MhaWay commited on
Commit
77877d8
·
1 Parent(s): afcea13

Veronica-Polymorphic 551M — Pretrained v1

Browse files
added_tokens.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "<|assistant|>": 50261,
3
+ "<|bos|>": 50257,
4
+ "<|code|>": 50268,
5
+ "<|document|>": 50264,
6
+ "<|endofchunk|>": 50270,
7
+ "<|observation|>": 50263,
8
+ "<|pad|>": 50258,
9
+ "<|retrieval|>": 50265,
10
+ "<|scratchpad|>": 50266,
11
+ "<|sep|>": 50269,
12
+ "<|system|>": 50259,
13
+ "<|thought|>": 50267,
14
+ "<|tool|>": 50262,
15
+ "<|user|>": 50260
16
+ }
config.json ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_comment": "Ultra-deep reasoning config: 24L/12H/768d, mlp_mult=4. Maximum reasoning depth for RAG + multi-step inference. ~11-12GB VRAM.",
3
+ "architectures": [
4
+ "VeronicaForCausalLM"
5
+ ],
6
+ "bos_token_id": 50257,
7
+ "dropout": 0.0,
8
+ "dtype": "bfloat16",
9
+ "eos_token_id": 50256,
10
+ "gradient_checkpointing": true,
11
+ "hidden_size": 768,
12
+ "layer_norm_epsilon": 1e-05,
13
+ "max_position_embeddings": 2048,
14
+ "mlp_mult": 4,
15
+ "model_type": "veronica",
16
+ "n_embd": 768,
17
+ "n_head": 12,
18
+ "n_layer": 24,
19
+ "num_attention_heads": 12,
20
+ "num_funcs": 3,
21
+ "num_hidden_layers": 24,
22
+ "pad_token_id": 50258,
23
+ "rope_theta": 10000.0,
24
+ "router_aux_weight": 0.015999763124389305,
25
+ "router_dim": 128,
26
+ "router_tau": 1.400028804954452,
27
+ "transformers_version": "4.57.0.dev0",
28
+ "use_cache": false,
29
+ "use_channel_attention": false,
30
+ "use_flash_attn": true,
31
+ "vocab_size": 50271
32
+ }
generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 50257,
4
+ "eos_token_id": [
5
+ 50256
6
+ ],
7
+ "pad_token_id": 50258,
8
+ "transformers_version": "4.57.0.dev0"
9
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:809ed2aff4a95bff277106992f002be0e789ad07b108eb62ca33dcd9e6952879
3
+ size 1102824111
trainer_state.json ADDED
@@ -0,0 +1,2657 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 0.9999925976919604,
6
+ "eval_steps": 1000,
7
+ "global_step": 33773,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.002960923215858705,
14
+ "grad_norm": 41.32149887084961,
15
+ "learning_rate": 1.98e-06,
16
+ "loss": 87.7468,
17
+ "step": 100
18
+ },
19
+ {
20
+ "epoch": 0.00592184643171741,
21
+ "grad_norm": 39.51910400390625,
22
+ "learning_rate": 3.98e-06,
23
+ "loss": 85.766,
24
+ "step": 200
25
+ },
26
+ {
27
+ "epoch": 0.008882769647576115,
28
+ "grad_norm": 26.557823181152344,
29
+ "learning_rate": 5.98e-06,
30
+ "loss": 80.1376,
31
+ "step": 300
32
+ },
33
+ {
34
+ "epoch": 0.01184369286343482,
35
+ "grad_norm": 21.80652618408203,
36
+ "learning_rate": 7.98e-06,
37
+ "loss": 74.3306,
38
+ "step": 400
39
+ },
40
+ {
41
+ "epoch": 0.014804616079293524,
42
+ "grad_norm": 16.2612247467041,
43
+ "learning_rate": 9.980000000000001e-06,
44
+ "loss": 72.3247,
45
+ "step": 500
46
+ },
47
+ {
48
+ "epoch": 0.01776553929515223,
49
+ "grad_norm": 17.281190872192383,
50
+ "learning_rate": 1.198e-05,
51
+ "loss": 71.0703,
52
+ "step": 600
53
+ },
54
+ {
55
+ "epoch": 0.020726462511010933,
56
+ "grad_norm": 16.140579223632812,
57
+ "learning_rate": 1.3980000000000002e-05,
58
+ "loss": 69.5824,
59
+ "step": 700
60
+ },
61
+ {
62
+ "epoch": 0.02368738572686964,
63
+ "grad_norm": 13.456184387207031,
64
+ "learning_rate": 1.598e-05,
65
+ "loss": 67.2135,
66
+ "step": 800
67
+ },
68
+ {
69
+ "epoch": 0.026648308942728342,
70
+ "grad_norm": 11.778711318969727,
71
+ "learning_rate": 1.798e-05,
72
+ "loss": 65.1397,
73
+ "step": 900
74
+ },
75
+ {
76
+ "epoch": 0.02960923215858705,
77
+ "grad_norm": 11.965922355651855,
78
+ "learning_rate": 1.9980000000000002e-05,
79
+ "loss": 63.204,
80
+ "step": 1000
81
+ },
82
+ {
83
+ "epoch": 0.02960923215858705,
84
+ "eval_loss": 7.691287040710449,
85
+ "eval_runtime": 37.9485,
86
+ "eval_samples_per_second": 28.486,
87
+ "eval_steps_per_second": 7.141,
88
+ "step": 1000
89
+ },
90
+ {
91
+ "epoch": 0.032570155374445756,
92
+ "grad_norm": 8.482980728149414,
93
+ "learning_rate": 2.198e-05,
94
+ "loss": 60.9402,
95
+ "step": 1100
96
+ },
97
+ {
98
+ "epoch": 0.03553107859030446,
99
+ "grad_norm": 49.948341369628906,
100
+ "learning_rate": 2.398e-05,
101
+ "loss": 58.9346,
102
+ "step": 1200
103
+ },
104
+ {
105
+ "epoch": 0.03849200180616316,
106
+ "grad_norm": 10.039616584777832,
107
+ "learning_rate": 2.5980000000000002e-05,
108
+ "loss": 57.309,
109
+ "step": 1300
110
+ },
111
+ {
112
+ "epoch": 0.041452925022021865,
113
+ "grad_norm": 8.930785179138184,
114
+ "learning_rate": 2.798e-05,
115
+ "loss": 56.3741,
116
+ "step": 1400
117
+ },
118
+ {
119
+ "epoch": 0.04441384823788057,
120
+ "grad_norm": 8.14844036102295,
121
+ "learning_rate": 2.998e-05,
122
+ "loss": 55.8969,
123
+ "step": 1500
124
+ },
125
+ {
126
+ "epoch": 0.04737477145373928,
127
+ "grad_norm": 16.170246124267578,
128
+ "learning_rate": 3.198e-05,
129
+ "loss": 55.12,
130
+ "step": 1600
131
+ },
132
+ {
133
+ "epoch": 0.05033569466959798,
134
+ "grad_norm": 10.525145530700684,
135
+ "learning_rate": 3.398e-05,
136
+ "loss": 54.5077,
137
+ "step": 1700
138
+ },
139
+ {
140
+ "epoch": 0.053296617885456685,
141
+ "grad_norm": 29.427160263061523,
142
+ "learning_rate": 3.5980000000000004e-05,
143
+ "loss": 54.2087,
144
+ "step": 1800
145
+ },
146
+ {
147
+ "epoch": 0.05625754110131539,
148
+ "grad_norm": 47.125083923339844,
149
+ "learning_rate": 3.798e-05,
150
+ "loss": 53.4388,
151
+ "step": 1900
152
+ },
153
+ {
154
+ "epoch": 0.0592184643171741,
155
+ "grad_norm": 10.07633113861084,
156
+ "learning_rate": 3.998e-05,
157
+ "loss": 53.5798,
158
+ "step": 2000
159
+ },
160
+ {
161
+ "epoch": 0.0592184643171741,
162
+ "eval_loss": 6.5744218826293945,
163
+ "eval_runtime": 38.8407,
164
+ "eval_samples_per_second": 27.832,
165
+ "eval_steps_per_second": 6.977,
166
+ "step": 2000
167
+ },
168
+ {
169
+ "epoch": 0.0621793875330328,
170
+ "grad_norm": 10.918025970458984,
171
+ "learning_rate": 4.198e-05,
172
+ "loss": 52.5816,
173
+ "step": 2100
174
+ },
175
+ {
176
+ "epoch": 0.06514031074889151,
177
+ "grad_norm": 12.68106460571289,
178
+ "learning_rate": 4.398e-05,
179
+ "loss": 52.6681,
180
+ "step": 2200
181
+ },
182
+ {
183
+ "epoch": 0.06810123396475021,
184
+ "grad_norm": 15.503605842590332,
185
+ "learning_rate": 4.598e-05,
186
+ "loss": 52.5443,
187
+ "step": 2300
188
+ },
189
+ {
190
+ "epoch": 0.07106215718060892,
191
+ "grad_norm": 10.995290756225586,
192
+ "learning_rate": 4.798e-05,
193
+ "loss": 51.4841,
194
+ "step": 2400
195
+ },
196
+ {
197
+ "epoch": 0.07402308039646761,
198
+ "grad_norm": 22.71038055419922,
199
+ "learning_rate": 4.998e-05,
200
+ "loss": 51.05,
201
+ "step": 2500
202
+ },
203
+ {
204
+ "epoch": 0.07698400361232632,
205
+ "grad_norm": 11.035310745239258,
206
+ "learning_rate": 5.198e-05,
207
+ "loss": 50.8273,
208
+ "step": 2600
209
+ },
210
+ {
211
+ "epoch": 0.07994492682818503,
212
+ "grad_norm": 14.107246398925781,
213
+ "learning_rate": 5.398e-05,
214
+ "loss": 50.9043,
215
+ "step": 2700
216
+ },
217
+ {
218
+ "epoch": 0.08290585004404373,
219
+ "grad_norm": 11.570377349853516,
220
+ "learning_rate": 5.598e-05,
221
+ "loss": 50.4862,
222
+ "step": 2800
223
+ },
224
+ {
225
+ "epoch": 0.08586677325990244,
226
+ "grad_norm": 16.12681770324707,
227
+ "learning_rate": 5.7980000000000004e-05,
228
+ "loss": 50.0248,
229
+ "step": 2900
230
+ },
231
+ {
232
+ "epoch": 0.08882769647576114,
233
+ "grad_norm": 14.925129890441895,
234
+ "learning_rate": 5.9980000000000005e-05,
235
+ "loss": 49.789,
236
+ "step": 3000
237
+ },
238
+ {
239
+ "epoch": 0.08882769647576114,
240
+ "eval_loss": 6.141844272613525,
241
+ "eval_runtime": 38.2773,
242
+ "eval_samples_per_second": 28.241,
