": 158.03059273422562, "grad_norm": 0.3939744532108307, "learning_rate": 8.006836441478683e-06, "loss": 0.0052, "step": 247950 }, { "epoch": 158.03696622052263, "grad_norm": 0.1756334751844406, "learning_rate": 8.003845029493407e-06, "loss": 0.0057, "step": 247960 }, { "epoch": 158.04333970681964, "grad_norm": 0.25501734018325806, "learning_rate": 8.00085412780106e-06, "loss": 0.0144, "step": 247970 }, { "epoch": 158.04971319311665, "grad_norm": 0.13021136820316315, "learning_rate": 7.997863736437989e-06, "loss": 0.008, "step": 247980 }, { "epoch": 158.05608667941365, "grad_norm": 0.14964549243450165, "learning_rate": 7.994873855440522e-06, "loss": 0.0035, "step": 247990 }, { "epoch": 158.06246016571063, "grad_norm": 0.09412121027708054, "learning_rate": 7.991884484845003e-06, "loss": 0.0078, "step": 248000 }, { "epoch": 158.06883365200764, "grad_norm": 0.34908902645111084, "learning_rate": 7.988895624687753e-06, "loss": 0.0065, "step": 248010 }, { "epoch": 158.07520713830465, "grad_norm": 0.08941590785980225, "learning_rate": 7.98590727500509e-06, "loss": 0.0058, "step": 248020 }, { "epoch": 158.08158062460166, "grad_norm": 0.12660452723503113, "learning_rate": 7.982919435833308e-06, "loss": 0.0041, "step": 248030 }, { "epoch": 158.08795411089866, "grad_norm": 0.09559118002653122, "learning_rate": 7.979932107208732e-06, "loss": 0.0068, "step": 248040 }, { "epoch": 158.09432759719567, "grad_norm": 0.15911369025707245, "learning_rate": 7.97694528916766e-06, "loss": 0.0082, "step": 248050 }, { "epoch": 158.10070108349268, "grad_norm": 0.2501651346683502, "learning_rate": 7.973958981746377e-06, "loss": 0.005, "step": 248060 }, { "epoch": 158.1070745697897, "grad_norm": 0.18535149097442627, "learning_rate": 7.970973184981162e-06, "loss": 0.0046, "step": 248070 }, { "epoch": 158.11344805608667, "grad_norm": 0.28055137395858765, "learning_rate": 7.967987898908325e-06, "loss": 0.0155, "step": 248080 }, { "epoch": 158.11982154238368, "grad_norm": 0.5511283874511719, "learning_rate": 7.965003123564096e-06, "loss": 0.0084, "step": 248090 }, { "epoch": 158.12619502868068, "grad_norm": 0.19054380059242249, "learning_rate": 7.962018858984782e-06, "loss": 0.004, "step": 248100 }, { "epoch": 158.1325685149777, "grad_norm": 0.1662248820066452, "learning_rate": 7.959035105206613e-06, "loss": 0.0057, "step": 248110 }, { "epoch": 158.1389420012747, "grad_norm": 0.14824172854423523, "learning_rate": 7.95605186226589e-06, "loss": 0.0027, "step": 248120 }, { "epoch": 158.1453154875717, "grad_norm": 0.14488065242767334, "learning_rate": 7.953069130198803e-06, "loss": 0.0085, "step": 248130 }, { "epoch": 158.15168897386872, "grad_norm": 0.25594961643218994, "learning_rate": 7.950086909041643e-06, "loss": 0.0077, "step": 248140 }, { "epoch": 158.15806246016572, "grad_norm": 0.20021477341651917, "learning_rate": 7.947105198830612e-06, "loss": 0.0032, "step": 248150 }, { "epoch": 158.1644359464627, "grad_norm": 0.36020907759666443, "learning_rate": 7.944123999601987e-06, "loss": 0.0084, "step": 248160 }, { "epoch": 158.1708094327597, "grad_norm": 0.11312484741210938, "learning_rate": 7.941143311391936e-06, "loss": 0.0049, "step": 248170 }, { "epoch": 158.17718291905672, "grad_norm": 0.20820462703704834, "learning_rate": 7.938163134236714e-06, "loss": 0.008, "step": 248180 }, { "epoch": 158.18355640535373, "grad_norm": 0.16161766648292542, "learning_rate": 7.935183468172518e-06, "loss": 0.0071, "step": 248190 }, { "epoch": 158.18992989165073, "grad_norm": 0.2258158028125763, "learning_rate": 7.93220431323557e-06, "loss": 0.0079, "step": 248200 }, { "epoch": 158.19630337794774, "grad_norm": 0.1485423892736435, "learning_rate": 7.929225669462036e-06, "loss": 0.0048, "step": 248210 }, { "epoch": 158.20267686424475, "grad_norm": 0.5452720522880554, "learning_rate": 7.926247536888148e-06, "loss": 0.009, "step": 248220 }, { "epoch": 158.20905035054176, "grad_norm": 0.16524797677993774, "learning_rate": 7.92326991555008e-06, "loss": 0.0056, "step": 248230 }, { "epoch": 158.21542383683874, "grad_norm": 0.15050587058067322, "learning_rate": 7.920292805484003e-06, "loss": 0.0048, "step": 248240 }, { "epoch": 158.22179732313575, "grad_norm": 0.19769279658794403, "learning_rate": 7.917316206726094e-06, "loss": 0.0025, "step": 248250 }, { "epoch": 158.22817080943275, "grad_norm": 0.09174544364213943, "learning_rate": 7.914340119312542e-06, "loss": 0.0066, "step": 248260 }, { "epoch": 158.23454429572976, "grad_norm": 0.1650516241788864, "learning_rate": 7.911364543279475e-06, "loss": 0.003, "step": 248270 }, { "epoch": 158.24091778202677, "grad_norm": 0.36226096749305725, "learning_rate": 7.908389478663076e-06, "loss": 0.0097, "step": 248280 }, { "epoch": 158.24729126832378, "grad_norm": 0.09549994021654129, "learning_rate": 7.905414925499471e-06, "loss": 0.0086, "step": 248290 }, { "epoch": 158.2536647546208, "grad_norm": 0.1750113070011139, "learning_rate": 7.902440883824847e-06, "loss": 0.0116, "step": 248300 }, { "epoch": 158.2600382409178, "grad_norm": 0.26195356249809265, "learning_rate": 7.899467353675288e-06, "loss": 0.0041, "step": 248310 }, { "epoch": 158.26641172721477, "grad_norm": 0.20074138045310974, "learning_rate": 7.896494335086962e-06, "loss": 0.007, "step": 248320 }, { "epoch": 158.27278521351178, "grad_norm": 0.12986037135124207, "learning_rate": 7.893521828095985e-06, "loss": 0.0056, "step": 248330 }, { "epoch": 158.2791586998088, "grad_norm": 0.12256233394145966, "learning_rate": 7.890549832738464e-06, "loss": 0.0042, "step": 248340 }, { "epoch": 158.2855321861058, "grad_norm": 0.3782998025417328, "learning_rate": 7.887578349050518e-06, "loss": 0.0041, "step": 248350 }, { "epoch": 158.2919056724028, "grad_norm": 0.10396416485309601, "learning_rate": 7.884607377068265e-06, "loss": 0.0043, "step": 248360 }, { "epoch": 158.2982791586998, "grad_norm": 0.497695654630661, "learning_rate": 7.881636916827794e-06, "loss": 0.0095, "step": 248370 }, { "epoch": 158.30465264499682, "grad_norm": 0.32669827342033386, "learning_rate": 7.878666968365201e-06, "loss": 0.0166, "step": 248380 }, { "epoch": 158.31102613129383, "grad_norm": 0.388284295797348, "learning_rate": 7.875697531716564e-06, "loss": 0.0047, "step": 248390 }, { "epoch": 158.3173996175908, "grad_norm": 0.07510823756456375, "learning_rate": 7.872728606917995e-06, "loss": 0.0225, "step": 248400 }, { "epoch": 158.32377310388782, "grad_norm": 0.05634007602930069, "learning_rate": 7.869760194005527e-06, "loss": 0.0071, "step": 248410 }, { "epoch": 158.33014659018482, "grad_norm": 0.13168446719646454, "learning_rate": 7.866792293015256e-06, "loss": 0.0028, "step": 248420 }, { "epoch": 158.33652007648183, "grad_norm": 0.14562135934829712, "learning_rate": 7.86382490398323e-06, "loss": 0.0047, "step": 248430 }, { "epoch": 158.34289356277884, "grad_norm": 0.07736492156982422, "learning_rate": 7.860858026945544e-06, "loss": 0.0073, "step": 248440 }, { "epoch": 158.34926704907585, "grad_norm": 0.20401638746261597, "learning_rate": 7.857891661938188e-06, "loss": 0.0051, "step": 248450 }, { "epoch": 158.35564053537286, "grad_norm": 0.07893229275941849, "learning_rate": 7.854925808997255e-06, "loss": 0.0026, "step": 248460 }, { "epoch": 158.36201402166986, "grad_norm": 0.2683872878551483, "learning_rate": 7.851960468158748e-06, "loss": 0.0162, "step": 248470 }, { "epoch": 158.36838750796684, "grad_norm": 0.09750226885080338, "learning_rate": 7.84899563945874e-06, "loss": 0.0066, "step": 248480 }, { "epoch": 158.37476099426385, "grad_norm": 0.054781682789325714, "learning_rate": 7.846031322933212e-06, "loss": 0.0062, "step": 248490 }, { "epoch": 158.38113448056086, "grad_norm": 0.1301402449607849, "learning_rate": 7.843067518618213e-06, "loss": 0.0063, "step": 248500 }, { "epoch": 158.38750796685787, "grad_norm": 0.1662915199995041, "learning_rate": 7.840104226549743e-06, "loss": 0.0053, "step": 248510 }, { "epoch": 158.39388145315488, "grad_norm": 0.08656777441501617, "learning_rate": 7.837141446763818e-06, "loss": 0.0055, "step": 248520 }, { "epoch": 158.40025493945188, "grad_norm": 0.5707467794418335, "learning_rate": 7.834179179296419e-06, "loss": 0.0162, "step": 248530 }, { "epoch": 158.4066284257489, "grad_norm": 0.14135120809078217, "learning_rate": 7.831217424183568e-06, "loss": 0.0104, "step": 248540 }, { "epoch": 158.4130019120459, "grad_norm": 0.14367984235286713, "learning_rate": 7.828256181461236e-06, "loss": 0.0041, "step": 248550 }, { "epoch": 158.4193753983429, "grad_norm": 0.3366682827472687, "learning_rate": 7.82529545116541e-06, "loss": 0.0041, "step": 248560 }, { "epoch": 158.4257488846399, "grad_norm": 0.26338768005371094, "learning_rate": 7.822335233332046e-06, "loss": 0.0052, "step": 248570 }, { "epoch": 158.4321223709369, "grad_norm": 0.1464126855134964, "learning_rate": 7.819375527997147e-06, "loss": 0.0033, "step": 248580 }, { "epoch": 158.4384958572339, "grad_norm": 0.10261106491088867, "learning_rate": 7.816416335196664e-06, "loss": 0.0044, "step": 248590 }, { "epoch": 158.4448693435309, "grad_norm": 0.09260949492454529, "learning_rate": 7.813457654966549e-06, "loss": 0.0026, "step": 248600 }, { "epoch": 158.45124282982792, "grad_norm": 0.0825587660074234, "learning_rate": 7.810499487342737e-06, "loss": 0.0036, "step": 248610 }, { "epoch": 158.45761631612493, "grad_norm": 0.10632538050413132, "learning_rate": 7.80754183236121e-06, "loss": 0.006, "step": 248620 }, { "epoch": 158.46398980242193, "grad_norm": 0.35693806409835815, "learning_rate": 7.804584690057882e-06, "loss": 0.014, "step": 248630 }, { "epoch": 158.47036328871894, "grad_norm": 0.06799792498350143, "learning_rate": 7.801628060468686e-06, "loss": 0.0058, "step": 248640 }, { "epoch": 158.47673677501592, "grad_norm": 0.10216580331325531, "learning_rate": 7.798671943629555e-06, "loss": 0.0108, "step": 248650 }, { "epoch": 158.48311026131293, "grad_norm": 0.09541383385658264, "learning_rate": 7.795716339576392e-06, "loss": 0.0063, "step": 248660 }, { "epoch": 158.48948374760994, "grad_norm": 0.16907505691051483, "learning_rate": 7.792761248345131e-06, "loss": 0.008, "step": 248670 }, { "epoch": 158.49585723390695, "grad_norm": 0.30975183844566345, "learning_rate": 7.789806669971678e-06, "loss": 0.0088, "step": 248680 }, { "epoch": 158.50223072020395, "grad_norm": 0.14192833006381989, "learning_rate": 7.786852604491922e-06, "loss": 0.0052, "step": 248690 }, { "epoch": 158.50860420650096, "grad_norm": 0.27906718850135803, "learning_rate": 7.783899051941752e-06, "loss": 0.0065, "step": 248700 }, { "epoch": 158.51497769279797, "grad_norm": 0.15150955319404602, "learning_rate": 7.780946012357077e-06, "loss": 0.007, "step": 248710 }, { "epoch": 158.52135117909498, "grad_norm": 0.18449664115905762, "learning_rate": 7.777993485773771e-06, "loss": 0.009, "step": 248720 }, { "epoch": 158.52772466539196, "grad_norm": 0.486555814743042, "learning_rate": 7.775041472227712e-06, "loss": 0.0093, "step": 248730 }, { "epoch": 158.53409815168897, "grad_norm": 0.3638744056224823, "learning_rate": 7.772089971754749e-06, "loss": 0.0039, "step": 248740 }, { "epoch": 158.54047163798597, "grad_norm": 0.0991995558142662, "learning_rate": 7.769138984390778e-06, "loss": 0.0069, "step": 248750 }, { "epoch": 158.54684512428298, "grad_norm": 0.10170572996139526, "learning_rate": 7.766188510171635e-06, "loss": 0.0056, "step": 248760 }, { "epoch": 158.55321861058, "grad_norm": 0.10038924962282181, "learning_rate": 7.763238549133184e-06, "loss": 0.0044, "step": 248770 }, { "epoch": 158.559592096877, "grad_norm": 0.12273125350475311, "learning_rate": 7.760289101311247e-06, "loss": 0.0066, "step": 248780 }, { "epoch": 158.565965583174, "grad_norm": 0.18988139927387238, "learning_rate": 7.757340166741689e-06, "loss": 0.0106, "step": 248790 }, { "epoch": 158.572339069471, "grad_norm": 0.09121110290288925, "learning_rate": 7.754391745460338e-06, "loss": 0.0027, "step": 248800 }, { "epoch": 158.578712555768, "grad_norm": 0.18253131210803986, "learning_rate": 7.751443837503008e-06, "loss": 0.0048, "step": 248810 }, { "epoch": 158.585086042065, "grad_norm": 0.24045825004577637, "learning_rate": 7.748496442905511e-06, "loss": 0.0039, "step": 248820 }, { "epoch": 158.591459528362, "grad_norm": 0.1065351665019989, "learning_rate": 7.745549561703702e-06, "loss": 0.0136, "step": 248830 }, { "epoch": 158.59783301465902, "grad_norm": 0.3877197802066803, "learning_rate": 7.742603193933339e-06, "loss": 0.0109, "step": 248840 }, { "epoch": 158.60420650095602, "grad_norm": 0.20235760509967804, "learning_rate": 7.739657339630251e-06, "loss": 0.0049, "step": 248850 }, { "epoch": 158.61057998725303, "grad_norm": 0.11442745476961136, "learning_rate": 7.73671199883022e-06, "loss": 0.005, "step": 248860 }, { "epoch": 158.61695347355004, "grad_norm": 0.8036231398582458, "learning_rate": 7.733767171569061e-06, "loss": 0.008, "step": 248870 }, { "epoch": 158.62332695984705, "grad_norm": 0.4426440894603729, "learning_rate": 7.730822857882519e-06, "loss": 0.0057, "step": 248880 }, { "epoch": 158.62970044614403, "grad_norm": 0.18215908110141754, "learning_rate": 7.727879057806392e-06, "loss": 0.0075, "step": 248890 }, { "epoch": 158.63607393244104, "grad_norm": 0.18590496480464935, "learning_rate": 7.724935771376451e-06, "loss": 0.0093, "step": 248900 }, { "epoch": 158.64244741873804, "grad_norm": 0.13591931760311127, "learning_rate": 7.721992998628453e-06, "loss": 0.0038, "step": 248910 }, { "epoch": 158.64882090503505, "grad_norm": 0.09362080693244934, "learning_rate": 7.719050739598143e-06, "loss": 0.0084, "step": 248920 }, { "epoch": 158.65519439133206, "grad_norm": 0.24309344589710236, "learning_rate": 7.716108994321303e-06, "loss": 0.0039, "step": 248930 }, { "epoch": 158.66156787762907, "grad_norm": 0.78606778383255, "learning_rate": 7.713167762833662e-06, "loss": 0.0102, "step": 248940 }, { "epoch": 158.66794136392608, "grad_norm": 0.2672649025917053, "learning_rate": 7.710227045170954e-06, "loss": 0.0043, "step": 248950 }, { "epoch": 158.67431485022308, "grad_norm": 0.23983211815357208, "learning_rate": 7.707286841368906e-06, "loss": 0.0079, "step": 248960 }, { "epoch": 158.68068833652006, "grad_norm": 0.07100934535264969, "learning_rate": 7.704347151463276e-06, "loss": 0.0124, "step": 248970 }, { "epoch": 158.68706182281707, "grad_norm": 0.12284216284751892, "learning_rate": 7.701407975489739e-06, "loss": 0.0101, "step": 248980 }, { "epoch": 158.69343530911408, "grad_norm": 0.16220316290855408, "learning_rate": 7.698469313484042e-06, "loss": 0.0051, "step": 248990 }, { "epoch": 158.6998087954111, "grad_norm": 0.17654123902320862, "learning_rate": 7.695531165481873e-06, "loss": 0.007, "step": 249000 }, { "epoch": 158.7061822817081, "grad_norm": 0.39543643593788147, "learning_rate": 7.69259353151896e-06, "loss": 0.0111, "step": 249010 }, { "epoch": 158.7125557680051, "grad_norm": 0.1059214323759079, "learning_rate": 7.689656411630963e-06, "loss": 0.0064, "step": 249020 }, { "epoch": 158.7189292543021, "grad_norm": 0.10164142400026321, "learning_rate": 7.686719805853598e-06, "loss": 0.0043, "step": 249030 }, { "epoch": 158.72530274059912, "grad_norm": 0.17925706505775452, "learning_rate": 7.683783714222526e-06, "loss": 0.0059, "step": 249040 }, { "epoch": 158.7316762268961, "grad_norm": 0.09556972235441208, "learning_rate": 7.68084813677346e-06, "loss": 0.0037, "step": 249050 }, { "epoch": 158.7380497131931, "grad_norm": 0.08758579939603806, "learning_rate": 7.67791307354202e-06, "loss": 0.0097, "step": 249060 }, { "epoch": 158.74442319949011, "grad_norm": 0.11782081425189972, "learning_rate": 7.674978524563903e-06, "loss": 0.0109, "step": 249070 }, { "epoch": 158.75079668578712, "grad_norm": 0.38956281542778015, "learning_rate": 7.672044489874763e-06, "loss": 0.0069, "step": 249080 }, { "epoch": 158.75717017208413, "grad_norm": 0.24538937211036682, "learning_rate": 7.669110969510245e-06, "loss": 0.0065, "step": 249090 }, { "epoch": 158.76354365838114, "grad_norm": 0.15265654027462006, "learning_rate": 7.666177963505988e-06, "loss": 0.0094, "step": 249100 }, { "epoch": 158.76991714467815, "grad_norm": 0.12139716744422913, "learning_rate": 7.663245471897645e-06, "loss": 0.0032, "step": 249110 }, { "epoch": 158.77629063097515, "grad_norm": 0.3779338300228119, "learning_rate": 7.660313494720844e-06, "loss": 0.0063, "step": 249120 }, { "epoch": 158.78266411727213, "grad_norm": 0.3085084855556488, "learning_rate": 7.657382032011206e-06, "loss": 0.0057, "step": 249130 }, { "epoch": 158.78903760356914, "grad_norm": 0.15884986519813538, "learning_rate": 7.654451083804347e-06, "loss": 0.0037, "step": 249140 }, { "epoch": 158.79541108986615, "grad_norm": 0.12672178447246552, "learning_rate": 7.651520650135913e-06, "loss": 0.0048, "step": 249150 }, { "epoch": 158.80178457616316, "grad_norm": 0.12246178090572357, "learning_rate": 7.648590731041461e-06, "loss": 0.0033, "step": 249160 }, { "epoch": 158.80815806246017, "grad_norm": 0.07038771361112595, "learning_rate": 7.64566132655663e-06, "loss": 0.0047, "step": 249170 }, { "epoch": 158.81453154875717, "grad_norm": 0.11836546659469604, "learning_rate": 7.642732436716987e-06, "loss": 0.0051, "step": 249180 }, { "epoch": 158.82090503505418, "grad_norm": 0.28124919533729553, "learning_rate": 7.639804061558164e-06, "loss": 0.0101, "step": 249190 }, { "epoch": 158.8272785213512, "grad_norm": 0.5282386541366577, "learning_rate": 7.636876201115695e-06, "loss": 0.012, "step": 249200 }, { "epoch": 158.8336520076482, "grad_norm": 0.08433591574430466, "learning_rate": 7.633948855425188e-06, "loss": 0.0054, "step": 249210 }, { "epoch": 158.84002549394518, "grad_norm": 0.26362982392311096, "learning_rate": 7.631022024522199e-06, "loss": 0.0052, "step": 249220 }, { "epoch": 158.84639898024218, "grad_norm": 0.13329406082630157, "learning_rate": 7.628095708442301e-06, "loss": 0.0077, "step": 249230 }, { "epoch": 158.8527724665392, "grad_norm": 0.08424153178930283, "learning_rate": 7.62516990722103e-06, "loss": 0.0025, "step": 249240 }, { "epoch": 158.8591459528362, "grad_norm": 0.20987550914287567, "learning_rate": 7.622244620893965e-06, "loss": 0.0102, "step": 249250 }, { "epoch": 158.8655194391332, "grad_norm": 0.1624990701675415, "learning_rate": 7.619319849496642e-06, "loss": 0.0019, "step": 249260 }, { "epoch": 158.87189292543022, "grad_norm": 0.27514171600341797, "learning_rate": 7.616395593064596e-06, "loss": 0.014, "step": 249270 }, { "epoch": 158.87826641172722, "grad_norm": 0.15646116435527802, "learning_rate": 7.613471851633347e-06, "loss": 0.0032, "step": 249280 }, { "epoch": 158.88463989802423, "grad_norm": 0.6572726368904114, "learning_rate": 7.610548625238445e-06, "loss": 0.0054, "step": 249290 }, { "epoch": 158.8910133843212, "grad_norm": 0.16170749068260193, "learning_rate": 7.6076259139154005e-06, "loss": 0.0052, "step": 249300 }, { "epoch": 158.89738687061822, "grad_norm": 0.14729906618595123, "learning_rate": 7.60470371769973e-06, "loss": 0.0057, "step": 249310 }, { "epoch": 158.90376035691523, "grad_norm": 0.2453431934118271, "learning_rate": 7.60178203662692e-06, "loss": 0.0111, "step": 249320 }, { "epoch": 158.91013384321224, "grad_norm": 0.2096860408782959, "learning_rate": 7.5988608707325e-06, "loss": 0.0059, "step": 249330 }, { "epoch": 158.91650732950924, "grad_norm": 0.13959702849388123, "learning_rate": 7.5959402200519555e-06, "loss": 0.003, "step": 249340 }, { "epoch": 158.92288081580625, "grad_norm": 0.10876677930355072, "learning_rate": 7.593020084620778e-06, "loss": 0.0111, "step": 249350 }, { "epoch": 158.92925430210326, "grad_norm": 0.19827571511268616, "learning_rate": 7.590100464474431e-06, "loss": 0.005, "step": 249360 }, { "epoch": 158.93562778840027, "grad_norm": 0.3271944522857666, "learning_rate": 7.587181359648415e-06, "loss": 0.0063, "step": 249370 }, { "epoch": 158.94200127469725, "grad_norm": 0.3308987617492676, "learning_rate": 7.584262770178191e-06, "loss": 0.0036, "step": 249380 }, { "epoch": 158.94837476099426, "grad_norm": 0.5363963842391968, "learning_rate": 7.581344696099218e-06, "loss": 0.0088, "step": 249390 }, { "epoch": 158.95474824729126, "grad_norm": 0.08367614448070526, "learning_rate": 7.578427137446959e-06, "loss": 0.0025, "step": 249400 }, { "epoch": 158.96112173358827, "grad_norm": 0.24221965670585632, "learning_rate": 7.575510094256854e-06, "loss": 0.007, "step": 249410 }, { "epoch": 158.96749521988528, "grad_norm": 0.21305213868618011, "learning_rate": 7.572593566564368e-06, "loss": 0.0046, "step": 249420 }, { "epoch": 158.9738687061823, "grad_norm": 0.1660011112689972, "learning_rate": 7.569677554404925e-06, "loss": 0.0071, "step": 249430 }, { "epoch": 158.9802421924793, "grad_norm": 0.20004883408546448, "learning_rate": 7.566762057813964e-06, "loss": 0.0081, "step": 249440 }, { "epoch": 158.9866156787763, "grad_norm": 0.38384100794792175, "learning_rate": 7.563847076826891e-06, "loss": 0.0183, "step": 249450 }, { "epoch": 158.99298916507328, "grad_norm": 0.2936384081840515, "learning_rate": 7.56093261147916e-06, "loss": 0.0054, "step": 249460 }, { "epoch": 158.9993626513703, "grad_norm": 0.11376946419477463, "learning_rate": 7.5580186618061595e-06, "loss": 0.0042, "step": 249470 }, { "epoch": 159.0057361376673, "grad_norm": 0.35287532210350037, "learning_rate": 7.555105227843312e-06, "loss": 0.0078, "step": 249480 }, { "epoch": 159.0121096239643, "grad_norm": 0.5610401630401611, "learning_rate": 7.552192309625994e-06, "loss": 0.0069, "step": 249490 }, { "epoch": 159.01848311026131, "grad_norm": 0.1794905811548233, "learning_rate": 7.549279907189633e-06, "loss": 0.0073, "step": 249500 }, { "epoch": 159.02485659655832, "grad_norm": 0.23902325332164764, "learning_rate": 7.546368020569594e-06, "loss": 0.0055, "step": 249510 }, { "epoch": 159.03123008285533, "grad_norm": 0.12742622196674347, "learning_rate": 7.5434566498012725e-06, "loss": 0.0075, "step": 249520 }, { "epoch": 159.03760356915234, "grad_norm": 0.3396504521369934, "learning_rate": 7.540545794920023e-06, "loss": 0.006, "step": 249530 }, { "epoch": 159.04397705544932, "grad_norm": 0.1442244052886963, "learning_rate": 7.537635455961256e-06, "loss": 0.0048, "step": 249540 }, { "epoch": 159.05035054174633, "grad_norm": 0.37791159749031067, "learning_rate": 7.534725632960288e-06, "loss": 0.0098, "step": 249550 }, { "epoch": 159.05672402804333, "grad_norm": 0.18770131468772888, "learning_rate": 7.531816325952507e-06, "loss": 0.0075, "step": 249560 }, { "epoch": 159.06309751434034, "grad_norm": 0.4076431393623352, "learning_rate": 7.52890753497324e-06, "loss": 0.0178, "step": 249570 }, { "epoch": 159.06947100063735, "grad_norm": 0.2045736163854599, "learning_rate": 7.525999260057875e-06, "loss": 0.01, "step": 249580 }, { "epoch": 159.07584448693436, "grad_norm": 0.10088489204645157, "learning_rate": 7.523091501241691e-06, "loss": 0.0063, "step": 249590 }, { "epoch": 159.08221797323137, "grad_norm": 0.1811029464006424, "learning_rate": 7.520184258560065e-06, "loss": 0.0038, "step": 249600 }, { "epoch": 159.08859145952837, "grad_norm": 0.2833595275878906, "learning_rate": 7.517277532048312e-06, "loss": 0.0092, "step": 249610 }, { "epoch": 159.09496494582535, "grad_norm": 0.2702428698539734, "learning_rate": 7.5143713217417426e-06, "loss": 0.0047, "step": 249620 }, { "epoch": 159.10133843212236, "grad_norm": 0.2947129011154175, "learning_rate": 7.511465627675662e-06, "loss": 0.0047, "step": 249630 }, { "epoch": 159.10771191841937, "grad_norm": 0.1190427765250206, "learning_rate": 7.508560449885405e-06, "loss": 0.0049, "step": 249640 }, { "epoch": 159.11408540471638, "grad_norm": 0.19735296070575714, "learning_rate": 7.5056557884062515e-06, "loss": 0.004, "step": 249650 }, { "epoch": 159.12045889101339, "grad_norm": 0.11383701115846634, "learning_rate": 7.502751643273509e-06, "loss": 0.0074, "step": 249660 }, { "epoch": 159.1268323773104, "grad_norm": 0.1664537638425827, "learning_rate": 7.499848014522443e-06, "loss": 0.0101, "step": 249670 }, { "epoch": 159.1332058636074, "grad_norm": 0.13321737945079803, "learning_rate": 7.496944902188358e-06, "loss": 0.0036, "step": 249680 }, { "epoch": 159.1395793499044, "grad_norm": 0.33050233125686646, "learning_rate": 7.494042306306526e-06, "loss": 0.0049, "step": 249690 }, { "epoch": 159.1459528362014, "grad_norm": 0.34184595942497253, "learning_rate": 7.491140226912213e-06, "loss": 0.0061, "step": 249700 }, { "epoch": 159.1523263224984, "grad_norm": 0.2605946958065033, "learning_rate": 7.48823866404067e-06, "loss": 0.0038, "step": 249710 }, { "epoch": 159.1586998087954, "grad_norm": 0.3185453712940216, "learning_rate": 7.485337617727184e-06, "loss": 0.0054, "step": 249720 }, { "epoch": 159.1650732950924, "grad_norm": 0.18494534492492676, "learning_rate": 7.482437088006966e-06, "loss": 0.0037, "step": 249730 }, { "epoch": 159.17144678138942, "grad_norm": 0.3244933784008026, "learning_rate": 7.479537074915293e-06, "loss": 0.0045, "step": 249740 }, { "epoch": 159.17782026768643, "grad_norm": 0.20264390110969543, "learning_rate": 7.476637578487378e-06, "loss": 0.008, "step": 249750 }, { "epoch": 159.18419375398344, "grad_norm": 0.16051369905471802, "learning_rate": 7.47373859875849e-06, "loss": 0.0056, "step": 249760 }, { "epoch": 159.19056724028044, "grad_norm": 0.09254167228937149, "learning_rate": 7.470840135763801e-06, "loss": 0.0185, "step": 249770 }, { "epoch": 159.19694072657742, "grad_norm": 0.09398631006479263, "learning_rate": 7.467942189538574e-06, "loss": 0.0141, "step": 249780 }, { "epoch": 159.20331421287443, "grad_norm": 0.2775622606277466, "learning_rate": 7.465044760118006e-06, "loss": 0.0055, "step": 249790 }, { "epoch": 159.20968769917144, "grad_norm": 0.2722986042499542, "learning_rate": 7.462147847537304e-06, "loss": 0.003, "step": 249800 }, { "epoch": 159.21606118546845, "grad_norm": 0.33228468894958496, "learning_rate": 7.459251451831661e-06, "loss": 0.0064, "step": 249810 }, { "epoch": 159.22243467176546, "grad_norm": 0.14252866804599762, "learning_rate": 7.4563555730362835e-06, "loss": 0.0039, "step": 249820 }, { "epoch": 159.22880815806246, "grad_norm": 0.10894136875867844, "learning_rate": 7.453460211186353e-06, "loss": 0.0035, "step": 249830 }, { "epoch": 159.23518164435947, "grad_norm": 0.3359614908695221, "learning_rate": 7.450565366317053e-06, "loss": 0.0085, "step": 249840 }, { "epoch": 159.24155513065648, "grad_norm": 0.09828076511621475, "learning_rate": 7.44767103846355e-06, "loss": 0.0062, "step": 249850 }, { "epoch": 159.2479286169535, "grad_norm": 0.13101297616958618, "learning_rate": 7.444777227661037e-06, "loss": 0.0028, "step": 249860 }, { "epoch": 159.25430210325047, "grad_norm": 0.10472159087657928, "learning_rate": 7.441883933944638e-06, "loss": 0.0035, "step": 249870 }, { "epoch": 159.26067558954747, "grad_norm": 0.1453990638256073, "learning_rate": 7.438991157349545e-06, "loss": 0.0056, "step": 249880 }, { "epoch": 159.26704907584448, "grad_norm": 0.3640722930431366, "learning_rate": 7.436098897910882e-06, "loss": 0.007, "step": 249890 }, { "epoch": 159.2734225621415, "grad_norm": 0.17825859785079956, "learning_rate": 7.433207155663824e-06, "loss": 0.0038, "step": 249900 }, { "epoch": 159.2797960484385, "grad_norm": 0.12859730422496796, "learning_rate": 7.430315930643467e-06, "loss": 0.005, "step": 249910 }, { "epoch": 159.2861695347355, "grad_norm": 0.15537217259407043, "learning_rate": 7.427425222884976e-06, "loss": 0.0032, "step": 249920 }, { "epoch": 159.29254302103251, "grad_norm": 0.072873055934906, "learning_rate": 7.424535032423452e-06, "loss": 0.0031, "step": 249930 }, { "epoch": 159.29891650732952, "grad_norm": 0.2240794152021408, "learning_rate": 7.421645359294049e-06, "loss": 0.0065, "step": 249940 }, { "epoch": 159.3052899936265, "grad_norm": 0.22331295907497406, "learning_rate": 7.41875620353183e-06, "loss": 0.0069, "step": 249950 }, { "epoch": 159.3116634799235, "grad_norm": 0.19556914269924164, "learning_rate": 7.415867565171941e-06, "loss": 0.0044, "step": 249960 }, { "epoch": 159.31803696622052, "grad_norm": 0.25062793493270874, "learning_rate": 7.412979444249463e-06, "loss": 0.0059, "step": 249970 }, { "epoch": 159.32441045251753, "grad_norm": 0.32566365599632263, "learning_rate": 7.410091840799499e-06, "loss": 0.0087, "step": 249980 }, { "epoch": 159.33078393881453, "grad_norm": 0.09942431002855301, "learning_rate": 7.407204754857117e-06, "loss": 0.0034, "step": 249990 }, { "epoch": 159.33715742511154, "grad_norm": 0.14535485208034515, "learning_rate": 7.404318186457421e-06, "loss": 0.0044, "step": 250000 }, { "epoch": 159.34353091140855, "grad_norm": 0.11465374380350113, "learning_rate": 7.401432135635477e-06, "loss": 0.0039, "step": 250010 }, { "epoch": 159.34990439770556, "grad_norm": 0.2386745810508728, "learning_rate": 7.398546602426354e-06, "loss": 0.0052, "step": 250020 }, { "epoch": 159.35627788400254, "grad_norm": 0.46362096071243286, "learning_rate": 7.395661586865099e-06, "loss": 0.0211, "step": 250030 }, { "epoch": 159.36265137029955, "grad_norm": 0.3598145842552185, "learning_rate": 7.3927770889867905e-06, "loss": 0.0061, "step": 250040 }, { "epoch": 159.36902485659655, "grad_norm": 0.12695041298866272, "learning_rate": 7.389893108826473e-06, "loss": 0.0032, "step": 250050 }, { "epoch": 159.37539834289356, "grad_norm": 0.3354473114013672, "learning_rate": 7.38700964641918e-06, "loss": 0.0089, "step": 250060 }, { "epoch": 159.38177182919057, "grad_norm": 0.2246335744857788, "learning_rate": 7.3841267017999484e-06, "loss": 0.0104, "step": 250070 }, { "epoch": 159.38814531548758, "grad_norm": 0.3188121020793915, "learning_rate": 7.381244275003818e-06, "loss": 0.0066, "step": 250080 }, { "epoch": 159.39451880178459, "grad_norm": 0.35523849725723267, "learning_rate": 7.378362366065811e-06, "loss": 0.0052, "step": 250090 }, { "epoch": 159.4008922880816, "grad_norm": 0.17169639468193054, "learning_rate": 7.375480975020943e-06, "loss": 0.0047, "step": 250100 }, { "epoch": 159.40726577437857, "grad_norm": 0.3614434003829956, "learning_rate": 7.372600101904226e-06, "loss": 0.0056, "step": 250110 }, { "epoch": 159.41363926067558, "grad_norm": 0.10486087203025818, "learning_rate": 7.369719746750653e-06, "loss": 0.0036, "step": 250120 }, { "epoch": 159.4200127469726, "grad_norm": 0.08325672149658203, "learning_rate": 7.366839909595247e-06, "loss": 0.0075, "step": 250130 }, { "epoch": 159.4263862332696, "grad_norm": 0.1329314112663269, "learning_rate": 7.363960590472985e-06, "loss": 0.0049, "step": 250140 }, { "epoch": 159.4327597195666, "grad_norm": 0.2294420301914215, "learning_rate": 7.361081789418861e-06, "loss": 0.0069, "step": 250150 }, { "epoch": 159.4391332058636, "grad_norm": 0.09352962672710419, "learning_rate": 7.3582035064678366e-06, "loss": 0.0046, "step": 250160 }, { "epoch": 159.44550669216062, "grad_norm": 0.4412665367126465, "learning_rate": 7.355325741654911e-06, "loss": 0.0059, "step": 250170 }, { "epoch": 159.45188017845763, "grad_norm": 0.04742329195141792, "learning_rate": 7.352448495015041e-06, "loss": 0.0081, "step": 250180 }, { "epoch": 159.4582536647546, "grad_norm": 0.1385142058134079, "learning_rate": 7.349571766583191e-06, "loss": 0.0072, "step": 250190 }, { "epoch": 159.46462715105162, "grad_norm": 0.2962905168533325, "learning_rate": 7.3466955563943005e-06, "loss": 0.005, "step": 250200 }, { "epoch": 159.47100063734862, "grad_norm": 0.20878073573112488, "learning_rate": 7.343819864483342e-06, "loss": 0.0081, "step": 250210 }, { "epoch": 159.47737412364563, "grad_norm": 0.3023134171962738, "learning_rate": 7.340944690885243e-06, "loss": 0.0077, "step": 250220 }, { "epoch": 159.48374760994264, "grad_norm": 0.5620205998420715, "learning_rate": 7.338070035634948e-06, "loss": 0.0037, "step": 250230 }, { "epoch": 159.49012109623965, "grad_norm": 0.1834968626499176, "learning_rate": 7.335195898767366e-06, "loss": 0.0057, "step": 250240 }, { "epoch": 159.49649458253666, "grad_norm": 0.10995211452245712, "learning_rate": 7.33232228031745e-06, "loss": 0.0047, "step": 250250 }, { "epoch": 159.50286806883366, "grad_norm": 0.06949854642152786, "learning_rate": 7.329449180320097e-06, "loss": 0.0078, "step": 250260 }, { "epoch": 159.50924155513064, "grad_norm": 0.18395619094371796, "learning_rate": 7.3265765988102305e-06, "loss": 0.0052, "step": 250270 }, { "epoch": 159.51561504142765, "grad_norm": 0.6266007423400879, "learning_rate": 7.323704535822734e-06, "loss": 0.0041, "step": 250280 }, { "epoch": 159.52198852772466, "grad_norm": 0.28362950682640076, "learning_rate": 7.320832991392546e-06, "loss": 0.005, "step": 250290 }, { "epoch": 159.52836201402167, "grad_norm": 0.13018836081027985, "learning_rate": 7.317961965554509e-06, "loss": 0.0054, "step": 250300 }, { "epoch": 159.53473550031867, "grad_norm": 0.40972501039505005, "learning_rate": 7.315091458343548e-06, "loss": 0.0089, "step": 250310 }, { "epoch": 159.54110898661568, "grad_norm": 0.10639822483062744, "learning_rate": 7.312221469794511e-06, "loss": 0.0074, "step": 250320 }, { "epoch": 159.5474824729127, "grad_norm": 0.2531736493110657, "learning_rate": 7.30935199994231e-06, "loss": 0.0083, "step": 250330 }, { "epoch": 159.5538559592097, "grad_norm": 0.3454510569572449, "learning_rate": 7.3064830488217666e-06, "loss": 0.0101, "step": 250340 }, { "epoch": 159.56022944550668, "grad_norm": 0.5118178725242615, "learning_rate": 7.303614616467774e-06, "loss": 0.0058, "step": 250350 }, { "epoch": 159.5666029318037, "grad_norm": 0.24266290664672852, "learning_rate": 7.300746702915179e-06, "loss": 0.0083, "step": 250360 }, { "epoch": 159.5729764181007, "grad_norm": 0.42674267292022705, "learning_rate": 7.297879308198824e-06, "loss": 0.0044, "step": 250370 }, { "epoch": 159.5793499043977, "grad_norm": 0.19440777599811554, "learning_rate": 7.295012432353543e-06, "loss": 0.0056, "step": 250380 }, { "epoch": 159.5857233906947, "grad_norm": 0.269379198551178, "learning_rate": 7.292146075414191e-06, "loss": 0.0045, "step": 250390 }, { "epoch": 159.59209687699172, "grad_norm": 0.0801854282617569, "learning_rate": 7.289280237415586e-06, "loss": 0.0062, "step": 250400 }, { "epoch": 159.59847036328873, "grad_norm": 0.244487926363945, "learning_rate": 7.2864149183925535e-06, "loss": 0.0053, "step": 250410 }, { "epoch": 159.60484384958573, "grad_norm": 0.0546245314180851, "learning_rate": 7.283550118379895e-06, "loss": 0.0039, "step": 250420 }, { "epoch": 159.61121733588274, "grad_norm": 0.14762599766254425, "learning_rate": 7.28068583741246e-06, "loss": 0.0028, "step": 250430 }, { "epoch": 159.61759082217972, "grad_norm": 0.2768120765686035, "learning_rate": 7.277822075524998e-06, "loss": 0.0096, "step": 250440 }, { "epoch": 159.62396430847673, "grad_norm": 0.3657853901386261, "learning_rate": 7.274958832752349e-06, "loss": 0.0046, "step": 250450 }, { "epoch": 159.63033779477374, "grad_norm": 0.1887391358613968, "learning_rate": 7.272096109129278e-06, "loss": 0.0037, "step": 250460 }, { "epoch": 159.63671128107075, "grad_norm": 0.15658588707447052, "learning_rate": 7.269233904690603e-06, "loss": 0.0036, "step": 250470 }, { "epoch": 159.64308476736775, "grad_norm": 0.10758309811353683, "learning_rate": 7.266372219471057e-06, "loss": 0.005, "step": 250480 }, { "epoch": 159.64945825366476, "grad_norm": 0.4715806543827057, "learning_rate": 7.263511053505451e-06, "loss": 0.0066, "step": 250490 }, { "epoch": 159.65583173996177, "grad_norm": 0.2541763186454773, "learning_rate": 7.260650406828528e-06, "loss": 0.0172, "step": 250500 }, { "epoch": 159.66220522625878, "grad_norm": 0.1523820161819458, "learning_rate": 7.257790279475063e-06, "loss": 0.0041, "step": 250510 }, { "epoch": 159.66857871255576, "grad_norm": 0.12045402079820633, "learning_rate": 7.254930671479787e-06, "loss": 0.0035, "step": 250520 }, { "epoch": 159.67495219885276, "grad_norm": 0.19282162189483643, "learning_rate": 7.252071582877474e-06, "loss": 0.0111, "step": 250530 }, { "epoch": 159.68132568514977, "grad_norm": 0.06649594008922577, "learning_rate": 7.249213013702849e-06, "loss": 0.0029, "step": 250540 }, { "epoch": 159.68769917144678, "grad_norm": 0.5768267512321472, "learning_rate": 7.246354963990653e-06, "loss": 0.0121, "step": 250550 }, { "epoch": 159.6940726577438, "grad_norm": 0.06330365687608719, "learning_rate": 7.243497433775598e-06, "loss": 0.0041, "step": 250560 }, { "epoch": 159.7004461440408, "grad_norm": 0.5605005025863647, "learning_rate": 7.240640423092426e-06, "loss": 0.0061, "step": 250570 }, { "epoch": 159.7068196303378, "grad_norm": 0.45237910747528076, "learning_rate": 7.2377839319758444e-06, "loss": 0.007, "step": 250580 }, { "epoch": 159.7131931166348, "grad_norm": 0.12632954120635986, "learning_rate": 7.234927960460569e-06, "loss": 0.006, "step": 250590 }, { "epoch": 159.7195666029318, "grad_norm": 0.2452673614025116, "learning_rate": 7.232072508581278e-06, "loss": 0.0074, "step": 250600 }, { "epoch": 159.7259400892288, "grad_norm": 0.07948341220617294, "learning_rate": 7.22921757637271e-06, "loss": 0.0062, "step": 250610 }, { "epoch": 159.7323135755258, "grad_norm": 0.34641143679618835, "learning_rate": 7.226363163869509e-06, "loss": 0.0074, "step": 250620 }, { "epoch": 159.73868706182282, "grad_norm": 0.15825015306472778, "learning_rate": 7.223509271106393e-06, "loss": 0.0091, "step": 250630 }, { "epoch": 159.74506054811982, "grad_norm": 0.2576844394207001, "learning_rate": 7.220655898118012e-06, "loss": 0.0078, "step": 250640 }, { "epoch": 159.75143403441683, "grad_norm": 0.26015278697013855, "learning_rate": 7.217803044939075e-06, "loss": 0.005, "step": 250650 }, { "epoch": 159.75780752071384, "grad_norm": 0.3324417173862457, "learning_rate": 7.214950711604207e-06, "loss": 0.0182, "step": 250660 }, { "epoch": 159.76418100701085, "grad_norm": 0.13331221044063568, "learning_rate": 7.212098898148089e-06, "loss": 0.0066, "step": 250670 }, { "epoch": 159.77055449330783, "grad_norm": 0.24628785252571106, "learning_rate": 7.2092476046053734e-06, "loss": 0.0055, "step": 250680 }, { "epoch": 159.77692797960484, "grad_norm": 0.13253282010555267, "learning_rate": 7.206396831010697e-06, "loss": 0.0051, "step": 250690 }, { "epoch": 159.78330146590184, "grad_norm": 0.19740617275238037, "learning_rate": 7.2035465773986895e-06, "loss": 0.0067, "step": 250700 }, { "epoch": 159.78967495219885, "grad_norm": 0.10888004302978516, "learning_rate": 7.200696843804011e-06, "loss": 0.0091, "step": 250710 }, { "epoch": 159.79604843849586, "grad_norm": 0.2690882980823517, "learning_rate": 7.1978476302612785e-06, "loss": 0.0108, "step": 250720 }, { "epoch": 159.80242192479287, "grad_norm": 0.07687271386384964, "learning_rate": 7.194998936805103e-06, "loss": 0.0074, "step": 250730 }, { "epoch": 159.80879541108987, "grad_norm": 0.29744720458984375, "learning_rate": 7.192150763470101e-06, "loss": 0.0057, "step": 250740 }, { "epoch": 159.81516889738688, "grad_norm": 0.1680261343717575, "learning_rate": 7.18930311029089e-06, "loss": 0.0121, "step": 250750 }, { "epoch": 159.82154238368386, "grad_norm": 0.31793051958084106, "learning_rate": 7.18645597730207e-06, "loss": 0.0049, "step": 250760 }, { "epoch": 159.82791586998087, "grad_norm": 0.21191316843032837, "learning_rate": 7.18360936453823e-06, "loss": 0.0085, "step": 250770 }, { "epoch": 159.83428935627788, "grad_norm": 0.15352608263492584, "learning_rate": 7.180763272033947e-06, "loss": 0.0045, "step": 250780 }, { "epoch": 159.8406628425749, "grad_norm": 0.15386345982551575, "learning_rate": 7.177917699823838e-06, "loss": 0.004, "step": 250790 }, { "epoch": 159.8470363288719, "grad_norm": 0.08112472295761108, "learning_rate": 7.175072647942454e-06, "loss": 0.0072, "step": 250800 }, { "epoch": 159.8534098151689, "grad_norm": 0.10813475400209427, "learning_rate": 7.172228116424373e-06, "loss": 0.0044, "step": 250810 }, { "epoch": 159.8597833014659, "grad_norm": 0.4317017197608948, "learning_rate": 7.169384105304156e-06, "loss": 0.0089, "step": 250820 }, { "epoch": 159.86615678776292, "grad_norm": 0.09605991095304489, "learning_rate": 7.166540614616346e-06, "loss": 0.0037, "step": 250830 }, { "epoch": 159.8725302740599, "grad_norm": 0.21936939656734467, "learning_rate": 7.163697644395523e-06, "loss": 0.0042, "step": 250840 }, { "epoch": 159.8789037603569, "grad_norm": 0.1444387137889862, "learning_rate": 7.16085519467622e-06, "loss": 0.0054, "step": 250850 }, { "epoch": 159.8852772466539, "grad_norm": 0.2992907166481018, "learning_rate": 7.158013265492969e-06, "loss": 0.0061, "step": 250860 }, { "epoch": 159.89165073295092, "grad_norm": 0.10022863000631332, "learning_rate": 7.155171856880299e-06, "loss": 0.0061, "step": 250870 }, { "epoch": 159.89802421924793, "grad_norm": 0.0930338203907013, "learning_rate": 7.1523309688727554e-06, "loss": 0.0032, "step": 250880 }, { "epoch": 159.90439770554494, "grad_norm": 0.15372538566589355, "learning_rate": 7.1494906015048435e-06, "loss": 0.0078, "step": 250890 }, { "epoch": 159.91077119184195, "grad_norm": 0.2737712264060974, "learning_rate": 7.146650754811074e-06, "loss": 0.008, "step": 250900 }, { "epoch": 159.91714467813895, "grad_norm": 0.1293330192565918, "learning_rate": 7.1438114288259525e-06, "loss": 0.0055, "step": 250910 }, { "epoch": 159.92351816443593, "grad_norm": 0.4688749313354492, "learning_rate": 7.140972623583991e-06, "loss": 0.0061, "step": 250920 }, { "epoch": 159.92989165073294, "grad_norm": 0.18740397691726685, "learning_rate": 7.1381343391196845e-06, "loss": 0.0049, "step": 250930 }, { "epoch": 159.93626513702995, "grad_norm": 0.20484456419944763, "learning_rate": 7.13529657546751e-06, "loss": 0.0035, "step": 250940 }, { "epoch": 159.94263862332696, "grad_norm": 0.33116331696510315, "learning_rate": 7.1324593326619446e-06, "loss": 0.009, "step": 250950 }, { "epoch": 159.94901210962396, "grad_norm": 0.07437436282634735, "learning_rate": 7.129622610737479e-06, "loss": 0.0041, "step": 250960 }, { "epoch": 159.95538559592097, "grad_norm": 0.2615228593349457, "learning_rate": 7.126786409728581e-06, "loss": 0.0083, "step": 250970 }, { "epoch": 159.96175908221798, "grad_norm": 0.2713069021701813, "learning_rate": 7.123950729669704e-06, "loss": 0.0104, "step": 250980 }, { "epoch": 159.968132568515, "grad_norm": 0.2173607498407364, "learning_rate": 7.121115570595294e-06, "loss": 0.0061, "step": 250990 }, { "epoch": 159.97450605481197, "grad_norm": 0.11684948951005936, "learning_rate": 7.1182809325398395e-06, "loss": 0.0055, "step": 251000 }, { "epoch": 159.98087954110898, "grad_norm": 0.23385071754455566, "learning_rate": 7.115446815537735e-06, "loss": 0.007, "step": 251010 }, { "epoch": 159.98725302740598, "grad_norm": 0.2111954540014267, "learning_rate": 7.1126132196234485e-06, "loss": 0.0089, "step": 251020 }, { "epoch": 159.993626513703, "grad_norm": 0.22450178861618042, "learning_rate": 7.109780144831396e-06, "loss": 0.0051, "step": 251030 }, { "epoch": 160.0, "grad_norm": 0.13612239062786102, "learning_rate": 7.106947591196034e-06, "loss": 0.0049, "step": 251040 }, { "epoch": 160.006373486297, "grad_norm": 0.07328265905380249, "learning_rate": 7.1041155587517335e-06, "loss": 0.0058, "step": 251050 }, { "epoch": 160.01274697259402, "grad_norm": 0.28869888186454773, "learning_rate": 7.101284047532941e-06, "loss": 0.0066, "step": 251060 }, { "epoch": 160.01912045889102, "grad_norm": 0.13697189092636108, "learning_rate": 7.0984530575740505e-06, "loss": 0.0043, "step": 251070 }, { "epoch": 160.02549394518803, "grad_norm": 0.3600170910358429, "learning_rate": 7.0956225889094615e-06, "loss": 0.0052, "step": 251080 }, { "epoch": 160.031867431485, "grad_norm": 0.19266460835933685, "learning_rate": 7.092792641573553e-06, "loss": 0.0048, "step": 251090 }, { "epoch": 160.03824091778202, "grad_norm": 0.1131901741027832, "learning_rate": 7.089963215600737e-06, "loss": 0.002, "step": 251100 }, { "epoch": 160.04461440407903, "grad_norm": 0.14111867547035217, "learning_rate": 7.087134311025384e-06, "loss": 0.0153, "step": 251110 }, { "epoch": 160.05098789037604, "grad_norm": 0.1465657651424408, "learning_rate": 7.084305927881862e-06, "loss": 0.006, "step": 251120 }, { "epoch": 160.05736137667304, "grad_norm": 0.08680373430252075, "learning_rate": 7.081478066204534e-06, "loss": 0.0074, "step": 251130 }, { "epoch": 160.06373486297005, "grad_norm": 0.1745128184556961, "learning_rate": 7.078650726027775e-06, "loss": 0.0108, "step": 251140 }, { "epoch": 160.07010834926706, "grad_norm": 0.5679028630256653, "learning_rate": 7.075823907385943e-06, "loss": 0.005, "step": 251150 }, { "epoch": 160.07648183556407, "grad_norm": 0.10388806462287903, "learning_rate": 7.07299761031337e-06, "loss": 0.0058, "step": 251160 }, { "epoch": 160.08285532186105, "grad_norm": 0.3319796323776245, "learning_rate": 7.070171834844397e-06, "loss": 0.0044, "step": 251170 }, { "epoch": 160.08922880815805, "grad_norm": 0.09408219158649445, "learning_rate": 7.06734658101339e-06, "loss": 0.0047, "step": 251180 }, { "epoch": 160.09560229445506, "grad_norm": 0.6246300339698792, "learning_rate": 7.064521848854638e-06, "loss": 0.0052, "step": 251190 }, { "epoch": 160.10197578075207, "grad_norm": 0.08743538707494736, "learning_rate": 7.061697638402492e-06, "loss": 0.0046, "step": 251200 }, { "epoch": 160.10834926704908, "grad_norm": 0.3924484848976135, "learning_rate": 7.058873949691252e-06, "loss": 0.0061, "step": 251210 }, { "epoch": 160.1147227533461, "grad_norm": 0.1997503638267517, "learning_rate": 7.056050782755258e-06, "loss": 0.0083, "step": 251220 }, { "epoch": 160.1210962396431, "grad_norm": 0.31214219331741333, "learning_rate": 7.053228137628765e-06, "loss": 0.0042, "step": 251230 }, { "epoch": 160.1274697259401, "grad_norm": 0.49071526527404785, "learning_rate": 7.0504060143461174e-06, "loss": 0.0088, "step": 251240 }, { "epoch": 160.13384321223708, "grad_norm": 0.08761759102344513, "learning_rate": 7.047584412941583e-06, "loss": 0.0038, "step": 251250 }, { "epoch": 160.1402166985341, "grad_norm": 0.07113423943519592, "learning_rate": 7.044763333449456e-06, "loss": 0.0055, "step": 251260 }, { "epoch": 160.1465901848311, "grad_norm": 0.19552356004714966, "learning_rate": 7.041942775903993e-06, "loss": 0.0081, "step": 251270 }, { "epoch": 160.1529636711281, "grad_norm": 0.1292421519756317, "learning_rate": 7.039122740339498e-06, "loss": 0.0099, "step": 251280 }, { "epoch": 160.1593371574251, "grad_norm": 0.3547952473163605, "learning_rate": 7.03630322679023e-06, "loss": 0.0149, "step": 251290 }, { "epoch": 160.16571064372212, "grad_norm": 0.08377642929553986, "learning_rate": 7.0334842352904385e-06, "loss": 0.0092, "step": 251300 }, { "epoch": 160.17208413001913, "grad_norm": 0.13415028154850006, "learning_rate": 7.030665765874367e-06, "loss": 0.0088, "step": 251310 }, { "epoch": 160.17845761631614, "grad_norm": 0.2545805871486664, "learning_rate": 7.0278478185762995e-06, "loss": 0.009, "step": 251320 }, { "epoch": 160.18483110261312, "grad_norm": 0.05671537294983864, "learning_rate": 7.025030393430432e-06, "loss": 0.0134, "step": 251330 }, { "epoch": 160.19120458891013, "grad_norm": 0.18117496371269226, "learning_rate": 7.02221349047103e-06, "loss": 0.008, "step": 251340 }, { "epoch": 160.19757807520713, "grad_norm": 0.08324745297431946, "learning_rate": 7.019397109732301e-06, "loss": 0.0036, "step": 251350 }, { "epoch": 160.20395156150414, "grad_norm": 0.36768046021461487, "learning_rate": 7.0165812512485e-06, "loss": 0.0054, "step": 251360 }, { "epoch": 160.21032504780115, "grad_norm": 0.11699370294809341, "learning_rate": 7.013765915053799e-06, "loss": 0.0042, "step": 251370 }, { "epoch": 160.21669853409816, "grad_norm": 0.18333274126052856, "learning_rate": 7.010951101182439e-06, "loss": 0.006, "step": 251380 }, { "epoch": 160.22307202039516, "grad_norm": 0.11829345673322678, "learning_rate": 7.008136809668609e-06, "loss": 0.0035, "step": 251390 }, { "epoch": 160.22944550669217, "grad_norm": 0.09237894415855408, "learning_rate": 7.0053230405465135e-06, "loss": 0.008, "step": 251400 }, { "epoch": 160.23581899298915, "grad_norm": 0.13596196472644806, "learning_rate": 7.00250979385032e-06, "loss": 0.0061, "step": 251410 }, { "epoch": 160.24219247928616, "grad_norm": 0.14207880198955536, "learning_rate": 6.9996970696142415e-06, "loss": 0.0048, "step": 251420 }, { "epoch": 160.24856596558317, "grad_norm": 0.08705073595046997, "learning_rate": 6.996884867872439e-06, "loss": 0.0036, "step": 251430 }, { "epoch": 160.25493945188018, "grad_norm": 0.10670893639326096, "learning_rate": 6.994073188659095e-06, "loss": 0.0028, "step": 251440 }, { "epoch": 160.26131293817718, "grad_norm": 0.1758333444595337, "learning_rate": 6.99126203200835e-06, "loss": 0.0073, "step": 251450 }, { "epoch": 160.2676864244742, "grad_norm": 0.13266649842262268, "learning_rate": 6.9884513979543875e-06, "loss": 0.0075, "step": 251460 }, { "epoch": 160.2740599107712, "grad_norm": 0.18306145071983337, "learning_rate": 6.985641286531353e-06, "loss": 0.0029, "step": 251470 }, { "epoch": 160.2804333970682, "grad_norm": 0.08503808081150055, "learning_rate": 6.98283169777339e-06, "loss": 0.0038, "step": 251480 }, { "epoch": 160.2868068833652, "grad_norm": 0.4147191345691681, "learning_rate": 6.980022631714622e-06, "loss": 0.023, "step": 251490 }, { "epoch": 160.2931803696622, "grad_norm": 0.140921488404274, "learning_rate": 6.977214088389211e-06, "loss": 0.004, "step": 251500 }, { "epoch": 160.2995538559592, "grad_norm": 0.30941301584243774, "learning_rate": 6.9744060678312686e-06, "loss": 0.0032, "step": 251510 }, { "epoch": 160.3059273422562, "grad_norm": 0.16516917943954468, "learning_rate": 6.971598570074917e-06, "loss": 0.0057, "step": 251520 }, { "epoch": 160.31230082855322, "grad_norm": 0.2071985900402069, "learning_rate": 6.968791595154256e-06, "loss": 0.0054, "step": 251530 }, { "epoch": 160.31867431485023, "grad_norm": 0.1899174153804779, "learning_rate": 6.965985143103415e-06, "loss": 0.0043, "step": 251540 }, { "epoch": 160.32504780114724, "grad_norm": 0.18032698333263397, "learning_rate": 6.963179213956489e-06, "loss": 0.0072, "step": 251550 }, { "epoch": 160.33142128744424, "grad_norm": 0.12875182926654816, "learning_rate": 6.9603738077475615e-06, "loss": 0.0057, "step": 251560 }, { "epoch": 160.33779477374122, "grad_norm": 0.3343532383441925, "learning_rate": 6.9575689245107325e-06, "loss": 0.0056, "step": 251570 }, { "epoch": 160.34416826003823, "grad_norm": 0.39787060022354126, "learning_rate": 6.954764564280069e-06, "loss": 0.0051, "step": 251580 }, { "epoch": 160.35054174633524, "grad_norm": 0.08457057923078537, "learning_rate": 6.951960727089674e-06, "loss": 0.0033, "step": 251590 }, { "epoch": 160.35691523263225, "grad_norm": 0.5707218050956726, "learning_rate": 6.9491574129735945e-06, "loss": 0.0065, "step": 251600 }, { "epoch": 160.36328871892925, "grad_norm": 0.22619688510894775, "learning_rate": 6.9463546219659e-06, "loss": 0.0101, "step": 251610 }, { "epoch": 160.36966220522626, "grad_norm": 0.31581196188926697, "learning_rate": 6.943552354100636e-06, "loss": 0.0051, "step": 251620 }, { "epoch": 160.37603569152327, "grad_norm": 0.15735186636447906, "learning_rate": 6.940750609411873e-06, "loss": 0.0051, "step": 251630 }, { "epoch": 160.38240917782028, "grad_norm": 0.11656533926725388, "learning_rate": 6.93794938793364e-06, "loss": 0.0056, "step": 251640 }, { "epoch": 160.38878266411726, "grad_norm": 0.1111409068107605, "learning_rate": 6.935148689699989e-06, "loss": 0.0055, "step": 251650 }, { "epoch": 160.39515615041427, "grad_norm": 0.23339366912841797, "learning_rate": 6.932348514744924e-06, "loss": 0.0037, "step": 251660 }, { "epoch": 160.40152963671127, "grad_norm": 0.12128827720880508, "learning_rate": 6.929548863102497e-06, "loss": 0.0104, "step": 251670 }, { "epoch": 160.40790312300828, "grad_norm": 0.29651957750320435, "learning_rate": 6.92674973480672e-06, "loss": 0.0076, "step": 251680 }, { "epoch": 160.4142766093053, "grad_norm": 0.25859540700912476, "learning_rate": 6.923951129891598e-06, "loss": 0.0041, "step": 251690 }, { "epoch": 160.4206500956023, "grad_norm": 0.097852922976017, "learning_rate": 6.9211530483911325e-06, "loss": 0.0041, "step": 251700 }, { "epoch": 160.4270235818993, "grad_norm": 0.23464754223823547, "learning_rate": 6.918355490339357e-06, "loss": 0.0049, "step": 251710 }, { "epoch": 160.4333970681963, "grad_norm": 0.12754546105861664, "learning_rate": 6.9155584557702105e-06, "loss": 0.0043, "step": 251720 }, { "epoch": 160.43977055449332, "grad_norm": 0.17841017246246338, "learning_rate": 6.912761944717716e-06, "loss": 0.0051, "step": 251730 }, { "epoch": 160.4461440407903, "grad_norm": 0.29430222511291504, "learning_rate": 6.909965957215836e-06, "loss": 0.0057, "step": 251740 }, { "epoch": 160.4525175270873, "grad_norm": 0.2750230133533478, "learning_rate": 6.907170493298576e-06, "loss": 0.0053, "step": 251750 }, { "epoch": 160.45889101338432, "grad_norm": 0.16285835206508636, "learning_rate": 6.904375552999859e-06, "loss": 0.0051, "step": 251760 }, { "epoch": 160.46526449968133, "grad_norm": 0.17326043546199799, "learning_rate": 6.90158113635368e-06, "loss": 0.0061, "step": 251770 }, { "epoch": 160.47163798597833, "grad_norm": 0.13423451781272888, "learning_rate": 6.898787243393972e-06, "loss": 0.006, "step": 251780 }, { "epoch": 160.47801147227534, "grad_norm": 0.19424574077129364, "learning_rate": 6.895993874154711e-06, "loss": 0.0078, "step": 251790 }, { "epoch": 160.48438495857235, "grad_norm": 0.20076701045036316, "learning_rate": 6.893201028669804e-06, "loss": 0.0053, "step": 251800 }, { "epoch": 160.49075844486936, "grad_norm": 0.12693341076374054, "learning_rate": 6.890408706973211e-06, "loss": 0.0044, "step": 251810 }, { "epoch": 160.49713193116634, "grad_norm": 0.3409349024295807, "learning_rate": 6.887616909098854e-06, "loss": 0.0059, "step": 251820 }, { "epoch": 160.50350541746334, "grad_norm": 0.16968142986297607, "learning_rate": 6.884825635080661e-06, "loss": 0.0104, "step": 251830 }, { "epoch": 160.50987890376035, "grad_norm": 0.22212164103984833, "learning_rate": 6.882034884952532e-06, "loss": 0.0038, "step": 251840 }, { "epoch": 160.51625239005736, "grad_norm": 0.0794161707162857, "learning_rate": 6.879244658748396e-06, "loss": 0.0042, "step": 251850 }, { "epoch": 160.52262587635437, "grad_norm": 0.5289539098739624, "learning_rate": 6.876454956502154e-06, "loss": 0.0045, "step": 251860 }, { "epoch": 160.52899936265138, "grad_norm": 0.34150099754333496, "learning_rate": 6.873665778247701e-06, "loss": 0.006, "step": 251870 }, { "epoch": 160.53537284894838, "grad_norm": 0.09941787272691727, "learning_rate": 6.870877124018909e-06, "loss": 0.0089, "step": 251880 }, { "epoch": 160.5417463352454, "grad_norm": 0.10306942462921143, "learning_rate": 6.8680889938497084e-06, "loss": 0.0076, "step": 251890 }, { "epoch": 160.54811982154237, "grad_norm": 0.11106495559215546, "learning_rate": 6.865301387773926e-06, "loss": 0.0022, "step": 251900 }, { "epoch": 160.55449330783938, "grad_norm": 0.1412331461906433, "learning_rate": 6.862514305825468e-06, "loss": 0.0109, "step": 251910 }, { "epoch": 160.5608667941364, "grad_norm": 0.1097356379032135, "learning_rate": 6.859727748038181e-06, "loss": 0.0114, "step": 251920 }, { "epoch": 160.5672402804334, "grad_norm": 0.19974298775196075, "learning_rate": 6.8569417144459525e-06, "loss": 0.0038, "step": 251930 }, { "epoch": 160.5736137667304, "grad_norm": 0.1436125934123993, "learning_rate": 6.854156205082596e-06, "loss": 0.0028, "step": 251940 }, { "epoch": 160.5799872530274, "grad_norm": 0.08685291558504105, "learning_rate": 6.851371219981989e-06, "loss": 0.0042, "step": 251950 }, { "epoch": 160.58636073932442, "grad_norm": 0.36819422245025635, "learning_rate": 6.848586759177961e-06, "loss": 0.005, "step": 251960 }, { "epoch": 160.59273422562143, "grad_norm": 0.20853659510612488, "learning_rate": 6.845802822704344e-06, "loss": 0.0031, "step": 251970 }, { "epoch": 160.5991077119184, "grad_norm": 0.1903805285692215, "learning_rate": 6.843019410594959e-06, "loss": 0.0047, "step": 251980 }, { "epoch": 160.60548119821541, "grad_norm": 0.3886750042438507, "learning_rate": 6.840236522883642e-06, "loss": 0.0041, "step": 251990 }, { "epoch": 160.61185468451242, "grad_norm": 0.14132730662822723, "learning_rate": 6.837454159604201e-06, "loss": 0.0135, "step": 252000 }, { "epoch": 160.61822817080943, "grad_norm": 0.5829475522041321, "learning_rate": 6.834672320790447e-06, "loss": 0.0042, "step": 252010 }, { "epoch": 160.62460165710644, "grad_norm": 0.2448253482580185, "learning_rate": 6.831891006476165e-06, "loss": 0.0032, "step": 252020 }, { "epoch": 160.63097514340345, "grad_norm": 0.40777602791786194, "learning_rate": 6.829110216695178e-06, "loss": 0.0069, "step": 252030 }, { "epoch": 160.63734862970045, "grad_norm": 0.1920197755098343, "learning_rate": 6.826329951481264e-06, "loss": 0.0033, "step": 252040 }, { "epoch": 160.64372211599746, "grad_norm": 0.27234524488449097, "learning_rate": 6.823550210868202e-06, "loss": 0.0055, "step": 252050 }, { "epoch": 160.65009560229444, "grad_norm": 0.42792510986328125, "learning_rate": 6.820770994889753e-06, "loss": 0.0037, "step": 252060 }, { "epoch": 160.65646908859145, "grad_norm": 0.1168457418680191, "learning_rate": 6.817992303579734e-06, "loss": 0.0096, "step": 252070 }, { "epoch": 160.66284257488846, "grad_norm": 0.11354458332061768, "learning_rate": 6.8152141369718534e-06, "loss": 0.0025, "step": 252080 }, { "epoch": 160.66921606118547, "grad_norm": 0.31367772817611694, "learning_rate": 6.812436495099911e-06, "loss": 0.0052, "step": 252090 }, { "epoch": 160.67558954748247, "grad_norm": 0.3092777132987976, "learning_rate": 6.809659377997623e-06, "loss": 0.0083, "step": 252100 }, { "epoch": 160.68196303377948, "grad_norm": 0.22014212608337402, "learning_rate": 6.806882785698776e-06, "loss": 0.003, "step": 252110 }, { "epoch": 160.6883365200765, "grad_norm": 0.2577904462814331, "learning_rate": 6.804106718237063e-06, "loss": 0.0089, "step": 252120 }, { "epoch": 160.6947100063735, "grad_norm": 0.2642727792263031, "learning_rate": 6.801331175646247e-06, "loss": 0.007, "step": 252130 }, { "epoch": 160.70108349267048, "grad_norm": 0.10220614075660706, "learning_rate": 6.7985561579600455e-06, "loss": 0.0053, "step": 252140 }, { "epoch": 160.70745697896749, "grad_norm": 0.1784929782152176, "learning_rate": 6.795781665212175e-06, "loss": 0.0051, "step": 252150 }, { "epoch": 160.7138304652645, "grad_norm": 0.2606440484523773, "learning_rate": 6.7930076974363375e-06, "loss": 0.0076, "step": 252160 }, { "epoch": 160.7202039515615, "grad_norm": 0.1568189263343811, "learning_rate": 6.7902342546662615e-06, "loss": 0.004, "step": 252170 }, { "epoch": 160.7265774378585, "grad_norm": 0.23314107954502106, "learning_rate": 6.787461336935635e-06, "loss": 0.0042, "step": 252180 }, { "epoch": 160.73295092415552, "grad_norm": 0.10724150389432907, "learning_rate": 6.784688944278156e-06, "loss": 0.0033, "step": 252190 }, { "epoch": 160.73932441045253, "grad_norm": 0.15821632742881775, "learning_rate": 6.781917076727495e-06, "loss": 0.0055, "step": 252200 }, { "epoch": 160.74569789674953, "grad_norm": 0.3371497392654419, "learning_rate": 6.7791457343173546e-06, "loss": 0.0105, "step": 252210 }, { "epoch": 160.7520713830465, "grad_norm": 0.189015194773674, "learning_rate": 6.776374917081407e-06, "loss": 0.0115, "step": 252220 }, { "epoch": 160.75844486934352, "grad_norm": 0.16158811748027802, "learning_rate": 6.773604625053309e-06, "loss": 0.0053, "step": 252230 }, { "epoch": 160.76481835564053, "grad_norm": 0.11727835237979889, "learning_rate": 6.770834858266717e-06, "loss": 0.0125, "step": 252240 }, { "epoch": 160.77119184193754, "grad_norm": 0.28191038966178894, "learning_rate": 6.768065616755309e-06, "loss": 0.0063, "step": 252250 }, { "epoch": 160.77756532823454, "grad_norm": 0.10923104733228683, "learning_rate": 6.76529690055272e-06, "loss": 0.0084, "step": 252260 }, { "epoch": 160.78393881453155, "grad_norm": 0.18837827444076538, "learning_rate": 6.762528709692595e-06, "loss": 0.0071, "step": 252270 }, { "epoch": 160.79031230082856, "grad_norm": 0.1750582903623581, "learning_rate": 6.759761044208568e-06, "loss": 0.0054, "step": 252280 }, { "epoch": 160.79668578712557, "grad_norm": 0.09089477360248566, "learning_rate": 6.756993904134262e-06, "loss": 0.0032, "step": 252290 }, { "epoch": 160.80305927342255, "grad_norm": 0.12141043692827225, "learning_rate": 6.754227289503318e-06, "loss": 0.0076, "step": 252300 }, { "epoch": 160.80943275971956, "grad_norm": 0.1118045374751091, "learning_rate": 6.7514612003493405e-06, "loss": 0.0061, "step": 252310 }, { "epoch": 160.81580624601656, "grad_norm": 0.33199042081832886, "learning_rate": 6.748695636705943e-06, "loss": 0.004, "step": 252320 }, { "epoch": 160.82217973231357, "grad_norm": 0.27624207735061646, "learning_rate": 6.74593059860672e-06, "loss": 0.0038, "step": 252330 }, { "epoch": 160.82855321861058, "grad_norm": 0.13051879405975342, "learning_rate": 6.7431660860852895e-06, "loss": 0.0043, "step": 252340 }, { "epoch": 160.8349267049076, "grad_norm": 0.18770897388458252, "learning_rate": 6.740402099175236e-06, "loss": 0.0058, "step": 252350 }, { "epoch": 160.8413001912046, "grad_norm": 0.19519595801830292, "learning_rate": 6.737638637910138e-06, "loss": 0.0041, "step": 252360 }, { "epoch": 160.8476736775016, "grad_norm": 0.16102729737758636, "learning_rate": 6.734875702323562e-06, "loss": 0.0034, "step": 252370 }, { "epoch": 160.8540471637986, "grad_norm": 0.3827328383922577, "learning_rate": 6.73211329244911e-06, "loss": 0.0159, "step": 252380 }, { "epoch": 160.8604206500956, "grad_norm": 0.4846457839012146, "learning_rate": 6.729351408320334e-06, "loss": 0.0066, "step": 252390 }, { "epoch": 160.8667941363926, "grad_norm": 0.4014611840248108, "learning_rate": 6.726590049970794e-06, "loss": 0.0062, "step": 252400 }, { "epoch": 160.8731676226896, "grad_norm": 0.29764965176582336, "learning_rate": 6.723829217434024e-06, "loss": 0.0172, "step": 252410 }, { "epoch": 160.87954110898661, "grad_norm": 0.2527816891670227, "learning_rate": 6.721068910743605e-06, "loss": 0.0045, "step": 252420 }, { "epoch": 160.88591459528362, "grad_norm": 0.09246676415205002, "learning_rate": 6.718309129933059e-06, "loss": 0.0058, "step": 252430 }, { "epoch": 160.89228808158063, "grad_norm": 0.18807396292686462, "learning_rate": 6.715549875035926e-06, "loss": 0.0086, "step": 252440 }, { "epoch": 160.89866156787764, "grad_norm": 0.06554216146469116, "learning_rate": 6.7127911460857115e-06, "loss": 0.0041, "step": 252450 }, { "epoch": 160.90503505417465, "grad_norm": 0.13966074585914612, "learning_rate": 6.710032943115979e-06, "loss": 0.0106, "step": 252460 }, { "epoch": 160.91140854047163, "grad_norm": 0.2845437526702881, "learning_rate": 6.707275266160201e-06, "loss": 0.0048, "step": 252470 }, { "epoch": 160.91778202676863, "grad_norm": 0.13424134254455566, "learning_rate": 6.704518115251912e-06, "loss": 0.0045, "step": 252480 }, { "epoch": 160.92415551306564, "grad_norm": 0.528134822845459, "learning_rate": 6.70176149042459e-06, "loss": 0.0063, "step": 252490 }, { "epoch": 160.93052899936265, "grad_norm": 0.11270952969789505, "learning_rate": 6.6990053917117714e-06, "loss": 0.0049, "step": 252500 }, { "epoch": 160.93690248565966, "grad_norm": 0.1047578826546669, "learning_rate": 6.696249819146899e-06, "loss": 0.0096, "step": 252510 }, { "epoch": 160.94327597195667, "grad_norm": 0.14316390454769135, "learning_rate": 6.693494772763486e-06, "loss": 0.0043, "step": 252520 }, { "epoch": 160.94964945825367, "grad_norm": 0.42974573373794556, "learning_rate": 6.690740252595002e-06, "loss": 0.0078, "step": 252530 }, { "epoch": 160.95602294455068, "grad_norm": 0.08633679896593094, "learning_rate": 6.687986258674917e-06, "loss": 0.0049, "step": 252540 }, { "epoch": 160.96239643084766, "grad_norm": 0.08872973173856735, "learning_rate": 6.685232791036677e-06, "loss": 0.0059, "step": 252550 }, { "epoch": 160.96876991714467, "grad_norm": 0.1279638260602951, "learning_rate": 6.6824798497137684e-06, "loss": 0.0038, "step": 252560 }, { "epoch": 160.97514340344168, "grad_norm": 0.11956082284450531, "learning_rate": 6.679727434739624e-06, "loss": 0.0056, "step": 252570 }, { "epoch": 160.98151688973869, "grad_norm": 0.21006576716899872, "learning_rate": 6.676975546147696e-06, "loss": 0.0043, "step": 252580 }, { "epoch": 160.9878903760357, "grad_norm": 0.22732862830162048, "learning_rate": 6.674224183971406e-06, "loss": 0.008, "step": 252590 }, { "epoch": 160.9942638623327, "grad_norm": 0.20140060782432556, "learning_rate": 6.671473348244217e-06, "loss": 0.0055, "step": 252600 }, { "epoch": 161.0006373486297, "grad_norm": 0.18626098334789276, "learning_rate": 6.668723038999519e-06, "loss": 0.0042, "step": 252610 }, { "epoch": 161.00701083492672, "grad_norm": 0.33952927589416504, "learning_rate": 6.665973256270758e-06, "loss": 0.0081, "step": 252620 }, { "epoch": 161.0133843212237, "grad_norm": 0.14344465732574463, "learning_rate": 6.663224000091323e-06, "loss": 0.006, "step": 252630 }, { "epoch": 161.0197578075207, "grad_norm": 0.10403929650783539, "learning_rate": 6.660475270494659e-06, "loss": 0.0028, "step": 252640 }, { "epoch": 161.0261312938177, "grad_norm": 0.10800457745790482, "learning_rate": 6.657727067514113e-06, "loss": 0.007, "step": 252650 }, { "epoch": 161.03250478011472, "grad_norm": 0.4187382161617279, "learning_rate": 6.654979391183119e-06, "loss": 0.0138, "step": 252660 }, { "epoch": 161.03887826641173, "grad_norm": 0.3019711971282959, "learning_rate": 6.652232241535039e-06, "loss": 0.0035, "step": 252670 }, { "epoch": 161.04525175270874, "grad_norm": 0.28545546531677246, "learning_rate": 6.649485618603285e-06, "loss": 0.006, "step": 252680 }, { "epoch": 161.05162523900574, "grad_norm": 0.13129472732543945, "learning_rate": 6.646739522421192e-06, "loss": 0.0043, "step": 252690 }, { "epoch": 161.05799872530275, "grad_norm": 0.16083818674087524, "learning_rate": 6.643993953022148e-06, "loss": 0.0034, "step": 252700 }, { "epoch": 161.06437221159973, "grad_norm": 0.10412372648715973, "learning_rate": 6.641248910439518e-06, "loss": 0.0144, "step": 252710 }, { "epoch": 161.07074569789674, "grad_norm": 0.15004868805408478, "learning_rate": 6.638504394706651e-06, "loss": 0.0028, "step": 252720 }, { "epoch": 161.07711918419375, "grad_norm": 0.19503192603588104, "learning_rate": 6.635760405856889e-06, "loss": 0.0029, "step": 252730 }, { "epoch": 161.08349267049076, "grad_norm": 0.13003942370414734, "learning_rate": 6.633016943923587e-06, "loss": 0.0067, "step": 252740 }, { "epoch": 161.08986615678776, "grad_norm": 0.12023241072893143, "learning_rate": 6.630274008940074e-06, "loss": 0.0054, "step": 252750 }, { "epoch": 161.09623964308477, "grad_norm": 0.3224061131477356, "learning_rate": 6.62753160093968e-06, "loss": 0.0068, "step": 252760 }, { "epoch": 161.10261312938178, "grad_norm": 0.20274734497070312, "learning_rate": 6.624789719955721e-06, "loss": 0.0034, "step": 252770 }, { "epoch": 161.1089866156788, "grad_norm": 0.10143863409757614, "learning_rate": 6.622048366021538e-06, "loss": 0.0037, "step": 252780 }, { "epoch": 161.11536010197577, "grad_norm": 0.10306897759437561, "learning_rate": 6.6193075391704e-06, "loss": 0.0035, "step": 252790 }, { "epoch": 161.12173358827278, "grad_norm": 0.1984298676252365, "learning_rate": 6.616567239435645e-06, "loss": 0.0054, "step": 252800 }, { "epoch": 161.12810707456978, "grad_norm": 0.34778255224227905, "learning_rate": 6.613827466850548e-06, "loss": 0.0061, "step": 252810 }, { "epoch": 161.1344805608668, "grad_norm": 0.5302714109420776, "learning_rate": 6.611088221448431e-06, "loss": 0.0112, "step": 252820 }, { "epoch": 161.1408540471638, "grad_norm": 0.2361699938774109, "learning_rate": 6.60834950326254e-06, "loss": 0.0035, "step": 252830 }, { "epoch": 161.1472275334608, "grad_norm": 0.09453482925891876, "learning_rate": 6.605611312326177e-06, "loss": 0.0065, "step": 252840 }, { "epoch": 161.15360101975782, "grad_norm": 0.140035018324852, "learning_rate": 6.602873648672614e-06, "loss": 0.0024, "step": 252850 }, { "epoch": 161.15997450605482, "grad_norm": 0.1106383204460144, "learning_rate": 6.600136512335103e-06, "loss": 0.0063, "step": 252860 }, { "epoch": 161.1663479923518, "grad_norm": 0.23034518957138062, "learning_rate": 6.597399903346901e-06, "loss": 0.0076, "step": 252870 }, { "epoch": 161.1727214786488, "grad_norm": 0.22979342937469482, "learning_rate": 6.594663821741281e-06, "loss": 0.0039, "step": 252880 }, { "epoch": 161.17909496494582, "grad_norm": 0.07598909735679626, "learning_rate": 6.591928267551478e-06, "loss": 0.0035, "step": 252890 }, { "epoch": 161.18546845124283, "grad_norm": 0.09656662493944168, "learning_rate": 6.589193240810731e-06, "loss": 0.0095, "step": 252900 }, { "epoch": 161.19184193753983, "grad_norm": 0.23786355555057526, "learning_rate": 6.586458741552259e-06, "loss": 0.0035, "step": 252910 }, { "epoch": 161.19821542383684, "grad_norm": 0.12815643846988678, "learning_rate": 6.5837247698093185e-06, "loss": 0.0095, "step": 252920 }, { "epoch": 161.20458891013385, "grad_norm": 0.24370881915092468, "learning_rate": 6.58099132561511e-06, "loss": 0.0084, "step": 252930 }, { "epoch": 161.21096239643086, "grad_norm": 0.19418533146381378, "learning_rate": 6.578258409002852e-06, "loss": 0.0041, "step": 252940 }, { "epoch": 161.21733588272787, "grad_norm": 0.20866377651691437, "learning_rate": 6.57552602000574e-06, "loss": 0.0054, "step": 252950 }, { "epoch": 161.22370936902485, "grad_norm": 0.1687077134847641, "learning_rate": 6.572794158656992e-06, "loss": 0.0053, "step": 252960 }, { "epoch": 161.23008285532185, "grad_norm": 0.12989316880702972, "learning_rate": 6.5700628249898036e-06, "loss": 0.0054, "step": 252970 }, { "epoch": 161.23645634161886, "grad_norm": 0.52376389503479, "learning_rate": 6.567332019037353e-06, "loss": 0.0115, "step": 252980 }, { "epoch": 161.24282982791587, "grad_norm": 0.14908303320407867, "learning_rate": 6.564601740832821e-06, "loss": 0.0055, "step": 252990 }, { "epoch": 161.24920331421288, "grad_norm": 0.1035449206829071, "learning_rate": 6.5618719904093906e-06, "loss": 0.0085, "step": 253000 }, { "epoch": 161.25557680050989, "grad_norm": 0.21071206033229828, "learning_rate": 6.559142767800236e-06, "loss": 0.0075, "step": 253010 }, { "epoch": 161.2619502868069, "grad_norm": 0.25245246291160583, "learning_rate": 6.556414073038503e-06, "loss": 0.0058, "step": 253020 }, { "epoch": 161.2683237731039, "grad_norm": 0.2313193380832672, "learning_rate": 6.553685906157364e-06, "loss": 0.0041, "step": 253030 }, { "epoch": 161.27469725940088, "grad_norm": 0.28023582696914673, "learning_rate": 6.550958267189944e-06, "loss": 0.0107, "step": 253040 }, { "epoch": 161.2810707456979, "grad_norm": 0.08412308990955353, "learning_rate": 6.548231156169421e-06, "loss": 0.0057, "step": 253050 }, { "epoch": 161.2874442319949, "grad_norm": 0.18665862083435059, "learning_rate": 6.545504573128908e-06, "loss": 0.0036, "step": 253060 }, { "epoch": 161.2938177182919, "grad_norm": 0.11711776256561279, "learning_rate": 6.5427785181015496e-06, "loss": 0.0038, "step": 253070 }, { "epoch": 161.3001912045889, "grad_norm": 0.3006271719932556, "learning_rate": 6.540052991120449e-06, "loss": 0.008, "step": 253080 }, { "epoch": 161.30656469088592, "grad_norm": 0.39344748854637146, "learning_rate": 6.5373279922187445e-06, "loss": 0.0102, "step": 253090 }, { "epoch": 161.31293817718293, "grad_norm": 0.4150097072124481, "learning_rate": 6.5346035214295444e-06, "loss": 0.0095, "step": 253100 }, { "epoch": 161.31931166347994, "grad_norm": 0.12918925285339355, "learning_rate": 6.531879578785949e-06, "loss": 0.0032, "step": 253110 }, { "epoch": 161.32568514977692, "grad_norm": 0.39680200815200806, "learning_rate": 6.529156164321043e-06, "loss": 0.0053, "step": 253120 }, { "epoch": 161.33205863607392, "grad_norm": 0.18534238636493683, "learning_rate": 6.52643327806795e-06, "loss": 0.0068, "step": 253130 }, { "epoch": 161.33843212237093, "grad_norm": 0.3637721836566925, "learning_rate": 6.523710920059733e-06, "loss": 0.0074, "step": 253140 }, { "epoch": 161.34480560866794, "grad_norm": 0.20477727055549622, "learning_rate": 6.520989090329482e-06, "loss": 0.0026, "step": 253150 }, { "epoch": 161.35117909496495, "grad_norm": 0.11809317767620087, "learning_rate": 6.518267788910248e-06, "loss": 0.0025, "step": 253160 }, { "epoch": 161.35755258126196, "grad_norm": 0.5145657658576965, "learning_rate": 6.515547015835139e-06, "loss": 0.0121, "step": 253170 }, { "epoch": 161.36392606755896, "grad_norm": 0.24371455609798431, "learning_rate": 6.512826771137171e-06, "loss": 0.0053, "step": 253180 }, { "epoch": 161.37029955385597, "grad_norm": 0.4193555414676666, "learning_rate": 6.510107054849423e-06, "loss": 0.0128, "step": 253190 }, { "epoch": 161.37667304015295, "grad_norm": 0.3621775507926941, "learning_rate": 6.507387867004927e-06, "loss": 0.0037, "step": 253200 }, { "epoch": 161.38304652644996, "grad_norm": 0.3174787163734436, "learning_rate": 6.504669207636749e-06, "loss": 0.0056, "step": 253210 }, { "epoch": 161.38942001274697, "grad_norm": 0.27153247594833374, "learning_rate": 6.50195107677789e-06, "loss": 0.0043, "step": 253220 }, { "epoch": 161.39579349904398, "grad_norm": 0.13234184682369232, "learning_rate": 6.499233474461402e-06, "loss": 0.0067, "step": 253230 }, { "epoch": 161.40216698534098, "grad_norm": 0.4092555642127991, "learning_rate": 6.496516400720293e-06, "loss": 0.0051, "step": 253240 }, { "epoch": 161.408540471638, "grad_norm": 0.2954556345939636, "learning_rate": 6.4937998555876016e-06, "loss": 0.0077, "step": 253250 }, { "epoch": 161.414913957935, "grad_norm": 0.2731178104877472, "learning_rate": 6.491083839096296e-06, "loss": 0.0059, "step": 253260 }, { "epoch": 161.421287444232, "grad_norm": 0.17777115106582642, "learning_rate": 6.488368351279417e-06, "loss": 0.0028, "step": 253270 }, { "epoch": 161.427660930529, "grad_norm": 0.23275259137153625, "learning_rate": 6.4856533921699384e-06, "loss": 0.0049, "step": 253280 }, { "epoch": 161.434034416826, "grad_norm": 0.14265017211437225, "learning_rate": 6.482938961800855e-06, "loss": 0.011, "step": 253290 }, { "epoch": 161.440407903123, "grad_norm": 0.07564099878072739, "learning_rate": 6.480225060205142e-06, "loss": 0.0028, "step": 253300 }, { "epoch": 161.44678138942, "grad_norm": 0.152714341878891, "learning_rate": 6.477511687415799e-06, "loss": 0.0036, "step": 253310 }, { "epoch": 161.45315487571702, "grad_norm": 0.43441271781921387, "learning_rate": 6.474798843465774e-06, "loss": 0.0058, "step": 253320 }, { "epoch": 161.45952836201403, "grad_norm": 0.0911206603050232, "learning_rate": 6.472086528388044e-06, "loss": 0.0041, "step": 253330 }, { "epoch": 161.46590184831103, "grad_norm": 0.0709262490272522, "learning_rate": 6.469374742215545e-06, "loss": 0.0039, "step": 253340 }, { "epoch": 161.47227533460804, "grad_norm": 0.0689813494682312, "learning_rate": 6.466663484981261e-06, "loss": 0.003, "step": 253350 }, { "epoch": 161.47864882090502, "grad_norm": 0.34797242283821106, "learning_rate": 6.4639527567181006e-06, "loss": 0.0051, "step": 253360 }, { "epoch": 161.48502230720203, "grad_norm": 0.13186340034008026, "learning_rate": 6.461242557459029e-06, "loss": 0.0035, "step": 253370 }, { "epoch": 161.49139579349904, "grad_norm": 0.2146262228488922, "learning_rate": 6.458532887236956e-06, "loss": 0.0071, "step": 253380 }, { "epoch": 161.49776927979605, "grad_norm": 0.09711022675037384, "learning_rate": 6.4558237460848435e-06, "loss": 0.0075, "step": 253390 }, { "epoch": 161.50414276609305, "grad_norm": 0.1504994034767151, "learning_rate": 6.4531151340355645e-06, "loss": 0.0081, "step": 253400 }, { "epoch": 161.51051625239006, "grad_norm": 0.31780338287353516, "learning_rate": 6.450407051122054e-06, "loss": 0.0035, "step": 253410 }, { "epoch": 161.51688973868707, "grad_norm": 0.16467295587062836, "learning_rate": 6.447699497377224e-06, "loss": 0.0071, "step": 253420 }, { "epoch": 161.52326322498408, "grad_norm": 0.35225215554237366, "learning_rate": 6.444992472833961e-06, "loss": 0.008, "step": 253430 }, { "epoch": 161.52963671128106, "grad_norm": 0.2938854694366455, "learning_rate": 6.442285977525153e-06, "loss": 0.006, "step": 253440 }, { "epoch": 161.53601019757807, "grad_norm": 0.15557554364204407, "learning_rate": 6.439580011483704e-06, "loss": 0.0044, "step": 253450 }, { "epoch": 161.54238368387507, "grad_norm": 0.2830461859703064, "learning_rate": 6.436874574742491e-06, "loss": 0.0063, "step": 253460 }, { "epoch": 161.54875717017208, "grad_norm": 0.09508409351110458, "learning_rate": 6.434169667334378e-06, "loss": 0.0056, "step": 253470 }, { "epoch": 161.5551306564691, "grad_norm": 0.10076386481523514, "learning_rate": 6.431465289292226e-06, "loss": 0.0044, "step": 253480 }, { "epoch": 161.5615041427661, "grad_norm": 0.13731041550636292, "learning_rate": 6.428761440648928e-06, "loss": 0.0038, "step": 253490 }, { "epoch": 161.5678776290631, "grad_norm": 0.21922741830348969, "learning_rate": 6.426058121437295e-06, "loss": 0.0079, "step": 253500 }, { "epoch": 161.5742511153601, "grad_norm": 0.14061985909938812, "learning_rate": 6.4233553316902065e-06, "loss": 0.004, "step": 253510 }, { "epoch": 161.5806246016571, "grad_norm": 0.10380091518163681, "learning_rate": 6.420653071440486e-06, "loss": 0.0098, "step": 253520 }, { "epoch": 161.5869980879541, "grad_norm": 0.1201348528265953, "learning_rate": 6.417951340720996e-06, "loss": 0.0063, "step": 253530 }, { "epoch": 161.5933715742511, "grad_norm": 0.12085968255996704, "learning_rate": 6.415250139564521e-06, "loss": 0.0034, "step": 253540 }, { "epoch": 161.59974506054812, "grad_norm": 0.10336624830961227, "learning_rate": 6.4125494680039255e-06, "loss": 0.0038, "step": 253550 }, { "epoch": 161.60611854684512, "grad_norm": 0.10207393020391464, "learning_rate": 6.409849326071992e-06, "loss": 0.0049, "step": 253560 }, { "epoch": 161.61249203314213, "grad_norm": 0.18328197300434113, "learning_rate": 6.407149713801564e-06, "loss": 0.0046, "step": 253570 }, { "epoch": 161.61886551943914, "grad_norm": 0.12356777489185333, "learning_rate": 6.404450631225406e-06, "loss": 0.0072, "step": 253580 }, { "epoch": 161.62523900573615, "grad_norm": 0.18557687103748322, "learning_rate": 6.40175207837635e-06, "loss": 0.0057, "step": 253590 }, { "epoch": 161.63161249203316, "grad_norm": 0.42815083265304565, "learning_rate": 6.399054055287162e-06, "loss": 0.0066, "step": 253600 }, { "epoch": 161.63798597833014, "grad_norm": 0.33512571454048157, "learning_rate": 6.39635656199064e-06, "loss": 0.0054, "step": 253610 }, { "epoch": 161.64435946462714, "grad_norm": 0.28320884704589844, "learning_rate": 6.393659598519542e-06, "loss": 0.0059, "step": 253620 }, { "epoch": 161.65073295092415, "grad_norm": 0.11752595007419586, "learning_rate": 6.390963164906661e-06, "loss": 0.0067, "step": 253630 }, { "epoch": 161.65710643722116, "grad_norm": 0.22324097156524658, "learning_rate": 6.38826726118475e-06, "loss": 0.0041, "step": 253640 }, { "epoch": 161.66347992351817, "grad_norm": 0.19764484465122223, "learning_rate": 6.385571887386565e-06, "loss": 0.0057, "step": 253650 }, { "epoch": 161.66985340981518, "grad_norm": 0.6442509293556213, "learning_rate": 6.382877043544855e-06, "loss": 0.0076, "step": 253660 }, { "epoch": 161.67622689611218, "grad_norm": 0.09648845344781876, "learning_rate": 6.38018272969238e-06, "loss": 0.0036, "step": 253670 }, { "epoch": 161.6826003824092, "grad_norm": 0.24914279580116272, "learning_rate": 6.377488945861865e-06, "loss": 0.0111, "step": 253680 }, { "epoch": 161.68897386870617, "grad_norm": 0.0768633782863617, "learning_rate": 6.37479569208605e-06, "loss": 0.0054, "step": 253690 }, { "epoch": 161.69534735500318, "grad_norm": 0.11585576832294464, "learning_rate": 6.372102968397647e-06, "loss": 0.0085, "step": 253700 }, { "epoch": 161.7017208413002, "grad_norm": 0.5975128412246704, "learning_rate": 6.369410774829393e-06, "loss": 0.0084, "step": 253710 }, { "epoch": 161.7080943275972, "grad_norm": 0.43086275458335876, "learning_rate": 6.366719111413993e-06, "loss": 0.0082, "step": 253720 }, { "epoch": 161.7144678138942, "grad_norm": 0.0930895060300827, "learning_rate": 6.364027978184151e-06, "loss": 0.0041, "step": 253730 }, { "epoch": 161.7208413001912, "grad_norm": 0.19688647985458374, "learning_rate": 6.361337375172566e-06, "loss": 0.0046, "step": 253740 }, { "epoch": 161.72721478648822, "grad_norm": 0.09625977277755737, "learning_rate": 6.358647302411924e-06, "loss": 0.0028, "step": 253750 }, { "epoch": 161.73358827278523, "grad_norm": 0.10741595178842545, "learning_rate": 6.3559577599349384e-06, "loss": 0.0079, "step": 253760 }, { "epoch": 161.7399617590822, "grad_norm": 0.1447523683309555, "learning_rate": 6.353268747774266e-06, "loss": 0.0031, "step": 253770 }, { "epoch": 161.7463352453792, "grad_norm": 0.27460092306137085, "learning_rate": 6.350580265962592e-06, "loss": 0.0105, "step": 253780 }, { "epoch": 161.75270873167622, "grad_norm": 0.14769232273101807, "learning_rate": 6.3478923145325666e-06, "loss": 0.0028, "step": 253790 }, { "epoch": 161.75908221797323, "grad_norm": 0.16715198755264282, "learning_rate": 6.345204893516876e-06, "loss": 0.0044, "step": 253800 }, { "epoch": 161.76545570427024, "grad_norm": 0.46376916766166687, "learning_rate": 6.342518002948162e-06, "loss": 0.013, "step": 253810 }, { "epoch": 161.77182919056725, "grad_norm": 0.1507163643836975, "learning_rate": 6.339831642859073e-06, "loss": 0.0095, "step": 253820 }, { "epoch": 161.77820267686425, "grad_norm": 0.11835210025310516, "learning_rate": 6.3371458132822415e-06, "loss": 0.0034, "step": 253830 }, { "epoch": 161.78457616316126, "grad_norm": 0.26916179060935974, "learning_rate": 6.3344605142503234e-06, "loss": 0.0053, "step": 253840 }, { "epoch": 161.79094964945824, "grad_norm": 0.1425647884607315, "learning_rate": 6.331775745795937e-06, "loss": 0.018, "step": 253850 }, { "epoch": 161.79732313575525, "grad_norm": 0.6334159970283508, "learning_rate": 6.329091507951707e-06, "loss": 0.004, "step": 253860 }, { "epoch": 161.80369662205226, "grad_norm": 0.0802663192152977, "learning_rate": 6.326407800750234e-06, "loss": 0.0049, "step": 253870 }, { "epoch": 161.81007010834927, "grad_norm": 0.1111469566822052, "learning_rate": 6.323724624224153e-06, "loss": 0.0064, "step": 253880 }, { "epoch": 161.81644359464627, "grad_norm": 0.12451963871717453, "learning_rate": 6.321041978406056e-06, "loss": 0.0032, "step": 253890 }, { "epoch": 161.82281708094328, "grad_norm": 0.15437188744544983, "learning_rate": 6.318359863328532e-06, "loss": 0.0048, "step": 253900 }, { "epoch": 161.8291905672403, "grad_norm": 0.2661720812320709, "learning_rate": 6.3156782790241735e-06, "loss": 0.003, "step": 253910 }, { "epoch": 161.8355640535373, "grad_norm": 0.15435320138931274, "learning_rate": 6.312997225525591e-06, "loss": 0.0066, "step": 253920 }, { "epoch": 161.84193753983428, "grad_norm": 0.14162756502628326, "learning_rate": 6.3103167028653155e-06, "loss": 0.0041, "step": 253930 }, { "epoch": 161.84831102613128, "grad_norm": 0.22836211323738098, "learning_rate": 6.307636711075954e-06, "loss": 0.0161, "step": 253940 }, { "epoch": 161.8546845124283, "grad_norm": 0.16064012050628662, "learning_rate": 6.304957250190047e-06, "loss": 0.0181, "step": 253950 }, { "epoch": 161.8610579987253, "grad_norm": 0.24012206494808197, "learning_rate": 6.3022783202401905e-06, "loss": 0.0169, "step": 253960 }, { "epoch": 161.8674314850223, "grad_norm": 0.11979655921459198, "learning_rate": 6.299599921258881e-06, "loss": 0.0059, "step": 253970 }, { "epoch": 161.87380497131932, "grad_norm": 0.1458776742219925, "learning_rate": 6.296922053278709e-06, "loss": 0.0054, "step": 253980 }, { "epoch": 161.88017845761632, "grad_norm": 0.08234068751335144, "learning_rate": 6.294244716332193e-06, "loss": 0.0087, "step": 253990 }, { "epoch": 161.88655194391333, "grad_norm": 0.08237392455339432, "learning_rate": 6.291567910451873e-06, "loss": 0.0035, "step": 254000 }, { "epoch": 161.8929254302103, "grad_norm": 0.30985498428344727, "learning_rate": 6.2888916356702575e-06, "loss": 0.0085, "step": 254010 }, { "epoch": 161.89929891650732, "grad_norm": 0.2585168778896332, "learning_rate": 6.286215892019886e-06, "loss": 0.0036, "step": 254020 }, { "epoch": 161.90567240280433, "grad_norm": 0.20364326238632202, "learning_rate": 6.283540679533267e-06, "loss": 0.0124, "step": 254030 }, { "epoch": 161.91204588910134, "grad_norm": 0.26355576515197754, "learning_rate": 6.280865998242908e-06, "loss": 0.0034, "step": 254040 }, { "epoch": 161.91841937539834, "grad_norm": 0.09487909078598022, "learning_rate": 6.278191848181287e-06, "loss": 0.0081, "step": 254050 }, { "epoch": 161.92479286169535, "grad_norm": 0.19497710466384888, "learning_rate": 6.275518229380939e-06, "loss": 0.0084, "step": 254060 }, { "epoch": 161.93116634799236, "grad_norm": 0.11735526472330093, "learning_rate": 6.272845141874306e-06, "loss": 0.0034, "step": 254070 }, { "epoch": 161.93753983428937, "grad_norm": 0.0974050983786583, "learning_rate": 6.270172585693901e-06, "loss": 0.0029, "step": 254080 }, { "epoch": 161.94391332058635, "grad_norm": 0.11214209347963333, "learning_rate": 6.267500560872175e-06, "loss": 0.0048, "step": 254090 }, { "epoch": 161.95028680688335, "grad_norm": 0.2171473205089569, "learning_rate": 6.26482906744163e-06, "loss": 0.0041, "step": 254100 }, { "epoch": 161.95666029318036, "grad_norm": 0.6265444755554199, "learning_rate": 6.2621581054346845e-06, "loss": 0.0073, "step": 254110 }, { "epoch": 161.96303377947737, "grad_norm": 0.10811761021614075, "learning_rate": 6.259487674883824e-06, "loss": 0.0053, "step": 254120 }, { "epoch": 161.96940726577438, "grad_norm": 0.470633327960968, "learning_rate": 6.256817775821483e-06, "loss": 0.0088, "step": 254130 }, { "epoch": 161.9757807520714, "grad_norm": 0.07899602502584457, "learning_rate": 6.254148408280114e-06, "loss": 0.0069, "step": 254140 }, { "epoch": 161.9821542383684, "grad_norm": 0.43521830439567566, "learning_rate": 6.251479572292124e-06, "loss": 0.0087, "step": 254150 }, { "epoch": 161.9885277246654, "grad_norm": 0.2782571017742157, "learning_rate": 6.2488112678899814e-06, "loss": 0.0099, "step": 254160 }, { "epoch": 161.99490121096238, "grad_norm": 0.15327459573745728, "learning_rate": 6.2461434951060825e-06, "loss": 0.0042, "step": 254170 }, { "epoch": 162.0012746972594, "grad_norm": 0.15110871195793152, "learning_rate": 6.243476253972858e-06, "loss": 0.0026, "step": 254180 }, { "epoch": 162.0076481835564, "grad_norm": 0.16806595027446747, "learning_rate": 6.240809544522697e-06, "loss": 0.0066, "step": 254190 }, { "epoch": 162.0140216698534, "grad_norm": 0.1342698484659195, "learning_rate": 6.238143366788024e-06, "loss": 0.0062, "step": 254200 }, { "epoch": 162.0203951561504, "grad_norm": 0.07840429991483688, "learning_rate": 6.235477720801225e-06, "loss": 0.0059, "step": 254210 }, { "epoch": 162.02676864244742, "grad_norm": 0.19659164547920227, "learning_rate": 6.232812606594696e-06, "loss": 0.006, "step": 254220 }, { "epoch": 162.03314212874443, "grad_norm": 0.3752891421318054, "learning_rate": 6.23014802420081e-06, "loss": 0.0091, "step": 254230 }, { "epoch": 162.03951561504144, "grad_norm": 0.13746392726898193, "learning_rate": 6.227483973651965e-06, "loss": 0.0052, "step": 254240 }, { "epoch": 162.04588910133845, "grad_norm": 0.19641567766666412, "learning_rate": 6.224820454980501e-06, "loss": 0.0115, "step": 254250 }, { "epoch": 162.05226258763543, "grad_norm": 0.2830568253993988, "learning_rate": 6.222157468218809e-06, "loss": 0.0031, "step": 254260 }, { "epoch": 162.05863607393243, "grad_norm": 0.24496178328990936, "learning_rate": 6.2194950133992235e-06, "loss": 0.0045, "step": 254270 }, { "epoch": 162.06500956022944, "grad_norm": 0.1181463822722435, "learning_rate": 6.2168330905541365e-06, "loss": 0.0073, "step": 254280 }, { "epoch": 162.07138304652645, "grad_norm": 0.1984294205904007, "learning_rate": 6.214171699715837e-06, "loss": 0.0054, "step": 254290 }, { "epoch": 162.07775653282346, "grad_norm": 0.1267748773097992, "learning_rate": 6.211510840916707e-06, "loss": 0.0056, "step": 254300 }, { "epoch": 162.08413001912047, "grad_norm": 0.3057089149951935, "learning_rate": 6.208850514189063e-06, "loss": 0.0168, "step": 254310 }, { "epoch": 162.09050350541747, "grad_norm": 0.5314648151397705, "learning_rate": 6.20619071956523e-06, "loss": 0.0134, "step": 254320 }, { "epoch": 162.09687699171448, "grad_norm": 0.09041726589202881, "learning_rate": 6.203531457077516e-06, "loss": 0.0039, "step": 254330 }, { "epoch": 162.10325047801146, "grad_norm": 0.19184601306915283, "learning_rate": 6.200872726758261e-06, "loss": 0.0047, "step": 254340 }, { "epoch": 162.10962396430847, "grad_norm": 0.21117322146892548, "learning_rate": 6.198214528639751e-06, "loss": 0.0097, "step": 254350 }, { "epoch": 162.11599745060548, "grad_norm": 0.185182124376297, "learning_rate": 6.195556862754292e-06, "loss": 0.0051, "step": 254360 }, { "epoch": 162.12237093690248, "grad_norm": 0.11998289078474045, "learning_rate": 6.192899729134166e-06, "loss": 0.0041, "step": 254370 }, { "epoch": 162.1287444231995, "grad_norm": 0.3938385844230652, "learning_rate": 6.190243127811679e-06, "loss": 0.0128, "step": 254380 }, { "epoch": 162.1351179094965, "grad_norm": 0.13166621327400208, "learning_rate": 6.1875870588191e-06, "loss": 0.005, "step": 254390 }, { "epoch": 162.1414913957935, "grad_norm": 0.19312089681625366, "learning_rate": 6.1849315221887025e-06, "loss": 0.0078, "step": 254400 }, { "epoch": 162.14786488209052, "grad_norm": 0.12611065804958344, "learning_rate": 6.182276517952751e-06, "loss": 0.0024, "step": 254410 }, { "epoch": 162.1542383683875, "grad_norm": 0.10172491520643234, "learning_rate": 6.179622046143513e-06, "loss": 0.0055, "step": 254420 }, { "epoch": 162.1606118546845, "grad_norm": 0.07439424097537994, "learning_rate": 6.176968106793246e-06, "loss": 0.0071, "step": 254430 }, { "epoch": 162.1669853409815, "grad_norm": 0.1928384155035019, "learning_rate": 6.174314699934191e-06, "loss": 0.0042, "step": 254440 }, { "epoch": 162.17335882727852, "grad_norm": 0.544295608997345, "learning_rate": 6.171661825598579e-06, "loss": 0.0064, "step": 254450 }, { "epoch": 162.17973231357553, "grad_norm": 0.2155836671590805, "learning_rate": 6.169009483818672e-06, "loss": 0.0032, "step": 254460 }, { "epoch": 162.18610579987254, "grad_norm": 0.21172761917114258, "learning_rate": 6.166357674626677e-06, "loss": 