243
+ "eval_steps_per_second": 7.08,
244
+ "step": 3000
245
+ },
246
+ {
247
+ "epoch": 0.09178861969161985,
248
+ "grad_norm": 16.748519897460938,
249
+ "learning_rate": 6.198e-05,
250
+ "loss": 49.4085,
251
+ "step": 3100
252
+ },
253
+ {
254
+ "epoch": 0.09474954290747856,
255
+ "grad_norm": 12.9891939163208,
256
+ "learning_rate": 6.398000000000001e-05,
257
+ "loss": 49.1003,
258
+ "step": 3200
259
+ },
260
+ {
261
+ "epoch": 0.09771046612333725,
262
+ "grad_norm": 11.456365585327148,
263
+ "learning_rate": 6.598e-05,
264
+ "loss": 48.6685,
265
+ "step": 3300
266
+ },
267
+ {
268
+ "epoch": 0.10067138933919596,
269
+ "grad_norm": 19.153154373168945,
270
+ "learning_rate": 6.798e-05,
271
+ "loss": 48.3342,
272
+ "step": 3400
273
+ },
274
+ {
275
+ "epoch": 0.10363231255505466,
276
+ "grad_norm": 13.501580238342285,
277
+ "learning_rate": 6.998e-05,
278
+ "loss": 47.7641,
279
+ "step": 3500
280
+ },
281
+ {
282
+ "epoch": 0.10659323577091337,
283
+ "grad_norm": 14.034686088562012,
284
+ "learning_rate": 7.198e-05,
285
+ "loss": 47.8053,
286
+ "step": 3600
287
+ },
288
+ {
289
+ "epoch": 0.10955415898677208,
290
+ "grad_norm": 14.023058891296387,
291
+ "learning_rate": 7.398e-05,
292
+ "loss": 47.4925,
293
+ "step": 3700
294
+ },
295
+ {
296
+ "epoch": 0.11251508220263078,
297
+ "grad_norm": 16.410221099853516,
298
+ "learning_rate": 7.598e-05,
299
+ "loss": 47.1501,
300
+ "step": 3800
301
+ },
302
+ {
303
+ "epoch": 0.11547600541848949,
304
+ "grad_norm": 17.547571182250977,
305
+ "learning_rate": 7.798000000000001e-05,
306
+ "loss": 47.3132,
307
+ "step": 3900
308
+ },
309
+ {
310
+ "epoch": 0.1184369286343482,
311
+ "grad_norm": 11.753161430358887,
312
+ "learning_rate": 7.998e-05,
313
+ "loss": 47.071,
314
+ "step": 4000
315
+ },
316
+ {
317
+ "epoch": 0.1184369286343482,
318
+ "eval_loss": 5.719655990600586,
319
+ "eval_runtime": 38.3575,
320
+ "eval_samples_per_second": 28.182,
321
+ "eval_steps_per_second": 7.065,
322
+ "step": 4000
323
+ },
324
+ {
325
+ "epoch": 0.12139785185020689,
326
+ "grad_norm": 51.37761306762695,
327
+ "learning_rate": 8.198000000000001e-05,
328
+ "loss": 46.1728,
329
+ "step": 4100
330
+ },
331
+ {
332
+ "epoch": 0.1243587750660656,
333
+ "grad_norm": 12.61581802368164,
334
+ "learning_rate": 8.398e-05,
335
+ "loss": 45.9941,
336
+ "step": 4200
337
+ },
338
+ {
339
+ "epoch": 0.1273196982819243,
340
+ "grad_norm": 12.667135238647461,
341
+ "learning_rate": 8.598e-05,
342
+ "loss": 46.1649,
343
+ "step": 4300
344
+ },
345
+ {
346
+ "epoch": 0.13028062149778302,
347
+ "grad_norm": 10.368157386779785,
348
+ "learning_rate": 8.798e-05,
349
+ "loss": 45.9172,
350
+ "step": 4400
351
+ },
352
+ {
353
+ "epoch": 0.13324154471364172,
354
+ "grad_norm": 12.5702543258667,
355
+ "learning_rate": 8.998e-05,
356
+ "loss": 45.5541,
357
+ "step": 4500
358
+ },
359
+ {
360
+ "epoch": 0.13620246792950041,
361
+ "grad_norm": 10.535380363464355,
362
+ "learning_rate": 9.198e-05,
363
+ "loss": 44.45,
364
+ "step": 4600
365
+ },
366
+ {
367
+ "epoch": 0.1391633911453591,
368
+ "grad_norm": 11.035446166992188,
369
+ "learning_rate": 9.398e-05,
370
+ "loss": 44.3243,
371
+ "step": 4700
372
+ },
373
+ {
374
+ "epoch": 0.14212431436121784,
375
+ "grad_norm": 10.718255043029785,
376
+ "learning_rate": 9.598e-05,
377
+ "loss": 43.8677,
378
+ "step": 4800
379
+ },
380
+ {
381
+ "epoch": 0.14508523757707653,
382
+ "grad_norm": 9.816108703613281,
383
+ "learning_rate": 9.798000000000001e-05,
384
+ "loss": 43.6948,
385
+ "step": 4900
386
+ },
387
+ {
388
+ "epoch": 0.14804616079293523,
389
+ "grad_norm": 9.837696075439453,
390
+ "learning_rate": 9.998000000000002e-05,
391
+ "loss": 43.6361,
392
+ "step": 5000
393
+ },
394
+ {
395
+ "epoch": 0.14804616079293523,
396
+ "eval_loss": 5.329010486602783,
397
+ "eval_runtime": 38.0693,
398
+ "eval_samples_per_second": 28.396,
399
+ "eval_steps_per_second": 7.119,
400
+ "step": 5000
401
+ },
402
+ {
403
+ "epoch": 0.15100708400879395,
404
+ "grad_norm": 10.140490531921387,
405
+ "learning_rate": 0.00010198,
406
+ "loss": 42.8106,
407
+ "step": 5100
408
+ },
409
+ {
410
+ "epoch": 0.15396800722465265,
411
+ "grad_norm": 9.330647468566895,
412
+ "learning_rate": 0.00010398,
413
+ "loss": 42.4439,
414
+ "step": 5200
415
+ },
416
+ {
417
+ "epoch": 0.15692893044051134,
418
+ "grad_norm": 9.038117408752441,
419
+ "learning_rate": 0.00010598,
420
+ "loss": 41.8943,
421
+ "step": 5300
422
+ },
423
+ {
424
+ "epoch": 0.15988985365637007,
425
+ "grad_norm": 10.28738021850586,
426
+ "learning_rate": 0.00010798,
427
+ "loss": 41.5117,
428
+ "step": 5400
429
+ },
430
+ {
431
+ "epoch": 0.16285077687222876,
432
+ "grad_norm": 9.869328498840332,
433
+ "learning_rate": 0.00010998,
434
+ "loss": 41.3489,
435
+ "step": 5500
436
+ },
437
+ {
438
+ "epoch": 0.16581170008808746,
439
+ "grad_norm": 10.985088348388672,
440
+ "learning_rate": 0.00011198000000000001,
441
+ "loss": 40.7585,
442
+ "step": 5600
443
+ },
444
+ {
445
+ "epoch": 0.16877262330394618,
446
+ "grad_norm": 11.46516227722168,
447
+ "learning_rate": 0.00011398,
448
+ "loss": 40.2893,
449
+ "step": 5700
450
+ },
451
+ {
452
+ "epoch": 0.17173354651980488,
453
+ "grad_norm": 9.691688537597656,
454
+ "learning_rate": 0.00011598000000000001,
455
+ "loss": 40.0513,
456
+ "step": 5800
457
+ },
458
+ {
459
+ "epoch": 0.17469446973566358,
460
+ "grad_norm": 10.819178581237793,
461
+ "learning_rate": 0.00011798,
462
+ "loss": 39.986,
463
+ "step": 5900
464
+ },
465
+ {
466
+ "epoch": 0.17765539295152227,
467
+ "grad_norm": 9.694029808044434,
468
+ "learning_rate": 0.00011998,
469
+ "loss": 39.3918,
470
+ "step": 6000
471
+ },
472
+ {
473
+ "epoch": 0.17765539295152227,
474
+ "eval_loss": 4.900777339935303,
475
+ "eval_runtime": 38.052,
476
+ "eval_samples_per_second": 28.409,
477
+ "eval_steps_per_second": 7.122,
478
+ "step": 6000
479
+ },
480
+ {
481
+ "epoch": 0.180616316167381,
482
+ "grad_norm": 9.988055229187012,
483
+ "learning_rate": 0.00011999900481764066,
484
+ "loss": 39.336,
485
+ "step": 6100
486
+ },
487
+ {
488
+ "epoch": 0.1835772393832397,
489
+ "grad_norm": 13.242379188537598,
490
+ "learning_rate": 0.00011999597899343296,
491
+ "loss": 39.0612,
492
+ "step": 6200
493
+ },
494
+ {
495
+ "epoch": 0.1865381625990984,
496
+ "grad_norm": 13.935639381408691,
497
+ "learning_rate": 0.00011999092252825071,
498
+ "loss": 38.585,
499
+ "step": 6300
500
+ },
501
+ {
502
+ "epoch": 0.1894990858149571,
503
+ "grad_norm": 9.50368881225586,
504
+ "learning_rate": 0.00011998383559323646,
505
+ "loss": 38.3112,
506
+ "step": 6400
507
+ },
508
+ {
509
+ "epoch": 0.1924600090308158,
510
+ "grad_norm": 10.878887176513672,
511
+ "learning_rate": 0.00011997471842825661,
512
+ "loss": 38.3219,
513
+ "step": 6500
514
+ },
515
+ {
516
+ "epoch": 0.1954209322466745,
517
+ "grad_norm": 10.517402648925781,
518
+ "learning_rate": 0.00011996357134189334,
519
+ "loss": 37.8246,
520
+ "step": 6600
521
+ },
522
+ {
523
+ "epoch": 0.19838185546253323,
524
+ "grad_norm": 10.922290802001953,
525
+ "learning_rate": 0.0001199503947114341,
526
+ "loss": 37.6387,
527
+ "step": 6700
528
+ },
529
+ {
530
+ "epoch": 0.20134277867839193,
531
+ "grad_norm": 11.845630645751953,
532
+ "learning_rate": 0.00011993518898285887,
533
+ "loss": 37.8343,
534
+ "step": 6800
535
+ },
536
+ {
537
+ "epoch": 0.20430370189425062,
538
+ "grad_norm": 8.628484725952148,
539
+ "learning_rate": 0.00011991795467082508,
540
+ "loss": 37.5011,
541
+ "step": 6900
542
+ },
543
+ {
544
+ "epoch": 0.20726462511010932,
545
+ "grad_norm": 9.489052772521973,
546
+ "learning_rate": 0.00011989869235865012,
547
+ "loss": 37.132,
548
+ "step": 7000
549
+ },
550
+ {
551
+ "epoch": 0.20726462511010932,