0.0052, "step": 254470 }, { "epoch": 162.19247928616954, "grad_norm": 0.17951487004756927, "learning_rate": 6.163706398054825e-06, "loss": 0.0025, "step": 254480 }, { "epoch": 162.19885277246655, "grad_norm": 0.16628625988960266, "learning_rate": 6.161055654135328e-06, "loss": 0.0054, "step": 254490 }, { "epoch": 162.20522625876353, "grad_norm": 0.14611080288887024, "learning_rate": 6.1584054429003836e-06, "loss": 0.0057, "step": 254500 }, { "epoch": 162.21159974506054, "grad_norm": 0.176676943898201, "learning_rate": 6.155755764382215e-06, "loss": 0.0092, "step": 254510 }, { "epoch": 162.21797323135755, "grad_norm": 0.19590415060520172, "learning_rate": 6.1531066186130135e-06, "loss": 0.0073, "step": 254520 }, { "epoch": 162.22434671765456, "grad_norm": 0.06949076056480408, "learning_rate": 6.150458005624965e-06, "loss": 0.0062, "step": 254530 }, { "epoch": 162.23072020395156, "grad_norm": 0.12326043844223022, "learning_rate": 6.147809925450243e-06, "loss": 0.004, "step": 254540 }, { "epoch": 162.23709369024857, "grad_norm": 0.11618439853191376, "learning_rate": 6.145162378121044e-06, "loss": 0.0045, "step": 254550 }, { "epoch": 162.24346717654558, "grad_norm": 0.29747405648231506, "learning_rate": 6.142515363669526e-06, "loss": 0.0029, "step": 254560 }, { "epoch": 162.2498406628426, "grad_norm": 0.06547699123620987, "learning_rate": 6.139868882127853e-06, "loss": 0.0066, "step": 254570 }, { "epoch": 162.25621414913957, "grad_norm": 0.49148035049438477, "learning_rate": 6.137222933528175e-06, "loss": 0.0117, "step": 254580 }, { "epoch": 162.26258763543657, "grad_norm": 0.4753580093383789, "learning_rate": 6.134577517902662e-06, "loss": 0.0054, "step": 254590 }, { "epoch": 162.26896112173358, "grad_norm": 0.06167599931359291, "learning_rate": 6.131932635283449e-06, "loss": 0.0041, "step": 254600 }, { "epoch": 162.2753346080306, "grad_norm": 0.12271980196237564, "learning_rate": 6.129288285702672e-06, "loss": 0.003, "step": 254610 }, { "epoch": 162.2817080943276, "grad_norm": 0.3528798818588257, "learning_rate": 6.126644469192455e-06, "loss": 0.0078, "step": 254620 }, { "epoch": 162.2880815806246, "grad_norm": 0.15964409708976746, "learning_rate": 6.12400118578495e-06, "loss": 0.0277, "step": 254630 }, { "epoch": 162.2944550669216, "grad_norm": 0.600974977016449, "learning_rate": 6.121358435512237e-06, "loss": 0.0049, "step": 254640 }, { "epoch": 162.30082855321862, "grad_norm": 0.481780469417572, "learning_rate": 6.118716218406456e-06, "loss": 0.0105, "step": 254650 }, { "epoch": 162.3072020395156, "grad_norm": 0.08706547319889069, "learning_rate": 6.116074534499694e-06, "loss": 0.003, "step": 254660 }, { "epoch": 162.3135755258126, "grad_norm": 0.10289126634597778, "learning_rate": 6.1134333838240795e-06, "loss": 0.0026, "step": 254670 }, { "epoch": 162.31994901210962, "grad_norm": 0.14628693461418152, "learning_rate": 6.110792766411666e-06, "loss": 0.0045, "step": 254680 }, { "epoch": 162.32632249840663, "grad_norm": 0.21782594919204712, "learning_rate": 6.108152682294565e-06, "loss": 0.0053, "step": 254690 }, { "epoch": 162.33269598470363, "grad_norm": 0.08645372092723846, "learning_rate": 6.1055131315048584e-06, "loss": 0.005, "step": 254700 }, { "epoch": 162.33906947100064, "grad_norm": 0.23734568059444427, "learning_rate": 6.102874114074603e-06, "loss": 0.0081, "step": 254710 }, { "epoch": 162.34544295729765, "grad_norm": 0.11232121288776398, "learning_rate": 6.100235630035866e-06, "loss": 0.0041, "step": 254720 }, { "epoch": 162.35181644359466, "grad_norm": 0.14696316421031952, "learning_rate": 6.097597679420719e-06, "loss": 0.0041, "step": 254730 }, { "epoch": 162.35818992989164, "grad_norm": 0.322108656167984, "learning_rate": 6.0949602622612175e-06, "loss": 0.0066, "step": 254740 }, { "epoch": 162.36456341618864, "grad_norm": 0.13060221076011658, "learning_rate": 6.0923233785893985e-06, "loss": 0.0033, "step": 254750 }, { "epoch": 162.37093690248565, "grad_norm": 0.2758280038833618, "learning_rate": 6.089687028437296e-06, "loss": 0.0064, "step": 254760 }, { "epoch": 162.37731038878266, "grad_norm": 0.07307418435811996, "learning_rate": 6.087051211836969e-06, "loss": 0.0152, "step": 254770 }, { "epoch": 162.38368387507967, "grad_norm": 0.1560930609703064, "learning_rate": 6.084415928820425e-06, "loss": 0.0096, "step": 254780 }, { "epoch": 162.39005736137668, "grad_norm": 0.2174450308084488, "learning_rate": 6.081781179419693e-06, "loss": 0.0036, "step": 254790 }, { "epoch": 162.39643084767368, "grad_norm": 0.23091496527194977, "learning_rate": 6.079146963666776e-06, "loss": 0.0056, "step": 254800 }, { "epoch": 162.4028043339707, "grad_norm": 0.2875189781188965, "learning_rate": 6.0765132815937095e-06, "loss": 0.0071, "step": 254810 }, { "epoch": 162.4091778202677, "grad_norm": 0.7181041836738586, "learning_rate": 6.0738801332324566e-06, "loss": 0.0098, "step": 254820 }, { "epoch": 162.41555130656468, "grad_norm": 0.2144593596458435, "learning_rate": 6.071247518615048e-06, "loss": 0.0171, "step": 254830 }, { "epoch": 162.4219247928617, "grad_norm": 0.11796090751886368, "learning_rate": 6.068615437773445e-06, "loss": 0.0061, "step": 254840 }, { "epoch": 162.4282982791587, "grad_norm": 0.2768888771533966, "learning_rate": 6.065983890739663e-06, "loss": 0.0076, "step": 254850 }, { "epoch": 162.4346717654557, "grad_norm": 0.18627695739269257, "learning_rate": 6.063352877545636e-06, "loss": 0.0048, "step": 254860 }, { "epoch": 162.4410452517527, "grad_norm": 0.20029132068157196, "learning_rate": 6.060722398223362e-06, "loss": 0.0096, "step": 254870 }, { "epoch": 162.44741873804972, "grad_norm": 0.14388957619667053, "learning_rate": 6.058092452804803e-06, "loss": 0.0071, "step": 254880 }, { "epoch": 162.45379222434673, "grad_norm": 0.12309885025024414, "learning_rate": 6.055463041321901e-06, "loss": 0.0038, "step": 254890 }, { "epoch": 162.46016571064374, "grad_norm": 0.13340157270431519, "learning_rate": 6.0528341638066085e-06, "loss": 0.0081, "step": 254900 }, { "epoch": 162.46653919694072, "grad_norm": 0.09973417967557907, "learning_rate": 6.050205820290883e-06, "loss": 0.0045, "step": 254910 }, { "epoch": 162.47291268323772, "grad_norm": 0.05448878929018974, "learning_rate": 6.047578010806654e-06, "loss": 0.0119, "step": 254920 }, { "epoch": 162.47928616953473, "grad_norm": 0.22304055094718933, "learning_rate": 6.044950735385852e-06, "loss": 0.0123, "step": 254930 }, { "epoch": 162.48565965583174, "grad_norm": 0.21640053391456604, "learning_rate": 6.0423239940603834e-06, "loss": 0.0098, "step": 254940 }, { "epoch": 162.49203314212875, "grad_norm": 0.30141687393188477, "learning_rate": 6.039697786862203e-06, "loss": 0.0066, "step": 254950 }, { "epoch": 162.49840662842576, "grad_norm": 0.5053894519805908, "learning_rate": 6.037072113823178e-06, "loss": 0.0054, "step": 254960 }, { "epoch": 162.50478011472276, "grad_norm": 0.09260153025388718, "learning_rate": 6.03444697497525e-06, "loss": 0.0048, "step": 254970 }, { "epoch": 162.51115360101977, "grad_norm": 0.2002430409193039, "learning_rate": 6.031822370350287e-06, "loss": 0.0042, "step": 254980 }, { "epoch": 162.51752708731675, "grad_norm": 0.1358792930841446, "learning_rate": 6.029198299980216e-06, "loss": 0.0111, "step": 254990 }, { "epoch": 162.52390057361376, "grad_norm": 0.27653035521507263, "learning_rate": 6.026574763896881e-06, "loss": 0.0058, "step": 255000 }, { "epoch": 162.53027405991077, "grad_norm": 0.4552176892757416, "learning_rate": 6.023951762132191e-06, "loss": 0.0047, "step": 255010 }, { "epoch": 162.53664754620777, "grad_norm": 0.30221667885780334, "learning_rate": 6.021329294718009e-06, "loss": 0.0073, "step": 255020 }, { "epoch": 162.54302103250478, "grad_norm": 0.10389645397663116, "learning_rate": 6.018707361686199e-06, "loss": 0.0089, "step": 255030 }, { "epoch": 162.5493945188018, "grad_norm": 0.13471311330795288, "learning_rate": 6.016085963068607e-06, "loss": 0.0039, "step": 255040 }, { "epoch": 162.5557680050988, "grad_norm": 0.16570836305618286, "learning_rate": 6.013465098897109e-06, "loss": 0.0059, "step": 255050 }, { "epoch": 162.5621414913958, "grad_norm": 0.1958315074443817, "learning_rate": 6.010844769203539e-06, "loss": 0.0116, "step": 255060 }, { "epoch": 162.5685149776928, "grad_norm": 0.26741161942481995, "learning_rate": 6.008224974019738e-06, "loss": 0.0046, "step": 255070 }, { "epoch": 162.5748884639898, "grad_norm": 0.2023056596517563, "learning_rate": 6.005605713377532e-06, "loss": 0.0058, "step": 255080 }, { "epoch": 162.5812619502868, "grad_norm": 0.17727826535701752, "learning_rate": 6.002986987308762e-06, "loss": 0.0088, "step": 255090 }, { "epoch": 162.5876354365838, "grad_norm": 0.13072745501995087, "learning_rate": 6.00036879584524e-06, "loss": 0.0052, "step": 255100 }, { "epoch": 162.59400892288082, "grad_norm": 0.28771916031837463, "learning_rate": 5.9977511390187816e-06, "loss": 0.0048, "step": 255110 }, { "epoch": 162.60038240917783, "grad_norm": 0.4040982127189636, "learning_rate": 5.995134016861176e-06, "loss": 0.0046, "step": 255120 }, { "epoch": 162.60675589547483, "grad_norm": 0.17650102078914642, "learning_rate": 5.992517429404254e-06, "loss": 0.0178, "step": 255130 }, { "epoch": 162.61312938177184, "grad_norm": 0.22778436541557312, "learning_rate": 5.989901376679796e-06, "loss": 0.0036, "step": 255140 }, { "epoch": 162.61950286806882, "grad_norm": 0.4495355486869812, "learning_rate": 5.987285858719588e-06, "loss": 0.0054, "step": 255150 }, { "epoch": 162.62587635436583, "grad_norm": 0.38899722695350647, "learning_rate": 5.984670875555404e-06, "loss": 0.0094, "step": 255160 }, { "epoch": 162.63224984066284, "grad_norm": 0.14581172168254852, "learning_rate": 5.9820564272190295e-06, "loss": 0.0133, "step": 255170 }, { "epoch": 162.63862332695984, "grad_norm": 0.11400624364614487, "learning_rate": 5.979442513742234e-06, "loss": 0.0051, "step": 255180 }, { "epoch": 162.64499681325685, "grad_norm": 0.1035243347287178, "learning_rate": 5.976829135156775e-06, "loss": 0.0055, "step": 255190 }, { "epoch": 162.65137029955386, "grad_norm": 0.26353582739830017, "learning_rate": 5.974216291494406e-06, "loss": 0.0105, "step": 255200 }, { "epoch": 162.65774378585087, "grad_norm": 0.08253810554742813, "learning_rate": 5.9716039827868665e-06, "loss": 0.005, "step": 255210 }, { "epoch": 162.66411727214788, "grad_norm": 0.2982611358165741, "learning_rate": 5.968992209065915e-06, "loss": 0.0038, "step": 255220 }, { "epoch": 162.67049075844486, "grad_norm": 0.12283468246459961, "learning_rate": 5.9663809703632835e-06, "loss": 0.0043, "step": 255230 }, { "epoch": 162.67686424474186, "grad_norm": 0.19307976961135864, "learning_rate": 5.963770266710694e-06, "loss": 0.0103, "step": 255240 }, { "epoch": 162.68323773103887, "grad_norm": 0.13388055562973022, "learning_rate": 5.961160098139868e-06, "loss": 0.0028, "step": 255250 }, { "epoch": 162.68961121733588, "grad_norm": 0.636547863483429, "learning_rate": 5.958550464682533e-06, "loss": 0.0085, "step": 255260 }, { "epoch": 162.6959847036329, "grad_norm": 0.17660517990589142, "learning_rate": 5.9559413663703935e-06, "loss": 0.0059, "step": 255270 }, { "epoch": 162.7023581899299, "grad_norm": 0.13637769222259521, "learning_rate": 5.953332803235151e-06, "loss": 0.0095, "step": 255280 }, { "epoch": 162.7087316762269, "grad_norm": 0.16808930039405823, "learning_rate": 5.950724775308491e-06, "loss": 0.0048, "step": 255290 }, { "epoch": 162.7151051625239, "grad_norm": 0.10062379390001297, "learning_rate": 5.948117282622123e-06, "loss": 0.0043, "step": 255300 }, { "epoch": 162.7214786488209, "grad_norm": 0.5152328610420227, "learning_rate": 5.945510325207726e-06, "loss": 0.012, "step": 255310 }, { "epoch": 162.7278521351179, "grad_norm": 0.35340917110443115, "learning_rate": 5.942903903096969e-06, "loss": 0.0039, "step": 255320 }, { "epoch": 162.7342256214149, "grad_norm": 0.11803494393825531, "learning_rate": 5.940298016321522e-06, "loss": 0.006, "step": 255330 }, { "epoch": 162.74059910771192, "grad_norm": 0.8124023079872131, "learning_rate": 5.937692664913069e-06, "loss": 0.0107, "step": 255340 }, { "epoch": 162.74697259400892, "grad_norm": 0.10941209644079208, "learning_rate": 5.935087848903231e-06, "loss": 0.0061, "step": 255350 }, { "epoch": 162.75334608030593, "grad_norm": 0.3564053773880005, "learning_rate": 5.932483568323688e-06, "loss": 0.0048, "step": 255360 }, { "epoch": 162.75971956660294, "grad_norm": 0.06911211460828781, "learning_rate": 5.9298798232060695e-06, "loss": 0.005, "step": 255370 }, { "epoch": 162.76609305289995, "grad_norm": 0.5044414401054382, "learning_rate": 5.927276613582039e-06, "loss": 0.0083, "step": 255380 }, { "epoch": 162.77246653919693, "grad_norm": 0.08415373414754868, "learning_rate": 5.924673939483194e-06, "loss": 0.0027, "step": 255390 }, { "epoch": 162.77884002549393, "grad_norm": 0.27202457189559937, "learning_rate": 5.922071800941181e-06, "loss": 0.0035, "step": 255400 }, { "epoch": 162.78521351179094, "grad_norm": 0.3415262997150421, "learning_rate": 5.919470197987598e-06, "loss": 0.0052, "step": 255410 }, { "epoch": 162.79158699808795, "grad_norm": 0.5920998454093933, "learning_rate": 5.9168691306540954e-06, "loss": 0.0083, "step": 255420 }, { "epoch": 162.79796048438496, "grad_norm": 0.12890274822711945, "learning_rate": 5.914268598972234e-06, "loss": 0.009, "step": 255430 }, { "epoch": 162.80433397068197, "grad_norm": 0.3040443956851959, "learning_rate": 5.911668602973641e-06, "loss": 0.004, "step": 255440 }, { "epoch": 162.81070745697897, "grad_norm": 0.22046145796775818, "learning_rate": 5.909069142689905e-06, "loss": 0.0024, "step": 255450 }, { "epoch": 162.81708094327598, "grad_norm": 0.15456300973892212, "learning_rate": 5.906470218152604e-06, "loss": 0.0079, "step": 255460 }, { "epoch": 162.823454429573, "grad_norm": 0.1829480081796646, "learning_rate": 5.903871829393315e-06, "loss": 0.0042, "step": 255470 }, { "epoch": 162.82982791586997, "grad_norm": 0.20426884293556213, "learning_rate": 5.901273976443622e-06, "loss": 0.0023, "step": 255480 }, { "epoch": 162.83620140216698, "grad_norm": 0.17241692543029785, "learning_rate": 5.89867665933509e-06, "loss": 0.0073, "step": 255490 }, { "epoch": 162.842574888464, "grad_norm": 0.07868527621030807, "learning_rate": 5.89607987809927e-06, "loss": 0.0029, "step": 255500 }, { "epoch": 162.848948374761, "grad_norm": 0.08100424706935883, "learning_rate": 5.893483632767715e-06, "loss": 0.0086, "step": 255510 }, { "epoch": 162.855321861058, "grad_norm": 0.07883677631616592, "learning_rate": 5.890887923371996e-06, "loss": 0.0135, "step": 255520 }, { "epoch": 162.861695347355, "grad_norm": 0.0867379680275917, "learning_rate": 5.888292749943613e-06, "loss": 0.0055, "step": 255530 }, { "epoch": 162.86806883365202, "grad_norm": 0.12726707756519318, "learning_rate": 5.885698112514137e-06, "loss": 0.0049, "step": 255540 }, { "epoch": 162.87444231994903, "grad_norm": 0.11352773755788803, "learning_rate": 5.883104011115065e-06, "loss": 0.007, "step": 255550 }, { "epoch": 162.880815806246, "grad_norm": 0.30590322613716125, "learning_rate": 5.8805104457779535e-06, "loss": 0.0062, "step": 255560 }, { "epoch": 162.887189292543, "grad_norm": 0.06058123707771301, "learning_rate": 5.877917416534273e-06, "loss": 0.0047, "step": 255570 }, { "epoch": 162.89356277884002, "grad_norm": 0.13767066597938538, "learning_rate": 5.875324923415571e-06, "loss": 0.0048, "step": 255580 }, { "epoch": 162.89993626513703, "grad_norm": 0.26372697949409485, "learning_rate": 5.872732966453326e-06, "loss": 0.0069, "step": 255590 }, { "epoch": 162.90630975143404, "grad_norm": 0.07822076231241226, "learning_rate": 5.870141545679042e-06, "loss": 0.0027, "step": 255600 }, { "epoch": 162.91268323773104, "grad_norm": 0.06513214856386185, "learning_rate": 5.867550661124188e-06, "loss": 0.0251, "step": 255610 }, { "epoch": 162.91905672402805, "grad_norm": 0.1588730663061142, "learning_rate": 5.8649603128202775e-06, "loss": 0.0095, "step": 255620 }, { "epoch": 162.92543021032506, "grad_norm": 0.5350372195243835, "learning_rate": 5.862370500798769e-06, "loss": 0.0121, "step": 255630 }, { "epoch": 162.93180369662204, "grad_norm": 0.11818215250968933, "learning_rate": 5.8597812250911265e-06, "loss": 0.0129, "step": 255640 }, { "epoch": 162.93817718291905, "grad_norm": 0.16536644101142883, "learning_rate": 5.857192485728813e-06, "loss": 0.0082, "step": 255650 }, { "epoch": 162.94455066921606, "grad_norm": 0.10287962853908539, "learning_rate": 5.8546042827432976e-06, "loss": 0.003, "step": 255660 }, { "epoch": 162.95092415551306, "grad_norm": 0.28882506489753723, "learning_rate": 5.852016616166017e-06, "loss": 0.0077, "step": 255670 }, { "epoch": 162.95729764181007, "grad_norm": 0.3826211392879486, "learning_rate": 5.849429486028418e-06, "loss": 0.0056, "step": 255680 }, { "epoch": 162.96367112810708, "grad_norm": 0.9047539830207825, "learning_rate": 5.846842892361931e-06, "loss": 0.0059, "step": 255690 }, { "epoch": 162.9700446144041, "grad_norm": 0.07459545880556107, "learning_rate": 5.844256835198003e-06, "loss": 0.0026, "step": 255700 }, { "epoch": 162.9764181007011, "grad_norm": 0.25320181250572205, "learning_rate": 5.8416713145680316e-06, "loss": 0.0044, "step": 255710 }, { "epoch": 162.98279158699808, "grad_norm": 0.15125368535518646, "learning_rate": 5.839086330503452e-06, "loss": 0.0029, "step": 255720 }, { "epoch": 162.98916507329508, "grad_norm": 0.6088195443153381, "learning_rate": 5.836501883035655e-06, "loss": 0.0101, "step": 255730 }, { "epoch": 162.9955385595921, "grad_norm": 0.3022219240665436, "learning_rate": 5.8339179721960845e-06, "loss": 0.0051, "step": 255740 }, { "epoch": 163.0019120458891, "grad_norm": 0.2332170307636261, "learning_rate": 5.831334598016086e-06, "loss": 0.0031, "step": 255750 }, { "epoch": 163.0082855321861, "grad_norm": 0.1965477019548416, "learning_rate": 5.828751760527084e-06, "loss": 0.0043, "step": 255760 }, { "epoch": 163.01465901848312, "grad_norm": 0.04447160288691521, "learning_rate": 5.826169459760456e-06, "loss": 0.0083, "step": 255770 }, { "epoch": 163.02103250478012, "grad_norm": 0.19541749358177185, "learning_rate": 5.823587695747573e-06, "loss": 0.0078, "step": 255780 }, { "epoch": 163.02740599107713, "grad_norm": 0.20130208134651184, "learning_rate": 5.821006468519796e-06, "loss": 0.0114, "step": 255790 }, { "epoch": 163.0337794773741, "grad_norm": 0.09088970720767975, "learning_rate": 5.818425778108516e-06, "loss": 0.0061, "step": 255800 }, { "epoch": 163.04015296367112, "grad_norm": 0.26332294940948486, "learning_rate": 5.81584562454508e-06, "loss": 0.0103, "step": 255810 }, { "epoch": 163.04652644996813, "grad_norm": 0.31411829590797424, "learning_rate": 5.8132660078608344e-06, "loss": 0.0067, "step": 255820 }, { "epoch": 163.05289993626513, "grad_norm": 0.3185410797595978, "learning_rate": 5.8106869280871115e-06, "loss": 0.0111, "step": 255830 }, { "epoch": 163.05927342256214, "grad_norm": 0.3001696467399597, "learning_rate": 5.8081083852552785e-06, "loss": 0.0068, "step": 255840 }, { "epoch": 163.06564690885915, "grad_norm": 0.12530452013015747, "learning_rate": 5.805530379396656e-06, "loss": 0.0032, "step": 255850 }, { "epoch": 163.07202039515616, "grad_norm": 0.1067061796784401, "learning_rate": 5.802952910542558e-06, "loss": 0.0093, "step": 255860 }, { "epoch": 163.07839388145317, "grad_norm": 0.2559948265552521, "learning_rate": 5.800375978724304e-06, "loss": 0.0045, "step": 255870 }, { "epoch": 163.08476736775015, "grad_norm": 0.12778380513191223, "learning_rate": 5.7977995839732295e-06, "loss": 0.008, "step": 255880 }, { "epoch": 163.09114085404715, "grad_norm": 0.15512946248054504, "learning_rate": 5.7952237263206146e-06, "loss": 0.0068, "step": 255890 }, { "epoch": 163.09751434034416, "grad_norm": 0.04337771609425545, "learning_rate": 5.792648405797773e-06, "loss": 0.0054, "step": 255900 }, { "epoch": 163.10388782664117, "grad_norm": 0.16043992340564728, "learning_rate": 5.790073622435987e-06, "loss": 0.0062, "step": 255910 }, { "epoch": 163.11026131293818, "grad_norm": 0.172862708568573, "learning_rate": 5.787499376266542e-06, "loss": 0.0038, "step": 255920 }, { "epoch": 163.1166347992352, "grad_norm": 0.06584963202476501, "learning_rate": 5.7849256673207344e-06, "loss": 0.0042, "step": 255930 }, { "epoch": 163.1230082855322, "grad_norm": 0.3083203434944153, "learning_rate": 5.782352495629822e-06, "loss": 0.0167, "step": 255940 }, { "epoch": 163.1293817718292, "grad_norm": 0.06963543593883514, "learning_rate": 5.779779861225077e-06, "loss": 0.0036, "step": 255950 }, { "epoch": 163.13575525812618, "grad_norm": 0.16217046976089478, "learning_rate": 5.777207764137749e-06, "loss": 0.0064, "step": 255960 }, { "epoch": 163.1421287444232, "grad_norm": 0.1956193447113037, "learning_rate": 5.7746362043991086e-06, "loss": 0.003, "step": 255970 }, { "epoch": 163.1485022307202, "grad_norm": 0.07985647767782211, "learning_rate": 5.772065182040393e-06, "loss": 0.0085, "step": 255980 }, { "epoch": 163.1548757170172, "grad_norm": 0.2088281512260437, "learning_rate": 5.769494697092842e-06, "loss": 0.0033, "step": 255990 }, { "epoch": 163.1612492033142, "grad_norm": 0.17611318826675415, "learning_rate": 5.766924749587682e-06, "loss": 0.0041, "step": 256000 }, { "epoch": 163.16762268961122, "grad_norm": 0.2214740812778473, "learning_rate": 5.76435533955616e-06, "loss": 0.0059, "step": 256010 }, { "epoch": 163.17399617590823, "grad_norm": 0.1725473701953888, "learning_rate": 5.76178646702949e-06, "loss": 0.005, "step": 256020 }, { "epoch": 163.18036966220524, "grad_norm": 0.24451856315135956, "learning_rate": 5.75921813203888e-06, "loss": 0.0038, "step": 256030 }, { "epoch": 163.18674314850222, "grad_norm": 0.5079136490821838, "learning_rate": 5.756650334615527e-06, "loss": 0.0103, "step": 256040 }, { "epoch": 163.19311663479922, "grad_norm": 0.16190728545188904, "learning_rate": 5.754083074790656e-06, "loss": 0.0051, "step": 256050 }, { "epoch": 163.19949012109623, "grad_norm": 0.11383629590272903, "learning_rate": 5.751516352595449e-06, "loss": 0.0026, "step": 256060 }, { "epoch": 163.20586360739324, "grad_norm": 0.08825641870498657, "learning_rate": 5.7489501680611016e-06, "loss": 0.007, "step": 256070 }, { "epoch": 163.21223709369025, "grad_norm": 0.1739524006843567, "learning_rate": 5.746384521218773e-06, "loss": 0.0037, "step": 256080 }, { "epoch": 163.21861057998726, "grad_norm": 0.17169907689094543, "learning_rate": 5.743819412099677e-06, "loss": 0.0034, "step": 256090 }, { "epoch": 163.22498406628426, "grad_norm": 0.06427735835313797, "learning_rate": 5.741254840734944e-06, "loss": 0.0098, "step": 256100 }, { "epoch": 163.23135755258127, "grad_norm": 0.11941289901733398, "learning_rate": 5.738690807155756e-06, "loss": 0.0058, "step": 256110 }, { "epoch": 163.23773103887828, "grad_norm": 0.1690865010023117, "learning_rate": 5.736127311393258e-06, "loss": 0.0063, "step": 256120 }, { "epoch": 163.24410452517526, "grad_norm": 0.2037336826324463, "learning_rate": 5.733564353478621e-06, "loss": 0.0063, "step": 256130 }, { "epoch": 163.25047801147227, "grad_norm": 0.44766923785209656, "learning_rate": 5.731001933442958e-06, "loss": 0.0059, "step": 256140 }, { "epoch": 163.25685149776928, "grad_norm": 0.33992063999176025, "learning_rate": 5.7284400513174245e-06, "loss": 0.0043, "step": 256150 }, { "epoch": 163.26322498406628, "grad_norm": 0.559888482093811, "learning_rate": 5.725878707133142e-06, "loss": 0.0087, "step": 256160 }, { "epoch": 163.2695984703633, "grad_norm": 0.3834857642650604, "learning_rate": 5.723317900921238e-06, "loss": 0.0033, "step": 256170 }, { "epoch": 163.2759719566603, "grad_norm": 0.11655734479427338, "learning_rate": 5.720757632712815e-06, "loss": 0.0062, "step": 256180 }, { "epoch": 163.2823454429573, "grad_norm": 0.622861921787262, "learning_rate": 5.718197902539002e-06, "loss": 0.0089, "step": 256190 }, { "epoch": 163.28871892925432, "grad_norm": 0.17041702568531036, "learning_rate": 5.715638710430892e-06, "loss": 0.0123, "step": 256200 }, { "epoch": 163.2950924155513, "grad_norm": 0.49818599224090576, "learning_rate": 5.71308005641959e-06, "loss": 0.014, "step": 256210 }, { "epoch": 163.3014659018483, "grad_norm": 0.10108380764722824, "learning_rate": 5.710521940536162e-06, "loss": 0.004, "step": 256220 }, { "epoch": 163.3078393881453, "grad_norm": 0.14596539735794067, "learning_rate": 5.707964362811735e-06, "loss": 0.0082, "step": 256230 }, { "epoch": 163.31421287444232, "grad_norm": 0.4607737362384796, "learning_rate": 5.705407323277334e-06, "loss": 0.0069, "step": 256240 }, { "epoch": 163.32058636073933, "grad_norm": 0.12185340374708176, "learning_rate": 5.702850821964068e-06, "loss": 0.003, "step": 256250 }, { "epoch": 163.32695984703633, "grad_norm": 0.2669285535812378, "learning_rate": 5.700294858902977e-06, "loss": 0.0067, "step": 256260 }, { "epoch": 163.33333333333334, "grad_norm": 0.16223564743995667, "learning_rate": 5.6977394341251485e-06, "loss": 0.0055, "step": 256270 }, { "epoch": 163.33970681963035, "grad_norm": 0.12329334020614624, "learning_rate": 5.695184547661592e-06, "loss": 0.0043, "step": 256280 }, { "epoch": 163.34608030592733, "grad_norm": 0.18224240839481354, "learning_rate": 5.692630199543386e-06, "loss": 0.0126, "step": 256290 }, { "epoch": 163.35245379222434, "grad_norm": 0.48968708515167236, "learning_rate": 5.690076389801546e-06, "loss": 0.0085, "step": 256300 }, { "epoch": 163.35882727852135, "grad_norm": 0.2431783527135849, "learning_rate": 5.6875231184671365e-06, "loss": 0.0048, "step": 256310 }, { "epoch": 163.36520076481835, "grad_norm": 0.2412695586681366, "learning_rate": 5.684970385571137e-06, "loss": 0.0034, "step": 256320 }, { "epoch": 163.37157425111536, "grad_norm": 0.07747001200914383, "learning_rate": 5.682418191144595e-06, "loss": 0.0207, "step": 256330 }, { "epoch": 163.37794773741237, "grad_norm": 0.40710878372192383, "learning_rate": 5.67986653521852e-06, "loss": 0.0052, "step": 256340 }, { "epoch": 163.38432122370938, "grad_norm": 0.09530401974916458, "learning_rate": 5.677315417823908e-06, "loss": 0.0044, "step": 256350 }, { "epoch": 163.3906947100064, "grad_norm": 0.4410642087459564, "learning_rate": 5.674764838991753e-06, "loss": 0.0102, "step": 256360 }, { "epoch": 163.39706819630337, "grad_norm": 0.30746424198150635, "learning_rate": 5.672214798753062e-06, "loss": 0.0065, "step": 256370 }, { "epoch": 163.40344168260037, "grad_norm": 0.12362642586231232, "learning_rate": 5.669665297138815e-06, "loss": 0.0045, "step": 256380 }, { "epoch": 163.40981516889738, "grad_norm": 0.06522151082754135, "learning_rate": 5.667116334179989e-06, "loss": 0.0065, "step": 256390 }, { "epoch": 163.4161886551944, "grad_norm": 0.1028171181678772, "learning_rate": 5.664567909907548e-06, "loss": 0.0064, "step": 256400 }, { "epoch": 163.4225621414914, "grad_norm": 0.32563987374305725, "learning_rate": 5.662020024352488e-06, "loss": 0.0087, "step": 256410 }, { "epoch": 163.4289356277884, "grad_norm": 0.10003465414047241, "learning_rate": 5.659472677545724e-06, "loss": 0.0031, "step": 256420 }, { "epoch": 163.4353091140854, "grad_norm": 0.09569908678531647, "learning_rate": 5.656925869518242e-06, "loss": 0.0061, "step": 256430 }, { "epoch": 163.44168260038242, "grad_norm": 0.14771334826946259, "learning_rate": 5.654379600300969e-06, "loss": 0.0091, "step": 256440 }, { "epoch": 163.4480560866794, "grad_norm": 0.1812470406293869, "learning_rate": 5.651833869924872e-06, "loss": 0.0029, "step": 256450 }, { "epoch": 163.4544295729764, "grad_norm": 0.6104326248168945, "learning_rate": 5.649288678420844e-06, "loss": 0.0092, "step": 256460 }, { "epoch": 163.46080305927342, "grad_norm": 0.20904602110385895, "learning_rate": 5.6467440258198445e-06, "loss": 0.004, "step": 256470 }, { "epoch": 163.46717654557042, "grad_norm": 0.18168988823890686, "learning_rate": 5.644199912152781e-06, "loss": 0.0034, "step": 256480 }, { "epoch": 163.47355003186743, "grad_norm": 0.08998741954565048, "learning_rate": 5.641656337450568e-06, "loss": 0.016, "step": 256490 }, { "epoch": 163.47992351816444, "grad_norm": 0.09619498997926712, "learning_rate": 5.639113301744098e-06, "loss": 0.0049, "step": 256500 }, { "epoch": 163.48629700446145, "grad_norm": 0.17140470445156097, "learning_rate": 5.636570805064301e-06, "loss": 0.0029, "step": 256510 }, { "epoch": 163.49267049075846, "grad_norm": 0.18277421593666077, "learning_rate": 5.634028847442047e-06, "loss": 0.0057, "step": 256520 }, { "epoch": 163.49904397705544, "grad_norm": 0.06734201312065125, "learning_rate": 5.631487428908239e-06, "loss": 0.0035, "step": 256530 }, { "epoch": 163.50541746335244, "grad_norm": 0.1728908121585846, "learning_rate": 5.628946549493735e-06, "loss": 0.0064, "step": 256540 }, { "epoch": 163.51179094964945, "grad_norm": 0.1962299942970276, "learning_rate": 5.626406209229434e-06, "loss": 0.0054, "step": 256550 }, { "epoch": 163.51816443594646, "grad_norm": 0.08031773567199707, "learning_rate": 5.623866408146194e-06, "loss": 0.0048, "step": 256560 }, { "epoch": 163.52453792224347, "grad_norm": 0.32563936710357666, "learning_rate": 5.621327146274874e-06, "loss": 0.0112, "step": 256570 }, { "epoch": 163.53091140854048, "grad_norm": 0.08865980803966522, "learning_rate": 5.61878842364632e-06, "loss": 0.0091, "step": 256580 }, { "epoch": 163.53728489483748, "grad_norm": 0.21349720656871796, "learning_rate": 5.616250240291399e-06, "loss": 0.0052, "step": 256590 }, { "epoch": 163.5436583811345, "grad_norm": 0.4223962128162384, "learning_rate": 5.613712596240944e-06, "loss": 0.0039, "step": 256600 }, { "epoch": 163.55003186743147, "grad_norm": 0.094746433198452, "learning_rate": 5.611175491525789e-06, "loss": 0.0026, "step": 256610 }, { "epoch": 163.55640535372848, "grad_norm": 0.33214956521987915, "learning_rate": 5.608638926176751e-06, "loss": 0.0042, "step": 256620 }, { "epoch": 163.5627788400255, "grad_norm": 0.11556455492973328, "learning_rate": 5.606102900224674e-06, "loss": 0.0059, "step": 256630 }, { "epoch": 163.5691523263225, "grad_norm": 0.09814174473285675, "learning_rate": 5.603567413700361e-06, "loss": 0.0032, "step": 256640 }, { "epoch": 163.5755258126195, "grad_norm": 0.2619894742965698, "learning_rate": 5.60103246663462e-06, "loss": 0.0054, "step": 256650 }, { "epoch": 163.5818992989165, "grad_norm": 0.22274108231067657, "learning_rate": 5.598498059058254e-06, "loss": 0.0037, "step": 256660 }, { "epoch": 163.58827278521352, "grad_norm": 0.20424790680408478, "learning_rate": 5.595964191002045e-06, "loss": 0.0082, "step": 256670 }, { "epoch": 163.59464627151053, "grad_norm": 0.15035083889961243, "learning_rate": 5.593430862496812e-06, "loss": 0.0067, "step": 256680 }, { "epoch": 163.60101975780753, "grad_norm": 0.1262453943490982, "learning_rate": 5.5908980735733196e-06, "loss": 0.0094, "step": 256690 }, { "epoch": 163.60739324410451, "grad_norm": 0.35680997371673584, "learning_rate": 5.5883658242623425e-06, "loss": 0.0121, "step": 256700 }, { "epoch": 163.61376673040152, "grad_norm": 0.19259117543697357, "learning_rate": 5.58583411459464e-06, "loss": 0.0043, "step": 256710 }, { "epoch": 163.62014021669853, "grad_norm": 0.12677693367004395, "learning_rate": 5.583302944600999e-06, "loss": 0.0055, "step": 256720 }, { "epoch": 163.62651370299554, "grad_norm": 0.4599088132381439, "learning_rate": 5.580772314312166e-06, "loss": 0.0108, "step": 256730 }, { "epoch": 163.63288718929255, "grad_norm": 0.09647348523139954, "learning_rate": 5.578242223758884e-06, "loss": 0.0079, "step": 256740 }, { "epoch": 163.63926067558955, "grad_norm": 0.4666845500469208, "learning_rate": 5.575712672971894e-06, "loss": 0.0054, "step": 256750 }, { "epoch": 163.64563416188656, "grad_norm": 0.08775439113378525, "learning_rate": 5.573183661981945e-06, "loss": 0.0041, "step": 256760 }, { "epoch": 163.65200764818357, "grad_norm": 0.18506452441215515, "learning_rate": 5.570655190819762e-06, "loss": 0.0054, "step": 256770 }, { "epoch": 163.65838113448055, "grad_norm": 0.2907041609287262, "learning_rate": 5.568127259516071e-06, "loss": 0.0078, "step": 256780 }, { "epoch": 163.66475462077756, "grad_norm": 0.5497888922691345, "learning_rate": 5.565599868101567e-06, "loss": 0.0083, "step": 256790 }, { "epoch": 163.67112810707457, "grad_norm": 0.16862691938877106, "learning_rate": 5.563073016607001e-06, "loss": 0.0079, "step": 256800 }, { "epoch": 163.67750159337157, "grad_norm": 0.11977215856313705, "learning_rate": 5.560546705063035e-06, "loss": 0.0082, "step": 256810 }, { "epoch": 163.68387507966858, "grad_norm": 0.19021689891815186, "learning_rate": 5.558020933500391e-06, "loss": 0.0095, "step": 256820 }, { "epoch": 163.6902485659656, "grad_norm": 0.14609019458293915, "learning_rate": 5.5554957019497475e-06, "loss": 0.0082, "step": 256830 }, { "epoch": 163.6966220522626, "grad_norm": 0.16185219585895538, "learning_rate": 5.5529710104418095e-06, "loss": 0.0046, "step": 256840 }, { "epoch": 163.7029955385596, "grad_norm": 0.3758481442928314, "learning_rate": 5.550446859007219e-06, "loss": 0.0053, "step": 256850 }, { "epoch": 163.70936902485658, "grad_norm": 0.10248728096485138, "learning_rate": 5.5479232476766785e-06, "loss": 0.0107, "step": 256860 }, { "epoch": 163.7157425111536, "grad_norm": 0.10536161810159683, "learning_rate": 5.54540017648083e-06, "loss": 0.0028, "step": 256870 }, { "epoch": 163.7221159974506, "grad_norm": 0.44218504428863525, "learning_rate": 5.5428776454503615e-06, "loss": 0.0062, "step": 256880 }, { "epoch": 163.7284894837476, "grad_norm": 0.06183084100484848, "learning_rate": 5.540355654615881e-06, "loss": 0.0033, "step": 256890 }, { "epoch": 163.73486297004462, "grad_norm": 0.17251522839069366, "learning_rate": 5.5378342040080686e-06, "loss": 0.0038, "step": 256900 }, { "epoch": 163.74123645634162, "grad_norm": 0.2002640813589096, "learning_rate": 5.535313293657552e-06, "loss": 0.0057, "step": 256910 }, { "epoch": 163.74760994263863, "grad_norm": 0.15268005430698395, "learning_rate": 5.532792923594954e-06, "loss": 0.0046, "step": 256920 }, { "epoch": 163.75398342893564, "grad_norm": 0.15917642414569855, "learning_rate": 5.530273093850902e-06, "loss": 0.0078, "step": 256930 }, { "epoch": 163.76035691523262, "grad_norm": 0.12408258765935898, "learning_rate": 5.527753804456026e-06, "loss": 0.0178, "step": 256940 }, { "epoch": 163.76673040152963, "grad_norm": 0.43169692158699036, "learning_rate": 5.5252350554409305e-06, "loss": 0.0063, "step": 256950 }, { "epoch": 163.77310388782664, "grad_norm": 0.04901335388422012, "learning_rate": 5.522716846836218e-06, "loss": 0.0089, "step": 256960 }, { "epoch": 163.77947737412364, "grad_norm": 0.22034524381160736, "learning_rate": 5.520199178672486e-06, "loss": 0.0056, "step": 256970 }, { "epoch": 163.78585086042065, "grad_norm": 0.14038220047950745, "learning_rate": 5.517682050980344e-06, "loss": 0.0088, "step": 256980 }, { "epoch": 163.79222434671766, "grad_norm": 0.25378137826919556, "learning_rate": 5.515165463790345e-06, "loss": 0.0037, "step": 256990 }, { "epoch": 163.79859783301467, "grad_norm": 0.0691761001944542, "learning_rate": 5.512649417133092e-06, "loss": 0.0077, "step": 257000 }, { "epoch": 163.80497131931168, "grad_norm": 0.3600837290287018, "learning_rate": 5.5101339110391506e-06, "loss": 0.0033, "step": 257010 }, { "epoch": 163.81134480560866, "grad_norm": 0.17920362949371338, "learning_rate": 5.5076189455391e-06, "loss": 0.0083, "step": 257020 }, { "epoch": 163.81771829190566, "grad_norm": 0.3474641740322113, "learning_rate": 5.505104520663468e-06, "loss": 0.0047, "step": 257030 }, { "epoch": 163.82409177820267, "grad_norm": 0.10171104222536087, "learning_rate": 5.502590636442839e-06, "loss": 0.009, "step": 257040 }, { "epoch": 163.83046526449968, "grad_norm": 0.16544021666049957, "learning_rate": 5.500077292907746e-06, "loss": 0.0063, "step": 257050 }, { "epoch": 163.8368387507967, "grad_norm": 0.048972442746162415, "learning_rate": 5.497564490088724e-06, "loss": 0.0106, "step": 257060 }, { "epoch": 163.8432122370937, "grad_norm": 0.11688055843114853, "learning_rate": 5.495052228016306e-06, "loss": 0.0047, "step": 257070 }, { "epoch": 163.8495857233907, "grad_norm": 0.5054823160171509, "learning_rate": 5.492540506721033e-06, "loss": 0.0115, "step": 257080 }, { "epoch": 163.8559592096877, "grad_norm": 0.08670678734779358, "learning_rate": 5.490029326233415e-06, "loss": 0.0061, "step": 257090 }, { "epoch": 163.8623326959847, "grad_norm": 0.1853478103876114, "learning_rate": 5.487518686583959e-06, "loss": 0.0038, "step": 257100 }, { "epoch": 163.8687061822817, "grad_norm": 0.4639529883861542, "learning_rate": 5.485008587803175e-06, "loss": 0.007, "step": 257110 }, { "epoch": 163.8750796685787, "grad_norm": 0.1613103598356247, "learning_rate": 5.482499029921584e-06, "loss": 0.0046, "step": 257120 }, { "epoch": 163.88145315487571, "grad_norm": 0.08303690701723099, "learning_rate": 5.479990012969638e-06, "loss": 0.0032, "step": 257130 }, { "epoch": 163.88782664117272, "grad_norm": 0.1778535693883896, "learning_rate": 5.477481536977858e-06, "loss": 0.0044, "step": 257140 }, { "epoch": 163.89420012746973, "grad_norm": 0.15204182267189026, "learning_rate": 5.474973601976707e-06, "loss": 0.0068, "step": 257150 }, { "epoch": 163.90057361376674, "grad_norm": 0.3371993601322174, "learning_rate": 5.472466207996685e-06, "loss": 0.0087, "step": 257160 }, { "epoch": 163.90694710006375, "grad_norm": 0.20287320017814636, "learning_rate": 5.469959355068216e-06, "loss": 0.0082, "step": 257170 }, { "epoch": 163.91332058636073, "grad_norm": 0.14586733281612396, "learning_rate": 5.467453043221799e-06, "loss": 0.0034, "step": 257180 }, { "epoch": 163.91969407265773, "grad_norm": 0.2277025431394577, "learning_rate": 5.464947272487858e-06, "loss": 0.0047, "step": 257190 }, { "epoch": 163.92606755895474, "grad_norm": 0.3015550374984741, "learning_rate": 5.462442042896876e-06, "loss": 0.0139, "step": 257200 }, { "epoch": 163.93244104525175, "grad_norm": 0.0801985040307045, "learning_rate": 5.459937354479255e-06, "loss": 0.0055, "step": 257210 }, { "epoch": 163.93881453154876, "grad_norm": 0.3735468089580536, "learning_rate": 5.457433207265461e-06, "loss": 0.0062, "step": 257220 }, { "epoch": 163.94518801784577, "grad_norm": 0.06923852860927582, "learning_rate": 5.454929601285908e-06, "loss": 0.0033, "step": 257230 }, { "epoch": 163.95156150414277, "grad_norm": 0.06086542084813118, "learning_rate": 5.452426536571015e-06, "loss": 0.0053, "step": 257240 }, { "epoch": 163.95793499043978, "grad_norm": 0.11003819853067398, "learning_rate": 5.449924013151187e-06, "loss": 0.0048, "step": 257250 }, { "epoch": 163.96430847673676, "grad_norm": 0.18505342304706573, "learning_rate": 5.447422031056859e-06, "loss": 0.0051, "step": 257260 }, { "epoch": 163.97068196303377, "grad_norm": 0.16658075153827667, "learning_rate": 5.44492059031842e-06, "loss": 0.0042, "step": 257270 }, { "epoch": 163.97705544933078, "grad_norm": 0.26365453004837036, "learning_rate": 5.442419690966255e-06, "loss": 0.0071, "step": 257280 }, { "epoch": 163.98342893562778, "grad_norm": 0.1697252094745636, "learning_rate": 5.439919333030757e-06, "loss": 0.0053, "step": 257290 }, { "epoch": 163.9898024219248, "grad_norm": 0.23287558555603027, "learning_rate": 5.4374195165423195e-06, "loss": 0.0031, "step": 257300 }, { "epoch": 163.9961759082218, "grad_norm": 0.10095398128032684, "learning_rate": 5.434920241531305e-06, "loss": 0.0076, "step": 257310 }, { "epoch": 164.0025493945188, "grad_norm": 0.16041290760040283, "learning_rate": 5.4324215080280905e-06, "loss": 0.0207, "step": 257320 }, { "epoch": 164.00892288081582, "grad_norm": 0.13025282323360443, "learning_rate": 5.429923316063023e-06, "loss": 0.0057, "step": 257330 }, { "epoch": 164.01529636711282, "grad_norm": 0.14998167753219604, "learning_rate": 5.4274256656664745e-06, "loss": 0.0064, "step": 257340 }, { "epoch": 164.0216698534098, "grad_norm": 0.18745973706245422, "learning_rate": 5.424928556868792e-06, "loss": 0.0039, "step": 257350 }, { "epoch": 164.0280433397068, "grad_norm": 0.22742147743701935, "learning_rate": 5.422431989700311e-06, "loss": 0.0066, "step": 257360 }, { "epoch": 164.03441682600382, "grad_norm": 0.08674534410238266, "learning_rate": 5.419935964191369e-06, "loss": 0.0024, "step": 257370 }, { "epoch": 164.04079031230083, "grad_norm": 0.12278445810079575, "learning_rate": 5.417440480372287e-06, "loss": 0.0031, "step": 257380 }, { "epoch": 164.04716379859784, "grad_norm": 0.16921620070934296, "learning_rate": 5.414945538273408e-06, "loss": 0.0033, "step": 257390 }, { "epoch": 164.05353728489484, "grad_norm": 0.25096291303634644, "learning_rate": 5.412451137925034e-06, "loss": 0.009, "step": 257400 }, { "epoch": 164.05991077119185, "grad_norm": 0.12529897689819336, "learning_rate": 5.409957279357475e-06, "loss": 0.0089, "step": 257410 }, { "epoch": 164.06628425748886, "grad_norm": 0.4850747585296631, "learning_rate": 5.407463962601028e-06, "loss": 0.0052, "step": 257420 }, { "epoch": 164.07265774378584, "grad_norm": 0.2715785801410675, "learning_rate": 5.404971187686003e-06, "loss": 0.0041, "step": 257430 }, { "epoch": 164.07903123008285, "grad_norm": 0.3925601840019226, "learning_rate": 5.402478954642687e-06, "loss": 0.0049, "step": 257440 }, { "epoch": 164.08540471637986, "grad_norm": 0.21780367195606232, "learning_rate": 5.399987263501355e-06, "loss": 0.0101, "step": 257450 }, { "epoch": 164.09177820267686, "grad_norm": 0.12038128823041916, "learning_rate": 5.397496114292278e-06, "loss": 0.0041, "step": 257460 }, { "epoch": 164.09815168897387, "grad_norm": 0.17592346668243408, "learning_rate": 5.395005507045742e-06, "loss": 0.0031, "step": 257470 }, { "epoch": 164.10452517527088, "grad_norm": 0.09981731325387955, "learning_rate": 5.392515441792006e-06, "loss": 0.0053, "step": 257480 }, { "epoch": 164.1108986615679, "grad_norm": 0.34536099433898926, "learning_rate": 5.390025918561326e-06, "loss": 0.0077, "step": 257490 }, { "epoch": 164.1172721478649, "grad_norm": 0.17588600516319275, "learning_rate": 5.387536937383936e-06, "loss": 0.0044, "step": 257500 }, { "epoch": 164.12364563416187, "grad_norm": 0.0992380753159523, "learning_rate": 5.385048498290108e-06, "loss": 0.0057, "step": 257510 }, { "epoch": 164.13001912045888, "grad_norm": 0.34837624430656433, "learning_rate": 5.38256060131006e-06, "loss": 0.0092, "step": 257520 }, { "epoch": 164.1363926067559, "grad_norm": 0.1258373260498047, "learning_rate": 5.380073246474027e-06, "loss": 0.005, "step": 257530 }, { "epoch": 164.1427660930529, "grad_norm": 0.17035992443561554, "learning_rate": 5.377586433812226e-06, "loss": 0.0035, "step": 257540 }, { "epoch": 164.1491395793499, "grad_norm": 0.10864932835102081, "learning_rate": 5.375100163354896e-06, "loss": 0.0071, "step": 257550 }, { "epoch": 164.15551306564691, "grad_norm": 0.14312076568603516, "learning_rate": 5.372614435132212e-06, "loss": 0.004, "step": 257560 }, { "epoch": 164.16188655194392, "grad_norm": 0.4198536276817322, "learning_rate": 5.37012924917441e-06, "loss": 0.006, "step": 257570 }, { "epoch": 164.16826003824093, "grad_norm": 0.06713751703500748, "learning_rate": 5.367644605511668e-06, "loss": 0.006, "step": 257580 }, { "epoch": 164.1746335245379, "grad_norm": 0.4905976355075836, "learning_rate": 5.365160504174199e-06, "loss": 0.0058, "step": 257590 }, { "epoch": 164.18100701083492, "grad_norm": 0.1592530459165573, "learning_rate": 5.362676945192152e-06, "loss": 0.0044, "step": 257600 }, { "epoch": 164.18738049713193, "grad_norm": 0.07130371779203415, "learning_rate": 5.360193928595736e-06, "loss": 0.0032, "step": 257610 }, { "epoch": 164.19375398342893, "grad_norm": 0.05496751144528389, "learning_rate": 5.357711454415115e-06, "loss": 0.0025, "step": 257620 }, { "epoch": 164.20012746972594, "grad_norm": 0.27107590436935425, "learning_rate": 5.355229522680444e-06, "loss": 0.0193, "step": 257630 }, { "epoch": 164.20650095602295, "grad_norm": 0.31263378262519836, "learning_rate": 5.352748133421875e-06, "loss": 0.0051, "step": 257640 }, { "epoch": 164.21287444231996, "grad_norm": 0.20321378111839294, "learning_rate": 5.350267286669586e-06, "loss": 0.0035, "step": 257650 }, { "epoch": 164.21924792861697, "grad_norm": 0.3528723418712616, "learning_rate": 5.347786982453701e-06, "loss": 0.0046, "step": 257660 }, { "epoch": 164.22562141491395, "grad_norm": 0.14583466947078705, "learning_rate": 5.3453072208043635e-06, "loss": 0.0025, "step": 257670 }, { "epoch": 164.23199490121095, "grad_norm": 0.1807813048362732, "learning_rate": 5.342828001751693e-06, "loss": 0.0064, "step": 257680 }, { "epoch": 164.23836838750796, "grad_norm": 0.1830977499485016, "learning_rate": 5.34034932532585e-06, "loss": 0.006, "step": 257690 }, { "epoch": 164.24474187380497, "grad_norm": 0.5254756808280945, "learning_rate": 5.337871191556903e-06, "loss": 0.0071, "step": 257700 }, { "epoch": 164.25111536010198, "grad_norm": 0.4062182605266571, "learning_rate": 5.335393600475003e-06, "loss": 0.0076, "step": 257710 }, { "epoch": 164.25748884639899, "grad_norm": 0.14651893079280853, "learning_rate": 5.332916552110229e-06, "loss": 0.0023, "step": 257720 }, { "epoch": 164.263862332696, "grad_norm": 0.2339746356010437, "learning_rate": 5.330440046492713e-06, "loss": 0.0041, "step": 257730 }, { "epoch": 164.270235818993, "grad_norm": 0.2618647515773773, "learning_rate": 5.327964083652509e-06, "loss": 0.005, "step": 257740 }, { "epoch": 164.27660930528998, "grad_norm": 0.12393735349178314, "learning_rate": 5.325488663619721e-06, "loss": 0.0087, "step": 257750 }, { "epoch": 164.282982791587, "grad_norm": 0.3317524492740631, "learning_rate": 5.32301378642443e-06, "loss": 0.0048, "step": 257760 }, { "epoch": 164.289356277884, "grad_norm": 0.4738103747367859, "learning_rate": 5.320539452096706e-06, "loss": 0.0065, "step": 257770 }, { "epoch": 164.295729764181, "grad_norm": 0.33550044894218445, "learning_rate": 5.318065660666599e-06, "loss": 0.0123, "step": 257780 }, { "epoch": 164.302103250478, "grad_norm": 0.4002983570098877, "learning_rate": 5.3155924121641884e-06, "loss": 0.0065, "step": 257790 }, { "epoch": 164.30847673677502, "grad_norm": 0.1312311291694641, "learning_rate": 5.313119706619523e-06, "loss": 0.0041, "step": 257800 }, { "epoch": 164.31485022307203, "grad_norm": 0.1975189447402954, "learning_rate": 5.310647544062642e-06, "loss": 0.0054, "step": 257810 }, { "epoch": 164.32122370936904, "grad_norm": 0.15069054067134857, "learning_rate": 5.308175924523579e-06, "loss": 0.0058, "step": 257820 }, { "epoch": 164.32759719566602, "grad_norm": 0.33294060826301575, "learning_rate": 5.3057048480323834e-06, "loss": 0.0056, "step": 257830 }, { "epoch": 164.33397068196302, "grad_norm": 0.1115935891866684, "learning_rate": 5.303234314619071e-06, "loss": 0.0055, "step": 257840 }, { "epoch": 164.34034416826003, "grad_norm": 0.2685803472995758, "learning_rate": 5.300764324313667e-06, "loss": 0.0076, "step": 257850 }, { "epoch": 164.34671765455704, "grad_norm": 0.2443392425775528, "learning_rate": 5.2982948771461645e-06, "loss": 0.0053, "step": 257860 }, { "epoch": 164.35309114085405, "grad_norm": 0.22434885799884796, "learning_rate": 5.2958259731466064e-06, "loss": 0.0054, "step": 257870 }, { "epoch": 164.35946462715106, "grad_norm": 0.12854470312595367, "learning_rate": 5.293357612344952e-06, "loss": 0.0044, "step": 257880 }, { "epoch": 164.36583811344806, "grad_norm": 0.2567087709903717, "learning_rate": 5.290889794771226e-06, "loss": 0.0045, "step": 257890 }, { "epoch": 164.37221159974507, "grad_norm": 0.16640830039978027, "learning_rate": 5.288422520455383e-06, "loss": 0.0125, "step": 257900 }, { "epoch": 164.37858508604205, "grad_norm": 0.2132343202829361, "learning_rate": 5.285955789427444e-06, "loss": 0.0097, "step": 257910 }, { "epoch": 164.38495857233906, "grad_norm": 0.08255568146705627, "learning_rate": 5.283489601717346e-06, "loss": 0.0036, "step": 257920 }, { "epoch": 164.39133205863607, "grad_norm": 0.2730201184749603, "learning_rate": 5.281023957355069e-06, "loss": 0.0058, "step": 257930 }, { "epoch": 164.39770554493307, "grad_norm": 0.32493913173675537, "learning_rate": 5.278558856370575e-06, "loss": 0.0042, "step": 257940 }, { "epoch": 164.40407903123008, "grad_norm": 0.19372259080410004, "learning_rate": 5.276094298793821e-06, "loss": 0.006, "step": 257950 }, { "epoch": 164.4104525175271, "grad_norm": 0.20991574227809906, "learning_rate": 5.273630284654729e-06, "loss": 0.0036, "step": 257960 }, { "epoch": 164.4168260038241, "grad_norm": 0.45106807351112366, "learning_rate": 5.271166813983275e-06, "loss": 0.0047, "step": 257970 }, { "epoch": 164.4231994901211, "grad_norm": 0.5484171509742737, "learning_rate": 5.268703886809373e-06, "loss": 0.0056, "step": 257980 }, { "epoch": 164.42957297641811, "grad_norm": 0.10748438537120819, "learning_rate": 5.26624150316295e-06, "loss": 0.003, "step": 257990 }, { "epoch": 164.4359464627151, "grad_norm": 0.1593407541513443, "learning_rate": 5.26377966307392e-06, "loss": 0.0047, "step": 258000 }, { "epoch": 164.4423199490121, "grad_norm": 0.2877505421638489, "learning_rate": 5.2613183665722155e-06, "loss": 0.0046, "step": 258010 }, { "epoch": 164.4486934353091, "grad_norm": 0.09447938948869705, "learning_rate": 5.258857613687729e-06, "loss": 0.0075, "step": 258020 }, { "epoch": 164.45506692160612, "grad_norm": 0.21147392690181732, "learning_rate": 5.25639740445037e-06, "loss": 0.0036, "step": 258030 }, { "epoch": 164.46144040790313, "grad_norm": 0.26208052039146423, "learning_rate": 5.253937738890019e-06, "loss": 0.005, "step": 258040 }, { "epoch": 164.46781389420013, "grad_norm": 0.16547757387161255, "learning_rate": 5.251478617036576e-06, "loss": 0.0071, "step": 258050 }, { "epoch": 164.47418738049714, "grad_norm": 0.2860775887966156, "learning_rate": 5.249020038919922e-06, "loss": 0.0066, "step": 258060 }, { "epoch": 164.48056086679415, "grad_norm": 0.38475921750068665, "learning_rate": 5.2465620045699216e-06, "loss": 0.0056, "step": 258070 }, { "epoch": 164.48693435309113, "grad_norm": 0.28849804401397705, "learning_rate": 5.2441045140164405e-06, "loss": 0.0049, "step": 258080 }, { "epoch": 164.49330783938814, "grad_norm": 0.19801412522792816, "learning_rate": 5.241647567289359e-06, "loss": 0.0114, "step": 258090 }, { "epoch": 164.49968132568515, "grad_norm": 0.1381465643644333, "learning_rate": 5.239191164418517e-06, "loss": 0.0035, "step": 258100 }, { "epoch": 164.50605481198215, "grad_norm": 0.5193448662757874, "learning_rate": 5.2367353054337595e-06, "loss": 0.007, "step": 258110 }, { "epoch": 164.51242829827916, "grad_norm": 0.17851291596889496, "learning_rate": 5.234279990364932e-06, "loss": 0.0045, "step": 258120 }, { "epoch": 164.51880178457617, "grad_norm": 0.20498885214328766, "learning_rate": 5.231825219241865e-06, "loss": 0.0046, "step": 258130 }, { "epoch": 164.52517527087318, "grad_norm": 0.12259351462125778, "learning_rate": 5.22937099209439e-06, "loss": 0.0032, "step": 258140 }, { "epoch": 164.53154875717019, "grad_norm": 0.0746709555387497, "learning_rate": 5.226917308952334e-06, "loss": 0.0069, "step": 258150 }, { "epoch": 164.53792224346716, "grad_norm": 0.12881967425346375, "learning_rate": 5.224464169845505e-06, "loss": 0.0039, "step": 258160 }, { "epoch": 164.54429572976417, "grad_norm": 0.2131449431180954, "learning_rate": 5.222011574803704e-06, "loss": 0.004, "step": 258170 }, { "epoch": 164.55066921606118, "grad_norm": 0.27697667479515076, "learning_rate": 5.219559523856743e-06, "loss": 0.009, "step": 258180 }, { "epoch": 164.5570427023582, "grad_norm": 0.19307634234428406, "learning_rate": 5.217108017034417e-06, "loss": 0.0059, "step": 258190 }, { "epoch": 164.5634161886552, "grad_norm": 0.2675957977771759, "learning_rate": 5.214657054366512e-06, "loss": 0.011, "step": 258200 }, { "epoch": 164.5697896749522, "grad_norm": 0.12959063053131104, "learning_rate": 5.212206635882799e-06, "loss": 0.0025, "step": 258210 }, { "epoch": 164.5761631612492, "grad_norm": 0.2123902440071106, "learning_rate": 5.209756761613071e-06, "loss": 0.0027, "step": 258220 }, { "epoch": 164.58253664754622, "grad_norm": 0.2587210237979889, "learning_rate": 5.207307431587083e-06, "loss": 0.0036, "step": 258230 }, { "epoch": 164.5889101338432, "grad_norm": 0.14274492859840393, "learning_rate": 5.204858645834604e-06, "loss": 0.0094, "step": 258240 }, { "epoch": 164.5952836201402, "grad_norm": 0.10909013450145721, "learning_rate": 5.2024104043853775e-06, "loss": 0.0097, "step": 258250 }, { "epoch": 164.60165710643722, "grad_norm": 0.13375383615493774, "learning_rate": 5.199962707269179e-06, "loss": 0.0034, "step": 258260 }, { "epoch": 164.60803059273422, "grad_norm": 0.5079490542411804, "learning_rate": 5.1975155545157136e-06, "loss": 0.0036, "step": 258270 }, { "epoch": 164.61440407903123, "grad_norm": 0.4369748532772064, "learning_rate": 5.195068946154746e-06, "loss": 0.0057, "step": 258280 }, { "epoch": 164.62077756532824, "grad_norm": 0.3147982060909271, "learning_rate": 5.1926228822159805e-06, "loss": 0.0052, "step": 258290 }, { "epoch": 164.62715105162525, "grad_norm": 0.14050085842609406, "learning_rate": 5.1901773627291706e-06, "loss": 0.0025, "step": 258300 }, { "epoch": 164.63352453792226, "grad_norm": 0.09992018342018127, "learning_rate": 5.187732387723998e-06, "loss": 0.0052, "step": 258310 }, { "epoch": 164.63989802421924, "grad_norm": 0.06881888955831528, "learning_rate": 5.185287957230195e-06, "loss": 0.015, "step": 258320 }, { "epoch": 164.64627151051624, "grad_norm": 0.38064444065093994, "learning_rate": 5.182844071277454e-06, "loss": 0.006, "step": 258330 }, { "epoch": 164.65264499681325, "grad_norm": 0.0513157993555069, "learning_rate": 5.180400729895474e-06, "loss": 0.0081, "step": 258340 }, { "epoch": 164.65901848311026, "grad_norm": 0.20887400209903717, "learning_rate": 5.17795793311393e-06, "loss": 0.0054, "step": 258350 }, { "epoch": 164.66539196940727, "grad_norm": 0.1649530529975891, "learning_rate": 5.1755156809625225e-06, "loss": 0.005, "step": 258360 }, { "epoch": 164.67176545570427, "grad_norm": 0.14890244603157043, "learning_rate": 5.173073973470927e-06, "loss": 0.0044, "step": 258370 }, { "epoch": 164.67813894200128, "grad_norm": 0.09060004353523254, "learning_rate": 5.1706328106688014e-06, "loss": 0.0049, "step": 258380 }, { "epoch": 164.6845124282983, "grad_norm": 0.14658865332603455, "learning_rate": 5.1681921925858075e-06, "loss": 0.0056, "step": 258390 }, { "epoch": 164.69088591459527, "grad_norm": 0.35552486777305603, "learning_rate": 5.165752119251616e-06, "loss": 0.0092, "step": 258400 }, { "epoch": 164.69725940089228, "grad_norm": 0.26763367652893066, "learning_rate": 5.1633125906958675e-06, "loss": 0.0042, "step": 258410 }, { "epoch": 164.7036328871893, "grad_norm": 0.22783832252025604, "learning_rate": 5.160873606948202e-06, "loss": 0.0054, "step": 258420 }, { "epoch": 164.7100063734863, "grad_norm": 0.257852166891098, "learning_rate": 5.1584351680382445e-06, "loss": 0.0119, "step": 258430 }, { "epoch": 164.7163798597833, "grad_norm": 0.27967631816864014, "learning_rate": 5.155997273995661e-06, "loss": 0.0116, "step": 258440 }, { "epoch": 164.7227533460803, "grad_norm": 0.12919431924819946, "learning_rate": 5.153559924850032e-06, "loss": 0.0051, "step": 258450 }, { "epoch": 164.72912683237732, "grad_norm": 0.08538161963224411, "learning_rate": 5.151123120630996e-06, "loss": 0.0049, "step": 258460 }, { "epoch": 164.73550031867433, "grad_norm": 0.29841184616088867, "learning_rate": 5.148686861368157e-06, "loss": 0.0072, "step": 258470 }, { "epoch": 164.7418738049713, "grad_norm": 0.44507887959480286, "learning_rate": 5.14625114709113e-06, "loss": 0.013, "step": 258480 }, { "epoch": 164.7482472912683, "grad_norm": 0.40844810009002686, "learning_rate": 5.143815977829486e-06, "loss": 0.0075, "step": 258490 }, { "epoch": 164.75462077756532, "grad_norm": 0.14156070351600647, "learning_rate": 5.141381353612839e-06, "loss": 0.006, "step": 258500 }, { "epoch": 164.76099426386233, "grad_norm": 0.12742547690868378, "learning_rate": 5.1389472744707625e-06, "loss": 0.0068, "step": 258510 }, { "epoch": 164.76736775015934, "grad_norm": 0.17851294577121735, "learning_rate": 5.13651374043283e-06, "loss": 0.0036, "step": 258520 }, { "epoch": 164.77374123645635, "grad_norm": 0.16409580409526825, "learning_rate": 5.134080751528603e-06, "loss": 0.0049, "step": 258530 }, { "epoch": 164.78011472275335, "grad_norm": 0.07926616072654724, "learning_rate": 5.131648307787668e-06, "loss": 0.0054, "step": 258540 }, { "epoch": 164.78648820905036, "grad_norm": 0.24728703498840332, "learning_rate": 5.129216409239568e-06, "loss": 0.0144, "step": 258550 }, { "epoch": 164.79286169534734, "grad_norm": 0.255546897649765, "learning_rate": 5.126785055913852e-06, "loss": 0.013, "step": 258560 }, { "epoch": 164.79923518164435, "grad_norm": 0.7114822864532471, "learning_rate": 5.124354247840058e-06, "loss": 0.0077, "step": 258570 }, { "epoch": 164.80560866794136, "grad_norm": 0.25658702850341797, "learning_rate": 5.12192398504775e-06, "loss": 0.0048, "step": 258580 }, { "epoch": 164.81198215423836, "grad_norm": 0.2296696901321411, "learning_rate": 5.1194942675664155e-06, "loss": 0.0045, "step": 258590 }, { "epoch": 164.81835564053537, "grad_norm": 0.10291329771280289, "learning_rate": 5.11706509542561e-06, "loss": 0.0031, "step": 258600 }, { "epoch": 164.82472912683238, "grad_norm": 0.43340492248535156, "learning_rate": 5.114636468654827e-06, "loss": 0.0039, "step": 258610 }, { "epoch": 164.8311026131294, "grad_norm": 0.11198335140943527, "learning_rate": 5.112208387283612e-06, "loss": 0.0049, "step": 258620 }, { "epoch": 164.8374760994264, "grad_norm": 0.15171585977077484, "learning_rate": 5.109780851341428e-06, "loss": 0.0044, "step": 258630 }, { "epoch": 164.8438495857234, "grad_norm": 0.09301751106977463, "learning_rate": 5.1073538608578e-06, "loss": 0.0032, "step": 258640 }, { "epoch": 164.85022307202038, "grad_norm": 0.07626781612634659, "learning_rate": 5.104927415862209e-06, "loss": 0.0068, "step": 258650 }, { "epoch": 164.8565965583174, "grad_norm": 0.1759454309940338, "learning_rate": 5.1025015163841335e-06, "loss": 0.0046, "step": 258660 }, { "epoch": 164.8629700446144, "grad_norm": 0.11396678537130356, "learning_rate": 5.100076162453049e-06, "loss": 0.006, "step": 258670 }, { "epoch": 164.8693435309114, "grad_norm": 0.49184849858283997, "learning_rate": 5.097651354098437e-06, "loss": 0.0088, "step": 258680 }, { "epoch": 164.87571701720842, "grad_norm": 0.14406883716583252, "learning_rate": 5.095227091349764e-06, "loss": 0.0077, "step": 258690 }, { "epoch": 164.88209050350542, "grad_norm": 0.1359294205904007, "learning_rate": 5.092803374236471e-06, "loss": 0.0031, "step": 258700 }, { "epoch": 164.88846398980243, "grad_norm": 0.07635313272476196, "learning_rate": 5.090380202788009e-06, "loss": 0.0035, "step": 258710 }, { "epoch": 164.89483747609944, "grad_norm": 0.3039734661579132, "learning_rate": 5.0879575770338415e-06, "loss": 0.0076, "step": 258720 }, { "epoch": 164.90121096239642, "grad_norm": 0.25346001982688904, "learning_rate": 5.085535497003391e-06, "loss": 0.0038, "step": 258730 }, { "epoch": 164.90758444869343, "grad_norm": 0.10415636003017426, "learning_rate": 5.083113962726088e-06, "loss": 0.0126, "step": 258740 }, { "epoch": 164.91395793499044, "grad_norm": 0.3466108441352844, "learning_rate": 5.080692974231355e-06, "loss": 0.0078, "step": 258750 }, { "epoch": 164.92033142128744, "grad_norm": 0.2121533453464508, "learning_rate": 5.078272531548617e-06, "loss": 0.0049, "step": 258760 }, { "epoch": 164.92670490758445, "grad_norm": 0.09551689773797989, "learning_rate": 5.075852634707284e-06, "loss": 0.005, "step": 258770 }, { "epoch": 164.93307839388146, "grad_norm": 0.2961551547050476, "learning_rate": 5.073433283736761e-06, "loss": 0.0053, "step": 258780 }, { "epoch": 164.93945188017847, "grad_norm": 0.3823767304420471, "learning_rate": 5.071014478666425e-06, "loss": 0.0073, "step": 258790 }, { "epoch": 164.94582536647547, "grad_norm": 0.3832308351993561, "learning_rate": 5.068596219525695e-06, "loss": 0.0227, "step": 258800 }, { "epoch": 164.95219885277245, "grad_norm": 0.303985595703125, "learning_rate": 5.066178506343944e-06, "loss": 0.0063, "step": 258810 }, { "epoch": 164.95857233906946, "grad_norm": 0.31026729941368103, "learning_rate": 5.063761339150547e-06, "loss": 0.0038, "step": 258820 }, { "epoch": 164.96494582536647, "grad_norm": 0.055518440902233124, "learning_rate": 5.06134471797488e-06, "loss": 0.0035, "step": 258830 }, { "epoch": 164.97131931166348, "grad_norm": 0.14402522146701813, "learning_rate": 5.058928642846289e-06, "loss": 0.0058, "step": 258840 }, { "epoch": 164.9776927979605, "grad_norm": 0.37867170572280884, "learning_rate": 5.056513113794159e-06, "loss": 0.0035, "step": 258850 }, { "epoch": 164.9840662842575, "grad_norm": 0.24633602797985077, "learning_rate": 5.0540981308478285e-06, "loss": 0.006, "step": 258860 }, { "epoch": 164.9904397705545, "grad_norm": 0.07491035759449005, "learning_rate": 5.051683694036647e-06, "loss": 0.0039, "step": 258870 }, { "epoch": 164.9968132568515, "grad_norm": 0.5281118750572205, "learning_rate": 5.049269803389927e-06, "loss": 0.0044, "step": 258880 }, { "epoch": 165.0031867431485, "grad_norm": 0.23052845895290375, "learning_rate": 5.046856458937038e-06, "loss": 0.0036, "step": 258890 }, { "epoch": 165.0095602294455, "grad_norm": 0.16067972779273987, "learning_rate": 5.044443660707277e-06, "loss": 0.0051, "step": 258900 }, { "epoch": 165.0159337157425, "grad_norm": 0.2125469148159027, "learning_rate": 5.0420314087299744e-06, "loss": 0.0117, "step": 258910 }, { "epoch": 165.0223072020395, "grad_norm": 0.12371458858251572, "learning_rate": 5.039619703034432e-06, "loss": 0.0049, "step": 258920 }, { "epoch": 165.02868068833652, "grad_norm": 0.09778059273958206, "learning_rate": 5.037208543649968e-06, "loss": 0.0042, "step": 258930 }, { "epoch": 165.03505417463353, "grad_norm": 0.15747591853141785, "learning_rate": 5.034797930605867e-06, "loss": 0.0069, "step": 258940 }, { "epoch": 165.04142766093054, "grad_norm": 0.07511842995882034, "learning_rate": 5.032387863931431e-06, "loss": 0.0054, "step": 258950 }, { "epoch": 165.04780114722755, "grad_norm": 0.0884171798825264, "learning_rate": 5.029978343655928e-06, "loss": 0.0052, "step": 258960 }, { "epoch": 165.05417463352453, "grad_norm": 0.19976815581321716, "learning_rate": 5.02756936980866e-06, "loss": 0.0027, "step": 258970 }, { "epoch": 165.06054811982153, "grad_norm": 0.06934987753629684, "learning_rate": 5.025160942418872e-06, "loss": 0.0027, "step": 258980 }, { "epoch": 165.06692160611854, "grad_norm": 0.15732988715171814, "learning_rate": 5.02275306151585e-06, "loss": 0.0037, "step": 258990 }, { "epoch": 165.07329509241555, "grad_norm": 1.052070140838623, "learning_rate": 5.020345727128839e-06, "loss": 0.0072, "step": 259000 }, { "epoch": 165.07966857871256, "grad_norm": 0.08470853418111801, "learning_rate": 5.017938939287109e-06, "loss": 0.0057, "step": 259010 }, { "epoch": 165.08604206500956, "grad_norm": 0.22379666566848755, "learning_rate": 5.01553269801987e-06, "loss": 0.0036, "step": 259020 }, { "epoch": 165.09241555130657, "grad_norm": 0.10877621918916702, "learning_rate": 5.0131270033563935e-06, "loss": 0.0062, "step": 259030 }, { "epoch": 165.09878903760358, "grad_norm": 0.18026691675186157, "learning_rate": 5.0107218553258895e-06, "loss": 0.0053, "step": 259040 }, { "epoch": 165.10516252390056, "grad_norm": 0.1167224571108818, "learning_rate": 5.008317253957612e-06, "loss": 0.0125, "step": 259050 }, { "epoch": 165.11153601019757, "grad_norm": 0.07492522895336151, "learning_rate": 5.005913199280737e-06, "loss": 0.003, "step": 259060 }, { "epoch": 165.11790949649458, "grad_norm": 0.12110288441181183, "learning_rate": 5.0035096913245095e-06, "loss": 0.0058, "step": 259070 }, { "epoch": 165.12428298279158, "grad_norm": 0.37994134426116943, "learning_rate": 5.001106730118121e-06, "loss": 0.0085, "step": 259080 }, { "epoch": 165.1306564690886, "grad_norm": 0.12560951709747314, "learning_rate": 4.998704315690777e-06, "loss": 0.0059, "step": 259090 }, { "epoch": 165.1370299553856, "grad_norm": 0.11187414079904556, "learning_rate": 4.996302448071649e-06, "loss": 0.0043, "step": 259100 }, { "epoch": 165.1434034416826, "grad_norm": 0.08922051638364792, "learning_rate": 4.993901127289946e-06, "loss": 0.0105, "step": 259110 }, { "epoch": 165.14977692797962, "grad_norm": 0.13184234499931335, "learning_rate": 4.9915003533748395e-06, "loss": 0.007, "step": 259120 }, { "epoch": 165.1561504142766, "grad_norm": 0.24347500503063202, "learning_rate": 4.989100126355501e-06, "loss": 0.0044, "step": 259130 }, { "epoch": 165.1625239005736, "grad_norm": 0.05517997965216637, "learning_rate": 4.986700446261078e-06, "loss": 0.0062, "step": 259140 }, { "epoch": 165.1688973868706, "grad_norm": 0.1336507350206375, "learning_rate": 4.984301313120771e-06, "loss": 0.0049, "step": 259150 }, { "epoch": 165.17527087316762, "grad_norm": 0.29912111163139343, "learning_rate": 4.9819027269636775e-06, "loss": 0.008, "step": 259160 }, { "epoch": 165.18164435946463, "grad_norm": 0.23914402723312378, "learning_rate": 4.979504687818987e-06, "loss": 0.0056, "step": 259170 }, { "epoch": 165.18801784576164, "grad_norm": 0.174147367477417, "learning_rate": 4.977107195715807e-06, "loss": 0.0041, "step": 259180 }, { "epoch": 165.19439133205864, "grad_norm": 0.12393385916948318, "learning_rate": 4.974710250683301e-06, "loss": 0.0068, "step": 259190 }, { "epoch": 165.20076481835565, "grad_norm": 0.13920527696609497, "learning_rate": 4.97231385275056e-06, "loss": 0.0074, "step": 259200 }, { "epoch": 165.20713830465266, "grad_norm": 0.4070513844490051, "learning_rate": 4.969918001946727e-06, "loss": 0.0047, "step": 259210 }, { "epoch": 165.21351179094964, "grad_norm": 0.11474504321813583, "learning_rate": 4.967522698300908e-06, "loss": 0.0084, "step": 259220 }, { "epoch": 165.21988527724665, "grad_norm": 0.3236656188964844, "learning_rate": 4.965127941842201e-06, "loss": 0.004, "step": 259230 }, { "epoch": 165.22625876354365, "grad_norm": 0.20114760100841522, "learning_rate": 4.962733732599706e-06, "loss": 0.0053, "step": 259240 }, { "epoch": 165.23263224984066, "grad_norm": 0.12130871415138245, "learning_rate": 4.960340070602526e-06, "loss": 0.0038, "step": 259250 }, { "epoch": 165.23900573613767, "grad_norm": 0.3080345392227173, "learning_rate": 4.957946955879733e-06, "loss": 0.0058, "step": 259260 }, { "epoch": 165.24537922243468, "grad_norm": 0.10207809507846832, "learning_rate": 4.955554388460421e-06, "loss": 0.008, "step": 259270 }, { "epoch": 165.2517527087317, "grad_norm": 0.10726969689130783, "learning_rate": 4.953162368373637e-06, "loss": 0.0038, "step": 259280 }, { "epoch": 165.2581261950287, "grad_norm": 0.11463458091020584, "learning_rate": 4.950770895648471e-06, "loss": 0.0057, "step": 259290 }, { "epoch": 165.26449968132567, "grad_norm": 0.14209693670272827, "learning_rate": 4.948379970313971e-06, "loss": 0.0058, "step": 259300 }, { "epoch": 165.27087316762268, "grad_norm": 0.14505165815353394, "learning_rate": 4.9459895923991915e-06, "loss": 0.0052, "step": 259310 }, { "epoch": 165.2772466539197, "grad_norm": 0.34082484245300293, "learning_rate": 4.943599761933171e-06, "loss": 0.0056, "step": 259320 }, { "epoch": 165.2836201402167, "grad_norm": 0.19580908119678497, "learning_rate": 4.941210478944969e-06, "loss": 0.007, "step": 259330 }, { "epoch": 165.2899936265137, "grad_norm": 0.3596380650997162, "learning_rate": 4.938821743463578e-06, "loss": 0.005, "step": 259340 }, { "epoch": 165.2963671128107, "grad_norm": 0.08784553408622742, "learning_rate": 4.936433555518066e-06, "loss": 0.0031, "step": 259350 }, { "epoch": 165.30274059910772, "grad_norm": 0.28097066283226013, "learning_rate": 4.9340459151374196e-06, "loss": 0.0061, "step": 259360 }, { "epoch": 165.30911408540473, "grad_norm": 0.18844088912010193, "learning_rate": 4.931658822350682e-06, "loss": 0.0313, "step": 259370 }, { "epoch": 165.3154875717017, "grad_norm": 0.153728649020195, "learning_rate": 4.929272277186825e-06, "loss": 0.0065, "step": 259380 }, { "epoch": 165.32186105799872, "grad_norm": 0.17568954825401306, "learning_rate": 4.926886279674875e-06, "loss": 0.0074, "step": 259390 }, { "epoch": 165.32823454429573, "grad_norm": 0.1523122936487198, "learning_rate": 4.924500829843809e-06, "loss": 0.0083, "step": 259400 }, { "epoch": 165.33460803059273, "grad_norm": 0.2550241947174072, "learning_rate": 4.9221159277226205e-06, "loss": 0.0053, "step": 259410 }, { "epoch": 165.34098151688974, "grad_norm": 0.13363972306251526, "learning_rate": 4.919731573340269e-06, "loss": 0.0227, "step": 259420 }, { "epoch": 165.34735500318675, "grad_norm": 0.31063684821128845, "learning_rate": 4.917347766725755e-06, "loss": 0.0071, "step": 259430 }, { "epoch": 165.35372848948376, "grad_norm": 0.1398000419139862, "learning_rate": 4.914964507908032e-06, "loss": 0.008, "step": 259440 }, { "epoch": 165.36010197578076, "grad_norm": 0.20035141706466675, "learning_rate": 4.912581796916055e-06, "loss": 0.0133, "step": 259450 }, { "epoch": 165.36647546207774, "grad_norm": 0.1927509605884552, "learning_rate": 4.910199633778767e-06, "loss": 0.0053, "step": 259460 }, { "epoch": 165.37284894837475, "grad_norm": 0.3175051510334015, "learning_rate": 4.90781801852514e-06, "loss": 0.0049, "step": 259470 }, { "epoch": 165.37922243467176, "grad_norm": 0.09951918572187424, "learning_rate": 4.9054369511840945e-06, "loss": 0.0078, "step": 259480 }, { "epoch": 165.38559592096877, "grad_norm": 0.4094008505344391, "learning_rate": 4.903056431784564e-06, "loss": 0.0114, "step": 259490 }, { "epoch": 165.39196940726578, "grad_norm": 0.0699811652302742, "learning_rate": 4.900676460355474e-06, "loss": 0.0059, "step": 259500 }, { "epoch": 165.39834289356278, "grad_norm": 0.3782654404640198, "learning_rate": 4.898297036925748e-06, "loss": 0.0063, "step": 259510 }, { "epoch": 165.4047163798598, "grad_norm": 0.0697590559720993, "learning_rate": 4.895918161524304e-06, "loss": 0.0052, "step": 259520 }, { "epoch": 165.4110898661568, "grad_norm": 0.3944982886314392, "learning_rate": 4.893539834180033e-06, "loss": 0.0043, "step": 259530 }, { "epoch": 165.41746335245378, "grad_norm": 0.11411939561367035, "learning_rate": 4.891162054921844e-06, "loss": 0.011, "step": 259540 }, { "epoch": 165.4238368387508, "grad_norm": 0.09024091064929962, "learning_rate": 4.888784823778614e-06, "loss": 0.0034, "step": 259550 }, { "epoch": 165.4302103250478, "grad_norm": 0.10290146619081497, "learning_rate": 4.886408140779253e-06, "loss": 0.0078, "step": 259560 }, { "epoch": 165.4365838113448, "grad_norm": 0.5569518804550171, "learning_rate": 4.884032005952627e-06, "loss": 0.006, "step": 259570 }, { "epoch": 165.4429572976418, "grad_norm": 0.10022032260894775, "learning_rate": 4.881656419327602e-06, "loss": 0.0034, "step": 259580 }, { "epoch": 165.44933078393882, "grad_norm": 0.25728046894073486, "learning_rate": 4.879281380933048e-06, "loss": 0.0127, "step": 259590 }, { "epoch": 165.45570427023583, "grad_norm": 0.24685755372047424, "learning_rate": 4.876906890797833e-06, "loss": 0.0029, "step": 259600 }, { "epoch": 165.46207775653284, "grad_norm": 0.1265050768852234, "learning_rate": 4.874532948950805e-06, "loss": 0.0049, "step": 259610 }, { "epoch": 165.46845124282981, "grad_norm": 0.1610865443944931, "learning_rate": 4.872159555420802e-06, "loss": 0.0054, "step": 259620 }, { "epoch": 165.47482472912682, "grad_norm": 0.1429799199104309, "learning_rate": 4.8697867102366625e-06, "loss": 0.0071, "step": 259630 }, { "epoch": 165.48119821542383, "grad_norm": 0.4075421094894409, "learning_rate": 4.867414413427235e-06, "loss": 0.0053, "step": 259640 }, { "epoch": 165.48757170172084, "grad_norm": 0.06281633675098419, "learning_rate": 4.865042665021335e-06, "loss": 0.0029, "step": 259650 }, { "epoch": 165.49394518801785, "grad_norm": 0.15708982944488525, "learning_rate": 4.86267146504778e-06, "loss": 0.0028, "step": 259660 }, { "epoch": 165.50031867431485, "grad_norm": 0.2767406702041626, "learning_rate": 4.860300813535373e-06, "loss": 0.0054, "step": 259670 }, { "epoch": 165.50669216061186, "grad_norm": 0.27407434582710266, "learning_rate": 4.857930710512936e-06, "loss": 0.0068, "step": 259680 }, { "epoch": 165.51306564690887, "grad_norm": 0.20909491181373596, "learning_rate": 4.85556115600927e-06, "loss": 0.0062, "step": 259690 }, { "epoch": 165.51943913320585, "grad_norm": 0.08569963276386261, "learning_rate": 4.853192150053154e-06, "loss": 0.0032, "step": 259700 }, { "epoch": 165.52581261950286, "grad_norm": 0.18202394247055054, "learning_rate": 4.850823692673368e-06, "loss": 0.0038, "step": 259710 }, { "epoch": 165.53218610579987, "grad_norm": 0.340250164270401, "learning_rate": 4.848455783898725e-06, "loss": 0.0064, "step": 259720 }, { "epoch": 165.53855959209687, "grad_norm": 0.4820033609867096, "learning_rate": 4.846088423757955e-06, "loss": 0.0036, "step": 259730 }, { "epoch": 165.54493307839388, "grad_norm": 0.26139146089553833, "learning_rate": 4.84372161227985e-06, "loss": 0.0081, "step": 259740 }, { "epoch": 165.5513065646909, "grad_norm": 0.07029587775468826, "learning_rate": 4.841355349493154e-06, "loss": 0.0059, "step": 259750 }, { "epoch": 165.5576800509879, "grad_norm": 0.13653564453125, "learning_rate": 4.838989635426644e-06, "loss": 0.0063, "step": 259760 }, { "epoch": 165.5640535372849, "grad_norm": 0.11357391625642776, "learning_rate": 4.836624470109036e-06, "loss": 0.0046, "step": 259770 }, { "epoch": 165.57042702358189, "grad_norm": 0.20583203434944153, "learning_rate": 4.8342598535690846e-06, "loss": 0.0108, "step": 259780 }, { "epoch": 165.5768005098789, "grad_norm": 0.13014844059944153, "learning_rate": 4.831895785835527e-06, "loss": 0.0034, "step": 259790 }, { "epoch": 165.5831739961759, "grad_norm": 0.3362135887145996, "learning_rate": 4.829532266937076e-06, "loss": 0.0058, "step": 259800 }, { "epoch": 165.5895474824729, "grad_norm": 0.5085921287536621, "learning_rate": 4.827169296902445e-06, "loss": 0.0104, "step": 259810 }, { "epoch": 165.59592096876992, "grad_norm": 0.2745073139667511, "learning_rate": 4.824806875760368e-06, "loss": 0.0058, "step": 259820 }, { "epoch": 165.60229445506693, "grad_norm": 0.17109791934490204, "learning_rate": 4.822445003539544e-06, "loss": 0.0046, "step": 259830 }, { "epoch": 165.60866794136393, "grad_norm": 0.13571420311927795, "learning_rate": 4.820083680268667e-06, "loss": 0.0032, "step": 259840 }, { "epoch": 165.61504142766094, "grad_norm": 0.11169199645519257, "learning_rate": 4.817722905976418e-06, "loss": 0.0058, "step": 259850 }, { "epoch": 165.62141491395795, "grad_norm": 0.223605677485466, "learning_rate": 4.815362680691515e-06, "loss": 0.0056, "step": 259860 }, { "epoch": 165.62778840025493, "grad_norm": 0.13921165466308594, "learning_rate": 4.813003004442596e-06, "loss": 0.0037, "step": 259870 }, { "epoch": 165.63416188655194, "grad_norm": 0.10869883000850677, "learning_rate": 4.810643877258364e-06, "loss": 0.0054, "step": 259880 }, { "epoch": 165.64053537284894, "grad_norm": 0.2365865856409073, "learning_rate": 4.808285299167465e-06, "loss": 0.0055, "step": 259890 }, { "epoch": 165.64690885914595, "grad_norm": 0.15174202620983124, "learning_rate": 4.80592727019859e-06, "loss": 0.0063, "step": 259900 }, { "epoch": 165.65328234544296, "grad_norm": 0.11004827171564102, "learning_rate": 4.8035697903803465e-06, "loss": 0.0042, "step": 259910 }, { "epoch": 165.65965583173997, "grad_norm": 0.10162004083395004, "learning_rate": 4.8012128597414164e-06, "loss": 0.0108, "step": 259920 }, { "epoch": 165.66602931803698, "grad_norm": 0.27953046560287476, "learning_rate": 4.798856478310409e-06, "loss": 0.0064, "step": 259930 }, { "epoch": 165.67240280433398, "grad_norm": 0.12063481658697128, "learning_rate": 4.7965006461159975e-06, "loss": 0.0033, "step": 259940 }, { "epoch": 165.67877629063096, "grad_norm": 0.10147932171821594, "learning_rate": 4.794145363186758e-06, "loss": 0.003, "step": 259950 }, { "epoch": 165.68514977692797, "grad_norm": 0.15382057428359985, "learning_rate": 4.791790629551346e-06, "loss": 0.0058, "step": 259960 }, { "epoch": 165.69152326322498, "grad_norm": 0.09055256843566895, "learning_rate": 4.78943644523836e-06, "loss": 0.0048, "step": 259970 }, { "epoch": 165.697896749522, "grad_norm": 0.14855964481830597, "learning_rate": 4.787082810276411e-06, "loss": 0.013, "step": 259980 }, { "epoch": 165.704270235819, "grad_norm": 0.14961279928684235, "learning_rate": 4.784729724694082e-06, "loss": 0.0131, "step": 259990 }, { "epoch": 165.710643722116, "grad_norm": 0.11250326782464981, "learning_rate": 4.782377188519982e-06, "loss": 0.0047, "step": 260000 }, { "epoch": 165.717017208413, "grad_norm": 0.32960352301597595, "learning_rate": 4.780025201782695e-06, "loss": 0.0083, "step": 260010 }, { "epoch": 165.72339069471002, "grad_norm": 0.16210196912288666, "learning_rate": 4.777673764510798e-06, "loss": 0.0093, "step": 260020 }, { "epoch": 165.729764181007, "grad_norm": 0.3730055093765259, "learning_rate": 4.7753228767328514e-06, "loss": 0.0137, "step": 260030 }, { "epoch": 165.736137667304, "grad_norm": 0.7107915878295898, "learning_rate": 4.772972538477455e-06, "loss": 0.0065, "step": 260040 }, { "epoch": 165.74251115360101, "grad_norm": 0.09989802539348602, "learning_rate": 4.770622749773118e-06, "loss": 0.0051, "step": 260050 }, { "epoch": 165.74888463989802, "grad_norm": 0.15095874667167664, "learning_rate": 4.76827351064843e-06, "loss": 0.0047, "step": 260060 }, { "epoch": 165.75525812619503, "grad_norm": 0.06583627313375473, "learning_rate": 4.7659248211319115e-06, "loss": 0.0075, "step": 260070 }, { "epoch": 165.76163161249204, "grad_norm": 0.1007314920425415, "learning_rate": 4.7635766812521406e-06, "loss": 0.0085, "step": 260080 }, { "epoch": 165.76800509878905, "grad_norm": 0.21446199715137482, "learning_rate": 4.7612290910376e-06, "loss": 0.0096, "step": 260090 }, { "epoch": 165.77437858508605, "grad_norm": 0.14196337759494781, "learning_rate": 4.7588820505168495e-06, "loss": 0.0069, "step": 260100 }, { "epoch": 165.78075207138303, "grad_norm": 0.20742954313755035, "learning_rate": 4.756535559718395e-06, "loss": 0.0061, "step": 260110 }, { "epoch": 165.78712555768004, "grad_norm": 0.15138085186481476, "learning_rate": 4.754189618670751e-06, "loss": 0.0029, "step": 260120 }, { "epoch": 165.79349904397705, "grad_norm": 0.162701815366745, "learning_rate": 4.751844227402413e-06, "loss": 0.0069, "step": 260130 }, { "epoch": 165.79987253027406, "grad_norm": 0.2947889268398285, "learning_rate": 4.749499385941897e-06, "loss": 0.0056, "step": 260140 }, { "epoch": 165.80624601657107, "grad_norm": 0.06515590846538544, "learning_rate": 4.74715509431769e-06, "loss": 0.0067, "step": 260150 }, { "epoch": 165.81261950286807, "grad_norm": 0.4914098381996155, "learning_rate": 4.7448113525582705e-06, "loss": 0.0218, "step": 260160 }, { "epoch": 165.81899298916508, "grad_norm": 0.10352760553359985, "learning_rate": 4.742468160692115e-06, "loss": 0.007, "step": 260170 }, { "epoch": 165.8253664754621, "grad_norm": 0.30511030554771423, "learning_rate": 4.740125518747707e-06, "loss": 0.004, "step": 260180 }, { "epoch": 165.83173996175907, "grad_norm": 0.23607932031154633, "learning_rate": 4.737783426753512e-06, "loss": 0.0046, "step": 260190 }, { "epoch": 165.83811344805608, "grad_norm": 0.08564259111881256, "learning_rate": 4.735441884737979e-06, "loss": 0.0066, "step": 260200 }, { "epoch": 165.84448693435309, "grad_norm": 0.4186568260192871, "learning_rate": 4.733100892729553e-06, "loss": 0.0092, "step": 260210 }, { "epoch": 165.8508604206501, "grad_norm": 0.11333822458982468, "learning_rate": 4.730760450756705e-06, "loss": 0.0039, "step": 260220 }, { "epoch": 165.8572339069471, "grad_norm": 0.12906615436077118, "learning_rate": 4.728420558847857e-06, "loss": 0.0024, "step": 260230 }, { "epoch": 165.8636073932441, "grad_norm": 0.09990564733743668, "learning_rate": 4.726081217031436e-06, "loss": 0.0107, "step": 260240 }, { "epoch": 165.86998087954112, "grad_norm": 0.11928517371416092, "learning_rate": 4.723742425335875e-06, "loss": 0.0079, "step": 260250 }, { "epoch": 165.87635436583813, "grad_norm": 0.08519551903009415, "learning_rate": 4.721404183789596e-06, "loss": 0.005, "step": 260260 }, { "epoch": 165.8827278521351, "grad_norm": 0.22131836414337158, "learning_rate": 4.719066492421004e-06, "loss": 0.0062, "step": 260270 }, { "epoch": 165.8891013384321, "grad_norm": 0.23593689501285553, "learning_rate": 4.716729351258509e-06, "loss": 0.0039, "step": 260280 }, { "epoch": 165.89547482472912, "grad_norm": 0.34942272305488586, "learning_rate": 4.714392760330505e-06, "loss": 0.0065, "step": 260290 }, { "epoch": 165.90184831102613, "grad_norm": 0.0789266973733902, "learning_rate": 4.712056719665381e-06, "loss": 0.007, "step": 260300 }, { "epoch": 165.90822179732314, "grad_norm": 0.14614880084991455, "learning_rate": 4.709721229291531e-06, "loss": 0.0057, "step": 260310 }, { "epoch": 165.91459528362014, "grad_norm": 0.08009956032037735, "learning_rate": 4.707386289237331e-06, "loss": 0.0033, "step": 260320 }, { "epoch": 165.92096876991715, "grad_norm": 0.15410782396793365, "learning_rate": 4.705051899531149e-06, "loss": 0.0052, "step": 260330 }, { "epoch": 165.92734225621416, "grad_norm": 0.15296819806098938, "learning_rate": 4.7027180602013446e-06, "loss": 0.0048, "step": 260340 }, { "epoch": 165.93371574251114, "grad_norm": 0.32129231095314026, "learning_rate": 4.700384771276295e-06, "loss": 0.0069, "step": 260350 }, { "epoch": 165.94008922880815, "grad_norm": 0.1048828586935997, "learning_rate": 4.6980520327843335e-06, "loss": 0.0048, "step": 260360 }, { "epoch": 165.94646271510516, "grad_norm": 0.11567486822605133, "learning_rate": 4.6957198447538206e-06, "loss": 0.0142, "step": 260370 }, { "epoch": 165.95283620140216, "grad_norm": 0.12445798516273499, "learning_rate": 4.693388207213073e-06, "loss": 0.0051, "step": 260380 }, { "epoch": 165.95920968769917, "grad_norm": 0.13689261674880981, "learning_rate": 4.6910571201904394e-06, "loss": 0.0089, "step": 260390 }, { "epoch": 165.96558317399618, "grad_norm": 0.21238097548484802, "learning_rate": 4.688726583714248e-06, "loss": 0.0052, "step": 260400 }, { "epoch": 165.9719566602932, "grad_norm": 0.07420655339956284, "learning_rate": 4.6863965978128086e-06, "loss": 0.0056, "step": 260410 }, { "epoch": 165.9783301465902, "grad_norm": 0.18276934325695038, "learning_rate": 4.684067162514422e-06, "loss": 0.0045, "step": 260420 }, { "epoch": 165.98470363288718, "grad_norm": 0.233768031001091, "learning_rate": 4.6817382778474194e-06, "loss": 0.0131, "step": 260430 }, { "epoch": 165.99107711918418, "grad_norm": 0.16450820863246918, "learning_rate": 4.679409943840074e-06, "loss": 0.0023, "step": 260440 }, { "epoch": 165.9974506054812, "grad_norm": 0.09753008931875229, "learning_rate": 4.677082160520691e-06, "loss": 0.0043, "step": 260450 }, { "epoch": 166.0038240917782, "grad_norm": 0.27308231592178345, "learning_rate": 4.674754927917546e-06, "loss": 0.0043, "step": 260460 }, { "epoch": 166.0101975780752, "grad_norm": 0.10401882231235504, "learning_rate": 4.672428246058935e-06, "loss": 0.0103, "step": 260470 }, { "epoch": 166.01657106437222, "grad_norm": 0.07066141068935394, "learning_rate": 4.670102114973107e-06, "loss": 0.0035, "step": 260480 }, { "epoch": 166.02294455066922, "grad_norm": 0.15170693397521973, "learning_rate": 4.66777653468834e-06, "loss": 0.0048, "step": 260490 }, { "epoch": 166.02931803696623, "grad_norm": 0.16167192161083221, "learning_rate": 4.665451505232882e-06, "loss": 0.0067, "step": 260500 }, { "epoch": 166.03569152326324, "grad_norm": 0.22420787811279297, "learning_rate": 4.663127026635011e-06, "loss": 0.0049, "step": 260510 }, { "epoch": 166.04206500956022, "grad_norm": 0.08520206809043884, "learning_rate": 4.660803098922928e-06, "loss": 0.007, "step": 260520 }, { "epoch": 166.04843849585723, "grad_norm": 0.06780404597520828, "learning_rate": 4.658479722124909e-06, "loss": 0.0182, "step": 260530 }, { "epoch": 166.05481198215423, "grad_norm": 0.15134762227535248, "learning_rate": 4.65615689626917e-06, "loss": 0.0062, "step": 260540 }, { "epoch": 166.06118546845124, "grad_norm": 0.08098101615905762, "learning_rate": 4.653834621383935e-06, "loss": 0.0086, "step": 260550 }, { "epoch": 166.06755895474825, "grad_norm": 0.09926863014698029, "learning_rate": 4.6515128974974175e-06, "loss": 0.0078, "step": 260560 }, { "epoch": 166.07393244104526, "grad_norm": 0.11767230927944183, "learning_rate": 4.649191724637836e-06, "loss": 0.0075, "step": 260570 }, { "epoch": 166.08030592734227, "grad_norm": 0.08925506472587585, "learning_rate": 4.6468711028334e-06, "loss": 0.0062, "step": 260580 }, { "epoch": 166.08667941363927, "grad_norm": 0.32041025161743164, "learning_rate": 4.6445510321122986e-06, "loss": 0.0047, "step": 260590 }, { "epoch": 166.09305289993625, "grad_norm": 0.1291348785161972, "learning_rate": 4.642231512502715e-06, "loss": 0.014, "step": 260600 }, { "epoch": 166.09942638623326, "grad_norm": 0.1383206695318222, "learning_rate": 4.639912544032865e-06, "loss": 0.0151, "step": 260610 }, { "epoch": 166.10579987253027, "grad_norm": 0.15583644807338715, "learning_rate": 4.637594126730882e-06, "loss": 0.0035, "step": 260620 }, { "epoch": 166.11217335882728, "grad_norm": 0.5288916230201721, "learning_rate": 4.635276260624971e-06, "loss": 0.0094, "step": 260630 }, { "epoch": 166.11854684512429, "grad_norm": 0.12999539077281952, "learning_rate": 4.6329589457432765e-06, "loss": 0.005, "step": 260640 }, { "epoch": 166.1249203314213, "grad_norm": 0.5240561962127686, "learning_rate": 4.630642182113981e-06, "loss": 0.0034, "step": 260650 }, { "epoch": 166.1312938177183, "grad_norm": 0.661456823348999, "learning_rate": 4.628325969765202e-06, "loss": 0.0053, "step": 260660 }, { "epoch": 166.1376673040153, "grad_norm": 0.12018027901649475, "learning_rate": 4.626010308725104e-06, "loss": 0.0039, "step": 260670 }, { "epoch": 166.1440407903123, "grad_norm": 0.12461700290441513, "learning_rate": 4.623695199021827e-06, "loss": 0.0039, "step": 260680 }, { "epoch": 166.1504142766093, "grad_norm": 0.13106806576251984, "learning_rate": 4.6213806406834925e-06, "loss": 0.0156, "step": 260690 }, { "epoch": 166.1567877629063, "grad_norm": 0.12585443258285522, "learning_rate": 4.619066633738223e-06, "loss": 0.0037, "step": 260700 }, { "epoch": 166.1631612492033, "grad_norm": 0.1142767071723938, "learning_rate": 4.616753178214145e-06, "loss": 0.0054, "step": 260710 }, { "epoch": 166.16953473550032, "grad_norm": 0.10673681646585464, "learning_rate": 4.6144402741393645e-06, "loss": 0.003, "step": 260720 }, { "epoch": 166.17590822179733, "grad_norm": 0.07986792922019958, "learning_rate": 4.612127921541987e-06, "loss": 0.0058, "step": 260730 }, { "epoch": 166.18228170809434, "grad_norm": 0.2455611526966095, "learning_rate": 4.6098161204501e-06, "loss": 0.0045, "step": 260740 }, { "epoch": 166.18865519439134, "grad_norm": 0.10788682848215103, "learning_rate": 4.607504870891821e-06, "loss": 0.0029, "step": 260750 }, { "epoch": 166.19502868068832, "grad_norm": 0.7634270191192627, "learning_rate": 4.6051941728951995e-06, "loss": 0.005, "step": 260760 }, { "epoch": 166.20140216698533, "grad_norm": 0.11021155118942261, "learning_rate": 4.6028840264883355e-06, "loss": 0.0047, "step": 260770 }, { "epoch": 166.20777565328234, "grad_norm": 0.5384775996208191, "learning_rate": 4.600574431699284e-06, "loss": 0.0041, "step": 260780 }, { "epoch": 166.21414913957935, "grad_norm": 0.24133382737636566, "learning_rate": 4.598265388556133e-06, "loss": 0.0034, "step": 260790 }, { "epoch": 166.22052262587636, "grad_norm": 0.3476819694042206, "learning_rate": 4.595956897086906e-06, "loss": 0.0047, "step": 260800 }, { "epoch": 166.22689611217336, "grad_norm": 0.12961427867412567, "learning_rate": 4.593648957319685e-06, "loss": 0.0085, "step": 260810 }, { "epoch": 166.23326959847037, "grad_norm": 0.4429541528224945, "learning_rate": 4.591341569282487e-06, "loss": 0.0108, "step": 260820 }, { "epoch": 166.23964308476738, "grad_norm": 0.11019319295883179, "learning_rate": 4.589034733003378e-06, "loss": 0.004, "step": 260830 }, { "epoch": 166.24601657106436, "grad_norm": 0.10624050348997116, "learning_rate": 4.586728448510358e-06, "loss": 0.006, "step": 260840 }, { "epoch": 166.25239005736137, "grad_norm": 0.26355302333831787, "learning_rate": 4.584422715831471e-06, "loss": 0.0042, "step": 260850 }, { "epoch": 166.25876354365838, "grad_norm": 0.11849760264158249, "learning_rate": 4.582117534994728e-06, "loss": 0.0018, "step": 260860 }, { "epoch": 166.26513702995538, "grad_norm": 0.2548305094242096, "learning_rate": 4.579812906028142e-06, "loss": 0.0071, "step": 260870 }, { "epoch": 166.2715105162524, "grad_norm": 0.11098156124353409, "learning_rate": 4.577508828959698e-06, "loss": 0.0102, "step": 260880 }, { "epoch": 166.2778840025494, "grad_norm": 0.3613797426223755, "learning_rate": 4.57520530381742e-06, "loss": 0.0057, "step": 260890 }, { "epoch": 166.2842574888464, "grad_norm": 0.08389043807983398, "learning_rate": 4.572902330629286e-06, "loss": 0.0088, "step": 260900 }, { "epoch": 166.29063097514342, "grad_norm": 0.12973567843437195, "learning_rate": 4.570599909423279e-06, "loss": 0.0046, "step": 260910 }, { "epoch": 166.2970044614404, "grad_norm": 0.09740287065505981, "learning_rate": 4.5682980402273644e-06, "loss": 0.0044, "step": 260920 }, { "epoch": 166.3033779477374, "grad_norm": 0.2687484920024872, "learning_rate": 4.565996723069532e-06, "loss": 0.0084, "step": 260930 }, { "epoch": 166.3097514340344, "grad_norm": 0.10688439756631851, "learning_rate": 4.563695957977737e-06, "loss": 0.0068, "step": 260940 }, { "epoch": 166.31612492033142, "grad_norm": 0.046998750418424606, "learning_rate": 4.561395744979941e-06, "loss": 0.0041, "step": 260950 }, { "epoch": 166.32249840662843, "grad_norm": 0.20698684453964233, "learning_rate": 4.55909608410407e-06, "loss": 0.0031, "step": 260960 }, { "epoch": 166.32887189292543, "grad_norm": 0.2649538815021515, "learning_rate": 4.556796975378097e-06, "loss": 0.013, "step": 260970 }, { "epoch": 166.33524537922244, "grad_norm": 0.09040538221597672, "learning_rate": 4.55449841882995e-06, "loss": 0.0041, "step": 260980 }, { "epoch": 166.34161886551945, "grad_norm": 0.19581446051597595, "learning_rate": 4.552200414487551e-06, "loss": 0.0062, "step": 260990 }, { "epoch": 166.34799235181643, "grad_norm": 0.7899379134178162, "learning_rate": 4.549902962378827e-06, "loss": 0.0067, "step": 261000 }, { "epoch": 166.35436583811344, "grad_norm": 0.19249923527240753, "learning_rate": 4.547606062531684e-06, "loss": 0.0063, "step": 261010 }, { "epoch": 166.36073932441045, "grad_norm": 0.11899774521589279, "learning_rate": 4.5453097149740555e-06, "loss": 0.0033, "step": 261020 }, { "epoch": 166.36711281070745, "grad_norm": 0.12384123355150223, "learning_rate": 4.543013919733824e-06, "loss": 0.0062, "step": 261030 }, { "epoch": 166.37348629700446, "grad_norm": 0.2283763438463211, "learning_rate": 4.5407186768388955e-06, "loss": 0.0075, "step": 261040 }, { "epoch": 166.37985978330147, "grad_norm": 0.08626466989517212, "learning_rate": 4.538423986317148e-06, "loss": 0.0065, "step": 261050 }, { "epoch": 166.38623326959848, "grad_norm": 0.27683210372924805, "learning_rate": 4.53612984819648e-06, "loss": 0.0064, "step": 261060 }, { "epoch": 166.39260675589549, "grad_norm": 0.10946382582187653, "learning_rate": 4.533836262504759e-06, "loss": 0.0055, "step": 261070 }, { "epoch": 166.39898024219247, "grad_norm": 0.4305162727832794, "learning_rate": 4.5315432292698526e-06, "loss": 0.0091, "step": 261080 }, { "epoch": 166.40535372848947, "grad_norm": 0.17397567629814148, "learning_rate": 4.5292507485196145e-06, "loss": 0.0035, "step": 261090 }, { "epoch": 166.41172721478648, "grad_norm": 0.24368153512477875, "learning_rate": 4.526958820281923e-06, "loss": 0.0033, "step": 261100 }, { "epoch": 166.4181007010835, "grad_norm": 0.11175033450126648, "learning_rate": 4.524667444584618e-06, "loss": 0.0022, "step": 261110 }, { "epoch": 166.4244741873805, "grad_norm": 0.04789979010820389, "learning_rate": 4.522376621455538e-06, "loss": 0.0032, "step": 261120 }, { "epoch": 166.4308476736775, "grad_norm": 0.10016809403896332, "learning_rate": 4.520086350922509e-06, "loss": 0.0028, "step": 261130 }, { "epoch": 166.4372211599745, "grad_norm": 0.13015033304691315, "learning_rate": 4.517796633013382e-06, "loss": 0.0049, "step": 261140 }, { "epoch": 166.44359464627152, "grad_norm": 0.6329812407493591, "learning_rate": 4.51550746775597e-06, "loss": 0.0091, "step": 261150 }, { "epoch": 166.44996813256853, "grad_norm": 0.31978297233581543, "learning_rate": 4.513218855178081e-06, "loss": 0.0038, "step": 261160 }, { "epoch": 166.4563416188655, "grad_norm": 0.5770478248596191, "learning_rate": 4.5109307953075275e-06, "loss": 0.0113, "step": 261170 }, { "epoch": 166.46271510516252, "grad_norm": 0.1073809415102005, "learning_rate": 4.508643288172126e-06, "loss": 0.0054, "step": 261180 }, { "epoch": 166.46908859145952, "grad_norm": 0.24443919956684113, "learning_rate": 4.506356333799644e-06, "loss": 0.0086, "step": 261190 }, { "epoch": 166.47546207775653, "grad_norm": 0.22539839148521423, "learning_rate": 4.504069932217891e-06, "loss": 0.0033, "step": 261200 }, { "epoch": 166.48183556405354, "grad_norm": 0.09930557012557983, "learning_rate": 4.501784083454641e-06, "loss": 0.0029, "step": 261210 }, { "epoch": 166.48820905035055, "grad_norm": 0.12506745755672455, "learning_rate": 4.499498787537687e-06, "loss": 0.0037, "step": 261220 }, { "epoch": 166.49458253664756, "grad_norm": 0.2743285894393921, "learning_rate": 4.497214044494763e-06, "loss": 0.0062, "step": 261230 }, { "epoch": 166.50095602294456, "grad_norm": 0.17748649418354034, "learning_rate": 4.4949298543536575e-06, "loss": 0.0134, "step": 261240 }, { "epoch": 166.50732950924154, "grad_norm": 0.43790674209594727, "learning_rate": 4.492646217142121e-06, "loss": 0.0065, "step": 261250 }, { "epoch": 166.51370299553855, "grad_norm": 0.15499696135520935, "learning_rate": 4.490363132887904e-06, "loss": 0.0107, "step": 261260 }, { "epoch": 166.52007648183556, "grad_norm": 0.1325688660144806, "learning_rate": 4.488080601618727e-06, "loss": 0.004, "step": 261270 }, { "epoch": 166.52644996813257, "grad_norm": 0.14819423854351044, "learning_rate": 4.485798623362353e-06, "loss": 0.0028, "step": 261280 }, { "epoch": 166.53282345442958, "grad_norm": 0.10679399967193604, "learning_rate": 4.483517198146497e-06, "loss": 0.0026, "step": 261290 }, { "epoch": 166.53919694072658, "grad_norm": 0.5452781915664673, "learning_rate": 4.481236325998878e-06, "loss": 0.0039, "step": 261300 }, { "epoch": 166.5455704270236, "grad_norm": 0.11789771914482117, "learning_rate": 4.47895600694721e-06, "loss": 0.0048, "step": 261310 }, { "epoch": 166.5519439133206, "grad_norm": 0.1618393510580063, "learning_rate": 4.476676241019223e-06, "loss": 0.0048, "step": 261320 }, { "epoch": 166.55831739961758, "grad_norm": 0.15042582154273987, "learning_rate": 4.474397028242583e-06, "loss": 0.0051, "step": 261330 }, { "epoch": 166.5646908859146, "grad_norm": 0.09723037481307983, "learning_rate": 4.472118368645012e-06, "loss": 0.0067, "step": 261340 }, { "epoch": 166.5710643722116, "grad_norm": 0.13430464267730713, "learning_rate": 4.4698402622541765e-06, "loss": 0.0063, "step": 261350 }, { "epoch": 166.5774378585086, "grad_norm": 0.18591152131557465, "learning_rate": 4.467562709097789e-06, "loss": 0.0099, "step": 261360 }, { "epoch": 166.5838113448056, "grad_norm": 0.12807925045490265, "learning_rate": 4.465285709203487e-06, "loss": 0.0074, "step": 261370 }, { "epoch": 166.59018483110262, "grad_norm": 0.15409676730632782, "learning_rate": 4.463009262598966e-06, "loss": 0.0056, "step": 261380 }, { "epoch": 166.59655831739963, "grad_norm": 0.3729633092880249, "learning_rate": 4.4607333693118646e-06, "loss": 0.011, "step": 261390 }, { "epoch": 166.60293180369663, "grad_norm": 0.22176885604858398, "learning_rate": 4.458458029369872e-06, "loss": 0.0046, "step": 261400 }, { "epoch": 166.6093052899936, "grad_norm": 0.08644382655620575, "learning_rate": 4.456183242800588e-06, "loss": 0.0027, "step": 261410 }, { "epoch": 166.61567877629062, "grad_norm": 0.5083773732185364, "learning_rate": 4.453909009631691e-06, "loss": 0.0058, "step": 261420 }, { "epoch": 166.62205226258763, "grad_norm": 0.28883203864097595, "learning_rate": 4.451635329890802e-06, "loss": 0.0102, "step": 261430 }, { "epoch": 166.62842574888464, "grad_norm": 0.48773321509361267, "learning_rate": 4.449362203605545e-06, "loss": 0.0048, "step": 261440 }, { "epoch": 166.63479923518165, "grad_norm": 0.06001149117946625, "learning_rate": 4.447089630803536e-06, "loss": 0.0033, "step": 261450 }, { "epoch": 166.64117272147865, "grad_norm": 0.13106806576251984, "learning_rate": 4.444817611512408e-06, "loss": 0.0022, "step": 261460 }, { "epoch": 166.64754620777566, "grad_norm": 0.18397527933120728, "learning_rate": 4.4425461457597575e-06, "loss": 0.0039, "step": 261470 }, { "epoch": 166.65391969407267, "grad_norm": 0.28997814655303955, "learning_rate": 4.440275233573188e-06, "loss": 0.0073, "step": 261480 }, { "epoch": 166.66029318036965, "grad_norm": 0.4116305410861969, "learning_rate": 4.438004874980273e-06, "loss": 0.0102, "step": 261490 }, { "epoch": 166.66666666666666, "grad_norm": 0.17706309258937836, "learning_rate": 4.43573507000864e-06, "loss": 0.0032, "step": 261500 }, { "epoch": 166.67304015296367, "grad_norm": 0.28326916694641113, "learning_rate": 4.433465818685828e-06, "loss": 0.0037, "step": 261510 }, { "epoch": 166.67941363926067, "grad_norm": 0.1694047451019287, "learning_rate": 4.431197121039432e-06, "loss": 0.0039, "step": 261520 }, { "epoch": 166.68578712555768, "grad_norm": 0.07282193750143051, "learning_rate": 4.428928977097013e-06, "loss": 0.0063, "step": 261530 }, { "epoch": 166.6921606118547, "grad_norm": 0.10834052413702011, "learning_rate": 4.426661386886144e-06, "loss": 0.0068, "step": 261540 }, { "epoch": 166.6985340981517, "grad_norm": 0.26237303018569946, "learning_rate": 4.424394350434352e-06, "loss": 0.0036, "step": 261550 }, { "epoch": 166.7049075844487, "grad_norm": 0.08684927225112915, "learning_rate": 4.42212786776921e-06, "loss": 0.0091, "step": 261560 }, { "epoch": 166.71128107074568, "grad_norm": 0.2613794207572937, "learning_rate": 4.419861938918246e-06, "loss": 0.0044, "step": 261570 }, { "epoch": 166.7176545570427, "grad_norm": 0.2183610200881958, "learning_rate": 4.417596563908999e-06, "loss": 0.0059, "step": 261580 }, { "epoch": 166.7240280433397, "grad_norm": 0.12743566930294037, "learning_rate": 4.415331742768975e-06, "loss": 0.0056, "step": 261590 }, { "epoch": 166.7304015296367, "grad_norm": 0.30415403842926025, "learning_rate": 4.413067475525723e-06, "loss": 0.0038, "step": 261600 }, { "epoch": 166.73677501593372, "grad_norm": 0.26345938444137573, "learning_rate": 4.410803762206739e-06, "loss": 0.0064, "step": 261610 }, { "epoch": 166.74314850223072, "grad_norm": 0.14461934566497803, "learning_rate": 4.408540602839534e-06, "loss": 0.0073, "step": 261620 }, { "epoch": 166.74952198852773, "grad_norm": 0.2696053981781006, "learning_rate": 4.406277997451597e-06, "loss": 0.0055, "step": 261630 }, { "epoch": 166.75589547482474, "grad_norm": 0.11135032773017883, "learning_rate": 4.404015946070439e-06, "loss": 0.0045, "step": 261640 }, { "epoch": 166.76226896112172, "grad_norm": 0.13947805762290955, "learning_rate": 4.401754448723533e-06, "loss": 0.0083, "step": 261650 }, { "epoch": 166.76864244741873, "grad_norm": 0.32606276869773865, "learning_rate": 4.399493505438368e-06, "loss": 0.0086, "step": 261660 }, { "epoch": 166.77501593371574, "grad_norm": 0.14309105277061462, "learning_rate": 4.3972331162424e-06, "loss": 0.0032, "step": 261670 }, { "epoch": 166.78138942001274, "grad_norm": 0.11789558082818985, "learning_rate": 4.394973281163112e-06, "loss": 0.0121, "step": 261680 }, { "epoch": 166.78776290630975, "grad_norm": 0.43312588334083557, "learning_rate": 4.39271400022796e-06, "loss": 0.0104, "step": 261690 }, { "epoch": 166.79413639260676, "grad_norm": 0.1003049910068512, "learning_rate": 4.390455273464395e-06, "loss": 0.0037, "step": 261700 }, { "epoch": 166.80050987890377, "grad_norm": 0.08118170499801636, "learning_rate": 4.388197100899849e-06, "loss": 0.0052, "step": 261710 }, { "epoch": 166.80688336520078, "grad_norm": 0.06971612572669983, "learning_rate": 4.385939482561785e-06, "loss": 0.0063, "step": 261720 }, { "epoch": 166.81325685149778, "grad_norm": 0.13642500340938568, "learning_rate": 4.3836824184776195e-06, "loss": 0.0036, "step": 261730 }, { "epoch": 166.81963033779476, "grad_norm": 0.0967281311750412, "learning_rate": 4.38142590867478e-06, "loss": 0.0067, "step": 261740 }, { "epoch": 166.82600382409177, "grad_norm": 0.2253541499376297, "learning_rate": 4.379169953180695e-06, "loss": 0.0069, "step": 261750 }, { "epoch": 166.83237731038878, "grad_norm": 0.17385751008987427, "learning_rate": 4.376914552022754e-06, "loss": 0.0093, "step": 261760 }, { "epoch": 166.8387507966858, "grad_norm": 0.18829001486301422, "learning_rate": 4.374659705228384e-06, "loss": 0.0028, "step": 261770 }, { "epoch": 166.8451242829828, "grad_norm": 0.3097306191921234, "learning_rate": 4.372405412824976e-06, "loss": 0.0052, "step": 261780 }, { "epoch": 166.8514977692798, "grad_norm": 0.12000580132007599, "learning_rate": 4.370151674839923e-06, "loss": 0.0028, "step": 261790 }, { "epoch": 166.8578712555768, "grad_norm": 0.497667521238327, "learning_rate": 4.367898491300603e-06, "loss": 0.008, "step": 261800 }, { "epoch": 166.86424474187382, "grad_norm": 0.15031205117702484, "learning_rate": 4.365645862234402e-06, "loss": 0.0032, "step": 261810 }, { "epoch": 166.8706182281708, "grad_norm": 0.11091279983520508, "learning_rate": 4.363393787668696e-06, "loss": 0.0042, "step": 261820 }, { "epoch": 166.8769917144678, "grad_norm": 0.09275887906551361, "learning_rate": 4.361142267630841e-06, "loss": 0.0044, "step": 261830 }, { "epoch": 166.8833652007648, "grad_norm": 0.4108527898788452, "learning_rate": 4.358891302148193e-06, "loss": 0.0038, "step": 261840 }, { "epoch": 166.88973868706182, "grad_norm": 0.08076123148202896, "learning_rate": 4.35664089124811e-06, "loss": 0.0036, "step": 261850 }, { "epoch": 166.89611217335883, "grad_norm": 0.13105756044387817, "learning_rate": 4.354391034957939e-06, "loss": 0.0155, "step": 261860 }, { "epoch": 166.90248565965584, "grad_norm": 0.09643613547086716, "learning_rate": 4.352141733305015e-06, "loss": 0.0087, "step": 261870 }, { "epoch": 166.90885914595285, "grad_norm": 0.27223777770996094, "learning_rate": 4.349892986316656e-06, "loss": 0.0088, "step": 261880 }, { "epoch": 166.91523263224985, "grad_norm": 0.438012957572937, "learning_rate": 4.347644794020217e-06, "loss": 0.0068, "step": 261890 }, { "epoch": 166.92160611854683, "grad_norm": 0.32441943883895874, "learning_rate": 4.345397156442976e-06, "loss": 0.0061, "step": 261900 }, { "epoch": 166.92797960484384, "grad_norm": 0.39061421155929565, "learning_rate": 4.343150073612273e-06, "loss": 0.0037, "step": 261910 }, { "epoch": 166.93435309114085, "grad_norm": 0.4426162540912628, "learning_rate": 4.340903545555403e-06, "loss": 0.0068, "step": 261920 }, { "epoch": 166.94072657743786, "grad_norm": 0.2742319107055664, "learning_rate": 4.3386575722996755e-06, "loss": 0.0035, "step": 261930 }, { "epoch": 166.94710006373487, "grad_norm": 0.3411758840084076, "learning_rate": 4.336412153872354e-06, "loss": 0.0139, "step": 261940 }, { "epoch": 166.95347355003187, "grad_norm": 0.06932718306779861, "learning_rate": 4.334167290300745e-06, "loss": 0.0051, "step": 261950 }, { "epoch": 166.95984703632888, "grad_norm": 0.1705956608057022, "learning_rate": 4.331922981612119e-06, "loss": 0.0068, "step": 261960 }, { "epoch": 166.9662205226259, "grad_norm": 0.0628056600689888, "learning_rate": 4.329679227833749e-06, "loss": 0.0035, "step": 261970 }, { "epoch": 166.97259400892287, "grad_norm": 0.3487556278705597, "learning_rate": 4.327436028992887e-06, "loss": 0.0048, "step": 261980 }, { "epoch": 166.97896749521988, "grad_norm": 0.12791483104228973, "learning_rate": 4.325193385116811e-06, "loss": 0.0038, "step": 261990 }, { "epoch": 166.98534098151688, "grad_norm": 0.4070602059364319, "learning_rate": 4.322951296232752e-06, "loss": 0.0077, "step": 262000 }, { "epoch": 166.9917144678139, "grad_norm": 0.27826645970344543, "learning_rate": 4.320709762367969e-06, "loss": 0.0111, "step": 262010 }, { "epoch": 166.9980879541109, "grad_norm": 0.188939169049263, "learning_rate": 4.318468783549679e-06, "loss": 0.0096, "step": 262020 }, { "epoch": 167.0044614404079, "grad_norm": 0.07763893157243729, "learning_rate": 4.31622835980513e-06, "loss": 0.0069, "step": 262030 }, { "epoch": 167.01083492670492, "grad_norm": 0.272368460893631, "learning_rate": 4.3139884911615414e-06, "loss": 0.0062, "step": 262040 }, { "epoch": 167.01720841300192, "grad_norm": 0.13531626760959625, "learning_rate": 4.311749177646129e-06, "loss": 0.0051, "step": 262050 }, { "epoch": 167.0235818992989, "grad_norm": 0.23607371747493744, "learning_rate": 4.309510419286089e-06, "loss": 0.0081, "step": 262060 }, { "epoch": 167.0299553855959, "grad_norm": 0.6824766993522644, "learning_rate": 4.307272216108654e-06, "loss": 0.0062, "step": 262070 }, { "epoch": 167.03632887189292, "grad_norm": 0.3305480480194092, "learning_rate": 4.305034568140986e-06, "loss": 0.0049, "step": 262080 }, { "epoch": 167.04270235818993, "grad_norm": 0.5475334525108337, "learning_rate": 4.302797475410303e-06, "loss": 0.0105, "step": 262090 }, { "epoch": 167.04907584448694, "grad_norm": 0.16314490139484406, "learning_rate": 4.300560937943759e-06, "loss": 0.0069, "step": 262100 }, { "epoch": 167.05544933078394, "grad_norm": 0.21804498136043549, "learning_rate": 4.298324955768573e-06, "loss": 0.0037, "step": 262110 }, { "epoch": 167.06182281708095, "grad_norm": 0.09037791192531586, "learning_rate": 4.296089528911868e-06, "loss": 0.0065, "step": 262120 }, { "epoch": 167.06819630337796, "grad_norm": 0.10970576852560043, "learning_rate": 4.293854657400836e-06, "loss": 0.0037, "step": 262130 }, { "epoch": 167.07456978967494, "grad_norm": 0.29297828674316406, "learning_rate": 4.291620341262626e-06, "loss": 0.0062, "step": 262140 }, { "epoch": 167.08094327597195, "grad_norm": 0.20597724616527557, "learning_rate": 4.2893865805243794e-06, "loss": 0.0064, "step": 262150 }, { "epoch": 167.08731676226896, "grad_norm": 0.08990170806646347, "learning_rate": 4.287153375213237e-06, "loss": 0.0061, "step": 262160 }, { "epoch": 167.09369024856596, "grad_norm": 0.563239574432373, "learning_rate": 4.284920725356351e-06, "loss": 0.0068, "step": 262170 }, { "epoch": 167.10006373486297, "grad_norm": 0.2656678557395935, "learning_rate": 4.2826886309808414e-06, "loss": 0.0082, "step": 262180 }, { "epoch": 167.10643722115998, "grad_norm": 0.15977084636688232, "learning_rate": 4.2804570921138265e-06, "loss": 0.0057, "step": 262190 }, { "epoch": 167.112810707457, "grad_norm": 0.09200692921876907, "learning_rate": 4.278226108782413e-06, "loss": 0.004, "step": 262200 }, { "epoch": 167.119184193754, "grad_norm": 0.4475238621234894, "learning_rate": 4.275995681013745e-06, "loss": 0.0066, "step": 262210 }, { "epoch": 167.12555768005097, "grad_norm": 0.3179242014884949, "learning_rate": 4.273765808834879e-06, "loss": 0.0029, "step": 262220 }, { "epoch": 167.13193116634798, "grad_norm": 0.18794633448123932, "learning_rate": 4.271536492272943e-06, "loss": 0.0082, "step": 262230 }, { "epoch": 167.138304652645, "grad_norm": 0.07375312596559525, "learning_rate": 4.269307731354999e-06, "loss": 0.0072, "step": 262240 }, { "epoch": 167.144678138942, "grad_norm": 0.4664427936077118, "learning_rate": 4.267079526108164e-06, "loss": 0.0115, "step": 262250 }, { "epoch": 167.151051625239, "grad_norm": 0.2568111717700958, "learning_rate": 4.264851876559473e-06, "loss": 0.0118, "step": 262260 }, { "epoch": 167.157425111536, "grad_norm": 0.16901959478855133, "learning_rate": 4.262624782736024e-06, "loss": 0.0028, "step": 262270 }, { "epoch": 167.16379859783302, "grad_norm": 0.17742820084095, "learning_rate": 4.260398244664865e-06, "loss": 0.0081, "step": 262280 }, { "epoch": 167.17017208413003, "grad_norm": 0.07638127356767654, "learning_rate": 4.258172262373056e-06, "loss": 0.0059, "step": 262290 }, { "epoch": 167.176545570427, "grad_norm": 0.28439924120903015, "learning_rate": 4.255946835887625e-06, "loss": 0.0056, "step": 262300 }, { "epoch": 167.18291905672402, "grad_norm": 0.14530442655086517, "learning_rate": 4.253721965235647e-06, "loss": 0.01, "step": 262310 }, { "epoch": 167.18929254302103, "grad_norm": 0.24897563457489014, "learning_rate": 4.251497650444136e-06, "loss": 0.0107, "step": 262320 }, { "epoch": 167.19566602931803, "grad_norm": 0.3481956124305725, "learning_rate": 4.249273891540118e-06, "loss": 0.0045, "step": 262330 }, { "epoch": 167.20203951561504, "grad_norm": 0.34355130791664124, "learning_rate": 4.247050688550613e-06, "loss": 0.0061, "step": 262340 }, { "epoch": 167.20841300191205, "grad_norm": 0.08291018009185791, "learning_rate": 4.244828041502652e-06, "loss": 0.0081, "step": 262350 }, { "epoch": 167.21478648820906, "grad_norm": 0.05863234028220177, "learning_rate": 4.242605950423223e-06, "loss": 0.0113, "step": 262360 }, { "epoch": 167.22115997450607, "grad_norm": 0.30612874031066895, "learning_rate": 4.240384415339338e-06, "loss": 0.0044, "step": 262370 }, { "epoch": 167.22753346080307, "grad_norm": 0.08788321912288666, "learning_rate": 4.238163436277981e-06, "loss": 0.0084, "step": 262380 }, { "epoch": 167.23390694710005, "grad_norm": 0.10845717042684555, "learning_rate": 4.235943013266147e-06, "loss": 0.003, "step": 262390 }, { "epoch": 167.24028043339706, "grad_norm": 0.16559560596942902, "learning_rate": 4.233723146330814e-06, "loss": 0.0065, "step": 262400 }, { "epoch": 167.24665391969407, "grad_norm": 0.11365080624818802, "learning_rate": 4.231503835498951e-06, "loss": 0.0048, "step": 262410 }, { "epoch": 167.25302740599108, "grad_norm": 0.13516530394554138, "learning_rate": 4.2292850807975344e-06, "loss": 0.0044, "step": 262420 }, { "epoch": 167.25940089228808, "grad_norm": 0.2789587676525116, "learning_rate": 4.227066882253522e-06, "loss": 0.0057, "step": 262430 }, { "epoch": 167.2657743785851, "grad_norm": 0.30834364891052246, "learning_rate": 4.224849239893863e-06, "loss": 0.004, "step": 262440 }, { "epoch": 167.2721478648821, "grad_norm": 0.10820165276527405, "learning_rate": 4.222632153745492e-06, "loss": 0.0037, "step": 262450 }, { "epoch": 167.2785213511791, "grad_norm": 0.3448079228401184, "learning_rate": 4.220415623835383e-06, "loss": 0.0049, "step": 262460 }, { "epoch": 167.2848948374761, "grad_norm": 0.09047827869653702, "learning_rate": 4.21819965019043e-06, "loss": 0.0051, "step": 262470 }, { "epoch": 167.2912683237731, "grad_norm": 0.41786015033721924, "learning_rate": 4.215984232837583e-06, "loss": 0.0104, "step": 262480 }, { "epoch": 167.2976418100701, "grad_norm": 0.05897887051105499, "learning_rate": 4.21376937180375e-06, "loss": 0.0044, "step": 262490 }, { "epoch": 167.3040152963671, "grad_norm": 0.0940186157822609, "learning_rate": 4.211555067115863e-06, "loss": 0.0045, "step": 262500 }, { "epoch": 167.31038878266412, "grad_norm": 0.3523818254470825, "learning_rate": 4.209341318800797e-06, "loss": 0.0076, "step": 262510 }, { "epoch": 167.31676226896113, "grad_norm": 0.11883025616407394, "learning_rate": 4.2071281268854735e-06, "loss": 0.0102, "step": 262520 }, { "epoch": 167.32313575525814, "grad_norm": 0.08769207447767258, "learning_rate": 4.204915491396777e-06, "loss": 0.0056, "step": 262530 }, { "epoch": 167.32950924155514, "grad_norm": 0.0686459019780159, "learning_rate": 4.202703412361602e-06, "loss": 0.0039, "step": 262540 }, { "epoch": 167.33588272785212, "grad_norm": 0.15308921039104462, "learning_rate": 4.200491889806807e-06, "loss": 0.0038, "step": 262550 }, { "epoch": 167.34225621414913, "grad_norm": 0.17835794389247894, "learning_rate": 4.198280923759284e-06, "loss": 0.0052, "step": 262560 }, { "epoch": 167.34862970044614, "grad_norm": 0.06277570873498917, "learning_rate": 4.196070514245898e-06, "loss": 0.0056, "step": 262570 }, { "epoch": 167.35500318674315, "grad_norm": 0.12601397931575775, "learning_rate": 4.193860661293497e-06, "loss": 0.0044, "step": 262580 }, { "epoch": 167.36137667304016, "grad_norm": 0.09370177984237671, "learning_rate": 4.191651364928934e-06, "loss": 0.0059, "step": 262590 }, { "epoch": 167.36775015933716, "grad_norm": 0.06639765202999115, "learning_rate": 4.189442625179058e-06, "loss": 0.0036, "step": 262600 }, { "epoch": 167.37412364563417, "grad_norm": 0.11527746170759201, "learning_rate": 4.187234442070709e-06, "loss": 0.0064, "step": 262610 }, { "epoch": 167.38049713193118, "grad_norm": 0.1851462721824646, "learning_rate": 4.185026815630721e-06, "loss": 0.0056, "step": 262620 }, { "epoch": 167.38687061822816, "grad_norm": 0.1837199479341507, "learning_rate": 4.1828197458858965e-06, "loss": 0.0048, "step": 262630 }, { "epoch": 167.39324410452517, "grad_norm": 0.15807466208934784, "learning_rate": 4.180613232863095e-06, "loss": 0.0065, "step": 262640 }, { "epoch": 167.39961759082217, "grad_norm": 0.2642914056777954, "learning_rate": 4.178407276589086e-06, "loss": 0.0062, "step": 262650 }, { "epoch": 167.40599107711918, "grad_norm": 0.22658033668994904, "learning_rate": 4.176201877090696e-06, "loss": 0.0038, "step": 262660 }, { "epoch": 167.4123645634162, "grad_norm": 0.06699913740158081, "learning_rate": 4.173997034394711e-06, "loss": 0.0036, "step": 262670 }, { "epoch": 167.4187380497132, "grad_norm": 0.3525974750518799, "learning_rate": 4.171792748527947e-06, "loss": 0.0063, "step": 262680 }, { "epoch": 167.4251115360102, "grad_norm": 0.06690289825201035, "learning_rate": 4.169589019517151e-06, "loss": 0.0028, "step": 262690 }, { "epoch": 167.4314850223072, "grad_norm": 0.20788182318210602, "learning_rate": 4.167385847389127e-06, "loss": 0.0035, "step": 262700 }, { "epoch": 167.4378585086042, "grad_norm": 0.16964253783226013, "learning_rate": 4.165183232170644e-06, "loss": 0.0044, "step": 262710 }, { "epoch": 167.4442319949012, "grad_norm": 0.19307322800159454, "learning_rate": 4.162981173888453e-06, "loss": 0.0087, "step": 262720 }, { "epoch": 167.4506054811982, "grad_norm": 0.15460516512393951, "learning_rate": 4.1607796725693146e-06, "loss": 0.0032, "step": 262730 }, { "epoch": 167.45697896749522, "grad_norm": 0.09549213200807571, "learning_rate": 4.158578728239987e-06, "loss": 0.0115, "step": 262740 }, { "epoch": 167.46335245379223, "grad_norm": 0.1747528463602066, "learning_rate": 4.156378340927214e-06, "loss": 0.0093, "step": 262750 }, { "epoch": 167.46972594008923, "grad_norm": 0.2244153469800949, "learning_rate": 4.154178510657725e-06, "loss": 0.0082, "step": 262760 }, { "epoch": 167.47609942638624, "grad_norm": 0.4032008945941925, "learning_rate": 4.1519792374582425e-06, "loss": 0.0049, "step": 262770 }, { "epoch": 167.48247291268325, "grad_norm": 0.12310336530208588, "learning_rate": 4.149780521355523e-06, "loss": 0.0069, "step": 262780 }, { "epoch": 167.48884639898023, "grad_norm": 0.3271171450614929, "learning_rate": 4.147582362376234e-06, "loss": 0.0052, "step": 262790 }, { "epoch": 167.49521988527724, "grad_norm": 0.23015275597572327, "learning_rate": 4.145384760547127e-06, "loss": 0.0134, "step": 262800 }, { "epoch": 167.50159337157424, "grad_norm": 0.1127963438630104, "learning_rate": 4.143187715894881e-06, "loss": 0.0035, "step": 262810 }, { "epoch": 167.50796685787125, "grad_norm": 0.11286415159702301, "learning_rate": 4.140991228446212e-06, "loss": 0.0067, "step": 262820 }, { "epoch": 167.51434034416826, "grad_norm": 0.14211130142211914, "learning_rate": 4.138795298227782e-06, "loss": 0.0088, "step": 262830 }, { "epoch": 167.52071383046527, "grad_norm": 0.19997185468673706, "learning_rate": 4.136599925266299e-06, "loss": 0.0077, "step": 262840 }, { "epoch": 167.52708731676228, "grad_norm": 0.12798044085502625, "learning_rate": 4.13440510958843e-06, "loss": 0.0072, "step": 262850 }, { "epoch": 167.53346080305928, "grad_norm": 0.06935226917266846, "learning_rate": 4.1322108512208415e-06, "loss": 0.0157, "step": 262860 }, { "epoch": 167.53983428935626, "grad_norm": 0.19314153492450714, "learning_rate": 4.130017150190191e-06, "loss": 0.004, "step": 262870 }, { "epoch": 167.54620777565327, "grad_norm": 0.24568234384059906, "learning_rate": 4.127824006523146e-06, "loss": 0.0048, "step": 262880 }, { "epoch": 167.55258126195028, "grad_norm": 0.06863165646791458, "learning_rate": 4.125631420246351e-06, "loss": 0.0018, "step": 262890 }, { "epoch": 167.5589547482473, "grad_norm": 0.07183882594108582, "learning_rate": 4.1234393913864525e-06, "loss": 0.0022, "step": 262900 }, { "epoch": 167.5653282345443, "grad_norm": 0.21690334379673004, "learning_rate": 4.121247919970061e-06, "loss": 0.0054, "step": 262910 }, { "epoch": 167.5717017208413, "grad_norm": 0.2778772711753845, "learning_rate": 4.119057006023841e-06, "loss": 0.0045, "step": 262920 }, { "epoch": 167.5780752071383, "grad_norm": 0.14633075892925262, "learning_rate": 4.116866649574397e-06, "loss": 0.0049, "step": 262930 }, { "epoch": 167.58444869343532, "grad_norm": 0.09882068634033203, "learning_rate": 4.114676850648335e-06, "loss": 0.0148, "step": 262940 }, { "epoch": 167.5908221797323, "grad_norm": 0.07859547436237335, "learning_rate": 4.112487609272276e-06, "loss": 0.0083, "step": 262950 }, { "epoch": 167.5971956660293, "grad_norm": 0.4076709747314453, "learning_rate": 4.110298925472828e-06, "loss": 0.0065, "step": 262960 }, { "epoch": 167.60356915232632, "grad_norm": 0.07110391557216644, "learning_rate": 4.108110799276554e-06, "loss": 0.0055, "step": 262970 }, { "epoch": 167.60994263862332, "grad_norm": 0.146065354347229, "learning_rate": 4.105923230710079e-06, "loss": 0.0066, "step": 262980 }, { "epoch": 167.61631612492033, "grad_norm": 0.057653557509183884, "learning_rate": 4.103736219799959e-06, "loss": 0.0031, "step": 262990 }, { "epoch": 167.62268961121734, "grad_norm": 0.23761601746082306, "learning_rate": 4.101549766572782e-06, "loss": 0.0061, "step": 263000 }, { "epoch": 167.62906309751435, "grad_norm": 0.32177427411079407, "learning_rate": 4.099363871055112e-06, "loss": 0.0122, "step": 263010 }, { "epoch": 167.63543658381136, "grad_norm": 0.17703378200531006, "learning_rate": 4.097178533273505e-06, "loss": 0.0107, "step": 263020 }, { "epoch": 167.64181007010836, "grad_norm": 0.1296294778585434, "learning_rate": 4.094993753254523e-06, "loss": 0.008, "step": 263030 }, { "epoch": 167.64818355640534, "grad_norm": 0.10786022245883942, "learning_rate": 4.0928095310246994e-06, "loss": 0.0039, "step": 263040 }, { "epoch": 167.65455704270235, "grad_norm": 0.3110397160053253, "learning_rate": 4.090625866610587e-06, "loss": 0.0031, "step": 263050 }, { "epoch": 167.66093052899936, "grad_norm": 0.5292169451713562, "learning_rate": 4.0884427600387244e-06, "loss": 0.0064, "step": 263060 }, { "epoch": 167.66730401529637, "grad_norm": 0.3552835285663605, "learning_rate": 4.086260211335624e-06, "loss": 0.0079, "step": 263070 }, { "epoch": 167.67367750159337, "grad_norm": 0.14786000549793243, "learning_rate": 4.084078220527804e-06, "loss": 0.0034, "step": 263080 }, { "epoch": 167.68005098789038, "grad_norm": 0.3935660123825073, "learning_rate": 4.081896787641793e-06, "loss": 0.0065, "step": 263090 }, { "epoch": 167.6864244741874, "grad_norm": 0.22720199823379517, "learning_rate": 4.079715912704096e-06, "loss": 0.0034, "step": 263100 }, { "epoch": 167.6927979604844, "grad_norm": 0.07402964681386948, "learning_rate": 4.0775355957411995e-06, "loss": 0.0052, "step": 263110 }, { "epoch": 167.69917144678138, "grad_norm": 0.14036816358566284, "learning_rate": 4.075355836779598e-06, "loss": 0.0041, "step": 263120 }, { "epoch": 167.7055449330784, "grad_norm": 0.10135030001401901, "learning_rate": 4.073176635845793e-06, "loss": 0.0137, "step": 263130 }, { "epoch": 167.7119184193754, "grad_norm": 0.3767858147621155, "learning_rate": 4.070997992966252e-06, "loss": 0.0079, "step": 263140 }, { "epoch": 167.7182919056724, "grad_norm": 0.07786650955677032, "learning_rate": 4.068819908167448e-06, "loss": 0.0029, "step": 263150 }, { "epoch": 167.7246653919694, "grad_norm": 0.16552364826202393, "learning_rate": 4.066642381475843e-06, "loss": 0.0026, "step": 263160 }, { "epoch": 167.73103887826642, "grad_norm": 0.08384661376476288, "learning_rate": 4.064465412917917e-06, "loss": 0.0063, "step": 263170 }, { "epoch": 167.73741236456343, "grad_norm": 0.42444920539855957, "learning_rate": 4.0622890025200865e-06, "loss": 0.0059, "step": 263180 }, { "epoch": 167.74378585086043, "grad_norm": 0.1502867043018341, "learning_rate": 4.060113150308831e-06, "loss": 0.0096, "step": 263190 }, { "epoch": 167.7501593371574, "grad_norm": 0.0741758644580841, "learning_rate": 4.057937856310562e-06, "loss": 0.006, "step": 263200 }, { "epoch": 167.75653282345442, "grad_norm": 0.19389455020427704, "learning_rate": 4.055763120551747e-06, "loss": 0.0074, "step": 263210 }, { "epoch": 167.76290630975143, "grad_norm": 0.09525268524885178, "learning_rate": 4.053588943058767e-06, "loss": 0.008, "step": 263220 }, { "epoch": 167.76927979604844, "grad_norm": 0.08976887166500092, "learning_rate": 4.051415323858071e-06, "loss": 0.0149, "step": 263230 }, { "epoch": 167.77565328234544, "grad_norm": 0.07497818022966385, "learning_rate": 4.049242262976055e-06, "loss": 0.0052, "step": 263240 }, { "epoch": 167.78202676864245, "grad_norm": 0.11827412992715836, "learning_rate": 4.047069760439149e-06, "loss": 0.0074, "step": 263250 }, { "epoch": 167.78840025493946, "grad_norm": 0.2430357038974762, "learning_rate": 4.044897816273707e-06, "loss": 0.0048, "step": 263260 }, { "epoch": 167.79477374123647, "grad_norm": 0.47984758019447327, "learning_rate": 4.0427264305061604e-06, "loss": 0.0145, "step": 263270 }, { "epoch": 167.80114722753345, "grad_norm": 0.2218986302614212, "learning_rate": 4.040555603162877e-06, "loss": 0.003, "step": 263280 }, { "epoch": 167.80752071383046, "grad_norm": 0.21813686192035675, "learning_rate": 4.03838533427024e-06, "loss": 0.0061, "step": 263290 }, { "epoch": 167.81389420012746, "grad_norm": 0.1043848991394043, "learning_rate": 4.0362156238546e-06, "loss": 0.0035, "step": 263300 }, { "epoch": 167.82026768642447, "grad_norm": 0.24538618326187134, "learning_rate": 4.034046471942349e-06, "loss": 0.0056, "step": 263310 }, { "epoch": 167.82664117272148, "grad_norm": 0.30349960923194885, "learning_rate": 4.031877878559836e-06, "loss": 0.0063, "step": 263320 }, { "epoch": 167.8330146590185, "grad_norm": 0.09505973756313324, "learning_rate": 4.0297098437334034e-06, "loss": 0.014, "step": 263330 }, { "epoch": 167.8393881453155, "grad_norm": 0.18451233208179474, "learning_rate": 4.027542367489395e-06, "loss": 0.0042, "step": 263340 }, { "epoch": 167.8457616316125, "grad_norm": 0.12347818166017532, "learning_rate": 4.025375449854163e-06, "loss": 0.0034, "step": 263350 }, { "epoch": 167.85213511790948, "grad_norm": 0.08797676861286163, "learning_rate": 4.023209090854013e-06, "loss": 0.0047, "step": 263360 }, { "epoch": 167.8585086042065, "grad_norm": 0.06785319745540619, "learning_rate": 4.021043290515292e-06, "loss": 0.0059, "step": 263370 }, { "epoch": 167.8648820905035, "grad_norm": 0.15170937776565552, "learning_rate": 4.018878048864294e-06, "loss": 0.0075, "step": 263380 }, { "epoch": 167.8712555768005, "grad_norm": 0.10825737565755844, "learning_rate": 4.016713365927361e-06, "loss": 0.0041, "step": 263390 }, { "epoch": 167.87762906309752, "grad_norm": 0.05993680655956268, "learning_rate": 4.014549241730759e-06, "loss": 0.0038, "step": 263400 }, { "epoch": 167.88400254939452, "grad_norm": 0.35140693187713623, "learning_rate": 4.012385676300806e-06, "loss": 0.006, "step": 263410 }, { "epoch": 167.89037603569153, "grad_norm": 0.15561021864414215, "learning_rate": 4.010222669663794e-06, "loss": 0.0042, "step": 263420 }, { "epoch": 167.89674952198854, "grad_norm": 0.20872017741203308, "learning_rate": 4.008060221845989e-06, "loss": 0.0031, "step": 263430 }, { "epoch": 167.90312300828552, "grad_norm": 0.2728160321712494, "learning_rate": 4.005898332873675e-06, "loss": 0.0119, "step": 263440 }, { "epoch": 167.90949649458253, "grad_norm": 0.26508939266204834, "learning_rate": 4.003737002773128e-06, "loss": 0.0049, "step": 263450 }, { "epoch": 167.91586998087953, "grad_norm": 0.890498697757721, "learning_rate": 4.001576231570603e-06, "loss": 0.0121, "step": 263460 }, { "epoch": 167.92224346717654, "grad_norm": 0.06399485468864441, "learning_rate": 3.999416019292357e-06, "loss": 0.0087, "step": 263470 }, { "epoch": 167.92861695347355, "grad_norm": 0.3429470658302307, "learning_rate": 3.997256365964625e-06, "loss": 0.0072, "step": 263480 }, { "epoch": 167.93499043977056, "grad_norm": 0.4658019244670868, "learning_rate": 3.995097271613685e-06, "loss": 0.0062, "step": 263490 }, { "epoch": 167.94136392606757, "grad_norm": 0.15588681399822235, "learning_rate": 3.992938736265728e-06, "loss": 0.0019, "step": 263500 }, { "epoch": 167.94773741236457, "grad_norm": 0.07771764695644379, "learning_rate": 3.990780759947016e-06, "loss": 0.0061, "step": 263510 }, { "epoch": 167.95411089866155, "grad_norm": 0.1509934365749359, "learning_rate": 3.988623342683745e-06, "loss": 0.0038, "step": 263520 }, { "epoch": 167.96048438495856, "grad_norm": 0.10708960145711899, "learning_rate": 3.986466484502166e-06, "loss": 0.0026, "step": 263530 }, { "epoch": 167.96685787125557, "grad_norm": 0.16425931453704834, "learning_rate": 3.984310185428441e-06, "loss": 0.0046, "step": 263540 }, { "epoch": 167.97323135755258, "grad_norm": 0.11623923480510712, "learning_rate": 3.9821544454888e-06, "loss": 0.0026, "step": 263550 }, { "epoch": 167.9796048438496, "grad_norm": 0.0544259138405323, "learning_rate": 3.979999264709427e-06, "loss": 0.0047, "step": 263560 }, { "epoch": 167.9859783301466, "grad_norm": 0.11400673538446426, "learning_rate": 3.977844643116524e-06, "loss": 0.0028, "step": 263570 }, { "epoch": 167.9923518164436, "grad_norm": 0.09135282784700394, "learning_rate": 3.975690580736252e-06, "loss": 0.009, "step": 263580 }, { "epoch": 167.9987253027406, "grad_norm": 0.16847291588783264, "learning_rate": 3.9735370775947966e-06, "loss": 0.005, "step": 263590 }, { "epoch": 168.00509878903762, "grad_norm": 0.2526135742664337, "learning_rate": 3.97138413371832e-06, "loss": 0.0133, "step": 263600 }, { "epoch": 168.0114722753346, "grad_norm": 0.8064946532249451, "learning_rate": 3.96923174913299e-06, "loss": 0.0061, "step": 263610 }, { "epoch": 168.0178457616316, "grad_norm": 0.7061338424682617, "learning_rate": 3.9670799238649416e-06, "loss": 0.0046, "step": 263620 }, { "epoch": 168.0242192479286, "grad_norm": 0.1293260157108307, "learning_rate": 3.964928657940342e-06, "loss": 0.005, "step": 263630 }, { "epoch": 168.03059273422562, "grad_norm": 0.3052131235599518, "learning_rate": 3.962777951385327e-06, "loss": 0.0043, "step": 263640 }, { "epoch": 168.03696622052263, "grad_norm": 0.1070639044046402, "learning_rate": 3.960627804226025e-06, "loss": 0.0064, "step": 263650 }, { "epoch": 168.04333970681964, "grad_norm": 0.09118987619876862, "learning_rate": 3.958478216488553e-06, "loss": 0.0053, "step": 263660 }, { "epoch": 168.04971319311665, "grad_norm": 0.13739557564258575, "learning_rate": 3.956329188199054e-06, "loss": 0.0158, "step": 263670 }, { "epoch": 168.05608667941365, "grad_norm": 0.0654212236404419, "learning_rate": 3.9541807193836265e-06, "loss": 0.0028, "step": 263680 }, { "epoch": 168.06246016571063, "grad_norm": 0.0635119080543518, "learning_rate": 3.952032810068373e-06, "loss": 0.0036, "step": 263690 }, { "epoch": 168.06883365200764, "grad_norm": 0.18913985788822174, "learning_rate": 3.949885460279396e-06, "loss": 0.0111, "step": 263700 }, { "epoch": 168.07520713830465, "grad_norm": 0.30679070949554443, "learning_rate": 3.947738670042794e-06, "loss": 0.0037, "step": 263710 }, { "epoch": 168.08158062460166, "grad_norm": 0.32915425300598145, "learning_rate": 3.945592439384649e-06, "loss": 0.0167, "step": 263720 }, { "epoch": 168.08795411089866, "grad_norm": 0.3985632061958313, "learning_rate": 3.943446768331033e-06, "loss": 0.0162, "step": 263730 }, { "epoch": 168.09432759719567, "grad_norm": 0.178201362490654, "learning_rate": 3.941301656908031e-06, "loss": 0.0046, "step": 263740 }, { "epoch": 168.10070108349268, "grad_norm": 0.5113154053688049, "learning_rate": 3.939157105141688e-06, "loss": 0.0141, "step": 263750 }, { "epoch": 168.1070745697897, "grad_norm": 0.26940926909446716, "learning_rate": 3.9370131130580855e-06, "loss": 0.0065, "step": 263760 }, { "epoch": 168.11344805608667, "grad_norm": 0.18365715444087982, "learning_rate": 3.934869680683262e-06, "loss": 0.0049, "step": 263770 }, { "epoch": 168.11982154238368, "grad_norm": 0.14877036213874817, "learning_rate": 3.932726808043263e-06, "loss": 0.0037, "step": 263780 }, { "epoch": 168.12619502868068, "grad_norm": 0.3661879897117615, "learning_rate": 3.930584495164125e-06, "loss": 0.0149, "step": 263790 }, { "epoch": 168.1325685149777, "grad_norm": 0.3100060224533081, "learning_rate": 3.928442742071886e-06, "loss": 0.0049, "step": 263800 }, { "epoch": 168.1389420012747, "grad_norm": 1.0656485557556152, "learning_rate": 3.9263015487925654e-06, "loss": 0.0046, "step": 263810 }, { "epoch": 168.1453154875717, "grad_norm": 0.4754037857055664, "learning_rate": 3.924160915352187e-06, "loss": 0.0063, "step": 263820 }, { "epoch": 168.15168897386872, "grad_norm": 0.1243157684803009, "learning_rate": 3.9220208417767466e-06, "loss": 0.0072, "step": 263830 }, { "epoch": 168.15806246016572, "grad_norm": 0.07905219495296478, "learning_rate": 3.919881328092262e-06, "loss": 0.0038, "step": 263840 }, { "epoch": 168.1644359464627, "grad_norm": 0.1211061105132103, "learning_rate": 3.917742374324729e-06, "loss": 0.0039, "step": 263850 }, { "epoch": 168.1708094327597, "grad_norm": 0.39989173412323, "learning_rate": 3.915603980500137e-06, "loss": 0.0028, "step": 263860 }, { "epoch": 168.17718291905672, "grad_norm": 0.08441076427698135, "learning_rate": 3.913466146644456e-06, "loss": 0.0026, "step": 263870 }, { "epoch": 168.18355640535373, "grad_norm": 0.4184037148952484, "learning_rate": 3.9113288727836815e-06, "loss": 0.0045, "step": 263880 }, { "epoch": 168.18992989165073, "grad_norm": 0.10345180332660675, "learning_rate": 3.9091921589437805e-06, "loss": 0.0042, "step": 263890 }, { "epoch": 168.19630337794774, "grad_norm": 0.1250694990158081, "learning_rate": 3.907056005150711e-06, "loss": 0.0071, "step": 263900 }, { "epoch": 168.20267686424475, "grad_norm": 0.10828761756420135, "learning_rate": 3.904920411430418e-06, "loss": 0.0128, "step": 263910 }, { "epoch": 168.20905035054176, "grad_norm": 0.17185744643211365, "learning_rate": 3.902785377808882e-06, "loss": 0.0037, "step": 263920 }, { "epoch": 168.21542383683874, "grad_norm": 0.1574988216161728, "learning_rate": 3.900650904312008e-06, "loss": 0.0045, "step": 263930 }, { "epoch": 168.22179732313575, "grad_norm": 0.06669218838214874, "learning_rate": 3.898516990965762e-06, "loss": 0.0028, "step": 263940 }, { "epoch": 168.22817080943275, "grad_norm": 0.09533914923667908, "learning_rate": 3.896383637796053e-06, "loss": 0.0104, "step": 263950 }, { "epoch": 168.23454429572976, "grad_norm": 0.36536073684692383, "learning_rate": 3.894250844828823e-06, "loss": 0.0135, "step": 263960 }, { "epoch": 168.24091778202677, "grad_norm": 0.5843457579612732, "learning_rate": 3.892118612089968e-06, "loss": 0.0127, "step": 263970 }, { "epoch": 168.24729126832378, "grad_norm": 0.09300880134105682, "learning_rate": 3.889986939605406e-06, "loss": 0.0051, "step": 263980 }, { "epoch": 168.2536647546208, "grad_norm": 0.4012986421585083, "learning_rate": 3.8878558274010435e-06, "loss": 0.0042, "step": 263990 }, { "epoch": 168.2600382409178, "grad_norm": 0.11414463818073273, "learning_rate": 3.885725275502766e-06, "loss": 0.0066, "step": 264000 }, { "epoch": 168.26641172721477, "grad_norm": 0.2913650572299957, "learning_rate": 3.883595283936459e-06, "loss": 0.0049, "step": 264010 }, { "epoch": 168.27278521351178, "grad_norm": 0.12095089256763458, "learning_rate": 3.881465852728017e-06, "loss": 0.0038, "step": 264020 }, { "epoch": 168.2791586998088, "grad_norm": 0.10482481867074966, "learning_rate": 3.879336981903309e-06, "loss": 0.0048, "step": 264030 }, { "epoch": 168.2855321861058, "grad_norm": 0.10426764190196991, "learning_rate": 3.877208671488202e-06, "loss": 0.0045, "step": 264040 }, { "epoch": 168.2919056724028, "grad_norm": 0.11138808727264404, "learning_rate": 3.87508092150855e-06, "loss": 0.0088, "step": 264050 }, { "epoch": 168.2982791586998, "grad_norm": 0.09660626202821732, "learning_rate": 3.8729537319902256e-06, "loss": 0.0092, "step": 264060 }, { "epoch": 168.30465264499682, "grad_norm": 0.1470949798822403, "learning_rate": 3.870827102959052e-06, "loss": 0.0049, "step": 264070 }, { "epoch": 168.31102613129383, "grad_norm": 0.2336309254169464, "learning_rate": 3.868701034440892e-06, "loss": 0.0056, "step": 264080 }, { "epoch": 168.3173996175908, "grad_norm": 0.10281974822282791, "learning_rate": 3.8665755264615575e-06, "loss": 0.0045, "step": 264090 }, { "epoch": 168.32377310388782, "grad_norm": 0.4022073745727539, "learning_rate": 3.864450579046907e-06, "loss": 0.0085, "step": 264100 }, { "epoch": 168.33014659018482, "grad_norm": 0.6880523562431335, "learning_rate": 3.86232619222272e-06, "loss": 0.0081, "step": 264110 }, { "epoch": 168.33652007648183, "grad_norm": 0.18753713369369507, "learning_rate": 3.860202366014842e-06, "loss": 0.0061, "step": 264120 }, { "epoch": 168.34289356277884, "grad_norm": 0.13584180176258087, "learning_rate": 3.858079100449063e-06, "loss": 0.006, "step": 264130 }, { "epoch": 168.34926704907585, "grad_norm": 0.09495005756616592, "learning_rate": 3.855956395551202e-06, "loss": 0.0039, "step": 264140 }, { "epoch": 168.35564053537286, "grad_norm": 0.33298519253730774, "learning_rate": 3.853834251347022e-06, "loss": 0.0075, "step": 264150 }, { "epoch": 168.36201402166986, "grad_norm": 0.13853074610233307, "learning_rate": 3.851712667862334e-06, "loss": 0.0042, "step": 264160 }, { "epoch": 168.36838750796684, "grad_norm": 0.40884238481521606, "learning_rate": 3.8495916451229075e-06, "loss": 0.0055, "step": 264170 }, { "epoch": 168.37476099426385, "grad_norm": 0.12152043730020523, "learning_rate": 3.847471183154516e-06, "loss": 0.003, "step": 264180 }, { "epoch": 168.38113448056086, "grad_norm": 0.1428890973329544, "learning_rate": 3.845351281982912e-06, "loss": 0.0033, "step": 264190 }, { "epoch": 168.38750796685787, "grad_norm": 0.210234597325325, "learning_rate": 3.843231941633879e-06, "loss": 0.0033, "step": 264200 }, { "epoch": 168.39388145315488, "grad_norm": 0.06190119683742523, "learning_rate": 3.841113162133158e-06, "loss": 0.005, "step": 264210 }, { "epoch": 168.40025493945188, "grad_norm": 0.2690955102443695, "learning_rate": 3.838994943506497e-06, "loss": 0.0043, "step": 264220 }, { "epoch": 168.4066284257489, "grad_norm": 0.1185227483510971, "learning_rate": 3.836877285779617e-06, "loss": 0.0084, "step": 264230 }, { "epoch": 168.4130019120459, "grad_norm": 0.29315075278282166, "learning_rate": 3.834760188978282e-06, "loss": 0.0119, "step": 264240 }, { "epoch": 168.4193753983429, "grad_norm": 0.10276800394058228, "learning_rate": 3.832643653128176e-06, "loss": 0.0037, "step": 264250 }, { "epoch": 168.4257488846399, "grad_norm": 0.1323411464691162, "learning_rate": 3.8305276782550525e-06, "loss": 0.005, "step": 264260 }, { "epoch": 168.4321223709369, "grad_norm": 0.3103278577327728, "learning_rate": 3.828412264384601e-06, "loss": 0.0078, "step": 264270 }, { "epoch": 168.4384958572339, "grad_norm": 0.08469058573246002, "learning_rate": 3.826297411542551e-06, "loss": 0.0056, "step": 264280 }, { "epoch": 168.4448693435309, "grad_norm": 0.6228512525558472, "learning_rate": 3.82418311975456e-06, "loss": 0.003, "step": 264290 }, { "epoch": 168.45124282982792, "grad_norm": 0.1925755739212036, "learning_rate": 3.8220693890463565e-06, "loss": 0.0042, "step": 264300 }, { "epoch": 168.45761631612493, "grad_norm": 0.07311905920505524, "learning_rate": 3.819956219443604e-06, "loss": 0.0044, "step": 264310 }, { "epoch": 168.46398980242193, "grad_norm": 0.06573481857776642, "learning_rate": 3.817843610971989e-06, "loss": 0.0041, "step": 264320 }, { "epoch": 168.47036328871894, "grad_norm": 0.08382797241210938, "learning_rate": 3.815731563657165e-06, "loss": 0.0044, "step": 264330 }, { "epoch": 168.47673677501592, "grad_norm": 0.061121243983507156, "learning_rate": 3.8136200775248154e-06, "loss": 0.0033, "step": 264340 }, { "epoch": 168.48311026131293, "grad_norm": 0.33529916405677795, "learning_rate": 3.811509152600595e-06, "loss": 0.0033, "step": 264350 }, { "epoch": 168.48948374760994, "grad_norm": 0.3218351900577545, "learning_rate": 3.8093987889101455e-06, "loss": 0.0048, "step": 264360 }, { "epoch": 168.49585723390695, "grad_norm": 0.3967501223087311, "learning_rate": 3.8072889864791018e-06, "loss": 0.0078, "step": 264370 }, { "epoch": 168.50223072020395, "grad_norm": 0.06297237426042557, "learning_rate": 3.805179745333115e-06, "loss": 0.0076, "step": 264380 }, { "epoch": 168.50860420650096, "grad_norm": 0.14660213887691498, "learning_rate": 3.8030710654978153e-06, "loss": 0.0053, "step": 264390 }, { "epoch": 168.51497769279797, "grad_norm": 0.18701115250587463, "learning_rate": 3.800962946998815e-06, "loss": 0.0038, "step": 264400 }, { "epoch": 168.52135117909498, "grad_norm": 0.07591491937637329, "learning_rate": 3.7988553898617273e-06, "loss": 0.0053, "step": 264410 }, { "epoch": 168.52772466539196, "grad_norm": 0.07612736523151398, "learning_rate": 3.7967483941121764e-06, "loss": 0.0022, "step": 264420 }, { "epoch": 168.53409815168897, "grad_norm": 0.33353129029273987, "learning_rate": 3.7946419597757576e-06, "loss": 0.0032, "step": 264430 }, { "epoch": 168.54047163798597, "grad_norm": 0.12125343084335327, "learning_rate": 3.7925360868780622e-06, "loss": 0.0049, "step": 264440 }, { "epoch": 168.54684512428298, "grad_norm": 0.08202077448368073, "learning_rate": 3.79043077544467e-06, "loss": 0.0118, "step": 264450 }, { "epoch": 168.55321861058, "grad_norm": 0.1825386881828308, "learning_rate": 3.7883260255011823e-06, "loss": 0.0093, "step": 264460 }, { "epoch": 168.559592096877, "grad_norm": 0.11405467242002487, "learning_rate": 3.7862218370731673e-06, "loss": 0.0059, "step": 264470 }, { "epoch": 168.565965583174, "grad_norm": 0.12056371569633484, "learning_rate": 3.784118210186188e-06, "loss": 0.0041, "step": 264480 }, { "epoch": 168.572339069471, "grad_norm": 0.1942806988954544, "learning_rate": 3.782015144865808e-06, "loss": 0.0049, "step": 264490 }, { "epoch": 168.578712555768, "grad_norm": 0.10742011666297913, "learning_rate": 3.779912641137573e-06, "loss": 0.0028, "step": 264500 }, { "epoch": 168.585086042065, "grad_norm": 0.18824833631515503, "learning_rate": 3.7778106990270458e-06, "loss": 0.0059, "step": 264510 }, { "epoch": 168.591459528362, "grad_norm": 0.1562764048576355, "learning_rate": 3.7757093185597616e-06, "loss": 0.0066, "step": 264520 }, { "epoch": 168.59783301465902, "grad_norm": 0.1134471595287323, "learning_rate": 3.77360849976125e-06, "loss": 0.0045, "step": 264530 }, { "epoch": 168.60420650095602, "grad_norm": 0.12372573465108871, "learning_rate": 3.7715082426570293e-06, "loss": 0.0039, "step": 264540 }, { "epoch": 168.61057998725303, "grad_norm": 0.1536703258752823, "learning_rate": 3.7694085472726405e-06, "loss": 0.0043, "step": 264550 }, { "epoch": 168.61695347355004, "grad_norm": 0.20244255661964417, "learning_rate": 3.767309413633591e-06, "loss": 0.0027, "step": 264560 }, { "epoch": 168.62332695984705, "grad_norm": 0.10961180180311203, "learning_rate": 3.7652108417653762e-06, "loss": 0.0038, "step": 264570 }, { "epoch": 168.62970044614403, "grad_norm": 0.2155916690826416, "learning_rate": 3.7631128316934994e-06, "loss": 0.0032, "step": 264580 }, { "epoch": 168.63607393244104, "grad_norm": 0.2988933324813843, "learning_rate": 3.7610153834434612e-06, "loss": 0.0032, "step": 264590 }, { "epoch": 168.64244741873804, "grad_norm": 0.1348562091588974, "learning_rate": 3.7589184970407476e-06, "loss": 0.0034, "step": 264600 }, { "epoch": 168.64882090503505, "grad_norm": 0.2891206443309784, "learning_rate": 3.756822172510832e-06, "loss": 0.0038, "step": 264610 }, { "epoch": 168.65519439133206, "grad_norm": 0.30462542176246643, "learning_rate": 3.754726409879178e-06, "loss": 0.0068, "step": 264620 }, { "epoch": 168.66156787762907, "grad_norm": 0.11555168032646179, "learning_rate": 3.7526312091712755e-06, "loss": 0.003, "step": 264630 }, { "epoch": 168.66794136392608, "grad_norm": 0.27956101298332214, "learning_rate": 3.750536570412555e-06, "loss": 0.0059, "step": 264640 }, { "epoch": 168.67431485022308, "grad_norm": 0.2525034248828888, "learning_rate": 3.7484424936284903e-06, "loss": 0.0035, "step": 264650 }, { "epoch": 168.68068833652006, "grad_norm": 0.12874895334243774, "learning_rate": 3.746348978844505e-06, "loss": 0.0077, "step": 264660 }, { "epoch": 168.68706182281707, "grad_norm": 0.09783555567264557, "learning_rate": 3.7442560260860736e-06, "loss": 0.0057, "step": 264670 }, { "epoch": 168.69343530911408, "grad_norm": 0.15809747576713562, "learning_rate": 3.7421636353785816e-06, "loss": 0.0051, "step": 264680 }, { "epoch": 168.6998087954111, "grad_norm": 0.28317269682884216, "learning_rate": 3.7400718067474862e-06, "loss": 0.0082, "step": 264690 }, { "epoch": 168.7061822817081, "grad_norm": 0.07486548274755478, "learning_rate": 3.7379805402182e-06, "loss": 0.0071, "step": 264700 }, { "epoch": 168.7125557680051, "grad_norm": 0.15418794751167297, "learning_rate": 3.7358898358161253e-06, "loss": 0.0055, "step": 264710 }, { "epoch": 168.7189292543021, "grad_norm": 0.1248030960559845, "learning_rate": 3.7337996935666585e-06, "loss": 0.005, "step": 264720 }, { "epoch": 168.72530274059912, "grad_norm": 0.296922892332077, "learning_rate": 3.731710113495218e-06, "loss": 0.0041, "step": 264730 }, { "epoch": 168.7316762268961, "grad_norm": 0.22343353927135468, "learning_rate": 3.7296210956271837e-06, "loss": 0.0091, "step": 264740 }, { "epoch": 168.7380497131931, "grad_norm": 0.08340167999267578, "learning_rate": 3.7275326399879405e-06, "loss": 0.006, "step": 264750 }, { "epoch": 168.74442319949011, "grad_norm": 0.3272117078304291, "learning_rate": 3.7254447466028573e-06, "loss": 0.0072, "step": 264760 }, { "epoch": 168.75079668578712, "grad_norm": 0.2020842283964157, "learning_rate": 3.7233574154973193e-06, "loss": 0.004, "step": 264770 }, { "epoch": 168.75717017208413, "grad_norm": 0.21968771517276764, "learning_rate": 3.7212706466966785e-06, "loss": 0.0045, "step": 264780 }, { "epoch": 168.76354365838114, "grad_norm": 0.6204662322998047, "learning_rate": 3.7191844402262923e-06, "loss": 0.0047, "step": 264790 }, { "epoch": 168.76991714467815, "grad_norm": 0.08425841480493546, "learning_rate": 3.7170987961115066e-06, "loss": 0.0064, "step": 264800 }, { "epoch": 168.77629063097515, "grad_norm": 0.30503109097480774, "learning_rate": 3.715013714377685e-06, "loss": 0.0078, "step": 264810 }, { "epoch": 168.78266411727213, "grad_norm": 0.20184747874736786, "learning_rate": 3.7129291950501297e-06, "loss": 0.0053, "step": 264820 }, { "epoch": 168.78903760356914, "grad_norm": 0.13870126008987427, "learning_rate": 3.7108452381541982e-06, "loss": 0.0033, "step": 264830 }, { "epoch": 168.79541108986615, "grad_norm": 0.16006410121917725, "learning_rate": 3.708761843715186e-06, "loss": 0.0048, "step": 264840 }, { "epoch": 168.80178457616316, "grad_norm": 0.08420804142951965, "learning_rate": 3.706679011758446e-06, "loss": 0.0041, "step": 264850 }, { "epoch": 168.80815806246017, "grad_norm": 0.3818349242210388, "learning_rate": 3.7045967423092465e-06, "loss": 0.0055, "step": 264860 }, { "epoch": 168.81453154875717, "grad_norm": 0.1470058262348175, "learning_rate": 3.7025150353929116e-06, "loss": 0.0099, "step": 264870 }, { "epoch": 168.82090503505418, "grad_norm": 0.06715121120214462, "learning_rate": 3.700433891034738e-06, "loss": 0.0048, "step": 264880 }, { "epoch": 168.8272785213512, "grad_norm": 0.11635758727788925, "learning_rate": 3.6983533092599997e-06, "loss": 0.0046, "step": 264890 }, { "epoch": 168.8336520076482, "grad_norm": 0.24924616515636444, "learning_rate": 3.696273290093982e-06, "loss": 0.0143, "step": 264900 }, { "epoch": 168.84002549394518, "grad_norm": 0.24830572307109833, "learning_rate": 3.6941938335619654e-06, "loss": 0.0037, "step": 264910 }, { "epoch": 168.84639898024218, "grad_norm": 0.16230735182762146, "learning_rate": 3.6921149396892174e-06, "loss": 0.0028, "step": 264920 }, { "epoch": 168.8527724665392, "grad_norm": 0.11237384378910065, "learning_rate": 3.690036608500991e-06, "loss": 0.0091, "step": 264930 }, { "epoch": 168.8591459528362, "grad_norm": 0.06148321181535721, "learning_rate": 3.6879588400225373e-06, "loss": 0.002, "step": 264940 }, { "epoch": 168.8655194391332, "grad_norm": 0.3406986892223358, "learning_rate": 3.685881634279126e-06, "loss": 0.0043, "step": 264950 }, { "epoch": 168.87189292543022, "grad_norm": 0.0836452841758728, "learning_rate": 3.6838049912959637e-06, "loss": 0.0046, "step": 264960 }, { "epoch": 168.87826641172722, "grad_norm": 0.1232709214091301, "learning_rate": 3.681728911098309e-06, "loss": 0.0062, "step": 264970 }, { "epoch": 168.88463989802423, "grad_norm": 0.2933559715747833, "learning_rate": 3.679653393711374e-06, "loss": 0.0043, "step": 264980 }, { "epoch": 168.8910133843212, "grad_norm": 0.19495436549186707, "learning_rate": 3.677578439160395e-06, "loss": 0.0034, "step": 264990 }, { "epoch": 168.89738687061822, "grad_norm": 0.07966489344835281, "learning_rate": 3.6755040474705626e-06, "loss": 0.0029, "step": 265000 }, { "epoch": 168.90376035691523, "grad_norm": 0.4643700122833252, "learning_rate": 3.6734302186670956e-06, "loss": 0.0087, "step": 265010 }, { "epoch": 168.91013384321224, "grad_norm": 0.09301911294460297, "learning_rate": 3.67135695277519e-06, "loss": 0.0023, "step": 265020 }, { "epoch": 168.91650732950924, "grad_norm": 0.13701646029949188, "learning_rate": 3.6692842498200487e-06, "loss": 0.0054, "step": 265030 }, { "epoch": 168.92288081580625, "grad_norm": 0.18944533169269562, "learning_rate": 3.667212109826834e-06, "loss": 0.0037, "step": 265040 }, { "epoch": 168.92925430210326, "grad_norm": 0.07432548701763153, "learning_rate": 3.665140532820749e-06, "loss": 0.0033, "step": 265050 }, { "epoch": 168.93562778840027, "grad_norm": 0.1559607982635498, "learning_rate": 3.6630695188269505e-06, "loss": 0.0034, "step": 265060 }, { "epoch": 168.94200127469725, "grad_norm": 0.17418749630451202, "learning_rate": 3.6609990678706074e-06, "loss": 0.013, "step": 265070 }, { "epoch": 168.94837476099426, "grad_norm": 0.3513387143611908, "learning_rate": 3.6589291799768665e-06, "loss": 0.0076, "step": 265080 }, { "epoch": 168.95474824729126, "grad_norm": 0.15179775655269623, "learning_rate": 3.6568598551708968e-06, "loss": 0.0041, "step": 265090 }, { "epoch": 168.96112173358827, "grad_norm": 0.08625699579715729, "learning_rate": 3.6547910934778383e-06, "loss": 0.0042, "step": 265100 }, { "epoch": 168.96749521988528, "grad_norm": 0.23267266154289246, "learning_rate": 3.6527228949228276e-06, "loss": 0.0049, "step": 265110 }, { "epoch": 168.9738687061823, "grad_norm": 0.21519982814788818, "learning_rate": 3.6506552595309775e-06, "loss": 0.0039, "step": 265120 }, { "epoch": 168.9802421924793, "grad_norm": 0.19201987981796265, "learning_rate": 3.6485881873274398e-06, "loss": 0.0034, "step": 265130 }, { "epoch": 168.9866156787763, "grad_norm": 0.33606767654418945, "learning_rate": 3.6465216783373167e-06, "loss": 0.0028, "step": 265140 }, { "epoch": 168.99298916507328, "grad_norm": 0.07193652540445328, "learning_rate": 3.644455732585722e-06, "loss": 0.003, "step": 265150 }, { "epoch": 168.9993626513703, "grad_norm": 0.6838586330413818, "learning_rate": 3.6423903500977463e-06, "loss": 0.0056, "step": 265160 }, { "epoch": 169.0057361376673, "grad_norm": 0.20339269936084747, "learning_rate": 3.6403255308985086e-06, "loss": 0.0047, "step": 265170 }, { "epoch": 169.0121096239643, "grad_norm": 0.05458047240972519, "learning_rate": 3.638261275013083e-06, "loss": 0.0029, "step": 265180 }, { "epoch": 169.01848311026131, "grad_norm": 0.20861414074897766, "learning_rate": 3.6361975824665605e-06, "loss": 0.0045, "step": 265190 }, { "epoch": 169.02485659655832, "grad_norm": 0.5653799772262573, "learning_rate": 3.634134453284005e-06, "loss": 0.0069, "step": 265200 }, { "epoch": 169.03123008285533, "grad_norm": 0.4034743905067444, "learning_rate": 3.63207188749049e-06, "loss": 0.0063, "step": 265210 }, { "epoch": 169.03760356915234, "grad_norm": 0.07446187734603882, "learning_rate": 3.630009885111085e-06, "loss": 0.0054, "step": 265220 }, { "epoch": 169.04397705544932, "grad_norm": 0.10674649477005005, "learning_rate": 3.6279484461708425e-06, "loss": 0.0083, "step": 265230 }, { "epoch": 169.05035054174633, "grad_norm": 0.3294195234775543, "learning_rate": 3.6258875706948083e-06, "loss": 0.0072, "step": 265240 }, { "epoch": 169.05672402804333, "grad_norm": 0.2165544033050537, "learning_rate": 3.623827258708018e-06, "loss": 0.0072, "step": 265250 }, { "epoch": 169.06309751434034, "grad_norm": 0.06634671241044998, "learning_rate": 3.6217675102355187e-06, "loss": 0.007, "step": 265260 }, { "epoch": 169.06947100063735, "grad_norm": 0.14656364917755127, "learning_rate": 3.6197083253023343e-06, "loss": 0.0053, "step": 265270 }, { "epoch": 169.07584448693436, "grad_norm": 0.14558161795139313, "learning_rate": 3.617649703933479e-06, "loss": 0.0054, "step": 265280 }, { "epoch": 169.08221797323137, "grad_norm": 0.05551554262638092, "learning_rate": 3.615591646153971e-06, "loss": 0.0041, "step": 265290 }, { "epoch": 169.08859145952837, "grad_norm": 0.16494213044643402, "learning_rate": 3.6135341519888234e-06, "loss": 0.0047, "step": 265300 }, { "epoch": 169.09496494582535, "grad_norm": 0.2423623651266098, "learning_rate": 3.6114772214630275e-06, "loss": 0.0057, "step": 265310 }, { "epoch": 169.10133843212236, "grad_norm": 0.06524115800857544, "learning_rate": 3.6094208546015863e-06, "loss": 0.0016, "step": 265320 }, { "epoch": 169.10771191841937, "grad_norm": 0.10726127028465271, "learning_rate": 3.607365051429473e-06, "loss": 0.0035, "step": 265330 }, { "epoch": 169.11408540471638, "grad_norm": 0.3541378080844879, "learning_rate": 3.6053098119716854e-06, "loss": 0.0068, "step": 265340 }, { "epoch": 169.12045889101339, "grad_norm": 0.16474391520023346, "learning_rate": 3.6032551362531864e-06, "loss": 0.0043, "step": 265350 }, { "epoch": 169.1268323773104, "grad_norm": 0.08658593893051147, "learning_rate": 3.601201024298945e-06, "loss": 0.0027, "step": 265360 }, { "epoch": 169.1332058636074, "grad_norm": 0.209693044424057, "learning_rate": 3.5991474761339073e-06, "loss": 0.0024, "step": 265370 }, { "epoch": 169.1395793499044, "grad_norm": 0.19996614754199982, "learning_rate": 3.597094491783054e-06, "loss": 0.0051, "step": 265380 }, { "epoch": 169.1459528362014, "grad_norm": 0.26973259449005127, "learning_rate": 3.5950420712712984e-06, "loss": 0.0058, "step": 265390 }, { "epoch": 169.1523263224984, "grad_norm": 0.434804230928421, "learning_rate": 3.5929902146236094e-06, "loss": 0.013, "step": 265400 }, { "epoch": 169.1586998087954, "grad_norm": 0.2944272458553314, "learning_rate": 3.590938921864889e-06, "loss": 0.0071, "step": 265410 }, { "epoch": 169.1650732950924, "grad_norm": 0.3729393780231476, "learning_rate": 3.5888881930200958e-06, "loss": 0.0081, "step": 265420 }, { "epoch": 169.17144678138942, "grad_norm": 0.1556173861026764, "learning_rate": 3.5868380281141144e-06, "loss": 0.0061, "step": 265430 }, { "epoch": 169.17782026768643, "grad_norm": 0.4696367383003235, "learning_rate": 3.5847884271718812e-06, "loss": 0.0124, "step": 265440 }, { "epoch": 169.18419375398344, "grad_norm": 1.206450343132019, "learning_rate": 3.5827393902182926e-06, "loss": 0.0064, "step": 265450 }, { "epoch": 169.19056724028044, "grad_norm": 0.15250933170318604, "learning_rate": 3.5806909172782455e-06, "loss": 0.0036, "step": 265460 }, { "epoch": 169.19694072657742, "grad_norm": 0.1525122970342636, "learning_rate": 3.5786430083766197e-06, "loss": 0.0031, "step": 265470 }, { "epoch": 169.20331421287443, "grad_norm": 0.2309175580739975, "learning_rate": 3.5765956635383236e-06, "loss": 0.0056, "step": 265480 }, { "epoch": 169.20968769917144, "grad_norm": 0.3274490535259247, "learning_rate": 3.57454888278822e-06, "loss": 0.0031, "step": 265490 }, { "epoch": 169.21606118546845, "grad_norm": 0.20015239715576172, "learning_rate": 3.5725026661511783e-06, "loss": 0.0047, "step": 265500 }, { "epoch": 169.22243467176546, "grad_norm": 0.12224364280700684, "learning_rate": 3.5704570136520567e-06, "loss": 0.0186, "step": 265510 }, { "epoch": 169.22880815806246, "grad_norm": 0.05739842355251312, "learning_rate": 3.5684119253157345e-06, "loss": 0.0042, "step": 265520 }, { "epoch": 169.23518164435947, "grad_norm": 0.11846014112234116, "learning_rate": 3.566367401167031e-06, "loss": 0.0037, "step": 265530 }, { "epoch": 169.24155513065648, "grad_norm": 0.1851559281349182, "learning_rate": 3.5643234412308103e-06, "loss": 0.0036, "step": 265540 }, { "epoch": 169.2479286169535, "grad_norm": 0.07203859835863113, "learning_rate": 3.562280045531896e-06, "loss": 0.0028, "step": 265550 }, { "epoch": 169.25430210325047, "grad_norm": 0.16202662885189056, "learning_rate": 3.5602372140951357e-06, "loss": 0.006, "step": 265560 }, { "epoch": 169.26067558954747, "grad_norm": 0.4121206998825073, "learning_rate": 3.5581949469453313e-06, "loss": 0.0083, "step": 265570 }, { "epoch": 169.26704907584448, "grad_norm": 0.24441799521446228, "learning_rate": 3.5561532441073074e-06, "loss": 0.0061, "step": 265580 }, { "epoch": 169.2734225621415, "grad_norm": 0.04652278497815132, "learning_rate": 3.5541121056058778e-06, "loss": 0.0038, "step": 265590 }, { "epoch": 169.2797960484385, "grad_norm": 0.0599835105240345, "learning_rate": 3.552071531465834e-06, "loss": 0.006, "step": 265600 }, { "epoch": 169.2861695347355, "grad_norm": 0.07833919674158096, "learning_rate": 3.5500315217119716e-06, "loss": 0.0041, "step": 265610 }, { "epoch": 169.29254302103251, "grad_norm": 0.11954798549413681, "learning_rate": 3.5479920763690834e-06, "loss": 0.0102, "step": 265620 }, { "epoch": 169.29891650732952, "grad_norm": 0.3989309072494507, "learning_rate": 3.545953195461954e-06, "loss": 0.0078, "step": 265630 }, { "epoch": 169.3052899936265, "grad_norm": 0.07762093842029572, "learning_rate": 3.5439148790153533e-06, "loss": 0.0065, "step": 265640 }, { "epoch": 169.3116634799235, "grad_norm": 0.07704069465398788, "learning_rate": 3.5418771270540386e-06, "loss": 0.0038, "step": 265650 }, { "epoch": 169.31803696622052, "grad_norm": 0.11121248453855515, "learning_rate": 3.5398399396027903e-06, "loss": 0.0048, "step": 265660 }, { "epoch": 169.32441045251753, "grad_norm": 0.7515981793403625, "learning_rate": 3.5378033166863554e-06, "loss": 0.007, "step": 265670 }, { "epoch": 169.33078393881453, "grad_norm": 0.24514417350292206, "learning_rate": 3.5357672583294697e-06, "loss": 0.0109, "step": 265680 }, { "epoch": 169.33715742511154, "grad_norm": 0.09400404989719391, "learning_rate": 3.5337317645568803e-06, "loss": 0.0141, "step": 265690 }, { "epoch": 169.34353091140855, "grad_norm": 0.10807454586029053, "learning_rate": 3.5316968353933334e-06, "loss": 0.0041, "step": 265700 }, { "epoch": 169.34990439770556, "grad_norm": 0.13893935084342957, "learning_rate": 3.5296624708635317e-06, "loss": 0.0054, "step": 265710 }, { "epoch": 169.35627788400254, "grad_norm": 0.16229350864887238, "learning_rate": 3.5276286709922114e-06, "loss": 0.0067, "step": 265720 }, { "epoch": 169.36265137029955, "grad_norm": 0.13658367097377777, "learning_rate": 3.5255954358040687e-06, "loss": 0.0047, "step": 265730 }, { "epoch": 169.36902485659655, "grad_norm": 0.27641093730926514, "learning_rate": 3.52356276532384e-06, "loss": 0.0043, "step": 265740 }, { "epoch": 169.37539834289356, "grad_norm": 0.17837728559970856, "learning_rate": 3.5215306595761887e-06, "loss": 0.004, "step": 265750 }, { "epoch": 169.38177182919057, "grad_norm": 0.45789989829063416, "learning_rate": 3.5194991185858337e-06, "loss": 0.0158, "step": 265760 }, { "epoch": 169.38814531548758, "grad_norm": 0.5702780485153198, "learning_rate": 3.5174681423774447e-06, "loss": 0.0074, "step": 265770 }, { "epoch": 169.39451880178459, "grad_norm": 0.0789378359913826, "learning_rate": 3.515437730975707e-06, "loss": 0.004, "step": 265780 }, { "epoch": 169.4008922880816, "grad_norm": 0.20251476764678955, "learning_rate": 3.5134078844052785e-06, "loss": 0.0051, "step": 265790 }, { "epoch": 169.40726577437857, "grad_norm": 1.1503922939300537, "learning_rate": 3.5113786026908456e-06, "loss": 0.0038, "step": 265800 }, { "epoch": 169.41363926067558, "grad_norm": 0.1344202160835266, "learning_rate": 3.509349885857055e-06, "loss": 0.0063, "step": 265810 }, { "epoch": 169.4200127469726, "grad_norm": 0.22665169835090637, "learning_rate": 3.5073217339285593e-06, "loss": 0.0033, "step": 265820 }, { "epoch": 169.4263862332696, "grad_norm": 0.11882618069648743, "learning_rate": 3.5052941469299883e-06, "loss": 0.0024, "step": 265830 }, { "epoch": 169.4327597195666, "grad_norm": 0.16850250959396362, "learning_rate": 3.503267124886006e-06, "loss": 0.0047, "step": 265840 }, { "epoch": 169.4391332058636, "grad_norm": 0.09451153874397278, "learning_rate": 3.5012406678212207e-06, "loss": 0.0071, "step": 265850 }, { "epoch": 169.44550669216062, "grad_norm": 0.30485936999320984, "learning_rate": 3.4992147757602678e-06, "loss": 0.0079, "step": 265860 }, { "epoch": 169.45188017845763, "grad_norm": 0.46703076362609863, "learning_rate": 3.49718944872775e-06, "loss": 0.0043, "step": 265870 }, { "epoch": 169.4582536647546, "grad_norm": 0.09110777825117111, "learning_rate": 3.4951646867482977e-06, "loss": 0.0051, "step": 265880 }, { "epoch": 169.46462715105162, "grad_norm": 0.19756875932216644, "learning_rate": 3.4931404898465026e-06, "loss": 0.0075, "step": 265890 }, { "epoch": 169.47100063734862, "grad_norm": 0.08584752678871155, "learning_rate": 3.4911168580469556e-06, "loss": 0.009, "step": 265900 }, { "epoch": 169.47737412364563, "grad_norm": 0.17139774560928345, "learning_rate": 3.4890937913742537e-06, "loss": 0.0023, "step": 265910 }, { "epoch": 169.48374760994264, "grad_norm": 0.15625527501106262, "learning_rate": 3.487071289852967e-06, "loss": 0.004, "step": 265920 }, { "epoch": 169.49012109623965, "grad_norm": 0.30070698261260986, "learning_rate": 3.4850493535076856e-06, "loss": 0.0049, "step": 265930 }, { "epoch": 169.49649458253666, "grad_norm": 0.11885728687047958, "learning_rate": 3.483027982362974e-06, "loss": 0.0031, "step": 265940 }, { "epoch": 169.50286806883366, "grad_norm": 0.11714938282966614, "learning_rate": 3.4810071764433904e-06, "loss": 0.0101, "step": 265950 }, { "epoch": 169.50924155513064, "grad_norm": 0.28615954518318176, "learning_rate": 3.4789869357734815e-06, "loss": 0.0046, "step": 265960 }, { "epoch": 169.51561504142765, "grad_norm": 0.12967653572559357, "learning_rate": 3.476967260377817e-06, "loss": 0.0097, "step": 265970 }, { "epoch": 169.52198852772466, "grad_norm": 0.1943012923002243, "learning_rate": 3.474948150280921e-06, "loss": 0.0128, "step": 265980 }, { "epoch": 169.52836201402167, "grad_norm": 0.11612482368946075, "learning_rate": 3.472929605507336e-06, "loss": 0.0047, "step": 265990 }, { "epoch": 169.53473550031867, "grad_norm": 0.1701166182756424, "learning_rate": 3.4709116260815745e-06, "loss": 0.0114, "step": 266000 }, { "epoch": 169.54110898661568, "grad_norm": 0.4290647804737091, "learning_rate": 3.4688942120281733e-06, "loss": 0.0128, "step": 266010 }, { "epoch": 169.5474824729127, "grad_norm": 0.20442771911621094, "learning_rate": 3.4668773633716454e-06, "loss": 0.0038, "step": 266020 }, { "epoch": 169.5538559592097, "grad_norm": 0.17298279702663422, "learning_rate": 3.464861080136489e-06, "loss": 0.0066, "step": 266030 }, { "epoch": 169.56022944550668, "grad_norm": 0.1349862962961197, "learning_rate": 3.4628453623472e-06, "loss": 0.0046, "step": 266040 }, { "epoch": 169.5666029318037, "grad_norm": 0.14954496920108795, "learning_rate": 3.4608302100282876e-06, "loss": 0.0037, "step": 266050 }, { "epoch": 169.5729764181007, "grad_norm": 0.060822755098342896, "learning_rate": 3.458815623204226e-06, "loss": 0.0025, "step": 266060 }, { "epoch": 169.5793499043977, "grad_norm": 0.13295258581638336, "learning_rate": 3.4568016018994955e-06, "loss": 0.0078, "step": 266070 }, { "epoch": 169.5857233906947, "grad_norm": 0.17803829908370972, "learning_rate": 3.45478814613856e-06, "loss": 0.0169, "step": 266080 }, { "epoch": 169.59209687699172, "grad_norm": 0.33216169476509094, "learning_rate": 3.452775255945917e-06, "loss": 0.0058, "step": 266090 }, { "epoch": 169.59847036328873, "grad_norm": 0.30343014001846313, "learning_rate": 3.45076293134598e-06, "loss": 0.0033, "step": 266100 }, { "epoch": 169.60484384958573, "grad_norm": 0.13739486038684845, "learning_rate": 3.4487511723632347e-06, "loss": 0.005, "step": 266110 }, { "epoch": 169.61121733588274, "grad_norm": 0.306586354970932, "learning_rate": 3.4467399790221065e-06, "loss": 0.0062, "step": 266120 }, { "epoch": 169.61759082217972, "grad_norm": 0.10321711003780365, "learning_rate": 3.444729351347059e-06, "loss": 0.0035, "step": 266130 }, { "epoch": 169.62396430847673, "grad_norm": 0.1062513142824173, "learning_rate": 3.4427192893624836e-06, "loss": 0.0021, "step": 266140 }, { "epoch": 169.63033779477374, "grad_norm": 0.15484170615673065, "learning_rate": 3.440709793092839e-06, "loss": 0.0092, "step": 266150 }, { "epoch": 169.63671128107075, "grad_norm": 0.2535690367221832, "learning_rate": 3.438700862562527e-06, "loss": 0.0056, "step": 266160 }, { "epoch": 169.64308476736775, "grad_norm": 0.12693758308887482, "learning_rate": 3.4366924977959626e-06, "loss": 0.0058, "step": 266170 }, { "epoch": 169.64945825366476, "grad_norm": 0.15143252909183502, "learning_rate": 3.434684698817542e-06, "loss": 0.005, "step": 266180 }, { "epoch": 169.65583173996177, "grad_norm": 0.3353555500507355, "learning_rate": 3.432677465651668e-06, "loss": 0.0086, "step": 266190 }, { "epoch": 169.66220522625878, "grad_norm": 0.19594988226890564, "learning_rate": 3.4306707983227325e-06, "loss": 0.0044, "step": 266200 }, { "epoch": 169.66857871255576, "grad_norm": 0.21864911913871765, "learning_rate": 3.4286646968551163e-06, "loss": 0.0067, "step": 266210 }, { "epoch": 169.67495219885276, "grad_norm": 0.13024543225765228, "learning_rate": 3.426659161273188e-06, "loss": 0.0087, "step": 266220 }, { "epoch": 169.68132568514977, "grad_norm": 0.09343008697032928, "learning_rate": 3.424654191601334e-06, "loss": 0.0038, "step": 266230 }, { "epoch": 169.68769917144678, "grad_norm": 0.05136138200759888, "learning_rate": 3.422649787863902e-06, "loss": 0.0027, "step": 266240 }, { "epoch": 169.6940726577438, "grad_norm": 0.09427068382501602, "learning_rate": 3.4206459500852496e-06, "loss": 0.0043, "step": 266250 }, { "epoch": 169.7004461440408, "grad_norm": 0.12430031597614288, "learning_rate": 3.4186426782897185e-06, "loss": 0.0043, "step": 266260 }, { "epoch": 169.7068196303378, "grad_norm": 0.07629089802503586, "learning_rate": 3.4166399725016783e-06, "loss": 0.0023, "step": 266270 }, { "epoch": 169.7131931166348, "grad_norm": 0.13971276581287384, "learning_rate": 3.414637832745432e-06, "loss": 0.0045, "step": 266280 }, { "epoch": 169.7195666029318, "grad_norm": 0.14081984758377075, "learning_rate": 3.412636259045321e-06, "loss": 0.0038, "step": 266290 }, { "epoch": 169.7259400892288, "grad_norm": 0.07415164262056351, "learning_rate": 3.410635251425659e-06, "loss": 0.01, "step": 266300 }, { "epoch": 169.7323135755258, "grad_norm": 0.08792652934789658, "learning_rate": 3.408634809910782e-06, "loss": 0.0065, "step": 266310 }, { "epoch": 169.73868706182282, "grad_norm": 0.08018506318330765, "learning_rate": 3.4066349345249715e-06, "loss": 0.0081, "step": 266320 }, { "epoch": 169.74506054811982, "grad_norm": 0.4257902204990387, "learning_rate": 3.4046356252925405e-06, "loss": 0.0055, "step": 266330 }, { "epoch": 169.75143403441683, "grad_norm": 0.18717248737812042, "learning_rate": 3.4026368822377808e-06, "loss": 0.0045, "step": 266340 }, { "epoch": 169.75780752071384, "grad_norm": 0.26126235723495483, "learning_rate": 3.400638705384979e-06, "loss": 0.0047, "step": 266350 }, { "epoch": 169.76418100701085, "grad_norm": 0.29350045323371887, "learning_rate": 3.3986410947584034e-06, "loss": 0.0108, "step": 266360 }, { "epoch": 169.77055449330783, "grad_norm": 0.2873431444168091, "learning_rate": 3.396644050382347e-06, "loss": 0.0099, "step": 266370 }, { "epoch": 169.77692797960484, "grad_norm": 0.39576366543769836, "learning_rate": 3.394647572281068e-06, "loss": 0.018, "step": 266380 }, { "epoch": 169.78330146590184, "grad_norm": 0.4843006432056427, "learning_rate": 3.3926516604788185e-06, "loss": 0.0067, "step": 266390 }, { "epoch": 169.78967495219885, "grad_norm": 0.19086851179599762, "learning_rate": 3.390656314999852e-06, "loss": 0.0067, "step": 266400 }, { "epoch": 169.79604843849586, "grad_norm": 0.15600870549678802, "learning_rate": 3.3886615358684372e-06, "loss": 0.0068, "step": 266410 }, { "epoch": 169.80242192479287, "grad_norm": 0.18484561145305634, "learning_rate": 3.386667323108772e-06, "loss": 0.0038, "step": 266420 }, { "epoch": 169.80879541108987, "grad_norm": 0.30101749300956726, "learning_rate": 3.3846736767451147e-06, "loss": 0.0051, "step": 266430 }, { "epoch": 169.81516889738688, "grad_norm": 0.08084551244974136, "learning_rate": 3.382680596801685e-06, "loss": 0.0041, "step": 266440 }, { "epoch": 169.82154238368386, "grad_norm": 0.19467680156230927, "learning_rate": 3.3806880833027077e-06, "loss": 0.0063, "step": 266450 }, { "epoch": 169.82791586998087, "grad_norm": 0.12596777081489563, "learning_rate": 3.3786961362723748e-06, "loss": 0.0045, "step": 266460 }, { "epoch": 169.83428935627788, "grad_norm": 0.07034245878458023, "learning_rate": 3.3767047557349106e-06, "loss": 0.0062, "step": 266470 }, { "epoch": 169.8406628425749, "grad_norm": 0.5259067416191101, "learning_rate": 3.374713941714497e-06, "loss": 0.0119, "step": 266480 }, { "epoch": 169.8470363288719, "grad_norm": 0.06533582508563995, "learning_rate": 3.372723694235336e-06, "loss": 0.0089, "step": 266490 }, { "epoch": 169.8534098151689, "grad_norm": 0.1915195882320404, "learning_rate": 3.370734013321597e-06, "loss": 0.0044, "step": 266500 }, { "epoch": 169.8597833014659, "grad_norm": 0.16169409453868866, "learning_rate": 3.3687448989974724e-06, "loss": 0.0026, "step": 266510 }, { "epoch": 169.86615678776292, "grad_norm": 0.07034560292959213, "learning_rate": 3.366756351287126e-06, "loss": 0.0127, "step": 266520 }, { "epoch": 169.8725302740599, "grad_norm": 0.3670799136161804, "learning_rate": 3.364768370214716e-06, "loss": 0.0078, "step": 266530 }, { "epoch": 169.8789037603569, "grad_norm": 0.09039058536291122, "learning_rate": 3.3627809558043953e-06, "loss": 0.0046, "step": 266540 }, { "epoch": 169.8852772466539, "grad_norm": 0.11003795266151428, "learning_rate": 3.3607941080803287e-06, "loss": 0.0061, "step": 266550 }, { "epoch": 169.89165073295092, "grad_norm": 0.15310680866241455, "learning_rate": 3.3588078270666454e-06, "loss": 0.0051, "step": 266560 }, { "epoch": 169.89802421924793, "grad_norm": 0.18552203476428986, "learning_rate": 3.356822112787483e-06, "loss": 0.0078, "step": 266570 }, { "epoch": 169.90439770554494, "grad_norm": 0.05254873260855675, "learning_rate": 3.354836965266961e-06, "loss": 0.0064, "step": 266580 }, { "epoch": 169.91077119184195, "grad_norm": 0.08891855925321579, "learning_rate": 3.3528523845292204e-06, "loss": 0.004, "step": 266590 }, { "epoch": 169.91714467813895, "grad_norm": 0.11971262097358704, "learning_rate": 3.35086837059837e-06, "loss": 0.0061, "step": 266600 }, { "epoch": 169.92351816443593, "grad_norm": 0.48878324031829834, "learning_rate": 3.3488849234985076e-06, "loss": 0.0111, "step": 266610 }, { "epoch": 169.92989165073294, "grad_norm": 0.07428846508264542, "learning_rate": 3.34690204325373e-06, "loss": 0.0034, "step": 266620 }, { "epoch": 169.93626513702995, "grad_norm": 0.41288548707962036, "learning_rate": 3.344919729888152e-06, "loss": 0.009, "step": 266630 }, { "epoch": 169.94263862332696, "grad_norm": 0.3132419288158417, "learning_rate": 3.342937983425848e-06, "loss": 0.0062, "step": 266640 }, { "epoch": 169.94901210962396, "grad_norm": 0.2883247435092926, "learning_rate": 3.340956803890899e-06, "loss": 0.006, "step": 266650 }, { "epoch": 169.95538559592097, "grad_norm": 0.15356311202049255, "learning_rate": 3.33897619130738e-06, "loss": 0.0037, "step": 266660 }, { "epoch": 169.96175908221798, "grad_norm": 0.18098139762878418, "learning_rate": 3.336996145699345e-06, "loss": 0.0072, "step": 266670 }, { "epoch": 169.968132568515, "grad_norm": 0.11661125719547272, "learning_rate": 3.3350166670908733e-06, "loss": 0.0045, "step": 266680 }, { "epoch": 169.97450605481197, "grad_norm": 0.13633987307548523, "learning_rate": 3.3330377555060077e-06, "loss": 0.0045, "step": 266690 }, { "epoch": 169.98087954110898, "grad_norm": 0.07199119031429291, "learning_rate": 3.3310594109687953e-06, "loss": 0.0033, "step": 266700 }, { "epoch": 169.98725302740598, "grad_norm": 0.13279585540294647, "learning_rate": 3.3290816335032614e-06, "loss": 0.0062, "step": 266710 }, { "epoch": 169.993626513703, "grad_norm": 0.13937294483184814, "learning_rate": 3.327104423133465e-06, "loss": 0.0028, "step": 266720 }, { "epoch": 170.0, "grad_norm": 0.2968294024467468, "learning_rate": 3.3251277798834137e-06, "loss": 0.0202, "step": 266730 }, { "epoch": 170.006373486297, "grad_norm": 0.43578359484672546, "learning_rate": 3.3231517037771278e-06, "loss": 0.0054, "step": 266740 }, { "epoch": 170.01274697259402, "grad_norm": 0.20321883261203766, "learning_rate": 3.32117619483861e-06, "loss": 0.0046, "step": 266750 }, { "epoch": 170.01912045889102, "grad_norm": 0.09692103415727615, "learning_rate": 3.3192012530918803e-06, "loss": 0.0044, "step": 266760 }, { "epoch": 170.02549394518803, "grad_norm": 0.05332019925117493, "learning_rate": 3.3172268785609307e-06, "loss": 0.0041, "step": 266770 }, { "epoch": 170.031867431485, "grad_norm": 0.3477380871772766, "learning_rate": 3.315253071269753e-06, "loss": 0.0036, "step": 266780 }, { "epoch": 170.03824091778202, "grad_norm": 0.12158728390932083, "learning_rate": 3.3132798312423165e-06, "loss": 0.0073, "step": 266790 }, { "epoch": 170.04461440407903, "grad_norm": 0.25577425956726074, "learning_rate": 3.31130715850263e-06, "loss": 0.0043, "step": 266800 }, { "epoch": 170.05098789037604, "grad_norm": 0.06479725241661072, "learning_rate": 3.3093350530746247e-06, "loss": 0.0065, "step": 266810 }, { "epoch": 170.05736137667304, "grad_norm": 0.24163004755973816, "learning_rate": 3.3073635149822866e-06, "loss": 0.0061, "step": 266820 }, { "epoch": 170.06373486297005, "grad_norm": 0.19615265727043152, "learning_rate": 3.3053925442495636e-06, "loss": 0.0221, "step": 266830 }, { "epoch": 170.07010834926706, "grad_norm": 0.1363798826932907, "learning_rate": 3.303422140900425e-06, "loss": 0.004, "step": 266840 }, { "epoch": 170.07648183556407, "grad_norm": 0.16386550664901733, "learning_rate": 3.3014523049587742e-06, "loss": 0.0038, "step": 266850 }, { "epoch": 170.08285532186105, "grad_norm": 0.12808936834335327, "learning_rate": 3.299483036448581e-06, "loss": 0.0134, "step": 266860 }, { "epoch": 170.08922880815805, "grad_norm": 0.09650857001543045, "learning_rate": 3.2975143353937532e-06, "loss": 0.0029, "step": 266870 }, { "epoch": 170.09560229445506, "grad_norm": 0.5694602727890015, "learning_rate": 3.2955462018182336e-06, "loss": 0.0093, "step": 266880 }, { "epoch": 170.10197578075207, "grad_norm": 0.07754583656787872, "learning_rate": 3.2935786357459144e-06, "loss": 0.0065, "step": 266890 }, { "epoch": 170.10834926704908, "grad_norm": 0.214237779378891, "learning_rate": 3.2916116372007143e-06, "loss": 0.0038, "step": 266900 }, { "epoch": 170.1147227533461, "grad_norm": 0.10117141902446747, "learning_rate": 3.2896452062065377e-06, "loss": 0.0042, "step": 266910 }, { "epoch": 170.1210962396431, "grad_norm": 0.15564841032028198, "learning_rate": 3.28767934278727e-06, "loss": 0.005, "step": 266920 }, { "epoch": 170.1274697259401, "grad_norm": 0.11349929124116898, "learning_rate": 3.285714046966798e-06, "loss": 0.0036, "step": 266930 }, { "epoch": 170.13384321223708, "grad_norm": 0.3338415026664734, "learning_rate": 3.2837493187690085e-06, "loss": 0.0104, "step": 266940 }, { "epoch": 170.1402166985341, "grad_norm": 0.09234866499900818, "learning_rate": 3.2817851582177762e-06, "loss": 0.0068, "step": 266950 }, { "epoch": 170.1465901848311, "grad_norm": 0.07690909504890442, "learning_rate": 3.279821565336966e-06, "loss": 0.0046, "step": 266960 }, { "epoch": 170.1529636711281, "grad_norm": 0.16718412935733795, "learning_rate": 3.277858540150419e-06, "loss": 0.0024, "step": 266970 }, { "epoch": 170.1593371574251, "grad_norm": 0.2603473663330078, "learning_rate": 3.275896082682028e-06, "loss": 0.0029, "step": 266980 }, { "epoch": 170.16571064372212, "grad_norm": 0.271703839302063, "learning_rate": 3.2739341929555913e-06, "loss": 0.0046, "step": 266990 }, { "epoch": 170.17208413001913, "grad_norm": 0.0681496411561966, "learning_rate": 3.2719728709949827e-06, "loss": 0.003, "step": 267000 }, { "epoch": 170.17845761631614, "grad_norm": 0.1441342979669571, "learning_rate": 3.2700121168240173e-06, "loss": 0.0044, "step": 267010 }, { "epoch": 170.18483110261312, "grad_norm": 0.22457657754421234, "learning_rate": 3.268051930466537e-06, "loss": 0.0048, "step": 267020 }, { "epoch": 170.19120458891013, "grad_norm": 0.0687578096985817, "learning_rate": 3.266092311946328e-06, "loss": 0.005, "step": 267030 }, { "epoch": 170.19757807520713, "grad_norm": 0.076683908700943, "learning_rate": 3.2641332612872387e-06, "loss": 0.0046, "step": 267040 }, { "epoch": 170.20395156150414, "grad_norm": 0.11216098070144653, "learning_rate": 3.2621747785130496e-06, "loss": 0.0076, "step": 267050 }, { "epoch": 170.21032504780115, "grad_norm": 0.1186259314417839, "learning_rate": 3.2602168636475694e-06, "loss": 0.0036, "step": 267060 }, { "epoch": 170.21669853409816, "grad_norm": 0.18263408541679382, "learning_rate": 3.2582595167145734e-06, "loss": 0.0035, "step": 267070 }, { "epoch": 170.22307202039516, "grad_norm": 0.13861727714538574, "learning_rate": 3.25630273773786e-06, "loss": 0.0068, "step": 267080 }, { "epoch": 170.22944550669217, "grad_norm": 0.10665234923362732, "learning_rate": 3.254346526741209e-06, "loss": 0.0054, "step": 267090 }, { "epoch": 170.23581899298915, "grad_norm": 1.0160781145095825, "learning_rate": 3.2523908837483795e-06, "loss": 0.0056, "step": 267100 }, { "epoch": 170.24219247928616, "grad_norm": 0.05909983441233635, "learning_rate": 3.2504358087831312e-06, "loss": 0.0028, "step": 267110 }, { "epoch": 170.24856596558317, "grad_norm": 0.06507214158773422, "learning_rate": 3.2484813018692385e-06, "loss": 0.0062, "step": 267120 }, { "epoch": 170.25493945188018, "grad_norm": 0.3286149799823761, "learning_rate": 3.246527363030427e-06, "loss": 0.008, "step": 267130 }, { "epoch": 170.26131293817718, "grad_norm": 0.1258501261472702, "learning_rate": 3.244573992290456e-06, "loss": 0.003, "step": 267140 }, { "epoch": 170.2676864244742, "grad_norm": 0.24030692875385284, "learning_rate": 3.242621189673051e-06, "loss": 0.0042, "step": 267150 }, { "epoch": 170.2740599107712, "grad_norm": 0.0556122325360775, "learning_rate": 3.240668955201959e-06, "loss": 0.0027, "step": 267160 }, { "epoch": 170.2804333970682, "grad_norm": 0.08572962135076523, "learning_rate": 3.2387172889008666e-06, "loss": 0.0034, "step": 267170 }, { "epoch": 170.2868068833652, "grad_norm": 0.13440410792827606, "learning_rate": 3.236766190793522e-06, "loss": 0.0042, "step": 267180 }, { "epoch": 170.2931803696622, "grad_norm": 0.25725752115249634, "learning_rate": 3.2348156609036118e-06, "loss": 0.004, "step": 267190 }, { "epoch": 170.2995538559592, "grad_norm": 0.18833619356155396, "learning_rate": 3.232865699254861e-06, "loss": 0.0064, "step": 267200 }, { "epoch": 170.3059273422562, "grad_norm": 0.17233020067214966, "learning_rate": 3.230916305870929e-06, "loss": 0.0051, "step": 267210 }, { "epoch": 170.31230082855322, "grad_norm": 0.39616987109184265, "learning_rate": 3.2289674807755354e-06, "loss": 0.0027, "step": 267220 }, { "epoch": 170.31867431485023, "grad_norm": 0.06524314731359482, "learning_rate": 3.227019223992339e-06, "loss": 0.0042, "step": 267230 }, { "epoch": 170.32504780114724, "grad_norm": 0.36287781596183777, "learning_rate": 3.2250715355450213e-06, "loss": 0.0049, "step": 267240 }, { "epoch": 170.33142128744424, "grad_norm": 0.06077964976429939, "learning_rate": 3.223124415457246e-06, "loss": 0.0066, "step": 267250 }, { "epoch": 170.33779477374122, "grad_norm": 0.3030097782611847, "learning_rate": 3.2211778637526722e-06, "loss": 0.0045, "step": 267260 }, { "epoch": 170.34416826003823, "grad_norm": 0.3563567101955414, "learning_rate": 3.2192318804549592e-06, "loss": 0.0068, "step": 267270 }, { "epoch": 170.35054174633524, "grad_norm": 0.1163388341665268, "learning_rate": 3.217286465587749e-06, "loss": 0.0098, "step": 267280 }, { "epoch": 170.35691523263225, "grad_norm": 0.19224640727043152, "learning_rate": 3.2153416191746665e-06, "loss": 0.0045, "step": 267290 }, { "epoch": 170.36328871892925, "grad_norm": 0.10764400660991669, "learning_rate": 3.213397341239366e-06, "loss": 0.0061, "step": 267300 }, { "epoch": 170.36966220522626, "grad_norm": 0.12369274348020554, "learning_rate": 3.2114536318054556e-06, "loss": 0.0072, "step": 267310 }, { "epoch": 170.37603569152327, "grad_norm": 0.13846614956855774, "learning_rate": 3.2095104908965667e-06, "loss": 0.0044, "step": 267320 }, { "epoch": 170.38240917782028, "grad_norm": 0.23946571350097656, "learning_rate": 3.2075679185362916e-06, "loss": 0.0042, "step": 267330 }, { "epoch": 170.38878266411726, "grad_norm": 0.5616464614868164, "learning_rate": 3.205625914748256e-06, "loss": 0.0053, "step": 267340 }, { "epoch": 170.39515615041427, "grad_norm": 0.11750643700361252, "learning_rate": 3.2036844795560463e-06, "loss": 0.0059, "step": 267350 }, { "epoch": 170.40152963671127, "grad_norm": 0.2021905928850174, "learning_rate": 3.201743612983249e-06, "loss": 0.004, "step": 267360 }, { "epoch": 170.40790312300828, "grad_norm": 0.14510981738567352, "learning_rate": 3.1998033150534568e-06, "loss": 0.0042, "step": 267370 }, { "epoch": 170.4142766093053, "grad_norm": 0.17545294761657715, "learning_rate": 3.1978635857902282e-06, "loss": 0.0028, "step": 267380 }, { "epoch": 170.4206500956023, "grad_norm": 0.3967234194278717, "learning_rate": 3.195924425217156e-06, "loss": 0.0077, "step": 267390 }, { "epoch": 170.4270235818993, "grad_norm": 0.1365600824356079, "learning_rate": 3.193985833357793e-06, "loss": 0.0038, "step": 267400 }, { "epoch": 170.4333970681963, "grad_norm": 0.25855764746665955, "learning_rate": 3.1920478102356934e-06, "loss": 0.004, "step": 267410 }, { "epoch": 170.43977055449332, "grad_norm": 0.40828219056129456, "learning_rate": 3.190110355874398e-06, "loss": 0.0109, "step": 267420 }, { "epoch": 170.4461440407903, "grad_norm": 0.2789449393749237, "learning_rate": 3.1881734702974673e-06, "loss": 0.0032, "step": 267430 }, { "epoch": 170.4525175270873, "grad_norm": 0.09161993116140366, "learning_rate": 3.1862371535284262e-06, "loss": 0.0069, "step": 267440 }, { "epoch": 170.45889101338432, "grad_norm": 0.12156790494918823, "learning_rate": 3.184301405590806e-06, "loss": 0.0077, "step": 267450 }, { "epoch": 170.46526449968133, "grad_norm": 0.22112639248371124, "learning_rate": 3.1823662265081155e-06, "loss": 0.0046, "step": 267460 }, { "epoch": 170.47163798597833, "grad_norm": 0.13055568933486938, "learning_rate": 3.180431616303886e-06, "loss": 0.0031, "step": 267470 }, { "epoch": 170.47801147227534, "grad_norm": 0.049689557403326035, "learning_rate": 3.1784975750016156e-06, "loss": 0.0137, "step": 267480 }, { "epoch": 170.48438495857235, "grad_norm": 0.05861978605389595, "learning_rate": 3.1765641026248073e-06, "loss": 0.0047, "step": 267490 }, { "epoch": 170.49075844486936, "grad_norm": 0.399427205324173, "learning_rate": 3.174631199196948e-06, "loss": 0.0063, "step": 267500 }, { "epoch": 170.49713193116634, "grad_norm": 