552
+ "eval_loss": 4.595886707305908,
553
+ "eval_runtime": 38.0814,
554
+ "eval_samples_per_second": 28.387,
555
+ "eval_steps_per_second": 7.116,
556
+ "step": 7000
557
+ },
558
+ {
559
+ "epoch": 0.21022554832596804,
560
+ "grad_norm": 9.687568664550781,
561
+ "learning_rate": 0.00011987740269829175,
562
+ "loss": 36.9362,
563
+ "step": 7100
564
+ },
565
+ {
566
+ "epoch": 0.21318647154182674,
567
+ "grad_norm": 8.676931381225586,
568
+ "learning_rate": 0.0001198540864103258,
569
+ "loss": 37.0267,
570
+ "step": 7200
571
+ },
572
+ {
573
+ "epoch": 0.21614739475768543,
574
+ "grad_norm": 9.232645988464355,
575
+ "learning_rate": 0.00011982874428392204,
576
+ "loss": 36.5181,
577
+ "step": 7300
578
+ },
579
+ {
580
+ "epoch": 0.21910831797354416,
581
+ "grad_norm": 8.917469024658203,
582
+ "learning_rate": 0.00011980137717681727,
583
+ "loss": 36.5812,
584
+ "step": 7400
585
+ },
586
+ {
587
+ "epoch": 0.22206924118940286,
588
+ "grad_norm": 8.593257904052734,
589
+ "learning_rate": 0.0001197719860152864,
590
+ "loss": 36.0672,
591
+ "step": 7500
592
+ },
593
+ {
594
+ "epoch": 0.22503016440526155,
595
+ "grad_norm": 10.630696296691895,
596
+ "learning_rate": 0.00011974057179411103,
597
+ "loss": 36.2405,
598
+ "step": 7600
599
+ },
600
+ {
601
+ "epoch": 0.22799108762112028,
602
+ "grad_norm": 9.975415229797363,
603
+ "learning_rate": 0.00011970713557654582,
604
+ "loss": 35.9903,
605
+ "step": 7700
606
+ },
607
+ {
608
+ "epoch": 0.23095201083697897,
609
+ "grad_norm": 8.622698783874512,
610
+ "learning_rate": 0.00011967167849428251,
611
+ "loss": 35.8196,
612
+ "step": 7800
613
+ },
614
+ {
615
+ "epoch": 0.23391293405283767,
616
+ "grad_norm": 14.828067779541016,
617
+ "learning_rate": 0.00011963420174741161,
618
+ "loss": 35.7946,
619
+ "step": 7900
620
+ },
621
+ {
622
+ "epoch": 0.2368738572686964,
623
+ "grad_norm": 9.303028106689453,
624
+ "learning_rate": 0.00011959470660438173,
625
+ "loss": 35.5493,
626
+ "step": 8000
627
+ },
628
+ {
629
+ "epoch": 0.2368738572686964,
630
+ "eval_loss": 4.408100128173828,
631
+ "eval_runtime": 37.807,
632
+ "eval_samples_per_second": 28.593,
633
+ "eval_steps_per_second": 7.168,
634
+ "step": 8000
635
+ },
636
+ {
637
+ "epoch": 0.2398347804845551,
638
+ "grad_norm": 11.987268447875977,
639
+ "learning_rate": 0.00011955319440195674,
640
+ "loss": 35.6014,
641
+ "step": 8100
642
+ },
643
+ {
644
+ "epoch": 0.24279570370041378,
645
+ "grad_norm": 10.032620429992676,
646
+ "learning_rate": 0.00011950966654517043,
647
+ "loss": 35.5302,
648
+ "step": 8200
649
+ },
650
+ {
651
+ "epoch": 0.24575662691627248,
652
+ "grad_norm": 9.362653732299805,
653
+ "learning_rate": 0.00011946412450727906,
654
+ "loss": 35.2124,
655
+ "step": 8300
656
+ },
657
+ {
658
+ "epoch": 0.2487175501321312,
659
+ "grad_norm": 9.706056594848633,
660
+ "learning_rate": 0.00011941656982971138,
661
+ "loss": 34.9229,
662
+ "step": 8400
663
+ },
664
+ {
665
+ "epoch": 0.25167847334798993,
666
+ "grad_norm": 10.424148559570312,
667
+ "learning_rate": 0.00011936700412201653,
668
+ "loss": 35.1602,
669
+ "step": 8500
670
+ },
671
+ {
672
+ "epoch": 0.2546393965638486,
673
+ "grad_norm": 10.900792121887207,
674
+ "learning_rate": 0.00011931542906180957,
675
+ "loss": 34.9212,
676
+ "step": 8600
677
+ },
678
+ {
679
+ "epoch": 0.2576003197797073,
680
+ "grad_norm": 10.541563034057617,
681
+ "learning_rate": 0.00011926184639471465,
682
+ "loss": 34.8347,
683
+ "step": 8700
684
+ },
685
+ {
686
+ "epoch": 0.26056124299556604,
687
+ "grad_norm": 8.576896667480469,
688
+ "learning_rate": 0.00011920625793430596,
689
+ "loss": 34.9933,
690
+ "step": 8800
691
+ },
692
+ {
693
+ "epoch": 0.2635221662114247,
694
+ "grad_norm": 10.162493705749512,
695
+ "learning_rate": 0.00011914866556204637,
696
+ "loss": 34.3925,
697
+ "step": 8900
698
+ },
699
+ {
700
+ "epoch": 0.26648308942728344,
701
+ "grad_norm": 11.247607231140137,
702
+ "learning_rate": 0.0001190890712272237,
703
+ "loss": 34.4828,
704
+ "step": 9000
705
+ },
706
+ {
707
+ "epoch": 0.26648308942728344,
708
+ "eval_loss": 4.2549567222595215,
709
+ "eval_runtime": 37.961,
710
+ "eval_samples_per_second": 28.477,
711
+ "eval_steps_per_second": 7.139,
712
+ "step": 9000
713
+ },
714
+ {
715
+ "epoch": 0.2694440126431421,
716
+ "grad_norm": 9.189545631408691,
717
+ "learning_rate": 0.00011902747694688472,
718
+ "loss": 34.3655,
719
+ "step": 9100
720
+ },
721
+ {
722
+ "epoch": 0.27240493585900083,
723
+ "grad_norm": 11.199912071228027,
724
+ "learning_rate": 0.000118963884805767,
725
+ "loss": 34.4358,
726
+ "step": 9200
727
+ },
728
+ {
729
+ "epoch": 0.27536585907485955,
730
+ "grad_norm": 9.673705101013184,
731
+ "learning_rate": 0.00011889829695622823,
732
+ "loss": 34.3689,
733
+ "step": 9300
734
+ },
735
+ {
736
+ "epoch": 0.2783267822907182,
737
+ "grad_norm": 10.03848934173584,
738
+ "learning_rate": 0.00011883071561817344,
739
+ "loss": 33.9158,
740
+ "step": 9400
741
+ },
742
+ {
743
+ "epoch": 0.28128770550657695,
744
+ "grad_norm": 11.581180572509766,
745
+ "learning_rate": 0.00011876114307897981,
746
+ "loss": 33.992,
747
+ "step": 9500
748
+ },
749
+ {
750
+ "epoch": 0.28424862872243567,
751
+ "grad_norm": 10.81711483001709,
752
+ "learning_rate": 0.00011868958169341929,
753
+ "loss": 34.1195,
754
+ "step": 9600
755
+ },
756
+ {
757
+ "epoch": 0.28720955193829434,
758
+ "grad_norm": 9.648648262023926,
759
+ "learning_rate": 0.00011861603388357893,
760
+ "loss": 34.1664,
761
+ "step": 9700
762
+ },
763
+ {
764
+ "epoch": 0.29017047515415306,
765
+ "grad_norm": 11.37558364868164,
766
+ "learning_rate": 0.00011854050213877877,
767
+ "loss": 33.9937,
768
+ "step": 9800
769
+ },
770
+ {
771
+ "epoch": 0.2931313983700118,
772
+ "grad_norm": 9.346961975097656,
773
+ "learning_rate": 0.0001184629890154878,
774
+ "loss": 33.6917,
775
+ "step": 9900
776
+ },
777
+ {
778
+ "epoch": 0.29609232158587045,
779
+ "grad_norm": 11.31644058227539,
780
+ "learning_rate": 0.0001183834971372372,
781
+ "loss": 33.7808,
782
+ "step": 10000
783
+ },
784
+ {
785
+ "epoch": 0.29609232158587045,
786
+ "eval_loss": 4.157764434814453,
787
+ "eval_runtime": 39.9196,
788
+ "eval_samples_per_second": 27.079,
789
+ "eval_steps_per_second": 6.789,
790
+ "step": 10000
791
+ },
792
+ {
793
+ "epoch": 0.2990532448017292,
794
+ "grad_norm": 18.920991897583008,
795
+ "learning_rate": 0.00011180531798567065,
796
+ "loss": 44.0571,
797
+ "step": 10100
798
+ },
799
+ {
800
+ "epoch": 0.3020141680175879,
801
+ "grad_norm": 15.393646240234375,
802
+ "learning_rate": 0.0001116461207502148,
803
+ "loss": 39.8888,
804
+ "step": 10200
805
+ },
806
+ {
807
+ "epoch": 0.30497509123344657,
808
+ "grad_norm": 13.2774076461792,
809
+ "learning_rate": 0.00011148550761026972,
810
+ "loss": 38.6529,
811
+ "step": 10300
812
+ },
813
+ {
814
+ "epoch": 0.3079360144493053,
815
+ "grad_norm": 13.391098976135254,
816
+ "learning_rate": 0.00011132348296912578,
817
+ "loss": 37.759,
818
+ "step": 10400
819
+ },
820
+ {
821
+ "epoch": 0.310896937665164,
822
+ "grad_norm": 11.917950630187988,
823
+ "learning_rate": 0.00011116005126877037,
824
+ "loss": 37.1968,
825
+ "step": 10500
826
+ },
827
+ {
828
+ "epoch": 0.3138578608810227,
829
+ "grad_norm": 11.100213050842285,
830
+ "learning_rate": 0.0001109952169897661,
831
+ "loss": 37.213,
832
+ "step": 10600
833
+ },
834
+ {
835
+ "epoch": 0.3168187840968814,
836
+ "grad_norm": 14.579487800598145,
837
+ "learning_rate": 0.00011082898465112802,
838
+ "loss": 36.7415,
839
+ "step": 10700
840
+ },
841
+ {
842
+ "epoch": 0.31977970731274014,
843
+ "grad_norm": 11.359614372253418,
844
+ "learning_rate": 0.00011066135881019965,
845
+ "loss": 36.361,
846
+ "step": 10800