0.12773185968399048, "learning_rate": 3.1726988647415357e-06, "loss": 0.0179, "step": 267510 }, { "epoch": 170.50350541746334, "grad_norm": 0.08638074994087219, "learning_rate": 3.170767099282046e-06, "loss": 0.007, "step": 267520 }, { "epoch": 170.50987890376035, "grad_norm": 0.42455437779426575, "learning_rate": 3.168835902841949e-06, "loss": 0.0105, "step": 267530 }, { "epoch": 170.51625239005736, "grad_norm": 0.2632969617843628, "learning_rate": 3.1669052754447036e-06, "loss": 0.012, "step": 267540 }, { "epoch": 170.52262587635437, "grad_norm": 0.29689115285873413, "learning_rate": 3.1649752171137912e-06, "loss": 0.006, "step": 267550 }, { "epoch": 170.52899936265138, "grad_norm": 0.06827318668365479, "learning_rate": 3.163045727872638e-06, "loss": 0.0049, "step": 267560 }, { "epoch": 170.53537284894838, "grad_norm": 0.2345457822084427, "learning_rate": 3.1611168077447073e-06, "loss": 0.0048, "step": 267570 }, { "epoch": 170.5417463352454, "grad_norm": 0.469937264919281, "learning_rate": 3.1591884567534204e-06, "loss": 0.0106, "step": 267580 }, { "epoch": 170.54811982154237, "grad_norm": 0.0719234049320221, "learning_rate": 3.1572606749222356e-06, "loss": 0.0097, "step": 267590 }, { "epoch": 170.55449330783938, "grad_norm": 0.2406892329454422, "learning_rate": 3.1553334622745455e-06, "loss": 0.0056, "step": 267600 }, { "epoch": 170.5608667941364, "grad_norm": 0.1622549295425415, "learning_rate": 3.153406818833787e-06, "loss": 0.0051, "step": 267610 }, { "epoch": 170.5672402804334, "grad_norm": 0.3329378664493561, "learning_rate": 3.1514807446233695e-06, "loss": 0.0045, "step": 267620 }, { "epoch": 170.5736137667304, "grad_norm": 0.18056797981262207, "learning_rate": 3.149555239666696e-06, "loss": 0.0086, "step": 267630 }, { "epoch": 170.5799872530274, "grad_norm": 0.2261197566986084, "learning_rate": 3.147630303987148e-06, "loss": 0.0067, "step": 267640 }, { "epoch": 170.58636073932442, "grad_norm": 0.07981722801923752, "learning_rate": 3.1457059376081346e-06, "loss": 0.0076, "step": 267650 }, { "epoch": 170.59273422562143, "grad_norm": 0.054046884179115295, "learning_rate": 3.1437821405530366e-06, "loss": 0.0028, "step": 267660 }, { "epoch": 170.5991077119184, "grad_norm": 0.09050631523132324, "learning_rate": 3.1418589128452245e-06, "loss": 0.0094, "step": 267670 }, { "epoch": 170.60548119821541, "grad_norm": 0.3131878077983856, "learning_rate": 3.1399362545080578e-06, "loss": 0.0082, "step": 267680 }, { "epoch": 170.61185468451242, "grad_norm": 0.32640641927719116, "learning_rate": 3.138014165564923e-06, "loss": 0.0041, "step": 267690 }, { "epoch": 170.61822817080943, "grad_norm": 0.506424069404602, "learning_rate": 3.136092646039146e-06, "loss": 0.0104, "step": 267700 }, { "epoch": 170.62460165710644, "grad_norm": 0.20463868975639343, "learning_rate": 3.1341716959541025e-06, "loss": 0.003, "step": 267710 }, { "epoch": 170.63097514340345, "grad_norm": 0.7307563424110413, "learning_rate": 3.1322513153331124e-06, "loss": 0.0051, "step": 267720 }, { "epoch": 170.63734862970045, "grad_norm": 0.13051526248455048, "learning_rate": 3.1303315041995295e-06, "loss": 0.0076, "step": 267730 }, { "epoch": 170.64372211599746, "grad_norm": 0.1586340069770813, "learning_rate": 3.128412262576663e-06, "loss": 0.0035, "step": 267740 }, { "epoch": 170.65009560229444, "grad_norm": 0.3078930675983429, "learning_rate": 3.1264935904878447e-06, "loss": 0.0033, "step": 267750 }, { "epoch": 170.65646908859145, "grad_norm": 0.4627077877521515, "learning_rate": 3.124575487956377e-06, "loss": 0.0048, "step": 267760 }, { "epoch": 170.66284257488846, "grad_norm": 0.10192520916461945, "learning_rate": 3.1226579550055924e-06, "loss": 0.009, "step": 267770 }, { "epoch": 170.66921606118547, "grad_norm": 0.12024418264627457, "learning_rate": 3.1207409916587606e-06, "loss": 0.0055, "step": 267780 }, { "epoch": 170.67558954748247, "grad_norm": 0.227926105260849, "learning_rate": 3.1188245979391904e-06, "loss": 0.004, "step": 267790 }, { "epoch": 170.68196303377948, "grad_norm": 0.2774147391319275, "learning_rate": 3.1169087738701643e-06, "loss": 0.0107, "step": 267800 }, { "epoch": 170.6883365200765, "grad_norm": 0.08970560878515244, "learning_rate": 3.1149935194749624e-06, "loss": 0.0042, "step": 267810 }, { "epoch": 170.6947100063735, "grad_norm": 0.2613808214664459, "learning_rate": 3.1130788347768503e-06, "loss": 0.0045, "step": 267820 }, { "epoch": 170.70108349267048, "grad_norm": 0.3105430006980896, "learning_rate": 3.111164719799109e-06, "loss": 0.0054, "step": 267830 }, { "epoch": 170.70745697896749, "grad_norm": 0.11914218217134476, "learning_rate": 3.109251174564981e-06, "loss": 0.0042, "step": 267840 }, { "epoch": 170.7138304652645, "grad_norm": 0.13686709105968475, "learning_rate": 3.1073381990977256e-06, "loss": 0.0069, "step": 267850 }, { "epoch": 170.7202039515615, "grad_norm": 0.060072995722293854, "learning_rate": 3.1054257934205745e-06, "loss": 0.0108, "step": 267860 }, { "epoch": 170.7265774378585, "grad_norm": 0.1783439815044403, "learning_rate": 3.1035139575567972e-06, "loss": 0.0068, "step": 267870 }, { "epoch": 170.73295092415552, "grad_norm": 0.14507132768630981, "learning_rate": 3.1016026915295813e-06, "loss": 0.0091, "step": 267880 }, { "epoch": 170.73932441045253, "grad_norm": 0.07395835220813751, "learning_rate": 3.0996919953621862e-06, "loss": 0.0066, "step": 267890 }, { "epoch": 170.74569789674953, "grad_norm": 0.43737709522247314, "learning_rate": 3.0977818690778038e-06, "loss": 0.005, "step": 267900 }, { "epoch": 170.7520713830465, "grad_norm": 0.08591892570257187, "learning_rate": 3.095872312699666e-06, "loss": 0.0086, "step": 267910 }, { "epoch": 170.75844486934352, "grad_norm": 0.19104596972465515, "learning_rate": 3.093963326250954e-06, "loss": 0.0071, "step": 267920 }, { "epoch": 170.76481835564053, "grad_norm": 0.21996469795703888, "learning_rate": 3.0920549097548824e-06, "loss": 0.0052, "step": 267930 }, { "epoch": 170.77119184193754, "grad_norm": 0.17133402824401855, "learning_rate": 3.0901470632346273e-06, "loss": 0.0133, "step": 267940 }, { "epoch": 170.77756532823454, "grad_norm": 0.08547700196504593, "learning_rate": 3.0882397867133818e-06, "loss": 0.0026, "step": 267950 }, { "epoch": 170.78393881453155, "grad_norm": 0.16984976828098297, "learning_rate": 3.086333080214304e-06, "loss": 0.0045, "step": 267960 }, { "epoch": 170.79031230082856, "grad_norm": 0.20028240978717804, "learning_rate": 3.0844269437605823e-06, "loss": 0.005, "step": 267970 }, { "epoch": 170.79668578712557, "grad_norm": 0.11772149056196213, "learning_rate": 3.0825213773753636e-06, "loss": 0.007, "step": 267980 }, { "epoch": 170.80305927342255, "grad_norm": 0.16247443854808807, "learning_rate": 3.0806163810818134e-06, "loss": 0.0037, "step": 267990 }, { "epoch": 170.80943275971956, "grad_norm": 0.19960026443004608, "learning_rate": 3.078711954903063e-06, "loss": 0.0055, "step": 268000 }, { "epoch": 170.81580624601656, "grad_norm": 0.12546494603157043, "learning_rate": 3.0768080988622718e-06, "loss": 0.004, "step": 268010 }, { "epoch": 170.82217973231357, "grad_norm": 0.10105164349079132, "learning_rate": 3.074904812982565e-06, "loss": 0.0064, "step": 268020 }, { "epoch": 170.82855321861058, "grad_norm": 0.08651826530694962, "learning_rate": 3.07300209728707e-06, "loss": 0.0046, "step": 268030 }, { "epoch": 170.8349267049076, "grad_norm": 0.30579274892807007, "learning_rate": 3.0710999517989004e-06, "loss": 0.0067, "step": 268040 }, { "epoch": 170.8413001912046, "grad_norm": 0.11335790157318115, "learning_rate": 3.0691983765411768e-06, "loss": 0.0055, "step": 268050 }, { "epoch": 170.8476736775016, "grad_norm": 0.20933446288108826, "learning_rate": 3.067297371537009e-06, "loss": 0.0037, "step": 268060 }, { "epoch": 170.8540471637986, "grad_norm": 0.06042187288403511, "learning_rate": 3.0653969368094893e-06, "loss": 0.0033, "step": 268070 }, { "epoch": 170.8604206500956, "grad_norm": 0.07059959322214127, "learning_rate": 3.0634970723816992e-06, "loss": 0.0088, "step": 268080 }, { "epoch": 170.8667941363926, "grad_norm": 0.16374851763248444, "learning_rate": 3.0615977782767424e-06, "loss": 0.0077, "step": 268090 }, { "epoch": 170.8731676226896, "grad_norm": 0.12831154465675354, "learning_rate": 3.0596990545176895e-06, "loss": 0.0043, "step": 268100 }, { "epoch": 170.87954110898661, "grad_norm": 0.09787213057279587, "learning_rate": 3.0578009011276165e-06, "loss": 0.0062, "step": 268110 }, { "epoch": 170.88591459528362, "grad_norm": 0.1671026051044464, "learning_rate": 3.055903318129577e-06, "loss": 0.0048, "step": 268120 }, { "epoch": 170.89228808158063, "grad_norm": 0.08023204654455185, "learning_rate": 3.0540063055466363e-06, "loss": 0.0118, "step": 268130 }, { "epoch": 170.89866156787764, "grad_norm": 0.28160566091537476, "learning_rate": 3.052109863401842e-06, "loss": 0.0067, "step": 268140 }, { "epoch": 170.90503505417465, "grad_norm": 0.09631697833538055, "learning_rate": 3.0502139917182425e-06, "loss": 0.0037, "step": 268150 }, { "epoch": 170.91140854047163, "grad_norm": 0.1823018193244934, "learning_rate": 3.04831869051887e-06, "loss": 0.0047, "step": 268160 }, { "epoch": 170.91778202676863, "grad_norm": 0.28508439660072327, "learning_rate": 3.04642395982675e-06, "loss": 0.0048, "step": 268170 }, { "epoch": 170.92415551306564, "grad_norm": 0.3914039731025696, "learning_rate": 3.044529799664919e-06, "loss": 0.0045, "step": 268180 }, { "epoch": 170.93052899936265, "grad_norm": 0.08011013269424438, "learning_rate": 3.042636210056382e-06, "loss": 0.0074, "step": 268190 }, { "epoch": 170.93690248565966, "grad_norm": 0.09446733444929123, "learning_rate": 3.040743191024148e-06, "loss": 0.0057, "step": 268200 }, { "epoch": 170.94327597195667, "grad_norm": 0.4170261323451996, "learning_rate": 3.038850742591215e-06, "loss": 0.0119, "step": 268210 }, { "epoch": 170.94964945825367, "grad_norm": 0.4570198059082031, "learning_rate": 3.0369588647805923e-06, "loss": 0.0052, "step": 268220 }, { "epoch": 170.95602294455068, "grad_norm": 0.3236539661884308, "learning_rate": 3.035067557615262e-06, "loss": 0.0033, "step": 268230 }, { "epoch": 170.96239643084766, "grad_norm": 0.25842779874801636, "learning_rate": 3.0331768211181998e-06, "loss": 0.0058, "step": 268240 }, { "epoch": 170.96876991714467, "grad_norm": 0.13439442217350006, "learning_rate": 3.0312866553123765e-06, "loss": 0.0036, "step": 268250 }, { "epoch": 170.97514340344168, "grad_norm": 0.0685834139585495, "learning_rate": 3.029397060220779e-06, "loss": 0.0096, "step": 268260 }, { "epoch": 170.98151688973869, "grad_norm": 0.24829597771167755, "learning_rate": 3.0275080358663443e-06, "loss": 0.0085, "step": 268270 }, { "epoch": 170.9878903760357, "grad_norm": 0.4902162551879883, "learning_rate": 3.0256195822720432e-06, "loss": 0.0083, "step": 268280 }, { "epoch": 170.9942638623327, "grad_norm": 0.13049903512001038, "learning_rate": 3.0237316994608023e-06, "loss": 0.0048, "step": 268290 }, { "epoch": 171.0006373486297, "grad_norm": 0.18429255485534668, "learning_rate": 3.0218443874555968e-06, "loss": 0.0035, "step": 268300 }, { "epoch": 171.00701083492672, "grad_norm": 0.08588650077581406, "learning_rate": 3.0199576462793144e-06, "loss": 0.0055, "step": 268310 }, { "epoch": 171.0133843212237, "grad_norm": 0.1348421722650528, "learning_rate": 3.0180714759549144e-06, "loss": 0.0099, "step": 268320 }, { "epoch": 171.0197578075207, "grad_norm": 0.08233744651079178, "learning_rate": 3.0161858765053064e-06, "loss": 0.0066, "step": 268330 }, { "epoch": 171.0261312938177, "grad_norm": 0.3354019820690155, "learning_rate": 3.0143008479533994e-06, "loss": 0.004, "step": 268340 }, { "epoch": 171.03250478011472, "grad_norm": 0.11348916590213776, "learning_rate": 3.012416390322087e-06, "loss": 0.0033, "step": 268350 }, { "epoch": 171.03887826641173, "grad_norm": 0.10532164573669434, "learning_rate": 3.0105325036342893e-06, "loss": 0.0093, "step": 268360 }, { "epoch": 171.04525175270874, "grad_norm": 0.08410369604825974, "learning_rate": 3.0086491879128873e-06, "loss": 0.0051, "step": 268370 }, { "epoch": 171.05162523900574, "grad_norm": 0.18458621203899384, "learning_rate": 3.0067664431807696e-06, "loss": 0.0022, "step": 268380 }, { "epoch": 171.05799872530275, "grad_norm": 0.17261512577533722, "learning_rate": 3.0048842694607946e-06, "loss": 0.0038, "step": 268390 }, { "epoch": 171.06437221159973, "grad_norm": 0.2556290030479431, "learning_rate": 3.003002666775856e-06, "loss": 0.0039, "step": 268400 }, { "epoch": 171.07074569789674, "grad_norm": 0.30877888202667236, "learning_rate": 3.001121635148807e-06, "loss": 0.0058, "step": 268410 }, { "epoch": 171.07711918419375, "grad_norm": 0.11545012891292572, "learning_rate": 2.999241174602502e-06, "loss": 0.0037, "step": 268420 }, { "epoch": 171.08349267049076, "grad_norm": 0.1023559495806694, "learning_rate": 2.997361285159789e-06, "loss": 0.0063, "step": 268430 }, { "epoch": 171.08986615678776, "grad_norm": 0.3643264174461365, "learning_rate": 2.9954819668435275e-06, "loss": 0.0059, "step": 268440 }, { "epoch": 171.09623964308477, "grad_norm": 0.20087502896785736, "learning_rate": 2.9936032196765273e-06, "loss": 0.0078, "step": 268450 }, { "epoch": 171.10261312938178, "grad_norm": 0.48430395126342773, "learning_rate": 2.991725043681637e-06, "loss": 0.0142, "step": 268460 }, { "epoch": 171.1089866156788, "grad_norm": 0.057502735406160355, "learning_rate": 2.98984743888166e-06, "loss": 0.0032, "step": 268470 }, { "epoch": 171.11536010197577, "grad_norm": 0.18174678087234497, "learning_rate": 2.9879704052994394e-06, "loss": 0.0059, "step": 268480 }, { "epoch": 171.12173358827278, "grad_norm": 0.27213701605796814, "learning_rate": 2.986093942957746e-06, "loss": 0.0042, "step": 268490 }, { "epoch": 171.12810707456978, "grad_norm": 0.1203882023692131, "learning_rate": 2.9842180518794117e-06, "loss": 0.0055, "step": 268500 }, { "epoch": 171.1344805608668, "grad_norm": 0.09915775060653687, "learning_rate": 2.982342732087212e-06, "loss": 0.0033, "step": 268510 }, { "epoch": 171.1408540471638, "grad_norm": 0.19827398657798767, "learning_rate": 2.980467983603946e-06, "loss": 0.0034, "step": 268520 }, { "epoch": 171.1472275334608, "grad_norm": 0.41412362456321716, "learning_rate": 2.978593806452379e-06, "loss": 0.0082, "step": 268530 }, { "epoch": 171.15360101975782, "grad_norm": 0.5084651112556458, "learning_rate": 2.976720200655303e-06, "loss": 0.0067, "step": 268540 }, { "epoch": 171.15997450605482, "grad_norm": 0.2819463014602661, "learning_rate": 2.974847166235467e-06, "loss": 0.0131, "step": 268550 }, { "epoch": 171.1663479923518, "grad_norm": 0.45625612139701843, "learning_rate": 2.972974703215642e-06, "loss": 0.0051, "step": 268560 }, { "epoch": 171.1727214786488, "grad_norm": 0.4513513445854187, "learning_rate": 2.971102811618565e-06, "loss": 0.0074, "step": 268570 }, { "epoch": 171.17909496494582, "grad_norm": 0.16005398333072662, "learning_rate": 2.9692314914670115e-06, "loss": 0.0051, "step": 268580 }, { "epoch": 171.18546845124283, "grad_norm": 0.07539145648479462, "learning_rate": 2.9673607427836815e-06, "loss": 0.0026, "step": 268590 }, { "epoch": 171.19184193753983, "grad_norm": 0.1127648875117302, "learning_rate": 2.965490565591328e-06, "loss": 0.0039, "step": 268600 }, { "epoch": 171.19821542383684, "grad_norm": 0.07706981152296066, "learning_rate": 2.963620959912672e-06, "loss": 0.0032, "step": 268610 }, { "epoch": 171.20458891013385, "grad_norm": 0.6745495200157166, "learning_rate": 2.9617519257704397e-06, "loss": 0.0072, "step": 268620 }, { "epoch": 171.21096239643086, "grad_norm": 0.2309757024049759, "learning_rate": 2.959883463187324e-06, "loss": 0.0058, "step": 268630 }, { "epoch": 171.21733588272787, "grad_norm": 0.1301955282688141, "learning_rate": 2.9580155721860402e-06, "loss": 0.0113, "step": 268640 }, { "epoch": 171.22370936902485, "grad_norm": 0.26949235796928406, "learning_rate": 2.956148252789276e-06, "loss": 0.0042, "step": 268650 }, { "epoch": 171.23008285532185, "grad_norm": 0.3807865083217621, "learning_rate": 2.954281505019746e-06, "loss": 0.0036, "step": 268660 }, { "epoch": 171.23645634161886, "grad_norm": 0.0576179064810276, "learning_rate": 2.9524153289000932e-06, "loss": 0.0027, "step": 268670 }, { "epoch": 171.24282982791587, "grad_norm": 0.16016754508018494, "learning_rate": 2.9505497244530224e-06, "loss": 0.0073, "step": 268680 }, { "epoch": 171.24920331421288, "grad_norm": 0.13445542752742767, "learning_rate": 2.948684691701198e-06, "loss": 0.0114, "step": 268690 }, { "epoch": 171.25557680050989, "grad_norm": 0.3806823492050171, "learning_rate": 2.94682023066728e-06, "loss": 0.0114, "step": 268700 }, { "epoch": 171.2619502868069, "grad_norm": 0.14922192692756653, "learning_rate": 2.9449563413739114e-06, "loss": 0.0061, "step": 268710 }, { "epoch": 171.2683237731039, "grad_norm": 0.1319970041513443, "learning_rate": 2.943093023843757e-06, "loss": 0.0045, "step": 268720 }, { "epoch": 171.27469725940088, "grad_norm": 0.3930418789386749, "learning_rate": 2.9412302780994495e-06, "loss": 0.0059, "step": 268730 }, { "epoch": 171.2810707456979, "grad_norm": 0.22731958329677582, "learning_rate": 2.939368104163631e-06, "loss": 0.0066, "step": 268740 }, { "epoch": 171.2874442319949, "grad_norm": 0.12615016102790833, "learning_rate": 2.937506502058912e-06, "loss": 0.0038, "step": 268750 }, { "epoch": 171.2938177182919, "grad_norm": 0.3119962513446808, "learning_rate": 2.9356454718079294e-06, "loss": 0.0056, "step": 268760 }, { "epoch": 171.3001912045889, "grad_norm": 0.10738582909107208, "learning_rate": 2.9337850134332933e-06, "loss": 0.0045, "step": 268770 }, { "epoch": 171.30656469088592, "grad_norm": 0.06568917632102966, "learning_rate": 2.9319251269576077e-06, "loss": 0.0037, "step": 268780 }, { "epoch": 171.31293817718293, "grad_norm": 0.4022819399833679, "learning_rate": 2.9300658124034598e-06, "loss": 0.0051, "step": 268790 }, { "epoch": 171.31931166347994, "grad_norm": 0.17864146828651428, "learning_rate": 2.9282070697934648e-06, "loss": 0.0044, "step": 268800 }, { "epoch": 171.32568514977692, "grad_norm": 0.19992861151695251, "learning_rate": 2.926348899150194e-06, "loss": 0.0038, "step": 268810 }, { "epoch": 171.33205863607392, "grad_norm": 0.09369894117116928, "learning_rate": 2.9244913004962295e-06, "loss": 0.0032, "step": 268820 }, { "epoch": 171.33843212237093, "grad_norm": 0.10680946707725525, "learning_rate": 2.9226342738541466e-06, "loss": 0.0054, "step": 268830 }, { "epoch": 171.34480560866794, "grad_norm": 0.16595250368118286, "learning_rate": 2.9207778192464952e-06, "loss": 0.0057, "step": 268840 }, { "epoch": 171.35117909496495, "grad_norm": 0.21787215769290924, "learning_rate": 2.9189219366958508e-06, "loss": 0.0046, "step": 268850 }, { "epoch": 171.35755258126196, "grad_norm": 0.10025911033153534, "learning_rate": 2.917066626224757e-06, "loss": 0.0073, "step": 268860 }, { "epoch": 171.36392606755896, "grad_norm": 0.197656512260437, "learning_rate": 2.9152118878557566e-06, "loss": 0.0039, "step": 268870 }, { "epoch": 171.37029955385597, "grad_norm": 0.16120634973049164, "learning_rate": 2.9133577216113763e-06, "loss": 0.0023, "step": 268880 }, { "epoch": 171.37667304015295, "grad_norm": 0.2361200898885727, "learning_rate": 2.9115041275141696e-06, "loss": 0.0049, "step": 268890 }, { "epoch": 171.38304652644996, "grad_norm": 0.24478642642498016, "learning_rate": 2.9096511055866472e-06, "loss": 0.0037, "step": 268900 }, { "epoch": 171.38942001274697, "grad_norm": 0.39057299494743347, "learning_rate": 2.907798655851324e-06, "loss": 0.0036, "step": 268910 }, { "epoch": 171.39579349904398, "grad_norm": 0.1037781834602356, "learning_rate": 2.9059467783306982e-06, "loss": 0.0039, "step": 268920 }, { "epoch": 171.40216698534098, "grad_norm": 0.17244872450828552, "learning_rate": 2.9040954730472914e-06, "loss": 0.0075, "step": 268930 }, { "epoch": 171.408540471638, "grad_norm": 0.16266943514347076, "learning_rate": 2.9022447400235962e-06, "loss": 0.0076, "step": 268940 }, { "epoch": 171.414913957935, "grad_norm": 0.1323779821395874, "learning_rate": 2.9003945792820896e-06, "loss": 0.0045, "step": 268950 }, { "epoch": 171.421287444232, "grad_norm": 0.3840689957141876, "learning_rate": 2.898544990845253e-06, "loss": 0.013, "step": 268960 }, { "epoch": 171.427660930529, "grad_norm": 0.109990194439888, "learning_rate": 2.8966959747355695e-06, "loss": 0.0041, "step": 268970 }, { "epoch": 171.434034416826, "grad_norm": 0.1094280555844307, "learning_rate": 2.8948475309755084e-06, "loss": 0.0086, "step": 268980 }, { "epoch": 171.440407903123, "grad_norm": 0.4220592975616455, "learning_rate": 2.8929996595875196e-06, "loss": 0.0043, "step": 268990 }, { "epoch": 171.44678138942, "grad_norm": 0.7417896389961243, "learning_rate": 2.891152360594057e-06, "loss": 0.0036, "step": 269000 }, { "epoch": 171.45315487571702, "grad_norm": 0.3100546598434448, "learning_rate": 2.889305634017586e-06, "loss": 0.0144, "step": 269010 }, { "epoch": 171.45952836201403, "grad_norm": 0.11442919075489044, "learning_rate": 2.8874594798805165e-06, "loss": 0.004, "step": 269020 }, { "epoch": 171.46590184831103, "grad_norm": 0.12671345472335815, "learning_rate": 2.8856138982053028e-06, "loss": 0.0035, "step": 269030 }, { "epoch": 171.47227533460804, "grad_norm": 0.1344417929649353, "learning_rate": 2.883768889014354e-06, "loss": 0.01, "step": 269040 }, { "epoch": 171.47864882090502, "grad_norm": 0.1618555784225464, "learning_rate": 2.88192445233012e-06, "loss": 0.0072, "step": 269050 }, { "epoch": 171.48502230720203, "grad_norm": 0.2351398766040802, "learning_rate": 2.8800805881749717e-06, "loss": 0.0041, "step": 269060 }, { "epoch": 171.49139579349904, "grad_norm": 0.512287974357605, "learning_rate": 2.8782372965713343e-06, "loss": 0.011, "step": 269070 }, { "epoch": 171.49776927979605, "grad_norm": 0.4585312306880951, "learning_rate": 2.8763945775416134e-06, "loss": 0.0059, "step": 269080 }, { "epoch": 171.50414276609305, "grad_norm": 0.1859171986579895, "learning_rate": 2.874552431108185e-06, "loss": 0.0037, "step": 269090 }, { "epoch": 171.51051625239006, "grad_norm": 0.34559178352355957, "learning_rate": 2.872710857293431e-06, "loss": 0.0144, "step": 269100 }, { "epoch": 171.51688973868707, "grad_norm": 0.0936204344034195, "learning_rate": 2.8708698561197457e-06, "loss": 0.0052, "step": 269110 }, { "epoch": 171.52326322498408, "grad_norm": 0.3104652762413025, "learning_rate": 2.8690294276094877e-06, "loss": 0.0042, "step": 269120 }, { "epoch": 171.52963671128106, "grad_norm": 0.4469491243362427, "learning_rate": 2.8671895717850227e-06, "loss": 0.0062, "step": 269130 }, { "epoch": 171.53601019757807, "grad_norm": 0.09633015841245651, "learning_rate": 2.8653502886686946e-06, "loss": 0.003, "step": 269140 }, { "epoch": 171.54238368387507, "grad_norm": 0.11865708976984024, "learning_rate": 2.863511578282879e-06, "loss": 0.0054, "step": 269150 }, { "epoch": 171.54875717017208, "grad_norm": 0.13318924605846405, "learning_rate": 2.8616734406498922e-06, "loss": 0.0143, "step": 269160 }, { "epoch": 171.5551306564691, "grad_norm": 0.13391681015491486, "learning_rate": 2.8598358757920774e-06, "loss": 0.0078, "step": 269170 }, { "epoch": 171.5615041427661, "grad_norm": 0.6560569405555725, "learning_rate": 2.857998883731766e-06, "loss": 0.0068, "step": 269180 }, { "epoch": 171.5678776290631, "grad_norm": 0.10632854700088501, "learning_rate": 2.8561624644912855e-06, "loss": 0.007, "step": 269190 }, { "epoch": 171.5742511153601, "grad_norm": 0.47993046045303345, "learning_rate": 2.8543266180929285e-06, "loss": 0.0091, "step": 269200 }, { "epoch": 171.5806246016571, "grad_norm": 0.10570710152387619, "learning_rate": 2.8524913445590275e-06, "loss": 0.0031, "step": 269210 }, { "epoch": 171.5869980879541, "grad_norm": 0.28848522901535034, "learning_rate": 2.85065664391187e-06, "loss": 0.0085, "step": 269220 }, { "epoch": 171.5933715742511, "grad_norm": 0.14422562718391418, "learning_rate": 2.8488225161737435e-06, "loss": 0.0071, "step": 269230 }, { "epoch": 171.59974506054812, "grad_norm": 0.09393922984600067, "learning_rate": 2.846988961366942e-06, "loss": 0.0047, "step": 269240 }, { "epoch": 171.60611854684512, "grad_norm": 0.15150390565395355, "learning_rate": 2.845155979513753e-06, "loss": 0.0061, "step": 269250 }, { "epoch": 171.61249203314213, "grad_norm": 0.3291664123535156, "learning_rate": 2.8433235706364357e-06, "loss": 0.0051, "step": 269260 }, { "epoch": 171.61886551943914, "grad_norm": 0.21048711240291595, "learning_rate": 2.8414917347572624e-06, "loss": 0.015, "step": 269270 }, { "epoch": 171.62523900573615, "grad_norm": 0.15354514122009277, "learning_rate": 2.839660471898481e-06, "loss": 0.0048, "step": 269280 }, { "epoch": 171.63161249203316, "grad_norm": 0.2230750024318695, "learning_rate": 2.8378297820823574e-06, "loss": 0.0033, "step": 269290 }, { "epoch": 171.63798597833014, "grad_norm": 0.08609641343355179, "learning_rate": 2.835999665331135e-06, "loss": 0.0021, "step": 269300 }, { "epoch": 171.64435946462714, "grad_norm": 0.17544466257095337, "learning_rate": 2.8341701216670404e-06, "loss": 0.0088, "step": 269310 }, { "epoch": 171.65073295092415, "grad_norm": 0.13584275543689728, "learning_rate": 2.832341151112311e-06, "loss": 0.0051, "step": 269320 }, { "epoch": 171.65710643722116, "grad_norm": 0.279723584651947, "learning_rate": 2.8305127536891797e-06, "loss": 0.0037, "step": 269330 }, { "epoch": 171.66347992351817, "grad_norm": 0.08312076330184937, "learning_rate": 2.828684929419839e-06, "loss": 0.004, "step": 269340 }, { "epoch": 171.66985340981518, "grad_norm": 0.17541077733039856, "learning_rate": 2.8268576783265223e-06, "loss": 0.0039, "step": 269350 }, { "epoch": 171.67622689611218, "grad_norm": 0.05293440446257591, "learning_rate": 2.825031000431416e-06, "loss": 0.0068, "step": 269360 }, { "epoch": 171.6826003824092, "grad_norm": 0.12119422852993011, "learning_rate": 2.823204895756737e-06, "loss": 0.0037, "step": 269370 }, { "epoch": 171.68897386870617, "grad_norm": 0.09058195352554321, "learning_rate": 2.82137936432465e-06, "loss": 0.0048, "step": 269380 }, { "epoch": 171.69534735500318, "grad_norm": 0.11940640956163406, "learning_rate": 2.8195544061573542e-06, "loss": 0.0022, "step": 269390 }, { "epoch": 171.7017208413002, "grad_norm": 0.09719692170619965, "learning_rate": 2.8177300212770154e-06, "loss": 0.0025, "step": 269400 }, { "epoch": 171.7080943275972, "grad_norm": 0.24422506988048553, "learning_rate": 2.815906209705804e-06, "loss": 0.0073, "step": 269410 }, { "epoch": 171.7144678138942, "grad_norm": 0.07837890833616257, "learning_rate": 2.8140829714658758e-06, "loss": 0.0085, "step": 269420 }, { "epoch": 171.7208413001912, "grad_norm": 0.3063873052597046, "learning_rate": 2.8122603065794006e-06, "loss": 0.0068, "step": 269430 }, { "epoch": 171.72721478648822, "grad_norm": 0.12778419256210327, "learning_rate": 2.8104382150685115e-06, "loss": 0.0035, "step": 269440 }, { "epoch": 171.73358827278523, "grad_norm": 0.6231788992881775, "learning_rate": 2.8086166969553516e-06, "loss": 0.0078, "step": 269450 }, { "epoch": 171.7399617590822, "grad_norm": 0.12566256523132324, "learning_rate": 2.806795752262048e-06, "loss": 0.003, "step": 269460 }, { "epoch": 171.7463352453792, "grad_norm": 0.0657537579536438, "learning_rate": 2.8049753810107435e-06, "loss": 0.0114, "step": 269470 }, { "epoch": 171.75270873167622, "grad_norm": 0.29903143644332886, "learning_rate": 2.803155583223549e-06, "loss": 0.0076, "step": 269480 }, { "epoch": 171.75908221797323, "grad_norm": 0.23944343626499176, "learning_rate": 2.801336358922574e-06, "loss": 0.0067, "step": 269490 }, { "epoch": 171.76545570427024, "grad_norm": 0.07101193070411682, "learning_rate": 2.799517708129912e-06, "loss": 0.0047, "step": 269500 }, { "epoch": 171.77182919056725, "grad_norm": 0.233189657330513, "learning_rate": 2.7976996308676896e-06, "loss": 0.0023, "step": 269510 }, { "epoch": 171.77820267686425, "grad_norm": 0.1378360241651535, "learning_rate": 2.7958821271579783e-06, "loss": 0.0059, "step": 269520 }, { "epoch": 171.78457616316126, "grad_norm": 0.07438676804304123, "learning_rate": 2.7940651970228715e-06, "loss": 0.0065, "step": 269530 }, { "epoch": 171.79094964945824, "grad_norm": 0.0907641351222992, "learning_rate": 2.7922488404844294e-06, "loss": 0.0029, "step": 269540 }, { "epoch": 171.79732313575525, "grad_norm": 0.29131200909614563, "learning_rate": 2.790433057564745e-06, "loss": 0.0045, "step": 269550 }, { "epoch": 171.80369662205226, "grad_norm": 0.12613698840141296, "learning_rate": 2.788617848285874e-06, "loss": 0.0045, "step": 269560 }, { "epoch": 171.81007010834927, "grad_norm": 0.09535332024097443, "learning_rate": 2.7868032126698696e-06, "loss": 0.0068, "step": 269570 }, { "epoch": 171.81644359464627, "grad_norm": 0.04612401872873306, "learning_rate": 2.7849891507387815e-06, "loss": 0.0076, "step": 269580 }, { "epoch": 171.82281708094328, "grad_norm": 0.08286865055561066, "learning_rate": 2.783175662514648e-06, "loss": 0.0036, "step": 269590 }, { "epoch": 171.8291905672403, "grad_norm": 0.12922091782093048, "learning_rate": 2.781362748019517e-06, "loss": 0.0187, "step": 269600 }, { "epoch": 171.8355640535373, "grad_norm": 0.0545278936624527, "learning_rate": 2.779550407275411e-06, "loss": 0.0083, "step": 269610 }, { "epoch": 171.84193753983428, "grad_norm": 0.49254196882247925, "learning_rate": 2.7777386403043505e-06, "loss": 0.012, "step": 269620 }, { "epoch": 171.84831102613128, "grad_norm": 0.09083165973424911, "learning_rate": 2.775927447128346e-06, "loss": 0.0085, "step": 269630 }, { "epoch": 171.8546845124283, "grad_norm": 0.14492090046405792, "learning_rate": 2.774116827769413e-06, "loss": 0.0039, "step": 269640 }, { "epoch": 171.8610579987253, "grad_norm": 0.256817102432251, "learning_rate": 2.7723067822495507e-06, "loss": 0.016, "step": 269650 }, { "epoch": 171.8674314850223, "grad_norm": 0.23338930308818817, "learning_rate": 2.7704973105907527e-06, "loss": 0.0041, "step": 269660 }, { "epoch": 171.87380497131932, "grad_norm": 0.13382293283939362, "learning_rate": 2.7686884128149958e-06, "loss": 0.0089, "step": 269670 }, { "epoch": 171.88017845761632, "grad_norm": 0.09454202651977539, "learning_rate": 2.766880088944279e-06, "loss": 0.0057, "step": 269680 }, { "epoch": 171.88655194391333, "grad_norm": 0.2529554069042206, "learning_rate": 2.7650723390005575e-06, "loss": 0.0059, "step": 269690 }, { "epoch": 171.8929254302103, "grad_norm": 0.2323484718799591, "learning_rate": 2.7632651630058126e-06, "loss": 0.009, "step": 269700 }, { "epoch": 171.89929891650732, "grad_norm": 0.06514400243759155, "learning_rate": 2.7614585609819833e-06, "loss": 0.0021, "step": 269710 }, { "epoch": 171.90567240280433, "grad_norm": 0.11988770216703415, "learning_rate": 2.759652532951046e-06, "loss": 0.0052, "step": 269720 }, { "epoch": 171.91204588910134, "grad_norm": 0.27170097827911377, "learning_rate": 2.757847078934922e-06, "loss": 0.0065, "step": 269730 }, { "epoch": 171.91841937539834, "grad_norm": 0.06246418505907059, "learning_rate": 2.7560421989555607e-06, "loss": 0.0048, "step": 269740 }, { "epoch": 171.92479286169535, "grad_norm": 0.08890467882156372, "learning_rate": 2.754237893034889e-06, "loss": 0.0068, "step": 269750 }, { "epoch": 171.93116634799236, "grad_norm": 0.21443898975849152, "learning_rate": 2.75243416119485e-06, "loss": 0.0074, "step": 269760 }, { "epoch": 171.93753983428937, "grad_norm": 0.13609153032302856, "learning_rate": 2.7506310034573325e-06, "loss": 0.0034, "step": 269770 }, { "epoch": 171.94391332058635, "grad_norm": 0.05655567720532417, "learning_rate": 2.748828419844263e-06, "loss": 0.0046, "step": 269780 }, { "epoch": 171.95028680688335, "grad_norm": 0.15491065382957458, "learning_rate": 2.74702641037754e-06, "loss": 0.0074, "step": 269790 }, { "epoch": 171.95666029318036, "grad_norm": 0.2779952585697174, "learning_rate": 2.745224975079064e-06, "loss": 0.0048, "step": 269800 }, { "epoch": 171.96303377947737, "grad_norm": 0.12023987621068954, "learning_rate": 2.7434241139707107e-06, "loss": 0.0059, "step": 269810 }, { "epoch": 171.96940726577438, "grad_norm": 0.22190964221954346, "learning_rate": 2.74162382707438e-06, "loss": 0.0112, "step": 269820 }, { "epoch": 171.9757807520714, "grad_norm": 0.23370610177516937, "learning_rate": 2.7398241144119376e-06, "loss": 0.005, "step": 269830 }, { "epoch": 171.9821542383684, "grad_norm": 0.08548101037740707, "learning_rate": 2.7380249760052544e-06, "loss": 0.003, "step": 269840 }, { "epoch": 171.9885277246654, "grad_norm": 0.1658645123243332, "learning_rate": 2.7362264118761806e-06, "loss": 0.0029, "step": 269850 }, { "epoch": 171.99490121096238, "grad_norm": 0.27435940504074097, "learning_rate": 2.7344284220465923e-06, "loss": 0.008, "step": 269860 }, { "epoch": 172.0012746972594, "grad_norm": 0.12645737826824188, "learning_rate": 2.732631006538322e-06, "loss": 0.0035, "step": 269870 }, { "epoch": 172.0076481835564, "grad_norm": 0.10861402750015259, "learning_rate": 2.730834165373214e-06, "loss": 0.0024, "step": 269880 }, { "epoch": 172.0140216698534, "grad_norm": 0.1797303706407547, "learning_rate": 2.729037898573089e-06, "loss": 0.0098, "step": 269890 }, { "epoch": 172.0203951561504, "grad_norm": 0.1689225733280182, "learning_rate": 2.7272422061598024e-06, "loss": 0.0068, "step": 269900 }, { "epoch": 172.02676864244742, "grad_norm": 0.12457641214132309, "learning_rate": 2.725447088155142e-06, "loss": 0.009, "step": 269910 }, { "epoch": 172.03314212874443, "grad_norm": 0.07622521370649338, "learning_rate": 2.72365254458094e-06, "loss": 0.0055, "step": 269920 }, { "epoch": 172.03951561504144, "grad_norm": 0.3470137417316437, "learning_rate": 2.721858575458991e-06, "loss": 0.0037, "step": 269930 }, { "epoch": 172.04588910133845, "grad_norm": 0.17527931928634644, "learning_rate": 2.7200651808111098e-06, "loss": 0.0039, "step": 269940 }, { "epoch": 172.05226258763543, "grad_norm": 0.15402622520923615, "learning_rate": 2.718272360659063e-06, "loss": 0.0033, "step": 269950 }, { "epoch": 172.05863607393243, "grad_norm": 0.12366283684968948, "learning_rate": 2.716480115024661e-06, "loss": 0.0054, "step": 269960 }, { "epoch": 172.06500956022944, "grad_norm": 0.5159522891044617, "learning_rate": 2.714688443929658e-06, "loss": 0.0206, "step": 269970 }, { "epoch": 172.07138304652645, "grad_norm": 0.47136154770851135, "learning_rate": 2.712897347395843e-06, "loss": 0.0042, "step": 269980 }, { "epoch": 172.07775653282346, "grad_norm": 0.46476322412490845, "learning_rate": 2.7111068254449645e-06, "loss": 0.016, "step": 269990 }, { "epoch": 172.08413001912047, "grad_norm": 0.09176459163427353, "learning_rate": 2.7093168780987887e-06, "loss": 0.0031, "step": 270000 }, { "epoch": 172.09050350541747, "grad_norm": 0.5861035585403442, "learning_rate": 2.707527505379065e-06, "loss": 0.0045, "step": 270010 }, { "epoch": 172.09687699171448, "grad_norm": 0.36909374594688416, "learning_rate": 2.7057387073075315e-06, "loss": 0.007, "step": 270020 }, { "epoch": 172.10325047801146, "grad_norm": 0.14851997792720795, "learning_rate": 2.7039504839059203e-06, "loss": 0.0035, "step": 270030 }, { "epoch": 172.10962396430847, "grad_norm": 0.08077064156532288, "learning_rate": 2.7021628351959815e-06, "loss": 0.0055, "step": 270040 }, { "epoch": 172.11599745060548, "grad_norm": 0.07881283760070801, "learning_rate": 2.7003757611994086e-06, "loss": 0.0067, "step": 270050 }, { "epoch": 172.12237093690248, "grad_norm": 0.15718914568424225, "learning_rate": 2.6985892619379283e-06, "loss": 0.0039, "step": 270060 }, { "epoch": 172.1287444231995, "grad_norm": 0.09305540472269058, "learning_rate": 2.696803337433246e-06, "loss": 0.0077, "step": 270070 }, { "epoch": 172.1351179094965, "grad_norm": 0.07864705473184586, "learning_rate": 2.695017987707077e-06, "loss": 0.0048, "step": 270080 }, { "epoch": 172.1414913957935, "grad_norm": 0.12063324451446533, "learning_rate": 2.6932332127810932e-06, "loss": 0.0059, "step": 270090 }, { "epoch": 172.14786488209052, "grad_norm": 0.1445685774087906, "learning_rate": 2.691449012676994e-06, "loss": 0.0033, "step": 270100 }, { "epoch": 172.1542383683875, "grad_norm": 0.32824915647506714, "learning_rate": 2.6896653874164566e-06, "loss": 0.0029, "step": 270110 }, { "epoch": 172.1606118546845, "grad_norm": 0.05860735476016998, "learning_rate": 2.6878823370211524e-06, "loss": 0.0034, "step": 270120 }, { "epoch": 172.1669853409815, "grad_norm": 0.1937219351530075, "learning_rate": 2.686099861512742e-06, "loss": 0.006, "step": 270130 }, { "epoch": 172.17335882727852, "grad_norm": 0.09472493827342987, "learning_rate": 2.6843179609128967e-06, "loss": 0.0137, "step": 270140 }, { "epoch": 172.17973231357553, "grad_norm": 0.18974363803863525, "learning_rate": 2.68253663524326e-06, "loss": 0.0049, "step": 270150 }, { "epoch": 172.18610579987254, "grad_norm": 0.03907783329486847, "learning_rate": 2.680755884525482e-06, "loss": 0.0033, "step": 270160 }, { "epoch": 172.19247928616954, "grad_norm": 0.14576202630996704, "learning_rate": 2.678975708781184e-06, "loss": 0.0031, "step": 270170 }, { "epoch": 172.19885277246655, "grad_norm": 0.3603346645832062, "learning_rate": 2.6771961080320206e-06, "loss": 0.0065, "step": 270180 }, { "epoch": 172.20522625876353, "grad_norm": 0.358776718378067, "learning_rate": 2.675417082299603e-06, "loss": 0.0039, "step": 270190 }, { "epoch": 172.21159974506054, "grad_norm": 0.17125540971755981, "learning_rate": 2.6736386316055517e-06, "loss": 0.005, "step": 270200 }, { "epoch": 172.21797323135755, "grad_norm": 0.1432906687259674, "learning_rate": 2.671860755971467e-06, "loss": 0.0098, "step": 270210 }, { "epoch": 172.22434671765456, "grad_norm": 0.14617817103862762, "learning_rate": 2.670083455418965e-06, "loss": 0.0032, "step": 270220 }, { "epoch": 172.23072020395156, "grad_norm": 0.13526326417922974, "learning_rate": 2.6683067299696385e-06, "loss": 0.0139, "step": 270230 }, { "epoch": 172.23709369024857, "grad_norm": 0.2362041026353836, "learning_rate": 2.6665305796450714e-06, "loss": 0.0057, "step": 270240 }, { "epoch": 172.24346717654558, "grad_norm": 0.18426333367824554, "learning_rate": 2.6647550044668402e-06, "loss": 0.0046, "step": 270250 }, { "epoch": 172.2498406628426, "grad_norm": 0.11622831225395203, "learning_rate": 2.6629800044565334e-06, "loss": 0.0063, "step": 270260 }, { "epoch": 172.25621414913957, "grad_norm": 0.19486652314662933, "learning_rate": 2.6612055796357173e-06, "loss": 0.0077, "step": 270270 }, { "epoch": 172.26258763543657, "grad_norm": 0.07600732892751694, "learning_rate": 2.6594317300259463e-06, "loss": 0.0068, "step": 270280 }, { "epoch": 172.26896112173358, "grad_norm": 0.0977218896150589, "learning_rate": 2.657658455648776e-06, "loss": 0.014, "step": 270290 }, { "epoch": 172.2753346080306, "grad_norm": 0.09960363060235977, "learning_rate": 2.65588575652575e-06, "loss": 0.0055, "step": 270300 }, { "epoch": 172.2817080943276, "grad_norm": 0.17253191769123077, "learning_rate": 2.6541136326784175e-06, "loss": 0.0047, "step": 270310 }, { "epoch": 172.2880815806246, "grad_norm": 0.2550829350948334, "learning_rate": 2.6523420841283066e-06, "loss": 0.0118, "step": 270320 }, { "epoch": 172.2944550669216, "grad_norm": 0.0987626239657402, "learning_rate": 2.650571110896943e-06, "loss": 0.0034, "step": 270330 }, { "epoch": 172.30082855321862, "grad_norm": 0.1030789166688919, "learning_rate": 2.648800713005839e-06, "loss": 0.0114, "step": 270340 }, { "epoch": 172.3072020395156, "grad_norm": 0.07414072006940842, "learning_rate": 2.6470308904765207e-06, "loss": 0.0058, "step": 270350 }, { "epoch": 172.3135755258126, "grad_norm": 0.06180542707443237, "learning_rate": 2.6452616433304824e-06, "loss": 0.0111, "step": 270360 }, { "epoch": 172.31994901210962, "grad_norm": 0.093619205057621, "learning_rate": 2.6434929715892285e-06, "loss": 0.0061, "step": 270370 }, { "epoch": 172.32632249840663, "grad_norm": 0.4441961944103241, "learning_rate": 2.6417248752742373e-06, "loss": 0.0043, "step": 270380 }, { "epoch": 172.33269598470363, "grad_norm": 0.05883072689175606, "learning_rate": 2.639957354407013e-06, "loss": 0.0031, "step": 270390 }, { "epoch": 172.33906947100064, "grad_norm": 0.17287220060825348, "learning_rate": 2.6381904090090224e-06, "loss": 0.0082, "step": 270400 }, { "epoch": 172.34544295729765, "grad_norm": 0.14165028929710388, "learning_rate": 2.6364240391017315e-06, "loss": 0.0066, "step": 270410 }, { "epoch": 172.35181644359466, "grad_norm": 0.38053685426712036, "learning_rate": 2.6346582447066005e-06, "loss": 0.0061, "step": 270420 }, { "epoch": 172.35818992989164, "grad_norm": 0.13389992713928223, "learning_rate": 2.6328930258451066e-06, "loss": 0.0085, "step": 270430 }, { "epoch": 172.36456341618864, "grad_norm": 0.15761539340019226, "learning_rate": 2.6311283825386722e-06, "loss": 0.0038, "step": 270440 }, { "epoch": 172.37093690248565, "grad_norm": 0.7084741592407227, "learning_rate": 2.6293643148087578e-06, "loss": 0.0101, "step": 270450 }, { "epoch": 172.37731038878266, "grad_norm": 0.18999183177947998, "learning_rate": 2.6276008226767847e-06, "loss": 0.0063, "step": 270460 }, { "epoch": 172.38368387507967, "grad_norm": 0.1100509911775589, "learning_rate": 2.625837906164208e-06, "loss": 0.0039, "step": 270470 }, { "epoch": 172.39005736137668, "grad_norm": 0.1098739430308342, "learning_rate": 2.6240755652924053e-06, "loss": 0.0033, "step": 270480 }, { "epoch": 172.39643084767368, "grad_norm": 0.18417085707187653, "learning_rate": 2.6223138000828317e-06, "loss": 0.0098, "step": 270490 }, { "epoch": 172.4028043339707, "grad_norm": 0.17646341025829315, "learning_rate": 2.6205526105568646e-06, "loss": 0.0052, "step": 270500 }, { "epoch": 172.4091778202677, "grad_norm": 0.18346349895000458, "learning_rate": 2.618791996735931e-06, "loss": 0.0071, "step": 270510 }, { "epoch": 172.41555130656468, "grad_norm": 0.08450762927532196, "learning_rate": 2.6170319586413973e-06, "loss": 0.0046, "step": 270520 }, { "epoch": 172.4219247928617, "grad_norm": 0.2984069287776947, "learning_rate": 2.6152724962946683e-06, "loss": 0.004, "step": 270530 }, { "epoch": 172.4282982791587, "grad_norm": 0.0794525220990181, "learning_rate": 2.6135136097171165e-06, "loss": 0.0091, "step": 270540 }, { "epoch": 172.4346717654557, "grad_norm": 0.08325503021478653, "learning_rate": 2.611755298930119e-06, "loss": 0.0041, "step": 270550 }, { "epoch": 172.4410452517527, "grad_norm": 0.27099061012268066, "learning_rate": 2.6099975639550243e-06, "loss": 0.0057, "step": 270560 }, { "epoch": 172.44741873804972, "grad_norm": 0.0921618714928627, "learning_rate": 2.6082404048132112e-06, "loss": 0.0026, "step": 270570 }, { "epoch": 172.45379222434673, "grad_norm": 0.11335483938455582, "learning_rate": 2.6064838215260236e-06, "loss": 0.0045, "step": 270580 }, { "epoch": 172.46016571064374, "grad_norm": 0.13079991936683655, "learning_rate": 2.6047278141148044e-06, "loss": 0.0051, "step": 270590 }, { "epoch": 172.46653919694072, "grad_norm": 0.05623256415128708, "learning_rate": 2.602972382600882e-06, "loss": 0.003, "step": 270600 }, { "epoch": 172.47291268323772, "grad_norm": 0.3711802363395691, "learning_rate": 2.6012175270056115e-06, "loss": 0.0036, "step": 270610 }, { "epoch": 172.47928616953473, "grad_norm": 0.4337255656719208, "learning_rate": 2.599463247350281e-06, "loss": 0.0096, "step": 270620 }, { "epoch": 172.48565965583174, "grad_norm": 0.21065567433834076, "learning_rate": 2.5977095436562406e-06, "loss": 0.0044, "step": 270630 }, { "epoch": 172.49203314212875, "grad_norm": 0.14485350251197815, "learning_rate": 2.5959564159447723e-06, "loss": 0.0085, "step": 270640 }, { "epoch": 172.49840662842576, "grad_norm": 0.17665669322013855, "learning_rate": 2.5942038642372045e-06, "loss": 0.0077, "step": 270650 }, { "epoch": 172.50478011472276, "grad_norm": 0.08905220776796341, "learning_rate": 2.5924518885548033e-06, "loss": 0.0052, "step": 270660 }, { "epoch": 172.51115360101977, "grad_norm": 0.16860100626945496, "learning_rate": 2.5907004889188847e-06, "loss": 0.0024, "step": 270670 }, { "epoch": 172.51752708731675, "grad_norm": 0.20809850096702576, "learning_rate": 2.5889496653507093e-06, "loss": 0.0147, "step": 270680 }, { "epoch": 172.52390057361376, "grad_norm": 0.08323971182107925, "learning_rate": 2.5871994178715663e-06, "loss": 0.0037, "step": 270690 }, { "epoch": 172.53027405991077, "grad_norm": 0.06779421865940094, "learning_rate": 2.585449746502705e-06, "loss": 0.0041, "step": 270700 }, { "epoch": 172.53664754620777, "grad_norm": 0.08001451939344406, "learning_rate": 2.583700651265403e-06, "loss": 0.0039, "step": 270710 }, { "epoch": 172.54302103250478, "grad_norm": 0.05636792629957199, "learning_rate": 2.581952132180904e-06, "loss": 0.006, "step": 270720 }, { "epoch": 172.5493945188018, "grad_norm": 0.16117379069328308, "learning_rate": 2.5802041892704578e-06, "loss": 0.0038, "step": 270730 }, { "epoch": 172.5557680050988, "grad_norm": 0.6309059262275696, "learning_rate": 2.5784568225552973e-06, "loss": 0.0077, "step": 270740 }, { "epoch": 172.5621414913958, "grad_norm": 0.06403902918100357, "learning_rate": 2.576710032056667e-06, "loss": 0.0074, "step": 270750 }, { "epoch": 172.5685149776928, "grad_norm": 0.05864082649350166, "learning_rate": 2.5749638177957835e-06, "loss": 0.0076, "step": 270760 }, { "epoch": 172.5748884639898, "grad_norm": 0.08672156929969788, "learning_rate": 2.5732181797938683e-06, "loss": 0.0058, "step": 270770 }, { "epoch": 172.5812619502868, "grad_norm": 0.2759556174278259, "learning_rate": 2.5714731180721206e-06, "loss": 0.0054, "step": 270780 }, { "epoch": 172.5876354365838, "grad_norm": 0.5777519941329956, "learning_rate": 2.569728632651769e-06, "loss": 0.0076, "step": 270790 }, { "epoch": 172.59400892288082, "grad_norm": 0.08355475217103958, "learning_rate": 2.567984723553979e-06, "loss": 0.0047, "step": 270800 }, { "epoch": 172.60038240917783, "grad_norm": 0.13045527040958405, "learning_rate": 2.5662413907999727e-06, "loss": 0.0053, "step": 270810 }, { "epoch": 172.60675589547483, "grad_norm": 0.1507883220911026, "learning_rate": 2.5644986344109055e-06, "loss": 0.0046, "step": 270820 }, { "epoch": 172.61312938177184, "grad_norm": 0.11371190845966339, "learning_rate": 2.562756454407983e-06, "loss": 0.0036, "step": 270830 }, { "epoch": 172.61950286806882, "grad_norm": 0.09576164931058884, "learning_rate": 2.561014850812343e-06, "loss": 0.0046, "step": 270840 }, { "epoch": 172.62587635436583, "grad_norm": 0.05482737347483635, "learning_rate": 2.5592738236451696e-06, "loss": 0.0077, "step": 270850 }, { "epoch": 172.63224984066284, "grad_norm": 0.11983622610569, "learning_rate": 2.557533372927612e-06, "loss": 0.0056, "step": 270860 }, { "epoch": 172.63862332695984, "grad_norm": 0.2698265612125397, "learning_rate": 2.5557934986808142e-06, "loss": 0.0067, "step": 270870 }, { "epoch": 172.64499681325685, "grad_norm": 0.06759651005268097, "learning_rate": 2.554054200925915e-06, "loss": 0.0041, "step": 270880 }, { "epoch": 172.65137029955386, "grad_norm": 0.11532813310623169, "learning_rate": 2.5523154796840588e-06, "loss": 0.0148, "step": 270890 }, { "epoch": 172.65774378585087, "grad_norm": 0.11270911246538162, "learning_rate": 2.550577334976367e-06, "loss": 0.007, "step": 270900 }, { "epoch": 172.66411727214788, "grad_norm": 0.2655410170555115, "learning_rate": 2.548839766823963e-06, "loss": 0.0096, "step": 270910 }, { "epoch": 172.67049075844486, "grad_norm": 0.17875851690769196, "learning_rate": 2.5471027752479504e-06, "loss": 0.0048, "step": 270920 }, { "epoch": 172.67686424474186, "grad_norm": 0.26957690715789795, "learning_rate": 2.5453663602694464e-06, "loss": 0.0033, "step": 270930 }, { "epoch": 172.68323773103887, "grad_norm": 0.06910556554794312, "learning_rate": 2.543630521909546e-06, "loss": 0.0061, "step": 270940 }, { "epoch": 172.68961121733588, "grad_norm": 0.5359303951263428, "learning_rate": 2.541895260189342e-06, "loss": 0.0041, "step": 270950 }, { "epoch": 172.6959847036329, "grad_norm": 0.1645900011062622, "learning_rate": 2.540160575129913e-06, "loss": 0.0035, "step": 270960 }, { "epoch": 172.7023581899299, "grad_norm": 0.4716991186141968, "learning_rate": 2.538426466752347e-06, "loss": 0.0161, "step": 270970 }, { "epoch": 172.7087316762269, "grad_norm": 0.09042959660291672, "learning_rate": 2.536692935077711e-06, "loss": 0.0057, "step": 270980 }, { "epoch": 172.7151051625239, "grad_norm": 0.06298601627349854, "learning_rate": 2.534959980127072e-06, "loss": 0.0064, "step": 270990 }, { "epoch": 172.7214786488209, "grad_norm": 0.11449669301509857, "learning_rate": 2.533227601921484e-06, "loss": 0.0022, "step": 271000 }, { "epoch": 172.7278521351179, "grad_norm": 0.4226139485836029, "learning_rate": 2.5314958004819866e-06, "loss": 0.0154, "step": 271010 }, { "epoch": 172.7342256214149, "grad_norm": 0.08017933368682861, "learning_rate": 2.529764575829635e-06, "loss": 0.0024, "step": 271020 }, { "epoch": 172.74059910771192, "grad_norm": 0.5582889914512634, "learning_rate": 2.5280339279854683e-06, "loss": 0.005, "step": 271030 }, { "epoch": 172.74697259400892, "grad_norm": 0.08579884469509125, "learning_rate": 2.5263038569705133e-06, "loss": 0.0062, "step": 271040 }, { "epoch": 172.75334608030593, "grad_norm": 0.442865788936615, "learning_rate": 2.524574362805776e-06, "loss": 0.0054, "step": 271050 }, { "epoch": 172.75971956660294, "grad_norm": 0.10581798851490021, "learning_rate": 2.522845445512295e-06, "loss": 0.0075, "step": 271060 }, { "epoch": 172.76609305289995, "grad_norm": 0.0714738592505455, "learning_rate": 2.5211171051110706e-06, "loss": 0.0029, "step": 271070 }, { "epoch": 172.77246653919693, "grad_norm": 0.19602595269680023, "learning_rate": 2.5193893416230962e-06, "loss": 0.0041, "step": 271080 }, { "epoch": 172.77884002549393, "grad_norm": 0.1470390111207962, "learning_rate": 2.517662155069367e-06, "loss": 0.0077, "step": 271090 }, { "epoch": 172.78521351179094, "grad_norm": 0.09142670035362244, "learning_rate": 2.5159355454708766e-06, "loss": 0.0043, "step": 271100 }, { "epoch": 172.79158699808795, "grad_norm": 0.4467517137527466, "learning_rate": 2.514209512848603e-06, "loss": 0.0192, "step": 271110 }, { "epoch": 172.79796048438496, "grad_norm": 0.22197148203849792, "learning_rate": 2.5124840572235186e-06, "loss": 0.0072, "step": 271120 }, { "epoch": 172.80433397068197, "grad_norm": 0.2566124200820923, "learning_rate": 2.5107591786165787e-06, "loss": 0.0079, "step": 271130 }, { "epoch": 172.81070745697897, "grad_norm": 0.08808477222919464, "learning_rate": 2.5090348770487605e-06, "loss": 0.006, "step": 271140 }, { "epoch": 172.81708094327598, "grad_norm": 0.14494968950748444, "learning_rate": 2.5073111525410088e-06, "loss": 0.0029, "step": 271150 }, { "epoch": 172.823454429573, "grad_norm": 0.5812640190124512, "learning_rate": 2.505588005114268e-06, "loss": 0.0065, "step": 271160 }, { "epoch": 172.82982791586997, "grad_norm": 0.8235348463058472, "learning_rate": 2.5038654347894652e-06, "loss": 0.0103, "step": 271170 }, { "epoch": 172.83620140216698, "grad_norm": 0.4063541889190674, "learning_rate": 2.5021434415875568e-06, "loss": 0.0048, "step": 271180 }, { "epoch": 172.842574888464, "grad_norm": 0.18254618346691132, "learning_rate": 2.5004220255294362e-06, "loss": 0.0038, "step": 271190 }, { "epoch": 172.848948374761, "grad_norm": 0.09034852683544159, "learning_rate": 2.498701186636043e-06, "loss": 0.0034, "step": 271200 }, { "epoch": 172.855321861058, "grad_norm": 0.13225007057189941, "learning_rate": 2.4969809249282717e-06, "loss": 0.0067, "step": 271210 }, { "epoch": 172.861695347355, "grad_norm": 0.4461700916290283, "learning_rate": 2.4952612404270436e-06, "loss": 0.0104, "step": 271220 }, { "epoch": 172.86806883365202, "grad_norm": 0.11747661232948303, "learning_rate": 2.4935421331532373e-06, "loss": 0.0187, "step": 271230 }, { "epoch": 172.87444231994903, "grad_norm": 0.2106698602437973, "learning_rate": 2.4918236031277463e-06, "loss": 0.0052, "step": 271240 }, { "epoch": 172.880815806246, "grad_norm": 0.19792938232421875, "learning_rate": 2.490105650371455e-06, "loss": 0.0047, "step": 271250 }, { "epoch": 172.887189292543, "grad_norm": 0.5742985606193542, "learning_rate": 2.48838827490524e-06, "loss": 0.0066, "step": 271260 }, { "epoch": 172.89356277884002, "grad_norm": 0.22324147820472717, "learning_rate": 2.4866714767499576e-06, "loss": 0.0163, "step": 271270 }, { "epoch": 172.89993626513703, "grad_norm": 0.12829527258872986, "learning_rate": 2.4849552559264854e-06, "loss": 0.0041, "step": 271280 }, { "epoch": 172.90630975143404, "grad_norm": 0.13596341013908386, "learning_rate": 2.4832396124556676e-06, "loss": 0.0028, "step": 271290 }, { "epoch": 172.91268323773104, "grad_norm": 0.12779884040355682, "learning_rate": 2.4815245463583493e-06, "loss": 0.0045, "step": 271300 }, { "epoch": 172.91905672402805, "grad_norm": 0.2015230357646942, "learning_rate": 2.4798100576553684e-06, "loss": 0.0055, "step": 271310 }, { "epoch": 172.92543021032506, "grad_norm": 0.19289977848529816, "learning_rate": 2.4780961463675757e-06, "loss": 0.0047, "step": 271320 }, { "epoch": 172.93180369662204, "grad_norm": 0.10521659255027771, "learning_rate": 2.476382812515765e-06, "loss": 0.0027, "step": 271330 }, { "epoch": 172.93817718291905, "grad_norm": 0.07805155962705612, "learning_rate": 2.4746700561207814e-06, "loss": 0.0061, "step": 271340 }, { "epoch": 172.94455066921606, "grad_norm": 0.07553989440202713, "learning_rate": 2.4729578772034246e-06, "loss": 0.004, "step": 271350 }, { "epoch": 172.95092415551306, "grad_norm": 0.22592495381832123, "learning_rate": 2.4712462757845113e-06, "loss": 0.0039, "step": 271360 }, { "epoch": 172.95729764181007, "grad_norm": 0.21784734725952148, "learning_rate": 2.469535251884819e-06, "loss": 0.0037, "step": 271370 }, { "epoch": 172.96367112810708, "grad_norm": 0.08211023360490799, "learning_rate": 2.4678248055251596e-06, "loss": 0.0032, "step": 271380 }, { "epoch": 172.9700446144041, "grad_norm": 0.10061118751764297, "learning_rate": 2.4661149367262994e-06, "loss": 0.0045, "step": 271390 }, { "epoch": 172.9764181007011, "grad_norm": 0.0892745852470398, "learning_rate": 2.4644056455090326e-06, "loss": 0.0039, "step": 271400 }, { "epoch": 172.98279158699808, "grad_norm": 0.04632900282740593, "learning_rate": 2.462696931894104e-06, "loss": 0.0038, "step": 271410 }, { "epoch": 172.98916507329508, "grad_norm": 0.18529081344604492, "learning_rate": 2.460988795902297e-06, "loss": 0.0051, "step": 271420 }, { "epoch": 172.9955385595921, "grad_norm": 0.3416810631752014, "learning_rate": 2.459281237554362e-06, "loss": 0.0055, "step": 271430 }, { "epoch": 173.0019120458891, "grad_norm": 0.11736394464969635, "learning_rate": 2.4575742568710482e-06, "loss": 0.0046, "step": 271440 }, { "epoch": 173.0082855321861, "grad_norm": 0.08840059489011765, "learning_rate": 2.455867853873084e-06, "loss": 0.0063, "step": 271450 }, { "epoch": 173.01465901848312, "grad_norm": 0.12299472093582153, "learning_rate": 2.454162028581225e-06, "loss": 0.0066, "step": 271460 }, { "epoch": 173.02103250478012, "grad_norm": 0.4551544785499573, "learning_rate": 2.4524567810161826e-06, "loss": 0.0041, "step": 271470 }, { "epoch": 173.02740599107713, "grad_norm": 0.09041951596736908, "learning_rate": 2.450752111198684e-06, "loss": 0.0064, "step": 271480 }, { "epoch": 173.0337794773741, "grad_norm": 0.20838747918605804, "learning_rate": 2.449048019149441e-06, "loss": 0.0084, "step": 271490 }, { "epoch": 173.04015296367112, "grad_norm": 0.13731631636619568, "learning_rate": 2.447344504889165e-06, "loss": 0.0042, "step": 271500 }, { "epoch": 173.04652644996813, "grad_norm": 0.2722484767436981, "learning_rate": 2.445641568438545e-06, "loss": 0.0056, "step": 271510 }, { "epoch": 173.05289993626513, "grad_norm": 0.22861045598983765, "learning_rate": 2.4439392098182803e-06, "loss": 0.0061, "step": 271520 }, { "epoch": 173.05927342256214, "grad_norm": 0.25500449538230896, "learning_rate": 2.442237429049049e-06, "loss": 0.0073, "step": 271530 }, { "epoch": 173.06564690885915, "grad_norm": 0.08281755447387695, "learning_rate": 2.440536226151546e-06, "loss": 0.0042, "step": 271540 }, { "epoch": 173.07202039515616, "grad_norm": 0.4991767406463623, "learning_rate": 2.438835601146422e-06, "loss": 0.0092, "step": 271550 }, { "epoch": 173.07839388145317, "grad_norm": 0.0817067101597786, "learning_rate": 2.437135554054354e-06, "loss": 0.0202, "step": 271560 }, { "epoch": 173.08476736775015, "grad_norm": 0.2559871971607208, "learning_rate": 2.435436084895998e-06, "loss": 0.0041, "step": 271570 }, { "epoch": 173.09114085404715, "grad_norm": 0.09267087280750275, "learning_rate": 2.433737193691998e-06, "loss": 0.0042, "step": 271580 }, { "epoch": 173.09751434034416, "grad_norm": 0.22460569441318512, "learning_rate": 2.4320388804630003e-06, "loss": 0.0049, "step": 271590 }, { "epoch": 173.10388782664117, "grad_norm": 0.2715873718261719, "learning_rate": 2.4303411452296475e-06, "loss": 0.0083, "step": 271600 }, { "epoch": 173.11026131293818, "grad_norm": 0.10288490355014801, "learning_rate": 2.428643988012558e-06, "loss": 0.01, "step": 271610 }, { "epoch": 173.1166347992352, "grad_norm": 0.15737468004226685, "learning_rate": 2.426947408832364e-06, "loss": 0.0028, "step": 271620 }, { "epoch": 173.1230082855322, "grad_norm": 0.09818506985902786, "learning_rate": 2.4252514077096666e-06, "loss": 0.0046, "step": 271630 }, { "epoch": 173.1293817718292, "grad_norm": 0.17907309532165527, "learning_rate": 2.423555984665088e-06, "loss": 0.0052, "step": 271640 }, { "epoch": 173.13575525812618, "grad_norm": 0.22302410006523132, "learning_rate": 2.421861139719228e-06, "loss": 0.0129, "step": 271650 }, { "epoch": 173.1421287444232, "grad_norm": 0.10434285551309586, "learning_rate": 2.420166872892671e-06, "loss": 0.0033, "step": 271660 }, { "epoch": 173.1485022307202, "grad_norm": 0.24601726233959198, "learning_rate": 2.4184731842060047e-06, "loss": 0.0075, "step": 271670 }, { "epoch": 173.1548757170172, "grad_norm": 0.10047566145658493, "learning_rate": 2.4167800736798197e-06, "loss": 0.0071, "step": 271680 }, { "epoch": 173.1612492033142, "grad_norm": 0.1724301129579544, "learning_rate": 2.4150875413346873e-06, "loss": 0.0052, "step": 271690 }, { "epoch": 173.16762268961122, "grad_norm": 0.19089257717132568, "learning_rate": 2.413395587191164e-06, "loss": 0.0033, "step": 271700 }, { "epoch": 173.17399617590823, "grad_norm": 0.34440097212791443, "learning_rate": 2.4117042112698108e-06, "loss": 0.0048, "step": 271710 }, { "epoch": 173.18036966220524, "grad_norm": 0.09950186312198639, "learning_rate": 2.4100134135911833e-06, "loss": 0.0082, "step": 271720 }, { "epoch": 173.18674314850222, "grad_norm": 0.8782897591590881, "learning_rate": 2.408323194175832e-06, "loss": 0.0061, "step": 271730 }, { "epoch": 173.19311663479922, "grad_norm": 0.13942952454090118, "learning_rate": 2.4066335530442795e-06, "loss": 0.0066, "step": 271740 }, { "epoch": 173.19949012109623, "grad_norm": 0.09997417032718658, "learning_rate": 2.4049444902170703e-06, "loss": 0.0046, "step": 271750 }, { "epoch": 173.20586360739324, "grad_norm": 0.2544388175010681, "learning_rate": 2.4032560057147157e-06, "loss": 0.0078, "step": 271760 }, { "epoch": 173.21223709369025, "grad_norm": 0.23767149448394775, "learning_rate": 2.4015680995577496e-06, "loss": 0.0065, "step": 271770 }, { "epoch": 173.21861057998726, "grad_norm": 0.12384625524282455, "learning_rate": 2.399880771766666e-06, "loss": 0.0034, "step": 271780 }, { "epoch": 173.22498406628426, "grad_norm": 0.11814012378454208, "learning_rate": 2.3981940223619714e-06, "loss": 0.0147, "step": 271790 }, { "epoch": 173.23135755258127, "grad_norm": 0.04053756967186928, "learning_rate": 2.3965078513641603e-06, "loss": 0.0044, "step": 271800 }, { "epoch": 173.23773103887828, "grad_norm": 0.11137016862630844, "learning_rate": 2.394822258793733e-06, "loss": 0.0034, "step": 271810 }, { "epoch": 173.24410452517526, "grad_norm": 0.10799925029277802, "learning_rate": 2.3931372446711563e-06, "loss": 0.0038, "step": 271820 }, { "epoch": 173.25047801147227, "grad_norm": 0.18465030193328857, "learning_rate": 2.391452809016914e-06, "loss": 0.008, "step": 271830 }, { "epoch": 173.25685149776928, "grad_norm": 0.13891452550888062, "learning_rate": 2.389768951851462e-06, "loss": 0.0064, "step": 271840 }, { "epoch": 173.26322498406628, "grad_norm": 0.43374302983283997, "learning_rate": 2.3880856731952784e-06, "loss": 0.0167, "step": 271850 }, { "epoch": 173.2695984703633, "grad_norm": 0.4810912609100342, "learning_rate": 2.386402973068802e-06, "loss": 0.0061, "step": 271860 }, { "epoch": 173.2759719566603, "grad_norm": 0.11076536774635315, "learning_rate": 2.384720851492489e-06, "loss": 0.0029, "step": 271870 }, { "epoch": 173.2823454429573, "grad_norm": 0.17901548743247986, "learning_rate": 2.383039308486762e-06, "loss": 0.0034, "step": 271880 }, { "epoch": 173.28871892925432, "grad_norm": 0.3842324912548065, "learning_rate": 2.381358344072082e-06, "loss": 0.006, "step": 271890 }, { "epoch": 173.2950924155513, "grad_norm": 0.18875254690647125, "learning_rate": 2.3796779582688446e-06, "loss": 0.0042, "step": 271900 }, { "epoch": 173.3014659018483, "grad_norm": 0.411583811044693, "learning_rate": 2.377998151097488e-06, "loss": 0.0042, "step": 271910 }, { "epoch": 173.3078393881453, "grad_norm": 0.07867658883333206, "learning_rate": 2.3763189225784077e-06, "loss": 0.0035, "step": 271920 }, { "epoch": 173.31421287444232, "grad_norm": 0.15063175559043884, "learning_rate": 2.3746402727320318e-06, "loss": 0.0046, "step": 271930 }, { "epoch": 173.32058636073933, "grad_norm": 0.15268753468990326, "learning_rate": 2.3729622015787268e-06, "loss": 0.0051, "step": 271940 }, { "epoch": 173.32695984703633, "grad_norm": 0.06087863817811012, "learning_rate": 2.3712847091389045e-06, "loss": 0.0035, "step": 271950 }, { "epoch": 173.33333333333334, "grad_norm": 0.16607217490673065, "learning_rate": 2.3696077954329378e-06, "loss": 0.0031, "step": 271960 }, { "epoch": 173.33970681963035, "grad_norm": 0.2342442125082016, "learning_rate": 2.367931460481221e-06, "loss": 0.0104, "step": 271970 }, { "epoch": 173.34608030592733, "grad_norm": 0.13542024791240692, "learning_rate": 2.3662557043040932e-06, "loss": 0.0035, "step": 271980 }, { "epoch": 173.35245379222434, "grad_norm": 0.2914607524871826, "learning_rate": 2.364580526921939e-06, "loss": 0.0072, "step": 271990 }, { "epoch": 173.35882727852135, "grad_norm": 0.07936330884695053, "learning_rate": 2.362905928355108e-06, "loss": 0.0093, "step": 272000 }, { "epoch": 173.36520076481835, "grad_norm": 0.24821829795837402, "learning_rate": 2.3612319086239453e-06, "loss": 0.0051, "step": 272010 }, { "epoch": 173.37157425111536, "grad_norm": 0.34836840629577637, "learning_rate": 2.3595584677487903e-06, "loss": 0.0046, "step": 272020 }, { "epoch": 173.37794773741237, "grad_norm": 0.14155495166778564, "learning_rate": 2.357885605749982e-06, "loss": 0.0025, "step": 272030 }, { "epoch": 173.38432122370938, "grad_norm": 0.15193726122379303, "learning_rate": 2.3562133226478487e-06, "loss": 0.0056, "step": 272040 }, { "epoch": 173.3906947100064, "grad_norm": 0.07349751144647598, "learning_rate": 2.354541618462708e-06, "loss": 0.0025, "step": 272050 }, { "epoch": 173.39706819630337, "grad_norm": 0.09204105287790298, "learning_rate": 2.352870493214859e-06, "loss": 0.0044, "step": 272060 }, { "epoch": 173.40344168260037, "grad_norm": 0.20340193808078766, "learning_rate": 2.351199946924637e-06, "loss": 0.0028, "step": 272070 }, { "epoch": 173.40981516889738, "grad_norm": 0.43706706166267395, "learning_rate": 2.3495299796123084e-06, "loss": 0.0058, "step": 272080 }, { "epoch": 173.4161886551944, "grad_norm": 0.12450247257947922, "learning_rate": 2.3478605912981898e-06, "loss": 0.0044, "step": 272090 }, { "epoch": 173.4225621414914, "grad_norm": 0.474947065114975, "learning_rate": 2.3461917820025493e-06, "loss": 0.0066, "step": 272100 }, { "epoch": 173.4289356277884, "grad_norm": 0.65901780128479, "learning_rate": 2.344523551745681e-06, "loss": 0.0069, "step": 272110 }, { "epoch": 173.4353091140854, "grad_norm": 0.1849173605442047, "learning_rate": 2.3428559005478357e-06, "loss": 0.0049, "step": 272120 }, { "epoch": 173.44168260038242, "grad_norm": 0.09728042781352997, "learning_rate": 2.3411888284292914e-06, "loss": 0.0049, "step": 272130 }, { "epoch": 173.4480560866794, "grad_norm": 0.19808237254619598, "learning_rate": 2.3395223354103045e-06, "loss": 0.0074, "step": 272140 }, { "epoch": 173.4544295729764, "grad_norm": 0.16241632401943207, "learning_rate": 2.3378564215111144e-06, "loss": 0.0054, "step": 272150 }, { "epoch": 173.46080305927342, "grad_norm": 0.1771589070558548, "learning_rate": 2.3361910867519656e-06, "loss": 0.0041, "step": 272160 }, { "epoch": 173.46717654557042, "grad_norm": 0.08281172811985016, "learning_rate": 2.334526331153103e-06, "loss": 0.0045, "step": 272170 }, { "epoch": 173.47355003186743, "grad_norm": 0.22410666942596436, "learning_rate": 2.3328621547347497e-06, "loss": 0.0021, "step": 272180 }, { "epoch": 173.47992351816444, "grad_norm": 0.03959900140762329, "learning_rate": 2.331198557517128e-06, "loss": 0.0068, "step": 272190 }, { "epoch": 173.48629700446145, "grad_norm": 0.17152039706707, "learning_rate": 2.329535539520439e-06, "loss": 0.0039, "step": 272200 }, { "epoch": 173.49267049075846, "grad_norm": 0.05197601765394211, "learning_rate": 2.327873100764921e-06, "loss": 0.0102, "step": 272210 }, { "epoch": 173.49904397705544, "grad_norm": 0.08690012246370316, "learning_rate": 2.326211241270737e-06, "loss": 0.003, "step": 272220 }, { "epoch": 173.50541746335244, "grad_norm": 0.13826140761375427, "learning_rate": 2.324549961058109e-06, "loss": 0.0027, "step": 272230 }, { "epoch": 173.51179094964945, "grad_norm": 0.23141559958457947, "learning_rate": 2.3228892601471985e-06, "loss": 0.0105, "step": 272240 }, { "epoch": 173.51816443594646, "grad_norm": 0.1529293805360794, "learning_rate": 2.3212291385582175e-06, "loss": 0.0047, "step": 272250 }, { "epoch": 173.52453792224347, "grad_norm": 0.47258347272872925, "learning_rate": 2.319569596311305e-06, "loss": 0.0077, "step": 272260 }, { "epoch": 173.53091140854048, "grad_norm": 0.3528272211551666, "learning_rate": 2.317910633426645e-06, "loss": 0.0091, "step": 272270 }, { "epoch": 173.53728489483748, "grad_norm": 0.11575314402580261, "learning_rate": 2.3162522499243832e-06, "loss": 0.0111, "step": 272280 }, { "epoch": 173.5436583811345, "grad_norm": 0.07021823525428772, "learning_rate": 2.3145944458246915e-06, "loss": 0.0039, "step": 272290 }, { "epoch": 173.55003186743147, "grad_norm": 0.81303870677948, "learning_rate": 2.312937221147682e-06, "loss": 0.008, "step": 272300 }, { "epoch": 173.55640535372848, "grad_norm": 0.07604850828647614, "learning_rate": 2.3112805759135214e-06, "loss": 0.0067, "step": 272310 }, { "epoch": 173.5627788400255, "grad_norm": 0.1665189415216446, "learning_rate": 2.3096245101423274e-06, "loss": 0.0047, "step": 272320 }, { "epoch": 173.5691523263225, "grad_norm": 0.10687810182571411, "learning_rate": 2.307969023854217e-06, "loss": 0.0026, "step": 272330 }, { "epoch": 173.5755258126195, "grad_norm": 0.28124329447746277, "learning_rate": 2.3063141170693137e-06, "loss": 0.0027, "step": 272340 }, { "epoch": 173.5818992989165, "grad_norm": 0.15540282428264618, "learning_rate": 2.304659789807723e-06, "loss": 0.01, "step": 272350 }, { "epoch": 173.58827278521352, "grad_norm": 0.2954860329627991, "learning_rate": 2.3030060420895504e-06, "loss": 0.0049, "step": 272360 }, { "epoch": 173.59464627151053, "grad_norm": 0.0659414529800415, "learning_rate": 2.301352873934892e-06, "loss": 0.0057, "step": 272370 }, { "epoch": 173.60101975780753, "grad_norm": 0.20491167902946472, "learning_rate": 2.2997002853638204e-06, "loss": 0.0073, "step": 272380 }, { "epoch": 173.60739324410451, "grad_norm": 0.07585316896438599, "learning_rate": 2.2980482763964305e-06, "loss": 0.0128, "step": 272390 }, { "epoch": 173.61376673040152, "grad_norm": 0.22048689424991608, "learning_rate": 2.296396847052801e-06, "loss": 0.0069, "step": 272400 }, { "epoch": 173.62014021669853, "grad_norm": 0.18023724853992462, "learning_rate": 2.2947459973529816e-06, "loss": 0.0152, "step": 272410 }, { "epoch": 173.62651370299554, "grad_norm": 0.1678423285484314, "learning_rate": 2.293095727317035e-06, "loss": 0.0079, "step": 272420 }, { "epoch": 173.63288718929255, "grad_norm": 0.061915718019008636, "learning_rate": 2.291446036965028e-06, "loss": 0.004, "step": 272430 }, { "epoch": 173.63926067558955, "grad_norm": 0.31424465775489807, "learning_rate": 2.2897969263169895e-06, "loss": 0.0036, "step": 272440 }, { "epoch": 173.64563416188656, "grad_norm": 0.2603558301925659, "learning_rate": 2.2881483953929696e-06, "loss": 0.006, "step": 272450 }, { "epoch": 173.65200764818357, "grad_norm": 0.24714961647987366, "learning_rate": 2.286500444212991e-06, "loss": 0.0123, "step": 272460 }, { "epoch": 173.65838113448055, "grad_norm": 0.11626645177602768, "learning_rate": 2.2848530727970774e-06, "loss": 0.0032, "step": 272470 }, { "epoch": 173.66475462077756, "grad_norm": 0.13488009572029114, "learning_rate": 2.283206281165257e-06, "loss": 0.0124, "step": 272480 }, { "epoch": 173.67112810707457, "grad_norm": 0.13402265310287476, "learning_rate": 2.2815600693375293e-06, "loss": 0.0069, "step": 272490 }, { "epoch": 173.67750159337157, "grad_norm": 0.2378217875957489, "learning_rate": 2.2799144373339022e-06, "loss": 0.0036, "step": 272500 }, { "epoch": 173.68387507966858, "grad_norm": 0.11173868924379349, "learning_rate": 2.2782693851743584e-06, "loss": 0.0042, "step": 272510 }, { "epoch": 173.6902485659656, "grad_norm": 0.05990833044052124, "learning_rate": 2.2766249128789107e-06, "loss": 0.0032, "step": 272520 }, { "epoch": 173.6966220522626, "grad_norm": 0.13415570557117462, "learning_rate": 2.2749810204675258e-06, "loss": 0.0025, "step": 272530 }, { "epoch": 173.7029955385596, "grad_norm": 0.2658144235610962, "learning_rate": 2.2733377079601826e-06, "loss": 0.0044, "step": 272540 }, { "epoch": 173.70936902485658, "grad_norm": 0.41377246379852295, "learning_rate": 2.2716949753768423e-06, "loss": 0.0089, "step": 272550 }, { "epoch": 173.7157425111536, "grad_norm": 0.21591347455978394, "learning_rate": 2.2700528227374784e-06, "loss": 0.0026, "step": 272560 }, { "epoch": 173.7221159974506, "grad_norm": 0.09779871255159378, "learning_rate": 2.2684112500620304e-06, "loss": 0.0044, "step": 272570 }, { "epoch": 173.7284894837476, "grad_norm": 0.3227836489677429, "learning_rate": 2.2667702573704596e-06, "loss": 0.0051, "step": 272580 }, { "epoch": 173.73486297004462, "grad_norm": 0.19844403862953186, "learning_rate": 2.2651298446826895e-06, "loss": 0.005, "step": 272590 }, { "epoch": 173.74123645634162, "grad_norm": 0.29198116064071655, "learning_rate": 2.2634900120186654e-06, "loss": 0.0036, "step": 272600 }, { "epoch": 173.74760994263863, "grad_norm": 0.15842439234256744, "learning_rate": 2.2618507593983097e-06, "loss": 0.0051, "step": 272610 }, { "epoch": 173.75398342893564, "grad_norm": 0.06949624419212341, "learning_rate": 2.26021208684154e-06, "loss": 0.0042, "step": 272620 }, { "epoch": 173.76035691523262, "grad_norm": 0.2671087384223938, "learning_rate": 2.258573994368257e-06, "loss": 0.0037, "step": 272630 }, { "epoch": 173.76673040152963, "grad_norm": 0.13947290182113647, "learning_rate": 2.2569364819983897e-06, "loss": 0.0037, "step": 272640 }, { "epoch": 173.77310388782664, "grad_norm": 0.09923495352268219, "learning_rate": 2.2552995497518105e-06, "loss": 0.0043, "step": 272650 }, { "epoch": 173.77947737412364, "grad_norm": 0.11706943064928055, "learning_rate": 2.2536631976484257e-06, "loss": 0.0057, "step": 272660 }, { "epoch": 173.78585086042065, "grad_norm": 0.20300249755382538, "learning_rate": 2.2520274257081086e-06, "loss": 0.004, "step": 272670 }, { "epoch": 173.79222434671766, "grad_norm": 0.08600405603647232, "learning_rate": 2.250392233950749e-06, "loss": 0.0025, "step": 272680 }, { "epoch": 173.79859783301467, "grad_norm": 0.2516833543777466, "learning_rate": 2.248757622396197e-06, "loss": 0.0068, "step": 272690 }, { "epoch": 173.80497131931168, "grad_norm": 0.2595096230506897, "learning_rate": 2.247123591064332e-06, "loss": 0.0031, "step": 272700 }, { "epoch": 173.81134480560866, "grad_norm": 0.11385609954595566, "learning_rate": 2.2454901399749983e-06, "loss": 0.0111, "step": 272710 }, { "epoch": 173.81771829190566, "grad_norm": 0.10752350091934204, "learning_rate": 2.2438572691480475e-06, "loss": 0.0052, "step": 272720 }, { "epoch": 173.82409177820267, "grad_norm": 0.22686269879341125, "learning_rate": 2.2422249786033133e-06, "loss": 0.0094, "step": 272730 }, { "epoch": 173.83046526449968, "grad_norm": 0.09595110267400742, "learning_rate": 2.240593268360641e-06, "loss": 0.0185, "step": 272740 }, { "epoch": 173.8368387507967, "grad_norm": 0.3810848295688629, "learning_rate": 2.2389621384398597e-06, "loss": 0.0045, "step": 272750 }, { "epoch": 173.8432122370937, "grad_norm": 0.5083868503570557, "learning_rate": 2.2373315888607748e-06, "loss": 0.0057, "step": 272760 }, { "epoch": 173.8495857233907, "grad_norm": 0.27697375416755676, "learning_rate": 2.2357016196432047e-06, "loss": 0.0037, "step": 272770 }, { "epoch": 173.8559592096877, "grad_norm": 0.12578821182250977, "learning_rate": 2.234072230806972e-06, "loss": 0.004, "step": 272780 }, { "epoch": 173.8623326959847, "grad_norm": 0.2191266417503357, "learning_rate": 2.232443422371844e-06, "loss": 0.0053, "step": 272790 }, { "epoch": 173.8687061822817, "grad_norm": 0.08605556190013885, "learning_rate": 2.2308151943576327e-06, "loss": 0.0039, "step": 272800 }, { "epoch": 173.8750796685787, "grad_norm": 0.10689648240804672, "learning_rate": 2.229187546784117e-06, "loss": 0.0038, "step": 272810 }, { "epoch": 173.88145315487571, "grad_norm": 0.10774046927690506, "learning_rate": 2.2275604796710813e-06, "loss": 0.0028, "step": 272820 }, { "epoch": 173.88782664117272, "grad_norm": 0.09269706159830093, "learning_rate": 2.2259339930382818e-06, "loss": 0.0061, "step": 272830 }, { "epoch": 173.89420012746973, "grad_norm": 0.14671802520751953, "learning_rate": 2.2243080869055023e-06, "loss": 0.0068, "step": 272840 }, { "epoch": 173.90057361376674, "grad_norm": 0.43508341908454895, "learning_rate": 2.222682761292477e-06, "loss": 0.0031, "step": 272850 }, { "epoch": 173.90694710006375, "grad_norm": 0.15173842012882233, "learning_rate": 2.221058016218974e-06, "loss": 0.008, "step": 272860 }, { "epoch": 173.91332058636073, "grad_norm": 0.1108187586069107, "learning_rate": 2.2194338517047165e-06, "loss": 0.0077, "step": 272870 }, { "epoch": 173.91969407265773, "grad_norm": 0.10217506438493729, "learning_rate": 2.2178102677694547e-06, "loss": 0.0123, "step": 272880 }, { "epoch": 173.92606755895474, "grad_norm": 0.418659508228302, "learning_rate": 2.2161872644329173e-06, "loss": 0.0117, "step": 272890 }, { "epoch": 173.93244104525175, "grad_norm": 0.3096332252025604, "learning_rate": 2.214564841714817e-06, "loss": 0.0033, "step": 272900 }, { "epoch": 173.93881453154876, "grad_norm": 0.07927604019641876, "learning_rate": 2.212942999634865e-06, "loss": 0.0043, "step": 272910 }, { "epoch": 173.94518801784577, "grad_norm": 0.49061742424964905, "learning_rate": 2.2113217382127795e-06, "loss": 0.0081, "step": 272920 }, { "epoch": 173.95156150414277, "grad_norm": 0.16097474098205566, "learning_rate": 2.209701057468255e-06, "loss": 0.0054, "step": 272930 }, { "epoch": 173.95793499043978, "grad_norm": 0.11269877105951309, "learning_rate": 2.2080809574209825e-06, "loss": 0.0037, "step": 272940 }, { "epoch": 173.96430847673676, "grad_norm": 0.6188797950744629, "learning_rate": 2.2064614380906457e-06, "loss": 0.0066, "step": 272950 }, { "epoch": 173.97068196303377, "grad_norm": 0.08210647851228714, "learning_rate": 2.2048424994969396e-06, "loss": 0.0028, "step": 272960 }, { "epoch": 173.97705544933078, "grad_norm": 0.20162124931812286, "learning_rate": 2.20322414165951e-06, "loss": 0.0049, "step": 272970 }, { "epoch": 173.98342893562778, "grad_norm": 0.15395276248455048, "learning_rate": 2.2016063645980413e-06, "loss": 0.0081, "step": 272980 }, { "epoch": 173.9898024219248, "grad_norm": 0.1884872019290924, "learning_rate": 2.199989168332178e-06, "loss": 0.0026, "step": 272990 }, { "epoch": 173.9961759082218, "grad_norm": 0.14954262971878052, "learning_rate": 2.198372552881589e-06, "loss": 0.0071, "step": 273000 }, { "epoch": 174.0025493945188, "grad_norm": 0.09492871165275574, "learning_rate": 2.196756518265891e-06, "loss": 0.0037, "step": 273010 }, { "epoch": 174.00892288081582, "grad_norm": 0.08575667440891266, "learning_rate": 2.1951410645047463e-06, "loss": 0.0048, "step": 273020 }, { "epoch": 174.01529636711282, "grad_norm": 0.12493041157722473, "learning_rate": 2.1935261916177675e-06, "loss": 0.0036, "step": 273030 }, { "epoch": 174.0216698534098, "grad_norm": 0.28276926279067993, "learning_rate": 2.191911899624588e-06, "loss": 0.0031, "step": 273040 }, { "epoch": 174.0280433397068, "grad_norm": 0.1387854516506195, "learning_rate": 2.1902981885448093e-06, "loss": 0.0043, "step": 273050 }, { "epoch": 174.03441682600382, "grad_norm": 0.2440875917673111, "learning_rate": 2.188685058398049e-06, "loss": 0.0041, "step": 273060 }, { "epoch": 174.04079031230083, "grad_norm": 0.21221517026424408, "learning_rate": 2.1870725092039133e-06, "loss": 0.0083, "step": 273070 }, { "epoch": 174.04716379859784, "grad_norm": 0.48944589495658875, "learning_rate": 2.185460540981982e-06, "loss": 0.006, "step": 273080 }, { "epoch": 174.05353728489484, "grad_norm": 0.2553650736808777, "learning_rate": 2.183849153751849e-06, "loss": 0.0091, "step": 273090 }, { "epoch": 174.05991077119185, "grad_norm": 0.08653663843870163, "learning_rate": 2.1822383475331e-06, "loss": 0.003, "step": 273100 }, { "epoch": 174.06628425748886, "grad_norm": 0.09330429136753082, "learning_rate": 2.180628122345302e-06, "loss": 0.0067, "step": 273110 }, { "epoch": 174.07265774378584, "grad_norm": 0.06234350800514221, "learning_rate": 2.1790184782080225e-06, "loss": 0.0034, "step": 273120 }, { "epoch": 174.07903123008285, "grad_norm": 0.28666695952415466, "learning_rate": 2.177409415140813e-06, "loss": 0.0074, "step": 273130 }, { "epoch": 174.08540471637986, "grad_norm": 0.22277973592281342, "learning_rate": 2.175800933163236e-06, "loss": 0.005, "step": 273140 }, { "epoch": 174.09177820267686, "grad_norm": 0.3791757822036743, "learning_rate": 2.17419303229483e-06, "loss": 0.0065, "step": 273150 }, { "epoch": 174.09815168897387, "grad_norm": 0.08405055850744247, "learning_rate": 2.1725857125551365e-06, "loss": 0.005, "step": 273160 }, { "epoch": 174.10452517527088, "grad_norm": 0.08912169188261032, "learning_rate": 2.1709789739636723e-06, "loss": 0.0083, "step": 273170 }, { "epoch": 174.1108986615679, "grad_norm": 0.20803749561309814, "learning_rate": 2.169372816539983e-06, "loss": 0.0065, "step": 273180 }, { "epoch": 174.1172721478649, "grad_norm": 0.24360232055187225, "learning_rate": 2.16776724030357e-06, "loss": 0.013, "step": 273190 }, { "epoch": 174.12364563416187, "grad_norm": 0.33427906036376953, "learning_rate": 2.1661622452739506e-06, "loss": 0.0078, "step": 273200 }, { "epoch": 174.13001912045888, "grad_norm": 0.043488673865795135, "learning_rate": 2.164557831470615e-06, "loss": 0.0029, "step": 273210 }, { "epoch": 174.1363926067559, "grad_norm": 0.22413760423660278, "learning_rate": 2.1629539989130643e-06, "loss": 0.0045, "step": 273220 }, { "epoch": 174.1427660930529, "grad_norm": 0.2708325684070587, "learning_rate": 2.1613507476207938e-06, "loss": 0.0066, "step": 273230 }, { "epoch": 174.1491395793499, "grad_norm": 0.1335286945104599, "learning_rate": 2.159748077613283e-06, "loss": 0.0044, "step": 273240 }, { "epoch": 174.15551306564691, "grad_norm": 0.05977649986743927, "learning_rate": 2.1581459889099985e-06, "loss": 0.0055, "step": 273250 }, { "epoch": 174.16188655194392, "grad_norm": 0.1344161331653595, "learning_rate": 2.1565444815304035e-06, "loss": 0.0025, "step": 273260 }, { "epoch": 174.16826003824093, "grad_norm": 0.16337323188781738, "learning_rate": 2.154943555493971e-06, "loss": 0.0104, "step": 273270 }, { "epoch": 174.1746335245379, "grad_norm": 0.13366316258907318, "learning_rate": 2.153343210820147e-06, "loss": 0.0056, "step": 273280 }, { "epoch": 174.18100701083492, "grad_norm": 0.121183842420578, "learning_rate": 2.151743447528376e-06, "loss": 0.0099, "step": 273290 }, { "epoch": 174.18738049713193, "grad_norm": 0.5867923498153687, "learning_rate": 2.1501442656380986e-06, "loss": 0.006, "step": 273300 }, { "epoch": 174.19375398342893, "grad_norm": 0.08859261870384216, "learning_rate": 2.148545665168744e-06, "loss": 0.0043, "step": 273310 }, { "epoch": 174.20012746972594, "grad_norm": 0.09174414724111557, "learning_rate": 2.1469476461397464e-06, "loss": 0.0033, "step": 273320 }, { "epoch": 174.20650095602295, "grad_norm": 0.272705614566803, "learning_rate": 2.1453502085705123e-06, "loss": 0.0041, "step": 273330 }, { "epoch": 174.21287444231996, "grad_norm": 0.15269261598587036, "learning_rate": 2.143753352480449e-06, "loss": 0.004, "step": 273340 }, { "epoch": 174.21924792861697, "grad_norm": 0.1658337414264679, "learning_rate": 2.142157077888979e-06, "loss": 0.0045, "step": 273350 }, { "epoch": 174.22562141491395, "grad_norm": 0.40724053978919983, "learning_rate": 2.140561384815476e-06, "loss": 0.0055, "step": 273360 }, { "epoch": 174.23199490121095, "grad_norm": 0.08837446570396423, "learning_rate": 2.1389662732793414e-06, "loss": 0.0033, "step": 273370 }, { "epoch": 174.23836838750796, "grad_norm": 0.1328115016222, "learning_rate": 2.1373717432999485e-06, "loss": 0.0062, "step": 273380 }, { "epoch": 174.24474187380497, "grad_norm": 0.39661920070648193, "learning_rate": 2.135777794896693e-06, "loss": 0.0068, "step": 273390 }, { "epoch": 174.25111536010198, "grad_norm": 0.2797715663909912, "learning_rate": 2.1341844280889144e-06, "loss": 0.0037, "step": 273400 }, { "epoch": 174.25748884639899, "grad_norm": 0.25838011503219604, "learning_rate": 2.132591642895998e-06, "loss": 0.0053, "step": 273410 }, { "epoch": 174.263862332696, "grad_norm": 0.1305437982082367, "learning_rate": 2.1309994393372832e-06, "loss": 0.0098, "step": 273420 }, { "epoch": 174.270235818993, "grad_norm": 0.061593640595674515, "learning_rate": 2.129407817432122e-06, "loss": 0.0058, "step": 273430 }, { "epoch": 174.27660930528998, "grad_norm": 0.09161899983882904, "learning_rate": 2.127816777199848e-06, "loss": 0.0044, "step": 273440 }, { "epoch": 174.282982791587, "grad_norm": 0.15213841199874878, "learning_rate": 2.1262263186598074e-06, "loss": 0.0032, "step": 273450 }, { "epoch": 174.289356277884, "grad_norm": 0.23842699825763702, "learning_rate": 2.1246364418313123e-06, "loss": 0.0174, "step": 273460 }, { "epoch": 174.295729764181, "grad_norm": 0.43927255272865295, "learning_rate": 2.1230471467336923e-06, "loss": 0.0037, "step": 273470 }, { "epoch": 174.302103250478, "grad_norm": 0.3616434335708618, "learning_rate": 2.1214584333862418e-06, "loss": 0.0071, "step": 273480 }, { "epoch": 174.30847673677502, "grad_norm": 0.46674060821533203, "learning_rate": 2.1198703018082854e-06, "loss": 0.0054, "step": 273490 }, { "epoch": 174.31485022307203, "grad_norm": 0.1764763593673706, "learning_rate": 2.118282752019107e-06, "loss": 0.0051, "step": 273500 }, { "epoch": 174.32122370936904, "grad_norm": 0.11812052875757217, "learning_rate": 2.116695784038003e-06, "loss": 0.0061, "step": 273510 }, { "epoch": 174.32759719566602, "grad_norm": 0.24129360914230347, "learning_rate": 2.115109397884252e-06, "loss": 0.0046, "step": 273520 }, { "epoch": 174.33397068196302, "grad_norm": 0.09591636061668396, "learning_rate": 2.113523593577138e-06, "loss": 0.0036, "step": 273530 }, { "epoch": 174.34034416826003, "grad_norm": 0.23651784658432007, "learning_rate": 2.1119383711359187e-06, "loss": 0.0071, "step": 273540 }, { "epoch": 174.34671765455704, "grad_norm": 0.07975887507200241, "learning_rate": 2.110353730579867e-06, "loss": 0.004, "step": 273550 }, { "epoch": 174.35309114085405, "grad_norm": 0.09688763320446014, "learning_rate": 2.1087696719282234e-06, "loss": 0.0046, "step": 273560 }, { "epoch": 174.35946462715106, "grad_norm": 0.22486229240894318, "learning_rate": 2.1071861952002614e-06, "loss": 0.0048, "step": 273570 }, { "epoch": 174.36583811344806, "grad_norm": 0.11126704514026642, "learning_rate": 2.1056033004151875e-06, "loss": 0.0047, "step": 273580 }, { "epoch": 174.37221159974507, "grad_norm": 0.17915402352809906, "learning_rate": 2.1040209875922646e-06, "loss": 0.0055, "step": 273590 }, { "epoch": 174.37858508604205, "grad_norm": 0.09436215460300446, "learning_rate": 2.1024392567507046e-06, "loss": 0.0135, "step": 273600 }, { "epoch": 174.38495857233906, "grad_norm": 0.1847650110721588, "learning_rate": 2.100858107909731e-06, "loss": 0.0111, "step": 273610 }, { "epoch": 174.39133205863607, "grad_norm": 0.06212270259857178, "learning_rate": 2.099277541088551e-06, "loss": 0.0029, "step": 273620 }, { "epoch": 174.39770554493307, "grad_norm": 0.20943816006183624, "learning_rate": 2.0976975563063827e-06, "loss": 0.0087, "step": 273630 }, { "epoch": 174.40407903123008, "grad_norm": 0.2110956907272339, "learning_rate": 2.096118153582416e-06, "loss": 0.0091, "step": 273640 }, { "epoch": 174.4104525175271, "grad_norm": 0.2327495813369751, "learning_rate": 2.094539332935841e-06, "loss": 0.0173, "step": 273650 }, { "epoch": 174.4168260038241, "grad_norm": 0.0716957226395607, "learning_rate": 2.0929610943858314e-06, "loss": 0.0067, "step": 273660 }, { "epoch": 174.4231994901211, "grad_norm": 0.25516390800476074, "learning_rate": 2.0913834379515995e-06, "loss": 0.0049, "step": 273670 }, { "epoch": 174.42957297641811, "grad_norm": 0.0853210985660553, "learning_rate": 2.089806363652269e-06, "loss": 0.0065, "step": 273680 }, { "epoch": 174.4359464627151, "grad_norm": 0.18366765975952148, "learning_rate": 2.0882298715070413e-06, "loss": 0.0095, "step": 273690 }, { "epoch": 174.4423199490121, "grad_norm": 0.357339471578598, "learning_rate": 2.0866539615350458e-06, "loss": 0.0082, "step": 273700 }, { "epoch": 174.4486934353091, "grad_norm": 0.12970216572284698, "learning_rate": 2.085078633755455e-06, "loss": 0.0048, "step": 273710 }, { "epoch": 174.45506692160612, "grad_norm": 0.1540210098028183, "learning_rate": 2.083503888187388e-06, "loss": 0.0053, "step": 273720 }, { "epoch": 174.46144040790313, "grad_norm": 0.14359676837921143, "learning_rate": 2.08192972484999e-06, "loss": 0.0036, "step": 273730 }, { "epoch": 174.46781389420013, "grad_norm": 0.1288544088602066, "learning_rate": 2.0803561437623964e-06, "loss": 0.0055, "step": 273740 }, { "epoch": 174.47418738049714, "grad_norm": 0.14659681916236877, "learning_rate": 2.0787831449437135e-06, "loss": 0.0039, "step": 273750 }, { "epoch": 174.48056086679415, "grad_norm": 0.1622399538755417, "learning_rate": 2.0772107284130537e-06, "loss": 0.009, "step": 273760 }, { "epoch": 174.48693435309113, "grad_norm": 0.13049942255020142, "learning_rate": 2.075638894189541e-06, "loss": 0.0027, "step": 273770 }, { "epoch": 174.49330783938814, "grad_norm": 0.09492146968841553, "learning_rate": 2.07406764229226e-06, "loss": 0.0054, "step": 273780 }, { "epoch": 174.49968132568515, "grad_norm": 0.08687897771596909, "learning_rate": 2.0724969727403065e-06, "loss": 0.0088, "step": 273790 }, { "epoch": 174.50605481198215, "grad_norm": 0.09402405470609665, "learning_rate": 2.0709268855527597e-06, "loss": 0.0031, "step": 273800 }, { "epoch": 174.51242829827916, "grad_norm": 0.46440067887306213, "learning_rate": 2.0693573807487153e-06, "loss": 0.0054, "step": 273810 }, { "epoch": 174.51880178457617, "grad_norm": 0.2231871634721756, "learning_rate": 2.0677884583472253e-06, "loss": 0.0038, "step": 273820 }, { "epoch": 174.52517527087318, "grad_norm": 0.400174081325531, "learning_rate": 2.066220118367368e-06, "loss": 0.0052, "step": 273830 }, { "epoch": 174.53154875717019, "grad_norm": 0.18241441249847412, "learning_rate": 2.0646523608281787e-06, "loss": 0.0045, "step": 273840 }, { "epoch": 174.53792224346716, "grad_norm": 0.11302852630615234, "learning_rate": 2.0630851857487366e-06, "loss": 0.0054, "step": 273850 }, { "epoch": 174.54429572976417, "grad_norm": 0.3091873526573181, "learning_rate": 2.0615185931480653e-06, "loss": 0.0081, "step": 273860 }, { "epoch": 174.55066921606118, "grad_norm": 0.07542041689157486, "learning_rate": 2.0599525830452047e-06, "loss": 0.0106, "step": 273870 }, { "epoch": 174.5570427023582, "grad_norm": 0.11909458786249161, "learning_rate": 2.0583871554591738e-06, "loss": 0.0047, "step": 273880 }, { "epoch": 174.5634161886552, "grad_norm": 0.07812432944774628, "learning_rate": 2.0568223104090125e-06, "loss": 0.0085, "step": 273890 }, { "epoch": 174.5697896749522, "grad_norm": 0.16436876356601715, "learning_rate": 2.0552580479137275e-06, "loss": 0.0069, "step": 273900 }, { "epoch": 174.5761631612492, "grad_norm": 0.297226220369339, "learning_rate": 2.053694367992326e-06, "loss": 0.0044, "step": 273910 }, { "epoch": 174.58253664754622, "grad_norm": 0.11736738681793213, "learning_rate": 2.052131270663804e-06, "loss": 0.0036, "step": 273920 }, { "epoch": 174.5889101338432, "grad_norm": 0.10593461990356445, "learning_rate": 2.0505687559471465e-06, "loss": 0.0133, "step": 273930 }, { "epoch": 174.5952836201402, "grad_norm": 0.22705978155136108, "learning_rate": 2.0490068238613657e-06, "loss": 0.0033, "step": 273940 }, { "epoch": 174.60165710643722, "grad_norm": 0.08452766388654709, "learning_rate": 2.047445474425419e-06, "loss": 0.0027, "step": 273950 }, { "epoch": 174.60803059273422, "grad_norm": 0.4793698787689209, "learning_rate": 2.0458847076582855e-06, "loss": 0.0063, "step": 273960 }, { "epoch": 174.61440407903123, "grad_norm": 0.07953763753175735, "learning_rate": 2.0443245235789222e-06, "loss": 0.006, "step": 273970 }, { "epoch": 174.62077756532824, "grad_norm": 0.2398182451725006, "learning_rate": 2.0427649222063027e-06, "loss": 0.0026, "step": 273980 }, { "epoch": 174.62715105162525, "grad_norm": 0.10045061260461807, "learning_rate": 2.041205903559368e-06, "loss": 0.0083, "step": 273990 }, { "epoch": 174.63352453792226, "grad_norm": 0.10105215013027191, "learning_rate": 2.0396474676570577e-06, "loss": 0.0028, "step": 274000 }, { "epoch": 174.63989802421924, "grad_norm": 