847
+ },
848
+ {
849
+ "epoch": 0.3227406305285988,
850
+ "grad_norm": 12.316486358642578,
851
+ "learning_rate": 0.00011049234406252809,
852
+ "loss": 36.0591,
853
+ "step": 10900
854
+ },
855
+ {
856
+ "epoch": 0.32570155374445753,
857
+ "grad_norm": 13.45693588256836,
858
+ "learning_rate": 0.00011032194504173804,
859
+ "loss": 35.6357,
860
+ "step": 11000
861
+ },
862
+ {
863
+ "epoch": 0.32570155374445753,
864
+ "eval_loss": 4.41249418258667,
865
+ "eval_runtime": 39.8412,
866
+ "eval_samples_per_second": 27.133,
867
+ "eval_steps_per_second": 6.802,
868
+ "step": 11000
869
+ },
870
+ {
871
+ "epoch": 0.32866247696031625,
872
+ "grad_norm": 10.018808364868164,
873
+ "learning_rate": 0.00011412090151135696,
874
+ "loss": 33.7098,
875
+ "step": 11100
876
+ },
877
+ {
878
+ "epoch": 0.3316234001761749,
879
+ "grad_norm": 10.30320930480957,
880
+ "learning_rate": 0.00011397731809339621,
881
+ "loss": 33.7831,
882
+ "step": 11200
883
+ },
884
+ {
885
+ "epoch": 0.33458432339203364,
886
+ "grad_norm": 9.650611877441406,
887
+ "learning_rate": 0.0001138320949911399,
888
+ "loss": 33.5415,
889
+ "step": 11300
890
+ },
891
+ {
892
+ "epoch": 0.33754524660789237,
893
+ "grad_norm": 8.77065372467041,
894
+ "learning_rate": 0.0001136852366160714,
895
+ "loss": 33.2261,
896
+ "step": 11400
897
+ },
898
+ {
899
+ "epoch": 0.34050616982375104,
900
+ "grad_norm": 12.062385559082031,
901
+ "learning_rate": 0.00011353674742934919,
902
+ "loss": 33.0819,
903
+ "step": 11500
904
+ },
905
+ {
906
+ "epoch": 0.34346709303960976,
907
+ "grad_norm": 10.947739601135254,
908
+ "learning_rate": 0.00011338663194167138,
909
+ "loss": 33.3451,
910
+ "step": 11600
911
+ },
912
+ {
913
+ "epoch": 0.34642801625546843,
914
+ "grad_norm": 9.377535820007324,
915
+ "learning_rate": 0.00011323489471313875,
916
+ "loss": 32.8928,
917
+ "step": 11700
918
+ },
919
+ {
920
+ "epoch": 0.34938893947132715,
921
+ "grad_norm": 8.902270317077637,
922
+ "learning_rate": 0.00011308154035311608,
923
+ "loss": 33.1756,
924
+ "step": 11800
925
+ },
926
+ {
927
+ "epoch": 0.3523498626871859,
928
+ "grad_norm": 10.434513092041016,
929
+ "learning_rate": 0.00011292657352009224,
930
+ "loss": 33.1595,
931
+ "step": 11900
932
+ },
933
+ {
934
+ "epoch": 0.35531078590304455,
935
+ "grad_norm": 11.084539413452148,
936
+ "learning_rate": 0.00011276999892153867,
937
+ "loss": 33.359,
938
+ "step": 12000
939
+ },
940
+ {
941
+ "epoch": 0.35531078590304455,
942
+ "eval_loss": 4.073917865753174,
943
+ "eval_runtime": 38.304,
944
+ "eval_samples_per_second": 28.222,
945
+ "eval_steps_per_second": 7.075,
946
+ "step": 12000
947
+ },
948
+ {
949
+ "epoch": 0.35827170911890327,
950
+ "grad_norm": 7.943862438201904,
951
+ "learning_rate": 9.143653002276282e-05,
952
+ "loss": 32.5648,
953
+ "step": 12100
954
+ },
955
+ {
956
+ "epoch": 0.361232632334762,
957
+ "grad_norm": 8.098073959350586,
958
+ "learning_rate": 9.09346201340685e-05,
959
+ "loss": 32.1551,
960
+ "step": 12200
961
+ },
962
+ {
963
+ "epoch": 0.36419355555062066,
964
+ "grad_norm": 7.46992826461792,
965
+ "learning_rate": 9.042974429385753e-05,
966
+ "loss": 32.3569,
967
+ "step": 12300
968
+ },
969
+ {
970
+ "epoch": 0.3671544787664794,
971
+ "grad_norm": 7.480947971343994,
972
+ "learning_rate": 8.992195090864853e-05,
973
+ "loss": 32.4467,
974
+ "step": 12400
975
+ },
976
+ {
977
+ "epoch": 0.3701154019823381,
978
+ "grad_norm": 7.488786220550537,
979
+ "learning_rate": 8.941128866468864e-05,
980
+ "loss": 32.4447,
981
+ "step": 12500
982
+ },
983
+ {
984
+ "epoch": 0.3730763251981968,
985
+ "grad_norm": 8.124217987060547,
986
+ "learning_rate": 8.889780652328559e-05,
987
+ "loss": 32.3657,
988
+ "step": 12600
989
+ },
990
+ {
991
+ "epoch": 0.3760372484140555,
992
+ "grad_norm": 8.322397232055664,
993
+ "learning_rate": 8.83815537161135e-05,
994
+ "loss": 31.9431,
995
+ "step": 12700
996
+ },
997
+ {
998
+ "epoch": 0.3789981716299142,
999
+ "grad_norm": 8.59915828704834,
1000
+ "learning_rate": 8.786257974049245e-05,
1001
+ "loss": 31.9211,
1002
+ "step": 12800
1003
+ },
1004
+ {
1005
+ "epoch": 0.3819590948457729,
1006
+ "grad_norm": 8.048558235168457,
1007
+ "learning_rate": 8.734093435464301e-05,
1008
+ "loss": 32.437,
1009
+ "step": 12900
1010
+ },
1011
+ {
1012
+ "epoch": 0.3849200180616316,
1013
+ "grad_norm": 7.816276550292969,
1014
+ "learning_rate": 8.681666757291531e-05,
1015
+ "loss": 32.0396,
1016
+ "step": 13000
1017
+ },
1018
+ {
1019
+ "epoch": 0.3849200180616316,
1020
+ "eval_loss": 3.9447479248046875,
1021
+ "eval_runtime": 112.3499,
1022
+ "eval_samples_per_second": 9.622,
1023
+ "eval_steps_per_second": 2.412,
1024
+ "step": 13000
1025
+ },
1026
+ {
1027
+ "epoch": 0.38788094127749034,
1028
+ "grad_norm": 8.613288879394531,
1029
+ "learning_rate": 8.628982966099388e-05,
1030
+ "loss": 31.874,
1031
+ "step": 13100
1032
+ },
1033
+ {
1034
+ "epoch": 0.390841864493349,
1035
+ "grad_norm": 7.478573799133301,
1036
+ "learning_rate": 8.576047113107821e-05,
1037
+ "loss": 31.7233,
1038
+ "step": 13200
1039
+ },
1040
+ {
1041
+ "epoch": 0.39380278770920774,
1042
+ "grad_norm": 7.845474720001221,
1043
+ "learning_rate": 8.52286427370398e-05,
1044
+ "loss": 31.628,
1045
+ "step": 13300
1046
+ },
1047
+ {
1048
+ "epoch": 0.39676371092506646,
1049
+ "grad_norm": 7.7132158279418945,
1050
+ "learning_rate": 8.469439546955592e-05,
1051
+ "loss": 31.8516,
1052
+ "step": 13400
1053
+ },
1054
+ {
1055
+ "epoch": 0.39972463414092513,
1056
+ "grad_norm": 9.245190620422363,
1057
+ "learning_rate": 8.415778055122073e-05,
1058
+ "loss": 31.8406,
1059
+ "step": 13500
1060
+ },
1061
+ {
1062
+ "epoch": 0.40268555735678385,
1063
+ "grad_norm": 8.426488876342773,
1064
+ "learning_rate": 8.361884943163423e-05,
1065
+ "loss": 31.7148,
1066
+ "step": 13600
1067
+ },
1068
+ {
1069
+ "epoch": 0.4056464805726426,
1070
+ "grad_norm": 7.879675388336182,
1071
+ "learning_rate": 8.307765378246925e-05,
1072
+ "loss": 31.9798,
1073
+ "step": 13700
1074
+ },
1075
+ {
1076
+ "epoch": 0.40860740378850124,
1077
+ "grad_norm": 8.469719886779785,
1078
+ "learning_rate": 8.253424549251735e-05,
1079
+ "loss": 31.6741,
1080
+ "step": 13800
1081
+ },
1082
+ {
1083
+ "epoch": 0.41156832700435997,
1084
+ "grad_norm": 8.198810577392578,
1085
+ "learning_rate": 8.198867666271385e-05,
1086
+ "loss": 31.6722,
1087
+ "step": 13900
1088
+ },
1089
+ {
1090
+ "epoch": 0.41452925022021864,
1091
+ "grad_norm": 7.881684303283691,
1092
+ "learning_rate": 8.144099960114239e-05,
1093
+ "loss": 31.8682,
1094
+ "step": 14000
1095
+ },
1096
+ {
1097
+ "epoch": 0.41452925022021864,
1098
+ "eval_loss": 3.904888153076172,
1099
+ "eval_runtime": 110.4703,
1100
+ "eval_samples_per_second": 9.785,
1101
+ "eval_steps_per_second": 2.453,
1102
+ "step": 14000
1103
+ },
1104
+ {
1105
+ "epoch": 0.41749017343607736,
1106
+ "grad_norm": 7.772391319274902,
1107
+ "learning_rate": 8.089126681801981e-05,
1108
+ "loss": 32.0349,
1109
+ "step": 14100
1110
+ },
1111
+ {
1112
+ "epoch": 0.4204510966519361,
1113
+ "grad_norm": 8.459504127502441,
1114
+ "learning_rate": 8.033953102066161e-05,
1115
+ "loss": 31.5844,
1116
+ "step": 14200
1117
+ },
1118
+ {
1119
+ "epoch": 0.42341201986779475,
1120
+ "grad_norm": 7.765544414520264,
1121
+ "learning_rate": 7.978584510842833e-05,
1122
+ "loss": 31.6879,
1123
+ "step": 14300
1124
+ },
1125
+ {
1126
+ "epoch": 0.4263729430836535,
1127
+ "grad_norm": 8.06749153137207,
1128
+ "learning_rate": 7.923026216765381e-05,
1129
+ "loss": 31.5893,
1130
+ "step": 14400
1131
+ },
1132
+ {
1133
+ "epoch": 0.4293338662995122,
1134
+ "grad_norm": 8.966425895690918,
1135
+ "learning_rate": 7.86728354665553e-05,
1136
+ "loss": 31.392,
1137
+ "step": 14500
1138
+ },
1139
+ {
1140
+ "epoch": 0.43229478951537087,
1141
+ "grad_norm": 8.47319221496582,
1142
+ "learning_rate": 7.81136184501262e-05,
1143
+ "loss": 31.3068,
1144
+ "step": 14600
1145
+ },
1146
+ {
1147
+ "epoch": 0.4352557127312296,
1148
+ "grad_norm": 8.642230033874512,
1149
+ "learning_rate": 7.755266473501193e-05,
1150
+ "loss": 31.5877,
1151
+ "step": 14700
1152
+ },
1153
+ {
1154
+ "epoch": 0.4382166359470883,
1155
+ "grad_norm": 8.412428855895996,
1156
+ "learning_rate": 7.699002810436915e-05,
1157
+ "loss": 31.6239,
1158
+ "step": 14800
1159
+ },
1160
+ {
1161
+ "epoch": 0.441177559162947,
1162
+ "grad_norm": 6.971558094024658,
1163
+ "learning_rate": 7.642576250270929e-05,
1164
+ "loss": 31.7946,
1165
+ "step": 14900
1166
+ },
1167
+ {
1168
+ "epoch": 0.4441384823788057,
1169
+ "grad_norm": 7.922480583190918,
1170
+ "learning_rate": 7.585992203072628e-05,
1171
+ "loss": 31.4474,
1172
+ "step": 15000
1173
+ },
1174
+ {
1175
+ "epoch": 0.4441384823788057,
1176
+ "eval_loss": 3.875948667526245,
1177
+ "eval_runtime": 109.3049,
1178
+ "eval_samples_per_second": 9.89,
1179
+ "eval_steps_per_second": 2.479,
1180
+ "step": 15000
1181
+ },
1182
+ {
1183
+ "epoch": 0.44709940559466443,
1184
+ "grad_norm": 8.747485160827637,
1185
+ "learning_rate": 7.529256094010965e-05,
1186
+ "loss": 31.6016,
1187
+ "step": 15100
1188
+ },
1189
+ {
1190
+ "epoch": 0.4500603288105231,
1191
+ "grad_norm": 8.723346710205078,
1192
+ "learning_rate": 7.472373362834283e-05,
1193
+ "loss": 31.2744,
1194
+ "step": 15200
1195
+ },
1196
+ {
1197
+ "epoch": 0.4530212520263818,
1198
+ "grad_norm": 8.310611724853516,
1199
+ "learning_rate": 7.415349463348775e-05,
1200
+ "loss": 31.7448,
1201
+ "step": 15300
1202
+ },
1203
+ {
1204
+ "epoch": 0.45598217524224055,
1205
+ "grad_norm": 8.236388206481934,
1206
+ "learning_rate": 7.358189862895577e-05,
1207
+ "loss": 30.9859,
1208
+ "step": 15400
1209
+ },
1210
+ {
1211
+ "epoch": 0.4589430984580992,
1212
+ "grad_norm": 8.104386329650879,
1213
+ "learning_rate": 7.300900041826566e-05,
1214
+ "loss": 31.1935,
1215
+ "step": 15500
1216
+ },
1217
+ {
1218
+ "epoch": 0.46190402167395794,
1219
+ "grad_norm": 8.219923973083496,
1220
+ "learning_rate": 7.243485492978928e-05,
1221
+ "loss": 30.9099,
1222
+ "step": 15600
1223
+ },
1224
+ {
1225
+ "epoch": 0.46486494488981667,
1226
+ "grad_norm": 8.872945785522461,
1227
+ "learning_rate": 7.185951721148502e-05,
1228
+ "loss": 31.3423,
1229
+ "step": 15700
1230
+ },
1231
+ {
1232
+ "epoch": 0.46782586810567534,
1233
+ "grad_norm": 8.087647438049316,
1234
+ "learning_rate": 7.128304242561999e-05,
1235
+ "loss": 31.1816,
1236
+ "step": 15800
1237
+ },
1238
+ {
1239
+ "epoch": 0.47078679132153406,
1240
+ "grad_norm": 8.805392265319824,
1241
+ "learning_rate": 7.070548584348108e-05,
1242
+ "loss": 31.0977,
1243
+ "step": 15900
1244
+ },
1245
+ {
1246
+ "epoch": 0.4737477145373928,
1247
+ "grad_norm": 8.469452857971191,
1248
+ "learning_rate": 7.012690284007577e-05,
1249
+ "loss": 31.5828,
1250
+ "step": 16000
1251
+ },
1252
+ {
1253
+ "epoch": 0.4737477145373928,
1254
+ "eval_loss": 3.8530030250549316,
1255
+ "eval_runtime": 109.275,
1256
+ "eval_samples_per_second": 9.892,
1257
+ "eval_steps_per_second": 2.48,
1258
+ "step": 16000
1259
+ },
1260
+ {
1261
+ "epoch": 0.47670863775325145,
1262
+ "grad_norm": 8.871159553527832,
1263
+ "learning_rate": 6.954734888882281e-05,
1264
+ "loss": 30.9753,
1265
+ "step": 16100
1266
+ },
1267
+ {
1268
+ "epoch": 0.4796695609691102,
1269
+ "grad_norm": 8.81116008758545,
1270
+ "learning_rate": 6.896687955623357e-05,
1271
+ "loss": 31.2067,
1272
+ "step": 16200
1273
+ },
1274
+ {
1275
+ "epoch": 0.4826304841849689,
1276
+ "grad_norm": 7.77982759475708,
1277
+ "learning_rate": 6.838555049658432e-05,
1278
+ "loss": 31.089,
1279
+ "step": 16300
1280
+ },
1281
+ {
1282
+ "epoch": 0.48559140740082757,
1283
+ "grad_norm": 8.370245933532715,
1284
+ "learning_rate": 6.780341744658044e-05,
1285
+ "loss": 30.9776,
1286
+ "step": 16400
1287
+ },
1288
+ {
1289
+ "epoch": 0.4885523306166863,
1290
+ "grad_norm": 8.41613483428955,
1291
+ "learning_rate": 6.722053622001221e-05,
1292
+ "loss": 31.1095,
1293
+ "step": 16500
1294
+ },
1295
+ {
1296
+ "epoch": 0.49151325383254496,
1297
+ "grad_norm": 7.951696395874023,
1298
+ "learning_rate": 6.663696270240373e-05,
1299
+ "loss": 31.1532,
1300
+ "step": 16600
1301
+ },
1302
+ {
1303
+ "epoch": 0.4944741770484037,
1304
+ "grad_norm": 9.02717113494873,
1305
+ "learning_rate": 6.60527528456546e-05,
1306
+ "loss": 31.0777,
1307
+ "step": 16700
1308
+ },
1309
+ {
1310
+ "epoch": 0.4974351002642624,
1311
+ "grad_norm": 8.57259750366211,
1312
+ "learning_rate": 6.546796266267535e-05,
1313
+ "loss": 31.3509,
1314
+ "step": 16800
1315
+ },
1316
+ {
1317
+ "epoch": 0.5003960234801211,
1318
+ "grad_norm": 9.129491806030273,
1319
+ "learning_rate": 6.488264822201711e-05,
1320
+ "loss": 30.7844,
1321
+ "step": 16900
1322
+ },
1323
+ {
1324
+ "epoch": 0.5033569466959799,
1325
+ "grad_norm": 8.600064277648926,
1326
+ "learning_rate": 6.429686564249579e-05,
1327
+ "loss": 31.1164,
1328
+ "step": 17000
1329
+ },
1330
+ {
1331
+ "epoch": 0.5033569466959799,
1332
+ "eval_loss": 3.836409091949463,
1333
+ "eval_runtime": 109.0903,
1334
+ "eval_samples_per_second": 9.909,
1335
+ "eval_steps_per_second": 2.484,
1336
+ "step": 17000
1337
+ },
1338
+ {
1339
+ "epoch": 0.5063178699118385,
1340
+ "grad_norm": 8.62096881866455,
1341
+ "learning_rate": 6.371067108781158e-05,
1342
+ "loss": 31.1944,
1343
+ "step": 17100
1344
+ },
1345
+ {
1346
+ "epoch": 0.5092787931276972,
1347
+ "grad_norm": 8.052851676940918,
1348
+ "learning_rate": 6.312412076116401e-05,
1349
+ "loss": 31.0126,
1350
+ "step": 17200
1351
+ },
1352
+ {
1353
+ "epoch": 0.5122397163435559,
1354
+ "grad_norm": 8.32268238067627,
1355
+ "learning_rate": 6.253727089986337e-05,
1356
+ "loss": 31.0692,
1357
+ "step": 17300
1358
+ },
1359
+ {
1360
+ "epoch": 0.5152006395594146,
1361
+ "grad_norm": 8.130902290344238,
1362
+ "learning_rate": 6.195017776993876e-05,
1363
+ "loss": 30.9143,
1364
+ "step": 17400
1365
+ },
1366
+ {
1367
+ "epoch": 0.5181615627752734,
1368
+ "grad_norm": 9.245232582092285,
1369
+ "learning_rate": 6.136289766074334e-05,
1370
+ "loss": 31.0029,
1371
+ "step": 17500
1372
+ },
1373
+ {
1374
+ "epoch": 0.5211224859911321,
1375
+ "grad_norm": 8.296626091003418,
1376
+ "learning_rate": 6.077548687955759e-05,
1377
+ "loss": 31.0624,
1378
+ "step": 17600
1379
+ },
1380
+ {
1381
+ "epoch": 0.5240834092069907,
1382
+ "grad_norm": 8.933104515075684,
1383
+ "learning_rate": 6.018800174619048e-05,
1384
+ "loss": 31.0619,
1385
+ "step": 17700
1386
+ },
1387
+ {
1388
+ "epoch": 0.5270443324228494,
1389
+ "grad_norm": 7.37945032119751,
1390
+ "learning_rate": 5.960049858757974e-05,
1391
+ "loss": 31.3181,
1392
+ "step": 17800
1393
+ },
1394
+ {
1395
+ "epoch": 0.5300052556387082,
1396
+ "grad_norm": 8.817550659179688,
1397
+ "learning_rate": 5.901303373239133e-05,
1398
+ "loss": 30.8424,
1399
+ "step": 17900
1400
+ },
1401
+ {
1402
+ "epoch": 0.5329661788545669,
1403
+ "grad_norm": 7.71854305267334,
1404
+ "learning_rate": 5.842566350561879e-05,
1405
+ "loss": 31.0376,
1406
+ "step": 18000
1407
+ },
1408
+ {
1409
+ "epoch": 0.5329661788545669,
1410
+ "eval_loss": 3.822613477706909,
1411
+ "eval_runtime": 112.0979,
1412
+ "eval_samples_per_second": 9.643,
1413
+ "eval_steps_per_second": 2.418,
1414
+ "step": 18000
1415
+ },
1416
+ {
1417
+ "epoch": 0.5359271020704256,
1418
+ "grad_norm": 8.84870719909668,
1419
+ "learning_rate": 5.7838444223182826e-05,
1420
+ "loss": 30.8901,
1421
+ "step": 18100
1422
+ },
1423
+ {
1424
+ "epoch": 0.5388880252862842,
1425