0.20501708984375, "learning_rate": 2.0380896145183124e-06, "loss": 0.0037, "step": 274010 }, { "epoch": 174.64627151051624, "grad_norm": 0.20322489738464355, "learning_rate": 2.036532344162068e-06, "loss": 0.0069, "step": 274020 }, { "epoch": 174.65264499681325, "grad_norm": 0.29902738332748413, "learning_rate": 2.034975656607235e-06, "loss": 0.0033, "step": 274030 }, { "epoch": 174.65901848311026, "grad_norm": 0.08109372109174728, "learning_rate": 2.0334195518727393e-06, "loss": 0.0033, "step": 274040 }, { "epoch": 174.66539196940727, "grad_norm": 0.2530292868614197, "learning_rate": 2.031864029977476e-06, "loss": 0.0053, "step": 274050 }, { "epoch": 174.67176545570427, "grad_norm": 0.409005731344223, "learning_rate": 2.0303090909403633e-06, "loss": 0.0046, "step": 274060 }, { "epoch": 174.67813894200128, "grad_norm": 0.3460387885570526, "learning_rate": 2.0287547347802803e-06, "loss": 0.0066, "step": 274070 }, { "epoch": 174.6845124282983, "grad_norm": 0.13095414638519287, "learning_rate": 2.0272009615161237e-06, "loss": 0.0054, "step": 274080 }, { "epoch": 174.69088591459527, "grad_norm": 0.22820571064949036, "learning_rate": 2.0256477711667608e-06, "loss": 0.004, "step": 274090 }, { "epoch": 174.69725940089228, "grad_norm": 0.11262455582618713, "learning_rate": 2.024095163751083e-06, "loss": 0.0055, "step": 274100 }, { "epoch": 174.7036328871893, "grad_norm": 0.1273442506790161, "learning_rate": 2.0225431392879412e-06, "loss": 0.0043, "step": 274110 }, { "epoch": 174.7100063734863, "grad_norm": 0.1816536784172058, "learning_rate": 2.0209916977961983e-06, "loss": 0.0157, "step": 274120 }, { "epoch": 174.7163798597833, "grad_norm": 0.18337921798229218, "learning_rate": 2.0194408392947007e-06, "loss": 0.0046, "step": 274130 }, { "epoch": 174.7227533460803, "grad_norm": 0.22678369283676147, "learning_rate": 2.0178905638023104e-06, "loss": 0.0068, "step": 274140 }, { "epoch": 174.72912683237732, "grad_norm": 0.07876009494066238, "learning_rate": 2.0163408713378406e-06, "loss": 0.004, "step": 274150 }, { "epoch": 174.73550031867433, "grad_norm": 0.16675293445587158, "learning_rate": 2.0147917619201427e-06, "loss": 0.0053, "step": 274160 }, { "epoch": 174.7418738049713, "grad_norm": 0.1723247915506363, "learning_rate": 2.0132432355680244e-06, "loss": 0.004, "step": 274170 }, { "epoch": 174.7482472912683, "grad_norm": 0.23964500427246094, "learning_rate": 2.011695292300314e-06, "loss": 0.0022, "step": 274180 }, { "epoch": 174.75462077756532, "grad_norm": 0.1337817758321762, "learning_rate": 2.010147932135803e-06, "loss": 0.0075, "step": 274190 }, { "epoch": 174.76099426386233, "grad_norm": 0.06666671484708786, "learning_rate": 2.0086011550933093e-06, "loss": 0.0102, "step": 274200 }, { "epoch": 174.76736775015934, "grad_norm": 0.07512985169887543, "learning_rate": 2.007054961191629e-06, "loss": 0.0047, "step": 274210 }, { "epoch": 174.77374123645635, "grad_norm": 0.19467902183532715, "learning_rate": 2.0055093504495415e-06, "loss": 0.0035, "step": 274220 }, { "epoch": 174.78011472275335, "grad_norm": 0.2237577885389328, "learning_rate": 2.003964322885821e-06, "loss": 0.0051, "step": 274230 }, { "epoch": 174.78648820905036, "grad_norm": 0.10638105124235153, "learning_rate": 2.0024198785192685e-06, "loss": 0.0033, "step": 274240 }, { "epoch": 174.79286169534734, "grad_norm": 0.25440582633018494, "learning_rate": 2.000876017368614e-06, "loss": 0.0074, "step": 274250 }, { "epoch": 174.79923518164435, "grad_norm": 0.24824173748493195, "learning_rate": 1.9993327394526416e-06, "loss": 0.0043, "step": 274260 }, { "epoch": 174.80560866794136, "grad_norm": 0.08335942029953003, "learning_rate": 1.997790044790088e-06, "loss": 0.0026, "step": 274270 }, { "epoch": 174.81198215423836, "grad_norm": 0.39347290992736816, "learning_rate": 1.996247933399725e-06, "loss": 0.0178, "step": 274280 }, { "epoch": 174.81835564053537, "grad_norm": 0.10592269152402878, "learning_rate": 1.9947064053002562e-06, "loss": 0.0042, "step": 274290 }, { "epoch": 174.82472912683238, "grad_norm": 0.25440075993537903, "learning_rate": 1.993165460510432e-06, "loss": 0.0056, "step": 274300 }, { "epoch": 174.8311026131294, "grad_norm": 0.13739362359046936, "learning_rate": 1.9916250990489824e-06, "loss": 0.0064, "step": 274310 }, { "epoch": 174.8374760994264, "grad_norm": 0.14957068860530853, "learning_rate": 1.9900853209346093e-06, "loss": 0.002, "step": 274320 }, { "epoch": 174.8438495857234, "grad_norm": 0.11128990352153778, "learning_rate": 1.988546126186025e-06, "loss": 0.0074, "step": 274330 }, { "epoch": 174.85022307202038, "grad_norm": 0.13964758813381195, "learning_rate": 1.9870075148219368e-06, "loss": 0.0043, "step": 274340 }, { "epoch": 174.8565965583174, "grad_norm": 0.25706154108047485, "learning_rate": 1.985469486861041e-06, "loss": 0.0146, "step": 274350 }, { "epoch": 174.8629700446144, "grad_norm": 0.2610946595668793, "learning_rate": 1.9839320423220285e-06, "loss": 0.0034, "step": 274360 }, { "epoch": 174.8693435309114, "grad_norm": 0.08072300255298615, "learning_rate": 1.9823951812235676e-06, "loss": 0.0029, "step": 274370 }, { "epoch": 174.87571701720842, "grad_norm": 0.21302884817123413, "learning_rate": 1.980858903584348e-06, "loss": 0.0047, "step": 274380 }, { "epoch": 174.88209050350542, "grad_norm": 0.07598202675580978, "learning_rate": 1.9793232094230286e-06, "loss": 0.0116, "step": 274390 }, { "epoch": 174.88846398980243, "grad_norm": 0.11135736852884293, "learning_rate": 1.977788098758271e-06, "loss": 0.0043, "step": 274400 }, { "epoch": 174.89483747609944, "grad_norm": 0.05119563639163971, "learning_rate": 1.9762535716087273e-06, "loss": 0.0034, "step": 274410 }, { "epoch": 174.90121096239642, "grad_norm": 0.14300604164600372, "learning_rate": 1.9747196279930492e-06, "loss": 0.0188, "step": 274420 }, { "epoch": 174.90758444869343, "grad_norm": 0.14497160911560059, "learning_rate": 1.9731862679298663e-06, "loss": 0.0078, "step": 274430 }, { "epoch": 174.91395793499044, "grad_norm": 0.1277652084827423, "learning_rate": 1.9716534914378136e-06, "loss": 0.0037, "step": 274440 }, { "epoch": 174.92033142128744, "grad_norm": 0.3568483889102936, "learning_rate": 1.9701212985355154e-06, "loss": 0.0069, "step": 274450 }, { "epoch": 174.92670490758445, "grad_norm": 0.07867226749658585, "learning_rate": 1.9685896892416067e-06, "loss": 0.0019, "step": 274460 }, { "epoch": 174.93307839388146, "grad_norm": 0.17153967916965485, "learning_rate": 1.967058663574667e-06, "loss": 0.0061, "step": 274470 }, { "epoch": 174.93945188017847, "grad_norm": 0.37027430534362793, "learning_rate": 1.96552822155332e-06, "loss": 0.0111, "step": 274480 }, { "epoch": 174.94582536647547, "grad_norm": 0.1311671882867813, "learning_rate": 1.9639983631961566e-06, "loss": 0.0063, "step": 274490 }, { "epoch": 174.95219885277245, "grad_norm": 0.08102986961603165, "learning_rate": 1.962469088521768e-06, "loss": 0.0051, "step": 274500 }, { "epoch": 174.95857233906946, "grad_norm": 0.10261695086956024, "learning_rate": 1.9609403975487275e-06, "loss": 0.0061, "step": 274510 }, { "epoch": 174.96494582536647, "grad_norm": 0.14586424827575684, "learning_rate": 1.9594122902956257e-06, "loss": 0.0042, "step": 274520 }, { "epoch": 174.97131931166348, "grad_norm": 0.09196797758340836, "learning_rate": 1.9578847667810206e-06, "loss": 0.0105, "step": 274530 }, { "epoch": 174.9776927979605, "grad_norm": 0.12426917999982834, "learning_rate": 1.9563578270234806e-06, "loss": 0.0169, "step": 274540 }, { "epoch": 174.9840662842575, "grad_norm": 0.5025028586387634, "learning_rate": 1.9548314710415404e-06, "loss": 0.0086, "step": 274550 }, { "epoch": 174.9904397705545, "grad_norm": 0.3020811975002289, "learning_rate": 1.953305698853769e-06, "loss": 0.0076, "step": 274560 }, { "epoch": 174.9968132568515, "grad_norm": 0.175157830119133, "learning_rate": 1.9517805104786955e-06, "loss": 0.0048, "step": 274570 }, { "epoch": 175.0031867431485, "grad_norm": 0.0986010804772377, "learning_rate": 1.9502559059348557e-06, "loss": 0.0065, "step": 274580 }, { "epoch": 175.0095602294455, "grad_norm": 0.0712924525141716, "learning_rate": 1.948731885240768e-06, "loss": 0.0142, "step": 274590 }, { "epoch": 175.0159337157425, "grad_norm": 0.10697848349809647, "learning_rate": 1.947208448414961e-06, "loss": 0.0059, "step": 274600 }, { "epoch": 175.0223072020395, "grad_norm": 0.3506622016429901, "learning_rate": 1.945685595475938e-06, "loss": 0.004, "step": 274610 }, { "epoch": 175.02868068833652, "grad_norm": 0.330706924200058, "learning_rate": 1.9441633264422055e-06, "loss": 0.0048, "step": 274620 }, { "epoch": 175.03505417463353, "grad_norm": 0.18329060077667236, "learning_rate": 1.9426416413322656e-06, "loss": 0.0062, "step": 274630 }, { "epoch": 175.04142766093054, "grad_norm": 0.1966279149055481, "learning_rate": 1.941120540164593e-06, "loss": 0.007, "step": 274640 }, { "epoch": 175.04780114722755, "grad_norm": 0.08370765298604965, "learning_rate": 1.9396000229576884e-06, "loss": 0.0219, "step": 274650 }, { "epoch": 175.05417463352453, "grad_norm": 0.07409495115280151, "learning_rate": 1.938080089730021e-06, "loss": 0.0031, "step": 274660 }, { "epoch": 175.06054811982153, "grad_norm": 0.30588480830192566, "learning_rate": 1.9365607405000596e-06, "loss": 0.0067, "step": 274670 }, { "epoch": 175.06692160611854, "grad_norm": 0.4093480706214905, "learning_rate": 1.935041975286256e-06, "loss": 0.01, "step": 274680 }, { "epoch": 175.07329509241555, "grad_norm": 0.13653390109539032, "learning_rate": 1.9335237941070784e-06, "loss": 0.0051, "step": 274690 }, { "epoch": 175.07966857871256, "grad_norm": 0.4086690843105316, "learning_rate": 1.9320061969809733e-06, "loss": 0.0068, "step": 274700 }, { "epoch": 175.08604206500956, "grad_norm": 0.5088330507278442, "learning_rate": 1.930489183926376e-06, "loss": 0.0128, "step": 274710 }, { "epoch": 175.09241555130657, "grad_norm": 0.17936040461063385, "learning_rate": 1.9289727549617108e-06, "loss": 0.008, "step": 274720 }, { "epoch": 175.09878903760358, "grad_norm": 0.0893910676240921, "learning_rate": 1.9274569101054187e-06, "loss": 0.0078, "step": 274730 }, { "epoch": 175.10516252390056, "grad_norm": 0.08355818688869476, "learning_rate": 1.9259416493759173e-06, "loss": 0.008, "step": 274740 }, { "epoch": 175.11153601019757, "grad_norm": 0.26890867948532104, "learning_rate": 1.9244269727916096e-06, "loss": 0.0074, "step": 274750 }, { "epoch": 175.11790949649458, "grad_norm": 0.09547010809183121, "learning_rate": 1.922912880370903e-06, "loss": 0.0052, "step": 274760 }, { "epoch": 175.12428298279158, "grad_norm": 0.3453105390071869, "learning_rate": 1.9213993721322043e-06, "loss": 0.0079, "step": 274770 }, { "epoch": 175.1306564690886, "grad_norm": 0.0937909260392189, "learning_rate": 1.919886448093894e-06, "loss": 0.0057, "step": 274780 }, { "epoch": 175.1370299553856, "grad_norm": 0.20827116072177887, "learning_rate": 1.9183741082743567e-06, "loss": 0.011, "step": 274790 }, { "epoch": 175.1434034416826, "grad_norm": 0.13633866608142853, "learning_rate": 1.916862352691967e-06, "loss": 0.0084, "step": 274800 }, { "epoch": 175.14977692797962, "grad_norm": 0.279620498418808, "learning_rate": 1.9153511813651103e-06, "loss": 0.0032, "step": 274810 }, { "epoch": 175.1561504142766, "grad_norm": 0.12400998175144196, "learning_rate": 1.9138405943121218e-06, "loss": 0.005, "step": 274820 }, { "epoch": 175.1625239005736, "grad_norm": 0.3955395221710205, "learning_rate": 1.912330591551381e-06, "loss": 0.0077, "step": 274830 }, { "epoch": 175.1688973868706, "grad_norm": 0.2608644664287567, "learning_rate": 1.9108211731012183e-06, "loss": 0.011, "step": 274840 }, { "epoch": 175.17527087316762, "grad_norm": 0.11162590235471725, "learning_rate": 1.9093123389799906e-06, "loss": 0.0168, "step": 274850 }, { "epoch": 175.18164435946463, "grad_norm": 0.5271188616752625, "learning_rate": 1.907804089206011e-06, "loss": 0.0078, "step": 274860 }, { "epoch": 175.18801784576164, "grad_norm": 0.06443272531032562, "learning_rate": 1.9062964237976322e-06, "loss": 0.0053, "step": 274870 }, { "epoch": 175.19439133205864, "grad_norm": 0.11076301336288452, "learning_rate": 1.9047893427731501e-06, "loss": 0.0058, "step": 274880 }, { "epoch": 175.20076481835565, "grad_norm": 0.17207011580467224, "learning_rate": 1.9032828461508946e-06, "loss": 0.0029, "step": 274890 }, { "epoch": 175.20713830465266, "grad_norm": 0.10474450141191483, "learning_rate": 1.9017769339491508e-06, "loss": 0.0048, "step": 274900 }, { "epoch": 175.21351179094964, "grad_norm": 0.0966150164604187, "learning_rate": 1.900271606186238e-06, "loss": 0.0076, "step": 274910 }, { "epoch": 175.21988527724665, "grad_norm": 0.3398136496543884, "learning_rate": 1.8987668628804412e-06, "loss": 0.0071, "step": 274920 }, { "epoch": 175.22625876354365, "grad_norm": 0.10009213536977768, "learning_rate": 1.8972627040500402e-06, "loss": 0.0043, "step": 274930 }, { "epoch": 175.23263224984066, "grad_norm": 0.13086828589439392, "learning_rate": 1.8957591297133093e-06, "loss": 0.0096, "step": 274940 }, { "epoch": 175.23900573613767, "grad_norm": 0.08056506514549255, "learning_rate": 1.8942561398885338e-06, "loss": 0.0027, "step": 274950 }, { "epoch": 175.24537922243468, "grad_norm": 0.08861684799194336, "learning_rate": 1.892753734593955e-06, "loss": 0.0042, "step": 274960 }, { "epoch": 175.2517527087317, "grad_norm": 0.07270756363868713, "learning_rate": 1.8912519138478412e-06, "loss": 0.0059, "step": 274970 }, { "epoch": 175.2581261950287, "grad_norm": 0.3415723443031311, "learning_rate": 1.8897506776684337e-06, "loss": 0.0041, "step": 274980 }, { "epoch": 175.26449968132567, "grad_norm": 0.13215404748916626, "learning_rate": 1.8882500260739955e-06, "loss": 0.0079, "step": 274990 }, { "epoch": 175.27087316762268, "grad_norm": 0.18855176866054535, "learning_rate": 1.8867499590827286e-06, "loss": 0.0068, "step": 275000 }, { "epoch": 175.2772466539197, "grad_norm": 0.07858499884605408, "learning_rate": 1.8852504767128853e-06, "loss": 0.0022, "step": 275010 }, { "epoch": 175.2836201402167, "grad_norm": 0.42222583293914795, "learning_rate": 1.8837515789826677e-06, "loss": 0.0063, "step": 275020 }, { "epoch": 175.2899936265137, "grad_norm": 0.1304078996181488, "learning_rate": 1.882253265910311e-06, "loss": 0.0031, "step": 275030 }, { "epoch": 175.2963671128107, "grad_norm": 0.08573833853006363, "learning_rate": 1.880755537513995e-06, "loss": 0.005, "step": 275040 }, { "epoch": 175.30274059910772, "grad_norm": 0.2484629899263382, "learning_rate": 1.879258393811939e-06, "loss": 0.0034, "step": 275050 }, { "epoch": 175.30911408540473, "grad_norm": 0.08860594779253006, "learning_rate": 1.8777618348223225e-06, "loss": 0.0028, "step": 275060 }, { "epoch": 175.3154875717017, "grad_norm": 0.10515493899583817, "learning_rate": 1.8762658605633366e-06, "loss": 0.0027, "step": 275070 }, { "epoch": 175.32186105799872, "grad_norm": 0.044670794159173965, "learning_rate": 1.8747704710531555e-06, "loss": 0.0073, "step": 275080 }, { "epoch": 175.32823454429573, "grad_norm": 0.3997303545475006, "learning_rate": 1.8732756663099539e-06, "loss": 0.0126, "step": 275090 }, { "epoch": 175.33460803059273, "grad_norm": 0.1324145495891571, "learning_rate": 1.871781446351889e-06, "loss": 0.0048, "step": 275100 }, { "epoch": 175.34098151688974, "grad_norm": 0.518002986907959, "learning_rate": 1.8702878111971245e-06, "loss": 0.0068, "step": 275110 }, { "epoch": 175.34735500318675, "grad_norm": 0.17404232919216156, "learning_rate": 1.8687947608637958e-06, "loss": 0.0091, "step": 275120 }, { "epoch": 175.35372848948376, "grad_norm": 0.055254094302654266, "learning_rate": 1.8673022953700658e-06, "loss": 0.0041, "step": 275130 }, { "epoch": 175.36010197578076, "grad_norm": 0.260132372379303, "learning_rate": 1.8658104147340484e-06, "loss": 0.0035, "step": 275140 }, { "epoch": 175.36647546207774, "grad_norm": 0.2412724494934082, "learning_rate": 1.864319118973884e-06, "loss": 0.0081, "step": 275150 }, { "epoch": 175.37284894837475, "grad_norm": 0.07422933727502823, "learning_rate": 1.8628284081076864e-06, "loss": 0.0041, "step": 275160 }, { "epoch": 175.37922243467176, "grad_norm": 0.15103279054164886, "learning_rate": 1.861338282153585e-06, "loss": 0.0035, "step": 275170 }, { "epoch": 175.38559592096877, "grad_norm": 0.14032185077667236, "learning_rate": 1.8598487411296605e-06, "loss": 0.0047, "step": 275180 }, { "epoch": 175.39196940726578, "grad_norm": 0.10797774791717529, "learning_rate": 1.8583597850540314e-06, "loss": 0.0034, "step": 275190 }, { "epoch": 175.39834289356278, "grad_norm": 0.5406715273857117, "learning_rate": 1.8568714139447885e-06, "loss": 0.0068, "step": 275200 }, { "epoch": 175.4047163798598, "grad_norm": 0.12635977566242218, "learning_rate": 1.8553836278200065e-06, "loss": 0.0102, "step": 275210 }, { "epoch": 175.4110898661568, "grad_norm": 0.17098656296730042, "learning_rate": 1.8538964266977655e-06, "loss": 0.0059, "step": 275220 }, { "epoch": 175.41746335245378, "grad_norm": 0.20403602719306946, "learning_rate": 1.8524098105961452e-06, "loss": 0.004, "step": 275230 }, { "epoch": 175.4238368387508, "grad_norm": 0.10768961906433105, "learning_rate": 1.850923779533209e-06, "loss": 0.0052, "step": 275240 }, { "epoch": 175.4302103250478, "grad_norm": 0.060229573398828506, "learning_rate": 1.8494383335270038e-06, "loss": 0.0168, "step": 275250 }, { "epoch": 175.4365838113448, "grad_norm": 0.0944160595536232, "learning_rate": 1.8479534725955816e-06, "loss": 0.0032, "step": 275260 }, { "epoch": 175.4429572976418, "grad_norm": 0.15908418595790863, "learning_rate": 1.8464691967569947e-06, "loss": 0.0145, "step": 275270 }, { "epoch": 175.44933078393882, "grad_norm": 0.16993117332458496, "learning_rate": 1.8449855060292676e-06, "loss": 0.0117, "step": 275280 }, { "epoch": 175.45570427023583, "grad_norm": 0.2445690631866455, "learning_rate": 1.8435024004304357e-06, "loss": 0.0037, "step": 275290 }, { "epoch": 175.46207775653284, "grad_norm": 0.10674241185188293, "learning_rate": 1.8420198799785072e-06, "loss": 0.0035, "step": 275300 }, { "epoch": 175.46845124282981, "grad_norm": 0.09347780048847198, "learning_rate": 1.8405379446915173e-06, "loss": 0.0095, "step": 275310 }, { "epoch": 175.47482472912682, "grad_norm": 0.20365147292613983, "learning_rate": 1.8390565945874571e-06, "loss": 0.0031, "step": 275320 }, { "epoch": 175.48119821542383, "grad_norm": 0.6415494084358215, "learning_rate": 1.837575829684335e-06, "loss": 0.01, "step": 275330 }, { "epoch": 175.48757170172084, "grad_norm": 0.16133737564086914, "learning_rate": 1.8360956500001302e-06, "loss": 0.0043, "step": 275340 }, { "epoch": 175.49394518801785, "grad_norm": 0.1766909509897232, "learning_rate": 1.8346160555528458e-06, "loss": 0.0092, "step": 275350 }, { "epoch": 175.50031867431485, "grad_norm": 0.16597358882427216, "learning_rate": 1.8331370463604502e-06, "loss": 0.0038, "step": 275360 }, { "epoch": 175.50669216061186, "grad_norm": 0.07603301107883453, "learning_rate": 1.8316586224409126e-06, "loss": 0.0042, "step": 275370 }, { "epoch": 175.51306564690887, "grad_norm": 0.669227123260498, "learning_rate": 1.8301807838122075e-06, "loss": 0.005, "step": 275380 }, { "epoch": 175.51943913320585, "grad_norm": 0.07873693108558655, "learning_rate": 1.8287035304922761e-06, "loss": 0.0041, "step": 275390 }, { "epoch": 175.52581261950286, "grad_norm": 0.3354049026966095, "learning_rate": 1.8272268624990874e-06, "loss": 0.0058, "step": 275400 }, { "epoch": 175.53218610579987, "grad_norm": 0.34977009892463684, "learning_rate": 1.8257507798505712e-06, "loss": 0.0042, "step": 275410 }, { "epoch": 175.53855959209687, "grad_norm": 0.09507279843091965, "learning_rate": 1.8242752825646692e-06, "loss": 0.0056, "step": 275420 }, { "epoch": 175.54493307839388, "grad_norm": 0.6243453025817871, "learning_rate": 1.8228003706593e-06, "loss": 0.0065, "step": 275430 }, { "epoch": 175.5513065646909, "grad_norm": 0.20713001489639282, "learning_rate": 1.8213260441523993e-06, "loss": 0.0024, "step": 275440 }, { "epoch": 175.5576800509879, "grad_norm": 0.11087159812450409, "learning_rate": 1.8198523030618809e-06, "loss": 0.0051, "step": 275450 }, { "epoch": 175.5640535372849, "grad_norm": 0.30287715792655945, "learning_rate": 1.818379147405641e-06, "loss": 0.0115, "step": 275460 }, { "epoch": 175.57042702358189, "grad_norm": 0.11132431030273438, "learning_rate": 1.8169065772015826e-06, "loss": 0.0034, "step": 275470 }, { "epoch": 175.5768005098789, "grad_norm": 0.051023997366428375, "learning_rate": 1.815434592467602e-06, "loss": 0.0064, "step": 275480 }, { "epoch": 175.5831739961759, "grad_norm": 0.09731710702180862, "learning_rate": 1.8139631932215905e-06, "loss": 0.0029, "step": 275490 }, { "epoch": 175.5895474824729, "grad_norm": 0.10462509840726852, "learning_rate": 1.8124923794814175e-06, "loss": 0.003, "step": 275500 }, { "epoch": 175.59592096876992, "grad_norm": 0.29803231358528137, "learning_rate": 1.8110221512649573e-06, "loss": 0.0057, "step": 275510 }, { "epoch": 175.60229445506693, "grad_norm": 0.15633639693260193, "learning_rate": 1.8095525085900843e-06, "loss": 0.0071, "step": 275520 }, { "epoch": 175.60866794136393, "grad_norm": 0.09688173234462738, "learning_rate": 1.8080834514746403e-06, "loss": 0.0042, "step": 275530 }, { "epoch": 175.61504142766094, "grad_norm": 0.11864329874515533, "learning_rate": 1.8066149799364828e-06, "loss": 0.0063, "step": 275540 }, { "epoch": 175.62141491395795, "grad_norm": 0.32410773634910583, "learning_rate": 1.8051470939934533e-06, "loss": 0.0072, "step": 275550 }, { "epoch": 175.62778840025493, "grad_norm": 0.05083056539297104, "learning_rate": 1.8036797936633986e-06, "loss": 0.0045, "step": 275560 }, { "epoch": 175.63416188655194, "grad_norm": 0.29283303022384644, "learning_rate": 1.8022130789641324e-06, "loss": 0.0079, "step": 275570 }, { "epoch": 175.64053537284894, "grad_norm": 0.16243165731430054, "learning_rate": 1.8007469499134843e-06, "loss": 0.003, "step": 275580 }, { "epoch": 175.64690885914595, "grad_norm": 0.08928197622299194, "learning_rate": 1.7992814065292685e-06, "loss": 0.0125, "step": 275590 }, { "epoch": 175.65328234544296, "grad_norm": 0.1385534405708313, "learning_rate": 1.7978164488292981e-06, "loss": 0.0048, "step": 275600 }, { "epoch": 175.65965583173997, "grad_norm": 0.056295085698366165, "learning_rate": 1.7963520768313591e-06, "loss": 0.0052, "step": 275610 }, { "epoch": 175.66602931803698, "grad_norm": 0.5644482970237732, "learning_rate": 1.7948882905532593e-06, "loss": 0.0075, "step": 275620 }, { "epoch": 175.67240280433398, "grad_norm": 0.08807476609945297, "learning_rate": 1.7934250900127792e-06, "loss": 0.0081, "step": 275630 }, { "epoch": 175.67877629063096, "grad_norm": 0.06414946168661118, "learning_rate": 1.7919624752277042e-06, "loss": 0.0046, "step": 275640 }, { "epoch": 175.68514977692797, "grad_norm": 0.08145014196634293, "learning_rate": 1.7905004462157926e-06, "loss": 0.0032, "step": 275650 }, { "epoch": 175.69152326322498, "grad_norm": 0.24482548236846924, "learning_rate": 1.789039002994819e-06, "loss": 0.0057, "step": 275660 }, { "epoch": 175.697896749522, "grad_norm": 0.15579721331596375, "learning_rate": 1.787578145582547e-06, "loss": 0.0043, "step": 275670 }, { "epoch": 175.704270235819, "grad_norm": 0.11467379331588745, "learning_rate": 1.7861178739967122e-06, "loss": 0.0038, "step": 275680 }, { "epoch": 175.710643722116, "grad_norm": 0.0670858845114708, "learning_rate": 1.7846581882550672e-06, "loss": 0.0043, "step": 275690 }, { "epoch": 175.717017208413, "grad_norm": 0.12399414926767349, "learning_rate": 1.783199088375359e-06, "loss": 0.0034, "step": 275700 }, { "epoch": 175.72339069471002, "grad_norm": 0.1436539739370346, "learning_rate": 1.7817405743752958e-06, "loss": 0.0042, "step": 275710 }, { "epoch": 175.729764181007, "grad_norm": 0.12766535580158234, "learning_rate": 1.780282646272613e-06, "loss": 0.007, "step": 275720 }, { "epoch": 175.736137667304, "grad_norm": 0.2412629872560501, "learning_rate": 1.778825304085019e-06, "loss": 0.0035, "step": 275730 }, { "epoch": 175.74251115360101, "grad_norm": 0.125607430934906, "learning_rate": 1.7773685478302383e-06, "loss": 0.0066, "step": 275740 }, { "epoch": 175.74888463989802, "grad_norm": 0.09349637478590012, "learning_rate": 1.7759123775259455e-06, "loss": 0.0077, "step": 275750 }, { "epoch": 175.75525812619503, "grad_norm": 0.15492604672908783, "learning_rate": 1.7744567931898548e-06, "loss": 0.0043, "step": 275760 }, { "epoch": 175.76163161249204, "grad_norm": 0.09421937167644501, "learning_rate": 1.773001794839646e-06, "loss": 0.0048, "step": 275770 }, { "epoch": 175.76800509878905, "grad_norm": 0.08395714312791824, "learning_rate": 1.7715473824929995e-06, "loss": 0.0025, "step": 275780 }, { "epoch": 175.77437858508605, "grad_norm": 0.15668924152851105, "learning_rate": 1.7700935561675848e-06, "loss": 0.0036, "step": 275790 }, { "epoch": 175.78075207138303, "grad_norm": 0.19329330325126648, "learning_rate": 1.7686403158810705e-06, "loss": 0.0052, "step": 275800 }, { "epoch": 175.78712555768004, "grad_norm": 0.28593674302101135, "learning_rate": 1.767187661651115e-06, "loss": 0.005, "step": 275810 }, { "epoch": 175.79349904397705, "grad_norm": 0.38054755330085754, "learning_rate": 1.7657355934953713e-06, "loss": 0.0094, "step": 275820 }, { "epoch": 175.79987253027406, "grad_norm": 0.20998823642730713, "learning_rate": 1.7642841114314746e-06, "loss": 0.0056, "step": 275830 }, { "epoch": 175.80624601657107, "grad_norm": 0.31031039357185364, "learning_rate": 1.7628332154770832e-06, "loss": 0.0068, "step": 275840 }, { "epoch": 175.81261950286807, "grad_norm": 0.11417978256940842, "learning_rate": 1.7613829056497945e-06, "loss": 0.0036, "step": 275850 }, { "epoch": 175.81899298916508, "grad_norm": 0.18772749602794647, "learning_rate": 1.759933181967255e-06, "loss": 0.0062, "step": 275860 }, { "epoch": 175.8253664754621, "grad_norm": 0.1490863561630249, "learning_rate": 1.7584840444470673e-06, "loss": 0.0026, "step": 275870 }, { "epoch": 175.83173996175907, "grad_norm": 0.20087669789791107, "learning_rate": 1.7570354931068623e-06, "loss": 0.0064, "step": 275880 }, { "epoch": 175.83811344805608, "grad_norm": 0.4685465693473816, "learning_rate": 1.7555875279642087e-06, "loss": 0.008, "step": 275890 }, { "epoch": 175.84448693435309, "grad_norm": 0.15830516815185547, "learning_rate": 1.754140149036726e-06, "loss": 0.0046, "step": 275900 }, { "epoch": 175.8508604206501, "grad_norm": 0.15877991914749146, "learning_rate": 1.7526933563419834e-06, "loss": 0.0056, "step": 275910 }, { "epoch": 175.8572339069471, "grad_norm": 0.13357149064540863, "learning_rate": 1.751247149897578e-06, "loss": 0.0048, "step": 275920 }, { "epoch": 175.8636073932441, "grad_norm": 0.2043164223432541, "learning_rate": 1.7498015297210678e-06, "loss": 0.0038, "step": 275930 }, { "epoch": 175.86998087954112, "grad_norm": 0.11902615427970886, "learning_rate": 1.748356495830028e-06, "loss": 0.0062, "step": 275940 }, { "epoch": 175.87635436583813, "grad_norm": 0.07849596440792084, "learning_rate": 1.7469120482420164e-06, "loss": 0.0035, "step": 275950 }, { "epoch": 175.8827278521351, "grad_norm": 0.06419183313846588, "learning_rate": 1.7454681869745803e-06, "loss": 0.0049, "step": 275960 }, { "epoch": 175.8891013384321, "grad_norm": 0.09776323288679123, "learning_rate": 1.744024912045261e-06, "loss": 0.0062, "step": 275970 }, { "epoch": 175.89547482472912, "grad_norm": 0.13812728226184845, "learning_rate": 1.742582223471606e-06, "loss": 0.0037, "step": 275980 }, { "epoch": 175.90184831102613, "grad_norm": 0.050246454775333405, "learning_rate": 1.7411401212711398e-06, "loss": 0.0066, "step": 275990 }, { "epoch": 175.90822179732314, "grad_norm": 0.27197006344795227, "learning_rate": 1.7396986054613818e-06, "loss": 0.0056, "step": 276000 }, { "epoch": 175.91459528362014, "grad_norm": 0.1113065779209137, "learning_rate": 1.7382576760598458e-06, "loss": 0.0091, "step": 276010 }, { "epoch": 175.92096876991715, "grad_norm": 0.23072966933250427, "learning_rate": 1.7368173330840565e-06, "loss": 0.0151, "step": 276020 }, { "epoch": 175.92734225621416, "grad_norm": 0.2518482208251953, "learning_rate": 1.7353775765515e-06, "loss": 0.0049, "step": 276030 }, { "epoch": 175.93371574251114, "grad_norm": 0.23538914322853088, "learning_rate": 1.7339384064796737e-06, "loss": 0.0103, "step": 276040 }, { "epoch": 175.94008922880815, "grad_norm": 0.32805290818214417, "learning_rate": 1.7324998228860633e-06, "loss": 0.0056, "step": 276050 }, { "epoch": 175.94646271510516, "grad_norm": 0.09572313725948334, "learning_rate": 1.7310618257881605e-06, "loss": 0.0027, "step": 276060 }, { "epoch": 175.95283620140216, "grad_norm": 0.10235362499952316, "learning_rate": 1.729624415203429e-06, "loss": 0.0061, "step": 276070 }, { "epoch": 175.95920968769917, "grad_norm": 0.12744803726673126, "learning_rate": 1.7281875911493328e-06, "loss": 0.0037, "step": 276080 }, { "epoch": 175.96558317399618, "grad_norm": 0.16465498507022858, "learning_rate": 1.7267513536433355e-06, "loss": 0.0088, "step": 276090 }, { "epoch": 175.9719566602932, "grad_norm": 0.21070727705955505, "learning_rate": 1.7253157027028788e-06, "loss": 0.0053, "step": 276100 }, { "epoch": 175.9783301465902, "grad_norm": 0.16451188921928406, "learning_rate": 1.7238806383454209e-06, "loss": 0.0064, "step": 276110 }, { "epoch": 175.98470363288718, "grad_norm": 0.06341242790222168, "learning_rate": 1.722446160588398e-06, "loss": 0.0033, "step": 276120 }, { "epoch": 175.99107711918418, "grad_norm": 0.20330853760242462, "learning_rate": 1.7210122694492293e-06, "loss": 0.0047, "step": 276130 }, { "epoch": 175.9974506054812, "grad_norm": 0.25456634163856506, "learning_rate": 1.7195789649453453e-06, "loss": 0.0055, "step": 276140 }, { "epoch": 176.0038240917782, "grad_norm": 0.3245943784713745, "learning_rate": 1.7181462470941656e-06, "loss": 0.0071, "step": 276150 }, { "epoch": 176.0101975780752, "grad_norm": 0.20628835260868073, "learning_rate": 1.716714115913093e-06, "loss": 0.0047, "step": 276160 }, { "epoch": 176.01657106437222, "grad_norm": 0.18600164353847504, "learning_rate": 1.7152825714195298e-06, "loss": 0.0042, "step": 276170 }, { "epoch": 176.02294455066922, "grad_norm": 0.12381718307733536, "learning_rate": 1.7138516136308624e-06, "loss": 0.0117, "step": 276180 }, { "epoch": 176.02931803696623, "grad_norm": 0.05407378822565079, "learning_rate": 1.7124212425644993e-06, "loss": 0.0027, "step": 276190 }, { "epoch": 176.03569152326324, "grad_norm": 0.3319127857685089, "learning_rate": 1.7109914582378095e-06, "loss": 0.0036, "step": 276200 }, { "epoch": 176.04206500956022, "grad_norm": 0.041496891528367996, "learning_rate": 1.7095622606681627e-06, "loss": 0.0093, "step": 276210 }, { "epoch": 176.04843849585723, "grad_norm": 0.11879713088274002, "learning_rate": 1.7081336498729227e-06, "loss": 0.002, "step": 276220 }, { "epoch": 176.05481198215423, "grad_norm": 0.0980495810508728, "learning_rate": 1.7067056258694648e-06, "loss": 0.0056, "step": 276230 }, { "epoch": 176.06118546845124, "grad_norm": 0.17400796711444855, "learning_rate": 1.7052781886751245e-06, "loss": 0.0034, "step": 276240 }, { "epoch": 176.06755895474825, "grad_norm": 0.0892915204167366, "learning_rate": 1.7038513383072495e-06, "loss": 0.0032, "step": 276250 }, { "epoch": 176.07393244104526, "grad_norm": 0.2017907053232193, "learning_rate": 1.7024250747831816e-06, "loss": 0.0058, "step": 276260 }, { "epoch": 176.08030592734227, "grad_norm": 0.35362645983695984, "learning_rate": 1.7009993981202565e-06, "loss": 0.0101, "step": 276270 }, { "epoch": 176.08667941363927, "grad_norm": 0.2972959578037262, "learning_rate": 1.6995743083357774e-06, "loss": 0.0051, "step": 276280 }, { "epoch": 176.09305289993625, "grad_norm": 0.08610859513282776, "learning_rate": 1.6981498054470856e-06, "loss": 0.0122, "step": 276290 }, { "epoch": 176.09942638623326, "grad_norm": 0.06317649036645889, "learning_rate": 1.6967258894714676e-06, "loss": 0.0036, "step": 276300 }, { "epoch": 176.10579987253027, "grad_norm": 0.07627560943365097, "learning_rate": 1.6953025604262484e-06, "loss": 0.008, "step": 276310 }, { "epoch": 176.11217335882728, "grad_norm": 0.3309505879878998, "learning_rate": 1.6938798183286975e-06, "loss": 0.0065, "step": 276320 }, { "epoch": 176.11854684512429, "grad_norm": 0.22768516838550568, "learning_rate": 1.692457663196123e-06, "loss": 0.0191, "step": 276330 }, { "epoch": 176.1249203314213, "grad_norm": 0.3158848285675049, "learning_rate": 1.6910360950457948e-06, "loss": 0.0054, "step": 276340 }, { "epoch": 176.1312938177183, "grad_norm": 0.18053795397281647, "learning_rate": 1.6896151138949934e-06, "loss": 0.0072, "step": 276350 }, { "epoch": 176.1376673040153, "grad_norm": 0.18740619719028473, "learning_rate": 1.6881947197609716e-06, "loss": 0.0039, "step": 276360 }, { "epoch": 176.1440407903123, "grad_norm": 0.10572820901870728, "learning_rate": 1.68677491266101e-06, "loss": 0.0032, "step": 276370 }, { "epoch": 176.1504142766093, "grad_norm": 0.5496734976768494, "learning_rate": 1.6853556926123394e-06, "loss": 0.0064, "step": 276380 }, { "epoch": 176.1567877629063, "grad_norm": 0.46469154953956604, "learning_rate": 1.683937059632218e-06, "loss": 0.0085, "step": 276390 }, { "epoch": 176.1631612492033, "grad_norm": 0.393588662147522, "learning_rate": 1.6825190137378765e-06, "loss": 0.0044, "step": 276400 }, { "epoch": 176.16953473550032, "grad_norm": 0.2762969136238098, "learning_rate": 1.681101554946557e-06, "loss": 0.0107, "step": 276410 }, { "epoch": 176.17590822179733, "grad_norm": 0.12323260307312012, "learning_rate": 1.6796846832754621e-06, "loss": 0.0037, "step": 276420 }, { "epoch": 176.18228170809434, "grad_norm": 0.09027928113937378, "learning_rate": 1.6782683987418279e-06, "loss": 0.004, "step": 276430 }, { "epoch": 176.18865519439134, "grad_norm": 0.04254785180091858, "learning_rate": 1.6768527013628465e-06, "loss": 0.0045, "step": 276440 }, { "epoch": 176.19502868068832, "grad_norm": 0.13480781018733978, "learning_rate": 1.6754375911557429e-06, "loss": 0.01, "step": 276450 }, { "epoch": 176.20140216698533, "grad_norm": 0.0534277968108654, "learning_rate": 1.6740230681376867e-06, "loss": 0.0104, "step": 276460 }, { "epoch": 176.20777565328234, "grad_norm": 0.09179238229990005, "learning_rate": 1.6726091323258864e-06, "loss": 0.003, "step": 276470 }, { "epoch": 176.21414913957935, "grad_norm": 0.09881261736154556, "learning_rate": 1.6711957837375114e-06, "loss": 0.0041, "step": 276480 }, { "epoch": 176.22052262587636, "grad_norm": 0.160806342959404, "learning_rate": 1.6697830223897371e-06, "loss": 0.0046, "step": 276490 }, { "epoch": 176.22689611217336, "grad_norm": 0.10054619610309601, "learning_rate": 1.6683708482997219e-06, "loss": 0.0039, "step": 276500 }, { "epoch": 176.23326959847037, "grad_norm": 0.06138325110077858, "learning_rate": 1.6669592614846462e-06, "loss": 0.0061, "step": 276510 }, { "epoch": 176.23964308476738, "grad_norm": 0.26037538051605225, "learning_rate": 1.6655482619616414e-06, "loss": 0.0076, "step": 276520 }, { "epoch": 176.24601657106436, "grad_norm": 0.7198311686515808, "learning_rate": 1.6641378497478656e-06, "loss": 0.0052, "step": 276530 }, { "epoch": 176.25239005736137, "grad_norm": 0.37532392144203186, "learning_rate": 1.6627280248604438e-06, "loss": 0.0047, "step": 276540 }, { "epoch": 176.25876354365838, "grad_norm": 0.50913405418396, "learning_rate": 1.6613187873165237e-06, "loss": 0.0071, "step": 276550 }, { "epoch": 176.26513702995538, "grad_norm": 0.08373554795980453, "learning_rate": 1.6599101371332192e-06, "loss": 0.004, "step": 276560 }, { "epoch": 176.2715105162524, "grad_norm": 0.0911097601056099, "learning_rate": 1.6585020743276446e-06, "loss": 0.0042, "step": 276570 }, { "epoch": 176.2778840025494, "grad_norm": 0.17452853918075562, "learning_rate": 1.6570945989169029e-06, "loss": 0.0025, "step": 276580 }, { "epoch": 176.2842574888464, "grad_norm": 0.06627876311540604, "learning_rate": 1.655687710918119e-06, "loss": 0.0036, "step": 276590 }, { "epoch": 176.29063097514342, "grad_norm": 0.5218528509140015, "learning_rate": 1.6542814103483684e-06, "loss": 0.0089, "step": 276600 }, { "epoch": 176.2970044614404, "grad_norm": 0.2919221818447113, "learning_rate": 1.6528756972247427e-06, "loss": 0.0035, "step": 276610 }, { "epoch": 176.3033779477374, "grad_norm": 0.10709366202354431, "learning_rate": 1.651470571564323e-06, "loss": 0.0042, "step": 276620 }, { "epoch": 176.3097514340344, "grad_norm": 0.07940628379583359, "learning_rate": 1.6500660333841955e-06, "loss": 0.0038, "step": 276630 }, { "epoch": 176.31612492033142, "grad_norm": 0.41280558705329895, "learning_rate": 1.648662082701402e-06, "loss": 0.0059, "step": 276640 }, { "epoch": 176.32249840662843, "grad_norm": 0.15187384188175201, "learning_rate": 1.6472587195330236e-06, "loss": 0.0055, "step": 276650 }, { "epoch": 176.32887189292543, "grad_norm": 0.22353307902812958, "learning_rate": 1.645855943896102e-06, "loss": 0.0076, "step": 276660 }, { "epoch": 176.33524537922244, "grad_norm": 0.635593593120575, "learning_rate": 1.6444537558076844e-06, "loss": 0.0098, "step": 276670 }, { "epoch": 176.34161886551945, "grad_norm": 0.095280721783638, "learning_rate": 1.643052155284802e-06, "loss": 0.0056, "step": 276680 }, { "epoch": 176.34799235181643, "grad_norm": 0.0418810173869133, "learning_rate": 1.6416511423445024e-06, "loss": 0.0053, "step": 276690 }, { "epoch": 176.35436583811344, "grad_norm": 0.2549346089363098, "learning_rate": 1.6402507170037994e-06, "loss": 0.0037, "step": 276700 }, { "epoch": 176.36073932441045, "grad_norm": 0.07393168658018112, "learning_rate": 1.6388508792797075e-06, "loss": 0.0067, "step": 276710 }, { "epoch": 176.36711281070745, "grad_norm": 0.17651785910129547, "learning_rate": 1.6374516291892295e-06, "loss": 0.0045, "step": 276720 }, { "epoch": 176.37348629700446, "grad_norm": 0.05120069906115532, "learning_rate": 1.636052966749385e-06, "loss": 0.0055, "step": 276730 }, { "epoch": 176.37985978330147, "grad_norm": 0.4586937129497528, "learning_rate": 1.6346548919771609e-06, "loss": 0.0147, "step": 276740 }, { "epoch": 176.38623326959848, "grad_norm": 0.1840190589427948, "learning_rate": 1.6332574048895434e-06, "loss": 0.0054, "step": 276750 }, { "epoch": 176.39260675589549, "grad_norm": 0.40674111247062683, "learning_rate": 1.6318605055035075e-06, "loss": 0.0096, "step": 276760 }, { "epoch": 176.39898024219247, "grad_norm": 0.40873751044273376, "learning_rate": 1.6304641938360399e-06, "loss": 0.0083, "step": 276770 }, { "epoch": 176.40535372848947, "grad_norm": 0.5901671051979065, "learning_rate": 1.6290684699040993e-06, "loss": 0.0087, "step": 276780 }, { "epoch": 176.41172721478648, "grad_norm": 0.1505087912082672, "learning_rate": 1.6276733337246496e-06, "loss": 0.0051, "step": 276790 }, { "epoch": 176.4181007010835, "grad_norm": 0.1966770440340042, "learning_rate": 1.6262787853146276e-06, "loss": 0.0086, "step": 276800 }, { "epoch": 176.4244741873805, "grad_norm": 0.41214901208877563, "learning_rate": 1.624884824691003e-06, "loss": 0.0071, "step": 276810 }, { "epoch": 176.4308476736775, "grad_norm": 0.09253805130720139, "learning_rate": 1.6234914518707012e-06, "loss": 0.0028, "step": 276820 }, { "epoch": 176.4372211599745, "grad_norm": 0.5277578830718994, "learning_rate": 1.6220986668706528e-06, "loss": 0.0054, "step": 276830 }, { "epoch": 176.44359464627152, "grad_norm": 0.3419092297554016, "learning_rate": 1.6207064697077778e-06, "loss": 0.0147, "step": 276840 }, { "epoch": 176.44996813256853, "grad_norm": 0.08389615267515182, "learning_rate": 1.619314860398996e-06, "loss": 0.0103, "step": 276850 }, { "epoch": 176.4563416188655, "grad_norm": 0.3483833372592926, "learning_rate": 1.6179238389612217e-06, "loss": 0.01, "step": 276860 }, { "epoch": 176.46271510516252, "grad_norm": 0.07819774746894836, "learning_rate": 1.6165334054113523e-06, "loss": 0.005, "step": 276870 }, { "epoch": 176.46908859145952, "grad_norm": 0.4185068905353546, "learning_rate": 1.6151435597662857e-06, "loss": 0.006, "step": 276880 }, { "epoch": 176.47546207775653, "grad_norm": 0.1371045857667923, "learning_rate": 1.6137543020428969e-06, "loss": 0.0097, "step": 276890 }, { "epoch": 176.48183556405354, "grad_norm": 0.06295493990182877, "learning_rate": 1.6123656322580893e-06, "loss": 0.0075, "step": 276900 }, { "epoch": 176.48820905035055, "grad_norm": 0.29193830490112305, "learning_rate": 1.6109775504287217e-06, "loss": 0.0054, "step": 276910 }, { "epoch": 176.49458253664756, "grad_norm": 0.06770236045122147, "learning_rate": 1.6095900565716637e-06, "loss": 0.0056, "step": 276920 }, { "epoch": 176.50095602294456, "grad_norm": 0.08254542946815491, "learning_rate": 1.6082031507037687e-06, "loss": 0.0049, "step": 276930 }, { "epoch": 176.50732950924154, "grad_norm": 0.24575506150722504, "learning_rate": 1.6068168328419009e-06, "loss": 0.0038, "step": 276940 }, { "epoch": 176.51370299553855, "grad_norm": 0.18614822626113892, "learning_rate": 1.6054311030028968e-06, "loss": 0.0042, "step": 276950 }, { "epoch": 176.52007648183556, "grad_norm": 0.22303631901741028, "learning_rate": 1.604045961203604e-06, "loss": 0.006, "step": 276960 }, { "epoch": 176.52644996813257, "grad_norm": 0.283275842666626, "learning_rate": 1.6026614074608315e-06, "loss": 0.0069, "step": 276970 }, { "epoch": 176.53282345442958, "grad_norm": 0.052384283393621445, "learning_rate": 1.6012774417914378e-06, "loss": 0.0028, "step": 276980 }, { "epoch": 176.53919694072658, "grad_norm": 0.165771022439003, "learning_rate": 1.599894064212204e-06, "loss": 0.0087, "step": 276990 }, { "epoch": 176.5455704270236, "grad_norm": 0.2171923667192459, "learning_rate": 1.5985112747399555e-06, "loss": 0.008, "step": 277000 }, { "epoch": 176.5519439133206, "grad_norm": 0.332753449678421, "learning_rate": 1.5971290733914957e-06, "loss": 0.0031, "step": 277010 }, { "epoch": 176.55831739961758, "grad_norm": 0.06422251462936401, "learning_rate": 1.595747460183622e-06, "loss": 0.0047, "step": 277020 }, { "epoch": 176.5646908859146, "grad_norm": 0.18673056364059448, "learning_rate": 1.59436643513311e-06, "loss": 0.0047, "step": 277030 }, { "epoch": 176.5710643722116, "grad_norm": 0.05436268076300621, "learning_rate": 1.5929859982567574e-06, "loss": 0.0038, "step": 277040 }, { "epoch": 176.5774378585086, "grad_norm": 0.0724327340722084, "learning_rate": 1.591606149571323e-06, "loss": 0.0073, "step": 277050 }, { "epoch": 176.5838113448056, "grad_norm": 0.48098692297935486, "learning_rate": 1.5902268890935824e-06, "loss": 0.0043, "step": 277060 }, { "epoch": 176.59018483110262, "grad_norm": 0.19839124381542206, "learning_rate": 1.5888482168402829e-06, "loss": 0.0025, "step": 277070 }, { "epoch": 176.59655831739963, "grad_norm": 0.26652219891548157, "learning_rate": 1.5874701328281948e-06, "loss": 0.0113, "step": 277080 }, { "epoch": 176.60293180369663, "grad_norm": 0.293529212474823, "learning_rate": 1.5860926370740491e-06, "loss": 0.0056, "step": 277090 }, { "epoch": 176.6093052899936, "grad_norm": 0.08972510695457458, "learning_rate": 1.584715729594588e-06, "loss": 0.0047, "step": 277100 }, { "epoch": 176.61567877629062, "grad_norm": 0.2224052995443344, "learning_rate": 1.5833394104065425e-06, "loss": 0.0044, "step": 277110 }, { "epoch": 176.62205226258763, "grad_norm": 0.35721954703330994, "learning_rate": 1.581963679526638e-06, "loss": 0.0044, "step": 277120 }, { "epoch": 176.62842574888464, "grad_norm": 0.12109968811273575, "learning_rate": 1.580588536971589e-06, "loss": 0.0047, "step": 277130 }, { "epoch": 176.63479923518165, "grad_norm": 0.13580960035324097, "learning_rate": 1.5792139827581042e-06, "loss": 0.0066, "step": 277140 }, { "epoch": 176.64117272147865, "grad_norm": 0.3673904240131378, "learning_rate": 1.5778400169028817e-06, "loss": 0.0031, "step": 277150 }, { "epoch": 176.64754620777566, "grad_norm": 0.3315570056438446, "learning_rate": 1.5764666394226356e-06, "loss": 0.0046, "step": 277160 }, { "epoch": 176.65391969407267, "grad_norm": 0.08797069638967514, "learning_rate": 1.5750938503340195e-06, "loss": 0.0081, "step": 277170 }, { "epoch": 176.66029318036965, "grad_norm": 0.11980016529560089, "learning_rate": 1.5737216496537476e-06, "loss": 0.0089, "step": 277180 }, { "epoch": 176.66666666666666, "grad_norm": 0.35164692997932434, "learning_rate": 1.5723500373984734e-06, "loss": 0.0048, "step": 277190 }, { "epoch": 176.67304015296367, "grad_norm": 0.18545228242874146, "learning_rate": 1.5709790135848778e-06, "loss": 0.0045, "step": 277200 }, { "epoch": 176.67941363926067, "grad_norm": 0.14580050110816956, "learning_rate": 1.5696085782295978e-06, "loss": 0.0044, "step": 277210 }, { "epoch": 176.68578712555768, "grad_norm": 0.24122366309165955, "learning_rate": 1.5682387313493086e-06, "loss": 0.0059, "step": 277220 }, { "epoch": 176.6921606118547, "grad_norm": 0.20346349477767944, "learning_rate": 1.566869472960647e-06, "loss": 0.0042, "step": 277230 }, { "epoch": 176.6985340981517, "grad_norm": 0.07641192525625229, "learning_rate": 1.5655008030802497e-06, "loss": 0.0039, "step": 277240 }, { "epoch": 176.7049075844487, "grad_norm": 0.076439268887043, "learning_rate": 1.564132721724737e-06, "loss": 0.0158, "step": 277250 }, { "epoch": 176.71128107074568, "grad_norm": 0.05529626086354256, "learning_rate": 1.5627652289107565e-06, "loss": 0.002, "step": 277260 }, { "epoch": 176.7176545570427, "grad_norm": 0.559586226940155, "learning_rate": 1.5613983246549058e-06, "loss": 0.0047, "step": 277270 }, { "epoch": 176.7240280433397, "grad_norm": 0.17414197325706482, "learning_rate": 1.5600320089737996e-06, "loss": 0.0056, "step": 277280 }, { "epoch": 176.7304015296367, "grad_norm": 0.09139235317707062, "learning_rate": 1.5586662818840303e-06, "loss": 0.0053, "step": 277290 }, { "epoch": 176.73677501593372, "grad_norm": 0.12896183133125305, "learning_rate": 1.5573011434022177e-06, "loss": 0.0054, "step": 277300 }, { "epoch": 176.74314850223072, "grad_norm": 0.21336528658866882, "learning_rate": 1.5559365935449211e-06, "loss": 0.0051, "step": 277310 }, { "epoch": 176.74952198852773, "grad_norm": 0.08669745922088623, "learning_rate": 1.5545726323287379e-06, "loss": 0.0053, "step": 277320 }, { "epoch": 176.75589547482474, "grad_norm": 0.2620621919631958, "learning_rate": 1.553209259770233e-06, "loss": 0.0036, "step": 277330 }, { "epoch": 176.76226896112172, "grad_norm": 0.18316194415092468, "learning_rate": 1.5518464758859874e-06, "loss": 0.004, "step": 277340 }, { "epoch": 176.76864244741873, "grad_norm": 0.22259753942489624, "learning_rate": 1.5504842806925434e-06, "loss": 0.0047, "step": 277350 }, { "epoch": 176.77501593371574, "grad_norm": 0.11987613141536713, "learning_rate": 1.5491226742064603e-06, "loss": 0.0079, "step": 277360 }, { "epoch": 176.78138942001274, "grad_norm": 0.32699429988861084, "learning_rate": 1.5477616564442799e-06, "loss": 0.0051, "step": 277370 }, { "epoch": 176.78776290630975, "grad_norm": 0.19016142189502716, "learning_rate": 1.5464012274225447e-06, "loss": 0.0072, "step": 277380 }, { "epoch": 176.79413639260676, "grad_norm": 0.18845343589782715, "learning_rate": 1.5450413871577808e-06, "loss": 0.0033, "step": 277390 }, { "epoch": 176.80050987890377, "grad_norm": 0.2902399003505707, "learning_rate": 1.543682135666513e-06, "loss": 0.0056, "step": 277400 }, { "epoch": 176.80688336520078, "grad_norm": 0.08449825644493103, "learning_rate": 1.542323472965257e-06, "loss": 0.0051, "step": 277410 }, { "epoch": 176.81325685149778, "grad_norm": 0.23298433423042297, "learning_rate": 1.5409653990705264e-06, "loss": 0.0028, "step": 277420 }, { "epoch": 176.81963033779476, "grad_norm": 0.11054684966802597, "learning_rate": 1.539607913998814e-06, "loss": 0.0084, "step": 277430 }, { "epoch": 176.82600382409177, "grad_norm": 0.11497845500707626, "learning_rate": 1.538251017766623e-06, "loss": 0.0056, "step": 277440 }, { "epoch": 176.83237731038878, "grad_norm": 0.39765968918800354, "learning_rate": 1.536894710390435e-06, "loss": 0.0097, "step": 277450 }, { "epoch": 176.8387507966858, "grad_norm": 0.15032115578651428, "learning_rate": 1.5355389918867368e-06, "loss": 0.0051, "step": 277460 }, { "epoch": 176.8451242829828, "grad_norm": 0.12738831341266632, "learning_rate": 1.5341838622719873e-06, "loss": 0.0065, "step": 277470 }, { "epoch": 176.8514977692798, "grad_norm": 0.3474059998989105, "learning_rate": 1.5328293215626733e-06, "loss": 0.0051, "step": 277480 }, { "epoch": 176.8578712555768, "grad_norm": 0.12211324274539948, "learning_rate": 1.5314753697752427e-06, "loss": 0.0032, "step": 277490 }, { "epoch": 176.86424474187382, "grad_norm": 0.1513955295085907, "learning_rate": 1.530122006926149e-06, "loss": 0.0037, "step": 277500 }, { "epoch": 176.8706182281708, "grad_norm": 0.1327822059392929, "learning_rate": 1.5287692330318293e-06, "loss": 0.0087, "step": 277510 }, { "epoch": 176.8769917144678, "grad_norm": 0.28009235858917236, "learning_rate": 1.5274170481087368e-06, "loss": 0.0056, "step": 277520 }, { "epoch": 176.8833652007648, "grad_norm": 0.39834654331207275, "learning_rate": 1.526065452173292e-06, "loss": 0.0048, "step": 277530 }, { "epoch": 176.88973868706182, "grad_norm": 0.1294848471879959, "learning_rate": 1.524714445241915e-06, "loss": 0.0048, "step": 277540 }, { "epoch": 176.89611217335883, "grad_norm": 0.07060711085796356, "learning_rate": 1.5233640273310257e-06, "loss": 0.006, "step": 277550 }, { "epoch": 176.90248565965584, "grad_norm": 0.045312050729990005, "learning_rate": 1.5220141984570336e-06, "loss": 0.0054, "step": 277560 }, { "epoch": 176.90885914595285, "grad_norm": 0.13335396349430084, "learning_rate": 1.5206649586363363e-06, "loss": 0.0066, "step": 277570 }, { "epoch": 176.91523263224985, "grad_norm": 0.14459702372550964, "learning_rate": 1.5193163078853379e-06, "loss": 0.0039, "step": 277580 }, { "epoch": 176.92160611854683, "grad_norm": 0.20667991042137146, "learning_rate": 1.5179682462204193e-06, "loss": 0.0041, "step": 277590 }, { "epoch": 176.92797960484384, "grad_norm": 0.6021701693534851, "learning_rate": 1.5166207736579562e-06, "loss": 0.0061, "step": 277600 }, { "epoch": 176.93435309114085, "grad_norm": 0.12168300151824951, "learning_rate": 1.5152738902143303e-06, "loss": 0.0047, "step": 277610 }, { "epoch": 176.94072657743786, "grad_norm": 0.3075573444366455, "learning_rate": 1.5139275959059063e-06, "loss": 0.004, "step": 277620 }, { "epoch": 176.94710006373487, "grad_norm": 0.5126577615737915, "learning_rate": 1.5125818907490375e-06, "loss": 0.0073, "step": 277630 }, { "epoch": 176.95347355003187, "grad_norm": 0.11813390254974365, "learning_rate": 1.511236774760072e-06, "loss": 0.0114, "step": 277640 }, { "epoch": 176.95984703632888, "grad_norm": 0.13056859374046326, "learning_rate": 1.509892247955369e-06, "loss": 0.0064, "step": 277650 }, { "epoch": 176.9662205226259, "grad_norm": 0.17040060460567474, "learning_rate": 1.5085483103512543e-06, "loss": 0.0033, "step": 277660 }, { "epoch": 176.97259400892287, "grad_norm": 0.19057022035121918, "learning_rate": 1.507204961964065e-06, "loss": 0.0052, "step": 277670 }, { "epoch": 176.97896749521988, "grad_norm": 0.12631544470787048, "learning_rate": 1.5058622028101154e-06, "loss": 0.0027, "step": 277680 }, { "epoch": 176.98534098151688, "grad_norm": 0.22371645271778107, "learning_rate": 1.5045200329057319e-06, "loss": 0.0034, "step": 277690 }, { "epoch": 176.9917144678139, "grad_norm": 0.1752142459154129, "learning_rate": 1.503178452267212e-06, "loss": 0.0044, "step": 277700 }, { "epoch": 176.9980879541109, "grad_norm": 0.522023618221283, "learning_rate": 1.5018374609108655e-06, "loss": 0.0052, "step": 277710 }, { "epoch": 177.0044614404079, "grad_norm": 0.33243516087532043, "learning_rate": 1.5004970588529732e-06, "loss": 0.0056, "step": 277720 }, { "epoch": 177.01083492670492, "grad_norm": 0.10905532538890839, "learning_rate": 1.4991572461098501e-06, "loss": 0.0052, "step": 277730 }, { "epoch": 177.01720841300192, "grad_norm": 0.09055198729038239, "learning_rate": 1.4978180226977445e-06, "loss": 0.004, "step": 277740 }, { "epoch": 177.0235818992989, "grad_norm": 0.11727466434240341, "learning_rate": 1.4964793886329542e-06, "loss": 0.0047, "step": 277750 }, { "epoch": 177.0299553855959, "grad_norm": 0.1288483440876007, "learning_rate": 1.4951413439317219e-06, "loss": 0.0081, "step": 277760 }, { "epoch": 177.03632887189292, "grad_norm": 0.10721693933010101, "learning_rate": 1.4938038886103346e-06, "loss": 0.0068, "step": 277770 }, { "epoch": 177.04270235818993, "grad_norm": 0.20702137053012848, "learning_rate": 1.492467022685018e-06, "loss": 0.0032, "step": 277780 }, { "epoch": 177.04907584448694, "grad_norm": 0.05790138989686966, "learning_rate": 1.4911307461720258e-06, "loss": 0.0041, "step": 277790 }, { "epoch": 177.05544933078394, "grad_norm": 0.35241472721099854, "learning_rate": 1.4897950590875954e-06, "loss": 0.0043, "step": 277800 }, { "epoch": 177.06182281708095, "grad_norm": 0.08241704851388931, "learning_rate": 1.4884599614479578e-06, "loss": 0.0071, "step": 277810 }, { "epoch": 177.06819630337796, "grad_norm": 0.17507579922676086, "learning_rate": 1.487125453269328e-06, "loss": 0.0081, "step": 277820 }, { "epoch": 177.07456978967494, "grad_norm": 0.19044117629528046, "learning_rate": 1.4857915345679373e-06, "loss": 0.0027, "step": 277830 }, { "epoch": 177.08094327597195, "grad_norm": 0.07574038207530975, "learning_rate": 1.4844582053599787e-06, "loss": 0.0047, "step": 277840 }, { "epoch": 177.08731676226896, "grad_norm": 0.49446627497673035, "learning_rate": 1.483125465661661e-06, "loss": 0.0073, "step": 277850 }, { "epoch": 177.09369024856596, "grad_norm": 0.1565517783164978, "learning_rate": 1.4817933154891716e-06, "loss": 0.0051, "step": 277860 }, { "epoch": 177.10006373486297, "grad_norm": 0.5084887146949768, "learning_rate": 1.480461754858714e-06, "loss": 0.0076, "step": 277870 }, { "epoch": 177.10643722115998, "grad_norm": 0.08204036951065063, "learning_rate": 1.4791307837864476e-06, "loss": 0.0059, "step": 277880 }, { "epoch": 177.112810707457, "grad_norm": 0.16896730661392212, "learning_rate": 1.4778004022885539e-06, "loss": 0.006, "step": 277890 }, { "epoch": 177.119184193754, "grad_norm": 0.3522171378135681, "learning_rate": 1.476470610381192e-06, "loss": 0.0024, "step": 277900 }, { "epoch": 177.12555768005097, "grad_norm": 0.15850816667079926, "learning_rate": 1.4751414080805437e-06, "loss": 0.0078, "step": 277910 }, { "epoch": 177.13193116634798, "grad_norm": 0.06688817590475082, "learning_rate": 1.4738127954027236e-06, "loss": 0.0106, "step": 277920 }, { "epoch": 177.138304652645, "grad_norm": 0.10653667896986008, "learning_rate": 1.4724847723639024e-06, "loss": 0.0055, "step": 277930 }, { "epoch": 177.144678138942, "grad_norm": 0.16535788774490356, "learning_rate": 1.4711573389802113e-06, "loss": 0.0108, "step": 277940 }, { "epoch": 177.151051625239, "grad_norm": 0.054850440472364426, "learning_rate": 1.4698304952677766e-06, "loss": 0.0056, "step": 277950 }, { "epoch": 177.157425111536, "grad_norm": 0.13785701990127563, "learning_rate": 1.468504241242713e-06, "loss": 0.008, "step": 277960 }, { "epoch": 177.16379859783302, "grad_norm": 0.11200380325317383, "learning_rate": 1.467178576921152e-06, "loss": 0.0033, "step": 277970 }, { "epoch": 177.17017208413003, "grad_norm": 0.08694561570882797, "learning_rate": 1.4658535023191922e-06, "loss": 0.0046, "step": 277980 }, { "epoch": 177.176545570427, "grad_norm": 0.15290725231170654, "learning_rate": 1.464529017452937e-06, "loss": 0.005, "step": 277990 }, { "epoch": 177.18291905672402, "grad_norm": 0.44820600748062134, "learning_rate": 1.4632051223384735e-06, "loss": 0.0067, "step": 278000 }, { "epoch": 177.18929254302103, "grad_norm": 0.28324854373931885, "learning_rate": 1.4618818169919002e-06, "loss": 0.0087, "step": 278010 }, { "epoch": 177.19566602931803, "grad_norm": 0.06135651841759682, "learning_rate": 1.460559101429293e-06, "loss": 0.0049, "step": 278020 }, { "epoch": 177.20203951561504, "grad_norm": 0.2432861179113388, "learning_rate": 1.4592369756667167e-06, "loss": 0.006, "step": 278030 }, { "epoch": 177.20841300191205, "grad_norm": 0.7529646754264832, "learning_rate": 1.4579154397202421e-06, "loss": 0.0048, "step": 278040 }, { "epoch": 177.21478648820906, "grad_norm": 0.11111296713352203, "learning_rate": 1.456594493605934e-06, "loss": 0.0123, "step": 278050 }, { "epoch": 177.22115997450607, "grad_norm": 0.23705941438674927, "learning_rate": 1.4552741373398239e-06, "loss": 0.0072, "step": 278060 }, { "epoch": 177.22753346080307, "grad_norm": 0.16615284979343414, "learning_rate": 1.453954370937971e-06, "loss": 0.0074, "step": 278070 }, { "epoch": 177.23390694710005, "grad_norm": 0.26131492853164673, "learning_rate": 1.4526351944164019e-06, "loss": 0.008, "step": 278080 }, { "epoch": 177.24028043339706, "grad_norm": 0.4765564799308777, "learning_rate": 1.4513166077911643e-06, "loss": 0.0035, "step": 278090 }, { "epoch": 177.24665391969407, "grad_norm": 0.19975383579730988, "learning_rate": 1.449998611078257e-06, "loss": 0.0096, "step": 278100 }, { "epoch": 177.25302740599108, "grad_norm": 0.5551490187644958, "learning_rate": 1.448681204293706e-06, "loss": 0.0246, "step": 278110 }, { "epoch": 177.25940089228808, "grad_norm": 0.12966430187225342, "learning_rate": 1.4473643874535258e-06, "loss": 0.0044, "step": 278120 }, { "epoch": 177.2657743785851, "grad_norm": 0.1731165498495102, "learning_rate": 1.446048160573704e-06, "loss": 0.0049, "step": 278130 }, { "epoch": 177.2721478648821, "grad_norm": 0.2632646858692169, "learning_rate": 1.4447325236702335e-06, "loss": 0.0037, "step": 278140 }, { "epoch": 177.2785213511791, "grad_norm": 0.0731944814324379, "learning_rate": 1.443417476759118e-06, "loss": 0.0064, "step": 278150 }, { "epoch": 177.2848948374761, "grad_norm": 0.21889908611774445, "learning_rate": 1.4421030198563223e-06, "loss": 0.0026, "step": 278160 }, { "epoch": 177.2912683237731, "grad_norm": 0.5779396295547485, "learning_rate": 1.4407891529778172e-06, "loss": 0.0105, "step": 278170 }, { "epoch": 177.2976418100701, "grad_norm": 0.21563124656677246, "learning_rate": 1.4394758761395733e-06, "loss": 0.0069, "step": 278180 }, { "epoch": 177.3040152963671, "grad_norm": 0.12517516314983368, "learning_rate": 1.4381631893575443e-06, "loss": 0.0019, "step": 278190 }, { "epoch": 177.31038878266412, "grad_norm": 0.11182703822851181, "learning_rate": 1.4368510926476897e-06, "loss": 0.0055, "step": 278200 }, { "epoch": 177.31676226896113, "grad_norm": 0.10348129272460938, "learning_rate": 1.4355395860259413e-06, "loss": 0.009, "step": 278210 }, { "epoch": 177.32313575525814, "grad_norm": 0.23372045159339905, "learning_rate": 1.4342286695082307e-06, "loss": 0.0078, "step": 278220 }, { "epoch": 177.32950924155514, "grad_norm": 0.1566370576620102, "learning_rate": 1.4329183431105065e-06, "loss": 0.0025, "step": 278230 }, { "epoch": 177.33588272785212, "grad_norm": 0.4672344923019409, "learning_rate": 1.4316086068486777e-06, "loss": 0.0067, "step": 278240 }, { "epoch": 177.34225621414913, "grad_norm": 0.456296443939209, "learning_rate": 1.43029946073866e-06, "loss": 0.0103, "step": 278250 }, { "epoch": 177.34862970044614, "grad_norm": 0.21482546627521515, "learning_rate": 1.4289909047963624e-06, "loss": 0.0047, "step": 278260 }, { "epoch": 177.35500318674315, "grad_norm": 0.08649458736181259, "learning_rate": 1.4276829390376779e-06, "loss": 0.0037, "step": 278270 }, { "epoch": 177.36137667304016, "grad_norm": 0.1796308308839798, "learning_rate": 1.4263755634785103e-06, "loss": 0.0051, "step": 278280 }, { "epoch": 177.36775015933716, "grad_norm": 0.10218016058206558, "learning_rate": 1.4250687781347415e-06, "loss": 0.0055, "step": 278290 }, { "epoch": 177.37412364563417, "grad_norm": 0.2459896355867386, "learning_rate": 1.4237625830222479e-06, "loss": 0.0041, "step": 278300 }, { "epoch": 177.38049713193118, "grad_norm": 0.2565355598926544, "learning_rate": 1.4224569781568997e-06, "loss": 0.0039, "step": 278310 }, { "epoch": 177.38687061822816, "grad_norm": 0.1411680281162262, "learning_rate": 1.4211519635545679e-06, "loss": 0.0051, "step": 278320 }, { "epoch": 177.39324410452517, "grad_norm": 0.17750480771064758, "learning_rate": 1.419847539231106e-06, "loss": 0.0035, "step": 278330 }, { "epoch": 177.39961759082217, "grad_norm": 0.31370770931243896, "learning_rate": 1.418543705202363e-06, "loss": 0.0088, "step": 278340 }, { "epoch": 177.40599107711918, "grad_norm": 0.3813159763813019, "learning_rate": 1.4172404614841762e-06, "loss": 0.0081, "step": 278350 }, { "epoch": 177.4123645634162, "grad_norm": 0.49862024188041687, "learning_rate": 1.4159378080923935e-06, "loss": 0.0102, "step": 278360 }, { "epoch": 177.4187380497132, "grad_norm": 0.2564702332019806, "learning_rate": 1.414635745042836e-06, "loss": 0.0089, "step": 278370 }, { "epoch": 177.4251115360102, "grad_norm": 0.08094695210456848, "learning_rate": 1.4133342723513299e-06, "loss": 0.0057, "step": 278380 }, { "epoch": 177.4314850223072, "grad_norm": 0.09824644774198532, "learning_rate": 1.412033390033679e-06, "loss": 0.0024, "step": 278390 }, { "epoch": 177.4378585086042, "grad_norm": 0.08092913031578064, "learning_rate": 1.4107330981057043e-06, "loss": 0.0083, "step": 278400 }, { "epoch": 177.4442319949012, "grad_norm": 0.2146797478199005, "learning_rate": 1.4094333965831985e-06, "loss": 0.0032, "step": 278410 }, { "epoch": 177.4506054811982, "grad_norm": 0.38433578610420227, "learning_rate": 1.4081342854819491e-06, "loss": 0.0063, "step": 278420 }, { "epoch": 177.45697896749522, "grad_norm": 0.16805467009544373, "learning_rate": 1.4068357648177433e-06, "loss": 0.0032, "step": 278430 }, { "epoch": 177.46335245379223, "grad_norm": 0.22070053219795227, "learning_rate": 1.4055378346063741e-06, "loss": 0.0078, "step": 278440 }, { "epoch": 177.46972594008923, "grad_norm": 0.18020984530448914, "learning_rate": 1.40424049486359e-06, "loss": 0.0043, "step": 278450 }, { "epoch": 177.47609942638624, "grad_norm": 0.2566033601760864, "learning_rate": 1.402943745605173e-06, "loss": 0.0035, "step": 278460 }, { "epoch": 177.48247291268325, "grad_norm": 0.4014528691768646, "learning_rate": 1.401647586846866e-06, "loss": 0.0101, "step": 278470 }, { "epoch": 177.48884639898023, "grad_norm": 0.07283765822649002, "learning_rate": 1.4003520186044338e-06, "loss": 0.0094, "step": 278480 }, { "epoch": 177.49521988527724, "grad_norm": 0.3976368010044098, "learning_rate": 1.3990570408936031e-06, "loss": 0.0079, "step": 278490 }, { "epoch": 177.50159337157424, "grad_norm": 0.09552770853042603, "learning_rate": 1.3977626537301224e-06, "loss": 0.0107, "step": 278500 }, { "epoch": 177.50796685787125, "grad_norm": 0.13229210674762726, "learning_rate": 1.3964688571297069e-06, "loss": 0.0039, "step": 278510 }, { "epoch": 177.51434034416826, "grad_norm": 0.12656205892562866, "learning_rate": 1.3951756511080883e-06, "loss": 0.0109, "step": 278520 }, { "epoch": 177.52071383046527, "grad_norm": 0.44508370757102966, "learning_rate": 1.3938830356809763e-06, "loss": 0.0062, "step": 278530 }, { "epoch": 177.52708731676228, "grad_norm": 0.13151854276657104, "learning_rate": 1.392591010864075e-06, "loss": 0.0034, "step": 278540 }, { "epoch": 177.53346080305928, "grad_norm": 0.17752259969711304, "learning_rate": 1.3912995766730885e-06, "loss": 0.0046, "step": 278550 }, { "epoch": 177.53983428935626, "grad_norm": 0.11268600821495056, "learning_rate": 1.3900087331237043e-06, "loss": 0.0051, "step": 278560 }, { "epoch": 177.54620777565327, "grad_norm": 0.19533635675907135, "learning_rate": 1.3887184802316045e-06, "loss": 0.0126, "step": 278570 }, { "epoch": 177.55258126195028, "grad_norm": 0.17456167936325073, "learning_rate": 1.3874288180124871e-06, "loss": 0.0037, "step": 278580 }, { "epoch": 177.5589547482473, "grad_norm": 0.10120189189910889, "learning_rate": 1.3861397464819903e-06, "loss": 0.0031, "step": 278590 }, { "epoch": 177.5653282345443, "grad_norm": 0.16055314242839813, "learning_rate": 1.3848512656558066e-06, "loss": 0.0024, "step": 278600 }, { "epoch": 177.5717017208413, "grad_norm": 0.16237913072109222, "learning_rate": 1.383563375549568e-06, "loss": 0.0054, "step": 278610 }, { "epoch": 177.5780752071383, "grad_norm": 0.39408883452415466, "learning_rate": 1.382276076178951e-06, "loss": 0.0057, "step": 278620 }, { "epoch": 177.58444869343532, "grad_norm": 0.06979204714298248, "learning_rate": 1.3809893675595708e-06, "loss": 0.0033, "step": 278630 }, { "epoch": 177.5908221797323, "grad_norm": 0.4642535150051117, "learning_rate": 1.3797032497070816e-06, "loss": 0.0102, "step": 278640 }, { "epoch": 177.5971956660293, "grad_norm": 0.3934544622898102, "learning_rate": 1.3784177226370932e-06, "loss": 0.0041, "step": 278650 }, { "epoch": 177.60356915232632, "grad_norm": 0.17313872277736664, "learning_rate": 1.3771327863652483e-06, "loss": 0.0065, "step": 278660 }, { "epoch": 177.60994263862332, "grad_norm": 0.09527195990085602, "learning_rate": 1.3758484409071348e-06, "loss": 0.0037, "step": 278670 }, { "epoch": 177.61631612492033, "grad_norm": 0.08915118128061295, "learning_rate": 1.374564686278379e-06, "loss": 0.0054, "step": 278680 }, { "epoch": 177.62268961121734, "grad_norm": 0.0871809720993042, "learning_rate": 1.3732815224945738e-06, "loss": 0.0033, "step": 278690 }, { "epoch": 177.62906309751435, "grad_norm": 0.17193709313869476, "learning_rate": 1.371998949571307e-06, "loss": 0.0097, "step": 278700 }, { "epoch": 177.63543658381136, "grad_norm": 0.36506861448287964, "learning_rate": 1.3707169675241661e-06, "loss": 0.0087, "step": 278710 }, { "epoch": 177.64181007010836, "grad_norm": 0.3809928894042969, "learning_rate": 1.3694355763687273e-06, "loss": 0.0022, "step": 278720 }, { "epoch": 177.64818355640534, "grad_norm": 0.10851477086544037, "learning_rate": 1.3681547761205615e-06, "loss": 0.004, "step": 278730 }, { "epoch": 177.65455704270235, "grad_norm": 0.27161905169487, "learning_rate": 1.3668745667952287e-06, "loss": 0.0043, "step": 278740 }, { "epoch": 177.66093052899936, "grad_norm": 0.11413377523422241, "learning_rate": 1.3655949484082885e-06, "loss": 0.0036, "step": 278750 }, { "epoch": 177.66730401529637, "grad_norm": 0.7036051154136658, "learning_rate": 1.3643159209752953e-06, "loss": 0.0117, "step": 278760 }, { "epoch": 177.67367750159337, "grad_norm": 0.16204515099525452, "learning_rate": 1.36303748451177e-06, "loss": 0.0038, "step": 278770 }, { "epoch": 177.68005098789038, "grad_norm": 0.10441179573535919, "learning_rate": 1.3617596390332666e-06, "loss": 0.0041, "step": 278780 }, { "epoch": 177.6864244741874, "grad_norm": 0.12868301570415497, "learning_rate": 1.360482384555306e-06, "loss": 0.0029, "step": 278790 }, { "epoch": 177.6927979604844, "grad_norm": 0.22290800511837006, "learning_rate": 1.3592057210934095e-06, "loss": 0.0033, "step": 278800 }, { "epoch": 177.69917144678138, "grad_norm": 0.46677109599113464, "learning_rate": 1.3579296486630865e-06, "loss": 0.0164, "step": 278810 }, { "epoch": 177.7055449330784, "grad_norm": 0.09975612163543701, "learning_rate": 1.3566541672798416e-06, "loss": 0.004, "step": 278820 }, { "epoch": 177.7119184193754, "grad_norm": 0.14627020061016083, "learning_rate": 1.3553792769591788e-06, "loss": 0.013, "step": 278830 }, { "epoch": 177.7182919056724, "grad_norm": 0.23135152459144592, "learning_rate": 1.3541049777165859e-06, "loss": 0.0042, "step": 278840 }, { "epoch": 177.7246653919694, "grad_norm": 0.12270554155111313, "learning_rate": 1.352831269567545e-06, "loss": 0.0022, "step": 278850 }, { "epoch": 177.73103887826642, "grad_norm": 0.22169189155101776, "learning_rate": 1.3515581525275323e-06, "loss": 0.0048, "step": 278860 }, { "epoch": 177.73741236456343, "grad_norm": 0.13511203229427338, "learning_rate": 1.3502856266120245e-06, "loss": 0.0032, "step": 278870 }, { "epoch": 177.74378585086043, "grad_norm": 0.44262295961380005, "learning_rate": 1.3490136918364816e-06, "loss": 0.0068, "step": 278880 }, { "epoch": 177.7501593371574, "grad_norm": 0.0633622407913208, "learning_rate": 1.3477423482163465e-06, "loss": 0.0045, "step": 278890 }, { "epoch": 177.75653282345442, "grad_norm": 0.08708839118480682, "learning_rate": 1.3464715957670848e-06, "loss": 0.0071, "step": 278900 }, { "epoch": 177.76290630975143, "grad_norm": 0.21958325803279877, "learning_rate": 1.3452014345041286e-06, "loss": 0.0152, "step": 278910 }, { "epoch": 177.76927979604844, "grad_norm": 0.3067914843559265, "learning_rate": 1.3439318644429156e-06, "loss": 0.0043, "step": 278920 }, { "epoch": 177.77565328234544, "grad_norm": 0.1548873484134674, "learning_rate": 1.342662885598861e-06, "loss": 0.0039, "step": 278930 }, { "epoch": 177.78202676864245, "grad_norm": 0.2416621446609497, "learning_rate": 1.3413944979873971e-06, "loss": 0.0161, "step": 278940 }, { "epoch": 177.78840025493946, "grad_norm": 0.19755223393440247, "learning_rate": 1.3401267016239339e-06, "loss": 0.0031, "step": 278950 }, { "epoch": 177.79477374123647, "grad_norm": 0.19986118376255035, "learning_rate": 1.3388594965238755e-06, "loss": 0.0029, "step": 278960 }, { "epoch": 177.80114722753345, "grad_norm": 0.21054910123348236, "learning_rate": 1.3375928827026097e-06, "loss": 0.008, "step": 278970 }, { "epoch": 177.80752071383046, "grad_norm": 0.15705551207065582, "learning_rate": 1.3363268601755408e-06, "loss": 0.0031, "step": 278980 }, { "epoch": 177.81389420012746, "grad_norm": 0.13030758500099182, "learning_rate": 1.3350614289580455e-06, "loss": 0.006, "step": 278990 }, { "epoch": 177.82026768642447, "grad_norm": 0.14373327791690826, "learning_rate": 1.3337965890655003e-06, "loss": 0.0054, "step": 279000 }, { "epoch": 177.82664117272148, "grad_norm": 0.13990718126296997, "learning_rate": 1.332532340513276e-06, "loss": 0.0038, "step": 279010 }, { "epoch": 177.8330146590185, "grad_norm": 0.20908012986183167, "learning_rate": 1.3312686833167332e-06, "loss": 0.0031, "step": 279020 }, { "epoch": 177.8393881453155, "grad_norm": 0.14301173388957977, "learning_rate": 1.3300056174912257e-06, "loss": 0.0041, "step": 279030 }, { "epoch": 177.8457616316125, "grad_norm": 0.06916928291320801, "learning_rate": 1.328743143052108e-06, "loss": 0.0028, "step": 279040 }, { "epoch": 177.85213511790948, "grad_norm": 0.2474038302898407, "learning_rate": 1.3274812600147068e-06, "loss": 0.0037, "step": 279050 }, { "epoch": 177.8585086042065, "grad_norm": 0.22769805788993835, "learning_rate": 1.3262199683943655e-06, "loss": 0.0046, "step": 279060 }, { "epoch": 177.8648820905035, "grad_norm": 0.19482101500034332, "learning_rate": 1.3249592682064106e-06, "loss": 0.0043, "step": 279070 }, { "epoch": 177.8712555768005, "grad_norm": 0.33337175846099854, "learning_rate": 1.3236991594661518e-06, "loss": 0.0082, "step": 279080 }, { "epoch": 177.87762906309752, "grad_norm": 0.09862909466028214, "learning_rate": 1.3224396421889106e-06, "loss": 0.0103, "step": 279090 }, { "epoch": 177.88400254939452, "grad_norm": 0.0754169374704361, "learning_rate": 1.3211807163899804e-06, "loss": 0.0075, "step": 279100 }, { "epoch": 177.89037603569153, "grad_norm": 0.07221657037734985, "learning_rate": 1.3199223820846706e-06, "loss": 0.0083, "step": 279110 }, { "epoch": 177.89674952198854, "grad_norm": 0.11409954726696014, "learning_rate": 1.3186646392882696e-06, "loss": 0.0023, "step": 279120 }, { "epoch": 177.90312300828552, "grad_norm": 0.1614426076412201, "learning_rate": 1.317407488016048e-06, "loss": 0.0056, "step": 279130 }, { "epoch": 177.90949649458253, "grad_norm": 0.2352873682975769, "learning_rate": 1.3161509282832885e-06, "loss": 0.0034, "step": 279140 }, { "epoch": 177.91586998087953, "grad_norm": 0.10599830746650696, "learning_rate": 1.3148949601052729e-06, "loss": 0.0052, "step": 279150 }, { "epoch": 177.92224346717654, "grad_norm": 0.08301091194152832, "learning_rate": 1.3136395834972392e-06, "loss": 0.0064, "step": 279160 }, { "epoch": 177.92861695347355, "grad_norm": 0.04234009608626366, "learning_rate": 1.312384798474453e-06, "loss": 0.0037, "step": 279170 }, { "epoch": 177.93499043977056, "grad_norm": 0.11570965498685837, "learning_rate": 1.3111306050521577e-06, "loss": 0.0069, "step": 279180 }, { "epoch": 177.94136392606757, "grad_norm": 0.4991712272167206, "learning_rate": 1.3098770032456076e-06, "loss": 0.0109, "step": 279190 }, { "epoch": 177.94773741236457, "grad_norm": 0.1517076939344406, "learning_rate": 1.3086239930700072e-06, "loss": 0.0038, "step": 279200 }, { "epoch": 177.95411089866155, "grad_norm": 0.23408539593219757, "learning_rate": 1.3073715745406056e-06, "loss": 0.0104, "step": 279210 }, { "epoch": 177.96048438495856, "grad_norm": 0.2167043536901474, "learning_rate": 1.3061197476726128e-06, "loss": 0.0041, "step": 279220 }, { "epoch": 177.96685787125557, "grad_norm": 0.1361575871706009, "learning_rate": 1.3048685124812442e-06, "loss": 0.002, "step": 279230 }, { "epoch": 177.97323135755258, "grad_norm": 0.11508876085281372, "learning_rate": 1.303617868981688e-06, "loss": 0.0074, "step": 279240 }, { "epoch": 177.9796048438496, "grad_norm": 0.15716420114040375, "learning_rate": 1.3023678171891597e-06, "loss": 0.0033, "step": 279250 }, { "epoch": 177.9859783301466, "grad_norm": 0.46710067987442017, "learning_rate": 1.3011183571188412e-06, "loss": 0.005, "step": 279260 }, { "epoch": 177.9923518164436, "grad_norm": 0.370906800031662, "learning_rate": 1.2998694887859097e-06, "loss": 0.0068, "step": 279270 }, { "epoch": 177.9987253027406, "grad_norm": 0.22323983907699585, "learning_rate": 1.2986212122055419e-06, "loss": 0.0021, "step": 279280 }, { "epoch": 178.00509878903762, "grad_norm": 0.13579869270324707, "learning_rate": 1.2973735273929143e-06, "loss": 0.004, "step": 279290 }, { "epoch": 178.0114722753346, "grad_norm": 0.10081005096435547, "learning_rate": 1.2961264343631818e-06, "loss": 0.0038, "step": 279300 }, { "epoch": 178.0178457616316, "grad_norm": 0.290127158164978, "learning_rate": 1.2948799331314932e-06, "loss": 0.003, "step": 279310 }, { "epoch": 178.0242192479286, "grad_norm": 0.20913036167621613, "learning_rate": 1.2936340237129975e-06, "loss": 0.0064, "step": 279320 }, { "epoch": 178.03059273422562, "grad_norm": 0.08173798769712448, "learning_rate": 1.2923887061228435e-06, "loss": 0.0089, "step": 279330 }, { "epoch": 178.03696622052263, "grad_norm": 0.4078877866268158, "learning_rate": 1.291143980376147e-06, "loss": 0.0077, "step": 279340 }, { "epoch": 178.04333970681964, "grad_norm": 0.08312412351369858, "learning_rate": 1.2898998464880407e-06, "loss": 0.0107, "step": 279350 }, { "epoch": 178.04971319311665, "grad_norm": 0.06007718667387962, "learning_rate": 1.28865630447364e-06, "loss": 0.0084, "step": 279360 }, { "epoch": 178.05608667941365, "grad_norm": 0.13051612675189972, "learning_rate": 1.2874133543480603e-06, "loss": 0.0036, "step": 279370 }, { "epoch": 178.06246016571063, "grad_norm": 0.1721290647983551, "learning_rate": 1.2861709961263956e-06, "loss": 0.0034, "step": 279380 }, { "epoch": 178.06883365200764, "grad_norm": 0.1080324575304985, "learning_rate": 1.2849292298237504e-06, "loss": 0.0026, "step": 279390 }, { "epoch": 178.07520713830465, "grad_norm": 0.0902484580874443, "learning_rate": 1.2836880554552123e-06, "loss": 0.0069, "step": 279400 }, { "epoch": 178.08158062460166, "grad_norm": 0.3405386805534363, "learning_rate": 1.2824474730358582e-06, "loss": 0.0069, "step": 279410 }, { "epoch": 178.08795411089866, "grad_norm": 0.09093481302261353, "learning_rate": 1.2812074825807596e-06, "loss": 0.0048, "step": 279420 }, { "epoch": 178.09432759719567, "grad_norm": 0.18735915422439575, "learning_rate": 1.2799680841049932e-06, "loss": 0.0046, "step": 279430 }, { "epoch": 178.10070108349268, "grad_norm": 0.07954783737659454, "learning_rate": 1.2787292776236138e-06, "loss": 0.0061, "step": 279440 }, { "epoch": 178.1070745697897, "grad_norm": 0.12051209807395935, "learning_rate": 1.27749106315167e-06, "loss": 0.0044, "step": 279450 }, { "epoch": 178.11344805608667, "grad_norm": 0.17441070079803467, "learning_rate": 1.2762534407042116e-06, "loss": 0.0087, "step": 279460 }, { "epoch": 178.11982154238368, "grad_norm": 0.13481305539608002, "learning_rate": 1.275016410296287e-06, "loss": 0.0038, "step": 279470 }, { "epoch": 178.12619502868068, "grad_norm": 0.17414361238479614, "learning_rate": 1.273779971942901e-06, "loss": 0.0072, "step": 279480 }, { "epoch": 178.1325685149777, "grad_norm": 0.19927211105823517, "learning_rate": 1.272544125659103e-06, "loss": 0.0027, "step": 279490 }, { "epoch": 178.1389420012747, "grad_norm": 0.11269427835941315, "learning_rate": 1.2713088714598975e-06, "loss": 0.0094, "step": 279500 }, { "epoch": 178.1453154875717, "grad_norm": 0.3192826807498932, "learning_rate": 1.2700742093603056e-06, "loss": 0.0071, "step": 279510 }, { "epoch": 178.15168897386872, "grad_norm": 0.09435617178678513, "learning_rate": 1.2688401393753102e-06, "loss": 0.0069, "step": 279520 }, { "epoch": 178.15806246016572, "grad_norm": 0.4342280924320221, "learning_rate": 1.2676066615199212e-06, "loss": 0.0059, "step": 279530 }, { "epoch": 178.1644359464627, "grad_norm": 0.10781070590019226, "learning_rate": 1.2663737758091154e-06, "loss": 0.0049, "step": 279540 }, { "epoch": 178.1708094327597, "grad_norm": 0.0694686770439148, "learning_rate": 1.2651414822578922e-06, "loss": 0.0074, "step": 279550 }, { "epoch": 178.17718291905672, "grad_norm": 0.380055695772171, "learning_rate": 1.2639097808812006e-06, "loss": 0.0039, "step": 279560 }, { "epoch": 178.18355640535373, "grad_norm": 0.17109206318855286, "learning_rate": 1.2626786716940288e-06, "loss": 0.0048, "step": 279570 }, { "epoch": 178.18992989165073, "grad_norm": 0.09975151717662811, "learning_rate": 1.2614481547113254e-06, "loss": 0.0045, "step": 279580 }, { "epoch": 178.19630337794774, "grad_norm": 0.136665940284729, "learning_rate": 1.2602182299480458e-06, "loss": 0.004, "step": 279590 }, { "epoch": 178.20267686424475, "grad_norm": 0.09873627871274948, "learning_rate": 1.2589888974191223e-06, "loss": 0.0116, "step": 279600 }, { "epoch": 178.20905035054176, "grad_norm": 0.1064174547791481, "learning_rate": 1.2577601571395148e-06, "loss": 0.0025, "step": 279610 }, { "epoch": 178.21542383683874, "grad_norm": 0.16958341002464294, "learning_rate": 1.2565320091241339e-06, "loss": 0.0027, "step": 279620 }, { "epoch": 178.22179732313575, "grad_norm": 0.3508605360984802, "learning_rate": 1.2553044533879176e-06, "loss": 0.0057, "step": 279630 }, { "epoch": 178.22817080943275, "grad_norm": 0.24132172763347626, "learning_rate": 1.2540774899457652e-06, "loss": 0.0065, "step": 279640 }, { "epoch": 178.23454429572976, "grad_norm": 0.1156081110239029, "learning_rate": 1.2528511188126036e-06, "loss": 0.0088, "step": 279650 }, { "epoch": 178.24091778202677, "grad_norm": 0.1166858896613121, "learning_rate": 1.2516253400033261e-06, "loss": 0.0034, "step": 279660 }, { "epoch": 178.24729126832378, "grad_norm": 0.08247307687997818, "learning_rate": 1.2504001535328214e-06, "loss": 0.0046, "step": 279670 }, { "epoch": 178.2536647546208, "grad_norm": 0.1856633573770523, "learning_rate": 1.2491755594159825e-06, "loss": 0.0026, "step": 279680 }, { "epoch": 178.2600382409178, "grad_norm": 0.4050697684288025, "learning_rate": 1.2479515576676927e-06, "loss": 0.0061, "step": 279690 }, { "epoch": 178.26641172721477, "grad_norm": 0.3014164865016937, "learning_rate": 1.2467281483028171e-06, "loss": 0.006, "step": 279700 }, { "epoch": 178.27278521351178, "grad_norm": 0.2529468536376953, "learning_rate": 1.2455053313362275e-06, "loss": 0.0051, "step": 279710 }, { "epoch": 178.2791586998088, "grad_norm": 0.2431836575269699, "learning_rate": 1.2442831067827842e-06, "loss": 0.0076, "step": 279720 }, { "epoch": 178.2855321861058, "grad_norm": 0.1492333710193634, "learning_rate": 1.2430614746573255e-06, "loss": 0.0035, "step": 279730 }, { "epoch": 178.2919056724028, "grad_norm": 0.30467426776885986, "learning_rate": 1.2418404349747058e-06, "loss": 0.0048, "step": 279740 }, { "epoch": 178.2982791586998, "grad_norm": 0.11203408986330032, "learning_rate": 1.2406199877497692e-06, "loss": 0.0043, "step": 279750 }, { "epoch": 178.30465264499682, "grad_norm": 0.06989684700965881, "learning_rate": 1.239400132997326e-06, "loss": 0.0041, "step": 279760 }, { "epoch": 178.31102613129383, "grad_norm": 0.15745815634727478, "learning_rate": 1.2381808707322084e-06, "loss": 0.0062, "step": 279770 }, { "epoch": 178.3173996175908, "grad_norm": 0.22073428332805634, "learning_rate": 1.2369622009692383e-06, "loss": 0.0182, "step": 279780 }, { "epoch": 178.32377310388782, "grad_norm": 0.12743312120437622, "learning_rate": 1.2357441237232204e-06, "loss": 0.005, "step": 279790 }, { "epoch": 178.33014659018482, "grad_norm": 0.11074240505695343, "learning_rate": 1.2345266390089482e-06, "loss": 0.0055, "step": 279800 }, { "epoch": 178.33652007648183, "grad_norm": 0.15236309170722961, "learning_rate": 1.2333097468412103e-06, "loss": 0.0114, "step": 279810 }, { "epoch": 178.34289356277884, "grad_norm": 0.17006699740886688, "learning_rate": 1.2320934472348112e-06, "loss": 0.0046, "step": 279820 }, { "epoch": 178.34926704907585, "grad_norm": 0.09563572704792023, "learning_rate": 1.2308777402045223e-06, "loss": 0.0069, "step": 279830 }, { "epoch": 178.35564053537286, "grad_norm": 0.10831533372402191, "learning_rate": 1.2296626257651157e-06, "loss": 0.0061, "step": 279840 }, { "epoch": 178.36201402166986, "grad_norm": 0.11425092816352844, "learning_rate": 1.2284481039313511e-06, "loss": 0.0071, "step": 279850 }, { "epoch": 178.36838750796684, "grad_norm": 0.07882004231214523, "learning_rate": 1.2272341747179893e-06, "loss": 0.0026, "step": 279860 }, { "epoch": 178.37476099426385, "grad_norm": 0.1373937875032425, "learning_rate": 1.2260208381397852e-06, "loss": 0.003, "step": 279870 }, { "epoch": 178.38113448056086, "grad_norm": 0.25624504685401917, "learning_rate": 1.224808094211477e-06, "loss": 0.0048, "step": 279880 }, { "epoch": 178.38750796685787, "grad_norm": 0.15343131124973297, "learning_rate": 1.2235959429477972e-06, "loss": 0.0046, "step": 279890 }, { "epoch": 178.39388145315488, "grad_norm": 0.15179425477981567, "learning_rate": 1.2223843843634896e-06, "loss": 0.0053, "step": 279900 }, { "epoch": 178.40025493945188, "grad_norm": 0.3232979476451874, "learning_rate": 1.2211734184732538e-06, "loss": 0.0035, "step": 279910 }, { "epoch": 178.4066284257489, "grad_norm": 0.2665908634662628, "learning_rate": 1.219963045291822e-06, "loss": 0.0065, "step": 279920 }, { "epoch": 178.4130019120459, "grad_norm": 0.2190011739730835, "learning_rate": 1.218753264833894e-06, "loss": 0.0045, "step": 279930 }, { "epoch": 178.4193753983429, "grad_norm": 0.16835132241249084, "learning_rate": 1.2175440771141744e-06, "loss": 0.0057, "step": 279940 }, { "epoch": 178.4257488846399, "grad_norm": 0.34108981490135193, "learning_rate": 1.2163354821473461e-06, "loss": 0.0043, "step": 279950 }, { "epoch": 178.4321223709369, "grad_norm": 0.10115467011928558, "learning_rate": 1.2151274799481082e-06, "loss": 0.0035, "step": 279960 }, { "epoch": 178.4384958572339, "grad_norm": 0.2859518229961395, "learning_rate": 1.213920070531127e-06, "loss": 0.0096, "step": 279970 }, { "epoch": 178.4448693435309, "grad_norm": 0.14651982486248016, "learning_rate": 1.2127132539110796e-06, "loss": 0.0062, "step": 279980 }, { "epoch": 178.45124282982792, "grad_norm": 0.21585333347320557, "learning_rate": 1.2115070301026266e-06, "loss": 0.0035, "step": 279990 }, { "epoch": 178.45761631612493, "grad_norm": 0.12804889678955078, "learning_rate": 1.2103013991204337e-06, "loss": 0.0118, "step": 280000 }, { "epoch": 178.46398980242193, "grad_norm": 0.08533932268619537, "learning_rate": 1.2090963609791395e-06, "loss": 0.0057, "step": 280010 }, { "epoch": 178.47036328871894, "grad_norm": 0.3135342001914978, "learning_rate": 1.2078919156933931e-06, "loss": 0.0034, "step": 280020 }, { "epoch": 178.47673677501592, "grad_norm": 0.16699498891830444, "learning_rate": 1.2066880632778222e-06, "loss": 0.0034, "step": 280030 }, { "epoch": 178.48311026131293, "grad_norm": 0.07513880729675293, "learning_rate": 1.2054848037470646e-06, "loss": 0.0054, "step": 280040 }, { "epoch": 178.48948374760994, "grad_norm": 0.06381925195455551, "learning_rate": 1.204282137115731e-06, "loss": 0.0043, "step": 280050 }, { "epoch": 178.49585723390695, "grad_norm": 0.13217493891716003, "learning_rate": 1.2030800633984485e-06, "loss": 0.0086, "step": 280060 }, { "epoch": 178.50223072020395, "grad_norm": 0.0962805449962616, "learning_rate": 1.2018785826098056e-06, "loss": 0.0048, "step": 280070 }, { "epoch": 178.50860420650096, "grad_norm": 0.1593610942363739, "learning_rate": 1.200677694764424e-06, "loss": 0.0032, "step": 280080 }, { "epoch": 178.51497769279797, "grad_norm": 0.3925822973251343, "learning_rate": 1.1994773998768695e-06, "loss": 0.0049, "step": 280090 }, { "epoch": 178.52135117909498, "grad_norm": 0.1294698417186737, "learning_rate": 1.1982776979617472e-06, "loss": 0.0074, "step": 280100 }, { "epoch": 178.52772466539196, "grad_norm": 0.11469677835702896, "learning_rate": 1.1970785890336178e-06, "loss": 0.0047, "step": 280110 }, { "epoch": 178.53409815168897, "grad_norm": 0.23556284606456757, "learning_rate": 1.1958800731070752e-06, "loss": 0.0045, "step": 280120 }, { "epoch": 178.54047163798597, "grad_norm": 0.12199951708316803, "learning_rate": 1.1946821501966576e-06, "loss": 0.0027, "step": 280130 }, { "epoch": 178.54684512428298, "grad_norm": 0.08611631393432617, "learning_rate": 1.193484820316937e-06, "loss": 0.0085, "step": 280140 }, { "epoch": 178.55321861058, "grad_norm": 0.12469654530286789, "learning_rate": 1.1922880834824567e-06, "loss": 0.0044, "step": 280150 }, { "epoch": 178.559592096877, "grad_norm": 0.09224969893693924, "learning_rate": 1.1910919397077612e-06, "loss": 0.0037, "step": 280160 }, { "epoch": 178.565965583174, "grad_norm": 0.2762269377708435, "learning_rate": 1.1898963890073721e-06, "loss": 0.0094, "step": 280170 }, { "epoch": 178.572339069471, "grad_norm": 0.10831842571496964, "learning_rate": 1.1887014313958333e-06, "loss": 0.0046, "step": 280180 }, { "epoch": 178.578712555768, "grad_norm": 0.33260273933410645, "learning_rate": 1.1875070668876552e-06, "loss": 0.0188, "step": 280190 }, { "epoch": 178.585086042065, "grad_norm": 0.057766616344451904, "learning_rate": 1.18631329549736e-06, "loss": 0.0023, "step": 280200 }, { "epoch": 178.591459528362, "grad_norm": 0.0986441895365715, "learning_rate": 1.1851201172394356e-06, "loss": 0.0038, "step": 280210 }, { "epoch": 178.59783301465902, "grad_norm": 0.4565417468547821, "learning_rate": 1.1839275321283983e-06, "loss": 0.0073, "step": 280220 }, { "epoch": 178.60420650095602, "grad_norm": 0.4047238528728485, "learning_rate": 1.1827355401787255e-06, "loss": 0.0185, "step": 280230 }, { "epoch": 178.61057998725303, "grad_norm": 0.4012536108493805, "learning_rate": 1.1815441414049111e-06, "loss": 0.015, "step": 280240 }, { "epoch": 178.61695347355004, "grad_norm": 0.20729325711727142, "learning_rate": 1.1803533358214214e-06, "loss": 0.0076, "step": 280250 }, { "epoch": 178.62332695984705, "grad_norm": 0.20590384304523468, "learning_rate": 1.1791631234427448e-06, "loss": 0.0027, "step": 280260 }, { "epoch": 178.62970044614403, "grad_norm": 0.4722278118133545, "learning_rate": 1.1779735042833196e-06, "loss": 0.0059, "step": 280270 }, { "epoch": 178.63607393244104, "grad_norm": 0.23734740912914276, "learning_rate": 1.1767844783576177e-06, "loss": 0.0063, "step": 280280 }, { "epoch": 178.64244741873804, "grad_norm": 0.08057398349046707, "learning_rate": 1.1755960456800774e-06, "loss": 0.003, "step": 280290 }, { "epoch": 178.64882090503505, "grad_norm": 0.16797040402889252, "learning_rate": 1.1744082062651485e-06, "loss": 0.0058, "step": 280300 }, { "epoch": 178.65519439133206, "grad_norm": 0.05439222231507301, "learning_rate": 1.173220960127247e-06, "loss": 0.0077, "step": 280310 }, { "epoch": 178.66156787762907, "grad_norm": 0.04730872064828873, "learning_rate": 1.1720343072808226e-06, "loss": 0.0024, "step": 280320 }, { "epoch": 178.66794136392608, "grad_norm": 0.09666468948125839, "learning_rate": 1.1708482477402804e-06, "loss": 0.0156, "step": 280330 }, { "epoch": 178.67431485022308, "grad_norm": 0.2647816836833954, "learning_rate": 1.1696627815200312e-06, "loss": 0.0045, "step": 280340 }, { "epoch": 178.68068833652006, "grad_norm": 0.06293470412492752, "learning_rate": 1.16847790863448e-06, "loss": 0.0015, "step": 280350 }, { "epoch": 178.68706182281707, "grad_norm": 0.47377610206604004, "learning_rate": 1.1672936290980374e-06, "loss": 0.0094, "step": 280360 }, { "epoch": 178.69343530911408, "grad_norm": 0.10751724988222122, "learning_rate": 1.1661099429250754e-06, "loss": 0.0026, "step": 280370 }, { "epoch": 178.6998087954111, "grad_norm": 0.08203781396150589, "learning_rate": 1.1649268501299826e-06, "loss": 0.0041, "step": 280380 }, { "epoch": 178.7061822817081, "grad_norm": 0.1419510692358017, "learning_rate": 1.1637443507271417e-06, "loss": 0.0043, "step": 280390 }, { "epoch": 178.7125557680051, "grad_norm": 0.09864455461502075, "learning_rate": 1.1625624447309137e-06, "loss": 0.0143, "step": 280400 }, { "epoch": 178.7189292543021, "grad_norm": 0.271968275308609, "learning_rate": 1.1613811321556645e-06, "loss": 0.0038, "step": 280410 }, { "epoch": 178.72530274059912, "grad_norm": 0.11585977673530579, "learning_rate": 1.1602004130157385e-06, "loss": 0.0021, "step": 280420 }, { "epoch": 178.7316762268961, "grad_norm": 0.10715748369693756, "learning_rate": 1.1590202873254962e-06, "loss": 0.0058, "step": 280430 }, { "epoch": 178.7380497131931, "grad_norm": 0.3994298279285431, "learning_rate": 1.1578407550992709e-06, "loss": 0.0066, "step": 280440 }, { "epoch": 178.74442319949011, "grad_norm": 0.08357731997966766, "learning_rate": 1.1566618163513953e-06, "loss": 0.0075, "step": 280450 }, { "epoch": 178.75079668578712, "grad_norm": 0.04926643520593643, "learning_rate": 1.155483471096186e-06, "loss": 0.0089, "step": 280460 }, { "epoch": 178.75717017208413, "grad_norm": 0.10636955499649048, "learning_rate": 1.1543057193479868e-06, "loss": 0.0028, "step": 280470 }, { "epoch": 178.76354365838114, "grad_norm": 0.10287278890609741, "learning_rate": 1.153128561121075e-06, "loss": 0.0038, "step": 280480 }, { "epoch": 178.76991714467815, "grad_norm": 0.11559917032718658, "learning_rate": 1.1519519964297787e-06, "loss": 0.0049, "step": 280490 }, { "epoch": 178.77629063097515, "grad_norm": 