+ "grad_norm": 7.48129415512085,
1426
+ "learning_rate": 5.725143218653187e-05,
1427
+ "loss": 31.0275,
1428
+ "step": 18200
1429
+ },
1430
+ {
1431
+ "epoch": 0.5418489485021429,
1432
+ "grad_norm": 8.218484878540039,
1433
+ "learning_rate": 5.666468367724412e-05,
1434
+ "loss": 31.1443,
1435
+ "step": 18300
1436
+ },
1437
+ {
1438
+ "epoch": 0.5448098717180017,
1439
+ "grad_norm": 9.589841842651367,
1440
+ "learning_rate": 5.607825495163119e-05,
1441
+ "loss": 30.9756,
1442
+ "step": 18400
1443
+ },
1444
+ {
1445
+ "epoch": 0.5477707949338604,
1446
+ "grad_norm": 8.583683013916016,
1447
+ "learning_rate": 5.549220223534451e-05,
1448
+ "loss": 31.0641,
1449
+ "step": 18500
1450
+ },
1451
+ {
1452
+ "epoch": 0.5507317181497191,
1453
+ "grad_norm": 7.978188991546631,
1454
+ "learning_rate": 5.490658171798439e-05,
1455
+ "loss": 30.8899,
1456
+ "step": 18600
1457
+ },
1458
+ {
1459
+ "epoch": 0.5536926413655778,
1460
+ "grad_norm": 8.130802154541016,
1461
+ "learning_rate": 5.432144954771287e-05,
1462
+ "loss": 31.0812,
1463
+ "step": 18700
1464
+ },
1465
+ {
1466
+ "epoch": 0.5566535645814364,
1467
+ "grad_norm": 8.981709480285645,
1468
+ "learning_rate": 5.37368618258701e-05,
1469
+ "loss": 31.0612,
1470
+ "step": 18800
1471
+ },
1472
+ {
1473
+ "epoch": 0.5596144877972952,
1474
+ "grad_norm": 7.87661075592041,
1475
+ "learning_rate": 5.315287460159561e-05,
1476
+ "loss": 30.8581,
1477
+ "step": 18900
1478
+ },
1479
+ {
1480
+ "epoch": 0.5625754110131539,
1481
+ "grad_norm": 8.329483032226562,
1482
+ "learning_rate": 5.256954386645438e-05,
1483
+ "loss": 31.1805,
1484
+ "step": 19000
1485
+ },
1486
+ {
1487
+ "epoch": 0.5625754110131539,
1488
+ "eval_loss": 3.8131661415100098,
1489
+ "eval_runtime": 111.5683,
1490
+ "eval_samples_per_second": 9.689,
1491
+ "eval_steps_per_second": 2.429,
1492
+ "step": 19000
1493
+ },
1494
+ {
1495
+ "epoch": 0.5655363342290126,
1496
+ "grad_norm": 8.833015441894531,
1497
+ "learning_rate": 5.198692554906851e-05,
1498
+ "loss": 30.9231,
1499
+ "step": 19100
1500
+ },
1501
+ {
1502
+ "epoch": 0.5684972574448713,
1503
+ "grad_norm": 7.966989994049072,
1504
+ "learning_rate": 5.1405075509754834e-05,
1505
+ "loss": 31.0225,
1506
+ "step": 19200
1507
+ },
1508
+ {
1509
+ "epoch": 0.5714581806607301,
1510
+ "grad_norm": 8.791169166564941,
1511
+ "learning_rate": 5.0824049535169166e-05,
1512
+ "loss": 31.1551,
1513
+ "step": 19300
1514
+ },
1515
+ {
1516
+ "epoch": 0.5744191038765887,
1517
+ "grad_norm": 7.9680023193359375,
1518
+ "learning_rate": 5.024390333295761e-05,
1519
+ "loss": 31.0498,
1520
+ "step": 19400
1521
+ },
1522
+ {
1523
+ "epoch": 0.5773800270924474,
1524
+ "grad_norm": 8.603718757629395,
1525
+ "learning_rate": 4.966469252641538e-05,
1526
+ "loss": 30.9017,
1527
+ "step": 19500
1528
+ },
1529
+ {
1530
+ "epoch": 0.5803409503083061,
1531
+ "grad_norm": 12.401627540588379,
1532
+ "learning_rate": 4.908647264915378e-05,
1533
+ "loss": 30.9988,
1534
+ "step": 19600
1535
+ },
1536
+ {
1537
+ "epoch": 0.5833018735241648,
1538
+ "grad_norm": 8.433266639709473,
1539
+ "learning_rate": 4.8509299139775734e-05,
1540
+ "loss": 30.9905,
1541
+ "step": 19700
1542
+ },
1543
+ {
1544
+ "epoch": 0.5862627967400236,
1545
+ "grad_norm": 7.99282693862915,
1546
+ "learning_rate": 4.7933227336560414e-05,
1547
+ "loss": 31.0604,
1548
+ "step": 19800
1549
+ },
1550
+ {
1551
+ "epoch": 0.5892237199558823,
1552
+ "grad_norm": 8.011063575744629,
1553
+ "learning_rate": 4.735831247215753e-05,
1554
+ "loss": 30.7471,
1555
+ "step": 19900
1556
+ },
1557
+ {
1558
+ "epoch": 0.5921846431717409,
1559
+ "grad_norm": 9.603862762451172,
1560
+ "learning_rate": 4.67846096682918e-05,
1561
+ "loss": 30.8428,
1562
+ "step": 20000
1563
+ },
1564
+ {
1565
+ "epoch": 0.5921846431717409,
1566
+ "eval_loss": 3.8060901165008545,
1567
+ "eval_runtime": 112.6154,
1568
+ "eval_samples_per_second": 9.599,
1569
+ "eval_steps_per_second": 2.406,
1570
+ "step": 20000
1571
+ },
1572
+ {
1573
+ "epoch": 0.5951455663875996,
1574
+ "grad_norm": 8.427188873291016,
1575
+ "learning_rate": 4.6212173930477874e-05,
1576
+ "loss": 30.8438,
1577
+ "step": 20100
1578
+ },
1579
+ {
1580
+ "epoch": 0.5981064896034584,
1581
+ "grad_norm": 7.692320346832275,
1582
+ "learning_rate": 4.5641060142746556e-05,
1583
+ "loss": 30.7664,
1584
+ "step": 20200
1585
+ },
1586
+ {
1587
+ "epoch": 0.6010674128193171,
1588
+ "grad_norm": 8.596179962158203,
1589
+ "learning_rate": 4.507132306238262e-05,
1590
+ "loss": 30.9387,
1591
+ "step": 20300
1592
+ },
1593
+ {
1594
+ "epoch": 0.6040283360351758,
1595
+ "grad_norm": 8.076534271240234,
1596
+ "learning_rate": 4.450301731467488e-05,
1597
+ "loss": 30.851,
1598
+ "step": 20400
1599
+ },
1600
+ {
1601
+ "epoch": 0.6069892592510344,
1602
+ "grad_norm": 9.05728816986084,
1603
+ "learning_rate": 4.3936197387678665e-05,
1604
+ "loss": 30.7486,
1605
+ "step": 20500
1606
+ },
1607
+ {
1608
+ "epoch": 0.6099501824668931,
1609
+ "grad_norm": 8.477595329284668,
1610
+ "learning_rate": 4.3370917626991706e-05,
1611
+ "loss": 30.6843,
1612
+ "step": 20600
1613
+ },
1614
+ {
1615
+ "epoch": 0.6129111056827519,
1616
+ "grad_norm": 8.171915054321289,
1617
+ "learning_rate": 4.2807232230543625e-05,
1618
+ "loss": 30.9551,
1619
+ "step": 20700
1620
+ },
1621
+ {
1622
+ "epoch": 0.6158720288986106,
1623
+ "grad_norm": 8.333806991577148,
1624
+ "learning_rate": 4.22451952433994e-05,
1625
+ "loss": 30.8566,
1626
+ "step": 20800
1627
+ },
1628
+ {
1629
+ "epoch": 0.6188329521144693,
1630
+ "grad_norm": 7.9477715492248535,
1631
+ "learning_rate": 4.168486055257777e-05,
1632
+ "loss": 30.8577,
1633
+ "step": 20900
1634
+ },
1635
+ {
1636
+ "epoch": 0.621793875330328,
1637
+ "grad_norm": 8.560218811035156,
1638
+ "learning_rate": 4.112628188188457e-05,
1639
+ "loss": 30.7203,
1640
+ "step": 21000
1641
+ },
1642
+ {
1643
+ "epoch": 0.621793875330328,
1644
+ "eval_loss": 3.7986109256744385,
1645
+ "eval_runtime": 109.4771,
1646
+ "eval_samples_per_second": 9.874,
1647
+ "eval_steps_per_second": 2.475,
1648
+ "step": 21000
1649
+ },
1650
+ {
1651
+ "epoch": 0.6247547985461867,
1652
+ "grad_norm": 8.963776588439941,
1653
+ "learning_rate": 4.056951278676187e-05,
1654
+ "loss": 30.9418,
1655
+ "step": 21100
1656
+ },
1657
+ {
1658
+ "epoch": 0.6277157217620454,
1659
+ "grad_norm": 8.338837623596191,
1660
+ "learning_rate": 4.001460664915308e-05,
1661
+ "loss": 30.756,
1662
+ "step": 21200
1663
+ },
1664
+ {
1665
+ "epoch": 0.6306766449779041,
1666
+ "grad_norm": 8.323155403137207,
1667
+ "learning_rate": 3.946161667238485e-05,
1668
+ "loss": 30.6959,
1669
+ "step": 21300
1670
+ },
1671
+ {
1672
+ "epoch": 0.6336375681937628,
1673
+ "grad_norm": 9.881996154785156,
1674
+ "learning_rate": 3.8910595876066085e-05,
1675
+ "loss": 30.9333,
1676
+ "step": 21400
1677
+ },
1678
+ {
1679
+ "epoch": 0.6365984914096215,
1680
+ "grad_norm": 8.089996337890625,
1681
+ "learning_rate": 3.836159709100446e-05,
1682
+ "loss": 30.6899,
1683
+ "step": 21500
1684
+ },
1685
+ {
1686
+ "epoch": 0.6395594146254803,
1687
+ "grad_norm": 7.9427289962768555,
1688
+ "learning_rate": 3.7814672954141055e-05,
1689
+ "loss": 30.8046,
1690
+ "step": 21600
1691
+ },
1692
+ {
1693
+ "epoch": 0.6425203378413389,
1694
+ "grad_norm": 8.468146324157715,
1695
+ "learning_rate": 3.7269875903503826e-05,
1696
+ "loss": 31.2292,
1697
+ "step": 21700
1698
+ },
1699
+ {
1700
+ "epoch": 0.6454812610571976,
1701
+ "grad_norm": 8.63842487335205,
1702
+ "learning_rate": 3.672725817317973e-05,
1703
+ "loss": 30.7721,
1704
+ "step": 21800
1705
+ },
1706
+ {
1707
+ "epoch": 0.6484421842730563,
1708
+ "grad_norm": 8.145241737365723,
1709
+ "learning_rate": 3.6186871788306674e-05,
1710
+ "loss": 30.5881,
1711