0.3895283639431, "learning_rate": 1.1507760252883804e-06, "loss": 0.0098, "step": 280500 }, { "epoch": 178.78266411727213, "grad_norm": 0.06992699950933456, "learning_rate": 1.1496006477111798e-06, "loss": 0.0039, "step": 280510 }, { "epoch": 178.78903760356914, "grad_norm": 0.06693917512893677, "learning_rate": 1.148425863712449e-06, "loss": 0.0153, "step": 280520 }, { "epoch": 178.79541108986615, "grad_norm": 0.21060092747211456, "learning_rate": 1.1472516733064709e-06, "loss": 0.0035, "step": 280530 }, { "epoch": 178.80178457616316, "grad_norm": 0.329528272151947, "learning_rate": 1.1460780765075119e-06, "loss": 0.0052, "step": 280540 }, { "epoch": 178.80815806246017, "grad_norm": 0.27303242683410645, "learning_rate": 1.1449050733298271e-06, "loss": 0.0058, "step": 280550 }, { "epoch": 178.81453154875717, "grad_norm": 0.16734565794467926, "learning_rate": 1.1437326637876667e-06, "loss": 0.0061, "step": 280560 }, { "epoch": 178.82090503505418, "grad_norm": 0.05370815470814705, "learning_rate": 1.142560847895291e-06, "loss": 0.0028, "step": 280570 }, { "epoch": 178.8272785213512, "grad_norm": 0.4712279438972473, "learning_rate": 1.1413896256669276e-06, "loss": 0.0073, "step": 280580 }, { "epoch": 178.8336520076482, "grad_norm": 0.16055147349834442, "learning_rate": 1.1402189971168098e-06, "loss": 0.0048, "step": 280590 }, { "epoch": 178.84002549394518, "grad_norm": 0.12025654315948486, "learning_rate": 1.1390489622591594e-06, "loss": 0.0053, "step": 280600 }, { "epoch": 178.84639898024218, "grad_norm": 0.09023137390613556, "learning_rate": 1.1378795211082039e-06, "loss": 0.0051, "step": 280610 }, { "epoch": 178.8527724665392, "grad_norm": 0.09243688732385635, "learning_rate": 1.1367106736781375e-06, "loss": 0.0054, "step": 280620 }, { "epoch": 178.8591459528362, "grad_norm": 0.05184109881520271, "learning_rate": 1.1355424199831766e-06, "loss": 0.0068, "step": 280630 }, { "epoch": 178.8655194391332, "grad_norm": 0.12598071992397308, "learning_rate": 1.1343747600375043e-06, "loss": 0.004, "step": 280640 }, { "epoch": 178.87189292543022, "grad_norm": 0.0976857990026474, "learning_rate": 1.1332076938553259e-06, "loss": 0.0074, "step": 280650 }, { "epoch": 178.87826641172722, "grad_norm": 0.13037675619125366, "learning_rate": 1.1320412214508024e-06, "loss": 0.0023, "step": 280660 }, { "epoch": 178.88463989802423, "grad_norm": 0.08792120963335037, "learning_rate": 1.1308753428381225e-06, "loss": 0.0058, "step": 280670 }, { "epoch": 178.8910133843212, "grad_norm": 0.08735863864421844, "learning_rate": 1.1297100580314468e-06, "loss": 0.0034, "step": 280680 }, { "epoch": 178.89738687061822, "grad_norm": 0.15088951587677002, "learning_rate": 1.1285453670449364e-06, "loss": 0.0039, "step": 280690 }, { "epoch": 178.90376035691523, "grad_norm": 0.0672035664319992, "learning_rate": 1.1273812698927356e-06, "loss": 0.0061, "step": 280700 }, { "epoch": 178.91013384321224, "grad_norm": 0.06456532329320908, "learning_rate": 1.1262177665889994e-06, "loss": 0.0058, "step": 280710 }, { "epoch": 178.91650732950924, "grad_norm": 0.3589201867580414, "learning_rate": 1.125054857147867e-06, "loss": 0.0042, "step": 280720 }, { "epoch": 178.92288081580625, "grad_norm": 0.09189694374799728, "learning_rate": 1.1238925415834601e-06, "loss": 0.0039, "step": 280730 }, { "epoch": 178.92925430210326, "grad_norm": 0.0649818703532219, "learning_rate": 1.1227308199099007e-06, "loss": 0.0076, "step": 280740 }, { "epoch": 178.93562778840027, "grad_norm": 0.38538068532943726, "learning_rate": 1.121569692141311e-06, "loss": 0.0049, "step": 280750 }, { "epoch": 178.94200127469725, "grad_norm": 0.6261150240898132, "learning_rate": 1.1204091582918019e-06, "loss": 0.0101, "step": 280760 }, { "epoch": 178.94837476099426, "grad_norm": 0.20101138949394226, "learning_rate": 1.1192492183754732e-06, "loss": 0.0031, "step": 280770 }, { "epoch": 178.95474824729126, "grad_norm": 0.27649345993995667, "learning_rate": 1.1180898724064136e-06, "loss": 0.0058, "step": 280780 }, { "epoch": 178.96112173358827, "grad_norm": 0.09787242114543915, "learning_rate": 1.1169311203987175e-06, "loss": 0.0021, "step": 280790 }, { "epoch": 178.96749521988528, "grad_norm": 0.2217566967010498, "learning_rate": 1.115772962366457e-06, "loss": 0.0038, "step": 280800 }, { "epoch": 178.9738687061823, "grad_norm": 0.10285264998674393, "learning_rate": 1.114615398323715e-06, "loss": 0.0075, "step": 280810 }, { "epoch": 178.9802421924793, "grad_norm": 0.38570699095726013, "learning_rate": 1.1134584282845416e-06, "loss": 0.0023, "step": 280820 }, { "epoch": 178.9866156787763, "grad_norm": 0.21072393655776978, "learning_rate": 1.11230205226302e-06, "loss": 0.0084, "step": 280830 }, { "epoch": 178.99298916507328, "grad_norm": 0.047864239662885666, "learning_rate": 1.1111462702731723e-06, "loss": 0.0054, "step": 280840 }, { "epoch": 178.9993626513703, "grad_norm": 0.07979422062635422, "learning_rate": 1.109991082329065e-06, "loss": 0.0056, "step": 280850 }, { "epoch": 179.0057361376673, "grad_norm": 0.1766345351934433, "learning_rate": 1.10883648844472e-06, "loss": 0.0088, "step": 280860 }, { "epoch": 179.0121096239643, "grad_norm": 0.08936595171689987, "learning_rate": 1.1076824886341763e-06, "loss": 0.0114, "step": 280870 }, { "epoch": 179.01848311026131, "grad_norm": 0.09187884628772736, "learning_rate": 1.1065290829114506e-06, "loss": 0.0053, "step": 280880 }, { "epoch": 179.02485659655832, "grad_norm": 0.04872800037264824, "learning_rate": 1.105376271290559e-06, "loss": 0.004, "step": 280890 }, { "epoch": 179.03123008285533, "grad_norm": 0.09736858308315277, "learning_rate": 1.104224053785513e-06, "loss": 0.0028, "step": 280900 }, { "epoch": 179.03760356915234, "grad_norm": 0.34526193141937256, "learning_rate": 1.1030724304103068e-06, "loss": 0.0056, "step": 280910 }, { "epoch": 179.04397705544932, "grad_norm": 0.16832871735095978, "learning_rate": 1.1019214011789347e-06, "loss": 0.0048, "step": 280920 }, { "epoch": 179.05035054174633, "grad_norm": 0.3566513657569885, "learning_rate": 1.100770966105391e-06, "loss": 0.0047, "step": 280930 }, { "epoch": 179.05672402804333, "grad_norm": 0.1370587795972824, "learning_rate": 1.0996211252036425e-06, "loss": 0.0089, "step": 280940 }, { "epoch": 179.06309751434034, "grad_norm": 0.1364271640777588, "learning_rate": 1.0984718784876724e-06, "loss": 0.0033, "step": 280950 }, { "epoch": 179.06947100063735, "grad_norm": 0.3845239281654358, "learning_rate": 1.0973232259714362e-06, "loss": 0.0065, "step": 280960 }, { "epoch": 179.07584448693436, "grad_norm": 0.1014973595738411, "learning_rate": 1.0961751676688947e-06, "loss": 0.0062, "step": 280970 }, { "epoch": 179.08221797323137, "grad_norm": 0.06845126301050186, "learning_rate": 1.0950277035940038e-06, "loss": 0.0039, "step": 280980 }, { "epoch": 179.08859145952837, "grad_norm": 0.09634266048669815, "learning_rate": 1.0938808337606964e-06, "loss": 0.0071, "step": 280990 }, { "epoch": 179.09496494582535, "grad_norm": 0.0808592438697815, "learning_rate": 1.0927345581829118e-06, "loss": 0.0056, "step": 281000 }, { "epoch": 179.10133843212236, "grad_norm": 0.2482740730047226, "learning_rate": 1.0915888768745718e-06, "loss": 0.0049, "step": 281010 }, { "epoch": 179.10771191841937, "grad_norm": 0.10970848053693771, "learning_rate": 1.0904437898496101e-06, "loss": 0.0037, "step": 281020 }, { "epoch": 179.11408540471638, "grad_norm": 0.14692166447639465, "learning_rate": 1.0892992971219374e-06, "loss": 0.0124, "step": 281030 }, { "epoch": 179.12045889101339, "grad_norm": 0.10220568627119064, "learning_rate": 1.0881553987054537e-06, "loss": 0.0039, "step": 281040 }, { "epoch": 179.1268323773104, "grad_norm": 0.05085622891783714, "learning_rate": 1.0870120946140595e-06, "loss": 0.0034, "step": 281050 }, { "epoch": 179.1332058636074, "grad_norm": 0.06549350172281265, "learning_rate": 1.0858693848616542e-06, "loss": 0.0055, "step": 281060 }, { "epoch": 179.1395793499044, "grad_norm": 0.1647864580154419, "learning_rate": 1.0847272694621214e-06, "loss": 0.0038, "step": 281070 }, { "epoch": 179.1459528362014, "grad_norm": 0.3274189531803131, "learning_rate": 1.0835857484293333e-06, "loss": 0.0036, "step": 281080 }, { "epoch": 179.1523263224984, "grad_norm": 0.1088041290640831, "learning_rate": 1.0824448217771566e-06, "loss": 0.0045, "step": 281090 }, { "epoch": 179.1586998087954, "grad_norm": 0.1736629456281662, "learning_rate": 1.081304489519469e-06, "loss": 0.0051, "step": 281100 }, { "epoch": 179.1650732950924, "grad_norm": 0.09260658919811249, "learning_rate": 1.0801647516701208e-06, "loss": 0.006, "step": 281110 }, { "epoch": 179.17144678138942, "grad_norm": 0.09972423315048218, "learning_rate": 1.0790256082429505e-06, "loss": 0.0022, "step": 281120 }, { "epoch": 179.17782026768643, "grad_norm": 0.2673929035663605, "learning_rate": 1.0778870592518143e-06, "loss": 0.0033, "step": 281130 }, { "epoch": 179.18419375398344, "grad_norm": 0.1840846836566925, "learning_rate": 1.0767491047105393e-06, "loss": 0.0048, "step": 281140 }, { "epoch": 179.19056724028044, "grad_norm": 0.10392381250858307, "learning_rate": 1.0756117446329538e-06, "loss": 0.0026, "step": 281150 }, { "epoch": 179.19694072657742, "grad_norm": 0.16279232501983643, "learning_rate": 1.0744749790328744e-06, "loss": 0.0024, "step": 281160 }, { "epoch": 179.20331421287443, "grad_norm": 0.14405913650989532, "learning_rate": 1.073338807924118e-06, "loss": 0.0047, "step": 281170 }, { "epoch": 179.20968769917144, "grad_norm": 0.5418949127197266, "learning_rate": 1.0722032313204955e-06, "loss": 0.005, "step": 281180 }, { "epoch": 179.21606118546845, "grad_norm": 0.11272456496953964, "learning_rate": 1.0710682492357904e-06, "loss": 0.009, "step": 281190 }, { "epoch": 179.22243467176546, "grad_norm": 0.17894600331783295, "learning_rate": 1.069933861683814e-06, "loss": 0.0103, "step": 281200 }, { "epoch": 179.22880815806246, "grad_norm": 0.16639865934848785, "learning_rate": 1.0688000686783274e-06, "loss": 0.0045, "step": 281210 }, { "epoch": 179.23518164435947, "grad_norm": 0.08146460354328156, "learning_rate": 1.0676668702331304e-06, "loss": 0.0026, "step": 281220 }, { "epoch": 179.24155513065648, "grad_norm": 0.2520894706249237, "learning_rate": 1.0665342663619682e-06, "loss": 0.0068, "step": 281230 }, { "epoch": 179.2479286169535, "grad_norm": 0.12118112295866013, "learning_rate": 1.0654022570786237e-06, "loss": 0.003, "step": 281240 }, { "epoch": 179.25430210325047, "grad_norm": 0.053088512271642685, "learning_rate": 1.0642708423968417e-06, "loss": 0.0028, "step": 281250 }, { "epoch": 179.26067558954747, "grad_norm": 0.11797679215669632, "learning_rate": 1.0631400223303722e-06, "loss": 0.0033, "step": 281260 }, { "epoch": 179.26704907584448, "grad_norm": 0.11263838410377502, "learning_rate": 1.0620097968929488e-06, "loss": 0.0056, "step": 281270 }, { "epoch": 179.2734225621415, "grad_norm": 0.16777442395687103, "learning_rate": 1.0608801660983214e-06, "loss": 0.0054, "step": 281280 }, { "epoch": 179.2797960484385, "grad_norm": 0.17581017315387726, "learning_rate": 1.0597511299602015e-06, "loss": 0.0068, "step": 281290 }, { "epoch": 179.2861695347355, "grad_norm": 0.12613365054130554, "learning_rate": 1.0586226884923168e-06, "loss": 0.0081, "step": 281300 }, { "epoch": 179.29254302103251, "grad_norm": 0.3140329420566559, "learning_rate": 1.0574948417083674e-06, "loss": 0.0037, "step": 281310 }, { "epoch": 179.29891650732952, "grad_norm": 0.16043810546398163, "learning_rate": 1.0563675896220703e-06, "loss": 0.0032, "step": 281320 }, { "epoch": 179.3052899936265, "grad_norm": 0.36259925365448, "learning_rate": 1.0552409322471146e-06, "loss": 0.0148, "step": 281330 }, { "epoch": 179.3116634799235, "grad_norm": 0.08808572590351105, "learning_rate": 1.0541148695971947e-06, "loss": 0.0023, "step": 281340 }, { "epoch": 179.31803696622052, "grad_norm": 0.09708543121814728, "learning_rate": 1.0529894016859888e-06, "loss": 0.0061, "step": 281350 }, { "epoch": 179.32441045251753, "grad_norm": 0.08820260316133499, "learning_rate": 1.0518645285271856e-06, "loss": 0.0034, "step": 281360 }, { "epoch": 179.33078393881453, "grad_norm": 0.23886646330356598, "learning_rate": 1.05074025013443e-06, "loss": 0.0054, "step": 281370 }, { "epoch": 179.33715742511154, "grad_norm": 0.06229911744594574, "learning_rate": 1.0496165665214054e-06, "loss": 0.0022, "step": 281380 }, { "epoch": 179.34353091140855, "grad_norm": 0.16473199427127838, "learning_rate": 1.0484934777017451e-06, "loss": 0.0035, "step": 281390 }, { "epoch": 179.34990439770556, "grad_norm": 0.15756934881210327, "learning_rate": 1.047370983689122e-06, "loss": 0.0035, "step": 281400 }, { "epoch": 179.35627788400254, "grad_norm": 0.1166437491774559, "learning_rate": 1.0462490844971474e-06, "loss": 0.0016, "step": 281410 }, { "epoch": 179.36265137029955, "grad_norm": 0.09039419144392014, "learning_rate": 1.0451277801394766e-06, "loss": 0.0028, "step": 281420 }, { "epoch": 179.36902485659655, "grad_norm": 0.11298984289169312, "learning_rate": 1.0440070706297155e-06, "loss": 0.0055, "step": 281430 }, { "epoch": 179.37539834289356, "grad_norm": 0.15802814066410065, "learning_rate": 1.042886955981498e-06, "loss": 0.0029, "step": 281440 }, { "epoch": 179.38177182919057, "grad_norm": 0.1624060422182083, "learning_rate": 1.0417674362084183e-06, "loss": 0.0082, "step": 281450 }, { "epoch": 179.38814531548758, "grad_norm": 0.05804400518536568, "learning_rate": 1.0406485113240993e-06, "loss": 0.0035, "step": 281460 }, { "epoch": 179.39451880178459, "grad_norm": 0.14341790974140167, "learning_rate": 1.0395301813421188e-06, "loss": 0.006, "step": 281470 }, { "epoch": 179.4008922880816, "grad_norm": 0.18028195202350616, "learning_rate": 1.038412446276077e-06, "loss": 0.0025, "step": 281480 }, { "epoch": 179.40726577437857, "grad_norm": 0.13385950028896332, "learning_rate": 1.0372953061395463e-06, "loss": 0.0069, "step": 281490 }, { "epoch": 179.41363926067558, "grad_norm": 0.30690646171569824, "learning_rate": 1.036178760946116e-06, "loss": 0.0048, "step": 281500 }, { "epoch": 179.4200127469726, "grad_norm": 0.23120707273483276, "learning_rate": 1.0350628107093308e-06, "loss": 0.0035, "step": 281510 }, { "epoch": 179.4263862332696, "grad_norm": 0.11027064919471741, "learning_rate": 1.0339474554427685e-06, "loss": 0.0027, "step": 281520 }, { "epoch": 179.4327597195666, "grad_norm": 0.1721785068511963, "learning_rate": 1.0328326951599742e-06, "loss": 0.0084, "step": 281530 }, { "epoch": 179.4391332058636, "grad_norm": 0.18513262271881104, "learning_rate": 1.0317185298745035e-06, "loss": 0.0038, "step": 281540 }, { "epoch": 179.44550669216062, "grad_norm": 0.11597725749015808, "learning_rate": 1.030604959599879e-06, "loss": 0.0027, "step": 281550 }, { "epoch": 179.45188017845763, "grad_norm": 0.3505341708660126, "learning_rate": 1.0294919843496397e-06, "loss": 0.0083, "step": 281560 }, { "epoch": 179.4582536647546, "grad_norm": 0.13899050652980804, "learning_rate": 1.0283796041373084e-06, "loss": 0.0046, "step": 281570 }, { "epoch": 179.46462715105162, "grad_norm": 0.088333860039711, "learning_rate": 1.0272678189764073e-06, "loss": 0.0036, "step": 281580 }, { "epoch": 179.47100063734862, "grad_norm": 0.05368173122406006, "learning_rate": 1.0261566288804315e-06, "loss": 0.0042, "step": 281590 }, { "epoch": 179.47737412364563, "grad_norm": 0.29844263195991516, "learning_rate": 1.0250460338628976e-06, "loss": 0.0064, "step": 281600 }, { "epoch": 179.48374760994264, "grad_norm": 0.12383807450532913, "learning_rate": 1.0239360339372894e-06, "loss": 0.0068, "step": 281610 }, { "epoch": 179.49012109623965, "grad_norm": 0.3786991238594055, "learning_rate": 1.0228266291171073e-06, "loss": 0.0044, "step": 281620 }, { "epoch": 179.49649458253666, "grad_norm": 0.19735057651996613, "learning_rate": 1.0217178194158128e-06, "loss": 0.0099, "step": 281630 }, { "epoch": 179.50286806883366, "grad_norm": 0.17408539354801178, "learning_rate": 1.0206096048469005e-06, "loss": 0.0025, "step": 281640 }, { "epoch": 179.50924155513064, "grad_norm": 0.4774239659309387, "learning_rate": 1.0195019854238208e-06, "loss": 0.007, "step": 281650 }, { "epoch": 179.51561504142765, "grad_norm": 0.16806985437870026, "learning_rate": 1.0183949611600408e-06, "loss": 0.0062, "step": 281660 }, { "epoch": 179.52198852772466, "grad_norm": 0.4795120656490326, "learning_rate": 1.0172885320689996e-06, "loss": 0.0089, "step": 281670 }, { "epoch": 179.52836201402167, "grad_norm": 0.06833856552839279, "learning_rate": 1.0161826981641586e-06, "loss": 0.0054, "step": 281680 }, { "epoch": 179.53473550031867, "grad_norm": 0.06866267323493958, "learning_rate": 1.0150774594589408e-06, "loss": 0.007, "step": 281690 }, { "epoch": 179.54110898661568, "grad_norm": 0.1924324780702591, "learning_rate": 1.0139728159667849e-06, "loss": 0.0029, "step": 281700 }, { "epoch": 179.5474824729127, "grad_norm": 0.23129688203334808, "learning_rate": 1.0128687677011084e-06, "loss": 0.0088, "step": 281710 }, { "epoch": 179.5538559592097, "grad_norm": 0.09236233681440353, "learning_rate": 1.011765314675328e-06, "loss": 0.0033, "step": 281720 }, { "epoch": 179.56022944550668, "grad_norm": 0.3997667729854584, "learning_rate": 1.01066245690285e-06, "loss": 0.0061, "step": 281730 }, { "epoch": 179.5666029318037, "grad_norm": 0.46003925800323486, "learning_rate": 1.00956019439708e-06, "loss": 0.0122, "step": 281740 }, { "epoch": 179.5729764181007, "grad_norm": 0.18153157830238342, "learning_rate": 1.0084585271714075e-06, "loss": 0.0066, "step": 281750 }, { "epoch": 179.5793499043977, "grad_norm": 0.13535325229167938, "learning_rate": 1.0073574552392163e-06, "loss": 0.0051, "step": 281760 }, { "epoch": 179.5857233906947, "grad_norm": 0.08941315114498138, "learning_rate": 1.0062569786138954e-06, "loss": 0.0061, "step": 281770 }, { "epoch": 179.59209687699172, "grad_norm": 0.09474664181470871, "learning_rate": 1.0051570973088064e-06, "loss": 0.0055, "step": 281780 }, { "epoch": 179.59847036328873, "grad_norm": 0.1437177062034607, "learning_rate": 1.0040578113373167e-06, "loss": 0.0059, "step": 281790 }, { "epoch": 179.60484384958573, "grad_norm": 0.23174208402633667, "learning_rate": 1.002959120712782e-06, "loss": 0.0044, "step": 281800 }, { "epoch": 179.61121733588274, "grad_norm": 0.13539722561836243, "learning_rate": 1.0018610254485582e-06, "loss": 0.003, "step": 281810 }, { "epoch": 179.61759082217972, "grad_norm": 0.07964546233415604, "learning_rate": 1.000763525557985e-06, "loss": 0.0032, "step": 281820 }, { "epoch": 179.62396430847673, "grad_norm": 0.1520102471113205, "learning_rate": 9.996666210544014e-07, "loss": 0.0073, "step": 281830 }, { "epoch": 179.63033779477374, "grad_norm": 0.06757012009620667, "learning_rate": 9.985703119511247e-07, "loss": 0.004, "step": 281840 }, { "epoch": 179.63671128107075, "grad_norm": 0.24750371277332306, "learning_rate": 9.97474598261483e-07, "loss": 0.0081, "step": 281850 }, { "epoch": 179.64308476736775, "grad_norm": 0.05345990136265755, "learning_rate": 9.963794799987991e-07, "loss": 0.0053, "step": 281860 }, { "epoch": 179.64945825366476, "grad_norm": 0.2851940393447876, "learning_rate": 9.95284957176368e-07, "loss": 0.0062, "step": 281870 }, { "epoch": 179.65583173996177, "grad_norm": 0.1295059621334076, "learning_rate": 9.941910298074898e-07, "loss": 0.0044, "step": 281880 }, { "epoch": 179.66220522625878, "grad_norm": 0.4127437174320221, "learning_rate": 9.930976979054652e-07, "loss": 0.0045, "step": 281890 }, { "epoch": 179.66857871255576, "grad_norm": 0.1575554460287094, "learning_rate": 9.92004961483567e-07, "loss": 0.0063, "step": 281900 }, { "epoch": 179.67495219885276, "grad_norm": 0.10100632160902023, "learning_rate": 9.9091282055509e-07, "loss": 0.0066, "step": 281910 }, { "epoch": 179.68132568514977, "grad_norm": 0.2636266350746155, "learning_rate": 9.898212751332847e-07, "loss": 0.0103, "step": 281920 }, { "epoch": 179.68769917144678, "grad_norm": 0.48814907670021057, "learning_rate": 9.887303252314295e-07, "loss": 0.0072, "step": 281930 }, { "epoch": 179.6940726577438, "grad_norm": 0.18679389357566833, "learning_rate": 9.876399708627694e-07, "loss": 0.0073, "step": 281940 }, { "epoch": 179.7004461440408, "grad_norm": 0.35127097368240356, "learning_rate": 9.865502120405657e-07, "loss": 0.0085, "step": 281950 }, { "epoch": 179.7068196303378, "grad_norm": 0.07202362269163132, "learning_rate": 9.854610487780469e-07, "loss": 0.0035, "step": 281960 }, { "epoch": 179.7131931166348, "grad_norm": 0.39090847969055176, "learning_rate": 9.843724810884636e-07, "loss": 0.0037, "step": 281970 }, { "epoch": 179.7195666029318, "grad_norm": 0.268439918756485, "learning_rate": 9.832845089850217e-07, "loss": 0.0049, "step": 281980 }, { "epoch": 179.7259400892288, "grad_norm": 0.312531977891922, "learning_rate": 9.821971324809608e-07, "loss": 0.005, "step": 281990 }, { "epoch": 179.7323135755258, "grad_norm": 0.15580786764621735, "learning_rate": 9.811103515894815e-07, "loss": 0.0058, "step": 282000 }, { "epoch": 179.73868706182282, "grad_norm": 0.07131128013134003, "learning_rate": 9.800241663237952e-07, "loss": 0.0062, "step": 282010 }, { "epoch": 179.74506054811982, "grad_norm": 0.1936279535293579, "learning_rate": 9.789385766970916e-07, "loss": 0.0039, "step": 282020 }, { "epoch": 179.75143403441683, "grad_norm": 0.22753258049488068, "learning_rate": 9.77853582722571e-07, "loss": 0.0055, "step": 282030 }, { "epoch": 179.75780752071384, "grad_norm": 0.10169927030801773, "learning_rate": 9.767691844134174e-07, "loss": 0.0054, "step": 282040 }, { "epoch": 179.76418100701085, "grad_norm": 0.07601547986268997, "learning_rate": 9.756853817828038e-07, "loss": 0.0041, "step": 282050 }, { "epoch": 179.77055449330783, "grad_norm": 0.12096378207206726, "learning_rate": 9.746021748438915e-07, "loss": 0.0048, "step": 282060 }, { "epoch": 179.77692797960484, "grad_norm": 0.4720210134983063, "learning_rate": 9.735195636098648e-07, "loss": 0.0064, "step": 282070 }, { "epoch": 179.78330146590184, "grad_norm": 0.15875092148780823, "learning_rate": 9.72437548093852e-07, "loss": 0.0032, "step": 282080 }, { "epoch": 179.78967495219885, "grad_norm": 0.21021117269992828, "learning_rate": 9.713561283090144e-07, "loss": 0.0044, "step": 282090 }, { "epoch": 179.79604843849586, "grad_norm": 0.2600403428077698, "learning_rate": 9.70275304268492e-07, "loss": 0.0145, "step": 282100 }, { "epoch": 179.80242192479287, "grad_norm": 0.14041563868522644, "learning_rate": 9.691950759854184e-07, "loss": 0.0033, "step": 282110 }, { "epoch": 179.80879541108987, "grad_norm": 0.23183897137641907, "learning_rate": 9.68115443472911e-07, "loss": 0.0073, "step": 282120 }, { "epoch": 179.81516889738688, "grad_norm": 0.16231004893779755, "learning_rate": 9.67036406744104e-07, "loss": 0.007, "step": 282130 }, { "epoch": 179.82154238368386, "grad_norm": 0.08211987465620041, "learning_rate": 9.65957965812092e-07, "loss": 0.0049, "step": 282140 }, { "epoch": 179.82791586998087, "grad_norm": 0.07780905067920685, "learning_rate": 9.648801206899926e-07, "loss": 0.0043, "step": 282150 }, { "epoch": 179.83428935627788, "grad_norm": 0.12781545519828796, "learning_rate": 9.638028713908897e-07, "loss": 0.004, "step": 282160 }, { "epoch": 179.8406628425749, "grad_norm": 0.08070769160985947, "learning_rate": 9.62726217927884e-07, "loss": 0.0045, "step": 282170 }, { "epoch": 179.8470363288719, "grad_norm": 0.09148270636796951, "learning_rate": 9.616501603140536e-07, "loss": 0.0028, "step": 282180 }, { "epoch": 179.8534098151689, "grad_norm": 0.13819344341754913, "learning_rate": 9.605746985624776e-07, "loss": 0.0067, "step": 282190 }, { "epoch": 179.8597833014659, "grad_norm": 0.16305455565452576, "learning_rate": 9.594998326862114e-07, "loss": 0.0034, "step": 282200 }, { "epoch": 179.86615678776292, "grad_norm": 0.13789261877536774, "learning_rate": 9.58425562698334e-07, "loss": 0.005, "step": 282210 }, { "epoch": 179.8725302740599, "grad_norm": 0.19726340472698212, "learning_rate": 9.573518886118793e-07, "loss": 0.0023, "step": 282220 }, { "epoch": 179.8789037603569, "grad_norm": 0.0998646542429924, "learning_rate": 9.562788104399034e-07, "loss": 0.0048, "step": 282230 }, { "epoch": 179.8852772466539, "grad_norm": 0.2641221880912781, "learning_rate": 9.55206328195446e-07, "loss": 0.0071, "step": 282240 }, { "epoch": 179.89165073295092, "grad_norm": 0.5552669763565063, "learning_rate": 9.541344418915466e-07, "loss": 0.0074, "step": 282250 }, { "epoch": 179.89802421924793, "grad_norm": 0.08979152143001556, "learning_rate": 9.530631515412114e-07, "loss": 0.0093, "step": 282260 }, { "epoch": 179.90439770554494, "grad_norm": 0.2766965627670288, "learning_rate": 9.519924571574689e-07, "loss": 0.0131, "step": 282270 }, { "epoch": 179.91077119184195, "grad_norm": 0.09880847483873367, "learning_rate": 9.509223587533256e-07, "loss": 0.0048, "step": 282280 }, { "epoch": 179.91714467813895, "grad_norm": 0.21862512826919556, "learning_rate": 9.49852856341793e-07, "loss": 0.0084, "step": 282290 }, { "epoch": 179.92351816443593, "grad_norm": 0.24121667444705963, "learning_rate": 9.487839499358498e-07, "loss": 0.0023, "step": 282300 }, { "epoch": 179.92989165073294, "grad_norm": 0.2283884584903717, "learning_rate": 9.477156395484965e-07, "loss": 0.0039, "step": 282310 }, { "epoch": 179.93626513702995, "grad_norm": 0.15008467435836792, "learning_rate": 9.466479251927174e-07, "loss": 0.0021, "step": 282320 }, { "epoch": 179.94263862332696, "grad_norm": 0.15938633680343628, "learning_rate": 9.455808068814742e-07, "loss": 0.0068, "step": 282330 }, { "epoch": 179.94901210962396, "grad_norm": 0.2572537362575531, "learning_rate": 9.445142846277344e-07, "loss": 0.0066, "step": 282340 }, { "epoch": 179.95538559592097, "grad_norm": 0.3247220516204834, "learning_rate": 9.434483584444709e-07, "loss": 0.0028, "step": 282350 }, { "epoch": 179.96175908221798, "grad_norm": 0.1558220535516739, "learning_rate": 9.423830283446233e-07, "loss": 0.0053, "step": 282360 }, { "epoch": 179.968132568515, "grad_norm": 0.11348523199558258, "learning_rate": 9.413182943411425e-07, "loss": 0.0052, "step": 282370 }, { "epoch": 179.97450605481197, "grad_norm": 0.1844884157180786, "learning_rate": 9.402541564469569e-07, "loss": 0.0063, "step": 282380 }, { "epoch": 179.98087954110898, "grad_norm": 0.2608223855495453, "learning_rate": 9.391906146750118e-07, "loss": 0.0043, "step": 282390 }, { "epoch": 179.98725302740598, "grad_norm": 0.34773969650268555, "learning_rate": 9.38127669038219e-07, "loss": 0.0098, "step": 282400 }, { "epoch": 179.993626513703, "grad_norm": 0.16598287224769592, "learning_rate": 9.370653195494961e-07, "loss": 0.0043, "step": 282410 }, { "epoch": 180.0, "grad_norm": 0.22029265761375427, "learning_rate": 9.360035662217492e-07, "loss": 0.0056, "step": 282420 }, { "epoch": 180.006373486297, "grad_norm": 0.17364367842674255, "learning_rate": 9.349424090678905e-07, "loss": 0.0034, "step": 282430 }, { "epoch": 180.01274697259402, "grad_norm": 0.09281522035598755, "learning_rate": 9.338818481007983e-07, "loss": 0.0097, "step": 282440 }, { "epoch": 180.01912045889102, "grad_norm": 0.10455110669136047, "learning_rate": 9.328218833333735e-07, "loss": 0.0044, "step": 282450 }, { "epoch": 180.02549394518803, "grad_norm": 0.08958447724580765, "learning_rate": 9.317625147784892e-07, "loss": 0.005, "step": 282460 }, { "epoch": 180.031867431485, "grad_norm": 0.2693827450275421, "learning_rate": 9.307037424490184e-07, "loss": 0.0121, "step": 282470 }, { "epoch": 180.03824091778202, "grad_norm": 0.13661964237689972, "learning_rate": 9.29645566357823e-07, "loss": 0.0038, "step": 282480 }, { "epoch": 180.04461440407903, "grad_norm": 0.09446237236261368, "learning_rate": 9.285879865177704e-07, "loss": 0.0017, "step": 282490 }, { "epoch": 180.05098789037604, "grad_norm": 0.2649889886379242, "learning_rate": 9.275310029417006e-07, "loss": 0.0032, "step": 282500 }, { "epoch": 180.05736137667304, "grad_norm": 0.391939252614975, "learning_rate": 9.264746156424587e-07, "loss": 0.0058, "step": 282510 }, { "epoch": 180.06373486297005, "grad_norm": 0.09230215102434158, "learning_rate": 9.254188246328898e-07, "loss": 0.003, "step": 282520 }, { "epoch": 180.07010834926706, "grad_norm": 0.13282045722007751, "learning_rate": 9.243636299258119e-07, "loss": 0.0036, "step": 282530 }, { "epoch": 180.07648183556407, "grad_norm": 0.3889681100845337, "learning_rate": 9.233090315340532e-07, "loss": 0.0032, "step": 282540 }, { "epoch": 180.08285532186105, "grad_norm": 0.08504128456115723, "learning_rate": 9.222550294704202e-07, "loss": 0.007, "step": 282550 }, { "epoch": 180.08922880815805, "grad_norm": 0.13456803560256958, "learning_rate": 9.212016237477305e-07, "loss": 0.0042, "step": 282560 }, { "epoch": 180.09560229445506, "grad_norm": 0.1250506341457367, "learning_rate": 9.201488143787795e-07, "loss": 0.0034, "step": 282570 }, { "epoch": 180.10197578075207, "grad_norm": 0.15112076699733734, "learning_rate": 9.190966013763624e-07, "loss": 0.0059, "step": 282580 }, { "epoch": 180.10834926704908, "grad_norm": 0.07551009207963943, "learning_rate": 9.180449847532524e-07, "loss": 0.0047, "step": 282590 }, { "epoch": 180.1147227533461, "grad_norm": 0.563389241695404, "learning_rate": 9.169939645222447e-07, "loss": 0.0108, "step": 282600 }, { "epoch": 180.1210962396431, "grad_norm": 0.36518192291259766, "learning_rate": 9.159435406961015e-07, "loss": 0.0035, "step": 282610 }, { "epoch": 180.1274697259401, "grad_norm": 0.11832883954048157, "learning_rate": 9.148937132875846e-07, "loss": 0.0032, "step": 282620 }, { "epoch": 180.13384321223708, "grad_norm": 0.0829467922449112, "learning_rate": 9.138444823094505e-07, "loss": 0.005, "step": 282630 }, { "epoch": 180.1402166985341, "grad_norm": 0.08319269865751266, "learning_rate": 9.127958477744613e-07, "loss": 0.0088, "step": 282640 }, { "epoch": 180.1465901848311, "grad_norm": 0.17942126095294952, "learning_rate": 9.117478096953347e-07, "loss": 0.0139, "step": 282650 }, { "epoch": 180.1529636711281, "grad_norm": 0.04812699556350708, "learning_rate": 9.107003680848269e-07, "loss": 0.0071, "step": 282660 }, { "epoch": 180.1593371574251, "grad_norm": 0.43587005138397217, "learning_rate": 9.096535229556557e-07, "loss": 0.0082, "step": 282670 }, { "epoch": 180.16571064372212, "grad_norm": 0.10634618252515793, "learning_rate": 9.086072743205498e-07, "loss": 0.0053, "step": 282680 }, { "epoch": 180.17208413001913, "grad_norm": 0.0814230889081955, "learning_rate": 9.075616221922101e-07, "loss": 0.0118, "step": 282690 }, { "epoch": 180.17845761631614, "grad_norm": 0.11800096929073334, "learning_rate": 9.065165665833486e-07, "loss": 0.0078, "step": 282700 }, { "epoch": 180.18483110261312, "grad_norm": 0.2102479189634323, "learning_rate": 9.054721075066608e-07, "loss": 0.0034, "step": 282710 }, { "epoch": 180.19120458891013, "grad_norm": 0.05022500082850456, "learning_rate": 9.04428244974842e-07, "loss": 0.007, "step": 282720 }, { "epoch": 180.19757807520713, "grad_norm": 0.0809963047504425, "learning_rate": 9.03384979000571e-07, "loss": 0.0071, "step": 282730 }, { "epoch": 180.20395156150414, "grad_norm": 0.1394580751657486, "learning_rate": 9.023423095965266e-07, "loss": 0.0042, "step": 282740 }, { "epoch": 180.21032504780115, "grad_norm": 0.10358662158250809, "learning_rate": 9.013002367753876e-07, "loss": 0.006, "step": 282750 }, { "epoch": 180.21669853409816, "grad_norm": 0.15106765925884247, "learning_rate": 9.002587605497992e-07, "loss": 0.0053, "step": 282760 }, { "epoch": 180.22307202039516, "grad_norm": 0.08154947310686111, "learning_rate": 8.992178809324237e-07, "loss": 0.0046, "step": 282770 }, { "epoch": 180.22944550669217, "grad_norm": 0.1134631335735321, "learning_rate": 8.981775979359175e-07, "loss": 0.0028, "step": 282780 }, { "epoch": 180.23581899298915, "grad_norm": 0.2255517691373825, "learning_rate": 8.971379115729095e-07, "loss": 0.0044, "step": 282790 }, { "epoch": 180.24219247928616, "grad_norm": 0.2963677644729614, "learning_rate": 8.96098821856034e-07, "loss": 0.0049, "step": 282800 }, { "epoch": 180.24856596558317, "grad_norm": 0.05892915278673172, "learning_rate": 8.950603287979197e-07, "loss": 0.004, "step": 282810 }, { "epoch": 180.25493945188018, "grad_norm": 0.6014948487281799, "learning_rate": 8.940224324111901e-07, "loss": 0.0095, "step": 282820 }, { "epoch": 180.26131293817718, "grad_norm": 0.1031215637922287, "learning_rate": 8.92985132708446e-07, "loss": 0.0055, "step": 282830 }, { "epoch": 180.2676864244742, "grad_norm": 0.06567951291799545, "learning_rate": 8.919484297023051e-07, "loss": 0.0043, "step": 282840 }, { "epoch": 180.2740599107712, "grad_norm": 0.09801530838012695, "learning_rate": 8.909123234053463e-07, "loss": 0.0028, "step": 282850 }, { "epoch": 180.2804333970682, "grad_norm": 0.2967608571052551, "learning_rate": 8.898768138301817e-07, "loss": 0.0045, "step": 282860 }, { "epoch": 180.2868068833652, "grad_norm": 0.3738230764865875, "learning_rate": 8.888419009893679e-07, "loss": 0.0044, "step": 282870 }, { "epoch": 180.2931803696622, "grad_norm": 0.2330240160226822, "learning_rate": 8.878075848955004e-07, "loss": 0.0052, "step": 282880 }, { "epoch": 180.2995538559592, "grad_norm": 0.10048506408929825, "learning_rate": 8.867738655611413e-07, "loss": 0.0044, "step": 282890 }, { "epoch": 180.3059273422562, "grad_norm": 0.07917457073926926, "learning_rate": 8.857407429988529e-07, "loss": 0.004, "step": 282900 }, { "epoch": 180.31230082855322, "grad_norm": 0.11106306314468384, "learning_rate": 8.847082172211752e-07, "loss": 0.0074, "step": 282910 }, { "epoch": 180.31867431485023, "grad_norm": 0.15473748743534088, "learning_rate": 8.836762882406757e-07, "loss": 0.0106, "step": 282920 }, { "epoch": 180.32504780114724, "grad_norm": 0.06859946250915527, "learning_rate": 8.826449560698779e-07, "loss": 0.0077, "step": 282930 }, { "epoch": 180.33142128744424, "grad_norm": 0.1738673895597458, "learning_rate": 8.816142207213218e-07, "loss": 0.0031, "step": 282940 }, { "epoch": 180.33779477374122, "grad_norm": 0.08864890784025192, "learning_rate": 8.805840822075251e-07, "loss": 0.0056, "step": 282950 }, { "epoch": 180.34416826003823, "grad_norm": 0.1432112604379654, "learning_rate": 8.79554540541011e-07, "loss": 0.0136, "step": 282960 }, { "epoch": 180.35054174633524, "grad_norm": 0.427723228931427, "learning_rate": 8.785255957342808e-07, "loss": 0.0034, "step": 282970 }, { "epoch": 180.35691523263225, "grad_norm": 0.4993727207183838, "learning_rate": 8.774972477998522e-07, "loss": 0.0066, "step": 282980 }, { "epoch": 180.36328871892925, "grad_norm": 0.10385897755622864, "learning_rate": 8.764694967501985e-07, "loss": 0.0059, "step": 282990 }, { "epoch": 180.36966220522626, "grad_norm": 0.1689101606607437, "learning_rate": 8.754423425978375e-07, "loss": 0.002, "step": 283000 }, { "epoch": 180.37603569152327, "grad_norm": 0.3835449516773224, "learning_rate": 8.744157853552204e-07, "loss": 0.0039, "step": 283010 }, { "epoch": 180.38240917782028, "grad_norm": 0.4145818054676056, "learning_rate": 8.733898250348371e-07, "loss": 0.0059, "step": 283020 }, { "epoch": 180.38878266411726, "grad_norm": 0.07040996104478836, "learning_rate": 8.7236446164915e-07, "loss": 0.0071, "step": 283030 }, { "epoch": 180.39515615041427, "grad_norm": 0.20127363502979279, "learning_rate": 8.713396952106212e-07, "loss": 0.0047, "step": 283040 }, { "epoch": 180.40152963671127, "grad_norm": 0.14771297574043274, "learning_rate": 8.703155257316964e-07, "loss": 0.0037, "step": 283050 }, { "epoch": 180.40790312300828, "grad_norm": 0.10801421850919724, "learning_rate": 8.692919532248267e-07, "loss": 0.0052, "step": 283060 }, { "epoch": 180.4142766093053, "grad_norm": 0.09718827903270721, "learning_rate": 8.682689777024522e-07, "loss": 0.0045, "step": 283070 }, { "epoch": 180.4206500956023, "grad_norm": 0.3556573688983917, "learning_rate": 8.672465991769907e-07, "loss": 0.0044, "step": 283080 }, { "epoch": 180.4270235818993, "grad_norm": 0.08757522702217102, "learning_rate": 8.662248176608712e-07, "loss": 0.0167, "step": 283090 }, { "epoch": 180.4333970681963, "grad_norm": 0.3083729147911072, "learning_rate": 8.652036331665115e-07, "loss": 0.0131, "step": 283100 }, { "epoch": 180.43977055449332, "grad_norm": 0.1146535724401474, "learning_rate": 8.641830457063238e-07, "loss": 0.003, "step": 283110 }, { "epoch": 180.4461440407903, "grad_norm": 0.20871618390083313, "learning_rate": 8.631630552926983e-07, "loss": 0.0032, "step": 283120 }, { "epoch": 180.4525175270873, "grad_norm": 0.20428618788719177, "learning_rate": 8.621436619380363e-07, "loss": 0.0064, "step": 283130 }, { "epoch": 180.45889101338432, "grad_norm": 0.09977416694164276, "learning_rate": 8.611248656547222e-07, "loss": 0.0035, "step": 283140 }, { "epoch": 180.46526449968133, "grad_norm": 0.10550451278686523, "learning_rate": 8.601066664551349e-07, "loss": 0.004, "step": 283150 }, { "epoch": 180.47163798597833, "grad_norm": 0.050487056374549866, "learning_rate": 8.59089064351648e-07, "loss": 0.0112, "step": 283160 }, { "epoch": 180.47801147227534, "grad_norm": 0.2873263657093048, "learning_rate": 8.580720593566238e-07, "loss": 0.0045, "step": 283170 }, { "epoch": 180.48438495857235, "grad_norm": 0.40878087282180786, "learning_rate": 8.570556514824246e-07, "loss": 0.004, "step": 283180 }, { "epoch": 180.49075844486936, "grad_norm": 0.17037856578826904, "learning_rate": 8.560398407414016e-07, "loss": 0.004, "step": 283190 }, { "epoch": 180.49713193116634, "grad_norm": 0.07699500769376755, "learning_rate": 8.550246271458895e-07, "loss": 0.0055, "step": 283200 }, { "epoch": 180.50350541746334, "grad_norm": 0.1147848591208458, "learning_rate": 8.540100107082283e-07, "loss": 0.0053, "step": 283210 }, { "epoch": 180.50987890376035, "grad_norm": 0.4062371551990509, "learning_rate": 8.52995991440747e-07, "loss": 0.0041, "step": 283220 }, { "epoch": 180.51625239005736, "grad_norm": 0.08063340187072754, "learning_rate": 8.519825693557692e-07, "loss": 0.0088, "step": 283230 }, { "epoch": 180.52262587635437, "grad_norm": 0.20724938809871674, "learning_rate": 8.509697444656018e-07, "loss": 0.0036, "step": 283240 }, { "epoch": 180.52899936265138, "grad_norm": 0.13882054388523102, "learning_rate": 8.499575167825624e-07, "loss": 0.0183, "step": 283250 }, { "epoch": 180.53537284894838, "grad_norm": 0.09129714220762253, "learning_rate": 8.489458863189414e-07, "loss": 0.004, "step": 283260 }, { "epoch": 180.5417463352454, "grad_norm": 0.11231055110692978, "learning_rate": 8.479348530870346e-07, "loss": 0.0074, "step": 283270 }, { "epoch": 180.54811982154237, "grad_norm": 0.4342598021030426, "learning_rate": 8.469244170991264e-07, "loss": 0.0108, "step": 283280 }, { "epoch": 180.55449330783938, "grad_norm": 0.28977271914482117, "learning_rate": 8.45914578367496e-07, "loss": 0.0043, "step": 283290 }, { "epoch": 180.5608667941364, "grad_norm": 0.18859584629535675, "learning_rate": 8.449053369044058e-07, "loss": 0.008, "step": 283300 }, { "epoch": 180.5672402804334, "grad_norm": 0.24921844899654388, "learning_rate": 8.438966927221348e-07, "loss": 0.0039, "step": 283310 }, { "epoch": 180.5736137667304, "grad_norm": 0.2348395138978958, "learning_rate": 8.428886458329288e-07, "loss": 0.0069, "step": 283320 }, { "epoch": 180.5799872530274, "grad_norm": 0.116031713783741, "learning_rate": 8.418811962490336e-07, "loss": 0.0035, "step": 283330 }, { "epoch": 180.58636073932442, "grad_norm": 0.08110268414020538, "learning_rate": 8.40874343982695e-07, "loss": 0.0102, "step": 283340 }, { "epoch": 180.59273422562143, "grad_norm": 0.5476280450820923, "learning_rate": 8.39868089046153e-07, "loss": 0.0038, "step": 283350 }, { "epoch": 180.5991077119184, "grad_norm": 0.38031139969825745, "learning_rate": 8.388624314516202e-07, "loss": 0.0123, "step": 283360 }, { "epoch": 180.60548119821541, "grad_norm": 0.07429445534944534, "learning_rate": 8.378573712113313e-07, "loss": 0.0084, "step": 283370 }, { "epoch": 180.61185468451242, "grad_norm": 0.1323961466550827, "learning_rate": 8.368529083374876e-07, "loss": 0.0025, "step": 283380 }, { "epoch": 180.61822817080943, "grad_norm": 0.15698552131652832, "learning_rate": 8.358490428423072e-07, "loss": 0.0042, "step": 283390 }, { "epoch": 180.62460165710644, "grad_norm": 0.26365983486175537, "learning_rate": 8.348457747379689e-07, "loss": 0.0037, "step": 283400 }, { "epoch": 180.63097514340345, "grad_norm": 0.10969981551170349, "learning_rate": 8.338431040366801e-07, "loss": 0.0053, "step": 283410 }, { "epoch": 180.63734862970045, "grad_norm": 0.18252550065517426, "learning_rate": 8.32841030750614e-07, "loss": 0.0043, "step": 283420 }, { "epoch": 180.64372211599746, "grad_norm": 0.2459532618522644, "learning_rate": 8.318395548919555e-07, "loss": 0.0087, "step": 283430 }, { "epoch": 180.65009560229444, "grad_norm": 0.3244854509830475, "learning_rate": 8.308386764728671e-07, "loss": 0.0063, "step": 283440 }, { "epoch": 180.65646908859145, "grad_norm": 0.20626172423362732, "learning_rate": 8.298383955055111e-07, "loss": 0.0042, "step": 283450 }, { "epoch": 180.66284257488846, "grad_norm": 0.13507282733917236, "learning_rate": 8.288387120020447e-07, "loss": 0.0115, "step": 283460 }, { "epoch": 180.66921606118547, "grad_norm": 0.08045528084039688, "learning_rate": 8.278396259746135e-07, "loss": 0.0063, "step": 283470 }, { "epoch": 180.67558954748247, "grad_norm": 0.14308752119541168, "learning_rate": 8.268411374353524e-07, "loss": 0.0031, "step": 283480 }, { "epoch": 180.68196303377948, "grad_norm": 0.11427893489599228, "learning_rate": 8.258432463964016e-07, "loss": 0.0032, "step": 283490 }, { "epoch": 180.6883365200765, "grad_norm": 0.48909899592399597, "learning_rate": 8.248459528698849e-07, "loss": 0.0104, "step": 283500 }, { "epoch": 180.6947100063735, "grad_norm": 0.3528842628002167, "learning_rate": 8.238492568679202e-07, "loss": 0.0058, "step": 283510 }, { "epoch": 180.70108349267048, "grad_norm": 0.11030014604330063, "learning_rate": 8.22853158402609e-07, "loss": 0.0149, "step": 283520 }, { "epoch": 180.70745697896749, "grad_norm": 0.49329161643981934, "learning_rate": 8.218576574860748e-07, "loss": 0.0046, "step": 283530 }, { "epoch": 180.7138304652645, "grad_norm": 0.08065935969352722, "learning_rate": 8.208627541303914e-07, "loss": 0.0037, "step": 283540 }, { "epoch": 180.7202039515615, "grad_norm": 0.3275332748889923, "learning_rate": 8.198684483476604e-07, "loss": 0.0031, "step": 283550 }, { "epoch": 180.7265774378585, "grad_norm": 0.25473257899284363, "learning_rate": 8.188747401499608e-07, "loss": 0.0037, "step": 283560 }, { "epoch": 180.73295092415552, "grad_norm": 0.0758567675948143, "learning_rate": 8.178816295493774e-07, "loss": 0.0041, "step": 283570 }, { "epoch": 180.73932441045253, "grad_norm": 0.31968337297439575, "learning_rate": 8.168891165579561e-07, "loss": 0.008, "step": 283580 }, { "epoch": 180.74569789674953, "grad_norm": 0.1274092197418213, "learning_rate": 8.158972011877763e-07, "loss": 0.0038, "step": 283590 }, { "epoch": 180.7520713830465, "grad_norm": 0.07166480273008347, "learning_rate": 8.149058834508838e-07, "loss": 0.0044, "step": 283600 }, { "epoch": 180.75844486934352, "grad_norm": 0.39930596947669983, "learning_rate": 8.139151633593189e-07, "loss": 0.0089, "step": 283610 }, { "epoch": 180.76481835564053, "grad_norm": 0.09144898504018784, "learning_rate": 8.129250409251222e-07, "loss": 0.004, "step": 283620 }, { "epoch": 180.77119184193754, "grad_norm": 0.12392685562372208, "learning_rate": 8.119355161603282e-07, "loss": 0.0038, "step": 283630 }, { "epoch": 180.77756532823454, "grad_norm": 0.2538257837295532, "learning_rate": 8.109465890769607e-07, "loss": 0.008, "step": 283640 }, { "epoch": 180.78393881453155, "grad_norm": 0.07050325721502304, "learning_rate": 8.099582596870325e-07, "loss": 0.0032, "step": 283650 }, { "epoch": 180.79031230082856, "grad_norm": 0.1844455450773239, "learning_rate": 8.089705280025562e-07, "loss": 0.0025, "step": 283660 }, { "epoch": 180.79668578712557, "grad_norm": 0.152289018034935, "learning_rate": 8.079833940355331e-07, "loss": 0.0125, "step": 283670 }, { "epoch": 180.80305927342255, "grad_norm": 0.09377269446849823, "learning_rate": 8.069968577979536e-07, "loss": 0.0068, "step": 283680 }, { "epoch": 180.80943275971956, "grad_norm": 0.19307103753089905, "learning_rate": 8.060109193018084e-07, "loss": 0.0101, "step": 283690 }, { "epoch": 180.81580624601656, "grad_norm": 0.12033426761627197, "learning_rate": 8.050255785590765e-07, "loss": 0.0039, "step": 283700 }, { "epoch": 180.82217973231357, "grad_norm": 0.10266333818435669, "learning_rate": 8.040408355817375e-07, "loss": 0.0046, "step": 283710 }, { "epoch": 180.82855321861058, "grad_norm": 0.39533883333206177, "learning_rate": 8.030566903817482e-07, "loss": 0.0021, "step": 283720 }, { "epoch": 180.8349267049076, "grad_norm": 0.15210437774658203, "learning_rate": 8.02073142971066e-07, "loss": 0.0027, "step": 283730 }, { "epoch": 180.8413001912046, "grad_norm": 0.06459658592939377, "learning_rate": 8.010901933616478e-07, "loss": 0.0047, "step": 283740 }, { "epoch": 180.8476736775016, "grad_norm": 0.11443714052438736, "learning_rate": 8.001078415654395e-07, "loss": 0.0046, "step": 283750 }, { "epoch": 180.8540471637986, "grad_norm": 0.1366180181503296, "learning_rate": 7.991260875943651e-07, "loss": 0.0053, "step": 283760 }, { "epoch": 180.8604206500956, "grad_norm": 0.5556504726409912, "learning_rate": 7.981449314603706e-07, "loss": 0.0054, "step": 283770 }, { "epoch": 180.8667941363926, "grad_norm": 0.2752401828765869, "learning_rate": 7.97164373175363e-07, "loss": 0.0043, "step": 283780 }, { "epoch": 180.8731676226896, "grad_norm": 0.08759760111570358, "learning_rate": 7.961844127512663e-07, "loss": 0.0035, "step": 283790 }, { "epoch": 180.87954110898661, "grad_norm": 0.15361998975276947, "learning_rate": 7.95205050199982e-07, "loss": 0.0141, "step": 283800 }, { "epoch": 180.88591459528362, "grad_norm": 0.06317306309938431, "learning_rate": 7.94226285533417e-07, "loss": 0.0036, "step": 283810 }, { "epoch": 180.89228808158063, "grad_norm": 0.11067775636911392, "learning_rate": 7.932481187634622e-07, "loss": 0.0057, "step": 283820 }, { "epoch": 180.89866156787764, "grad_norm": 0.16723714768886566, "learning_rate": 7.922705499019967e-07, "loss": 0.0023, "step": 283830 }, { "epoch": 180.90503505417465, "grad_norm": 0.10286586731672287, "learning_rate": 7.912935789609055e-07, "loss": 0.0029, "step": 283840 }, { "epoch": 180.91140854047163, "grad_norm": 0.21236582100391388, "learning_rate": 7.903172059520569e-07, "loss": 0.0056, "step": 283850 }, { "epoch": 180.91778202676863, "grad_norm": 0.17049480974674225, "learning_rate": 7.893414308873193e-07, "loss": 0.0072, "step": 283860 }, { "epoch": 180.92415551306564, "grad_norm": 0.16619837284088135, "learning_rate": 7.883662537785441e-07, "loss": 0.0027, "step": 283870 }, { "epoch": 180.93052899936265, "grad_norm": 0.1667666733264923, "learning_rate": 7.873916746375831e-07, "loss": 0.0042, "step": 283880 }, { "epoch": 180.93690248565966, "grad_norm": 0.12147871404886246, "learning_rate": 7.864176934762768e-07, "loss": 0.0048, "step": 283890 }, { "epoch": 180.94327597195667, "grad_norm": 0.07349622994661331, "learning_rate": 7.854443103064602e-07, "loss": 0.0052, "step": 283900 }, { "epoch": 180.94964945825367, "grad_norm": 0.35153403878211975, "learning_rate": 7.844715251399626e-07, "loss": 0.0073, "step": 283910 }, { "epoch": 180.95602294455068, "grad_norm": 0.06902259588241577, "learning_rate": 7.834993379886025e-07, "loss": 0.0026, "step": 283920 }, { "epoch": 180.96239643084766, "grad_norm": 0.21851374208927155, "learning_rate": 7.825277488641924e-07, "loss": 0.0048, "step": 283930 }, { "epoch": 180.96876991714467, "grad_norm": 0.10074127465486526, "learning_rate": 7.815567577785399e-07, "loss": 0.0058, "step": 283940 }, { "epoch": 180.97514340344168, "grad_norm": 0.21603845059871674, "learning_rate": 7.805863647434464e-07, "loss": 0.0076, "step": 283950 }, { "epoch": 180.98151688973869, "grad_norm": 0.14313453435897827, "learning_rate": 7.796165697706969e-07, "loss": 0.0052, "step": 283960 }, { "epoch": 180.9878903760357, "grad_norm": 0.07417789101600647, "learning_rate": 7.786473728720767e-07, "loss": 0.0022, "step": 283970 }, { "epoch": 180.9942638623327, "grad_norm": 0.4260280430316925, "learning_rate": 7.77678774059365e-07, "loss": 0.007, "step": 283980 }, { "epoch": 181.0006373486297, "grad_norm": 0.10817644745111465, "learning_rate": 7.767107733443301e-07, "loss": 0.0031, "step": 283990 }, { "epoch": 181.00701083492672, "grad_norm": 0.27483728528022766, "learning_rate": 7.757433707387297e-07, "loss": 0.0043, "step": 284000 }, { "epoch": 181.0133843212237, "grad_norm": 0.16335858404636383, "learning_rate": 7.747765662543261e-07, "loss": 0.0055, "step": 284010 }, { "epoch": 181.0197578075207, "grad_norm": 0.1774452030658722, "learning_rate": 7.738103599028601e-07, "loss": 0.0025, "step": 284020 }, { "epoch": 181.0261312938177, "grad_norm": 0.042599570006132126, "learning_rate": 7.728447516960835e-07, "loss": 0.0024, "step": 284030 }, { "epoch": 181.03250478011472, "grad_norm": 0.22808566689491272, "learning_rate": 7.718797416457147e-07, "loss": 0.004, "step": 284040 }, { "epoch": 181.03887826641173, "grad_norm": 0.29800063371658325, "learning_rate": 7.70915329763483e-07, "loss": 0.0073, "step": 284050 }, { "epoch": 181.04525175270874, "grad_norm": 1.3304952383041382, "learning_rate": 7.699515160611181e-07, "loss": 0.0111, "step": 284060 }, { "epoch": 181.05162523900574, "grad_norm": 0.09408178180456161, "learning_rate": 7.689883005503217e-07, "loss": 0.0026, "step": 284070 }, { "epoch": 181.05799872530275, "grad_norm": 0.1284089982509613, "learning_rate": 7.680256832427956e-07, "loss": 0.0023, "step": 284080 }, { "epoch": 181.06437221159973, "grad_norm": 0.08782655000686646, "learning_rate": 7.670636641502416e-07, "loss": 0.0031, "step": 284090 }, { "epoch": 181.07074569789674, "grad_norm": 0.1191529855132103, "learning_rate": 7.661022432843501e-07, "loss": 0.0051, "step": 284100 }, { "epoch": 181.07711918419375, "grad_norm": 0.2526791989803314, "learning_rate": 7.651414206567953e-07, "loss": 0.0045, "step": 284110 }, { "epoch": 181.08349267049076, "grad_norm": 0.12946709990501404, "learning_rate": 7.641811962792622e-07, "loss": 0.0039, "step": 284120 }, { "epoch": 181.08986615678776, "grad_norm": 0.1270047426223755, "learning_rate": 7.632215701634083e-07, "loss": 0.0026, "step": 284130 }, { "epoch": 181.09623964308477, "grad_norm": 0.3377130329608917, "learning_rate": 7.622625423209129e-07, "loss": 0.0076, "step": 284140 }, { "epoch": 181.10261312938178, "grad_norm": 0.19176101684570312, "learning_rate": 7.613041127634002e-07, "loss": 0.0162, "step": 284150 }, { "epoch": 181.1089866156788, "grad_norm": 0.12073685228824615, "learning_rate": 7.603462815025386e-07, "loss": 0.0094, "step": 284160 }, { "epoch": 181.11536010197577, "grad_norm": 0.47645092010498047, "learning_rate": 7.593890485499633e-07, "loss": 0.0084, "step": 284170 }, { "epoch": 181.12173358827278, "grad_norm": 0.5809662342071533, "learning_rate": 7.584324139172982e-07, "loss": 0.005, "step": 284180 }, { "epoch": 181.12810707456978, "grad_norm": 0.12863041460514069, "learning_rate": 7.574763776161676e-07, "loss": 0.0101, "step": 284190 }, { "epoch": 181.1344805608668, "grad_norm": 0.1664825677871704, "learning_rate": 7.565209396581951e-07, "loss": 0.0088, "step": 284200 }, { "epoch": 181.1408540471638, "grad_norm": 0.12182512879371643, "learning_rate": 7.55566100054983e-07, "loss": 0.0053, "step": 284210 }, { "epoch": 181.1472275334608, "grad_norm": 0.14808738231658936, "learning_rate": 7.546118588181439e-07, "loss": 0.0117, "step": 284220 }, { "epoch": 181.15360101975782, "grad_norm": 0.34625253081321716, "learning_rate": 7.536582159592575e-07, "loss": 0.0057, "step": 284230 }, { "epoch": 181.15997450605482, "grad_norm": 0.2031942456960678, "learning_rate": 7.527051714899314e-07, "loss": 0.0048, "step": 284240 }, { "epoch": 181.1663479923518, "grad_norm": 0.12676677107810974, "learning_rate": 7.517527254217283e-07, "loss": 0.0034, "step": 284250 }, { "epoch": 181.1727214786488, "grad_norm": 0.11197886615991592, "learning_rate": 7.508008777662279e-07, "loss": 0.0043, "step": 284260 }, { "epoch": 181.17909496494582, "grad_norm": 0.09675311297178268, "learning_rate": 7.498496285349931e-07, "loss": 0.0062, "step": 284270 }, { "epoch": 181.18546845124283, "grad_norm": 0.21600723266601562, "learning_rate": 7.488989777395927e-07, "loss": 0.0044, "step": 284280 }, { "epoch": 181.19184193753983, "grad_norm": 0.2684369683265686, "learning_rate": 7.479489253915617e-07, "loss": 0.0044, "step": 284290 }, { "epoch": 181.19821542383684, "grad_norm": 0.07874896377325058, "learning_rate": 7.469994715024575e-07, "loss": 0.0042, "step": 284300 }, { "epoch": 181.20458891013385, "grad_norm": 0.4950924813747406, "learning_rate": 7.460506160838098e-07, "loss": 0.015, "step": 284310 }, { "epoch": 181.21096239643086, "grad_norm": 0.42185065150260925, "learning_rate": 7.45102359147154e-07, "loss": 0.0073, "step": 284320 }, { "epoch": 181.21733588272787, "grad_norm": 0.1022353321313858, "learning_rate": 7.441547007040029e-07, "loss": 0.0035, "step": 284330 }, { "epoch": 181.22370936902485, "grad_norm": 0.3768746554851532, "learning_rate": 7.432076407658805e-07, "loss": 0.0035, "step": 284340 }, { "epoch": 181.23008285532185, "grad_norm": 0.10719943791627884, "learning_rate": 7.422611793442946e-07, "loss": 0.0023, "step": 284350 }, { "epoch": 181.23645634161886, "grad_norm": 0.07626024633646011, "learning_rate": 7.413153164507414e-07, "loss": 0.0032, "step": 284360 }, { "epoch": 181.24282982791587, "grad_norm": 0.37505677342414856, "learning_rate": 7.403700520967116e-07, "loss": 0.0043, "step": 284370 }, { "epoch": 181.24920331421288, "grad_norm": 0.4781358540058136, "learning_rate": 7.394253862936962e-07, "loss": 0.0063, "step": 284380 }, { "epoch": 181.25557680050989, "grad_norm": 0.13376176357269287, "learning_rate": 7.384813190531747e-07, "loss": 0.0102, "step": 284390 }, { "epoch": 181.2619502868069, "grad_norm": 0.13807670772075653, "learning_rate": 7.375378503866159e-07, "loss": 0.0063, "step": 284400 }, { "epoch": 181.2683237731039, "grad_norm": 0.15799888968467712, "learning_rate": 7.365949803054772e-07, "loss": 0.0036, "step": 284410 }, { "epoch": 181.27469725940088, "grad_norm": 0.124306321144104, "learning_rate": 7.356527088212328e-07, "loss": 0.0069, "step": 284420 }, { "epoch": 181.2810707456979, "grad_norm": 0.23608365654945374, "learning_rate": 7.347110359453069e-07, "loss": 0.0034, "step": 284430 }, { "epoch": 181.2874442319949, "grad_norm": 0.10190767794847488, "learning_rate": 7.33769961689168e-07, "loss": 0.0032, "step": 284440 }, { "epoch": 181.2938177182919, "grad_norm": 0.1735294610261917, "learning_rate": 7.328294860642293e-07, "loss": 0.0044, "step": 284450 }, { "epoch": 181.3001912045889, "grad_norm": 0.23149830102920532, "learning_rate": 7.318896090819371e-07, "loss": 0.0029, "step": 284460 }, { "epoch": 181.30656469088592, "grad_norm": 0.12495731562376022, "learning_rate": 7.309503307536991e-07, "loss": 0.008, "step": 284470 }, { "epoch": 181.31293817718293, "grad_norm": 0.05770658701658249, "learning_rate": 7.300116510909339e-07, "loss": 0.0025, "step": 284480 }, { "epoch": 181.31931166347994, "grad_norm": 0.16036541759967804, "learning_rate": 7.290735701050433e-07, "loss": 0.0144, "step": 284490 }, { "epoch": 181.32568514977692, "grad_norm": 0.11825509369373322, "learning_rate": 7.281360878074295e-07, "loss": 0.0065, "step": 284500 }, { "epoch": 181.33205863607392, "grad_norm": 0.18686670064926147, "learning_rate": 7.271992042094833e-07, "loss": 0.0098, "step": 284510 }, { "epoch": 181.33843212237093, "grad_norm": 0.07677283138036728, "learning_rate": 7.262629193225846e-07, "loss": 0.0064, "step": 284520 }, { "epoch": 181.34480560866794, "grad_norm": 0.1562020182609558, "learning_rate": 7.253272331581185e-07, "loss": 0.0063, "step": 284530 }, { "epoch": 181.35117909496495, "grad_norm": 0.14274370670318604, "learning_rate": 7.243921457274483e-07, "loss": 0.0029, "step": 284540 }, { "epoch": 181.35755258126196, "grad_norm": 0.2780081629753113, "learning_rate": 7.234576570419316e-07, "loss": 0.0043, "step": 284550 }, { "epoch": 181.36392606755896, "grad_norm": 0.3994666635990143, "learning_rate": 7.22523767112937e-07, "loss": 0.0086, "step": 284560 }, { "epoch": 181.37029955385597, "grad_norm": 0.07213214039802551, "learning_rate": 7.215904759518e-07, "loss": 0.0061, "step": 284570 }, { "epoch": 181.37667304015295, "grad_norm": 0.1054803878068924, "learning_rate": 7.206577835698669e-07, "loss": 0.0034, "step": 284580 }, { "epoch": 181.38304652644996, "grad_norm": 0.08615777641534805, "learning_rate": 7.197256899784677e-07, "loss": 0.0045, "step": 284590 }, { "epoch": 181.38942001274697, "grad_norm": 0.10119548439979553, "learning_rate": 7.187941951889321e-07, "loss": 0.0025, "step": 284600 }, { "epoch": 181.39579349904398, "grad_norm": 0.09280990809202194, "learning_rate": 7.178632992125733e-07, "loss": 0.0035, "step": 284610 }, { "epoch": 181.40216698534098, "grad_norm": 0.3856104910373688, "learning_rate": 7.169330020607102e-07, "loss": 0.0066, "step": 284620 }, { "epoch": 181.408540471638, "grad_norm": 0.480433851480484, "learning_rate": 7.16003303744639e-07, "loss": 0.0068, "step": 284630 }, { "epoch": 181.414913957935, "grad_norm": 0.1141522079706192, "learning_rate": 7.150742042756565e-07, "loss": 0.0078, "step": 284640 }, { "epoch": 181.421287444232, "grad_norm": 0.12121639400720596, "learning_rate": 7.14145703665059e-07, "loss": 0.0114, "step": 284650 }, { "epoch": 181.427660930529, "grad_norm": 0.5552188754081726, "learning_rate": 7.132178019241209e-07, "loss": 0.0049, "step": 284660 }, { "epoch": 181.434034416826, "grad_norm": 0.09602152556180954, "learning_rate": 7.122904990641277e-07, "loss": 0.0046, "step": 284670 }, { "epoch": 181.440407903123, "grad_norm": 0.36732879281044006, "learning_rate": 7.113637950963314e-07, "loss": 0.0024, "step": 284680 }, { "epoch": 181.44678138942, "grad_norm": 0.3551980257034302, "learning_rate": 7.104376900320064e-07, "loss": 0.0055, "step": 284690 }, { "epoch": 181.45315487571702, "grad_norm": 0.08077206462621689, "learning_rate": 7.095121838824048e-07, "loss": 0.0082, "step": 284700 }, { "epoch": 181.45952836201403, "grad_norm": 0.6833856105804443, "learning_rate": 7.08587276658762e-07, "loss": 0.0063, "step": 284710 }, { "epoch": 181.46590184831103, "grad_norm": 0.1102493479847908, "learning_rate": 7.076629683723191e-07, "loss": 0.0047, "step": 284720 }, { "epoch": 181.47227533460804, "grad_norm": 0.29043102264404297, "learning_rate": 7.06739259034317e-07, "loss": 0.0047, "step": 284730 }, { "epoch": 181.47864882090502, "grad_norm": 0.1269208937883377, "learning_rate": 7.058161486559744e-07, "loss": 0.0043, "step": 284740 }, { "epoch": 181.48502230720203, "grad_norm": 0.07093789428472519, "learning_rate": 7.048936372485049e-07, "loss": 0.0057, "step": 284750 }, { "epoch": 181.49139579349904, "grad_norm": 0.14779305458068848, "learning_rate": 7.039717248231159e-07, "loss": 0.0036, "step": 284760 }, { "epoch": 181.49776927979605, "grad_norm": 0.42781808972358704, "learning_rate": 7.030504113910208e-07, "loss": 0.0034, "step": 284770 }, { "epoch": 181.50414276609305, "grad_norm": 0.1433442085981369, "learning_rate": 7.021296969634051e-07, "loss": 0.0056, "step": 284780 }, { "epoch": 181.51051625239006, "grad_norm": 0.0908217504620552, "learning_rate": 7.012095815514541e-07, "loss": 0.004, "step": 284790 }, { "epoch": 181.51688973868707, "grad_norm": 0.2210044115781784, "learning_rate": 7.002900651663535e-07, "loss": 0.0063, "step": 284800 }, { "epoch": 181.52326322498408, "grad_norm": 0.6275880336761475, "learning_rate": 6.993711478192833e-07, "loss": 0.0088, "step": 284810 }, { "epoch": 181.52963671128106, "grad_norm": 0.1683920919895172, "learning_rate": 6.984528295213899e-07, "loss": 0.0032, "step": 284820 }, { "epoch": 181.53601019757807, "grad_norm": 0.317792147397995, "learning_rate": 6.975351102838534e-07, "loss": 0.0067, "step": 284830 }, { "epoch": 181.54238368387507, "grad_norm": 0.13096566498279572, "learning_rate": 6.966179901178038e-07, "loss": 0.0045, "step": 284840 }, { "epoch": 181.54875717017208, "grad_norm": 0.11571305245161057, "learning_rate": 6.957014690344099e-07, "loss": 0.0041, "step": 284850 }, { "epoch": 181.5551306564691, "grad_norm": 0.10242888331413269, "learning_rate": 6.947855470447795e-07, "loss": 0.0037, "step": 284860 }, { "epoch": 181.5615041427661, "grad_norm": 0.072715163230896, "learning_rate": 6.938702241600648e-07, "loss": 0.0059, "step": 284870 }, { "epoch": 181.5678776290631, "grad_norm": 0.09375341981649399, "learning_rate": 6.92955500391379e-07, "loss": 0.0042, "step": 284880 }, { "epoch": 181.5742511153601, "grad_norm": 0.1847035437822342, "learning_rate": 6.920413757498357e-07, "loss": 0.0029, "step": 284890 }, { "epoch": 181.5806246016571, "grad_norm": 0.36118248105049133, "learning_rate": 6.911278502465423e-07, "loss": 0.0053, "step": 284900 }, { "epoch": 181.5869980879541, "grad_norm": 0.09706626832485199, "learning_rate": 6.902149238926014e-07, "loss": 0.0038, "step": 284910 }, { "epoch": 181.5933715742511, "grad_norm": 0.20630891621112823, "learning_rate": 6.893025966991096e-07, "loss": 0.0072, "step": 284920 }, { "epoch": 181.59974506054812, "grad_norm": 0.06795644760131836, "learning_rate": 6.883908686771468e-07, "loss": 0.007, "step": 284930 }, { "epoch": 181.60611854684512, "grad_norm": 0.07098829001188278, "learning_rate": 6.874797398377874e-07, "loss": 0.006, "step": 284940 }, { "epoch": 181.61249203314213, "grad_norm": 0.060112740844488144, "learning_rate": 6.865692101921118e-07, "loss": 0.0094, "step": 284950 }, { "epoch": 181.61886551943914, "grad_norm": 0.23165912926197052, "learning_rate": 6.85659279751183e-07, "loss": 0.0053, "step": 284960 }, { "epoch": 181.62523900573615, "grad_norm": 0.05435090884566307, "learning_rate": 6.847499485260533e-07, "loss": 0.0087, "step": 284970 }, { "epoch": 181.63161249203316, "grad_norm": 0.09703074395656586, "learning_rate": 6.838412165277641e-07, "loss": 0.003, "step": 284980 }, { "epoch": 181.63798597833014, "grad_norm": 0.14175784587860107, "learning_rate": 6.829330837673786e-07, "loss": 0.0065, "step": 284990 }, { "epoch": 181.64435946462714, "grad_norm": 0.18883325159549713, "learning_rate": 6.820255502559103e-07, "loss": 0.006, "step": 285000 }, { "epoch": 181.65073295092415, "grad_norm": 0.07028353959321976, "learning_rate": 6.811186160044003e-07, "loss": 0.0042, "step": 285010 }, { "epoch": 181.65710643722116, "grad_norm": 0.10823452472686768, "learning_rate": 6.80212281023862e-07, "loss": 0.0068, "step": 285020 }, { "epoch": 181.66347992351817, "grad_norm": 0.09742851555347443, "learning_rate": 6.793065453253145e-07, "loss": 0.0041, "step": 285030 }, { "epoch": 181.66985340981518, "grad_norm": 0.08939331024885178, "learning_rate": 6.784014089197544e-07, "loss": 0.0046, "step": 285040 }, { "epoch": 181.67622689611218, "grad_norm": 0.19768621027469635, "learning_rate": 6.77496871818184e-07, "loss": 0.0059, "step": 285050 }, { "epoch": 181.6826003824092, "grad_norm": 0.04410964995622635, "learning_rate": 6.765929340315946e-07, "loss": 0.0035, "step": 285060 }, { "epoch": 181.68897386870617, "grad_norm": 0.05919486656785011, "learning_rate": 6.756895955709774e-07, "loss": 0.0046, "step": 285070 }, { "epoch": 181.69534735500318, "grad_norm": 0.11023827642202377, "learning_rate": 6.747868564472903e-07, "loss": 0.0058, "step": 285080 }, { "epoch": 181.7017208413002, "grad_norm": 0.08848599344491959, "learning_rate": 6.73884716671519e-07, "loss": 0.0046, "step": 285090 }, { "epoch": 181.7080943275972, "grad_norm": 0.18287792801856995, "learning_rate": 6.729831762546213e-07, "loss": 0.0023, "step": 285100 }, { "epoch": 181.7144678138942, "grad_norm": 0.0848299115896225, "learning_rate": 6.720822352075496e-07, "loss": 0.0044, "step": 285110 }, { "epoch": 181.7208413001912, "grad_norm": 0.39753830432891846, "learning_rate": 6.711818935412507e-07, "loss": 0.0084, "step": 285120 }, { "epoch": 181.72721478648822, "grad_norm": 0.25329527258872986, "learning_rate": 6.702821512666713e-07, "loss": 0.0094, "step": 285130 }, { "epoch": 181.73358827278523, "grad_norm": 0.12948699295520782, "learning_rate": 6.693830083947306e-07, "loss": 0.0039, "step": 285140 }, { "epoch": 181.7399617590822, "grad_norm": 0.1031964048743248, "learning_rate": 6.684844649363642e-07, "loss": 0.0033, "step": 285150 }, { "epoch": 181.7463352453792, "grad_norm": 0.334703654050827, "learning_rate": 6.675865209024856e-07, "loss": 0.0106, "step": 285160 }, { "epoch": 181.75270873167622, "grad_norm": 0.2015695720911026, "learning_rate": 6.666891763040195e-07, "loss": 0.0037, "step": 285170 }, { "epoch": 181.75908221797323, "grad_norm": 0.3145545721054077, "learning_rate": 6.65792431151846e-07, "loss": 0.0052, "step": 285180 }, { "epoch": 181.76545570427024, "grad_norm": 0.18825393915176392, "learning_rate": 6.648962854568785e-07, "loss": 0.0031, "step": 285190 }, { "epoch": 181.77182919056725, "grad_norm": 0.3418183922767639, "learning_rate": 6.64000739230003e-07, "loss": 0.0043, "step": 285200 }, { "epoch": 181.77820267686425, "grad_norm": 0.524255096912384, "learning_rate": 6.631057924820994e-07, "loss": 0.0074, "step": 285210 }, { "epoch": 181.78457616316126, "grad_norm": 0.596217930316925, "learning_rate": 6.622114452240369e-07, "loss": 0.0043, "step": 285220 }, { "epoch": 181.79094964945824, "grad_norm": 0.24858249723911285, "learning_rate": 6.613176974666957e-07, "loss": 0.0062, "step": 285230 }, { "epoch": 181.79732313575525, "grad_norm": 0.11980405449867249, "learning_rate": 6.604245492209227e-07, "loss": 0.0039, "step": 285240 }, { "epoch": 181.80369662205226, "grad_norm": 0.23569448292255402, "learning_rate": 6.595320004975814e-07, "loss": 0.005, "step": 285250 }, { "epoch": 181.81007010834927, "grad_norm": 0.3243686556816101, "learning_rate": 6.586400513075019e-07, "loss": 0.0096, "step": 285260 }, { "epoch": 181.81644359464627, "grad_norm": 0.3738023042678833, "learning_rate": 6.577487016615369e-07, "loss": 0.0081, "step": 285270 }, { "epoch": 181.82281708094328, "grad_norm": 0.10441657900810242, "learning_rate": 6.568579515705108e-07, "loss": 0.0062, "step": 285280 }, { "epoch": 181.8291905672403, "grad_norm": 0.16112655401229858, "learning_rate": 6.559678010452541e-07, "loss": 0.0027, "step": 285290 }, { "epoch": 181.8355640535373, "grad_norm": 0.17569918930530548, "learning_rate": 6.550782500965691e-07, "loss": 0.0042, "step": 285300 }, { "epoch": 181.84193753983428, "grad_norm": 0.2088160216808319, "learning_rate": 6.54189298735275e-07, "loss": 0.0076, "step": 285310 }, { "epoch": 181.84831102613128, "grad_norm": 0.07225137948989868, "learning_rate": 6.53300946972174e-07, "loss": 0.0076, "step": 285320 }, { "epoch": 181.8546845124283, "grad_norm": 0.09014533460140228, "learning_rate": 6.524131948180579e-07, "loss": 0.0055, "step": 285330 }, { "epoch": 181.8610579987253, "grad_norm": 0.10902392119169235, "learning_rate": 6.515260422837066e-07, "loss": 0.0049, "step": 285340 }, { "epoch": 181.8674314850223, "grad_norm": 0.1757165938615799, "learning_rate": 6.506394893799117e-07, "loss": 0.0048, "step": 285350 }, { "epoch": 181.87380497131932, "grad_norm": 0.2751997411251068, "learning_rate": 6.497535361174367e-07, "loss": 0.0046, "step": 285360 }, { "epoch": 181.88017845761632, "grad_norm": 0.12532134354114532, "learning_rate": 6.488681825070564e-07, "loss": 0.0026, "step": 285370 }, { "epoch": 181.88655194391333, "grad_norm": 0.11113253235816956, "learning_rate": 6.479834285595232e-07, "loss": 0.0053, "step": 285380 }, { "epoch": 181.8929254302103, "grad_norm": 0.2410665899515152, "learning_rate": 6.470992742855786e-07, "loss": 0.0081, "step": 285390 }, { "epoch": 181.89929891650732, "grad_norm": 0.12697866559028625, "learning_rate": 6.462157196959806e-07, "loss": 0.0082, "step": 285400 }, { "epoch": 181.90567240280433, "grad_norm": 0.2346000373363495, "learning_rate": 6.453327648014595e-07, "loss": 0.0037, "step": 285410 }, { "epoch": 181.91204588910134, "grad_norm": 0.2358279824256897, "learning_rate": 6.444504096127457e-07, "loss": 0.0036, "step": 285420 }, { "epoch": 181.91841937539834, "grad_norm": 0.19714738428592682, "learning_rate": 6.435686541405528e-07, "loss": 0.0066, "step": 285430 }, { "epoch": 181.92479286169535, "grad_norm": 0.40226638317108154, "learning_rate": 6.426874983956055e-07, "loss": 0.0112, "step": 285440 }, { "epoch": 181.93116634799236, "grad_norm": 0.1965879499912262, "learning_rate": 6.418069423886064e-07, "loss": 0.0053, "step": 285450 }, { "epoch": 181.93753983428937, "grad_norm": 0.1856701821088791, "learning_rate": 6.409269861302525e-07, "loss": 0.0025, "step": 285460 }, { "epoch": 181.94391332058635, "grad_norm": 0.44870424270629883, "learning_rate": 6.400476296312408e-07, "loss": 0.003, "step": 285470 }, { "epoch": 181.95028680688335, "grad_norm": 0.15933750569820404, "learning_rate": 6.391688729022516e-07, "loss": 0.0065, "step": 285480 }, { "epoch": 181.95666029318036, "grad_norm": 0.06054285541176796, "learning_rate": 6.382907159539653e-07, "loss": 0.005, "step": 285490 }, { "epoch": 181.96303377947737, "grad_norm": 0.3498568534851074, "learning_rate": 6.374131587970566e-07, "loss": 0.0048, "step": 285500 }, { "epoch": 181.96940726577438, "grad_norm": 0.11112350225448608, "learning_rate": 6.365362014421728e-07, "loss": 0.0035, "step": 285510 }, { "epoch": 181.9757807520714, "grad_norm": 0.18361306190490723, "learning_rate": 6.356598438999939e-07, "loss": 0.0114, "step": 285520 }, { "epoch": 181.9821542383684, "grad_norm": 0.6440020203590393, "learning_rate": 6.347840861811505e-07, "loss": 0.0078, "step": 285530 }, { "epoch": 181.9885277246654, "grad_norm": 0.25295957922935486, "learning_rate": 6.339089282962895e-07, "loss": 0.0059, "step": 285540 }, { "epoch": 181.99490121096238, "grad_norm": 0.28969642519950867, "learning_rate": 6.33034370256036e-07, "loss": 0.0055, "step": 285550 }, { "epoch": 182.0012746972594, "grad_norm": 0.31911736726760864, "learning_rate": 6.321604120710367e-07, "loss": 0.003, "step": 285560 }, { "epoch": 182.0076481835564, "grad_norm": 0.3081730008125305, "learning_rate": 6.312870537518945e-07, "loss": 0.0032, "step": 285570 }, { "epoch": 182.0140216698534, "grad_norm": 0.05464997515082359, "learning_rate": 6.304142953092284e-07, "loss": 0.0037, "step": 285580 }, { "epoch": 182.0203951561504, "grad_norm": 0.05679759010672569, "learning_rate": 6.295421367536358e-07, "loss": 0.0092, "step": 285590 }, { "epoch": 182.02676864244742, "grad_norm": 0.30354568362236023, "learning_rate": 6.286705780957303e-07, "loss": 0.0051, "step": 285600 }, { "epoch": 182.03314212874443, "grad_norm": 0.2712688148021698, "learning_rate": 6.277996193460811e-07, "loss": 0.0054, "step": 285610 }, { "epoch": 182.03951561504144, "grad_norm": 0.09028390049934387, "learning_rate": 6.269292605152854e-07, "loss": 0.0039, "step": 285620 }, { "epoch": 182.04588910133845, "grad_norm": 0.22192902863025665, "learning_rate": 6.260595016139126e-07, "loss": 0.0055, "step": 285630 }, { "epoch": 182.05226258763543, "grad_norm": 0.6023141741752625, "learning_rate": 6.251903426525374e-07, "loss": 0.0063, "step": 285640 }, { "epoch": 182.05863607393243, "grad_norm": 0.1627834439277649, "learning_rate": 6.243217836417126e-07, "loss": 0.0019, "step": 285650 }, { "epoch": 182.06500956022944, "grad_norm": 0.18293185532093048, "learning_rate": 6.234538245920019e-07, "loss": 0.0043, "step": 285660 }, { "epoch": 182.07138304652645, "grad_norm": 0.06533025950193405, "learning_rate": 6.225864655139469e-07, "loss": 0.003, "step": 285670 }, { "epoch": 182.07775653282346, "grad_norm": 0.14663268625736237, "learning_rate": 6.217197064180835e-07, "loss": 0.0133, "step": 285680 }, { "epoch": 182.08413001912047, "grad_norm": 0.2593053877353668, "learning_rate": 6.208535473149424e-07, "loss": 0.0149, "step": 285690 }, { "epoch": 182.09050350541747, "grad_norm": 0.11353737860918045, "learning_rate": 6.199879882150595e-07, "loss": 0.0056, "step": 285700 }, { "epoch": 182.09687699171448, "grad_norm": 0.22123058140277863, "learning_rate": 6.191230291289374e-07, "loss": 0.0071, "step": 285710 }, { "epoch": 182.10325047801146, "grad_norm": 0.09207047522068024, "learning_rate": 6.182586700671012e-07, "loss": 0.0031, "step": 285720 }, { "epoch": 182.10962396430847, "grad_norm": 0.0846748799085617, "learning_rate": 6.173949110400423e-07, "loss": 0.0025, "step": 285730 }, { "epoch": 182.11599745060548, "grad_norm": 0.31268784403800964, "learning_rate": 6.165317520582636e-07, "loss": 0.0042, "step": 285740 }, { "epoch": 182.12237093690248, "grad_norm": 0.2163887321949005, "learning_rate": 6.156691931322456e-07, "loss": 0.0046, "step": 285750 }, { "epoch": 182.1287444231995, "grad_norm": 0.2594603896141052, "learning_rate": 6.148072342724742e-07, "loss": 0.0148, "step": 285760 }, { "epoch": 182.1351179094965, "grad_norm": 0.24828343093395233, "learning_rate": 6.139458754894245e-07, "loss": 0.0035, "step": 285770 }, { "epoch": 182.1414913957935, "grad_norm": 0.18198786675930023, "learning_rate": 6.130851167935603e-07, "loss": 0.0078, "step": 285780 }, { "epoch": 182.14786488209052, "grad_norm": 0.17073999345302582, "learning_rate": 6.122249581953398e-07, "loss": 0.0048, "step": 285790 }, { "epoch": 182.1542383683875, "grad_norm": 0.1906946897506714, "learning_rate": 6.11365399705216e-07, "loss": 0.0048, "step": 285800 }, { "epoch": 182.1606118546845, "grad_norm": 0.47806933522224426, "learning_rate": 6.105064413336304e-07, "loss": 0.0069, "step": 285810 }, { "epoch": 182.1669853409815, "grad_norm": 0.5740478038787842, "learning_rate": 6.096480830910301e-07, "loss": 0.006, "step": 285820 }, { "epoch": 182.17335882727852, "grad_norm": 0.12941613793373108, "learning_rate": 6.08790324987829e-07, "loss": 0.003, "step": 285830 }, { "epoch": 182.17973231357553, "grad_norm": 0.1934204399585724, "learning_rate": 6.07933167034469e-07, "loss": 0.0066, "step": 285840 }, { "epoch": 182.18610579987254, "grad_norm": 0.11532354354858398, "learning_rate": 6.070766092413415e-07, "loss": 0.0029, "step": 285850 }, { "epoch": 182.19247928616954, "grad_norm": 0.08493057638406754, "learning_rate": 6.062206516188773e-07, "loss": 0.0057, "step": 285860 }, { "epoch": 182.19885277246655, "grad_norm": 0.2970784902572632, "learning_rate": 6.053652941774624e-07, "loss": 0.0078, "step": 285870 }, { "epoch": 182.20522625876353, "grad_norm": 0.09885190427303314, "learning_rate": 6.04510536927505e-07, "loss": 0.0061, "step": 285880 }, { "epoch": 182.21159974506054, "grad_norm": 0.03998027741909027, "learning_rate": 6.036563798793693e-07, "loss": 0.0023, "step": 285890 }, { "epoch": 182.21797323135755, "grad_norm": 0.1920248568058014, "learning_rate": 6.028028230434579e-07, "loss": 0.0043, "step": 285900 }, { "epoch": 182.22434671765456, "grad_norm": 0.16426847875118256, "learning_rate": 6.019498664301182e-07, "loss": 0.0089, "step": 285910 }, { "epoch": 182.23072020395156, "grad_norm": 0.1900763064622879, "learning_rate": 6.010975100497418e-07, "loss": 0.0026, "step": 285920 }, { "epoch": 182.23709369024857, "grad_norm": 0.17028427124023438, "learning_rate": 6.002457539126594e-07, "loss": 0.0031, "step": 285930 }, { "epoch": 182.24346717654558, "grad_norm": 0.10899844765663147, "learning_rate": 5.993945980292348e-07, "loss": 0.0044, "step": 285940 }, { "epoch": 182.2498406628426, "grad_norm": 0.1130935326218605, "learning_rate": 5.985440424098098e-07, "loss": 0.0042, "step": 285950 }, { "epoch": 182.25621414913957, "grad_norm": 0.160164937376976, "learning_rate": 5.976940870647208e-07, "loss": 0.0082, "step": 285960 }, { "epoch": 182.26258763543657, "grad_norm": 0.31590721011161804, "learning_rate": 5.968447320042813e-07, "loss": 0.0042, "step": 285970 }, { "epoch": 182.26896112173358, "grad_norm": 0.33922579884529114, "learning_rate": 5.959959772388334e-07, "loss": 0.0036, "step": 285980 }, { "epoch": 182.2753346080306, "grad_norm": 0.2744465470314026, "learning_rate": 5.951478227786744e-07, "loss": 0.0036, "step": 285990 }, { "epoch": 182.2817080943276, "grad_norm": 0.3157724440097809, "learning_rate": 5.943002686341182e-07, "loss": 0.008, "step": 286000 }, { "epoch": 182.2880815806246, "grad_norm": 0.1198098435997963, "learning_rate": 5.934533148154564e-07, "loss": 0.0072, "step": 286010 }, { "epoch": 182.2944550669216, "grad_norm": 0.3370143473148346, "learning_rate": 5.926069613329865e-07, "loss": 0.0074, "step": 286020 }, { "epoch": 182.30082855321862, "grad_norm": 0.07725319266319275, "learning_rate": 5.917612081969893e-07, "loss": 0.0022, "step": 286030 }, { "epoch": 182.3072020395156, "grad_norm": 0.11859340965747833, "learning_rate": 5.90916055417745e-07, "loss": 0.0037, "step": 286040 }, { "epoch": 182.3135755258126, "grad_norm": 0.10533207654953003, "learning_rate": 5.900715030055181e-07, "loss": 0.0045, "step": 286050 }, { "epoch": 182.31994901210962, "grad_norm": 0.14384938776493073, "learning_rate": 5.892275509705725e-07, "loss": 0.0083, "step": 286060 }, { "epoch": 182.32632249840663, "grad_norm": 0.29695925116539, "learning_rate": 5.883841993231665e-07, "loss": 0.0045, "step": 286070 }, { "epoch": 182.33269598470363, "grad_norm": 0.5339444279670715, "learning_rate": 5.875414480735475e-07, "loss": 0.0059, "step": 286080 }, { "epoch": 182.33906947100064, "grad_norm": 0.331181138753891, "learning_rate": 5.866992972319462e-07, "loss": 0.0095, "step": 286090 }, { "epoch": 182.34544295729765, "grad_norm": 0.4812847971916199, "learning_rate": 5.858577468086047e-07, "loss": 0.0041, "step": 286100 }, { "epoch": 182.35181644359466, "grad_norm": 0.06256017833948135, "learning_rate": 5.850167968137477e-07, "loss": 0.0051, "step": 286110 }, { "epoch": 182.35818992989164, "grad_norm": 0.06311715394258499, "learning_rate": 5.841764472575895e-07, "loss": 0.0069, "step": 286120 }, { "epoch": 182.36456341618864, "grad_norm": 0.11964637041091919, "learning_rate": 5.833366981503496e-07, "loss": 0.0021, "step": 286130 }, { "epoch": 182.37093690248565, "grad_norm": 0.07343468815088272, "learning_rate": 5.824975495022144e-07, "loss": 0.0051, "step": 286140 }, { "epoch": 182.37731038878266, "grad_norm": 0.13966967165470123, "learning_rate": 5.816590013234035e-07, "loss": 0.0044, "step": 286150 }, { "epoch": 182.38368387507967, "grad_norm": 0.2086213231086731, "learning_rate": 5.808210536240921e-07, "loss": 0.0045, "step": 286160 }, { "epoch": 182.39005736137668, "grad_norm": 0.19424930214881897, "learning_rate": 5.799837064144609e-07, "loss": 0.005, "step": 286170 }, { "epoch": 182.39643084767368, "grad_norm": 0.23497064411640167, "learning_rate": 5.79146959704685e-07, "loss": 0.0037, "step": 286180 }, { "epoch": 182.4028043339707, "grad_norm": 0.2484838217496872, "learning_rate": 5.783108135049397e-07, "loss": 0.0036, "step": 286190 }, { "epoch": 182.4091778202677, "grad_norm": 0.17587019503116608, "learning_rate": 5.774752678253781e-07, "loss": 0.0043, "step": 286200 }, { "epoch": 182.41555130656468, "grad_norm": 0.07411734759807587, "learning_rate": 5.766403226761585e-07, "loss": 0.0058, "step": 286210 }, { "epoch": 182.4219247928617, "grad_norm": 0.0946289598941803, "learning_rate": 5.758059780674174e-07, "loss": 0.0068, "step": 286220 }, { "epoch": 182.4282982791587, "grad_norm": 0.0832279771566391, "learning_rate": 5.749722340092967e-07, "loss": 0.0038, "step": 286230 }, { "epoch": 182.4346717654557, "grad_norm": 0.08566837757825851, "learning_rate": 5.741390905119326e-07, "loss": 0.0045, "step": 286240 }, { "epoch": 182.4410452517527, "grad_norm": 0.15808293223381042, "learning_rate": 5.733065475854449e-07, "loss": 0.0033, "step": 286250 }, { "epoch": 182.44741873804972, "grad_norm": 0.1939474195241928, "learning_rate": 5.724746052399476e-07, "loss": 0.0073, "step": 286260 }, { "epoch": 182.45379222434673, "grad_norm": 0.32454320788383484, "learning_rate": 5.71643263485555e-07, "loss": 0.0049, "step": 286270 }, { "epoch": 182.46016571064374, "grad_norm": 0.12170456349849701, "learning_rate": 5.708125223323535e-07, "loss": 0.0055, "step": 286280 }, { "epoch": 182.46653919694072, "grad_norm": 0.29743391275405884, "learning_rate": 5.699823817904571e-07, "loss": 0.0064, "step": 286290 }, { "epoch": 182.47291268323772, "grad_norm": 0.15489386022090912, "learning_rate": 5.69152841869941e-07, "loss": 0.0052, "step": 286300 }, { "epoch": 182.47928616953473, "grad_norm": 0.19168105721473694, "learning_rate": 5.683239025808973e-07, "loss": 0.0051, "step": 286310 }, { "epoch": 182.48565965583174, "grad_norm": 0.11913325637578964, "learning_rate": 5.67495563933379e-07, "loss": 0.0043, "step": 286320 }, { "epoch": 182.49203314212875, "grad_norm": 0.10748308897018433, "learning_rate": 5.666678259374669e-07, "loss": 0.0024, "step": 286330 }, { "epoch": 182.49840662842576, "grad_norm": 0.14634555578231812, "learning_rate": 5.658406886032141e-07, "loss": 0.0036, "step": 286340 }, { "epoch": 182.50478011472276, "grad_norm": 0.0897197499871254, "learning_rate": 5.650141519406682e-07, "loss": 0.0092, "step": 286350 }, { "epoch": 182.51115360101977, "grad_norm": 0.10432253777980804, "learning_rate": 5.64188215959871e-07, "loss": 0.0056, "step": 286360 }, { "epoch": 182.51752708731675, "grad_norm": 0.20340417325496674, "learning_rate": 5.633628806708646e-07, "loss": 0.0039, "step": 286370 }, { "epoch": 182.52390057361376, "grad_norm": 0.11213963478803635, "learning_rate": 5.625381460836799e-07, "loss": 0.0029, "step": 286380 }, { "epoch": 182.53027405991077, "grad_norm": 0.2167142778635025, "learning_rate": 5.617140122083309e-07, "loss": 0.0032, "step": 286390 }, { "epoch": 182.53664754620777, "grad_norm": 0.1242891475558281, "learning_rate": 5.608904790548319e-07, "loss": 0.0029, "step": 286400 }, { "epoch": 182.54302103250478, "grad_norm": 0.1602160632610321, "learning_rate": 5.600675466331973e-07, "loss": 0.0025, "step": 286410 }, { "epoch": 182.5493945188018, "grad_norm": 0.39208802580833435, "learning_rate": 5.592452149534133e-07, "loss": 0.0094, "step": 286420 }, { "epoch": 182.5557680050988, "grad_norm": 0.1328674554824829, "learning_rate": 5.584234840254832e-07, "loss": 0.0055, "step": 286430 }, { "epoch": 182.5621414913958, "grad_norm": 0.19881759583950043, "learning_rate": 5.576023538593878e-07, "loss": 0.0034, "step": 286440 }, { "epoch": 182.5685149776928, "grad_norm": 0.21700060367584229, "learning_rate": 5.56781824465108e-07, "loss": 0.0053, "step": 286450 }, { "epoch": 182.5748884639898, "grad_norm": 0.09297648072242737, "learning_rate": 5.55961895852608e-07, "loss": 0.0079, "step": 286460 }, { "epoch": 182.5812619502868, "grad_norm": 0.3872019648551941, "learning_rate": 5.551425680318523e-07, "loss": 0.0085, "step": 286470 }, { "epoch": 182.5876354365838, "grad_norm": 0.1469523012638092, "learning_rate": 5.543238410127993e-07, "loss": 0.0056, "step": 286480 }, { "epoch": 182.59400892288082, "grad_norm": 0.06906117498874664, "learning_rate": 5.535057148054024e-07, "loss": 0.0135, "step": 286490 }, { "epoch": 182.60038240917783, "grad_norm": 0.08550548553466797, "learning_rate": 5.526881894195868e-07, "loss": 0.0056, "step": 286500 }, { "epoch": 182.60675589547483, "grad_norm": 0.1315353661775589, "learning_rate": 5.518712648653002e-07, "loss": 0.0037, "step": 286510 }, { "epoch": 182.61312938177184, "grad_norm": 0.2788107693195343, "learning_rate": 5.510549411524679e-07, "loss": 0.0032, "step": 286520 }, { "epoch": 182.61950286806882, "grad_norm": 0.06590504199266434, "learning_rate": 5.502392182910043e-07, "loss": 0.0024, "step": 286530 }, { "epoch": 182.62587635436583, "grad_norm": 0.055550482124090195, "learning_rate": 5.494240962908181e-07, "loss": 0.0026, "step": 286540 }, { "epoch": 182.63224984066284, "grad_norm": 0.06930024176836014, "learning_rate": 5.486095751618181e-07, "loss": 0.0094, "step": 286550 }, { "epoch": 182.63862332695984, "grad_norm": 0.09155640006065369, "learning_rate": 5.477956549139073e-07, "loss": 0.0054, "step": 286560 }, { "epoch": 182.64499681325685, "grad_norm": 0.12426075339317322, "learning_rate": 5.46982335556967e-07, "loss": 0.012, "step": 286570 }, { "epoch": 182.65137029955386, "grad_norm": 0.12620167434215546, "learning_rate": 5.461696171008779e-07, "loss": 0.003, "step": 286580 }, { "epoch": 182.65774378585087, "grad_norm": 0.09339947253465652, "learning_rate": 5.453574995555267e-07, "loss": 0.0045, "step": 286590 }, { "epoch": 182.66411727214788, "grad_norm": 0.35090938210487366, "learning_rate": 5.44545982930772e-07, "loss": 0.0025, "step": 286600 }, { "epoch": 182.67049075844486, "grad_norm": 0.16223260760307312, "learning_rate": 5.437350672364783e-07, "loss": 0.0036, "step": 286610 }, { "epoch": 182.67686424474186, "grad_norm": 0.2794647514820099, "learning_rate": 5.429247524824988e-07, "loss": 0.0051, "step": 286620 }, { "epoch": 182.68323773103887, "grad_norm": 0.20917052030563354, "learning_rate": 5.421150386786811e-07, "loss": 0.0094, "step": 286630 }, { "epoch": 182.68961121733588, "grad_norm": 0.6320255994796753, "learning_rate": 5.413059258348563e-07, "loss": 0.0065, "step": 286640 }, { "epoch": 182.6959847036329, "grad_norm": 0.15353935956954956, "learning_rate": 5.404974139608609e-07, "loss": 0.0032, "step": 286650 }, { "epoch": 182.7023581899299, "grad_norm": 0.08098075538873672, "learning_rate": 5.396895030665206e-07, "loss": 0.004, "step": 286660 }, { "epoch": 182.7087316762269, "grad_norm": 0.20220310986042023, "learning_rate": 5.38882193161655e-07, "loss": 0.0062, "step": 286670 }, { "epoch": 182.7151051625239, "grad_norm": 0.07227865606546402, "learning_rate": 5.380754842560621e-07, "loss": 0.0065, "step": 286680 }, { "epoch": 182.7214786488209, "grad_norm": 0.2524735927581787, "learning_rate": 5.372693763595616e-07, "loss": 0.0034, "step": 286690 }, { "epoch": 182.7278521351179, "grad_norm": 0.21386775374412537, "learning_rate": 5.364638694819346e-07, "loss": 0.0045, "step": 286700 }, { "epoch": 182.7342256214149, "grad_norm": 0.05783062055706978, "learning_rate": 5.356589636329734e-07, "loss": 0.0032, "step": 286710 }, { "epoch": 182.74059910771192, "grad_norm": 0.4846315383911133, "learning_rate": 5.348546588224534e-07, "loss": 0.0041, "step": 286720 }, { "epoch": 182.74697259400892, "grad_norm": 0.2520621120929718, "learning_rate": 5.340509550601558e-07, "loss": 0.0042, "step": 286730 }, { "epoch": 182.75334608030593, "grad_norm": 0.07457812130451202, "learning_rate": 5.332478523558449e-07, "loss": 0.0043, "step": 286740 }, { "epoch": 182.75971956660294, "grad_norm": 0.2733611762523651, "learning_rate": 5.32445350719274e-07, "loss": 0.0109, "step": 286750 }, { "epoch": 182.76609305289995, "grad_norm": 0.11455442011356354, "learning_rate": 5.316434501601907e-07, "loss": 0.0041, "step": 286760 }, { "epoch": 182.77246653919693, "grad_norm": 0.42293182015419006, "learning_rate": 5.308421506883543e-07, "loss": 0.0051, "step": 286770 }, { "epoch": 182.77884002549393, "grad_norm": 0.06686533987522125, "learning_rate": 5.3004145231349e-07, "loss": 0.0106, "step": 286780 }, { "epoch": 182.78521351179094, "grad_norm": 0.19365857541561127, "learning_rate": 5.292413550453345e-07, "loss": 0.0029, "step": 286790 }, { "epoch": 182.79158699808795, "grad_norm": 0.2193979024887085, "learning_rate": 5.284418588935969e-07, "loss": 0.0035, "step": 286800 }, { "epoch": 182.79796048438496, "grad_norm": 0.07553121447563171, "learning_rate": 5.276429638680024e-07, "loss": 0.004, "step": 286810 }, { "epoch": 182.80433397068197, "grad_norm": 0.3948060870170593, "learning_rate": 5.268446699782548e-07, "loss": 0.0049, "step": 286820 }, { "epoch": 182.81070745697897, "grad_norm": 0.10596098005771637, "learning_rate": 5.260469772340571e-07, "loss": 0.0171, "step": 286830 }, { "epoch": 182.81708094327598, "grad_norm": 0.2047543227672577, "learning_rate": 5.252498856451015e-07, "loss": 0.0041, "step": 286840 }, { "epoch": 182.823454429573, "grad_norm": 0.3193299174308777, "learning_rate": 5.24453395221064e-07, "loss": 0.0061, "step": 286850 }, { "epoch": 182.82982791586997, "grad_norm": 0.2547261118888855, "learning_rate": 5.236575059716364e-07, "loss": 0.0041, "step": 286860 }, { "epoch": 182.83620140216698, "grad_norm": 0.1982000321149826, "learning_rate": 5.228622179064835e-07, "loss": 0.0068, "step": 286870 }, { "epoch": 182.842574888464, "grad_norm": 0.07274234294891357, "learning_rate": 5.220675310352696e-07, "loss": 0.0021, "step": 286880 }, { "epoch": 182.848948374761, "grad_norm": 0.1181144192814827, "learning_rate": 5.212734453676427e-07, "loss": 0.0038, "step": 286890 }, { "epoch": 182.855321861058, "grad_norm": 0.12025543302297592, "learning_rate": 5.204799609132671e-07, "loss": 0.0049, "step": 286900 }, { "epoch": 182.861695347355, "grad_norm": 0.13788433372974396, "learning_rate": 5.196870776817742e-07, "loss": 0.0063, "step": 286910 }, { "epoch": 182.86806883365202, "grad_norm": 0.062396105378866196, "learning_rate": 5.188947956828006e-07, "loss": 0.0082, "step": 286920 }, { "epoch": 182.87444231994903, "grad_norm": 0.12385780364274979, "learning_rate": 5.181031149259719e-07, "loss": 0.004, "step": 286930 }, { "epoch": 182.880815806246, "grad_norm": 0.2583842873573303, "learning_rate": 5.173120354209082e-07, "loss": 0.0032, "step": 286940 }, { "epoch": 182.887189292543, "grad_norm": 0.2518474757671356, "learning_rate": 5.165215571772242e-07, "loss": 0.0039, "step": 286950 }, { "epoch": 182.89356277884002, "grad_norm": 0.12912799417972565, "learning_rate": 5.157316802045231e-07, "loss": 0.0064, "step": 286960 }, { "epoch": 182.89993626513703, "grad_norm": 0.21076087653636932, "learning_rate": 5.14942404512403e-07, "loss": 0.0137, "step": 286970 }, { "epoch": 182.90630975143404, "grad_norm": 0.05115387588739395, "learning_rate": 5.14153730110456e-07, "loss": 0.0049, "step": 286980 }, { "epoch": 182.91268323773104, "grad_norm": 0.056075818836688995, "learning_rate": 5.133656570082634e-07, "loss": 0.0053, "step": 286990 }, { "epoch": 182.91905672402805, "grad_norm": 0.2901186943054199, "learning_rate": 5.125781852153954e-07, "loss": 0.006, "step": 287000 }, { "epoch": 182.92543021032506, "grad_norm": 0.27426877617836, "learning_rate": 5.117913147414333e-07, "loss": 0.0071, "step": 287010 }, { "epoch": 182.93180369662204, "grad_norm": 0.15838827192783356, "learning_rate": 5.110050455959303e-07, "loss": 