+ "step": 21900
1712
+ },
1713
+ {
1714
+ "epoch": 0.6514031074889151,
1715
+ "grad_norm": 8.194993019104004,
1716
+ "learning_rate": 3.5648768560085604e-05,
1717
+ "loss": 30.9425,
1718
+ "step": 22000
1719
+ },
1720
+ {
1721
+ "epoch": 0.6514031074889151,
1722
+ "eval_loss": 3.7950870990753174,
1723
+ "eval_runtime": 109.4264,
1724
+ "eval_samples_per_second": 9.879,
1725
+ "eval_steps_per_second": 2.477,
1726
+ "step": 22000
1727
+ },
1728
+ {
1729
+ "epoch": 0.6543640307047738,
1730
+ "grad_norm": 9.304323196411133,
1731
+ "learning_rate": 3.511300008081273e-05,
1732
+ "loss": 30.722,
1733
+ "step": 22100
1734
+ },
1735
+ {
1736
+ "epoch": 0.6573249539206325,
1737
+ "grad_norm": 7.82930850982666,
1738
+ "learning_rate": 3.4579617718933054e-05,
1739
+ "loss": 30.7943,
1740
+ "step": 22200
1741
+ },
1742
+ {
1743
+ "epoch": 0.6602858771364911,
1744
+ "grad_norm": 7.912548542022705,
1745
+ "learning_rate": 3.4048672614115294e-05,
1746
+ "loss": 30.8451,
1747
+ "step": 22300
1748
+ },
1749
+ {
1750
+ "epoch": 0.6632468003523498,
1751
+ "grad_norm": 8.46181583404541,
1752
+ "learning_rate": 3.352021567234869e-05,
1753
+ "loss": 30.9009,
1754
+ "step": 22400
1755
+ },
1756
+ {
1757
+ "epoch": 0.6662077235682086,
1758
+ "grad_norm": 7.727646827697754,
1759
+ "learning_rate": 3.299429756106215e-05,
1760
+ "loss": 30.8281,
1761
+ "step": 22500
1762
+ },
1763
+ {
1764
+ "epoch": 0.6691686467840673,
1765
+ "grad_norm": 8.119136810302734,
1766
+ "learning_rate": 3.247096870426649e-05,
1767
+ "loss": 30.7757,
1768
+ "step": 22600
1769
+ },
1770
+ {
1771
+ "epoch": 0.672129569999926,
1772
+ "grad_norm": 8.091607093811035,
1773
+ "learning_rate": 3.195027927771982e-05,
1774
+ "loss": 30.8661,
1775
+ "step": 22700
1776
+ },
1777
+ {
1778
+ "epoch": 0.6750904932157847,
1779
+ "grad_norm": 7.598474979400635,
1780
+ "learning_rate": 3.1432279204116776e-05,
1781
+ "loss": 30.6257,
1782
+ "step": 22800
1783
+ },
1784
+ {
1785
+ "epoch": 0.6780514164316434,
1786
+ "grad_norm": 9.547100067138672,
1787
+ "learning_rate": 3.091701814830198e-05,
1788
+ "loss": 30.8582,
1789
+ "step": 22900
1790
+ },
1791
+ {
1792
+ "epoch": 0.6810123396475021,
1793
+ "grad_norm": 7.637078762054443,
1794
+ "learning_rate": 3.0404545512508415e-05,
1795
+ "loss": 30.9432,
1796
+ "step": 23000
1797
+ },
1798
+ {
1799
+ "epoch": 0.6810123396475021,
1800
+ "eval_loss": 3.791748285293579,
1801
+ "eval_runtime": 109.2867,
1802
+ "eval_samples_per_second": 9.891,
1803
+ "eval_steps_per_second": 2.48,
1804
+ "step": 23000
1805
+ },
1806
+ {
1807
+ "epoch": 0.6839732628633608,
1808
+ "grad_norm": 8.485209465026855,
1809
+ "learning_rate": 2.98949104316207e-05,
1810
+ "loss": 30.921,
1811
+ "step": 23100
1812
+ },
1813
+ {
1814
+ "epoch": 0.6869341860792195,
1815
+ "grad_norm": 7.777042865753174,
1816
+ "learning_rate": 2.938816176846421e-05,
1817
+ "loss": 30.8116,
1818
+ "step": 23200
1819
+ },
1820
+ {
1821
+ "epoch": 0.6898951092950782,
1822
+ "grad_norm": 7.6587138175964355,
1823
+ "learning_rate": 2.8884348109120106e-05,
1824
+ "loss": 30.7965,
1825
+ "step": 23300
1826
+ },
1827
+ {
1828
+ "epoch": 0.6928560325109369,
1829
+ "grad_norm": 8.276775360107422,
1830
+ "learning_rate": 2.8383517758267178e-05,
1831
+ "loss": 30.6582,
1832
+ "step": 23400
1833
+ },
1834
+ {
1835
+ "epoch": 0.6958169557267956,
1836
+ "grad_norm": 7.5494771003723145,
1837
+ "learning_rate": 2.7885718734550257e-05,
1838
+ "loss": 30.6483,
1839
+ "step": 23500
1840
+ },
1841
+ {
1842
+ "epoch": 0.6987778789426543,
1843
+ "grad_norm": 7.938130855560303,
1844
+ "learning_rate": 2.739099876597646e-05,
1845
+ "loss": 30.529,
1846
+ "step": 23600
1847
+ },
1848
+ {
1849
+ "epoch": 0.701738802158513,
1850
+ "grad_norm": 8.202885627746582,
1851
+ "learning_rate": 2.6899405285339026e-05,
1852
+ "loss": 30.825,
1853
+ "step": 23700
1854
+ },
1855
+ {
1856
+ "epoch": 0.7046997253743718,
1857
+ "grad_norm": 8.393240928649902,
1858
+ "learning_rate": 2.6410985425669622e-05,
1859
+ "loss": 30.7867,
1860
+ "step": 23800
1861
+ },
1862
+ {
1863
+ "epoch": 0.7076606485902305,
1864
+ "grad_norm": 8.32459831237793,
1865
+ "learning_rate": 2.5925786015719207e-05,
1866
+ "loss": 30.7898,
1867
+ "step": 23900
1868
+ },
1869
+ {
1870
+ "epoch": 0.7106215718060891,
1871
+ "grad_norm": 9.323598861694336,
1872
+ "learning_rate": 2.544385357546831e-05,
1873
+ "loss": 30.5684,
1874
+ "step": 24000
1875
+ },
1876
+ {
1877
+ "epoch": 0.7106215718060891,
1878
+ "eval_loss": 3.789947271347046,
1879
+ "eval_runtime": 110.7686,
1880
+ "eval_samples_per_second": 9.759,
1881
+ "eval_steps_per_second": 2.447,
1882
+ "step": 24000
1883
+ },
1884
+ {
1885
+ "epoch": 0.7135824950219478,
1886
+ "grad_norm": 8.184738159179688,
1887
+ "learning_rate": 2.4965234311666717e-05,
1888
+ "loss": 30.7187,
1889
+ "step": 24100
1890
+ },
1891
+ {
1892
+ "epoch": 0.7165434182378065,
1893
+ "grad_norm": 7.82784366607666,
1894
+ "learning_rate": 2.4489974113403275e-05,
1895
+ "loss": 30.705,
1896
+ "step": 24200
1897
+ },
1898
+ {
1899
+ "epoch": 0.7195043414536653,
1900
+ "grad_norm": 7.945186614990234,
1901
+ "learning_rate": 2.4018118547706078e-05,
1902
+ "loss": 30.4846,
1903
+ "step": 24300
1904
+ },
1905
+ {
1906
+ "epoch": 0.722465264669524,
1907
+ "grad_norm": 9.277371406555176,
1908
+ "learning_rate": 2.3549712855173688e-05,
1909
+ "loss": 30.6765,
1910
+ "step": 24400
1911
+ },
1912
+ {
1913
+ "epoch": 0.7254261878853827,
1914
+ "grad_norm": 8.619938850402832,
1915
+ "learning_rate": 2.3084801945637512e-05,
1916
+ "loss": 30.6503,
1917
+ "step": 24500
1918
+ },
1919
+ {
1920
+ "epoch": 0.7283871111012413,
1921
+ "grad_norm": 8.467925071716309,
1922
+ "learning_rate": 2.262343039385585e-05,
1923
+ "loss": 30.957,
1924
+ "step": 24600
1925
+ },
1926
+ {
1927
+ "epoch": 0.7313480343171,
1928
+ "grad_norm": 8.035057067871094,
1929
+ "learning_rate": 2.216564243524035e-05,
1930
+ "loss": 30.6764,
1931
+ "step": 24700
1932
+ },
1933
+ {
1934
+ "epoch": 0.7343089575329588,
1935
+ "grad_norm": 7.555221080780029,
1936
+ "learning_rate": 2.1711481961614565e-05,
1937
+ "loss": 30.7666,
1938
+ "step": 24800
1939
+ },
1940
+ {
1941
+ "epoch": 0.7372698807488175,
1942
+ "grad_norm": 7.959348201751709,
1943
+ "learning_rate": 2.1260992517005892e-05,
1944
+ "loss": 30.8212,
1945
+ "step": 24900
1946
+ },
1947
+ {
1948
+ "epoch": 0.7402308039646762,
1949
+ "grad_norm": 7.882981300354004,
1950
+ "learning_rate": 2.0814217293470476e-05,
1951
+ "loss": 30.8312,
1952
+ "step": 25000
1953
+ },
1954
+ {
1955
+ "epoch": 0.7402308039646762,
1956
+ "eval_loss": 3.7874350547790527,
1957
+ "eval_runtime": 107.8316,
1958
+ "eval_samples_per_second": 10.025,
1959
+ "eval_steps_per_second": 2.513,
1960
+ "step": 25000
1961
+ },
1962
+ {
1963
+ "epoch": 0.743191727180535,
1964
+ "grad_norm": 7.499105930328369,
1965
+ "learning_rate": 2.0371199126952268e-05,
1966
+ "loss": 30.9958,
1967
+ "step": 25100
1968
+ },
1969
+ {
1970
+ "epoch": 0.7461526503963936,
1971
+ "grad_norm": 7.973631381988525,
1972
+ "learning_rate": 1.9931980493175735e-05,
1973
+ "loss": 30.6469,
1974
+ "step": 25200
1975
+ },
1976
+ {
1977
+ "epoch": 0.7491135736122523,
1978
+ "grad_norm": 7.996872425079346,
1979
+ "learning_rate": 1.949660350357356e-05,
1980
+ "loss": 30.6363,
1981
+ "step": 25300
1982
+ },
1983
+ {
1984
+ "epoch": 0.752074496828111,
1985
+ "grad_norm": 8.139349937438965,
1986
+ "learning_rate": 1.9065109901249e-05,
1987
+ "loss": 30.924,
1988
+ "step": 25400
1989
+ },
1990
+ {
1991
+ "epoch": 0.7550354200439697,
1992
+ "grad_norm": 8.981887817382812,
1993
+ "learning_rate": 1.863754105697369e-05,
1994
+ "loss": 30.9555,
1995
+ "step": 25500
1996
+ },
1997
+ {
1998
+ "epoch": 0.7579963432598285,