0.0083, "step": 287020 }, { "epoch": 182.93817718291905, "grad_norm": 0.11427615582942963, "learning_rate": 5.102193777884401e-07, "loss": 0.0041, "step": 287030 }, { "epoch": 182.94455066921606, "grad_norm": 0.08265263587236404, "learning_rate": 5.094343113285105e-07, "loss": 0.0025, "step": 287040 }, { "epoch": 182.95092415551306, "grad_norm": 0.1035715788602829, "learning_rate": 5.086498462256783e-07, "loss": 0.0067, "step": 287050 }, { "epoch": 182.95729764181007, "grad_norm": 0.19879783689975739, "learning_rate": 5.078659824894804e-07, "loss": 0.0016, "step": 287060 }, { "epoch": 182.96367112810708, "grad_norm": 0.15469376742839813, "learning_rate": 5.07082720129437e-07, "loss": 0.0029, "step": 287070 }, { "epoch": 182.9700446144041, "grad_norm": 0.14914141595363617, "learning_rate": 5.063000591550738e-07, "loss": 0.0051, "step": 287080 }, { "epoch": 182.9764181007011, "grad_norm": 0.20058876276016235, "learning_rate": 5.055179995758885e-07, "loss": 0.005, "step": 287090 }, { "epoch": 182.98279158699808, "grad_norm": 0.0733494684100151, "learning_rate": 5.04736541401396e-07, "loss": 0.0052, "step": 287100 }, { "epoch": 182.98916507329508, "grad_norm": 0.10971666872501373, "learning_rate": 5.039556846410776e-07, "loss": 0.0046, "step": 287110 }, { "epoch": 182.9955385595921, "grad_norm": 0.048507511615753174, "learning_rate": 5.031754293044366e-07, "loss": 0.0044, "step": 287120 }, { "epoch": 183.0019120458891, "grad_norm": 0.19494035840034485, "learning_rate": 5.023957754009434e-07, "loss": 0.0028, "step": 287130 }, { "epoch": 183.0082855321861, "grad_norm": 0.330787718296051, "learning_rate": 5.016167229400737e-07, "loss": 0.0059, "step": 287140 }, { "epoch": 183.01465901848312, "grad_norm": 0.29703083634376526, "learning_rate": 5.008382719312976e-07, "loss": 0.0041, "step": 287150 }, { "epoch": 183.02103250478012, "grad_norm": 0.13840417563915253, "learning_rate": 5.000604223840744e-07, "loss": 0.0031, "step": 287160 }, { "epoch": 183.02740599107713, "grad_norm": 0.06553860753774643, "learning_rate": 4.992831743078464e-07, "loss": 0.005, "step": 287170 }, { "epoch": 183.0337794773741, "grad_norm": 0.31960606575012207, "learning_rate": 4.985065277120671e-07, "loss": 0.0048, "step": 287180 }, { "epoch": 183.04015296367112, "grad_norm": 0.1373695582151413, "learning_rate": 4.97730482606168e-07, "loss": 0.0061, "step": 287190 }, { "epoch": 183.04652644996813, "grad_norm": 0.18637147545814514, "learning_rate": 4.969550389995914e-07, "loss": 0.0054, "step": 287200 }, { "epoch": 183.05289993626513, "grad_norm": 0.075704425573349, "learning_rate": 4.96180196901741e-07, "loss": 0.0027, "step": 287210 }, { "epoch": 183.05927342256214, "grad_norm": 0.05931151658296585, "learning_rate": 4.954059563220425e-07, "loss": 0.0044, "step": 287220 }, { "epoch": 183.06564690885915, "grad_norm": 0.2039949893951416, "learning_rate": 4.946323172698997e-07, "loss": 0.0035, "step": 287230 }, { "epoch": 183.07202039515616, "grad_norm": 0.18285998702049255, "learning_rate": 4.938592797547159e-07, "loss": 0.0079, "step": 287240 }, { "epoch": 183.07839388145317, "grad_norm": 0.11882677674293518, "learning_rate": 4.930868437858837e-07, "loss": 0.0079, "step": 287250 }, { "epoch": 183.08476736775015, "grad_norm": 0.06617877632379532, "learning_rate": 4.923150093727902e-07, "loss": 0.0079, "step": 287260 }, { "epoch": 183.09114085404715, "grad_norm": 0.11042705178260803, "learning_rate": 4.91543776524811e-07, "loss": 0.0024, "step": 287270 }, { "epoch": 183.09751434034416, "grad_norm": 0.6137973070144653, "learning_rate": 4.90773145251322e-07, "loss": 0.0068, "step": 287280 }, { "epoch": 183.10388782664117, "grad_norm": 0.21940509974956512, "learning_rate": 4.900031155616769e-07, "loss": 0.0049, "step": 287290 }, { "epoch": 183.11026131293818, "grad_norm": 0.10059979557991028, "learning_rate": 4.892336874652459e-07, "loss": 0.0024, "step": 287300 }, { "epoch": 183.1166347992352, "grad_norm": 0.17015470564365387, "learning_rate": 4.88464860971366e-07, "loss": 0.0048, "step": 287310 }, { "epoch": 183.1230082855322, "grad_norm": 0.07596730440855026, "learning_rate": 4.876966360893909e-07, "loss": 0.0042, "step": 287320 }, { "epoch": 183.1293817718292, "grad_norm": 0.13587266206741333, "learning_rate": 4.869290128286407e-07, "loss": 0.0047, "step": 287330 }, { "epoch": 183.13575525812618, "grad_norm": 0.18964192271232605, "learning_rate": 4.861619911984583e-07, "loss": 0.0134, "step": 287340 }, { "epoch": 183.1421287444232, "grad_norm": 0.047697778791189194, "learning_rate": 4.853955712081526e-07, "loss": 0.0097, "step": 287350 }, { "epoch": 183.1485022307202, "grad_norm": 0.6617637872695923, "learning_rate": 4.84629752867044e-07, "loss": 0.0153, "step": 287360 }, { "epoch": 183.1548757170172, "grad_norm": 0.08141399174928665, "learning_rate": 4.838645361844252e-07, "loss": 0.0025, "step": 287370 }, { "epoch": 183.1612492033142, "grad_norm": 0.5516194701194763, "learning_rate": 4.830999211696163e-07, "loss": 0.0088, "step": 287380 }, { "epoch": 183.16762268961122, "grad_norm": 0.11743229627609253, "learning_rate": 4.823359078318878e-07, "loss": 0.0056, "step": 287390 }, { "epoch": 183.17399617590823, "grad_norm": 0.11133517324924469, "learning_rate": 4.815724961805323e-07, "loss": 0.0031, "step": 287400 }, { "epoch": 183.18036966220524, "grad_norm": 0.35277462005615234, "learning_rate": 4.808096862248201e-07, "loss": 0.0091, "step": 287410 }, { "epoch": 183.18674314850222, "grad_norm": 0.16787497699260712, "learning_rate": 4.800474779740271e-07, "loss": 0.004, "step": 287420 }, { "epoch": 183.19311663479922, "grad_norm": 0.061073821038007736, "learning_rate": 4.792858714374127e-07, "loss": 0.0053, "step": 287430 }, { "epoch": 183.19949012109623, "grad_norm": 0.05595304071903229, "learning_rate": 4.785248666242304e-07, "loss": 0.0099, "step": 287440 }, { "epoch": 183.20586360739324, "grad_norm": 0.042885083705186844, "learning_rate": 4.777644635437229e-07, "loss": 0.0023, "step": 287450 }, { "epoch": 183.21223709369025, "grad_norm": 0.05050312355160713, "learning_rate": 4.770046622051383e-07, "loss": 0.0096, "step": 287460 }, { "epoch": 183.21861057998726, "grad_norm": 0.13165102899074554, "learning_rate": 4.7624546261770264e-07, "loss": 0.0029, "step": 287470 }, { "epoch": 183.22498406628426, "grad_norm": 0.07005344331264496, "learning_rate": 4.7548686479064187e-07, "loss": 0.0035, "step": 287480 }, { "epoch": 183.23135755258127, "grad_norm": 0.17022530734539032, "learning_rate": 4.7472886873317635e-07, "loss": 0.004, "step": 287490 }, { "epoch": 183.23773103887828, "grad_norm": 0.29874539375305176, "learning_rate": 4.739714744545154e-07, "loss": 0.0104, "step": 287500 }, { "epoch": 183.24410452517526, "grad_norm": 0.16734035313129425, "learning_rate": 4.732146819638572e-07, "loss": 0.0099, "step": 287510 }, { "epoch": 183.25047801147227, "grad_norm": 0.07086575031280518, "learning_rate": 4.7245849127040554e-07, "loss": 0.0029, "step": 287520 }, { "epoch": 183.25685149776928, "grad_norm": 0.11699583381414413, "learning_rate": 4.717029023833475e-07, "loss": 0.004, "step": 287530 }, { "epoch": 183.26322498406628, "grad_norm": 0.1934352070093155, "learning_rate": 4.70947915311859e-07, "loss": 0.0026, "step": 287540 }, { "epoch": 183.2695984703633, "grad_norm": 0.11695881932973862, "learning_rate": 4.7019353006511613e-07, "loss": 0.0028, "step": 287550 }, { "epoch": 183.2759719566603, "grad_norm": 0.1292366236448288, "learning_rate": 4.6943974665228375e-07, "loss": 0.0077, "step": 287560 }, { "epoch": 183.2823454429573, "grad_norm": 0.24092069268226624, "learning_rate": 4.6868656508252116e-07, "loss": 0.0041, "step": 287570 }, { "epoch": 183.28871892925432, "grad_norm": 0.06768403947353363, "learning_rate": 4.679339853649878e-07, "loss": 0.0078, "step": 287580 }, { "epoch": 183.2950924155513, "grad_norm": 0.37637653946876526, "learning_rate": 4.671820075088207e-07, "loss": 0.0071, "step": 287590 }, { "epoch": 183.3014659018483, "grad_norm": 0.4262920320034027, "learning_rate": 4.66430631523157e-07, "loss": 0.0046, "step": 287600 }, { "epoch": 183.3078393881453, "grad_norm": 0.12871065735816956, "learning_rate": 4.656798574171284e-07, "loss": 0.0038, "step": 287610 }, { "epoch": 183.31421287444232, "grad_norm": 0.15357224643230438, "learning_rate": 4.649296851998608e-07, "loss": 0.0056, "step": 287620 }, { "epoch": 183.32058636073933, "grad_norm": 0.41050660610198975, "learning_rate": 4.641801148804692e-07, "loss": 0.0102, "step": 287630 }, { "epoch": 183.32695984703633, "grad_norm": 0.19880850613117218, "learning_rate": 4.6343114646805187e-07, "loss": 0.0104, "step": 287640 }, { "epoch": 183.33333333333334, "grad_norm": 0.164036825299263, "learning_rate": 4.6268277997171814e-07, "loss": 0.0048, "step": 287650 }, { "epoch": 183.33970681963035, "grad_norm": 0.3448396325111389, "learning_rate": 4.619350154005608e-07, "loss": 0.0073, "step": 287660 }, { "epoch": 183.34608030592733, "grad_norm": 0.16359589993953705, "learning_rate": 4.6118785276366707e-07, "loss": 0.0049, "step": 287670 }, { "epoch": 183.35245379222434, "grad_norm": 0.1686665564775467, "learning_rate": 4.604412920701073e-07, "loss": 0.0033, "step": 287680 }, { "epoch": 183.35882727852135, "grad_norm": 0.4334869980812073, "learning_rate": 4.5969533332896333e-07, "loss": 0.005, "step": 287690 }, { "epoch": 183.36520076481835, "grad_norm": 0.22472313046455383, "learning_rate": 4.589499765492944e-07, "loss": 0.0045, "step": 287700 }, { "epoch": 183.37157425111536, "grad_norm": 0.09778814017772675, "learning_rate": 4.5820522174016e-07, "loss": 0.005, "step": 287710 }, { "epoch": 183.37794773741237, "grad_norm": 0.14297905564308167, "learning_rate": 4.5746106891059735e-07, "loss": 0.0027, "step": 287720 }, { "epoch": 183.38432122370938, "grad_norm": 0.12500977516174316, "learning_rate": 4.5671751806967145e-07, "loss": 0.0038, "step": 287730 }, { "epoch": 183.3906947100064, "grad_norm": 0.09174348413944244, "learning_rate": 4.5597456922639727e-07, "loss": 0.0089, "step": 287740 }, { "epoch": 183.39706819630337, "grad_norm": 0.07157566398382187, "learning_rate": 4.55232222389812e-07, "loss": 0.0043, "step": 287750 }, { "epoch": 183.40344168260037, "grad_norm": 0.33189693093299866, "learning_rate": 4.544904775689307e-07, "loss": 0.0137, "step": 287760 }, { "epoch": 183.40981516889738, "grad_norm": 0.10700064152479172, "learning_rate": 4.537493347727739e-07, "loss": 0.0032, "step": 287770 }, { "epoch": 183.4161886551944, "grad_norm": 0.09993835538625717, "learning_rate": 4.5300879401033446e-07, "loss": 0.0057, "step": 287780 }, { "epoch": 183.4225621414914, "grad_norm": 0.17196179926395416, "learning_rate": 4.522688552906218e-07, "loss": 0.0072, "step": 287790 }, { "epoch": 183.4289356277884, "grad_norm": 0.2541394829750061, "learning_rate": 4.515295186226232e-07, "loss": 0.0149, "step": 287800 }, { "epoch": 183.4353091140854, "grad_norm": 0.11364297568798065, "learning_rate": 4.507907840153258e-07, "loss": 0.002, "step": 287810 }, { "epoch": 183.44168260038242, "grad_norm": 0.09769927710294724, "learning_rate": 4.5005265147769484e-07, "loss": 0.0064, "step": 287820 }, { "epoch": 183.4480560866794, "grad_norm": 0.1316215991973877, "learning_rate": 4.493151210187174e-07, "loss": 0.0064, "step": 287830 }, { "epoch": 183.4544295729764, "grad_norm": 0.0894237756729126, "learning_rate": 4.485781926473365e-07, "loss": 0.0044, "step": 287840 }, { "epoch": 183.46080305927342, "grad_norm": 0.19791921973228455, "learning_rate": 4.4784186637252256e-07, "loss": 0.0077, "step": 287850 }, { "epoch": 183.46717654557042, "grad_norm": 0.15549039840698242, "learning_rate": 4.4710614220320746e-07, "loss": 0.0126, "step": 287860 }, { "epoch": 183.47355003186743, "grad_norm": 0.09619654715061188, "learning_rate": 4.463710201483451e-07, "loss": 0.0067, "step": 287870 }, { "epoch": 183.47992351816444, "grad_norm": 0.08741132915019989, "learning_rate": 4.4563650021685057e-07, "loss": 0.0031, "step": 287880 }, { "epoch": 183.48629700446145, "grad_norm": 0.36212754249572754, "learning_rate": 4.4490258241766667e-07, "loss": 0.008, "step": 287890 }, { "epoch": 183.49267049075846, "grad_norm": 0.2107536941766739, "learning_rate": 4.4416926675970303e-07, "loss": 0.0047, "step": 287900 }, { "epoch": 183.49904397705544, "grad_norm": 0.8226764798164368, "learning_rate": 4.434365532518747e-07, "loss": 0.01, "step": 287910 }, { "epoch": 183.50541746335244, "grad_norm": 0.10358156263828278, "learning_rate": 4.4270444190308014e-07, "loss": 0.0036, "step": 287920 }, { "epoch": 183.51179094964945, "grad_norm": 0.10768315196037292, "learning_rate": 4.419729327222122e-07, "loss": 0.0072, "step": 287930 }, { "epoch": 183.51816443594646, "grad_norm": 0.13726195693016052, "learning_rate": 4.412420257181693e-07, "loss": 0.0025, "step": 287940 }, { "epoch": 183.52453792224347, "grad_norm": 0.25523972511291504, "learning_rate": 4.4051172089982217e-07, "loss": 0.0077, "step": 287950 }, { "epoch": 183.53091140854048, "grad_norm": 0.20039640367031097, "learning_rate": 4.397820182760526e-07, "loss": 0.0051, "step": 287960 }, { "epoch": 183.53728489483748, "grad_norm": 0.3969191312789917, "learning_rate": 4.3905291785572567e-07, "loss": 0.0054, "step": 287970 }, { "epoch": 183.5436583811345, "grad_norm": 0.37133288383483887, "learning_rate": 4.3832441964769546e-07, "loss": 0.0055, "step": 287980 }, { "epoch": 183.55003186743147, "grad_norm": 0.16088901460170746, "learning_rate": 4.375965236608159e-07, "loss": 0.0045, "step": 287990 }, { "epoch": 183.55640535372848, "grad_norm": 0.44373440742492676, "learning_rate": 4.3686922990393564e-07, "loss": 0.0043, "step": 288000 }, { "epoch": 183.5627788400255, "grad_norm": 0.07898295670747757, "learning_rate": 4.3614253838589194e-07, "loss": 0.0197, "step": 288010 }, { "epoch": 183.5691523263225, "grad_norm": 0.11206202954053879, "learning_rate": 4.3541644911551105e-07, "loss": 0.0091, "step": 288020 }, { "epoch": 183.5755258126195, "grad_norm": 0.04656016826629639, "learning_rate": 4.346909621016193e-07, "loss": 0.0056, "step": 288030 }, { "epoch": 183.5818992989165, "grad_norm": 0.1940499097108841, "learning_rate": 4.3396607735302075e-07, "loss": 0.0109, "step": 288040 }, { "epoch": 183.58827278521352, "grad_norm": 0.10056433826684952, "learning_rate": 4.3324179487854165e-07, "loss": 0.0076, "step": 288050 }, { "epoch": 183.59464627151053, "grad_norm": 0.10925811529159546, "learning_rate": 4.3251811468696945e-07, "loss": 0.0046, "step": 288060 }, { "epoch": 183.60101975780753, "grad_norm": 0.4568643569946289, "learning_rate": 4.3179503678710265e-07, "loss": 0.0095, "step": 288070 }, { "epoch": 183.60739324410451, "grad_norm": 0.10878607630729675, "learning_rate": 4.310725611877231e-07, "loss": 0.0045, "step": 288080 }, { "epoch": 183.61376673040152, "grad_norm": 0.11507976800203323, "learning_rate": 4.303506878976238e-07, "loss": 0.002, "step": 288090 }, { "epoch": 183.62014021669853, "grad_norm": 0.1276884377002716, "learning_rate": 4.2962941692555323e-07, "loss": 0.0045, "step": 288100 }, { "epoch": 183.62651370299554, "grad_norm": 0.6080954074859619, "learning_rate": 4.2890874828029337e-07, "loss": 0.0054, "step": 288110 }, { "epoch": 183.63288718929255, "grad_norm": 0.2109002321958542, "learning_rate": 4.281886819705927e-07, "loss": 0.0025, "step": 288120 }, { "epoch": 183.63926067558955, "grad_norm": 0.13122348487377167, "learning_rate": 4.2746921800520536e-07, "loss": 0.0037, "step": 288130 }, { "epoch": 183.64563416188656, "grad_norm": 0.12189992517232895, "learning_rate": 4.267503563928632e-07, "loss": 0.0032, "step": 288140 }, { "epoch": 183.65200764818357, "grad_norm": 0.422730028629303, "learning_rate": 4.260320971423204e-07, "loss": 0.0069, "step": 288150 }, { "epoch": 183.65838113448055, "grad_norm": 0.08225367963314056, "learning_rate": 4.253144402622866e-07, "loss": 0.01, "step": 288160 }, { "epoch": 183.66475462077756, "grad_norm": 0.525806188583374, "learning_rate": 4.245973857614882e-07, "loss": 0.007, "step": 288170 }, { "epoch": 183.67112810707457, "grad_norm": 0.0826062336564064, "learning_rate": 4.2388093364864047e-07, "loss": 0.0045, "step": 288180 }, { "epoch": 183.67750159337157, "grad_norm": 0.19470445811748505, "learning_rate": 4.231650839324475e-07, "loss": 0.0072, "step": 288190 }, { "epoch": 183.68387507966858, "grad_norm": 0.061726104468107224, "learning_rate": 4.2244983662160785e-07, "loss": 0.0025, "step": 288200 }, { "epoch": 183.6902485659656, "grad_norm": 0.25673776865005493, "learning_rate": 4.2173519172480915e-07, "loss": 0.0061, "step": 288210 }, { "epoch": 183.6966220522626, "grad_norm": 0.059695493429899216, "learning_rate": 4.210211492507388e-07, "loss": 0.0037, "step": 288220 }, { "epoch": 183.7029955385596, "grad_norm": 0.24205482006072998, "learning_rate": 4.2030770920807315e-07, "loss": 0.007, "step": 288230 }, { "epoch": 183.70936902485658, "grad_norm": 0.18837976455688477, "learning_rate": 4.1959487160547763e-07, "loss": 0.0099, "step": 288240 }, { "epoch": 183.7157425111536, "grad_norm": 0.6058289408683777, "learning_rate": 4.188826364516174e-07, "loss": 0.0084, "step": 288250 }, { "epoch": 183.7221159974506, "grad_norm": 0.05866431072354317, "learning_rate": 4.181710037551412e-07, "loss": 0.0037, "step": 288260 }, { "epoch": 183.7284894837476, "grad_norm": 0.4570545554161072, "learning_rate": 4.174599735247031e-07, "loss": 0.0056, "step": 288270 }, { "epoch": 183.73486297004462, "grad_norm": 0.3548975884914398, "learning_rate": 4.167495457689463e-07, "loss": 0.0064, "step": 288280 }, { "epoch": 183.74123645634162, "grad_norm": 0.11041148006916046, "learning_rate": 4.1603972049649165e-07, "loss": 0.004, "step": 288290 }, { "epoch": 183.74760994263863, "grad_norm": 0.28031107783317566, "learning_rate": 4.1533049771596556e-07, "loss": 0.0033, "step": 288300 }, { "epoch": 183.75398342893564, "grad_norm": 0.3529419004917145, "learning_rate": 4.1462187743599446e-07, "loss": 0.006, "step": 288310 }, { "epoch": 183.76035691523262, "grad_norm": 0.14740808308124542, "learning_rate": 4.139138596651826e-07, "loss": 0.0058, "step": 288320 }, { "epoch": 183.76673040152963, "grad_norm": 0.36207932233810425, "learning_rate": 4.132064444121342e-07, "loss": 0.0052, "step": 288330 }, { "epoch": 183.77310388782664, "grad_norm": 0.13025546073913574, "learning_rate": 4.124996316854479e-07, "loss": 0.0031, "step": 288340 }, { "epoch": 183.77947737412364, "grad_norm": 0.15625500679016113, "learning_rate": 4.117934214937058e-07, "loss": 0.0046, "step": 288350 }, { "epoch": 183.78585086042065, "grad_norm": 0.056162722408771515, "learning_rate": 4.1108781384549544e-07, "loss": 0.0046, "step": 288360 }, { "epoch": 183.79222434671766, "grad_norm": 0.10858796536922455, "learning_rate": 4.103828087493822e-07, "loss": 0.007, "step": 288370 }, { "epoch": 183.79859783301467, "grad_norm": 0.2128080129623413, "learning_rate": 4.096784062139425e-07, "loss": 0.0071, "step": 288380 }, { "epoch": 183.80497131931168, "grad_norm": 0.06728693097829819, "learning_rate": 4.0897460624773064e-07, "loss": 0.0032, "step": 288390 }, { "epoch": 183.81134480560866, "grad_norm": 0.15886534750461578, "learning_rate": 4.082714088593009e-07, "loss": 0.0078, "step": 288400 }, { "epoch": 183.81771829190566, "grad_norm": 0.10389498621225357, "learning_rate": 4.075688140571965e-07, "loss": 0.0051, "step": 288410 }, { "epoch": 183.82409177820267, "grad_norm": 0.06793231517076492, "learning_rate": 4.068668218499494e-07, "loss": 0.0087, "step": 288420 }, { "epoch": 183.83046526449968, "grad_norm": 0.09273064136505127, "learning_rate": 4.0616543224609726e-07, "loss": 0.0023, "step": 288430 }, { "epoch": 183.8368387507967, "grad_norm": 0.4495030641555786, "learning_rate": 4.0546464525416107e-07, "loss": 0.0108, "step": 288440 }, { "epoch": 183.8432122370937, "grad_norm": 0.08112737536430359, "learning_rate": 4.0476446088265064e-07, "loss": 0.0027, "step": 288450 }, { "epoch": 183.8495857233907, "grad_norm": 0.07267000526189804, "learning_rate": 4.0406487914007585e-07, "loss": 0.0035, "step": 288460 }, { "epoch": 183.8559592096877, "grad_norm": 0.6540042161941528, "learning_rate": 4.03365900034941e-07, "loss": 0.0068, "step": 288470 }, { "epoch": 183.8623326959847, "grad_norm": 0.10806455463171005, "learning_rate": 4.0266752357574487e-07, "loss": 0.0073, "step": 288480 }, { "epoch": 183.8687061822817, "grad_norm": 0.13843056559562683, "learning_rate": 4.0196974977095846e-07, "loss": 0.011, "step": 288490 }, { "epoch": 183.8750796685787, "grad_norm": 0.07165723294019699, "learning_rate": 4.0127257862906943e-07, "loss": 0.0049, "step": 288500 }, { "epoch": 183.88145315487571, "grad_norm": 0.22803887724876404, "learning_rate": 4.005760101585487e-07, "loss": 0.0052, "step": 288510 }, { "epoch": 183.88782664117272, "grad_norm": 0.10218014568090439, "learning_rate": 3.998800443678563e-07, "loss": 0.003, "step": 288520 }, { "epoch": 183.89420012746973, "grad_norm": 0.06958536803722382, "learning_rate": 3.9918468126545207e-07, "loss": 0.0038, "step": 288530 }, { "epoch": 183.90057361376674, "grad_norm": 0.30735963582992554, "learning_rate": 3.9848992085978475e-07, "loss": 0.0041, "step": 288540 }, { "epoch": 183.90694710006375, "grad_norm": 0.059504684060811996, "learning_rate": 3.977957631592977e-07, "loss": 0.0034, "step": 288550 }, { "epoch": 183.91332058636073, "grad_norm": 0.1646568477153778, "learning_rate": 3.971022081724285e-07, "loss": 0.003, "step": 288560 }, { "epoch": 183.91969407265773, "grad_norm": 0.08739833533763885, "learning_rate": 3.964092559075927e-07, "loss": 0.0174, "step": 288570 }, { "epoch": 183.92606755895474, "grad_norm": 0.1067264974117279, "learning_rate": 3.9571690637322246e-07, "loss": 0.0043, "step": 288580 }, { "epoch": 183.93244104525175, "grad_norm": 0.12659607827663422, "learning_rate": 3.950251595777277e-07, "loss": 0.0039, "step": 288590 }, { "epoch": 183.93881453154876, "grad_norm": 0.08292935788631439, "learning_rate": 3.943340155295072e-07, "loss": 0.0047, "step": 288600 }, { "epoch": 183.94518801784577, "grad_norm": 0.23183473944664001, "learning_rate": 3.9364347423696543e-07, "loss": 0.0063, "step": 288610 }, { "epoch": 183.95156150414277, "grad_norm": 0.08023490756750107, "learning_rate": 3.9295353570849567e-07, "loss": 0.0037, "step": 288620 }, { "epoch": 183.95793499043978, "grad_norm": 0.15911181271076202, "learning_rate": 3.92264199952469e-07, "loss": 0.0042, "step": 288630 }, { "epoch": 183.96430847673676, "grad_norm": 0.192198246717453, "learning_rate": 3.9157546697727864e-07, "loss": 0.0047, "step": 288640 }, { "epoch": 183.97068196303377, "grad_norm": 0.1472061574459076, "learning_rate": 3.9088733679127907e-07, "loss": 0.0054, "step": 288650 }, { "epoch": 183.97705544933078, "grad_norm": 0.27484235167503357, "learning_rate": 3.901998094028414e-07, "loss": 0.0173, "step": 288660 }, { "epoch": 183.98342893562778, "grad_norm": 0.09546966850757599, "learning_rate": 3.895128848203089e-07, "loss": 0.0075, "step": 288670 }, { "epoch": 183.9898024219248, "grad_norm": 0.07788197696208954, "learning_rate": 3.888265630520416e-07, "loss": 0.0046, "step": 288680 }, { "epoch": 183.9961759082218, "grad_norm": 0.10810893028974533, "learning_rate": 3.881408441063661e-07, "loss": 0.0035, "step": 288690 }, { "epoch": 184.0025493945188, "grad_norm": 0.16508083045482635, "learning_rate": 3.874557279916258e-07, "loss": 0.004, "step": 288700 }, { "epoch": 184.00892288081582, "grad_norm": 0.6294320821762085, "learning_rate": 3.867712147161362e-07, "loss": 0.008, "step": 288710 }, { "epoch": 184.01529636711282, "grad_norm": 0.328871488571167, "learning_rate": 3.8608730428822403e-07, "loss": 0.0085, "step": 288720 }, { "epoch": 184.0216698534098, "grad_norm": 0.3897271454334259, "learning_rate": 3.8540399671619377e-07, "loss": 0.0057, "step": 288730 }, { "epoch": 184.0280433397068, "grad_norm": 0.1943090856075287, "learning_rate": 3.847212920083443e-07, "loss": 0.0021, "step": 288740 }, { "epoch": 184.03441682600382, "grad_norm": 0.07417171448469162, "learning_rate": 3.840391901729801e-07, "loss": 0.0092, "step": 288750 }, { "epoch": 184.04079031230083, "grad_norm": 0.5352672934532166, "learning_rate": 3.8335769121838336e-07, "loss": 0.008, "step": 288760 }, { "epoch": 184.04716379859784, "grad_norm": 0.3217476010322571, "learning_rate": 3.826767951528365e-07, "loss": 0.0039, "step": 288770 }, { "epoch": 184.05353728489484, "grad_norm": 0.2724103331565857, "learning_rate": 3.8199650198461055e-07, "loss": 0.0035, "step": 288780 }, { "epoch": 184.05991077119185, "grad_norm": 0.0823209136724472, "learning_rate": 3.8131681172197674e-07, "loss": 0.0026, "step": 288790 }, { "epoch": 184.06628425748886, "grad_norm": 0.5108993053436279, "learning_rate": 3.806377243731951e-07, "loss": 0.0062, "step": 288800 }, { "epoch": 184.07265774378584, "grad_norm": 0.22699199616909027, "learning_rate": 3.7995923994650904e-07, "loss": 0.0126, "step": 288810 }, { "epoch": 184.07903123008285, "grad_norm": 0.08870507031679153, "learning_rate": 3.7928135845016757e-07, "loss": 0.0093, "step": 288820 }, { "epoch": 184.08540471637986, "grad_norm": 0.2062404602766037, "learning_rate": 3.7860407989240844e-07, "loss": 0.004, "step": 288830 }, { "epoch": 184.09177820267686, "grad_norm": 0.06512412428855896, "learning_rate": 3.7792740428145846e-07, "loss": 0.0028, "step": 288840 }, { "epoch": 184.09815168897387, "grad_norm": 0.08110380172729492, "learning_rate": 3.7725133162553884e-07, "loss": 0.003, "step": 288850 }, { "epoch": 184.10452517527088, "grad_norm": 0.14377562701702118, "learning_rate": 3.7657586193287076e-07, "loss": 0.0062, "step": 288860 }, { "epoch": 184.1108986615679, "grad_norm": 0.1702236384153366, "learning_rate": 3.7590099521166436e-07, "loss": 0.0043, "step": 288870 }, { "epoch": 184.1172721478649, "grad_norm": 0.12389246374368668, "learning_rate": 3.752267314701075e-07, "loss": 0.0084, "step": 288880 }, { "epoch": 184.12364563416187, "grad_norm": 0.10323214530944824, "learning_rate": 3.7455307071639357e-07, "loss": 0.0036, "step": 288890 }, { "epoch": 184.13001912045888, "grad_norm": 0.09691846370697021, "learning_rate": 3.7388001295872165e-07, "loss": 0.0038, "step": 288900 }, { "epoch": 184.1363926067559, "grad_norm": 0.37433722615242004, "learning_rate": 3.73207558205263e-07, "loss": 0.0082, "step": 288910 }, { "epoch": 184.1427660930529, "grad_norm": 0.12456043064594269, "learning_rate": 3.725357064641888e-07, "loss": 0.0043, "step": 288920 }, { "epoch": 184.1491395793499, "grad_norm": 0.371355801820755, "learning_rate": 3.7186445774365365e-07, "loss": 0.0095, "step": 288930 }, { "epoch": 184.15551306564691, "grad_norm": 0.09400203078985214, "learning_rate": 3.7119381205182877e-07, "loss": 0.005, "step": 288940 }, { "epoch": 184.16188655194392, "grad_norm": 0.11261101812124252, "learning_rate": 3.705237693968577e-07, "loss": 0.0029, "step": 288950 }, { "epoch": 184.16826003824093, "grad_norm": 0.14365807175636292, "learning_rate": 3.698543297868784e-07, "loss": 0.0033, "step": 288960 }, { "epoch": 184.1746335245379, "grad_norm": 0.05602431669831276, "learning_rate": 3.691854932300232e-07, "loss": 0.0043, "step": 288970 }, { "epoch": 184.18100701083492, "grad_norm": 0.13621903955936432, "learning_rate": 3.6851725973442995e-07, "loss": 0.0058, "step": 288980 }, { "epoch": 184.18738049713193, "grad_norm": 0.06211492046713829, "learning_rate": 3.67849629308209e-07, "loss": 0.0041, "step": 288990 }, { "epoch": 184.19375398342893, "grad_norm": 0.09304916858673096, "learning_rate": 3.67182601959476e-07, "loss": 0.0045, "step": 289000 }, { "epoch": 184.20012746972594, "grad_norm": 0.058029163628816605, "learning_rate": 3.665161776963355e-07, "loss": 0.0028, "step": 289010 }, { "epoch": 184.20650095602295, "grad_norm": 0.08219657838344574, "learning_rate": 3.658503565268867e-07, "loss": 0.0051, "step": 289020 }, { "epoch": 184.21287444231996, "grad_norm": 0.2431332767009735, "learning_rate": 3.6518513845921753e-07, "loss": 0.005, "step": 289030 }, { "epoch": 184.21924792861697, "grad_norm": 0.07024046778678894, "learning_rate": 3.645205235014104e-07, "loss": 0.0051, "step": 289040 }, { "epoch": 184.22562141491395, "grad_norm": 0.1578882932662964, "learning_rate": 3.6385651166154223e-07, "loss": 0.0043, "step": 289050 }, { "epoch": 184.23199490121095, "grad_norm": 0.27455100417137146, "learning_rate": 3.6319310294768427e-07, "loss": 0.0084, "step": 289060 }, { "epoch": 184.23836838750796, "grad_norm": 0.10918472707271576, "learning_rate": 3.6253029736789677e-07, "loss": 0.0065, "step": 289070 }, { "epoch": 184.24474187380497, "grad_norm": 0.1349605768918991, "learning_rate": 3.6186809493022886e-07, "loss": 0.0035, "step": 289080 }, { "epoch": 184.25111536010198, "grad_norm": 0.7727872133255005, "learning_rate": 3.612064956427297e-07, "loss": 0.0062, "step": 289090 }, { "epoch": 184.25748884639899, "grad_norm": 0.09245073795318604, "learning_rate": 3.605454995134372e-07, "loss": 0.0066, "step": 289100 }, { "epoch": 184.263862332696, "grad_norm": 0.0729587972164154, "learning_rate": 3.5988510655038386e-07, "loss": 0.0039, "step": 289110 }, { "epoch": 184.270235818993, "grad_norm": 0.13627447187900543, "learning_rate": 3.592253167615967e-07, "loss": 0.0019, "step": 289120 }, { "epoch": 184.27660930528998, "grad_norm": 0.6004162430763245, "learning_rate": 3.585661301550913e-07, "loss": 0.0108, "step": 289130 }, { "epoch": 184.282982791587, "grad_norm": 0.2312997430562973, "learning_rate": 3.579075467388726e-07, "loss": 0.0026, "step": 289140 }, { "epoch": 184.289356277884, "grad_norm": 0.11355355381965637, "learning_rate": 3.572495665209563e-07, "loss": 0.0066, "step": 289150 }, { "epoch": 184.295729764181, "grad_norm": 0.10058372467756271, "learning_rate": 3.5659218950931937e-07, "loss": 0.0062, "step": 289160 }, { "epoch": 184.302103250478, "grad_norm": 0.20743867754936218, "learning_rate": 3.55935415711961e-07, "loss": 0.0036, "step": 289170 }, { "epoch": 184.30847673677502, "grad_norm": 0.052928756922483444, "learning_rate": 3.552792451368581e-07, "loss": 0.0041, "step": 289180 }, { "epoch": 184.31485022307203, "grad_norm": 0.10528891533613205, "learning_rate": 3.546236777919876e-07, "loss": 0.004, "step": 289190 }, { "epoch": 184.32122370936904, "grad_norm": 0.15189634263515472, "learning_rate": 3.539687136853098e-07, "loss": 0.0099, "step": 289200 }, { "epoch": 184.32759719566602, "grad_norm": 0.31898462772369385, "learning_rate": 3.533143528247851e-07, "loss": 0.0031, "step": 289210 }, { "epoch": 184.33397068196302, "grad_norm": 0.23696140944957733, "learning_rate": 3.5266059521836813e-07, "loss": 0.0075, "step": 289220 }, { "epoch": 184.34034416826003, "grad_norm": 0.1592373102903366, "learning_rate": 3.520074408740026e-07, "loss": 0.0036, "step": 289230 }, { "epoch": 184.34671765455704, "grad_norm": 0.07395996898412704, "learning_rate": 3.513548897996155e-07, "loss": 0.0047, "step": 289240 }, { "epoch": 184.35309114085405, "grad_norm": 0.0634901225566864, "learning_rate": 3.507029420031449e-07, "loss": 0.004, "step": 289250 }, { "epoch": 184.35946462715106, "grad_norm": 0.13075041770935059, "learning_rate": 3.5005159749251224e-07, "loss": 0.0026, "step": 289260 }, { "epoch": 184.36583811344806, "grad_norm": 0.22886337339878082, "learning_rate": 3.494008562756279e-07, "loss": 0.0069, "step": 289270 }, { "epoch": 184.37221159974507, "grad_norm": 0.49991872906684875, "learning_rate": 3.4875071836039664e-07, "loss": 0.0103, "step": 289280 }, { "epoch": 184.37858508604205, "grad_norm": 0.25665903091430664, "learning_rate": 3.4810118375472875e-07, "loss": 0.0041, "step": 289290 }, { "epoch": 184.38495857233906, "grad_norm": 0.20942577719688416, "learning_rate": 3.4745225246651247e-07, "loss": 0.0049, "step": 289300 }, { "epoch": 184.39133205863607, "grad_norm": 0.30691492557525635, "learning_rate": 3.468039245036303e-07, "loss": 0.0035, "step": 289310 }, { "epoch": 184.39770554493307, "grad_norm": 0.19646045565605164, "learning_rate": 3.4615619987395377e-07, "loss": 0.0029, "step": 289320 }, { "epoch": 184.40407903123008, "grad_norm": 0.09090633690357208, "learning_rate": 3.4550907858537094e-07, "loss": 0.0051, "step": 289330 }, { "epoch": 184.4104525175271, "grad_norm": 0.0813029482960701, "learning_rate": 3.448625606457256e-07, "loss": 0.0051, "step": 289340 }, { "epoch": 184.4168260038241, "grad_norm": 0.2801307141780853, "learning_rate": 3.4421664606288373e-07, "loss": 0.0063, "step": 289350 }, { "epoch": 184.4231994901211, "grad_norm": 0.20390504598617554, "learning_rate": 3.435713348446945e-07, "loss": 0.0031, "step": 289360 }, { "epoch": 184.42957297641811, "grad_norm": 0.17617247998714447, "learning_rate": 3.429266269989961e-07, "loss": 0.003, "step": 289370 }, { "epoch": 184.4359464627151, "grad_norm": 0.6958492398262024, "learning_rate": 3.422825225336268e-07, "loss": 0.008, "step": 289380 }, { "epoch": 184.4423199490121, "grad_norm": 0.3983565866947174, "learning_rate": 3.4163902145640245e-07, "loss": 0.0047, "step": 289390 }, { "epoch": 184.4486934353091, "grad_norm": 0.28066179156303406, "learning_rate": 3.4099612377515576e-07, "loss": 0.0083, "step": 289400 }, { "epoch": 184.45506692160612, "grad_norm": 0.12210903316736221, "learning_rate": 3.403538294976916e-07, "loss": 0.0036, "step": 289410 }, { "epoch": 184.46144040790313, "grad_norm": 0.34045860171318054, "learning_rate": 3.3971213863180917e-07, "loss": 0.0111, "step": 289420 }, { "epoch": 184.46781389420013, "grad_norm": 0.335490345954895, "learning_rate": 3.390710511853135e-07, "loss": 0.0068, "step": 289430 }, { "epoch": 184.47418738049714, "grad_norm": 0.07878798246383667, "learning_rate": 3.3843056716599817e-07, "loss": 0.0024, "step": 289440 }, { "epoch": 184.48056086679415, "grad_norm": 0.4445968270301819, "learning_rate": 3.3779068658163493e-07, "loss": 0.0103, "step": 289450 }, { "epoch": 184.48693435309113, "grad_norm": 0.22293296456336975, "learning_rate": 3.371514094400008e-07, "loss": 0.0039, "step": 289460 }, { "epoch": 184.49330783938814, "grad_norm": 0.1040337085723877, "learning_rate": 3.365127357488729e-07, "loss": 0.004, "step": 289470 }, { "epoch": 184.49968132568515, "grad_norm": 0.1501828283071518, "learning_rate": 3.3587466551600054e-07, "loss": 0.007, "step": 289480 }, { "epoch": 184.50605481198215, "grad_norm": 0.5350468754768372, "learning_rate": 3.352371987491443e-07, "loss": 0.0082, "step": 289490 }, { "epoch": 184.51242829827916, "grad_norm": 0.1786755919456482, "learning_rate": 3.346003354560479e-07, "loss": 0.0111, "step": 289500 }, { "epoch": 184.51880178457617, "grad_norm": 0.33376824855804443, "learning_rate": 3.339640756444495e-07, "loss": 0.0056, "step": 289510 }, { "epoch": 184.52517527087318, "grad_norm": 0.20320026576519012, "learning_rate": 3.3332841932207646e-07, "loss": 0.0022, "step": 289520 }, { "epoch": 184.53154875717019, "grad_norm": 0.1296105980873108, "learning_rate": 3.326933664966614e-07, "loss": 0.0023, "step": 289530 }, { "epoch": 184.53792224346716, "grad_norm": 0.4669922888278961, "learning_rate": 3.320589171759092e-07, "loss": 0.0097, "step": 289540 }, { "epoch": 184.54429572976417, "grad_norm": 0.09033814817667007, "learning_rate": 3.3142507136754153e-07, "loss": 0.0026, "step": 289550 }, { "epoch": 184.55066921606118, "grad_norm": 0.06948623806238174, "learning_rate": 3.307918290792467e-07, "loss": 0.015, "step": 289560 }, { "epoch": 184.5570427023582, "grad_norm": 0.11041051149368286, "learning_rate": 3.301591903187351e-07, "loss": 0.0023, "step": 289570 }, { "epoch": 184.5634161886552, "grad_norm": 0.16174502670764923, "learning_rate": 3.2952715509368403e-07, "loss": 0.0052, "step": 289580 }, { "epoch": 184.5697896749522, "grad_norm": 0.06599589437246323, "learning_rate": 3.2889572341177065e-07, "loss": 0.0041, "step": 289590 }, { "epoch": 184.5761631612492, "grad_norm": 0.2643928527832031, "learning_rate": 3.282648952806722e-07, "loss": 0.0048, "step": 289600 }, { "epoch": 184.58253664754622, "grad_norm": 0.2668929100036621, "learning_rate": 3.2763467070805467e-07, "loss": 0.0034, "step": 289610 }, { "epoch": 184.5889101338432, "grad_norm": 0.0444699190557003, "learning_rate": 3.270050497015731e-07, "loss": 0.0087, "step": 289620 }, { "epoch": 184.5952836201402, "grad_norm": 0.0853947177529335, "learning_rate": 3.2637603226888245e-07, "loss": 0.0026, "step": 289630 }, { "epoch": 184.60165710643722, "grad_norm": 0.09721074998378754, "learning_rate": 3.2574761841761557e-07, "loss": 0.004, "step": 289640 }, { "epoch": 184.60803059273422, "grad_norm": 0.09490261971950531, "learning_rate": 3.2511980815542184e-07, "loss": 0.0046, "step": 289650 }, { "epoch": 184.61440407903123, "grad_norm": 0.27700841426849365, "learning_rate": 3.24492601489923e-07, "loss": 0.0061, "step": 289660 }, { "epoch": 184.62077756532824, "grad_norm": 0.08081678301095963, "learning_rate": 3.2386599842874064e-07, "loss": 0.0058, "step": 289670 }, { "epoch": 184.62715105162525, "grad_norm": 0.09262195229530334, "learning_rate": 3.232399989794854e-07, "loss": 0.0048, "step": 289680 }, { "epoch": 184.63352453792226, "grad_norm": 0.1320497840642929, "learning_rate": 3.226146031497679e-07, "loss": 0.0121, "step": 289690 }, { "epoch": 184.63989802421924, "grad_norm": 0.07701773941516876, "learning_rate": 3.219898109471875e-07, "loss": 0.0047, "step": 289700 }, { "epoch": 184.64627151051624, "grad_norm": 0.10271897912025452, "learning_rate": 3.213656223793382e-07, "loss": 0.0064, "step": 289710 }, { "epoch": 184.65264499681325, "grad_norm": 0.1893012523651123, "learning_rate": 3.207420374537973e-07, "loss": 0.0037, "step": 289720 }, { "epoch": 184.65901848311026, "grad_norm": 0.09091983735561371, "learning_rate": 3.20119056178142e-07, "loss": 0.0036, "step": 289730 }, { "epoch": 184.66539196940727, "grad_norm": 0.07254134863615036, "learning_rate": 3.194966785599496e-07, "loss": 0.005, "step": 289740 }, { "epoch": 184.67176545570427, "grad_norm": 0.11614732444286346, "learning_rate": 3.1887490460677517e-07, "loss": 0.0049, "step": 289750 }, { "epoch": 184.67813894200128, "grad_norm": 0.18051700294017792, "learning_rate": 3.182537343261849e-07, "loss": 0.0033, "step": 289760 }, { "epoch": 184.6845124282983, "grad_norm": 0.10568675398826599, "learning_rate": 3.176331677257116e-07, "loss": 0.0095, "step": 289770 }, { "epoch": 184.69088591459527, "grad_norm": 0.40448036789894104, "learning_rate": 3.1701320481290486e-07, "loss": 0.0041, "step": 289780 }, { "epoch": 184.69725940089228, "grad_norm": 0.07743049412965775, "learning_rate": 3.163938455952975e-07, "loss": 0.0026, "step": 289790 }, { "epoch": 184.7036328871893, "grad_norm": 0.3599030673503876, "learning_rate": 3.1577509008041685e-07, "loss": 0.0065, "step": 289800 }, { "epoch": 184.7100063734863, "grad_norm": 0.1278051733970642, "learning_rate": 3.1515693827577353e-07, "loss": 0.0039, "step": 289810 }, { "epoch": 184.7163798597833, "grad_norm": 0.1284254789352417, "learning_rate": 3.1453939018888376e-07, "loss": 0.0041, "step": 289820 }, { "epoch": 184.7227533460803, "grad_norm": 0.07220626622438431, "learning_rate": 3.1392244582725274e-07, "loss": 0.0102, "step": 289830 }, { "epoch": 184.72912683237732, "grad_norm": 0.15798111259937286, "learning_rate": 3.133061051983799e-07, "loss": 0.0043, "step": 289840 }, { "epoch": 184.73550031867433, "grad_norm": 0.23855635523796082, "learning_rate": 3.126903683097426e-07, "loss": 0.0038, "step": 289850 }, { "epoch": 184.7418738049713, "grad_norm": 0.17826679348945618, "learning_rate": 3.1207523516883495e-07, "loss": 0.0064, "step": 289860 }, { "epoch": 184.7482472912683, "grad_norm": 0.13155198097229004, "learning_rate": 3.114607057831231e-07, "loss": 0.0049, "step": 289870 }, { "epoch": 184.75462077756532, "grad_norm": 0.04790632054209709, "learning_rate": 3.1084678016007893e-07, "loss": 0.0117, "step": 289880 }, { "epoch": 184.76099426386233, "grad_norm": 0.049866050481796265, "learning_rate": 3.102334583071631e-07, "loss": 0.0035, "step": 289890 }, { "epoch": 184.76736775015934, "grad_norm": 0.13297773897647858, "learning_rate": 3.096207402318252e-07, "loss": 0.0051, "step": 289900 }, { "epoch": 184.77374123645635, "grad_norm": 0.37479424476623535, "learning_rate": 3.090086259415037e-07, "loss": 0.0056, "step": 289910 }, { "epoch": 184.78011472275335, "grad_norm": 0.3393571078777313, "learning_rate": 3.083971154436538e-07, "loss": 0.0109, "step": 289920 }, { "epoch": 184.78648820905036, "grad_norm": 0.08752891421318054, "learning_rate": 3.077862087456862e-07, "loss": 0.0061, "step": 289930 }, { "epoch": 184.79286169534734, "grad_norm": 0.08219563961029053, "learning_rate": 3.071759058550394e-07, "loss": 0.0054, "step": 289940 }, { "epoch": 184.79923518164435, "grad_norm": 0.278680682182312, "learning_rate": 3.065662067791186e-07, "loss": 0.0037, "step": 289950 }, { "epoch": 184.80560866794136, "grad_norm": 1.0001219511032104, "learning_rate": 3.0595711152534013e-07, "loss": 0.005, "step": 289960 }, { "epoch": 184.81198215423836, "grad_norm": 0.5082775950431824, "learning_rate": 3.0534862010110353e-07, "loss": 0.0032, "step": 289970 }, { "epoch": 184.81835564053537, "grad_norm": 0.06341010332107544, "learning_rate": 3.047407325137974e-07, "loss": 0.0184, "step": 289980 }, { "epoch": 184.82472912683238, "grad_norm": 0.17697374522686005, "learning_rate": 3.041334487708103e-07, "loss": 0.0055, "step": 289990 }, { "epoch": 184.8311026131294, "grad_norm": 0.24596352875232697, "learning_rate": 3.035267688795251e-07, "loss": 0.006, "step": 290000 }, { "epoch": 184.8374760994264, "grad_norm": 0.31702330708503723, "learning_rate": 3.0292069284731383e-07, "loss": 0.0051, "step": 290010 }, { "epoch": 184.8438495857234, "grad_norm": 0.08183570951223373, "learning_rate": 3.0231522068153716e-07, "loss": 0.0039, "step": 290020 }, { "epoch": 184.85022307202038, "grad_norm": 0.06828751415014267, "learning_rate": 3.017103523895448e-07, "loss": 0.0031, "step": 290030 }, { "epoch": 184.8565965583174, "grad_norm": 0.24386285245418549, "learning_rate": 3.011060879787031e-07, "loss": 0.0034, "step": 290040 }, { "epoch": 184.8629700446144, "grad_norm": 0.07285426557064056, "learning_rate": 3.00502427456345e-07, "loss": 0.005, "step": 290050 }, { "epoch": 184.8693435309114, "grad_norm": 0.16393078863620758, "learning_rate": 2.998993708298037e-07, "loss": 0.0112, "step": 290060 }, { "epoch": 184.87571701720842, "grad_norm": 0.15870900452136993, "learning_rate": 2.992969181064065e-07, "loss": 0.0065, "step": 290070 }, { "epoch": 184.88209050350542, "grad_norm": 0.07148761302232742, "learning_rate": 2.9869506929348644e-07, "loss": 0.0045, "step": 290080 }, { "epoch": 184.88846398980243, "grad_norm": 0.15362554788589478, "learning_rate": 2.980938243983378e-07, "loss": 0.0048, "step": 290090 }, { "epoch": 184.89483747609944, "grad_norm": 0.3956873118877411, "learning_rate": 2.9749318342828236e-07, "loss": 0.0046, "step": 290100 }, { "epoch": 184.90121096239642, "grad_norm": 0.19328153133392334, "learning_rate": 2.968931463906033e-07, "loss": 0.004, "step": 290110 }, { "epoch": 184.90758444869343, "grad_norm": 0.2982752323150635, "learning_rate": 2.9629371329261133e-07, "loss": 0.0092, "step": 290120 }, { "epoch": 184.91395793499044, "grad_norm": 0.1329202651977539, "learning_rate": 2.9569488414156743e-07, "loss": 0.0039, "step": 290130 }, { "epoch": 184.92033142128744, "grad_norm": 0.27125775814056396, "learning_rate": 2.9509665894476566e-07, "loss": 0.0044, "step": 290140 }, { "epoch": 184.92670490758445, "grad_norm": 0.04344259947538376, "learning_rate": 2.944990377094725e-07, "loss": 0.0028, "step": 290150 }, { "epoch": 184.93307839388146, "grad_norm": 0.09883449971675873, "learning_rate": 2.939020204429377e-07, "loss": 0.0039, "step": 290160 }, { "epoch": 184.93945188017847, "grad_norm": 0.0877203643321991, "learning_rate": 2.933056071524276e-07, "loss": 0.0038, "step": 290170 }, { "epoch": 184.94582536647547, "grad_norm": 0.2967046797275543, "learning_rate": 2.927097978451865e-07, "loss": 0.0055, "step": 290180 }, { "epoch": 184.95219885277245, "grad_norm": 0.060098737478256226, "learning_rate": 2.9211459252845297e-07, "loss": 0.0097, "step": 290190 }, { "epoch": 184.95857233906946, "grad_norm": 0.04740971326828003, "learning_rate": 2.915199912094546e-07, "loss": 0.0046, "step": 290200 }, { "epoch": 184.96494582536647, "grad_norm": 0.06681862473487854, "learning_rate": 2.909259938954245e-07, "loss": 0.0063, "step": 290210 }, { "epoch": 184.97131931166348, "grad_norm": 0.1293976902961731, "learning_rate": 2.9033260059357913e-07, "loss": 0.0048, "step": 290220 }, { "epoch": 184.9776927979605, "grad_norm": 0.26579952239990234, "learning_rate": 2.897398113111183e-07, "loss": 0.0125, "step": 290230 }, { "epoch": 184.9840662842575, "grad_norm": 0.17375050485134125, "learning_rate": 2.8914762605525835e-07, "loss": 0.0047, "step": 290240 }, { "epoch": 184.9904397705545, "grad_norm": 0.20056365430355072, "learning_rate": 2.885560448331881e-07, "loss": 0.0059, "step": 290250 }, { "epoch": 184.9968132568515, "grad_norm": 0.45037636160850525, "learning_rate": 2.8796506765210176e-07, "loss": 0.0172, "step": 290260 }, { "epoch": 185.0031867431485, "grad_norm": 0.4398123025894165, "learning_rate": 2.873746945191713e-07, "loss": 0.0083, "step": 290270 }, { "epoch": 185.0095602294455, "grad_norm": 0.09031985700130463, "learning_rate": 2.867849254415744e-07, "loss": 0.006, "step": 290280 }, { "epoch": 185.0159337157425, "grad_norm": 0.34482720494270325, "learning_rate": 2.861957604264831e-07, "loss": 0.0035, "step": 290290 }, { "epoch": 185.0223072020395, "grad_norm": 0.23276205360889435, "learning_rate": 2.856071994810472e-07, "loss": 0.004, "step": 290300 }, { "epoch": 185.02868068833652, "grad_norm": 0.22577446699142456, "learning_rate": 2.850192426124165e-07, "loss": 0.0065, "step": 290310 }, { "epoch": 185.03505417463353, "grad_norm": 0.13116554915905, "learning_rate": 2.8443188982774647e-07, "loss": 0.0082, "step": 290320 }, { "epoch": 185.04142766093054, "grad_norm": 0.5872001647949219, "learning_rate": 2.838451411341703e-07, "loss": 0.0074, "step": 290330 }, { "epoch": 185.04780114722755, "grad_norm": 0.160243958234787, "learning_rate": 2.8325899653881547e-07, "loss": 0.0037, "step": 290340 }, { "epoch": 185.05417463352453, "grad_norm": 0.463890939950943, "learning_rate": 2.826734560487987e-07, "loss": 0.0073, "step": 290350 }, { "epoch": 185.06054811982153, "grad_norm": 0.24963873624801636, "learning_rate": 2.8208851967124193e-07, "loss": 0.0071, "step": 290360 }, { "epoch": 185.06692160611854, "grad_norm": 0.07148229330778122, "learning_rate": 2.815041874132562e-07, "loss": 0.0116, "step": 290370 }, { "epoch": 185.07329509241555, "grad_norm": 0.19996747374534607, "learning_rate": 2.8092045928193034e-07, "loss": 0.0066, "step": 290380 }, { "epoch": 185.07966857871256, "grad_norm": 0.06001577526330948, "learning_rate": 2.803373352843641e-07, "loss": 0.0037, "step": 290390 }, { "epoch": 185.08604206500956, "grad_norm": 0.13204964995384216, "learning_rate": 2.797548154276464e-07, "loss": 0.0062, "step": 290400 }, { "epoch": 185.09241555130657, "grad_norm": 0.19787274301052094, "learning_rate": 2.791728997188492e-07, "loss": 0.0026, "step": 290410 }, { "epoch": 185.09878903760358, "grad_norm": 0.30198439955711365, "learning_rate": 2.7859158816504474e-07, "loss": 0.0027, "step": 290420 }, { "epoch": 185.10516252390056, "grad_norm": 0.10928566008806229, "learning_rate": 2.780108807732995e-07, "loss": 0.0042, "step": 290430 }, { "epoch": 185.11153601019757, "grad_norm": 0.09055133163928986, "learning_rate": 2.774307775506635e-07, "loss": 0.0049, "step": 290440 }, { "epoch": 185.11790949649458, "grad_norm": 0.08218362182378769, "learning_rate": 2.7685127850419766e-07, "loss": 0.0044, "step": 290450 }, { "epoch": 185.12428298279158, "grad_norm": 0.26944881677627563, "learning_rate": 2.7627238364092975e-07, "loss": 0.0119, "step": 290460 }, { "epoch": 185.1306564690886, "grad_norm": 0.11814144998788834, "learning_rate": 2.756940929678986e-07, "loss": 0.0045, "step": 290470 }, { "epoch": 185.1370299553856, "grad_norm": 0.4961004853248596, "learning_rate": 2.751164064921319e-07, "loss": 0.0063, "step": 290480 }, { "epoch": 185.1434034416826, "grad_norm": 0.1362456977367401, "learning_rate": 2.745393242206462e-07, "loss": 0.0057, "step": 290490 }, { "epoch": 185.14977692797962, "grad_norm": 0.06203081086277962, "learning_rate": 2.739628461604638e-07, "loss": 0.0091, "step": 290500 }, { "epoch": 185.1561504142766, "grad_norm": 0.18736007809638977, "learning_rate": 2.733869723185789e-07, "loss": 0.0031, "step": 290510 }, { "epoch": 185.1625239005736, "grad_norm": 0.33256688714027405, "learning_rate": 2.7281170270198607e-07, "loss": 0.0065, "step": 290520 }, { "epoch": 185.1688973868706, "grad_norm": 0.23050044476985931, "learning_rate": 2.722370373176852e-07, "loss": 0.0035, "step": 290530 }, { "epoch": 185.17527087316762, "grad_norm": 0.06943032145500183, "learning_rate": 2.7166297617265965e-07, "loss": 0.0036, "step": 290540 }, { "epoch": 185.18164435946463, "grad_norm": 0.493923544883728, "learning_rate": 2.7108951927387606e-07, "loss": 0.0075, "step": 290550 }, { "epoch": 185.18801784576164, "grad_norm": 0.10889046639204025, "learning_rate": 2.7051666662830653e-07, "loss": 0.0044, "step": 290560 }, { "epoch": 185.19439133205864, "grad_norm": 0.22211125493049622, "learning_rate": 2.699444182429123e-07, "loss": 0.004, "step": 290570 }, { "epoch": 185.20076481835565, "grad_norm": 0.08589188754558563, "learning_rate": 2.6937277412464877e-07, "loss": 0.0039, "step": 290580 }, { "epoch": 185.20713830465266, "grad_norm": 0.19198748469352722, "learning_rate": 2.6880173428045495e-07, "loss": 0.0025, "step": 290590 }, { "epoch": 185.21351179094964, "grad_norm": 0.07247523218393326, "learning_rate": 2.6823129871727524e-07, "loss": 0.0035, "step": 290600 }, { "epoch": 185.21988527724665, "grad_norm": 0.17256326973438263, "learning_rate": 2.67661467442043e-07, "loss": 0.0043, "step": 290610 }, { "epoch": 185.22625876354365, "grad_norm": 0.06538109481334686, "learning_rate": 2.670922404616749e-07, "loss": 0.0088, "step": 290620 }, { "epoch": 185.23263224984066, "grad_norm": 0.22380755841732025, "learning_rate": 2.6652361778309877e-07, "loss": 0.0031, "step": 290630 }, { "epoch": 185.23900573613767, "grad_norm": 0.18896864354610443, "learning_rate": 2.6595559941320904e-07, "loss": 0.005, "step": 290640 }, { "epoch": 185.24537922243468, "grad_norm": 0.272243469953537, "learning_rate": 2.653881853589224e-07, "loss": 0.0079, "step": 290650 }, { "epoch": 185.2517527087317, "grad_norm": 0.22098560631275177, "learning_rate": 2.6482137562711675e-07, "loss": 0.0108, "step": 290660 }, { "epoch": 185.2581261950287, "grad_norm": 0.22269800305366516, "learning_rate": 2.642551702246976e-07, "loss": 0.005, "step": 290670 }, { "epoch": 185.26449968132567, "grad_norm": 0.12557797133922577, "learning_rate": 2.636895691585317e-07, "loss": 0.0109, "step": 290680 }, { "epoch": 185.27087316762268, "grad_norm": 0.28435590863227844, "learning_rate": 2.6312457243550247e-07, "loss": 0.0033, "step": 290690 }, { "epoch": 185.2772466539197, "grad_norm": 0.08552192896604538, "learning_rate": 2.625601800624655e-07, "loss": 0.0097, "step": 290700 }, { "epoch": 185.2836201402167, "grad_norm": 0.29775407910346985, "learning_rate": 2.61996392046282e-07, "loss": 0.0049, "step": 290710 }, { "epoch": 185.2899936265137, "grad_norm": 0.08600527793169022, "learning_rate": 2.6143320839380205e-07, "loss": 0.0034, "step": 290720 }, { "epoch": 185.2963671128107, "grad_norm": 0.19570870697498322, "learning_rate": 2.608706291118701e-07, "loss": 0.0039, "step": 290730 }, { "epoch": 185.30274059910772, "grad_norm": 0.08156052976846695, "learning_rate": 2.6030865420731965e-07, "loss": 0.0025, "step": 290740 }, { "epoch": 185.30911408540473, "grad_norm": 0.23871131241321564, "learning_rate": 2.597472836869841e-07, "loss": 0.0092, "step": 290750 }, { "epoch": 185.3154875717017, "grad_norm": 0.22568942606449127, "learning_rate": 2.591865175576802e-07, "loss": 0.0025, "step": 290760 }, { "epoch": 185.32186105799872, "grad_norm": 0.05933267995715141, "learning_rate": 2.5862635582622473e-07, "loss": 0.0054, "step": 290770 }, { "epoch": 185.32823454429573, "grad_norm": 0.25143858790397644, "learning_rate": 2.580667984994234e-07, "loss": 0.0032, "step": 290780 }, { "epoch": 185.33460803059273, "grad_norm": 0.06820507347583771, "learning_rate": 2.575078455840763e-07, "loss": 0.0057, "step": 290790 }, { "epoch": 185.34098151688974, "grad_norm": 0.2944341003894806, "learning_rate": 2.569494970869668e-07, "loss": 0.0094, "step": 290800 }, { "epoch": 185.34735500318675, "grad_norm": 0.19391904771327972, "learning_rate": 2.563917530148896e-07, "loss": 0.0064, "step": 290810 }, { "epoch": 185.35372848948376, "grad_norm": 0.1257154643535614, "learning_rate": 2.55834613374617e-07, "loss": 0.0082, "step": 290820 }, { "epoch": 185.36010197578076, "grad_norm": 0.13993310928344727, "learning_rate": 2.5527807817292136e-07, "loss": 0.0063, "step": 290830 }, { "epoch": 185.36647546207774, "grad_norm": 0.3170294165611267, "learning_rate": 2.54722147416564e-07, "loss": 0.0067, "step": 290840 }, { "epoch": 185.37284894837475, "grad_norm": 0.2984769344329834, "learning_rate": 2.541668211123005e-07, "loss": 0.0053, "step": 290850 }, { "epoch": 185.37922243467176, "grad_norm": 0.18794798851013184, "learning_rate": 2.536120992668811e-07, "loss": 0.0039, "step": 290860 }, { "epoch": 185.38559592096877, "grad_norm": 0.07969775050878525, "learning_rate": 2.530579818870393e-07, "loss": 0.0022, "step": 290870 }, { "epoch": 185.39196940726578, "grad_norm": 0.10214447975158691, "learning_rate": 2.5250446897950866e-07, "loss": 0.002, "step": 290880 }, { "epoch": 185.39834289356278, "grad_norm": 0.2977011501789093, "learning_rate": 2.5195156055102254e-07, "loss": 0.0056, "step": 290890 }, { "epoch": 185.4047163798598, "grad_norm": 0.08958638459444046, "learning_rate": 2.5139925660829236e-07, "loss": 0.0046, "step": 290900 }, { "epoch": 185.4110898661568, "grad_norm": 0.17122530937194824, "learning_rate": 2.508475571580349e-07, "loss": 0.0034, "step": 290910 }, { "epoch": 185.41746335245378, "grad_norm": 0.12403066456317902, "learning_rate": 2.502964622069448e-07, "loss": 0.003, "step": 290920 }, { "epoch": 185.4238368387508, "grad_norm": 0.33144423365592957, "learning_rate": 2.497459717617334e-07, "loss": 0.0052, "step": 290930 }, { "epoch": 185.4302103250478, "grad_norm": 0.31275153160095215, "learning_rate": 2.4919608582907317e-07, "loss": 0.006, "step": 290940 }, { "epoch": 185.4365838113448, "grad_norm": 0.1927715539932251, "learning_rate": 2.486468044156587e-07, "loss": 0.0038, "step": 290950 }, { "epoch": 185.4429572976418, "grad_norm": 0.16450943052768707, "learning_rate": 2.4809812752815134e-07, "loss": 0.0068, "step": 290960 }, { "epoch": 185.44933078393882, "grad_norm": 0.1533139944076538, "learning_rate": 2.4755005517322904e-07, "loss": 0.0035, "step": 290970 }, { "epoch": 185.45570427023583, "grad_norm": 0.47796526551246643, "learning_rate": 2.470025873575477e-07, "loss": 0.005, "step": 290980 }, { "epoch": 185.46207775653284, "grad_norm": 0.1635933518409729, "learning_rate": 2.46455724087763e-07, "loss": 0.0032, "step": 290990 }, { "epoch": 185.46845124282981, "grad_norm": 0.39078593254089355, "learning_rate": 2.4590946537050854e-07, "loss": 0.0241, "step": 291000 }, { "epoch": 185.47482472912682, "grad_norm": 0.08698996901512146, "learning_rate": 2.4536381121243457e-07, "loss": 0.0042, "step": 291010 }, { "epoch": 185.48119821542383, "grad_norm": 0.37984997034072876, "learning_rate": 2.4481876162016915e-07, "loss": 0.0069, "step": 291020 }, { "epoch": 185.48757170172084, "grad_norm": 0.34721630811691284, "learning_rate": 2.442743166003292e-07, "loss": 0.0086, "step": 291030 }, { "epoch": 185.49394518801785, "grad_norm": 0.22775481641292572, "learning_rate": 2.437304761595316e-07, "loss": 0.0094, "step": 291040 }, { "epoch": 185.50031867431485, "grad_norm": 0.13609904050827026, "learning_rate": 2.431872403043878e-07, "loss": 0.0041, "step": 291050 }, { "epoch": 185.50669216061186, "grad_norm": 0.06900975853204727, "learning_rate": 2.42644609041498e-07, "loss": 0.0025, "step": 291060 }, { "epoch": 185.51306564690887, "grad_norm": 0.18953898549079895, "learning_rate": 2.4210258237745697e-07, "loss": 0.0032, "step": 291070 }, { "epoch": 185.51943913320585, "grad_norm": 0.30948683619499207, "learning_rate": 2.415611603188428e-07, "loss": 0.0051, "step": 291080 }, { "epoch": 185.52581261950286, "grad_norm": 0.191049262881279, "learning_rate": 2.410203428722446e-07, "loss": 0.0192, "step": 291090 }, { "epoch": 185.53218610579987, "grad_norm": 0.047689393162727356, "learning_rate": 2.404801300442294e-07, "loss": 0.002, "step": 291100 }, { "epoch": 185.53855959209687, "grad_norm": 0.46798738837242126, "learning_rate": 2.399405218413642e-07, "loss": 0.0068, "step": 291110 }, { "epoch": 185.54493307839388, "grad_norm": 0.2127700299024582, "learning_rate": 2.394015182701992e-07, "loss": 0.0045, "step": 291120 }, { "epoch": 185.5513065646909, "grad_norm": 0.18199217319488525, "learning_rate": 2.3886311933728476e-07, "loss": 0.0043, "step": 291130 }, { "epoch": 185.5576800509879, "grad_norm": 0.0881122425198555, "learning_rate": 2.3832532504917127e-07, "loss": 0.0044, "step": 291140 }, { "epoch": 185.5640535372849, "grad_norm": 0.28642377257347107, "learning_rate": 2.3778813541238675e-07, "loss": 0.01, "step": 291150 }, { "epoch": 185.57042702358189, "grad_norm": 0.3190871775150299, "learning_rate": 2.3725155043345936e-07, "loss": 0.0042, "step": 291160 }, { "epoch": 185.5768005098789, "grad_norm": 0.24020305275917053, "learning_rate": 2.367155701189061e-07, "loss": 0.0121, "step": 291170 }, { "epoch": 185.5831739961759, "grad_norm": 0.16514429450035095, "learning_rate": 2.3618019447524399e-07, "loss": 0.0213, "step": 291180 }, { "epoch": 185.5895474824729, "grad_norm": 0.2077716439962387, "learning_rate": 2.3564542350897888e-07, "loss": 0.0044, "step": 291190 }, { "epoch": 185.59592096876992, "grad_norm": 0.1795341968536377, "learning_rate": 2.351112572266112e-07, "loss": 0.0072, "step": 291200 }, { "epoch": 185.60229445506693, "grad_norm": 0.2248963862657547, "learning_rate": 2.3457769563461906e-07, "loss": 0.0041, "step": 291210 }, { "epoch": 185.60866794136393, "grad_norm": 0.5218859910964966, "learning_rate": 2.3404473873950283e-07, "loss": 0.0111, "step": 291220 }, { "epoch": 185.61504142766094, "grad_norm": 0.40001237392425537, "learning_rate": 2.33512386547724e-07, "loss": 0.0047, "step": 291230 }, { "epoch": 185.62141491395795, "grad_norm": 0.39643394947052, "learning_rate": 2.3298063906576073e-07, "loss": 0.0075, "step": 291240 }, { "epoch": 185.62778840025493, "grad_norm": 0.2540898323059082, "learning_rate": 2.3244949630006896e-07, "loss": 0.0069, "step": 291250 }, { "epoch": 185.63416188655194, "grad_norm": 0.06941527873277664, "learning_rate": 2.3191895825710463e-07, "loss": 0.0105, "step": 291260 }, { "epoch": 185.64053537284894, "grad_norm": 0.170094832777977, "learning_rate": 2.3138902494330705e-07, "loss": 0.0072, "step": 291270 }, { "epoch": 185.64690885914595, "grad_norm": 0.11320376396179199, "learning_rate": 2.3085969636513216e-07, "loss": 0.0042, "step": 291280 }, { "epoch": 185.65328234544296, "grad_norm": 0.2201387584209442, "learning_rate": 2.3033097252899704e-07, "loss": 0.0023, "step": 291290 }, { "epoch": 185.65965583173997, "grad_norm": 0.3502888083457947, "learning_rate": 2.2980285344133546e-07, "loss": 0.0053, "step": 291300 }, { "epoch": 185.66602931803698, "grad_norm": 0.18620391190052032, "learning_rate": 2.2927533910855337e-07, "loss": 0.0066, "step": 291310 }, { "epoch": 185.67240280433398, "grad_norm": 0.18747642636299133, "learning_rate": 2.2874842953706788e-07, "loss": 0.0069, "step": 291320 }, { "epoch": 185.67877629063096, "grad_norm": 0.4576979875564575, "learning_rate": 2.2822212473328496e-07, "loss": 0.0046, "step": 291330 }, { "epoch": 185.68514977692797, "grad_norm": 0.10099223256111145, "learning_rate": 2.2769642470359398e-07, "loss": 0.0031, "step": 291340 }, { "epoch": 185.69152326322498, "grad_norm": 0.26900240778923035, "learning_rate": 2.2717132945438425e-07, "loss": 0.0037, "step": 291350 }, { "epoch": 185.697896749522, "grad_norm": 0.5082555413246155, "learning_rate": 2.2664683899203953e-07, "loss": 0.0052, "step": 291360 }, { "epoch": 185.704270235819, "grad_norm": 0.12500350177288055, "learning_rate": 2.2612295332292145e-07, "loss": 0.0059, "step": 291370 }, { "epoch": 185.710643722116, "grad_norm": 0.32830241322517395, "learning_rate": 2.255996724534082e-07, "loss": 0.007, "step": 291380 }, { "epoch": 185.717017208413, "grad_norm": 0.11577259749174118, "learning_rate": 2.250769963898558e-07, "loss": 0.0071, "step": 291390 }, { "epoch": 185.72339069471002, "grad_norm": 0.17236974835395813, "learning_rate": 2.2455492513861475e-07, "loss": 0.0032, "step": 291400 }, { "epoch": 185.729764181007, "grad_norm": 0.17355944216251373, "learning_rate": 2.2403345870601887e-07, "loss": 0.0059, "step": 291410 }, { "epoch": 185.736137667304, "grad_norm": 0.0788334533572197, "learning_rate": 2.2351259709841866e-07, "loss": 0.0064, "step": 291420 }, { "epoch": 185.74251115360101, "grad_norm": 0.14657831192016602, "learning_rate": 2.2299234032213677e-07, "loss": 0.0091, "step": 291430 }, { "epoch": 185.74888463989802, "grad_norm": 0.09061586111783981, "learning_rate": 2.22472688383496e-07, "loss": 0.0023, "step": 291440 }, { "epoch": 185.75525812619503, "grad_norm": 0.10473926365375519, "learning_rate": 2.219536412888079e-07, "loss": 0.0037, "step": 291450 }, { "epoch": 185.76163161249204, "grad_norm": 0.1865634173154831, "learning_rate": 2.2143519904438413e-07, "loss": 0.0024, "step": 291460 }, { "epoch": 185.76800509878905, "grad_norm": 0.219227597117424, "learning_rate": 2.2091736165651965e-07, "loss": 0.0054, "step": 291470 }, { "epoch": 185.77437858508605, "grad_norm": 0.3612596392631531, "learning_rate": 2.2040012913150942e-07, "loss": 0.0061, "step": 291480 }, { "epoch": 185.78075207138303, "grad_norm": 0.32340195775032043, "learning_rate": 2.1988350147563174e-07, "loss": 0.0027, "step": 291490 }, { "epoch": 185.78712555768004, "grad_norm": 0.20477785170078278, "learning_rate": 2.1936747869517605e-07, "loss": 0.0072, "step": 291500 }, { "epoch": 185.79349904397705, "grad_norm": 0.1607261747121811, "learning_rate": 2.1885206079640397e-07, "loss": 0.0048, "step": 291510 }, { "epoch": 185.79987253027406, "grad_norm": 0.06678816676139832, "learning_rate": 2.1833724778558272e-07, "loss": 0.0048, "step": 291520 }, { "epoch": 185.80624601657107, "grad_norm": 0.11029834300279617, "learning_rate": 2.1782303966895733e-07, "loss": 0.0066, "step": 291530 }, { "epoch": 185.81261950286807, "grad_norm": 0.08193044364452362, "learning_rate": 2.17309436452795e-07, "loss": 0.0066, "step": 291540 }, { "epoch": 185.81899298916508, "grad_norm": 0.1033463254570961, "learning_rate": 2.167964381433185e-07, "loss": 0.0026, "step": 291550 }, { "epoch": 185.8253664754621, "grad_norm": 0.12715524435043335, "learning_rate": 2.162840447467729e-07, "loss": 0.007, "step": 291560 }, { "epoch": 185.83173996175907, "grad_norm": 0.6350979804992676, "learning_rate": 2.1577225626938092e-07, "loss": 0.0098, "step": 291570 }, { "epoch": 185.83811344805608, "grad_norm": 0.10467188060283661, "learning_rate": 2.152610727173543e-07, "loss": 0.0034, "step": 291580 }, { "epoch": 185.84448693435309, "grad_norm": 0.4742167294025421, "learning_rate": 2.1475049409691582e-07, "loss": 0.0064, "step": 291590 }, { "epoch": 185.8508604206501, "grad_norm": 0.1805267184972763, "learning_rate": 2.1424052041426057e-07, "loss": 0.0046, "step": 291600 }, { "epoch": 185.8572339069471, "grad_norm": 0.1900811791419983, "learning_rate": 2.137311516755891e-07, "loss": 0.0052, "step": 291610 }, { "epoch": 185.8636073932441, "grad_norm": 0.13793990015983582, "learning_rate": 2.132223878870909e-07, "loss": 0.0035, "step": 291620 }, { "epoch": 185.86998087954112, "grad_norm": 0.5593563914299011, "learning_rate": 2.1271422905494442e-07, "loss": 0.0069, "step": 291630 }, { "epoch": 185.87635436583813, "grad_norm": 0.5420768857002258, "learning_rate": 2.12206675185328e-07, "loss": 0.0076, "step": 291640 }, { "epoch": 185.8827278521351, "grad_norm": 0.16045229136943817, "learning_rate": 2.1169972628440892e-07, "loss": 0.0044, "step": 291650 }, { "epoch": 185.8891013384321, "grad_norm": 0.37411537766456604, "learning_rate": 2.1119338235834895e-07, "loss": 0.0079, "step": 291660 }, { "epoch": 185.89547482472912, "grad_norm": 0.0731230229139328, "learning_rate": 2.1068764341329317e-07, "loss": 0.0033, "step": 291670 }, { "epoch": 185.90184831102613, "grad_norm": 0.05873050540685654, "learning_rate": 2.1018250945539775e-07, "loss": 0.0071, "step": 291680 }, { "epoch": 185.90822179732314, "grad_norm": 0.39518141746520996, "learning_rate": 2.0967798049078557e-07, "loss": 0.0075, "step": 291690 }, { "epoch": 185.91459528362014, "grad_norm": 0.2601073086261749, "learning_rate": 2.0917405652560174e-07, "loss": 0.0126, "step": 291700 }, { "epoch": 185.92096876991715, "grad_norm": 0.16774052381515503, "learning_rate": 2.08670737565958e-07, "loss": 0.0038, "step": 291710 }, { "epoch": 185.92734225621416, "grad_norm": 0.05002216994762421, "learning_rate": 2.0816802361798282e-07, "loss": 0.002, "step": 291720 }, { "epoch": 185.93371574251114, "grad_norm": 0.4895937144756317, "learning_rate": 2.076659146877713e-07, "loss": 0.0073, "step": 291730 }, { "epoch": 185.94008922880815, "grad_norm": 0.18163461983203888, "learning_rate": 2.0716441078143521e-07, "loss": 0.004, "step": 291740 }, { "epoch": 185.94646271510516, "grad_norm": 0.2137087881565094, "learning_rate": 2.0666351190506416e-07, "loss": 0.007, "step": 291750 }, { "epoch": 185.95283620140216, "grad_norm": 0.13396598398685455, "learning_rate": 2.0616321806474216e-07, "loss": 0.0069, "step": 291760 }, { "epoch": 185.95920968769917, "grad_norm": 0.087526336312294, "learning_rate": 2.0566352926654763e-07, "loss": 0.0067, "step": 291770 }, { "epoch": 185.96558317399618, "grad_norm": 0.30097028613090515, "learning_rate": 2.0516444551655912e-07, "loss": 0.0042, "step": 291780 }, { "epoch": 185.9719566602932, "grad_norm": 0.08376029878854752, "learning_rate": 2.0466596682083284e-07, "loss": 0.0045, "step": 291790 }, { "epoch": 185.9783301465902, "grad_norm": 0.07889550179243088, "learning_rate": 2.0416809318543063e-07, "loss": 0.0056, "step": 291800 }, { "epoch": 185.98470363288718, "grad_norm": 0.11689627915620804, "learning_rate": 2.0367082461640318e-07, "loss": 0.0014, "step": 291810 }, { "epoch": 185.99107711918418, "grad_norm": 0.13378308713436127, "learning_rate": 2.0317416111979014e-07, "loss": 0.0025, "step": 291820 }, { "epoch": 185.9974506054812, "grad_norm": 0.1587834507226944, "learning_rate": 2.0267810270162556e-07, "loss": 0.0054, "step": 291830 }, { "epoch": 186.0038240917782, "grad_norm": 0.07657632231712341, "learning_rate": 2.0218264936794352e-07, "loss": 0.0042, "step": 291840 }, { "epoch": 186.0101975780752, "grad_norm": 0.19734233617782593, "learning_rate": 2.0168780112475026e-07, "loss": 0.0063, "step": 291850 }, { "epoch": 186.01657106437222, "grad_norm": 0.34313178062438965, "learning_rate": 2.0119355797807437e-07, "loss": 0.0091, "step": 291860 }, { "epoch": 186.02294455066922, "grad_norm": 0.36424073576927185, "learning_rate": 2.0069991993391101e-07, "loss": 0.0047, "step": 291870 }, { "epoch": 186.02931803696623, "grad_norm": 0.1857503205537796, "learning_rate": 2.002068869982665e-07, "loss": 0.0049, "step": 291880 }, { "epoch": 186.03569152326324, "grad_norm": 0.18912501633167267, "learning_rate": 1.9971445917712495e-07, "loss": 0.005, "step": 291890 }, { "epoch": 186.04206500956022, "grad_norm": 0.11153353005647659, "learning_rate": 1.9922263647647045e-07, "loss": 0.0039, "step": 291900 }, { "epoch": 186.04843849585723, "grad_norm": 0.09614957869052887, "learning_rate": 1.9873141890228153e-07, "loss": 0.0058, "step": 291910 }, { "epoch": 186.05481198215423, "grad_norm": 0.18438301980495453, "learning_rate": 1.9824080646052567e-07, "loss": 0.0131, "step": 291920 }, { "epoch": 186.06118546845124, "grad_norm": 0.10231900215148926, "learning_rate": 1.9775079915716476e-07, "loss": 0.0046, "step": 291930 }, { "epoch": 186.06755895474825, "grad_norm": 0.29264703392982483, "learning_rate": 1.9726139699815516e-07, "loss": 0.0052, "step": 291940 }, { "epoch": 186.07393244104526, "grad_norm": 0.13334235548973083, "learning_rate": 1.9677259998943654e-07, "loss": 0.0076, "step": 291950 }, { "epoch": 186.08030592734227, "grad_norm": 0.7273305058479309, "learning_rate": 1.962844081369597e-07, "loss": 0.009, "step": 291960 }, { "epoch": 186.08667941363927, "grad_norm": 0.37757501006126404, "learning_rate": 1.9579682144664214e-07, "loss": 0.0118, "step": 291970 }, { "epoch": 186.09305289993625, "grad_norm": 0.43778252601623535, "learning_rate": 1.9530983992441797e-07, "loss": 0.0056, "step": 291980 }, { "epoch": 186.09942638623326, "grad_norm": 0.19727805256843567, "learning_rate": 1.9482346357620473e-07, "loss": 0.0032, "step": 291990 }, { "epoch": 186.10579987253027, "grad_norm": 0.08712508529424667, "learning_rate": 1.9433769240791434e-07, "loss": 0.0029, "step": 292000 }, { "epoch": 186.11217335882728, "grad_norm": 0.17571458220481873, "learning_rate": 1.938525264254365e-07, "loss": 0.0094, "step": 292010 }, { "epoch": 186.11854684512429, "grad_norm": 0.1943484991788864, "learning_rate": 1.9336796563468318e-07, "loss": 0.0034, "step": 292020 }, { "epoch": 186.1249203314213, "grad_norm": 0.1331087201833725, "learning_rate": 1.9288401004152746e-07, "loss": 0.0057, "step": 292030 }, { "epoch": 186.1312938177183, "grad_norm": 0.06149834394454956, "learning_rate": 1.9240065965185904e-07, "loss": 0.0027, "step": 292040 }, { "epoch": 186.1376673040153, "grad_norm": 0.11453074216842651, "learning_rate": 1.9191791447155104e-07, "loss": 0.0075, "step": 292050 }, { "epoch": 186.1440407903123, "grad_norm": 0.44207242131233215, "learning_rate": 1.9143577450646543e-07, "loss": 0.005, "step": 292060 }, { "epoch": 186.1504142766093, "grad_norm": 0.2213781177997589, "learning_rate": 1.9095423976246417e-07, "loss": 0.012, "step": 292070 }, { "epoch": 186.1567877629063, "grad_norm": 0.18707826733589172, "learning_rate": 1.9047331024539262e-07, "loss": 0.0039, "step": 292080 }, { "epoch": 186.1631612492033, "grad_norm": 0.13709978759288788, "learning_rate": 1.899929859611016e-07, "loss": 0.0041, "step": 292090 }, { "epoch": 186.16953473550032, "grad_norm": 0.07813901454210281, "learning_rate": 1.8951326691542537e-07, "loss": 0.0074, "step": 292100 }, { "epoch": 186.17590822179733, "grad_norm": 0.1511165201663971, "learning_rate": 1.8903415311418705e-07, "loss": 0.0039, "step": 292110 }, { "epoch": 186.18228170809434, "grad_norm": 0.1631588637828827, "learning_rate": 1.8855564456321528e-07, "loss": 0.0048, "step": 292120 }, { "epoch": 186.18865519439134, "grad_norm": 0.2962399423122406, "learning_rate": 1.8807774126832213e-07, "loss": 0.0034, "step": 292130 }, { "epoch": 186.19502868068832, "grad_norm": 0.2609523832798004, "learning_rate": 1.8760044323531955e-07, "loss": 0.0063, "step": 292140 }, { "epoch": 186.20140216698533, "grad_norm": 0.1056441068649292, "learning_rate": 1.8712375046999743e-07, "loss": 0.0078, "step": 292150 }, { "epoch": 186.20777565328234, "grad_norm": 0.12404213100671768, "learning_rate": 1.8664766297815106e-07, "loss": 0.0025, "step": 292160 }, { "epoch": 186.21414913957935, "grad_norm": 0.09677920490503311, "learning_rate": 1.861721807655703e-07, "loss": 0.0031, "step": 292170 }, { "epoch": 186.22052262587636, "grad_norm": 0.15987084805965424, "learning_rate": 1.856973038380283e-07, "loss": 0.0077, "step": 292180 }, { "epoch": 186.22689611217336, "grad_norm": 0.25730741024017334, "learning_rate": 1.8522303220129823e-07, "loss": 0.004, "step": 292190 }, { "epoch": 186.23326959847037, "grad_norm": 0.08574479818344116, "learning_rate": 1.8474936586113656e-07, "loss": 0.0061, "step": 292200 }, { "epoch": 186.23964308476738, "grad_norm": 0.16530801355838776, "learning_rate": 1.8427630482331094e-07, "loss": 0.0021, "step": 292210 }, { "epoch": 186.24601657106436, "grad_norm": 0.06958524137735367, "learning_rate": 1.8380384909355564e-07, "loss": 0.0038, "step": 292220 }, { "epoch": 186.25239005736137, "grad_norm": 0.22153106331825256, "learning_rate": 1.8333199867762163e-07, "loss": 0.0096, "step": 292230 }, { "epoch": 186.25876354365838, "grad_norm": 0.08938434720039368, "learning_rate": 1.8286075358123768e-07, "loss": 0.0077, "step": 292240 }, { "epoch": 186.26513702995538, "grad_norm": 0.050239596515893936, "learning_rate": 1.8239011381013248e-07, "loss": 0.0121, "step": 292250 }, { "epoch": 186.2715105162524, "grad_norm": 0.2271101325750351, "learning_rate": 1.8192007937001822e-07, "loss": 0.0059, "step": 292260 }, { "epoch": 186.2778840025494, "grad_norm": 0.059725482016801834, "learning_rate": 1.8145065026661246e-07, "loss": 0.0035, "step": 292270 }, { "epoch": 186.2842574888464, "grad_norm": 0.10103576630353928, "learning_rate": 1.809818265056218e-07, "loss": 0.007, "step": 292280 }, { "epoch": 186.29063097514342, "grad_norm": 0.13001064956188202, "learning_rate": 1.805136080927361e-07, "loss": 0.013, "step": 292290 }, { "epoch": 186.2970044614404, "grad_norm": 0.4808099567890167, "learning_rate": 1.8004599503364527e-07, "loss": 0.0052, "step": 292300 }, { "epoch": 186.3033779477374, "grad_norm": 0.06423653662204742, "learning_rate": 1.7957898733403922e-07, "loss": 0.0028, "step": 292310 }, { "epoch": 186.3097514340344, "grad_norm": 0.353520005941391, "learning_rate": 1.7911258499958006e-07, "loss": 0.0092, "step": 292320 }, { "epoch": 186.31612492033142, "grad_norm": 0.3368816077709198, "learning_rate": 1.786467880359466e-07, "loss": 0.0044, "step": 292330 }, { "epoch": 186.32249840662843, "grad_norm": 0.43014615774154663, "learning_rate": 1.7818159644878984e-07, "loss": 0.0054, "step": 292340 }, { "epoch": 186.32887189292543, "grad_norm": 0.05435171723365784, "learning_rate": 1.7771701024377197e-07, "loss": 0.0021, "step": 292350 }, { "epoch": 186.33524537922244, "grad_norm": 0.0550537072122097, "learning_rate": 1.7725302942652734e-07, "loss": 0.0033, "step": 292360 }, { "epoch": 186.34161886551945, "grad_norm": 0.08948510885238647, "learning_rate": 1.7678965400270143e-07, "loss": 0.0037, "step": 292370 }, { "epoch": 186.34799235181643, "grad_norm": 0.14999991655349731, "learning_rate": 1.7632688397791752e-07, "loss": 0.0046, "step": 292380 }, { "epoch": 186.35436583811344, "grad_norm": 0.2071230113506317, "learning_rate": 1.7586471935781002e-07, "loss": 0.0063, "step": 292390 }, { "epoch": 186.36073932441045, "grad_norm": 0.2946870028972626, "learning_rate": 1.7540316014798552e-07, "loss": 0.0091, "step": 292400 }, { "epoch": 186.36711281070745, "grad_norm": 0.2819874882698059, "learning_rate": 1.7494220635405624e-07, "loss": 0.0038, "step": 292410 }, { "epoch": 186.37348629700446, "grad_norm": 0.10974999517202377, "learning_rate": 1.7448185798161764e-07, "loss": 0.004, "step": 292420 }, { "epoch": 186.37985978330147, "grad_norm": 0.2469206303358078, "learning_rate": 1.7402211503627087e-07, "loss": 0.0084, "step": 292430 }, { "epoch": 186.38623326959848, "grad_norm": 0.2352783977985382, "learning_rate": 1.7356297752359474e-07, "loss": 0.0073, "step": 292440 }, { "epoch": 186.39260675589549, "grad_norm": 0.17599108815193176, "learning_rate": 1.7310444544917925e-07, "loss": 0.0046, "step": 292450 }, { "epoch": 186.39898024219247, "grad_norm": 0.10660089552402496, "learning_rate": 1.7264651881858108e-07, "loss": 0.005, "step": 292460 }, { "epoch": 186.40535372848947, "grad_norm": 0.10907768458127975, "learning_rate": 1.7218919763737907e-07, "loss": 0.005, "step": 292470 }, { "epoch": 186.41172721478648, "grad_norm": 0.040015075355768204, "learning_rate": 1.717324819111188e-07, "loss": 0.0028, "step": 292480 }, { "epoch": 186.4181007010835, "grad_norm": 0.13007961213588715, "learning_rate": 1.7127637164535692e-07, "loss": 0.0023, "step": 292490 }, { "epoch": 186.4244741873805, "grad_norm": 0.07927694171667099, "learning_rate": 1.7082086684563347e-07, "loss": 0.0043, "step": 292500 }, { "epoch": 186.4308476736775, "grad_norm": 0.3887336552143097, "learning_rate": 1.703659675174829e-07, "loss": 0.0041, "step": 292510 }, { "epoch": 186.4372211599745, "grad_norm": 0.2643521726131439, "learning_rate": 1.699116736664341e-07, "loss": 0.0061, "step": 292520 }, { "epoch": 186.44359464627152, "grad_norm": 0.09103967994451523, "learning_rate": 1.6945798529800494e-07, "loss": 0.0034, "step": 292530 }, { "epoch": 186.44996813256853, "grad_norm": 0.1827927976846695, "learning_rate": 1.690049024177076e-07, "loss": 0.0054, "step": 292540 }, { "epoch": 186.4563416188655, "grad_norm": 0.05691580846905708, "learning_rate": 1.685524250310544e-07, "loss": 0.0023, "step": 292550 }, { "epoch": 186.46271510516252, "grad_norm": 0.2770381569862366, "learning_rate": 1.6810055314352978e-07, "loss": 0.008, "step": 292560 }, { "epoch": 186.46908859145952, "grad_norm": 0.5321934223175049, "learning_rate": 1.6764928676063495e-07, "loss": 0.0084, "step": 292570 }, { "epoch": 186.47546207775653, "grad_norm": 0.13042433559894562, "learning_rate": 1.671986258878544e-07, "loss": 0.0057, "step": 292580 }, { "epoch": 186.48183556405354, "grad_norm": 0.4261491000652313, "learning_rate": 1.6674857053065596e-07, "loss": 0.0042, "step": 292590 }, { "epoch": 186.48820905035055, "grad_norm": 0.06248163431882858, "learning_rate": 1.662991206945186e-07, "loss": 0.0034, "step": 292600 }, { "epoch": 186.49458253664756, "grad_norm": 0.2253660410642624, "learning_rate": 1.6585027638489348e-07, "loss": 0.0033, "step": 292610 }, { "epoch": 186.50095602294456, "grad_norm": 0.13169483840465546, "learning_rate": 1.6540203760723738e-07, "loss": 0.0039, "step": 292620 }, { "epoch": 186.50732950924154, "grad_norm": 0.32767847180366516, "learning_rate": 1.649544043670015e-07, "loss": 0.0049, "step": 292630 }, { "epoch": 186.51370299553855, "grad_norm": 0.5656651854515076, "learning_rate": 1.6450737666962035e-07, "loss": 0.0083, "step": 292640 }, { "epoch": 186.52007648183556, "grad_norm": 0.13807065784931183, "learning_rate": 1.640609545205285e-07, "loss": 0.0105, "step": 292650 }, { "epoch": 186.52644996813257, "grad_norm": 0.08594509959220886, "learning_rate": 1.6361513792514936e-07, "loss": 0.0074, "step": 292660 }, { "epoch": 186.53282345442958, "grad_norm": 0.17508846521377563, "learning_rate": 1.6316992688889532e-07, "loss": 0.0026, "step": 292670 }, { "epoch": 186.53919694072658, "grad_norm": 0.13708950579166412, "learning_rate": 1.627253214171842e-07, "loss": 0.0031, "step": 292680 }, { "epoch": 186.5455704270236, "grad_norm": 0.1401226669549942, "learning_rate": 1.6228132151541176e-07, "loss": 0.0039, "step": 292690 }, { "epoch": 186.5519439133206, "grad_norm": 0.08000784367322922, "learning_rate": 1.6183792718897916e-07, "loss": 0.0017, "step": 292700 }, { "epoch": 186.55831739961758, "grad_norm": 0.21372418105602264, "learning_rate": 1.6139513844327102e-07, "loss": 0.0098, "step": 292710 }, { "epoch": 186.5646908859146, "grad_norm": 0.0666375383734703, "learning_rate": 1.6095295528366638e-07, "loss": 0.0023, "step": 292720 }, { "epoch": 186.5710643722116, "grad_norm": 0.27728310227394104, "learning_rate": 1.6051137771553314e-07, "loss": 0.0154, "step": 292730 }, { "epoch": 186.5774378585086, "grad_norm": 0.07730256766080856, "learning_rate": 1.6007040574425036e-07, "loss": 0.0046, "step": 292740 }, { "epoch": 186.5838113448056, "grad_norm": 0.1762198805809021, "learning_rate": 1.596300393751693e-07, "loss": 0.0053, "step": 292750 }, { "epoch": 186.59018483110262, "grad_norm": 0.11476919800043106, "learning_rate": 1.5919027861363568e-07, "loss": 0.0038, "step": 292760 }, { "epoch": 186.59655831739963, "grad_norm": 0.08810577541589737, "learning_rate": 1.5875112346500077e-07, "loss": 0.004, "step": 292770 }, { "epoch": 186.60293180369663, "grad_norm": 0.11959867179393768, "learning_rate": 1.583125739345992e-07, "loss": 0.0038, "step": 292780 }, { "epoch": 186.6093052899936, "grad_norm": 0.18735234439373016, "learning_rate": 1.5787463002774894e-07, "loss": 0.005, "step": 292790 }, { "epoch": 186.61567877629062, "grad_norm": 0.22036412358283997, "learning_rate": 1.5743729174979016e-07, "loss": 0.0049, "step": 292800 }, { "epoch": 186.62205226258763, "grad_norm": 0.3888501226902008, "learning_rate": 1.5700055910602418e-07, "loss": 0.0075, "step": 292810 }, { "epoch": 186.62842574888464, "grad_norm": 0.1994324028491974, "learning_rate": 1.5656443210176342e-07, "loss": 0.0071, "step": 292820 }, { "epoch": 186.63479923518165, "grad_norm": 0.06037644296884537, "learning_rate": 1.5612891074230362e-07, "loss": 0.0027, "step": 292830 }, { "epoch": 186.64117272147865, "grad_norm": 0.08761489391326904, "learning_rate": 1.5569399503293502e-07, "loss": 0.0025, "step": 292840 }, { "epoch": 186.64754620777566, "grad_norm": 0.09517098218202591, "learning_rate": 1.5525968497894784e-07, "loss": 0.0063, "step": 292850 }, { "epoch": 186.65391969407267, "grad_norm": 0.06592141836881638, "learning_rate": 1.548259805856156e-07, "loss": 0.0043, "step": 292860 }, { "epoch": 186.66029318036965, "grad_norm": 0.3064422607421875, "learning_rate": 1.543928818582119e-07, "loss": 0.0055, "step": 292870 }, { "epoch": 186.66666666666666, "grad_norm": 0.17993789911270142, "learning_rate": 1.539603888019936e-07, "loss": 0.0047, "step": 292880 }, { "epoch": 186.67304015296367, "grad_norm": 0.31010204553604126, "learning_rate": 1.535285014222232e-07, "loss": 0.0055, "step": 292890 }, { "epoch": 186.67941363926067, "grad_norm": 0.13510440289974213, "learning_rate": 1.5309721972414092e-07, "loss": 0.0039, "step": 292900 }, { "epoch": 186.68578712555768, "grad_norm": 0.1539820283651352, "learning_rate": 1.5266654371298705e-07, "loss": 0.0035, "step": 292910 }, { "epoch": 186.6921606118547, "grad_norm": 0.13013656437397003, "learning_rate": 1.5223647339400736e-07, "loss": 0.0035, "step": 292920 }, { "epoch": 186.6985340981517, "grad_norm": 0.1096758097410202, "learning_rate": 1.5180700877241438e-07, "loss": 0.0069, "step": 292930 }, { "epoch": 186.7049075844487, "grad_norm": 0.11130701005458832, "learning_rate": 1.5137814985342615e-07, "loss": 0.003, "step": 292940 }, { "epoch": 186.71128107074568, "grad_norm": 0.04903775826096535, "learning_rate": 1.5094989664226067e-07, "loss": 0.0038, "step": 292950 }, { "epoch": 186.7176545570427, "grad_norm": 0.3370482623577118, "learning_rate": 1.5052224914412494e-07, "loss": 0.0044, "step": 292960 }, { "epoch": 186.7240280433397, "grad_norm": 0.3406929671764374, "learning_rate": 1.5009520736420368e-07, "loss": 0.0085, "step": 292970 }, { "epoch": 186.7304015296367, "grad_norm": 0.38929781317710876, "learning_rate": 1.4966877130769274e-07, "loss": 0.0083, "step": 292980 }, { "epoch": 186.73677501593372, "grad_norm": 0.19282303750514984, "learning_rate": 1.4924294097977686e-07, "loss": 0.0025, "step": 292990 }, { "epoch": 186.74314850223072, "grad_norm": 0.21280987560749054, "learning_rate": 1.4881771638562414e-07, "loss": 0.0032, "step": 293000 }, { "epoch": 186.74952198852773, "grad_norm": 0.08164457976818085, "learning_rate": 1.483930975303971e-07, "loss": 0.0068, "step": 293010 }, { "epoch": 186.75589547482474, "grad_norm": 0.05364447459578514, "learning_rate": 1.4796908441926938e-07, "loss": 0.0064, "step": 293020 }, { "epoch": 186.76226896112172, "grad_norm": 0.14942039549350739, "learning_rate": 1.475456770573813e-07, "loss": 0.0035, "step": 293030 }, { "epoch": 186.76864244741873, "grad_norm": 0.10492528975009918, "learning_rate": 1.4712287544988435e-07, "loss": 0.0029, "step": 293040 }, { "epoch": 186.77501593371574, "grad_norm": 0.25303831696510315, "learning_rate": 1.4670067960191324e-07, "loss": 0.0043, "step": 293050 }, { "epoch": 186.78138942001274, "grad_norm": 0.19165261089801788, "learning_rate": 1.4627908951859726e-07, "loss": 0.0241, "step": 293060 }, { "epoch": 186.78776290630975, "grad_norm": 0.12989972531795502, "learning_rate": 1.4585810520506005e-07, "loss": 0.004, "step": 293070 }, { "epoch": 186.79413639260676, "grad_norm": 0.34855443239212036, "learning_rate": 1.4543772666641976e-07, "loss": 0.0071, "step": 293080 }, { "epoch": 186.80050987890377, "grad_norm": 0.09278681874275208, "learning_rate": 1.4501795390777783e-07, "loss": 0.0035, "step": 293090 }, { "epoch": 186.80688336520078, "grad_norm": 0.37080493569374084, "learning_rate": 1.4459878693423578e-07, "loss": 0.0053, "step": 293100 }, { "epoch": 186.81325685149778, "grad_norm": 0.44499292969703674, "learning_rate": 1.4418022575089506e-07, "loss": 0.0058, "step": 293110 }, { "epoch": 186.81963033779476, "grad_norm": 0.4031830430030823, "learning_rate": 1.4376227036283497e-07, "loss": 0.0075, "step": 293120 }, { "epoch": 186.82600382409177, "grad_norm": 0.08908754587173462, "learning_rate": 1.4334492077513472e-07, "loss": 0.0035, "step": 293130 }, { "epoch": 186.83237731038878, "grad_norm": 0.5132228136062622, "learning_rate": 1.429281769928681e-07, "loss": 0.024, "step": 293140 }, { "epoch": 186.8387507966858, "grad_norm": 0.09282772988080978, "learning_rate": 1.4251203902109767e-07, "loss": 0.0036, "step": 293150 }, { "epoch": 186.8451242829828, "grad_norm": 0.0702122300863266, "learning_rate": 1.4209650686487497e-07, "loss": 0.0045, "step": 293160 }, { "epoch": 186.8514977692798, "grad_norm": 0.12971140444278717, "learning_rate": 1.4168158052925707e-07, "loss": 0.0047, "step": 293170 }, { "epoch": 186.8578712555768, "grad_norm": 0.5272262692451477, "learning_rate": 1.4126726001927882e-07, "loss": 0.0134, "step": 293180 }, { "epoch": 186.86424474187382, "grad_norm": 0.10888497531414032, "learning_rate": 1.4085354533998062e-07, "loss": 0.0052, "step": 293190 }, { "epoch": 186.8706182281708, "grad_norm": 0.21098867058753967, "learning_rate": 1.4044043649638627e-07, "loss": 0.0076, "step": 293200 }, { "epoch": 186.8769917144678, "grad_norm": 0.046404264867305756, "learning_rate": 1.4002793349351396e-07, "loss": 0.0032, "step": 293210 }, { "epoch": 186.8833652007648, "grad_norm": 0.14221236109733582, "learning_rate": 1.3961603633638187e-07, "loss": 0.0036, "step": 293220 }, { "epoch": 186.88973868706182, "grad_norm": 0.06501346081495285, "learning_rate": 1.3920474502998603e-07, "loss": 0.0051, "step": 293230 }, { "epoch": 186.89611217335883, "grad_norm": 0.08222560584545135, "learning_rate": 1.3879405957933355e-07, "loss": 0.0061, "step": 293240 }, { "epoch": 186.90248565965584, "grad_norm": 0.11412344127893448, "learning_rate": 1.383839799894038e-07, "loss": 0.0034, "step": 293250 }, { "epoch": 186.90885914595285, "grad_norm": 0.18679320812225342, "learning_rate": 1.379745062651927e-07, "loss": 0.0041, "step": 293260 }, { "epoch": 186.91523263224985, "grad_norm": 0.06392285227775574, "learning_rate": 1.3756563841166303e-07, "loss": 0.011, "step": 293270 }, { "epoch": 186.92160611854683, "grad_norm": 0.13000980019569397, "learning_rate": 1.371573764337941e-07, "loss": 0.0047, "step": 293280 }, { "epoch": 186.92797960484384, "grad_norm": 0.12098698318004608, "learning_rate": 1.3674972033653754e-07, "loss": 0.0047, "step": 293290 }, { "epoch": 186.93435309114085, "grad_norm": 0.13511714339256287, "learning_rate": 1.3634267012485603e-07, "loss": 0.0055, "step": 293300 }, { "epoch": 186.94072657743786, "grad_norm": 0.17707671225070953, "learning_rate": 1.3593622580368448e-07, "loss": 0.0028, "step": 293310 }, { "epoch": 186.94710006373487, "grad_norm": 0.32490989565849304, "learning_rate": 1.3553038737796897e-07, "loss": 0.0047, "step": 293320 }, { "epoch": 186.95347355003187, "grad_norm": 0.10145716369152069, "learning_rate": 1.351251548526389e-07, "loss": 0.0038, "step": 293330 }, { "epoch": 186.95984703632888, "grad_norm": 0.1118275597691536, "learning_rate": 1.3472052823261805e-07, "loss": 0.0063, "step": 293340 }, { "epoch": 186.9662205226259, "grad_norm": 0.24913419783115387, "learning_rate": 1.3431650752282477e-07, "loss": 0.0042, "step": 293350 }, { "epoch": 186.97259400892287, "grad_norm": 0.17025712132453918, "learning_rate": 1.339130927281662e-07, "loss": 0.004, "step": 293360 }, { "epoch": 186.97896749521988, "grad_norm": 0.0731760635972023, "learning_rate": 1.3351028385354404e-07, "loss": 0.0034, "step": 293370 }, { "epoch": 186.98534098151688, "grad_norm": 0.22064927220344543, "learning_rate": 1.331080809038543e-07, "loss": 0.0033, "step": 293380 }, { "epoch": 186.9917144678139, "grad_norm": 0.1256624311208725, "learning_rate": 1.3270648388398198e-07, "loss": 0.0026, "step": 293390 }, { "epoch": 186.9980879541109, "grad_norm": 0.23346738517284393, "learning_rate": 1.323054927988121e-07, "loss": 0.0044, "step": 293400 }, { "epoch": 187.0044614404079, "grad_norm": 0.2609887421131134, "learning_rate": 1.3190510765321297e-07, "loss": 0.0037, "step": 293410 }, { "epoch": 187.01083492670492, "grad_norm": 0.07947564125061035, "learning_rate": 1.3150532845204733e-07, "loss": 0.0068, "step": 293420 }, { "epoch": 187.01720841300192, "grad_norm": 0.1339130848646164, "learning_rate": 1.31106155200178e-07, "loss": 0.0075, "step": 293430 }, { "epoch": 187.0235818992989, "grad_norm": 0.4036775231361389, "learning_rate": 1.3070758790245108e-07, "loss": 0.0097, "step": 293440 }, { "epoch": 187.0299553855959, "grad_norm": 0.20986732840538025, "learning_rate": 1.303096265637127e-07, "loss": 0.005, "step": 293450 }, { "epoch": 187.03632887189292, "grad_norm": 0.1570766121149063, "learning_rate": 1.2991227118879235e-07, "loss": 0.0051, "step": 293460 }, { "epoch": 187.04270235818993, "grad_norm": 0.08658257871866226, "learning_rate": 1.2951552178253056e-07, "loss": 0.0057, "step": 293470 }, { "epoch": 187.04907584448694, "grad_norm": 0.23488439619541168, "learning_rate": 1.2911937834973464e-07, "loss": 0.0034, "step": 293480 }, { "epoch": 187.05544933078394, "grad_norm": 0.11466704308986664, "learning_rate": 1.287238408952285e-07, "loss": 0.0029, "step": 293490 }, { "epoch": 187.06182281708095, "grad_norm": 0.22154605388641357, "learning_rate": 1.283289094238138e-07, "loss": 0.005, "step": 293500 }, { "epoch": 187.06819630337796, "grad_norm": 0.15452761948108673, "learning_rate": 1.2793458394028678e-07, "loss": 0.0046, "step": 293510 }, { "epoch": 187.07456978967494, "grad_norm": 0.15844203531742096, "learning_rate": 1.2754086444944357e-07, "loss": 0.0037, "step": 293520 }, { "epoch": 187.08094327597195, "grad_norm": 0.07734151929616928, "learning_rate": 1.2714775095606368e-07, "loss": 0.0032, "step": 293530 }, { "epoch": 187.08731676226896, "grad_norm": 0.255555123090744, "learning_rate": 1.2675524346492662e-07, "loss": 0.0042, "step": 293540 }, { "epoch": 187.09369024856596, "grad_norm": 0.16847750544548035, "learning_rate": 1.2636334198080634e-07, "loss": 0.0036, "step": 293550 }, { "epoch": 187.10006373486297, "grad_norm": 0.11168022453784943, "learning_rate": 1.2597204650845463e-07, "loss": 0.0059, "step": 293560 }, { "epoch": 187.10643722115998, "grad_norm": 0.040835361927747726, "learning_rate": 1.255813570526343e-07, "loss": 0.0037, "step": 293570 }, { "epoch": 187.112810707457, "grad_norm": 0.20376832783222198, "learning_rate": 1.2519127361809157e-07, "loss": 0.0051, "step": 293580 }, { "epoch": 187.119184193754, "grad_norm": 0.20400331914424896, "learning_rate": 1.2480179620956157e-07, "loss": 0.0055, "step": 293590 }, { "epoch": 187.12555768005097, "grad_norm": 0.4150937795639038, "learning_rate": 1.2441292483177935e-07, "loss": 0.0042, "step": 293600 }, { "epoch": 187.13193116634798, "grad_norm": 0.09765041619539261, "learning_rate": 1.2402465948947451e-07, "loss": 0.0039, "step": 293610 }, { "epoch": 187.138304652645, "grad_norm": 0.13020369410514832, "learning_rate": 1.2363700018735435e-07, "loss": 0.0072, "step": 293620 }, { "epoch": 187.144678138942, "grad_norm": 0.12707877159118652, "learning_rate": 1.2324994693013737e-07, "loss": 0.0059, "step": 293630 }, { "epoch": 187.151051625239, "grad_norm": 0.19918791949748993, "learning_rate": 1.2286349972253087e-07, "loss": 0.004, "step": 293640 }, { "epoch": 187.157425111536, "grad_norm": 0.1392500400543213, "learning_rate": 1.2247765856922e-07, "loss": 0.0041, "step": 293650 }, { "epoch": 187.16379859783302, "grad_norm": 0.279185026884079, "learning_rate": 1.2209242347489546e-07, "loss": 0.0146, "step": 293660 }, { "epoch": 187.17017208413003, "grad_norm": 0.11704249680042267, "learning_rate": 1.217077944442424e-07, "loss": 0.0075, "step": 293670 }, { "epoch": 187.176545570427, "grad_norm": 0.09673374891281128, "learning_rate": 1.2132377148193485e-07, "loss": 0.0025, "step": 293680 }, { "epoch": 187.18291905672402, "grad_norm": 0.5871257185935974, "learning_rate": 1.2094035459263574e-07, "loss": 0.0198, "step": 293690 }, { "epoch": 187.18929254302103, "grad_norm": 0.3233766555786133, "learning_rate": 1.205575437810025e-07, "loss": 0.0103, "step": 293700 }, { "epoch": 187.19566602931803, "grad_norm": 0.280526727437973, "learning_rate": 1.2017533905169242e-07, "loss": 0.0042, "step": 293710 }, { "epoch": 187.20203951561504, "grad_norm": 0.21011652052402496, "learning_rate": 1.1979374040934633e-07, "loss": 0.0049, "step": 293720 }, { "epoch": 187.20841300191205, "grad_norm": 0.14680112898349762, "learning_rate": 1.1941274785859936e-07, "loss": 0.0029, "step": 293730 }, { "epoch": 187.21478648820906, "grad_norm": 0.24374547600746155, "learning_rate": 1.1903236140408113e-07, "loss": 0.0141, "step": 293740 }, { "epoch": 187.22115997450607, "grad_norm": 0.0614871121942997, "learning_rate": 1.1865258105041577e-07, "loss": 0.0091, "step": 293750 }, { "epoch": 187.22753346080307, "grad_norm": 0.09334322810173035, "learning_rate": 1.1827340680222176e-07, "loss": 0.0029, "step": 293760 }, { "epoch": 187.23390694710005, "grad_norm": 0.12376750260591507, "learning_rate": 1.1789483866409545e-07, "loss": 0.0038, "step": 293770 }, { "epoch": 187.24028043339706, "grad_norm": 0.14477285742759705, "learning_rate": 1.175168766406498e-07, "loss": 0.0031, "step": 293780 }, { "epoch": 187.24665391969407, "grad_norm": 0.18278983235359192, "learning_rate": 1.171395207364645e-07, "loss": 0.0054, "step": 293790 }, { "epoch": 187.25302740599108, "grad_norm": 0.10807893425226212, "learning_rate": 1.1676277095613586e-07, "loss": 0.0042, "step": 293800 }, { "epoch": 187.25940089228808, "grad_norm": 0.16416390240192413, "learning_rate": 1.1638662730423799e-07, "loss": 0.0043, "step": 293810 }, { "epoch": 187.2657743785851, "grad_norm": 0.35943156480789185, "learning_rate": 1.1601108978533947e-07, "loss": 0.006, "step": 293820 }, { "epoch": 187.2721478648821, "grad_norm": 0.0941913053393364, "learning_rate": 1.1563615840400333e-07, "loss": 0.0042, "step": 293830 }, { "epoch": 187.2785213511791, "grad_norm": 0.098558709025383, "learning_rate": 1.1526183316478145e-07, "loss": 0.0036, "step": 293840 }, { "epoch": 187.2848948374761, "grad_norm": 0.2571232318878174, "learning_rate": 1.1488811407223687e-07, "loss": 0.0077, "step": 293850 }, { "epoch": 187.2912683237731, "grad_norm": 0.08476009219884872, "learning_rate": 1.1451500113089376e-07, "loss": 0.0103, "step": 293860 }, { "epoch": 187.2976418100701, "grad_norm": 0.2070610225200653, "learning_rate": 1.1414249434529844e-07, "loss": 0.0039, "step": 293870 }, { "epoch": 187.3040152963671, "grad_norm": 0.3443935215473175, "learning_rate": 1.1377059371996956e-07, "loss": 0.0039, "step": 293880 }, { "epoch": 187.31038878266412, "grad_norm": 0.1356041133403778, "learning_rate": 1.1339929925942571e-07, "loss": 0.004, "step": 293890 }, { "epoch": 187.31676226896113, "grad_norm": 0.13355813920497894, "learning_rate": 1.1302861096818551e-07, "loss": 0.0047, "step": 293900 }, { "epoch": 187.32313575525814, "grad_norm": 0.2761976718902588, "learning_rate": 1.1265852885074535e-07, "loss": 0.0082, "step": 293910 }, { "epoch": 187.32950924155514, "grad_norm": 0.22920289635658264, "learning_rate": 1.1228905291160719e-07, "loss": 0.0031, "step": 293920 }, { "epoch": 187.33588272785212, "grad_norm": 0.23525120317935944, "learning_rate": 1.1192018315526187e-07, "loss": 0.005, "step": 293930 }, { "epoch": 187.34225621414913, "grad_norm": 0.10257823765277863, "learning_rate": 1.1155191958618361e-07, "loss": 0.0076, "step": 293940 }, { "epoch": 187.34862970044614, "grad_norm": 0.17632749676704407, "learning_rate": 1.1118426220885214e-07, "loss": 0.0074, "step": 293950 }, { "epoch": 187.35500318674315, "grad_norm": 0.08103132992982864, "learning_rate": 1.1081721102773612e-07, "loss": 0.0041, "step": 293960 }, { "epoch": 187.36137667304016, "grad_norm": 0.924004852771759, "learning_rate": 1.1045076604729865e-07, "loss": 0.0036, "step": 293970 }, { "epoch": 187.36775015933716, "grad_norm": 0.1294523924589157, "learning_rate": 1.1008492727198061e-07, "loss": 0.0084, "step": 293980 }, { "epoch": 187.37412364563417, "grad_norm": 0.45787593722343445, "learning_rate": 1.0971969470623955e-07, "loss": 0.0031, "step": 293990 }, { "epoch": 187.38049713193118, "grad_norm": 0.10188322514295578, "learning_rate": 1.0935506835450527e-07, "loss": 0.0027, "step": 294000 }, { "epoch": 187.38687061822816, "grad_norm": 0.0715041384100914, "learning_rate": 1.0899104822121309e-07, "loss": 0.0038, "step": 294010 }, { "epoch": 187.39324410452517, "grad_norm": 0.3668554723262787, "learning_rate": 1.0862763431078172e-07, "loss": 0.0088, "step": 294020 }, { "epoch": 187.39961759082217, "grad_norm": 0.2933976352214813, "learning_rate": 1.0826482662763537e-07, "loss": 0.004, "step": 294030 }, { "epoch": 187.40599107711918, "grad_norm": 0.23425757884979248, "learning_rate": 1.0790262517617056e-07, "loss": 0.0046, "step": 294040 }, { "epoch": 187.4123645634162, "grad_norm": 0.11445076763629913, "learning_rate": 1.075410299608004e-07, "loss": 0.0055, "step": 294050 }, { "epoch": 187.4187380497132, "grad_norm": 0.10531330853700638, "learning_rate": 1.0718004098591028e-07, "loss": 0.0149, "step": 294060 }, { "epoch": 187.4251115360102, "grad_norm": 0.11015193909406662, "learning_rate": 1.0681965825589113e-07, "loss": 0.009, "step": 294070 }, { "epoch": 187.4314850223072, "grad_norm": 0.184393972158432, "learning_rate": 1.0645988177511724e-07, "loss": 0.007, "step": 294080 }, { "epoch": 187.4378585086042, "grad_norm": 0.6601685881614685, "learning_rate": 1.0610071154796286e-07, "loss": 0.0059, "step": 294090 }, { "epoch": 187.4442319949012, "grad_norm": 0.09550119191408157, "learning_rate": 1.0574214757879675e-07, "loss": 0.0062, "step": 294100 }, { "epoch": 187.4506054811982, "grad_norm": 0.10961782932281494, "learning_rate": 1.0538418987197096e-07, "loss": 0.0038, "step": 294110 }, { "epoch": 187.45697896749522, "grad_norm": 0.10626889765262604, "learning_rate": 1.0502683843183203e-07, "loss": 0.0042, "step": 294120 }, { "epoch": 187.46335245379223, "grad_norm": 0.10246331989765167, "learning_rate": 1.0467009326272648e-07, "loss": 0.0042, "step": 294130 }, { "epoch": 187.46972594008923, "grad_norm": 0.10563123226165771, "learning_rate": 1.0431395436898972e-07, "loss": 0.0032, "step": 294140 }, { "epoch": 187.47609942638624, "grad_norm": 0.1623450219631195, "learning_rate": 1.0395842175494608e-07, "loss": 0.0032, "step": 294150 }, { "epoch": 187.48247291268325, "grad_norm": 0.04151136800646782, "learning_rate": 1.0360349542491988e-07, "loss": 0.0073, "step": 294160 }, { "epoch": 187.48884639898023, "grad_norm": 0.18196795880794525, "learning_rate": 1.0324917538321876e-07, "loss": 0.0057, "step": 294170 }, { "epoch": 187.49521988527724, "grad_norm": 0.06662037968635559, "learning_rate": 1.0289546163415042e-07, "loss": 0.0035, "step": 294180 }, { "epoch": 187.50159337157424, "grad_norm": 0.13203613460063934, "learning_rate": 1.0254235418201697e-07, "loss": 0.0046, "step": 294190 }, { "epoch": 187.50796685787125, "grad_norm": 0.17916704714298248, "learning_rate": 1.021898530310983e-07, "loss": 0.0061, "step": 294200 }, { "epoch": 187.51434034416826, "grad_norm": 0.09490329027175903, "learning_rate": 1.0183795818569098e-07, "loss": 0.0025, "step": 294210 }, { "epoch": 187.52071383046527, "grad_norm": 0.42184415459632874, "learning_rate": 1.0148666965005826e-07, "loss": 0.0107, "step": 294220 }, { "epoch": 187.52708731676228, "grad_norm": 0.07832296937704086, "learning_rate": 1.0113598742848007e-07, "loss": 0.0137, "step": 294230 }, { "epoch": 187.53346080305928, "grad_norm": 0.20923128724098206, "learning_rate": 1.0078591152520856e-07, "loss": 0.003, "step": 294240 }, { "epoch": 187.53983428935626, "grad_norm": 0.2776281237602234, "learning_rate": 1.0043644194450141e-07, "loss": 0.0044, "step": 294250 }, { "epoch": 187.54620777565327, "grad_norm": 0.2730754315853119, "learning_rate": 1.0008757869060525e-07, "loss": 0.0066, "step": 294260 }, { "epoch": 187.55258126195028, "grad_norm": 0.2031715214252472, "learning_rate": 9.973932176775558e-08, "loss": 0.0074, "step": 294270 }, { "epoch": 187.5589547482473, "grad_norm": 0.35826635360717773, "learning_rate": 9.939167118019344e-08, "loss": 0.0095, "step": 294280 }, { "epoch": 187.5653282345443, "grad_norm": 0.10653795301914215, "learning_rate": 9.904462693212657e-08, "loss": 0.0027, "step": 294290 }, { "epoch": 187.5717017208413, "grad_norm": 0.06593932211399078, "learning_rate": 9.869818902779049e-08, "loss": 0.0027, "step": 294300 }, { "epoch": 187.5780752071383, "grad_norm": 0.20707176625728607, "learning_rate": 9.835235747138183e-08, "loss": 0.0072, "step": 294310 }, { "epoch": 187.58444869343532, "grad_norm": 0.17939040064811707, "learning_rate": 9.800713226710833e-08, "loss": 0.0075, "step": 294320 }, { "epoch": 187.5908221797323, "grad_norm": 0.043705470860004425, "learning_rate": 9.766251341916111e-08, "loss": 0.0036, "step": 294330 }, { "epoch": 187.5971956660293, "grad_norm": 0.209797203540802, "learning_rate": 9.731850093173122e-08, "loss": 0.0041, "step": 294340 }, { "epoch": 187.60356915232632, "grad_norm": 0.12744607031345367, "learning_rate": 9.69750948089987e-08, "loss": 0.0186, "step": 294350 }, { "epoch": 187.60994263862332, "grad_norm": 0.14781217277050018, "learning_rate": 9.663229505513794e-08, "loss": 0.0045, "step": 294360 }, { "epoch": 187.61631612492033, "grad_norm": 0.3810962736606598, "learning_rate": 9.629010167430675e-08, "loss": 0.0076, "step": 294370 }, { "epoch": 187.62268961121734, "grad_norm": 0.4313236176967621, "learning_rate": 9.594851467066845e-08, "loss": 0.0055, "step": 294380 }, { "epoch": 187.62906309751435, "grad_norm": 0.1475953608751297, "learning_rate": 9.56075340483753e-08, "loss": 0.0048, "step": 294390 }, { "epoch": 187.63543658381136, "grad_norm": 0.1762630045413971, "learning_rate": 9.526715981156287e-08, "loss": 0.0043, "step": 294400 }, { "epoch": 187.64181007010836, "grad_norm": 0.1733626425266266, "learning_rate": 9.492739196437784e-08, "loss": 0.0043, "step": 294410 }, { "epoch": 187.64818355640534, "grad_norm": 0.17791426181793213, "learning_rate": 9.458823051093913e-08, "loss": 0.0024, "step": 294420 }, { "epoch": 187.65455704270235, "grad_norm": 0.0828753113746643, "learning_rate": 9.424967545537677e-08, "loss": 0.0035, "step": 294430 }, { "epoch": 187.66093052899936, "grad_norm": 0.055254168808460236, "learning_rate": 9.391172680179305e-08, "loss": 0.0036, "step": 294440 }, { "epoch": 187.66730401529637, "grad_norm": 0.12010600417852402, "learning_rate": 9.357438455430689e-08, "loss": 0.0033, "step": 294450 }, { "epoch": 187.67367750159337, "grad_norm": 0.055947814136743546, "learning_rate": 9.323764871700946e-08, "loss": 0.0071, "step": 294460 }, { "epoch": 187.68005098789038, "grad_norm": 0.115305095911026, "learning_rate": 9.290151929399194e-08, "loss": 0.0032, "step": 294470 }, { "epoch": 187.6864244741874, "grad_norm": 0.10760972648859024, "learning_rate": 9.256599628934548e-08, "loss": 0.0039, "step": 294480 }, { "epoch": 187.6927979604844, "grad_norm": 0.09039295464754105, "learning_rate": 9.223107970713907e-08, "loss": 0.0041, "step": 294490 }, { "epoch": 187.69917144678138, "grad_norm": 0.32530319690704346, "learning_rate": 9.189676955144721e-08, "loss": 0.008, "step": 294500 }, { "epoch": 187.7055449330784, "grad_norm": 0.20005853474140167, "learning_rate": 9.156306582633334e-08, "loss": 0.0055, "step": 294510 }, { "epoch": 187.7119184193754, "grad_norm": 0.06862766295671463, "learning_rate": 9.12299685358442e-08, "loss": 0.0034, "step": 294520 }, { "epoch": 187.7182919056724, "grad_norm": 0.1022002100944519, "learning_rate": 9.089747768403766e-08, "loss": 0.0275, "step": 294530 }, { "epoch": 187.7246653919694, "grad_norm": 0.34136223793029785, "learning_rate": 9.056559327494385e-08, "loss": 0.0067, "step": 294540 }, { "epoch": 187.73103887826642, "grad_norm": 0.4585473835468292, "learning_rate": 9.023431531260396e-08, "loss": 0.0058, "step": 294550 }, { "epoch": 187.73741236456343, "grad_norm": 0.21391963958740234, "learning_rate": 8.990364380104255e-08, "loss": 0.0029, "step": 294560 }, { "epoch": 187.74378585086043, "grad_norm": 0.06573081016540527, "learning_rate": 8.957357874426753e-08, "loss": 0.0045, "step": 294570 }, { "epoch": 187.7501593371574, "grad_norm": 0.46628549695014954, "learning_rate": 8.92441201462979e-08, "loss": 0.0061, "step": 294580 }, { "epoch": 187.75653282345442, "grad_norm": 0.25217658281326294, "learning_rate": 8.891526801114159e-08, "loss": 0.0042, "step": 294590 }, { "epoch": 187.76290630975143, "grad_norm": 0.07519945502281189, "learning_rate": 8.858702234278427e-08, "loss": 0.0056, "step": 294600 }, { "epoch": 187.76927979604844, "grad_norm": 0.17305169999599457, "learning_rate": 8.82593831452172e-08, "loss": 0.0042, "step": 294610 }, { "epoch": 187.77565328234544, "grad_norm": 0.1263207346200943, "learning_rate": 8.793235042242609e-08, "loss": 0.0036, "step": 294620 }, { "epoch": 187.78202676864245, "grad_norm": 0.3185008764266968, "learning_rate": 8.760592417837999e-08, "loss": 0.0087, "step": 294630 }, { "epoch": 187.78840025493946, "grad_norm": 0.115807443857193, "learning_rate": 8.728010441705347e-08, "loss": 0.0028, "step": 294640 }, { "epoch": 187.79477374123647, "grad_norm": 0.27276426553726196, "learning_rate": 8.695489114239341e-08, "loss": 0.0037, "step": 294650 }, { "epoch": 187.80114722753345, "grad_norm": 0.09784265607595444, "learning_rate": 8.663028435835774e-08, "loss": 0.0024, "step": 294660 }, { "epoch": 187.80752071383046, "grad_norm": 0.1445794552564621, "learning_rate": 8.630628406888775e-08, "loss": 0.0032, "step": 294670 }, { "epoch": 187.81389420012746, "grad_norm": 0.447489857673645, "learning_rate": 8.598289027792473e-08, "loss": 0.0054, "step": 294680 }, { "epoch": 187.82026768642447, "grad_norm": 0.07989800721406937, "learning_rate": 8.566010298939886e-08, "loss": 0.0043, "step": 294690 }, { "epoch": 187.82664117272148, "grad_norm": 0.09422675520181656, "learning_rate": 8.53379222072237e-08, "loss": 0.0031, "step": 294700 }, { "epoch": 187.8330146590185, "grad_norm": 0.4925830364227295, "learning_rate": 8.501634793532942e-08, "loss": 0.0076, "step": 294710 }, { "epoch": 187.8393881453155, "grad_norm": 0.09039095789194107, "learning_rate": 8.469538017760736e-08, "loss": 0.0041, "step": 294720 }, { "epoch": 187.8457616316125, "grad_norm": 0.13551568984985352, "learning_rate": 8.43750189379655e-08, "loss": 0.026, "step": 294730 }, { "epoch": 187.85213511790948, "grad_norm": 0.11405115574598312, "learning_rate": 8.405526422029519e-08, "loss": 0.0036, "step": 294740 }, { "epoch": 187.8585086042065, "grad_norm": 0.11608605831861496, "learning_rate": 8.37361160284822e-08, "loss": 0.0027, "step": 294750 }, { "epoch": 187.8648820905035, "grad_norm": 0.09264861792325974, "learning_rate": 8.341757436640119e-08, "loss": 0.003, "step": 294760 }, { "epoch": 187.8712555768005, "grad_norm": 0.11290473490953445, "learning_rate": 8.309963923792685e-08, "loss": 0.0021, "step": 294770 }, { "epoch": 187.87762906309752, "grad_norm": 0.17770390212535858, "learning_rate": 8.278231064692277e-08, "loss": 0.0023, "step": 294780 }, { "epoch": 187.88400254939452, "grad_norm": 0.0710185095667839, "learning_rate": 8.246558859724696e-08, "loss": 0.0061, "step": 294790 }, { "epoch": 187.89037603569153, "grad_norm": 0.07758022099733353, "learning_rate": 8.214947309274079e-08, "loss": 0.0026, "step": 294800 }, { "epoch": 187.89674952198854, "grad_norm": 0.15122823417186737, "learning_rate": 8.183396413724564e-08, "loss": 0.0032, "step": 294810 }, { "epoch": 187.90312300828552, "grad_norm": 0.0907803550362587, "learning_rate": 8.151906173460289e-08, "loss": 0.0043, "step": 294820 }, { "epoch": 187.90949649458253, "grad_norm": 0.28325414657592773, "learning_rate": 8.120476588863169e-08, "loss": 0.0049, "step": 294830 }, { "epoch": 187.91586998087953, "grad_norm": 0.1871730387210846, "learning_rate": 8.089107660315676e-08, "loss": 0.0026, "step": 294840 }, { "epoch": 187.92224346717654, "grad_norm": 0.07337921112775803, "learning_rate": 8.057799388198062e-08, "loss": 0.0085, "step": 294850 }, { "epoch": 187.92861695347355, "grad_norm": 0.13477280735969543, "learning_rate": 8.026551772891688e-08, "loss": 0.0036, "step": 294860 }, { "epoch": 187.93499043977056, "grad_norm": 0.10975028574466705, "learning_rate": 7.995364814776251e-08, "loss": 0.0056, "step": 294870 }, { "epoch": 187.94136392606757, "grad_norm": 0.06922036409378052, "learning_rate": 7.964238514230337e-08, "loss": 0.0131, "step": 294880 }, { "epoch": 187.94773741236457, "grad_norm": 0.09341148287057877, "learning_rate": 7.933172871631978e-08, "loss": 0.0045, "step": 294890 }, { "epoch": 187.95411089866155, "grad_norm": 0.1459399312734604, "learning_rate": 7.902167887358647e-08, "loss": 0.0051, "step": 294900 }, { "epoch": 187.96048438495856, "grad_norm": 0.08532652258872986, "learning_rate": 7.871223561787266e-08, "loss": 0.0121, "step": 294910 }, { "epoch": 187.96685787125557, "grad_norm": 0.15932036936283112, "learning_rate": 7.840339895294202e-08, "loss": 0.0059, "step": 294920 }, { "epoch": 187.97323135755258, "grad_norm": 0.4135689437389374, "learning_rate": 7.809516888254709e-08, "loss": 0.0083, "step": 294930 }, { "epoch": 187.9796048438496, "grad_norm": 0.056321997195482254, "learning_rate": 7.778754541042376e-08, "loss": 0.0031, "step": 294940 }, { "epoch": 187.9859783301466, "grad_norm": 0.07440821826457977, "learning_rate": 7.748052854031906e-08, "loss": 0.0053, "step": 294950 }, { "epoch": 187.9923518164436, "grad_norm": 0.19221241772174835, "learning_rate": 7.717411827595778e-08, "loss": 0.0041, "step": 294960 }, { "epoch": 187.9987253027406, "grad_norm": 0.26603591442108154, "learning_rate": 7.686831462107024e-08, "loss": 0.0069, "step": 294970 }, { "epoch": 188.00509878903762, "grad_norm": 0.08058816194534302, "learning_rate": 7.656311757936463e-08, "loss": 0.0035, "step": 294980 }, { "epoch": 188.0114722753346, "grad_norm": 0.4287916421890259, "learning_rate": 7.625852715455462e-08, "loss": 0.0089, "step": 294990 }, { "epoch": 188.0178457616316, "grad_norm": 0.09966188669204712, "learning_rate": 7.595454335033725e-08, "loss": 0.0043, "step": 295000 }, { "epoch": 188.0242192479286, "grad_norm": 0.08318217098712921, "learning_rate": 7.565116617040957e-08, "loss": 0.014, "step": 295010 }, { "epoch": 188.03059273422562, "grad_norm": 0.37601685523986816, "learning_rate": 7.534839561845197e-08, "loss": 0.0038, "step": 295020 }, { "epoch": 188.03696622052263, "grad_norm": 0.5108857750892639, "learning_rate": 7.504623169815595e-08, "loss": 0.0052, "step": 295030 }, { "epoch": 188.04333970681964, "grad_norm": 0.1377566158771515, "learning_rate": 7.474467441317967e-08, "loss": 0.0035, "step": 295040 }, { "epoch": 188.04971319311665, "grad_norm": 0.12987074255943298, "learning_rate": 7.444372376719799e-08, "loss": 0.0068, "step": 295050 }, { "epoch": 188.05608667941365, "grad_norm": 0.13519679009914398, "learning_rate": 7.4143379763858e-08, "loss": 0.0072, "step": 295060 }, { "epoch": 188.06246016571063, "grad_norm": 0.43723395466804504, "learning_rate": 7.384364240681229e-08, "loss": 0.0068, "step": 295070 }, { "epoch": 188.06883365200764, "grad_norm": 0.10904563218355179, "learning_rate": 7.354451169970799e-08, "loss": 0.0057, "step": 295080 }, { "epoch": 188.07520713830465, "grad_norm": 0.24413445591926575, "learning_rate": 7.324598764617552e-08, "loss": 0.0063, "step": 295090 }, { "epoch": 188.08158062460166, "grad_norm": 0.22447265684604645, "learning_rate": 7.294807024984529e-08, "loss": 0.003, "step": 295100 }, { "epoch": 188.08795411089866, "grad_norm": 0.22095820307731628, "learning_rate": 7.265075951433109e-08, "loss": 0.0047, "step": 295110 }, { "epoch": 188.09432759719567, "grad_norm": 0.7751568555831909, "learning_rate": 7.235405544325225e-08, "loss": 0.0109, "step": 295120 }, { "epoch": 188.10070108349268, "grad_norm": 0.23575827479362488, "learning_rate": 7.205795804021142e-08, "loss": 0.0036, "step": 295130 }, { "epoch": 188.1070745697897, "grad_norm": 0.08099804818630219, "learning_rate": 7.176246730880021e-08, "loss": 0.0045, "step": 295140 }, { "epoch": 188.11344805608667, "grad_norm": 0.09268167614936829, "learning_rate": 7.146758325262126e-08, "loss": 0.0049, "step": 295150 }, { "epoch": 188.11982154238368, "grad_norm": 0.17445628345012665, "learning_rate": 7.117330587524951e-08, "loss": 0.0049, "step": 295160 }, { "epoch": 188.12619502868068, "grad_norm": 0.08274617791175842, "learning_rate": 7.087963518026542e-08, "loss": 0.0091, "step": 295170 }, { "epoch": 188.1325685149777, "grad_norm": 0.33495014905929565, "learning_rate": 7.05865711712328e-08, "loss": 0.0065, "step": 295180 }, { "epoch": 188.1389420012747, "grad_norm": 0.2296888828277588, "learning_rate": 7.029411385170993e-08, "loss": 0.0088, "step": 295190 }, { "epoch": 188.1453154875717, "grad_norm": 0.18068398535251617, "learning_rate": 7.00022632252606e-08, "loss": 0.0079, "step": 295200 }, { "epoch": 188.15168897386872, "grad_norm": 0.10087314993143082, "learning_rate": 6.971101929542645e-08, "loss": 0.0061, "step": 295210 }, { "epoch": 188.15806246016572, "grad_norm": 0.12287931144237518, "learning_rate": 6.942038206574353e-08, "loss": 0.0047, "step": 295220 }, { "epoch": 188.1644359464627, "grad_norm": 0.24633905291557312, "learning_rate": 6.913035153974234e-08, "loss": 0.0036, "step": 295230 }, { "epoch": 188.1708094327597, "grad_norm": 0.187620609998703, "learning_rate": 6.88409277209534e-08, "loss": 0.0025, "step": 295240 }, { "epoch": 188.17718291905672, "grad_norm": 0.06881016492843628, "learning_rate": 6.855211061289057e-08, "loss": 0.0221, "step": 295250 }, { "epoch": 188.18355640535373, "grad_norm": 0.26699116826057434, "learning_rate": 6.826390021905659e-08, "loss": 0.0046, "step": 295260 }, { "epoch": 188.18992989165073, "grad_norm": 0.13279935717582703, "learning_rate": 6.797629654296534e-08, "loss": 0.0044, "step": 295270 }, { "epoch": 188.19630337794774, "grad_norm": 0.1487402319908142, "learning_rate": 6.768929958810288e-08, "loss": 0.0051, "step": 295280 }, { "epoch": 188.20267686424475, "grad_norm": 0.12523776292800903, "learning_rate": 6.74029093579609e-08, "loss": 0.0043, "step": 295290 }, { "epoch": 188.20905035054176, "grad_norm": 0.3794594407081604, "learning_rate": 6.711712585601992e-08, "loss": 0.0085, "step": 295300 }, { "epoch": 188.21542383683874, "grad_norm": 0.17093224823474884, "learning_rate": 6.68319490857494e-08, "loss": 0.0075, "step": 295310 }, { "epoch": 188.22179732313575, "grad_norm": 0.4699123501777649, "learning_rate": 6.654737905061325e-08, "loss": 0.0058, "step": 295320 }, { "epoch": 188.22817080943275, "grad_norm": 0.1589963734149933, "learning_rate": 6.62634157540698e-08, "loss": 0.0033, "step": 295330 }, { "epoch": 188.23454429572976, "grad_norm": 0.18165434896945953, "learning_rate": 6.598005919957739e-08, "loss": 0.0038, "step": 295340 }, { "epoch": 188.24091778202677, "grad_norm": 0.04870198294520378, "learning_rate": 6.569730939056662e-08, "loss": 0.0068, "step": 295350 }, { "epoch": 188.24729126832378, "grad_norm": 0.2547279894351959, "learning_rate": 6.541516633048473e-08, "loss": 0.0055, "step": 295360 }, { "epoch": 188.2536647546208, "grad_norm": 0.12706395983695984, "learning_rate": 6.513363002275119e-08, "loss": 0.0038, "step": 295370 }, { "epoch": 188.2600382409178, "grad_norm": 0.08851028233766556, "learning_rate": 6.485270047079107e-08, "loss": 0.0037, "step": 295380 }, { "epoch": 188.26641172721477, "grad_norm": 0.4315407872200012, "learning_rate": 6.457237767801826e-08, "loss": 0.0086, "step": 295390 }, { "epoch": 188.27278521351178, "grad_norm": 0.12141354382038116, "learning_rate": 6.429266164783565e-08, "loss": 0.0058, "step": 295400 }, { "epoch": 188.2791586998088, "grad_norm": 0.10914934426546097, "learning_rate": 6.401355238364049e-08, "loss": 0.004, "step": 295410 }, { "epoch": 188.2855321861058, "grad_norm": 0.11121917515993118, "learning_rate": 6.373504988883561e-08, "loss": 0.0055, "step": 295420 }, { "epoch": 188.2919056724028, "grad_norm": 0.06309094280004501, "learning_rate": 6.34571541667961e-08, "loss": 0.0081, "step": 295430 }, { "epoch": 188.2982791586998, "grad_norm": 0.11542899906635284, "learning_rate": 6.317986522089702e-08, "loss": 0.003, "step": 295440 }, { "epoch": 188.30465264499682, "grad_norm": 0.37451672554016113, "learning_rate": 6.290318305451349e-08, "loss": 0.0052, "step": 295450 }, { "epoch": 188.31102613129383, "grad_norm": 0.3461536765098572, "learning_rate": 6.262710767100388e-08, "loss": 0.0029, "step": 295460 }, { "epoch": 188.3173996175908, "grad_norm": 0.19895809888839722, "learning_rate": 6.23516390737211e-08, "loss": 0.0079, "step": 295470 }, { "epoch": 188.32377310388782, "grad_norm": 0.13327553868293762, "learning_rate": 6.207677726601802e-08, "loss": 0.0032, "step": 295480 }, { "epoch": 188.33014659018482, "grad_norm": 0.10693284124135971, "learning_rate": 6.180252225123085e-08, "loss": 0.0035, "step": 295490 }, { "epoch": 188.33652007648183, "grad_norm": 0.31649723649024963, "learning_rate": 6.152887403269026e-08, "loss": 0.0103, "step": 295500 }, { "epoch": 188.34289356277884, "grad_norm": 0.04787103831768036, "learning_rate": 6.125583261372691e-08, "loss": 0.004, "step": 295510 }, { "epoch": 188.34926704907585, "grad_norm": 0.11611207574605942, "learning_rate": 6.098339799765484e-08, "loss": 0.006, "step": 295520 }, { "epoch": 188.35564053537286, "grad_norm": 0.10738407075405121, "learning_rate": 6.07115701877825e-08, "loss": 0.0034, "step": 295530 }, { "epoch": 188.36201402166986, "grad_norm": 0.11662016808986664, "learning_rate": 6.044034918741836e-08, "loss": 0.0018, "step": 295540 }, { "epoch": 188.36838750796684, "grad_norm": 0.27026739716529846, "learning_rate": 6.016973499985424e-08, "loss": 0.0098, "step": 295550 }, { "epoch": 188.37476099426385, "grad_norm": 0.5926682949066162, "learning_rate": 5.989972762837637e-08, "loss": 0.0076, "step": 295560 }, { "epoch": 188.38113448056086, "grad_norm": 0.08358152955770493, "learning_rate": 5.963032707626548e-08, "loss": 0.0099, "step": 295570 }, { "epoch": 188.38750796685787, "grad_norm": 0.06280877441167831, "learning_rate": 5.936153334680228e-08, "loss": 0.005, "step": 295580 }, { "epoch": 188.39388145315488, "grad_norm": 0.14708372950553894, "learning_rate": 5.909334644324527e-08, "loss": 0.0036, "step": 295590 }, { "epoch": 188.40025493945188, "grad_norm": 0.17525508999824524, "learning_rate": 5.88257663688585e-08, "loss": 0.0052, "step": 295600 }, { "epoch": 188.4066284257489, "grad_norm": 0.2126612365245819, "learning_rate": 5.8558793126889386e-08, "loss": 0.0058, "step": 295610 }, { "epoch": 188.4130019120459, "grad_norm": 0.09304589033126831, "learning_rate": 5.829242672058532e-08, "loss": 0.0033, "step": 295620 }, { "epoch": 188.4193753983429, "grad_norm": 0.06024056673049927, "learning_rate": 5.802666715318261e-08, "loss": 0.012, "step": 295630 }, { "epoch": 188.4257488846399, "grad_norm": 0.20656679570674896, "learning_rate": 5.776151442790645e-08, "loss": 0.004, "step": 295640 }, { "epoch": 188.4321223709369, "grad_norm": 0.540976881980896, "learning_rate": 5.7496968547982034e-08, "loss": 0.0033, "step": 295650 }, { "epoch": 188.4384958572339, "grad_norm": 0.06529674679040909, "learning_rate": 5.723302951662346e-08, "loss": 0.006, "step": 295660 }, { "epoch": 188.4448693435309, "grad_norm": 0.12150214612483978, "learning_rate": 5.696969733703372e-08, "loss": 0.0039, "step": 295670 }, { "epoch": 188.45124282982792, "grad_norm": 0.3586105406284332, "learning_rate": 5.6706972012421365e-08, "loss": 0.0068, "step": 295680 }, { "epoch": 188.45761631612493, "grad_norm": 0.13047437369823456, "learning_rate": 5.644485354596718e-08, "loss": 0.0071, "step": 295690 }, { "epoch": 188.46398980242193, "grad_norm": 0.13749222457408905, "learning_rate": 5.618334194086861e-08, "loss": 0.0039, "step": 295700 }, { "epoch": 188.47036328871894, "grad_norm": 0.08114586770534515, "learning_rate": 5.592243720029533e-08, "loss": 0.0034, "step": 295710 }, { "epoch": 188.47673677501592, "grad_norm": 0.14725512266159058, "learning_rate": 5.566213932741704e-08, "loss": 0.004, "step": 295720 }, { "epoch": 188.48311026131293, "grad_norm": 0.06351584941148758, "learning_rate": 5.540244832539787e-08, "loss": 0.0035, "step": 295730 }, { "epoch": 188.48948374760994, "grad_norm": 0.44636431336402893, "learning_rate": 5.514336419740196e-08, "loss": 0.0118, "step": 295740 }, { "epoch": 188.49585723390695, "grad_norm": 0.20406177639961243, "learning_rate": 5.4884886946565686e-08, "loss": 0.0036, "step": 295750 }, { "epoch": 188.50223072020395, "grad_norm": 0.07975152879953384, "learning_rate": 5.462701657603098e-08, "loss": 0.0029, "step": 295760 }, { "epoch": 188.50860420650096, "grad_norm": 0.2631688117980957, "learning_rate": 5.436975308893977e-08, "loss": 0.0049, "step": 295770 }, { "epoch": 188.51497769279797, "grad_norm": 0.4030792713165283, "learning_rate": 5.411309648840623e-08, "loss": 0.0067, "step": 295780 }, { "epoch": 188.52135117909498, "grad_norm": 0.24631871283054352, "learning_rate": 5.38570467775612e-08, "loss": 0.0042, "step": 295790 }, { "epoch": 188.52772466539196, "grad_norm": 0.12496693432331085, "learning_rate": 5.3601603959507754e-08, "loss": 0.0033, "step": 295800 }, { "epoch": 188.53409815168897, "grad_norm": 0.09657290577888489, "learning_rate": 5.334676803734895e-08, "loss": 0.0031, "step": 295810 }, { "epoch": 188.54047163798597, "grad_norm": 0.17515216767787933, "learning_rate": 5.309253901418787e-08, "loss": 0.0054, "step": 295820 }, { "epoch": 188.54684512428298, "grad_norm": 0.10204564779996872, "learning_rate": 5.283891689311093e-08, "loss": 0.0057, "step": 295830 }, { "epoch": 188.55321861058, "grad_norm": 0.07632917165756226, "learning_rate": 5.258590167719901e-08, "loss": 0.0062, "step": 295840 }, { "epoch": 188.559592096877, "grad_norm": 0.24164707958698273, "learning_rate": 5.233349336952742e-08, "loss": 0.0097, "step": 295850 }, { "epoch": 188.565965583174, "grad_norm": 0.48255032300949097, "learning_rate": 5.2081691973160375e-08, "loss": 0.0043, "step": 295860 }, { "epoch": 188.572339069471, "grad_norm": 0.3553536832332611, "learning_rate": 5.183049749115654e-08, "loss": 0.0049, "step": 295870 }, { "epoch": 188.578712555768, "grad_norm": 0.1604592204093933, "learning_rate": 5.157990992657458e-08, "loss": 0.0025, "step": 295880 }, { "epoch": 188.585086042065, "grad_norm": 0.24173489212989807, "learning_rate": 5.13299292824565e-08, "loss": 0.004, "step": 295890 }, { "epoch": 188.591459528362, "grad_norm": 0.4132849872112274, "learning_rate": 5.1080555561838773e-08, "loss": 0.0038, "step": 295900 }, { "epoch": 188.59783301465902, "grad_norm": 0.1932496279478073, "learning_rate": 5.0831788767746746e-08, "loss": 0.0038, "step": 295910 }, { "epoch": 188.60420650095602, "grad_norm": 0.183511883020401, "learning_rate": 5.0583628903216886e-08, "loss": 0.0079, "step": 295920 }, { "epoch": 188.61057998725303, "grad_norm": 0.09754496812820435, "learning_rate": 5.033607597124679e-08, "loss": 0.0018, "step": 295930 }, { "epoch": 188.61695347355004, "grad_norm": 0.21658620238304138, "learning_rate": 5.008912997486182e-08, "loss": 0.0032, "step": 295940 }, { "epoch": 188.62332695984705, "grad_norm": 0.08972279727458954, "learning_rate": 4.984279091704846e-08, "loss": 0.0041, "step": 295950 }, { "epoch": 188.62970044614403, "grad_norm": 0.11722669005393982, "learning_rate": 4.959705880080989e-08, "loss": 0.0055, "step": 295960 }, { "epoch": 188.63607393244104, "grad_norm": 0.09642281383275986, "learning_rate": 4.9351933629127044e-08, "loss": 0.0038, "step": 295970 }, { "epoch": 188.64244741873804, "grad_norm": 0.27795320749282837, "learning_rate": 4.910741540498087e-08, "loss": 0.0032, "step": 295980 }, { "epoch": 188.64882090503505, "grad_norm": 0.2522614896297455, "learning_rate": 4.886350413133567e-08, "loss": 0.0039, "step": 295990 }, { "epoch": 188.65519439133206, "grad_norm": 0.12799060344696045, "learning_rate": 4.8620199811166834e-08, "loss": 0.0048, "step": 296000 }, { "epoch": 188.66156787762907, "grad_norm": 0.10515277087688446, "learning_rate": 4.8377502447422005e-08, "loss": 0.0031, "step": 296010 }, { "epoch": 188.66794136392608, "grad_norm": 0.10882411152124405, "learning_rate": 4.8135412043054386e-08, "loss": 0.0057, "step": 296020 }, { "epoch": 188.67431485022308, "grad_norm": 0.1288839876651764, "learning_rate": 4.7893928601000504e-08, "loss": 0.004, "step": 296030 }, { "epoch": 188.68068833652006, "grad_norm": 0.34715911746025085, "learning_rate": 4.765305212419691e-08, "loss": 0.0048, "step": 296040 }, { "epoch": 188.68706182281707, "grad_norm": 0.14633290469646454, "learning_rate": 4.741278261557458e-08, "loss": 0.0079, "step": 296050 }, { "epoch": 188.69343530911408, "grad_norm": 0.31823664903640747, "learning_rate": 4.717312007805341e-08, "loss": 0.0155, "step": 296060 }, { "epoch": 188.6998087954111, "grad_norm": 0.20926980674266815, "learning_rate": 4.693406451453664e-08, "loss": 0.0036, "step": 296070 }, { "epoch": 188.7061822817081, "grad_norm": 0.19819225370883942, "learning_rate": 4.6695615927933034e-08, "loss": 0.0076, "step": 296080 }, { "epoch": 188.7125557680051, "grad_norm": 0.09732785075902939, "learning_rate": 4.6457774321145844e-08, "loss": 0.0042, "step": 296090 }, { "epoch": 188.7189292543021, "grad_norm": 0.06799308210611343, "learning_rate": 4.622053969706164e-08, "loss": 0.0036, "step": 296100 }, { "epoch": 188.72530274059912, "grad_norm": 0.1708095371723175, "learning_rate": 4.59839120585559e-08, "loss": 0.0075, "step": 296110 }, { "epoch": 188.7316762268961, "grad_norm": 0.08786923438310623, "learning_rate": 4.574789140851521e-08, "loss": 0.0044, "step": 296120 }, { "epoch": 188.7380497131931, "grad_norm": 0.3216920793056488, "learning_rate": 4.551247774980394e-08, "loss": 0.0086, "step": 296130 }, { "epoch": 188.74442319949011, "grad_norm": 0.12989778816699982, "learning_rate": 4.527767108527536e-08, "loss": 0.0049, "step": 296140 }, { "epoch": 188.75079668578712, "grad_norm": 0.17835195362567902, "learning_rate": 4.5043471417793856e-08, "loss": 0.0045, "step": 296150 }, { "epoch": 188.75717017208413, "grad_norm": 0.06947202235460281, "learning_rate": 4.480987875019604e-08, "loss": 0.0026, "step": 296160 }, { "epoch": 188.76354365838114, "grad_norm": 0.15983383357524872, "learning_rate": 4.457689308532409e-08, "loss": 0.0046, "step": 296170 }, { "epoch": 188.76991714467815, "grad_norm": 0.21221354603767395, "learning_rate": 4.434451442600907e-08, "loss": 0.003, "step": 296180 }, { "epoch": 188.77629063097515, "grad_norm": 0.06803441792726517, "learning_rate": 4.411274277507094e-08, "loss": 0.0046, "step": 296190 }, { "epoch": 188.78266411727213, "grad_norm": 0.32324761152267456, "learning_rate": 4.388157813533522e-08, "loss": 0.0035, "step": 296200 }, { "epoch": 188.78903760356914, "grad_norm": 0.7037439942359924, "learning_rate": 4.365102050959968e-08, "loss": 0.006, "step": 296210 }, { "epoch": 188.79541108986615, "grad_norm": 0.12136375159025192, "learning_rate": 4.342106990067318e-08, "loss": 0.0123, "step": 296220 }, { "epoch": 188.80178457616316, "grad_norm": 0.34366655349731445, "learning_rate": 4.3191726311342384e-08, "loss": 0.003, "step": 296230 }, { "epoch": 188.80815806246017, "grad_norm": 0.14055880904197693, "learning_rate": 4.2962989744405044e-08, "loss": 0.0071, "step": 296240 }, { "epoch": 188.81453154875717, "grad_norm": 0.07311122119426727, "learning_rate": 4.273486020263673e-08, "loss": 0.0034, "step": 296250 }, { "epoch": 188.82090503505418, "grad_norm": 0.12178167700767517, "learning_rate": 4.250733768880188e-08, "loss": 0.0029, "step": 296260 }, { "epoch": 188.8272785213512, "grad_norm": 0.06960569322109222, "learning_rate": 4.228042220567607e-08, "loss": 0.0068, "step": 296270 }, { "epoch": 188.8336520076482, "grad_norm": 0.12163999676704407, "learning_rate": 4.205411375600154e-08, "loss": 0.0055, "step": 296280 }, { "epoch": 188.84002549394518, "grad_norm": 0.13302293419837952, "learning_rate": 4.1828412342548304e-08, "loss": 0.007, "step": 296290 }, { "epoch": 188.84639898024218, "grad_norm": 0.23550483584403992, "learning_rate": 4.160331796804196e-08, "loss": 0.0063, "step": 296300 }, { "epoch": 188.8527724665392, "grad_norm": 0.14021587371826172, "learning_rate": 4.1378830635224744e-08, "loss": 0.0028, "step": 296310 }, { "epoch": 188.8591459528362, "grad_norm": 0.20070265233516693, "learning_rate": 4.1154950346822263e-08, "loss": 0.007, "step": 296320 }, { "epoch": 188.8655194391332, "grad_norm": 0.35944369435310364, "learning_rate": 4.0931677105560116e-08, "loss": 0.0073, "step": 296330 }, { "epoch": 188.87189292543022, "grad_norm": 0.17874853312969208, "learning_rate": 4.070901091414725e-08, "loss": 0.0124, "step": 296340 }, { "epoch": 188.87826641172722, "grad_norm": 0.08079739660024643, "learning_rate": 4.048695177528705e-08, "loss": 0.0041, "step": 296350 }, { "epoch": 188.88463989802423, "grad_norm": 0.06870477646589279, "learning_rate": 4.0265499691677364e-08, "loss": 0.0158, "step": 296360 }, { "epoch": 188.8910133843212, "grad_norm": 0.239071324467659, "learning_rate": 4.004465466601603e-08, "loss": 0.0061, "step": 296370 }, { "epoch": 188.89738687061822, "grad_norm": 0.24987728893756866, "learning_rate": 3.982441670097869e-08, "loss": 0.0085, "step": 296380 }, { "epoch": 188.90376035691523, "grad_norm": 0.09074851870536804, "learning_rate": 3.9604785799246535e-08, "loss": 0.0032, "step": 296390 }, { "epoch": 188.91013384321224, "grad_norm": 0.10825735330581665, "learning_rate": 3.938576196348964e-08, "loss": 0.0028, "step": 296400 }, { "epoch": 188.91650732950924, "grad_norm": 0.20088697969913483, "learning_rate": 3.916734519636145e-08, "loss": 0.0052, "step": 296410 }, { "epoch": 188.92288081580625, "grad_norm": 0.17865224182605743, "learning_rate": 3.894953550052649e-08, "loss": 0.0031, "step": 296420 }, { "epoch": 188.92925430210326, "grad_norm": 0.0817057341337204, "learning_rate": 3.8732332878621544e-08, "loss": 0.003, "step": 296430 }, { "epoch": 188.93562778840027, "grad_norm": 0.16942498087882996, "learning_rate": 3.851573733328895e-08, "loss": 0.0023, "step": 296440 }, { "epoch": 188.94200127469725, "grad_norm": 0.1419515162706375, "learning_rate": 3.8299748867165476e-08, "loss": 0.002, "step": 296450 }, { "epoch": 188.94837476099426, "grad_norm": 0.1676950752735138, "learning_rate": 3.80843674828657e-08, "loss": 0.0081, "step": 296460 }, { "epoch": 188.95474824729126, "grad_norm": 0.1914488822221756, "learning_rate": 3.786959318302086e-08, "loss": 0.0045, "step": 296470 }, { "epoch": 188.96112173358827, "grad_norm": 0.07594010978937149, "learning_rate": 3.765542597022331e-08, "loss": 0.0048, "step": 296480 }, { "epoch": 188.96749521988528, "grad_norm": 0.5187348127365112, "learning_rate": 3.744186584709319e-08, "loss": 0.0099, "step": 296490 }, { "epoch": 188.9738687061823, "grad_norm": 0.22376735508441925, "learning_rate": 3.722891281621177e-08, "loss": 0.0095, "step": 296500 }, { "epoch": 188.9802421924793, "grad_norm": 0.1068815290927887, "learning_rate": 3.701656688017141e-08, "loss": 0.008, "step": 296510 }, { "epoch": 188.9866156787763, "grad_norm": 0.06730739027261734, "learning_rate": 3.680482804155894e-08, "loss": 0.0049, "step": 296520 }, { "epoch": 188.99298916507328, "grad_norm": 0.24257709085941315, "learning_rate": 3.659369630293341e-08, "loss": 0.0061, "step": 296530 }, { "epoch": 188.9993626513703, "grad_norm": 0.21727845072746277, "learning_rate": 3.6383171666870555e-08, "loss": 0.0051, "step": 296540 }, { "epoch": 189.0057361376673, "grad_norm": 0.5117260217666626, "learning_rate": 3.617325413592942e-08, "loss": 0.0216, "step": 296550 }, { "epoch": 189.0121096239643, "grad_norm": 0.33810722827911377, "learning_rate": 3.5963943712652436e-08, "loss": 0.0068, "step": 296560 }, { "epoch": 189.01848311026131, "grad_norm": 0.09816924482584, "learning_rate": 3.575524039958755e-08, "loss": 0.0111, "step": 296570 }, { "epoch": 189.02485659655832, "grad_norm": 0.3048410415649414, "learning_rate": 3.5547144199271635e-08, "loss": 0.0053, "step": 296580 }, { "epoch": 189.03123008285533, "grad_norm": 0.11536870151758194, "learning_rate": 3.533965511423043e-08, "loss": 0.004, "step": 296590 }, { "epoch": 189.03760356915234, "grad_norm": 0.08250776678323746, "learning_rate": 3.513277314698971e-08, "loss": 0.0032, "step": 296600 }, { "epoch": 189.04397705544932, "grad_norm": 0.11058985441923141, "learning_rate": 3.492649830005856e-08, "loss": 0.004, "step": 296610 }, { "epoch": 189.05035054174633, "grad_norm": 0.08930636942386627, "learning_rate": 3.47208305759461e-08, "loss": 0.0066, "step": 296620 }, { "epoch": 189.05672402804333, "grad_norm": 0.05109309405088425, "learning_rate": 3.4515769977150336e-08, "loss": 0.0038, "step": 296630 }, { "epoch": 189.06309751434034, "grad_norm": 0.6605996489524841, "learning_rate": 3.4311316506163705e-08, "loss": 0.0098, "step": 296640 }, { "epoch": 189.06947100063735, "grad_norm": 0.3073866367340088, "learning_rate": 3.410747016546756e-08, "loss": 0.004, "step": 296650 }, { "epoch": 189.07584448693436, "grad_norm": 0.10154100507497787, "learning_rate": 3.390423095754325e-08, "loss": 0.0064, "step": 296660 }, { "epoch": 189.08221797323137, "grad_norm": 0.2128845900297165, "learning_rate": 3.370159888486102e-08, "loss": 0.0028, "step": 296670 }, { "epoch": 189.08859145952837, "grad_norm": 0.3198198080062866, "learning_rate": 3.349957394987446e-08, "loss": 0.0033, "step": 296680 }, { "epoch": 189.09496494582535, "grad_norm": 0.14880989491939545, "learning_rate": 3.329815615504828e-08, "loss": 0.0049, "step": 296690 }, { "epoch": 189.10133843212236, "grad_norm": 0.10029733926057816, "learning_rate": 3.309734550282495e-08, "loss": 0.0045, "step": 296700 }, { "epoch": 189.10771191841937, "grad_norm": 0.12677347660064697, "learning_rate": 3.289714199564697e-08, "loss": 0.0043, "step": 296710 }, { "epoch": 189.11408540471638, "grad_norm": 0.13277117908000946, "learning_rate": 3.2697545635940184e-08, "loss": 0.0032, "step": 296720 }, { "epoch": 189.12045889101339, "grad_norm": 0.1619090884923935, "learning_rate": 3.2498556426141524e-08, "loss": 0.0053, "step": 296730 }, { "epoch": 189.1268323773104, "grad_norm": 0.22254951298236847, "learning_rate": 3.230017436866017e-08, "loss": 0.0034, "step": 296740 }, { "epoch": 189.1332058636074, "grad_norm": 0.4065954089164734, "learning_rate": 3.210239946591087e-08, "loss": 0.0043, "step": 296750 }, { "epoch": 189.1395793499044, "grad_norm": 0.5222809910774231, "learning_rate": 3.1905231720291696e-08, "loss": 0.0048, "step": 296760 }, { "epoch": 189.1459528362014, "grad_norm": 0.16400320827960968, "learning_rate": 3.170867113420073e-08, "loss": 0.006, "step": 296770 }, { "epoch": 189.1523263224984, "grad_norm": 0.06497401744127274, "learning_rate": 3.15127177100305e-08, "loss": 0.0043, "step": 296780 }, { "epoch": 189.1586998087954, "grad_norm": 0.1681327074766159, "learning_rate": 3.1317371450156895e-08, "loss": 0.0045, "step": 296790 }, { "epoch": 189.1650732950924, "grad_norm": 0.07617761939764023, "learning_rate": 3.1122632356955786e-08, "loss": 0.0073, "step": 296800 }, { "epoch": 189.17144678138942, "grad_norm": 0.07546544820070267, "learning_rate": 3.092850043279749e-08, "loss": 0.0035, "step": 296810 }, { "epoch": 189.17782026768643, "grad_norm": 0.1941872090101242, "learning_rate": 3.0734975680030144e-08, "loss": 0.0026, "step": 296820 }, { "epoch": 189.18419375398344, "grad_norm": 0.1047024354338646, "learning_rate": 3.054205810101296e-08, "loss": 0.0065, "step": 296830 }, { "epoch": 189.19056724028044, "grad_norm": 0.05354887992143631, "learning_rate": 3.034974769809407e-08, "loss": 0.0032, "step": 296840 }, { "epoch": 189.19694072657742, "grad_norm": 0.06441653519868851, "learning_rate": 3.015804447359938e-08, "loss": 0.0026, "step": 296850 }, { "epoch": 189.20331421287443, "grad_norm": 0.39483365416526794, "learning_rate": 2.9966948429865915e-08, "loss": 0.0046, "step": 296860 }, { "epoch": 189.20968769917144, "grad_norm": 0.11388028413057327, "learning_rate": 2.9776459569214044e-08, "loss": 0.0038, "step": 296870 }, { "epoch": 189.21606118546845, "grad_norm": 0.26432958245277405, "learning_rate": 2.9586577893953027e-08, "loss": 0.0032, "step": 296880 }, { "epoch": 189.22243467176546, "grad_norm": 0.11879128217697144, "learning_rate": 2.939730340640323e-08, "loss": 0.0033, "step": 296890 }, { "epoch": 189.22880815806246, "grad_norm": 0.0584251768887043, "learning_rate": 2.9208636108851718e-08, "loss": 0.0028, "step": 296900 }, { "epoch": 189.23518164435947, "grad_norm": 0.25674968957901, "learning_rate": 2.9020576003591095e-08, "loss": 0.0067, "step": 296910 }, { "epoch": 189.24155513065648, "grad_norm": 0.6235970258712769, "learning_rate": 2.883312309291952e-08, "loss": 0.0086, "step": 296920 }, { "epoch": 189.2479286169535, "grad_norm": 0.3863191604614258, "learning_rate": 2.864627737910186e-08, "loss": 0.0093, "step": 296930 }, { "epoch": 189.25430210325047, "grad_norm": 0.07355433702468872, "learning_rate": 2.8460038864414064e-08, "loss": 0.0031, "step": 296940 }, { "epoch": 189.26067558954747, "grad_norm": 0.14742104709148407, "learning_rate": 2.8274407551115432e-08, "loss": 0.0051, "step": 296950 }, { "epoch": 189.26704907584448, "grad_norm": 0.29686832427978516, "learning_rate": 2.8089383441465278e-08, "loss": 0.0077, "step": 296960 }, { "epoch": 189.2734225621415, "grad_norm": 0.16078005731105804, "learning_rate": 2.7904966537706246e-08, "loss": 0.0042, "step": 296970 }, { "epoch": 189.2797960484385, "grad_norm": 0.20683087408542633, "learning_rate": 2.772115684209209e-08, "loss": 0.0044, "step": 296980 }, { "epoch": 189.2861695347355, "grad_norm": 0.10745426267385483, "learning_rate": 2.7537954356837703e-08, "loss": 0.012, "step": 296990 }, { "epoch": 189.29254302103251, "grad_norm": 0.11685596406459808, "learning_rate": 2.735535908418574e-08, "loss": 0.0045, "step": 297000 }, { "epoch": 189.29891650732952, "grad_norm": 0.19323167204856873, "learning_rate": 2.717337102634554e-08, "loss": 0.0072, "step": 297010 }, { "epoch": 189.3052899936265, "grad_norm": 0.10523601621389389, "learning_rate": 2.6991990185532e-08, "loss": 0.0061, "step": 297020 }, { "epoch": 189.3116634799235, "grad_norm": 0.07446424663066864, "learning_rate": 2.6811216563948915e-08, "loss": 0.0051, "step": 297030 }, { "epoch": 189.31803696622052, "grad_norm": 0.2056524157524109, "learning_rate": 2.6631050163794524e-08, "loss": 0.0035, "step": 297040 }, { "epoch": 189.32441045251753, "grad_norm": 0.14821042120456696, "learning_rate": 2.6451490987250415e-08, "loss": 0.0052, "step": 297050 }, { "epoch": 189.33078393881453, "grad_norm": 0.10646683722734451, "learning_rate": 2.627253903650373e-08, "loss": 0.0096, "step": 297060 }, { "epoch": 189.33715742511154, "grad_norm": 0.17626622319221497, "learning_rate": 2.6094194313730503e-08, "loss": 0.0163, "step": 297070 }, { "epoch": 189.34353091140855, "grad_norm": 0.12089025974273682, "learning_rate": 2.591645682109567e-08, "loss": 0.008, "step": 297080 }, { "epoch": 189.34990439770556, "grad_norm": 0.35958871245384216, "learning_rate": 2.5739326560758616e-08, "loss": 0.0065, "step": 297090 }, { "epoch": 189.35627788400254, "grad_norm": 0.21629081666469574, "learning_rate": 2.556280353487317e-08, "loss": 0.0043, "step": 297100 }, { "epoch": 189.36265137029955, "grad_norm": 0.38515836000442505, "learning_rate": 2.5386887745582065e-08, "loss": 0.0062, "step": 297110 }, { "epoch": 189.36902485659655, "grad_norm": 0.35913747549057007, "learning_rate": 2.5211579195028035e-08, "loss": 0.0041, "step": 297120 }, { "epoch": 189.37539834289356, "grad_norm": 0.17255820333957672, "learning_rate": 2.50368778853316e-08, "loss": 0.0034, "step": 297130 }, { "epoch": 189.38177182919057, "grad_norm": 0.22658850252628326, "learning_rate": 2.486278381862439e-08, "loss": 0.0054, "step": 297140 }, { "epoch": 189.38814531548758, "grad_norm": 0.1359238475561142, "learning_rate": 2.468929699702138e-08, "loss": 0.0021, "step": 297150 }, { "epoch": 189.39451880178459, "grad_norm": 0.18132919073104858, "learning_rate": 2.4516417422626447e-08, "loss": 0.004, "step": 297160 }, { "epoch": 189.4008922880816, "grad_norm": 0.14493589103221893, "learning_rate": 2.4344145097537908e-08, "loss": 0.0029, "step": 297170 }, { "epoch": 189.40726577437857, "grad_norm": 0.22701983153820038, "learning_rate": 2.4172480023854083e-08, "loss": 0.0027, "step": 297180 }, { "epoch": 189.41363926067558, "grad_norm": 0.07092025876045227, "learning_rate": 2.4001422203662193e-08, "loss": 0.0053, "step": 297190 }, { "epoch": 189.4200127469726, "grad_norm": 0.07895837724208832, "learning_rate": 2.3830971639038357e-08, "loss": 0.0053, "step": 297200 }, { "epoch": 189.4263862332696, "grad_norm": 0.11558322608470917, "learning_rate": 2.3661128332053138e-08, "loss": 0.0052, "step": 297210 }, { "epoch": 189.4327597195666, "grad_norm": 0.059126488864421844, "learning_rate": 2.3491892284766003e-08, "loss": 0.0062, "step": 297220 }, { "epoch": 189.4391332058636, "grad_norm": 0.3055422306060791, "learning_rate": 2.3323263499241964e-08, "loss": 0.0043, "step": 297230 }, { "epoch": 189.44550669216062, "grad_norm": 0.06021801382303238, "learning_rate": 2.3155241977529386e-08, "loss": 0.0036, "step": 297240 }, { "epoch": 189.45188017845763, "grad_norm": 0.2529100477695465, "learning_rate": 2.2987827721659973e-08, "loss": 0.0025, "step": 297250 }, { "epoch": 189.4582536647546, "grad_norm": 0.12444552779197693, "learning_rate": 2.2821020733676536e-08, "loss": 0.0036, "step": 297260 }, { "epoch": 189.46462715105162, "grad_norm": 0.08223500847816467, "learning_rate": 2.265482101560523e-08, "loss": 0.0026, "step": 297270 }, { "epoch": 189.47100063734862, "grad_norm": 0.23236192762851715, "learning_rate": 2.248922856946667e-08, "loss": 0.0073, "step": 297280 }, { "epoch": 189.47737412364563, "grad_norm": 0.12105612456798553, "learning_rate": 2.23242433972648e-08, "loss": 0.0054, "step": 297290 }, { "epoch": 189.48374760994264, "grad_norm": 0.2328159362077713, "learning_rate": 2.2159865501014676e-08, "loss": 0.0062, "step": 297300 }, { "epoch": 189.49012109623965, "grad_norm": 0.19029003381729126, "learning_rate": 2.1996094882709152e-08, "loss": 0.0081, "step": 297310 }, { "epoch": 189.49649458253666, "grad_norm": 0.4160686731338501, "learning_rate": 2.183293154433552e-08, "loss": 0.0067, "step": 297320 }, { "epoch": 189.50286806883366, "grad_norm": 0.3011624813079834, "learning_rate": 2.167037548788109e-08, "loss": 0.0055, "step": 297330 }, { "epoch": 189.50924155513064, "grad_norm": 0.23313963413238525, "learning_rate": 2.1508426715316498e-08, "loss": 0.0047, "step": 297340 }, { "epoch": 189.51561504142765, "grad_norm": 0.056116849184036255, "learning_rate": 2.1347085228612396e-08, "loss": 0.0041, "step": 297350 }, { "epoch": 189.52198852772466, "grad_norm": 0.07466873526573181, "learning_rate": 2.1186351029733875e-08, "loss": 0.0038, "step": 297360 }, { "epoch": 189.52836201402167, "grad_norm": 0.2180560827255249, "learning_rate": 2.102622412062383e-08, "loss": 0.0064, "step": 297370 }, { "epoch": 189.53473550031867, "grad_norm": 0.18217788636684418, "learning_rate": 2.0866704503236246e-08, "loss": 0.0032, "step": 297380 }, { "epoch": 189.54110898661568, "grad_norm": 0.1609777808189392, "learning_rate": 2.0707792179502917e-08, "loss": 0.0043, "step": 297390 }, { "epoch": 189.5474824729127, "grad_norm": 0.09558706730604172, "learning_rate": 2.054948715136118e-08, "loss": 0.0108, "step": 297400 }, { "epoch": 189.5538559592097, "grad_norm": 0.1306294947862625, "learning_rate": 2.0391789420726172e-08, "loss": 0.0055, "step": 297410 }, { "epoch": 189.56022944550668, "grad_norm": 0.5064882040023804, "learning_rate": 2.0234698989524127e-08, "loss": 0.0069, "step": 297420 }, { "epoch": 189.5666029318037, "grad_norm": 0.07215956598520279, "learning_rate": 2.0078215859659077e-08, "loss": 0.0015, "step": 297430 }, { "epoch": 189.5729764181007, "grad_norm": 0.044256675988435745, "learning_rate": 1.992234003303506e-08, "loss": 0.0021, "step": 297440 }, { "epoch": 189.5793499043977, "grad_norm": 0.06045926734805107, "learning_rate": 1.9767071511539446e-08, "loss": 0.0032, "step": 297450 }, { "epoch": 189.5857233906947, "grad_norm": 0.11077629774808884, "learning_rate": 1.9612410297065177e-08, "loss": 0.0118, "step": 297460 }, { "epoch": 189.59209687699172, "grad_norm": 0.06335186213254929, "learning_rate": 1.945835639148852e-08, "loss": 0.0104, "step": 297470 }, { "epoch": 189.59847036328873, "grad_norm": 0.056402985006570816, "learning_rate": 1.9304909796680204e-08, "loss": 0.0039, "step": 297480 }, { "epoch": 189.60484384958573, "grad_norm": 0.13328664004802704, "learning_rate": 1.9152070514510954e-08, "loss": 0.0042, "step": 297490 }, { "epoch": 189.61121733588274, "grad_norm": 0.15019544959068298, "learning_rate": 1.8999838546829296e-08, "loss": 0.0067, "step": 297500 }, { "epoch": 189.61759082217972, "grad_norm": 0.2580767571926117, "learning_rate": 1.8848213895489296e-08, "loss": 0.0082, "step": 297510 }, { "epoch": 189.62396430847673, "grad_norm": 0.552155613899231, "learning_rate": 1.869719656233948e-08, "loss": 0.0061, "step": 297520 }, { "epoch": 189.63033779477374, "grad_norm": 0.4356333315372467, "learning_rate": 1.8546786549206163e-08, "loss": 0.0063, "step": 297530 }, { "epoch": 189.63671128107075, "grad_norm": 0.22825190424919128, "learning_rate": 1.839698385791566e-08, "loss": 0.0044, "step": 297540 }, { "epoch": 189.64308476736775, "grad_norm": 0.13234975934028625, "learning_rate": 1.8247788490299844e-08, "loss": 0.0043, "step": 297550 }, { "epoch": 189.64945825366476, "grad_norm": 0.5078578591346741, "learning_rate": 1.809920044815727e-08, "loss": 0.0129, "step": 297560 }, { "epoch": 189.65583173996177, "grad_norm": 0.2242719829082489, "learning_rate": 1.795121973330316e-08, "loss": 0.0118, "step": 297570 }, { "epoch": 189.66220522625878, "grad_norm": 0.05047623813152313, "learning_rate": 1.7803846347530517e-08, "loss": 0.0046, "step": 297580 }, { "epoch": 189.66857871255576, "grad_norm": 0.20666278898715973, "learning_rate": 1.7657080292637905e-08, "loss": 0.0172, "step": 297590 }, { "epoch": 189.67495219885276, "grad_norm": 0.49902671575546265, "learning_rate": 1.7510921570396134e-08, "loss": 0.0163, "step": 297600 }, { "epoch": 189.68132568514977, "grad_norm": 0.10809300094842911, "learning_rate": 1.7365370182592655e-08, "loss": 0.0086, "step": 297610 }, { "epoch": 189.68769917144678, "grad_norm": 0.10590340942144394, "learning_rate": 1.7220426130987178e-08, "loss": 0.0114, "step": 297620 }, { "epoch": 189.6940726577438, "grad_norm": 0.1953423023223877, "learning_rate": 1.7076089417350506e-08, "loss": 0.0024, "step": 297630 }, { "epoch": 189.7004461440408, "grad_norm": 0.1930435597896576, "learning_rate": 1.6932360043425686e-08, "loss": 0.004, "step": 297640 }, { "epoch": 189.7068196303378, "grad_norm": 0.10636623948812485, "learning_rate": 1.678923801096688e-08, "loss": 0.0036, "step": 297650 }, { "epoch": 189.7131931166348, "grad_norm": 0.0865987166762352, "learning_rate": 1.6646723321711578e-08, "loss": 0.0192, "step": 297660 }, { "epoch": 189.7195666029318, "grad_norm": 0.10517627745866776, "learning_rate": 1.6504815977386178e-08, "loss": 0.005, "step": 297670 }, { "epoch": 189.7259400892288, "grad_norm": 0.1058163046836853, "learning_rate": 1.636351597972263e-08, "loss": 0.0078, "step": 297680 }, { "epoch": 189.7323135755258, "grad_norm": 0.46996310353279114, "learning_rate": 1.622282333043623e-08, "loss": 0.0087, "step": 297690 }, { "epoch": 189.73868706182282, "grad_norm": 0.1587182879447937, "learning_rate": 1.608273803123672e-08, "loss": 0.0063, "step": 297700 }, { "epoch": 189.74506054811982, "grad_norm": 0.11727356165647507, "learning_rate": 1.594326008382274e-08, "loss": 0.0065, "step": 297710 }, { "epoch": 189.75143403441683, "grad_norm": 0.08796380460262299, "learning_rate": 1.5804389489887383e-08, "loss": 0.0061, "step": 297720 }, { "epoch": 189.75780752071384, "grad_norm": 0.37621790170669556, "learning_rate": 1.5666126251123736e-08, "loss": 0.0037, "step": 297730 }, { "epoch": 189.76418100701085, "grad_norm": 0.16402481496334076, "learning_rate": 1.5528470369208238e-08, "loss": 0.006, "step": 297740 }, { "epoch": 189.77055449330783, "grad_norm": 0.08071665465831757, "learning_rate": 1.5391421845817323e-08, "loss": 0.0024, "step": 297750 }, { "epoch": 189.77692797960484, "grad_norm": 0.11878548562526703, "learning_rate": 1.5254980682616325e-08, "loss": 0.0034, "step": 297760 }, { "epoch": 189.78330146590184, "grad_norm": 0.13533630967140198, "learning_rate": 1.5119146881259482e-08, "loss": 0.0049, "step": 297770 }, { "epoch": 189.78967495219885, "grad_norm": 0.22621577978134155, "learning_rate": 1.498392044339547e-08, "loss": 0.0047, "step": 297780 }, { "epoch": 189.79604843849586, "grad_norm": 0.07533727586269379, "learning_rate": 1.4849301370672974e-08, "loss": 0.0061, "step": 297790 }, { "epoch": 189.80242192479287, "grad_norm": 0.11585631966590881, "learning_rate": 1.4715289664724019e-08, "loss": 0.0102, "step": 297800 }, { "epoch": 189.80879541108987, "grad_norm": 0.47667375206947327, "learning_rate": 1.4581885327175083e-08, "loss": 0.0065, "step": 297810 }, { "epoch": 189.81516889738688, "grad_norm": 0.42049142718315125, "learning_rate": 1.4449088359658192e-08, "loss": 0.0051, "step": 297820 }, { "epoch": 189.82154238368386, "grad_norm": 0.08424077183008194, "learning_rate": 1.431689876377207e-08, "loss": 0.0089, "step": 297830 }, { "epoch": 189.82791586998087, "grad_norm": 0.18294285237789154, "learning_rate": 1.4185316541132088e-08, "loss": 0.0044, "step": 297840 }, { "epoch": 189.83428935627788, "grad_norm": 0.07341707497835159, "learning_rate": 1.405434169333697e-08, "loss": 0.005, "step": 297850 }, { "epoch": 189.8406628425749, "grad_norm": 0.06044583395123482, "learning_rate": 1.3923974221974334e-08, "loss": 0.0048, "step": 297860 }, { "epoch": 189.8470363288719, "grad_norm": 0.13367526233196259, "learning_rate": 1.37942141286318e-08, "loss": 0.0044, "step": 297870 }, { "epoch": 189.8534098151689, "grad_norm": 0.08339710533618927, "learning_rate": 1.366506141488033e-08, "loss": 0.0034, "step": 297880 }, { "epoch": 189.8597833014659, "grad_norm": 0.2560681402683258, "learning_rate": 1.3536516082296446e-08, "loss": 0.0062, "step": 297890 }, { "epoch": 189.86615678776292, "grad_norm": 0.06594772636890411, "learning_rate": 1.3408578132434457e-08, "loss": 0.0116, "step": 297900 }, { "epoch": 189.8725302740599, "grad_norm": 0.10840126127004623, "learning_rate": 1.3281247566859777e-08, "loss": 0.002, "step": 297910 }, { "epoch": 189.8789037603569, "grad_norm": 0.28327152132987976, "learning_rate": 1.3154524387110068e-08, "loss": 0.0057, "step": 297920 }, { "epoch": 189.8852772466539, "grad_norm": 0.08006695657968521, "learning_rate": 1.3028408594728536e-08, "loss": 0.0023, "step": 297930 }, { "epoch": 189.89165073295092, "grad_norm": 0.15620437264442444, "learning_rate": 1.2902900191247292e-08, "loss": 0.005, "step": 297940 }, { "epoch": 189.89802421924793, "grad_norm": 0.17057324945926666, "learning_rate": 1.277799917819289e-08, "loss": 0.0072, "step": 297950 }, { "epoch": 189.90439770554494, "grad_norm": 0.16260184347629547, "learning_rate": 1.2653705557080787e-08, "loss": 0.0084, "step": 297960 }, { "epoch": 189.91077119184195, "grad_norm": 0.2160799354314804, "learning_rate": 1.2530019329420884e-08, "loss": 0.0043, "step": 297970 }, { "epoch": 189.91714467813895, "grad_norm": 0.048006750643253326, "learning_rate": 1.2406940496717535e-08, "loss": 0.0046, "step": 297980 }, { "epoch": 189.92351816443593, "grad_norm": 0.2791106700897217, "learning_rate": 1.2284469060469538e-08, "loss": 0.005, "step": 297990 }, { "epoch": 189.92989165073294, "grad_norm": 0.11563897132873535, "learning_rate": 1.2162605022159046e-08, "loss": 0.003, "step": 298000 }, { "epoch": 189.93626513702995, "grad_norm": 0.12068942189216614, "learning_rate": 1.2041348383268203e-08, "loss": 0.0042, "step": 298010 }, { "epoch": 189.94263862332696, "grad_norm": 0.1406150460243225, "learning_rate": 1.1920699145273606e-08, "loss": 0.0034, "step": 298020 }, { "epoch": 189.94901210962396, "grad_norm": 0.39291033148765564, "learning_rate": 1.1800657309635199e-08, "loss": 0.0051, "step": 298030 }, { "epoch": 189.95538559592097, "grad_norm": 0.0682714655995369, "learning_rate": 1.1681222877818476e-08, "loss": 0.0048, "step": 298040 }, { "epoch": 189.96175908221798, "grad_norm": 0.055531758815050125, "learning_rate": 1.1562395851272279e-08, "loss": 0.0046, "step": 298050 }, { "epoch": 189.968132568515, "grad_norm": 0.08511833846569061, "learning_rate": 1.1444176231434345e-08, "loss": 0.0036, "step": 298060 }, { "epoch": 189.97450605481197, "grad_norm": 0.10260163247585297, "learning_rate": 1.1326564019753516e-08, "loss": 0.0035, "step": 298070 }, { "epoch": 189.98087954110898, "grad_norm": 0.11630035191774368, "learning_rate": 1.1209559217645327e-08, "loss": 0.0171, "step": 298080 }, { "epoch": 189.98725302740598, "grad_norm": 0.33356866240501404, "learning_rate": 1.1093161826547515e-08, "loss": 0.0028, "step": 298090 }, { "epoch": 189.993626513703, "grad_norm": 0.44927462935447693, "learning_rate": 1.0977371847858964e-08, "loss": 0.0075, "step": 298100 }, { "epoch": 190.0, "grad_norm": 0.24879717826843262, "learning_rate": 1.0862189282989655e-08, "loss": 0.0063, "step": 298110 }, { "epoch": 190.006373486297, "grad_norm": 0.12567779421806335, "learning_rate": 1.074761413334957e-08, "loss": 0.0043, "step": 298120 }, { "epoch": 190.01274697259402, "grad_norm": 0.4815015494823456, "learning_rate": 1.063364640032094e-08, "loss": 0.0066, "step": 298130 }, { "epoch": 190.01912045889102, "grad_norm": 0.07385817170143127, "learning_rate": 1.0520286085291542e-08, "loss": 0.0034, "step": 298140 }, { "epoch": 190.02549394518803, "grad_norm": 0.0824502557516098, "learning_rate": 1.0407533189638052e-08, "loss": 0.0103, "step": 298150 }, { "epoch": 190.031867431485, "grad_norm": 0.27674591541290283, "learning_rate": 1.0295387714731597e-08, "loss": 0.0031, "step": 298160 }, { "epoch": 190.03824091778202, "grad_norm": 0.0869186595082283, "learning_rate": 1.0183849661932199e-08, "loss": 0.0042, "step": 298170 }, { "epoch": 190.04461440407903, "grad_norm": 0.0969652310013771, "learning_rate": 1.0072919032599882e-08, "loss": 0.0044, "step": 298180 }, { "epoch": 190.05098789037604, "grad_norm": 0.04986559599637985, "learning_rate": 9.962595828078014e-09, "loss": 0.0058, "step": 298190 }, { "epoch": 190.05736137667304, "grad_norm": 0.5371344685554504, "learning_rate": 9.852880049709967e-09, "loss": 0.0101, "step": 298200 }, { "epoch": 190.06373486297005, "grad_norm": 0.4130677878856659, "learning_rate": 9.743771698828009e-09, "loss": 0.0117, "step": 298210 }, { "epoch": 190.07010834926706, "grad_norm": 0.34763428568840027, "learning_rate": 9.635270776753302e-09, "loss": 0.0062, "step": 298220 }, { "epoch": 190.07648183556407, "grad_norm": 0.07397058606147766, "learning_rate": 9.527377284812566e-09, "loss": 0.0029, "step": 298230 }, { "epoch": 190.08285532186105, "grad_norm": 0.5087558031082153, "learning_rate": 9.420091224315863e-09, "loss": 0.006, "step": 298240 }, { "epoch": 190.08922880815805, "grad_norm": 0.2543173134326935, "learning_rate": 9.313412596562154e-09, "loss": 0.0034, "step": 298250 }, { "epoch": 190.09560229445506, "grad_norm": 0.06641823798418045, "learning_rate": 9.207341402844848e-09, "loss": 0.0038, "step": 298260 }, { "epoch": 190.10197578075207, "grad_norm": 0.0928073525428772, "learning_rate": 9.101877644462908e-09, "loss": 0.0092, "step": 298270 }, { "epoch": 190.10834926704908, "grad_norm": 0.1313711553812027, "learning_rate": 8.99702132269309e-09, "loss": 0.0038, "step": 298280 }, { "epoch": 190.1147227533461, "grad_norm": 0.6698604226112366, "learning_rate": 8.892772438806595e-09, "loss": 0.0065, "step": 298290 }, { "epoch": 190.1210962396431, "grad_norm": 0.08089077472686768, "learning_rate": 8.789130994069084e-09, "loss": 0.0084, "step": 298300 }, { "epoch": 190.1274697259401, "grad_norm": 0.046150580048561096, "learning_rate": 8.686096989751757e-09, "loss": 0.0085, "step": 298310 }, { "epoch": 190.13384321223708, "grad_norm": 0.14509516954421997, "learning_rate": 8.583670427092516e-09, "loss": 0.0052, "step": 298320 }, { "epoch": 190.1402166985341, "grad_norm": 0.05925793573260307, "learning_rate": 8.481851307340361e-09, "loss": 0.0026, "step": 298330 }, { "epoch": 190.1465901848311, "grad_norm": 0.18509458005428314, "learning_rate": 8.38063963173319e-09, "loss": 0.0088, "step": 298340 }, { "epoch": 190.1529636711281, "grad_norm": 0.08067155629396439, "learning_rate": 8.28003540150335e-09, "loss": 0.008, "step": 298350 }, { "epoch": 190.1593371574251, "grad_norm": 0.5658901929855347, "learning_rate": 8.180038617872088e-09, "loss": 0.007, "step": 298360 }, { "epoch": 190.16571064372212, "grad_norm": 0.19293543696403503, "learning_rate": 8.080649282049546e-09, "loss": 0.0018, "step": 298370 }, { "epoch": 190.17208413001913, "grad_norm": 0.5889506340026855, "learning_rate": 7.981867395251418e-09, "loss": 0.0088, "step": 298380 }, { "epoch": 190.17845761631614, "grad_norm": 0.3897763788700104, "learning_rate": 7.883692958671196e-09, "loss": 0.0191, "step": 298390 }, { "epoch": 190.18483110261312, "grad_norm": 0.4962092936038971, "learning_rate": 7.786125973502367e-09, "loss": 0.0063, "step": 298400 }, { "epoch": 190.19120458891013, "grad_norm": 0.05955678969621658, "learning_rate": 7.689166440938423e-09, "loss": 0.0083, "step": 298410 }, { "epoch": 190.19757807520713, "grad_norm": 0.12326288968324661, "learning_rate": 7.592814362150646e-09, "loss": 0.0044, "step": 298420 }, { "epoch": 190.20395156150414, "grad_norm": 0.13224326074123383, "learning_rate": 7.497069738304774e-09, "loss": 0.0064, "step": 298430 }, { "epoch": 190.21032504780115, "grad_norm": 0.07244327664375305, "learning_rate": 7.401932570577641e-09, "loss": 0.0086, "step": 298440 }, { "epoch": 190.21669853409816, "grad_norm": 0.11097578704357147, "learning_rate": 7.3074028601127775e-09, "loss": 0.0026, "step": 298450 }, { "epoch": 190.22307202039516, "grad_norm": 0.4774100184440613, "learning_rate": 7.213480608064816e-09, "loss": 0.0095, "step": 298460 }, { "epoch": 190.22944550669217, "grad_norm": 0.35835495591163635, "learning_rate": 7.120165815577284e-09, "loss": 0.0059, "step": 298470 }, { "epoch": 190.23581899298915, "grad_norm": 0.34317103028297424, "learning_rate": 7.0274584837770604e-09, "loss": 0.0083, "step": 298480 }, { "epoch": 190.24219247928616, "grad_norm": 0.3800944983959198, "learning_rate": 6.935358613796572e-09, "loss": 0.0042, "step": 298490 }, { "epoch": 190.24856596558317, "grad_norm": 0.6119922399520874, "learning_rate": 6.843866206751592e-09, "loss": 0.0132, "step": 298500 }, { "epoch": 190.25493945188018, "grad_norm": 0.1342450976371765, "learning_rate": 6.752981263757896e-09, "loss": 0.0062, "step": 298510 }, { "epoch": 190.26131293817718, "grad_norm": 0.2742069959640503, "learning_rate": 6.662703785914604e-09, "loss": 0.0029, "step": 298520 }, { "epoch": 190.2676864244742, "grad_norm": 0.20354066789150238, "learning_rate": 6.573033774320836e-09, "loss": 0.007, "step": 298530 }, { "epoch": 190.2740599107712, "grad_norm": 0.08015038818120956, "learning_rate": 6.483971230064612e-09, "loss": 0.009, "step": 298540 }, { "epoch": 190.2804333970682, "grad_norm": 0.12158428132534027, "learning_rate": 6.395516154233949e-09, "loss": 0.003, "step": 298550 }, { "epoch": 190.2868068833652, "grad_norm": 0.08538033813238144, "learning_rate": 6.307668547894663e-09, "loss": 0.0025, "step": 298560 }, { "epoch": 190.2931803696622, "grad_norm": 0.1410306841135025, "learning_rate": 6.220428412123669e-09, "loss": 0.0082, "step": 298570 }, { "epoch": 190.2995538559592, "grad_norm": 0.11275601387023926, "learning_rate": 6.133795747970128e-09, "loss": 0.0064, "step": 298580 }, { "epoch": 190.3059273422562, "grad_norm": 0.2000451683998108, "learning_rate": 6.047770556499854e-09, "loss": 0.0033, "step": 298590 }, { "epoch": 190.31230082855322, "grad_norm": 0.081319659948349, "learning_rate": 5.962352838750907e-09, "loss": 0.0065, "step": 298600 }, { "epoch": 190.31867431485023, "grad_norm": 0.43258631229400635, "learning_rate": 5.877542595755792e-09, "loss": 0.005, "step": 298610 }, { "epoch": 190.32504780114724, "grad_norm": 0.32405292987823486, "learning_rate": 5.79333982855812e-09, "loss": 0.0036, "step": 298620 }, { "epoch": 190.33142128744424, "grad_norm": 0.11890079081058502, "learning_rate": 5.709744538168194e-09, "loss": 0.0033, "step": 298630 }, { "epoch": 190.33779477374122, "grad_norm": 0.1518031805753708, "learning_rate": 5.626756725607418e-09, "loss": 0.0031, "step": 298640 }, { "epoch": 190.34416826003823, "grad_norm": 0.3691232204437256, "learning_rate": 5.544376391886097e-09, "loss": 0.0043, "step": 298650 }, { "epoch": 190.35054174633524, "grad_norm": 0.2165735810995102, "learning_rate": 5.46260353800343e-09, "loss": 0.0049, "step": 298660 }, { "epoch": 190.35691523263225, "grad_norm": 0.1342117190361023, "learning_rate": 5.381438164953067e-09, "loss": 0.0046, "step": 298670 }, { "epoch": 190.36328871892925, "grad_norm": 0.07925551384687424, "learning_rate": 5.300880273717557e-09, "loss": 0.0057, "step": 298680 }, { "epoch": 190.36966220522626, "grad_norm": 0.09302566200494766, "learning_rate": 5.220929865284996e-09, "loss": 0.012, "step": 298690 }, { "epoch": 190.37603569152327, "grad_norm": 0.11416547745466232, "learning_rate": 5.141586940615728e-09, "loss": 0.0073, "step": 298700 }, { "epoch": 190.38240917782028, "grad_norm": 0.2860948145389557, "learning_rate": 5.062851500681198e-09, "loss": 0.0068, "step": 298710 }, { "epoch": 190.38878266411726, "grad_norm": 0.11029336601495743, "learning_rate": 4.984723546436198e-09, "loss": 0.0057, "step": 298720 }, { "epoch": 190.39515615041427, "grad_norm": 0.19135640561580658, "learning_rate": 4.90720307882997e-09, "loss": 0.0025, "step": 298730 }, { "epoch": 190.40152963671127, "grad_norm": 0.08969959616661072, "learning_rate": 4.8302900988062004e-09, "loss": 0.004, "step": 298740 }, { "epoch": 190.40790312300828, "grad_norm": 0.07961215078830719, "learning_rate": 4.753984607291928e-09, "loss": 0.0041, "step": 298750 }, { "epoch": 190.4142766093053, "grad_norm": 0.45909225940704346, "learning_rate": 4.67828660522529e-09, "loss": 0.0089, "step": 298760 }, { "epoch": 190.4206500956023, "grad_norm": 0.20675383508205414, "learning_rate": 4.603196093522222e-09, "loss": 0.0065, "step": 298770 }, { "epoch": 190.4270235818993, "grad_norm": 0.23445729911327362, "learning_rate": 4.528713073087554e-09, "loss": 0.0028, "step": 298780 }, { "epoch": 190.4333970681963, "grad_norm": 0.3120797872543335, "learning_rate": 4.45483754483722e-09, "loss": 0.0098, "step": 298790 }, { "epoch": 190.43977055449332, "grad_norm": 0.13731467723846436, "learning_rate": 4.381569509664951e-09, "loss": 0.004, "step": 298800 }, { "epoch": 190.4461440407903, "grad_norm": 0.4516731798648834, "learning_rate": 4.308908968458925e-09, "loss": 0.0076, "step": 298810 }, { "epoch": 190.4525175270873, "grad_norm": 0.41705456376075745, "learning_rate": 4.236855922101768e-09, "loss": 0.0042, "step": 298820 }, { "epoch": 190.45889101338432, "grad_norm": 0.041996248066425323, "learning_rate": 4.165410371470557e-09, "loss": 0.0038, "step": 298830 }, { "epoch": 190.46526449968133, "grad_norm": 0.04862324520945549, "learning_rate": 4.094572317436818e-09, "loss": 0.0082, "step": 298840 }, { "epoch": 190.47163798597833, "grad_norm": 0.07590573281049728, "learning_rate": 4.024341760855421e-09, "loss": 0.0085, "step": 298850 }, { "epoch": 190.47801147227534, "grad_norm": 0.3595457673072815, "learning_rate": 3.95471870258124e-09, "loss": 0.0066, "step": 298860 }, { "epoch": 190.48438495857235, "grad_norm": 0.22825458645820618, "learning_rate": 3.885703143463593e-09, "loss": 0.007, "step": 298870 }, { "epoch": 190.49075844486936, "grad_norm": 0.4076245427131653, "learning_rate": 3.817295084335148e-09, "loss": 0.0064, "step": 298880 }, { "epoch": 190.49713193116634, "grad_norm": 0.08570587635040283, "learning_rate": 3.749494526034125e-09, "loss": 0.0041, "step": 298890 }, { "epoch": 190.50350541746334, "grad_norm": 0.05060943588614464, "learning_rate": 3.682301469382088e-09, "loss": 0.0033, "step": 298900 }, { "epoch": 190.50987890376035, "grad_norm": 0.1578666716814041, "learning_rate": 3.6157159151950505e-09, "loss": 0.0035, "step": 298910 }, { "epoch": 190.51625239005736, "grad_norm": 0.056203484535217285, "learning_rate": 3.5497378642779244e-09, "loss": 0.0028, "step": 298920 }, { "epoch": 190.52262587635437, "grad_norm": 0.07751420885324478, "learning_rate": 3.484367317435622e-09, "loss": 0.0051, "step": 298930 }, { "epoch": 190.52899936265138, "grad_norm": 0.2465314120054245, "learning_rate": 3.419604275467503e-09, "loss": 0.0042, "step": 298940 }, { "epoch": 190.53537284894838, "grad_norm": 0.2539895176887512, "learning_rate": 3.355448739150724e-09, "loss": 0.0051, "step": 298950 }, { "epoch": 190.5417463352454, "grad_norm": 0.0759672224521637, "learning_rate": 3.2919007092735433e-09, "loss": 0.0074, "step": 298960 }, { "epoch": 190.54811982154237, "grad_norm": 0.16150176525115967, "learning_rate": 3.2289601866020146e-09, "loss": 0.0047, "step": 298970 }, { "epoch": 190.55449330783938, "grad_norm": 0.0907793939113617, "learning_rate": 3.1666271719021923e-09, "loss": 0.0048, "step": 298980 }, { "epoch": 190.5608667941364, "grad_norm": 0.3689658045768738, "learning_rate": 3.1049016659345786e-09, "loss": 0.0063, "step": 298990 }, { "epoch": 190.5672402804334, "grad_norm": 0.12464717030525208, "learning_rate": 3.0437836694485745e-09, "loss": 0.0038, "step": 299000 }, { "epoch": 190.5736137667304, "grad_norm": 0.2692871689796448, "learning_rate": 2.983273183188029e-09, "loss": 0.0062, "step": 299010 }, { "epoch": 190.5799872530274, "grad_norm": 0.06143957003951073, "learning_rate": 2.9233702078801382e-09, "loss": 0.0066, "step": 299020 }, { "epoch": 190.58636073932442, "grad_norm": 0.352961003780365, "learning_rate": 2.864074744263201e-09, "loss": 0.0053, "step": 299030 }, { "epoch": 190.59273422562143, "grad_norm": 0.17137175798416138, "learning_rate": 2.8053867930477597e-09, "loss": 0.0041, "step": 299040 }, { "epoch": 190.5991077119184, "grad_norm": 0.08922956883907318, "learning_rate": 2.747306354955459e-09, "loss": 0.0053, "step": 299050 }, { "epoch": 190.60548119821541, "grad_norm": 0.20621588826179504, "learning_rate": 2.68983343068574e-09, "loss": 0.0049, "step": 299060 }, { "epoch": 190.61185468451242, "grad_norm": 0.12091609835624695, "learning_rate": 2.6329680209435936e-09, "loss": 0.0065, "step": 299070 }, { "epoch": 190.61822817080943, "grad_norm": 0.1347047984600067, "learning_rate": 2.576710126411808e-09, "loss": 0.0053, "step": 299080 }, { "epoch": 190.62460165710644, "grad_norm": 0.2161485105752945, "learning_rate": 2.5210597477842713e-09, "loss": 0.0056, "step": 299090 }, { "epoch": 190.63097514340345, "grad_norm": 0.2271895408630371, "learning_rate": 2.466016885727118e-09, "loss": 0.0082, "step": 299100 }, { "epoch": 190.63734862970045, "grad_norm": 0.06518982350826263, "learning_rate": 2.4115815409120335e-09, "loss": 0.0031, "step": 299110 }, { "epoch": 190.64372211599746, "grad_norm": 0.041325055062770844, "learning_rate": 2.357753714005151e-09, "loss": 0.0019, "step": 299120 }, { "epoch": 190.65009560229444, "grad_norm": 0.2879883348941803, "learning_rate": 2.3045334056559508e-09, "loss": 0.0048, "step": 299130 }, { "epoch": 190.65646908859145, "grad_norm": 0.3299157917499542, "learning_rate": 2.2519206165083628e-09, "loss": 0.0035, "step": 299140 }, { "epoch": 190.66284257488846, "grad_norm": 0.4483194947242737, "learning_rate": 2.199915347211867e-09, "loss": 0.0062, "step": 299150 }, { "epoch": 190.66921606118547, "grad_norm": 0.24789638817310333, "learning_rate": 2.148517598388189e-09, "loss": 0.0151, "step": 299160 }, { "epoch": 190.67558954748247, "grad_norm": 0.12132579833269119, "learning_rate": 2.097727370670155e-09, "loss": 0.0066, "step": 299170 }, { "epoch": 190.68196303377948, "grad_norm": 0.08883944153785706, "learning_rate": 2.0475446646683884e-09, "loss": 0.0029, "step": 299180 }, { "epoch": 190.6883365200765, "grad_norm": 0.078717440366745, "learning_rate": 1.9979694809935114e-09, "loss": 0.0034, "step": 299190 }, { "epoch": 190.6947100063735, "grad_norm": 0.33708080649375916, "learning_rate": 1.9490018202505956e-09, "loss": 0.0058, "step": 299200 }, { "epoch": 190.70108349267048, "grad_norm": 0.2094375044107437, "learning_rate": 1.9006416830336107e-09, "loss": 0.0041, "step": 299210 }, { "epoch": 190.70745697896749, "grad_norm": 0.13471972942352295, "learning_rate": 1.852889069925423e-09, "loss": 0.0072, "step": 299220 }, { "epoch": 190.7138304652645, "grad_norm": 0.11414216458797455, "learning_rate": 1.8057439815144516e-09, "loss": 0.0063, "step": 299230 }, { "epoch": 190.7202039515615, "grad_norm": 0.550111711025238, "learning_rate": 1.759206418372461e-09, "loss": 0.0077, "step": 299240 }, { "epoch": 190.7265774378585, "grad_norm": 0.5713498592376709, "learning_rate": 1.7132763810601137e-09, "loss": 0.0075, "step": 299250 }, { "epoch": 190.73295092415552, "grad_norm": 0.09618548303842545, "learning_rate": 1.6679538701325215e-09, "loss": 0.0034, "step": 299260 }, { "epoch": 190.73932441045253, "grad_norm": 0.054806314408779144, "learning_rate": 1.6232388861503466e-09, "loss": 0.0054, "step": 299270 }, { "epoch": 190.74569789674953, "grad_norm": 0.07290177792310715, "learning_rate": 1.5791314296520476e-09, "loss": 0.0033, "step": 299280 }, { "epoch": 190.7520713830465, "grad_norm": 0.09063385426998138, "learning_rate": 1.5356315011760824e-09, "loss": 0.0045, "step": 299290 }, { "epoch": 190.75844486934352, "grad_norm": 0.12616568803787231, "learning_rate": 1.492739101244256e-09, "loss": 0.0056, "step": 299300 }, { "epoch": 190.76481835564053, "grad_norm": 0.10972274839878082, "learning_rate": 1.4504542303839242e-09, "loss": 0.0038, "step": 299310 }, { "epoch": 190.77119184193754, "grad_norm": 0.08713392913341522, "learning_rate": 1.4087768891057896e-09, "loss": 0.0031, "step": 299320 }, { "epoch": 190.77756532823454, "grad_norm": 0.20210908353328705, "learning_rate": 1.3677070779205548e-09, "loss": 0.0059, "step": 299330 }, { "epoch": 190.78393881453155, "grad_norm": 0.17630724608898163, "learning_rate": 1.327244797322269e-09, "loss": 0.005, "step": 299340 }, { "epoch": 190.79031230082856, "grad_norm": 0.1653313785791397, "learning_rate": 1.2873900478049816e-09, "loss": 0.0104, "step": 299350 }, { "epoch": 190.79668578712557, "grad_norm": 0.14582833647727966, "learning_rate": 1.2481428298516396e-09, "loss": 0.0034, "step": 299360 }, { "epoch": 190.80305927342255, "grad_norm": 0.18078850209712982, "learning_rate": 1.2095031439396387e-09, "loss": 0.0021, "step": 299370 }, { "epoch": 190.80943275971956, "grad_norm": 0.06810852140188217, "learning_rate": 1.1714709905408239e-09, "loss": 0.0099, "step": 299380 }, { "epoch": 190.81580624601656, "grad_norm": 0.057779572904109955, "learning_rate": 1.1340463701103864e-09, "loss": 0.0072, "step": 299390 }, { "epoch": 190.82217973231357, "grad_norm": 0.2715069055557251, "learning_rate": 1.0972292831146202e-09, "loss": 0.0081, "step": 299400 }, { "epoch": 190.82855321861058, "grad_norm": 0.18574830889701843, "learning_rate": 1.0610197299920632e-09, "loss": 0.009, "step": 299410 }, { "epoch": 190.8349267049076, "grad_norm": 0.255311518907547, "learning_rate": 1.0254177111812535e-09, "loss": 0.005, "step": 299420 }, { "epoch": 190.8413001912046, "grad_norm": 0.0776253193616867, "learning_rate": 9.904232271207291e-10, "loss": 0.0067, "step": 299430 }, { "epoch": 190.8476736775016, "grad_norm": 0.07140463590621948, "learning_rate": 9.560362782379261e-10, "loss": 0.0045, "step": 299440 }, { "epoch": 190.8540471637986, "grad_norm": 0.13913467526435852, "learning_rate": 9.222568649380759e-10, "loss": 0.0023, "step": 299450 }, { "epoch": 190.8604206500956, "grad_norm": 0.34046614170074463, "learning_rate": 8.89084987648614e-10, "loss": 0.0052, "step": 299460 }, { "epoch": 190.8667941363926, "grad_norm": 0.35147544741630554, "learning_rate": 8.565206467581189e-10, "loss": 0.0042, "step": 299470 }, { "epoch": 190.8731676226896, "grad_norm": 0.05638090893626213, "learning_rate": 8.245638426662705e-10, "loss": 0.009, "step": 299480 }, { "epoch": 190.87954110898661, "grad_norm": 0.07813289761543274, "learning_rate": 7.932145757671982e-10, "loss": 0.0034, "step": 299490 }, { "epoch": 190.88591459528362, "grad_norm": 0.1334853321313858, "learning_rate": 7.624728464383779e-10, "loss": 0.006, "step": 299500 }, { "epoch": 190.89228808158063, "grad_norm": 0.11588063836097717, "learning_rate": 7.32338655046183e-10, "loss": 0.0067, "step": 299510 }, { "epoch": 190.89866156787764, "grad_norm": 0.10758955031633377, "learning_rate": 7.028120019680895e-10, "loss": 0.0049, "step": 299520 }, { "epoch": 190.90503505417465, "grad_norm": 0.08821549266576767, "learning_rate": 6.738928875593686e-10, "loss": 0.0046, "step": 299530 }, { "epoch": 190.91140854047163, "grad_norm": 0.2237098067998886, "learning_rate": 6.455813121697408e-10, "loss": 0.004, "step": 299540 }, { "epoch": 190.91778202676863, "grad_norm": 0.10068634897470474, "learning_rate": 6.178772761378237e-10, "loss": 0.007, "step": 299550 }, { "epoch": 190.92415551306564, "grad_norm": 0.123094841837883, "learning_rate": 5.90780779813338e-10, "loss": 0.0063, "step": 299560 }, { "epoch": 190.93052899936265, "grad_norm": 0.12959125638008118, "learning_rate": 5.642918235126971e-10, "loss": 0.0067, "step": 299570 }, { "epoch": 190.93690248565966, "grad_norm": 0.16912412643432617, "learning_rate": 5.384104075634166e-10, "loss": 0.0022, "step": 299580 }, { "epoch": 190.94327597195667, "grad_norm": 0.18281897902488708, "learning_rate": 5.131365322819104e-10, "loss": 0.0049, "step": 299590 }, { "epoch": 190.94964945825367, "grad_norm": 0.32844460010528564, "learning_rate": 4.884701979734896e-10, "loss": 0.0115, "step": 299600 }, { "epoch": 190.95602294455068, "grad_norm": 0.06573081016540527, "learning_rate": 4.6441140493791447e-10, "loss": 0.0077, "step": 299610 }, { "epoch": 190.96239643084766, "grad_norm": 0.05992361530661583, "learning_rate": 4.409601534693941e-10, "loss": 0.0057, "step": 299620 }, { "epoch": 190.96876991714467, "grad_norm": 0.4019034206867218, "learning_rate": 4.181164438454843e-10, "loss": 0.0125, "step": 299630 }, { "epoch": 190.97514340344168, "grad_norm": 0.14805282652378082, "learning_rate": 3.958802763492919e-10, "loss": 0.0041, "step": 299640 }, { "epoch": 190.98151688973869, "grad_norm": 0.08191647380590439, "learning_rate": 3.742516512528216e-10, "loss": 0.0047, "step": 299650 }, { "epoch": 190.9878903760357, "grad_norm": 0.05929013714194298, "learning_rate": 3.532305688169757e-10, "loss": 0.005, "step": 299660 }, { "epoch": 190.9942638623327, "grad_norm": 0.30080413818359375, "learning_rate": 3.3281702929155444e-10, "loss": 0.0037, "step": 299670 }, { "epoch": 191.0006373486297, "grad_norm": 0.1730039417743683, "learning_rate": 3.1301103293746024e-10, "loss": 0.003, "step": 299680 }, { "epoch": 191.00701083492672, "grad_norm": 0.17973355948925018, "learning_rate": 2.9381257998228883e-10, "loss": 0.0077, "step": 299690 }, { "epoch": 191.0133843212237, "grad_norm": 0.12715354561805725, "learning_rate": 2.7522167067028927e-10, "loss": 0.004, "step": 299700 }, { "epoch": 191.0197578075207, "grad_norm": 0.126768097281456, "learning_rate": 2.57238305217955e-10, "loss": 0.0146, "step": 299710 }, { "epoch": 191.0261312938177, "grad_norm": 0.46471333503723145, "learning_rate": 2.398624838473307e-10, "loss": 0.0076, "step": 299720 }, { "epoch": 191.03250478011472, "grad_norm": 0.36523085832595825, "learning_rate": 2.2309420676935866e-10, "loss": 0.0048, "step": 299730 }, { "epoch": 191.03887826641173, "grad_norm": 0.18512141704559326, "learning_rate": 2.069334741894302e-10, "loss": 0.0061, "step": 299740 }, { "epoch": 191.04525175270874, "grad_norm": 0.08375099301338196, "learning_rate": 1.9138028630738547e-10, "loss": 0.0048, "step": 299750 }, { "epoch": 191.05162523900574, "grad_norm": 0.07719286531209946, "learning_rate": 1.7643464330086012e-10, "loss": 0.0045, "step": 299760 }, { "epoch": 191.05799872530275, "grad_norm": 0.5265268683433533, "learning_rate": 1.620965453641432e-10, "loss": 0.0067, "step": 299770 }, { "epoch": 191.06437221159973, "grad_norm": 0.2872610092163086, "learning_rate": 1.4836599266376815e-10, "loss": 0.0041, "step": 299780 }, { "epoch": 191.07074569789674, "grad_norm": 0.07976324111223221, "learning_rate": 1.3524298536626844e-10, "loss": 0.0079, "step": 299790 }, { "epoch": 191.07711918419375, "grad_norm": 0.22643019258975983, "learning_rate": 1.227275236326264e-10, "loss": 0.0033, "step": 299800 }, { "epoch": 191.08349267049076, "grad_norm": 0.04248817637562752, "learning_rate": 1.1081960761827326e-10, "loss": 0.005, "step": 299810 }, { "epoch": 191.08986615678776, "grad_norm": 0.5904961824417114, "learning_rate": 9.951923746198687e-11, "loss": 0.0078, "step": 299820 }, { "epoch": 191.09623964308477, "grad_norm": 0.10548438131809235, "learning_rate": 8.882641330809626e-11, "loss": 0.0044, "step": 299830 }, { "epoch": 191.10261312938178, "grad_norm": 0.18525660037994385, "learning_rate": 7.874113528427706e-11, "loss": 0.0054, "step": 299840 }, { "epoch": 191.1089866156788, "grad_norm": 0.100853830575943, "learning_rate": 6.92634035071027e-11, "loss": 0.0078, "step": 299850 }, { "epoch": 191.11536010197577, "grad_norm": 0.06576357781887054, "learning_rate": 6.039321809869769e-11, "loss": 0.0042, "step": 299860 }, { "epoch": 191.12173358827278, "grad_norm": 0.135863795876503, "learning_rate": 5.213057915898212e-11, "loss": 0.0029, "step": 299870 }, { "epoch": 191.12810707456978, "grad_norm": 0.04510011151432991, "learning_rate": 4.447548679897828e-11, "loss": 0.0024, "step": 299880 }, { "epoch": 191.1344805608668, "grad_norm": 0.06996582448482513, "learning_rate": 3.7427941107504025e-11, "loss": 0.0031, "step": 299890 }, { "epoch": 191.1408540471638, "grad_norm": 0.15793561935424805, "learning_rate": 3.0987942167826076e-11, "loss": 0.0037, "step": 299900 }, { "epoch": 191.1472275334608, "grad_norm": 0.09069343656301498, "learning_rate": 2.5155490057660047e-11, "loss": 0.0052, "step": 299910 }, { "epoch": 191.15360101975782, "grad_norm": 0.32630816102027893, "learning_rate": 1.993058484917043e-11, "loss": 0.0045, "step": 299920 }, { "epoch": 191.15997450605482, "grad_norm": 0.09634612500667572, "learning_rate": 1.531322660897061e-11, "loss": 0.0037, "step": 299930 }, { "epoch": 191.1663479923518, "grad_norm": 0.19892928004264832, "learning_rate": 1.1303415387020621e-11, "loss": 0.0042, "step": 299940 }, { "epoch": 191.1727214786488, "grad_norm": 0.3828645646572113, "learning_rate": 7.901151238831616e-12, "loss": 0.0051, "step": 299950 }, { "epoch": 191.17909496494582, "grad_norm": 0.13747739791870117, "learning_rate": 5.106434197710286e-12, "loss": 0.0036, "step": 299960 }, { "epoch": 191.18546845124283, "grad_norm": 0.08791064471006393, "learning_rate": 2.91926430806555e-12, "loss": 0.0034, "step": 299970 }, { "epoch": 191.19184193753983, "grad_norm": 0.5657444000244141, "learning_rate": 1.339641586550755e-12, "loss": 0.0075, "step": 299980 }, { "epoch": 191.19821542383684, "grad_norm": 0.08235937356948853, "learning_rate": 3.675660609214759e-13, "loss": 0.0063, "step": 299990 }, { "epoch": 191.20458891013385, "grad_norm": 0.07190953940153122, "learning_rate": 3.0377367288281224e-15, "loss": 0.0073, "step": 300000 }, { "epoch": 191.20458891013385, "step": 300000, "total_flos": 0.0, "train_loss": 0.019851835373515883, "train_runtime": 124705.1687, "train_samples_per_second": 19.245, "train_steps_per_second": 2.406 } ], "logging_steps": 10, "max_steps": 300000, "num_input_tokens_seen": 0, "num_train_epochs": 192, "save_steps": 150000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 8, "trial_name": null, "trial_params": null }