1999
+ "grad_norm": 7.660996913909912,
2000
+ "learning_rate": 1.821393796522096e-05,
2001
+ "loss": 30.8007,
2002
+ "step": 25600
2003
+ },
2004
+ {
2005
+ "epoch": 0.7609572664756871,
2006
+ "grad_norm": 7.750844955444336,
2007
+ "learning_rate": 1.7794341240235615e-05,
2008
+ "loss": 30.7227,
2009
+ "step": 25700
2010
+ },
2011
+ {
2012
+ "epoch": 0.7639181896915458,
2013
+ "grad_norm": 7.581575870513916,
2014
+ "learning_rate": 1.737879111213961e-05,
2015
+ "loss": 30.6509,
2016
+ "step": 25800
2017
+ },
2018
+ {
2019
+ "epoch": 0.7668791129074045,
2020
+ "grad_norm": 8.771635055541992,
2021
+ "learning_rate": 1.6967327423075142e-05,
2022
+ "loss": 30.7893,
2023
+ "step": 25900
2024
+ },
2025
+ {
2026
+ "epoch": 0.7698400361232632,
2027
+ "grad_norm": 8.594512939453125,
2028
+ "learning_rate": 1.6559989623384456e-05,
2029
+ "loss": 30.6874,
2030
+ "step": 26000
2031
+ },
2032
+ {
2033
+ "epoch": 0.7698400361232632,
2034
+ "eval_loss": 3.7861363887786865,
2035
+ "eval_runtime": 112.2096,
2036
+ "eval_samples_per_second": 9.634,
2037
+ "eval_steps_per_second": 2.415,
2038
+ "step": 26000
2039
+ },
2040
+ {
2041
+ "epoch": 0.772800959339122,
2042
+ "grad_norm": 7.919267177581787,
2043
+ "learning_rate": 1.615681676782755e-05,
2044
+ "loss": 30.7685,
2045
+ "step": 26100
2046
+ },
2047
+ {
2048
+ "epoch": 0.7757618825549807,
2049
+ "grad_norm": 7.744143009185791,
2050
+ "learning_rate": 1.5757847511837648e-05,
2051
+ "loss": 30.7558,
2052
+ "step": 26200
2053
+ },
2054
+ {
2055
+ "epoch": 0.7787228057708393,
2056
+ "grad_norm": 7.894962787628174,
2057
+ "learning_rate": 1.5363120107814955e-05,
2058
+ "loss": 30.7543,
2059
+ "step": 26300
2060
+ },
2061
+ {
2062
+ "epoch": 0.781683728986698,
2063
+ "grad_norm": 9.573600769042969,
2064
+ "learning_rate": 1.4972672401459143e-05,
2065
+ "loss": 30.808,
2066
+ "step": 26400
2067
+ },
2068
+ {
2069
+ "epoch": 0.7846446522025567,
2070
+ "grad_norm": 7.708218574523926,
2071
+ "learning_rate": 1.4586541828140706e-05,
2072
+ "loss": 30.6115,
2073
+ "step": 26500
2074
+ },
2075
+ {
2076
+ "epoch": 0.7876055754184155,
2077
+ "grad_norm": 8.170422554016113,
2078
+ "learning_rate": 1.4204765409311852e-05,
2079
+ "loss": 30.8811,
2080
+ "step": 26600
2081
+ },
2082
+ {
2083
+ "epoch": 0.7905664986342742,
2084
+ "grad_norm": 8.293937683105469,
2085
+ "learning_rate": 1.3827379748956783e-05,
2086
+ "loss": 30.8484,
2087
+ "step": 26700
2088
+ },
2089
+ {
2090
+ "epoch": 0.7935274218501329,
2091
+ "grad_norm": 7.64206075668335,
2092
+ "learning_rate": 1.3454421030082402e-05,
2093
+ "loss": 30.7768,
2094
+ "step": 26800
2095
+ },
2096
+ {
2097
+ "epoch": 0.7964883450659915,
2098
+ "grad_norm": 7.780085563659668,
2099
+ "learning_rate": 1.3085925011248902e-05,
2100
+ "loss": 30.6903,
2101
+ "step": 26900
2102
+ },
2103
+ {
2104
+ "epoch": 0.7994492682818503,
2105
+ "grad_norm": 7.651244640350342,
2106
+ "learning_rate": 1.2721927023141509e-05,
2107
+ "loss": 30.8888,
2108
+ "step": 27000
2109
+ },
2110
+ {
2111
+ "epoch": 0.7994492682818503,
2112
+ "eval_loss": 3.7866110801696777,
2113
+ "eval_runtime": 111.3993,
2114
+ "eval_samples_per_second": 9.704,
2115
+ "eval_steps_per_second": 2.433,
2116
+ "step": 27000
2117
+ },
2118
+ {
2119
+ "epoch": 0.802410191497709,
2120
+ "grad_norm": 7.893172740936279,
2121
+ "learning_rate": 1.2362461965182951e-05,
2122
+ "loss": 30.8551,
2123
+ "step": 27100
2124
+ },
2125
+ {
2126
+ "epoch": 0.8053711147135677,
2127
+ "grad_norm": 8.348461151123047,
2128
+ "learning_rate": 1.2007564302187395e-05,
2129
+ "loss": 30.9086,
2130
+ "step": 27200
2131
+ },
2132
+ {
2133
+ "epoch": 0.8083320379294264,
2134
+ "grad_norm": 8.005925178527832,
2135
+ "learning_rate": 1.1657268061055954e-05,
2136
+ "loss": 30.6258,
2137
+ "step": 27300
2138
+ },
2139
+ {
2140
+ "epoch": 0.8112929611452852,
2141
+ "grad_norm": 7.919161319732666,
2142
+ "learning_rate": 1.1311606827514432e-05,
2143
+ "loss": 30.4614,
2144
+ "step": 27400
2145
+ },
2146
+ {
2147
+ "epoch": 0.8142538843611438,
2148
+ "grad_norm": 8.806751251220703,
2149
+ "learning_rate": 1.0970613742892959e-05,
2150
+ "loss": 30.9882,
2151
+ "step": 27500
2152
+ },
2153
+ {
2154
+ "epoch": 0.8172148075770025,
2155
+ "grad_norm": 8.126434326171875,
2156
+ "learning_rate": 1.0634321500948665e-05,
2157
+ "loss": 30.6459,
2158
+ "step": 27600
2159
+ },
2160
+ {
2161
+ "epoch": 0.8201757307928612,
2162
+ "grad_norm": 7.643808364868164,
2163
+ "learning_rate": 1.0302762344730893e-05,
2164
+ "loss": 30.6614,
2165
+ "step": 27700
2166
+ },
2167
+ {
2168
+ "epoch": 0.8231366540087199,
2169
+ "grad_norm": 8.046734809875488,
2170
+ "learning_rate": 9.97596806349001e-06,
2171
+ "loss": 30.6958,
2172
+ "step": 27800
2173
+ },
2174
+ {
2175
+ "epoch": 0.8260975772245787,
2176
+ "grad_norm": 8.094582557678223,
2177
+ "learning_rate": 9.653969989629268e-06,
2178
+ "loss": 30.5807,
2179
+ "step": 27900
2180
+ },
2181
+ {
2182
+ "epoch": 0.8290585004404373,
2183
+ "grad_norm": 8.062453269958496,
2184
+ "learning_rate": 9.336798995700899e-06,
2185
+ "loss": 30.8323,
2186
+ "step": 28000
2187
+ },
2188
+ {
2189
+ "epoch": 0.8290585004404373,
2190
+ "eval_loss": 3.786661386489868,
2191
+ "eval_runtime": 106.9454,
2192
+ "eval_samples_per_second": 10.108,
2193
+ "eval_steps_per_second": 2.534,
2194
+ "step": 28000
2195
+ },
2196
+ {
2197
+ "epoch": 0.832019423656296,
2198
+ "grad_norm": 7.68911075592041,
2199
+ "learning_rate": 9.024485491446045e-06,
2200
+ "loss": 30.9853,
2201
+ "step": 28100
2202
+ },
2203
+ {
2204
+ "epoch": 0.8349803468721547,
2205
+ "grad_norm": 7.82414436340332,
2206
+ "learning_rate": 8.717059420879143e-06,
2207
+ "loss": 30.5061,
2208
+ "step": 28200
2209
+ },
2210
+ {
2211
+ "epoch": 0.8379412700880134,
2212
+ "grad_norm": 7.392062664031982,
2213
+ "learning_rate": 8.414550259416917e-06,
2214
+ "loss": 30.9525,
2215
+ "step": 28300
2216
+ },
2217
+ {
2218
+ "epoch": 0.8409021933038722,
2219
+ "grad_norm": 7.675992965698242,
2220
+ "learning_rate": 8.116987011052387e-06,
2221
+ "loss": 30.8296,
2222
+ "step": 28400
2223
+ },
2224
+ {
2225
+ "epoch": 0.8438631165197309,
2226
+ "grad_norm": 8.038030624389648,
2227
+ "learning_rate": 7.824398205574006e-06,
2228
+ "loss": 30.8155,
2229
+ "step": 28500
2230
+ },
2231
+ {
2232
+ "epoch": 0.8468240397355895,
2233
+ "grad_norm": 7.427101135253906,
2234
+ "learning_rate": 7.536811895830222e-06,
2235
+ "loss": 30.9259,
2236
+ "step": 28600
2237
+ },
2238
+ {
2239
+ "epoch": 0.8497849629514482,
2240
+ "grad_norm": 8.095186233520508,
2241
+ "learning_rate": 7.254255655039919e-06,
2242
+ "loss": 30.824,
2243
+ "step": 28700
2244
+ },
2245
+ {
2246
+ "epoch": 0.852745886167307,
2247
+ "grad_norm": 7.521733283996582,
2248
+ "learning_rate": 6.9767565741486815e-06,
2249
+ "loss": 30.7226,
2250
+ "step": 28800
2251
+ },
2252
+ {
2253
+ "epoch": 0.8557068093831657,
2254
+ "grad_norm": 7.494954586029053,
2255
+ "learning_rate": 6.704341259231415e-06,
2256
+ "loss": 30.7789,
2257
+ "step": 28900
2258
+ },
2259
+ {
2260
+ "epoch": 0.8586677325990244,
2261
+ "grad_norm": 7.641082763671875,
2262
+ "learning_rate": 6.437035828941324e-06,
2263
+ "loss": 30.8001,
2264
+ "step": 29000
2265
+ },
2266
+ {
2267
+ "epoch": 0.8586677325990244,
2268
+ "eval_loss": 3.786005973815918,
2269
+ "eval_runtime": 111.1521,
2270
+ "eval_samples_per_second": 9.725,
2271
+ "eval_steps_per_second": 2.438,
2272
+ "step": 29000
2273
+ },
2274
+ {
2275
+ "epoch": 0.8616286558148831,
2276
+ "grad_norm": 7.96475887298584,
2277
+ "learning_rate": 6.1748659120058386e-06,
2278
+ "loss": 30.8879,
2279
+ "step": 29100
2280
+ },
2281
+ {
2282
+ "epoch": 0.8645895790307417,
2283
+ "grad_norm": 6.990954875946045,
2284
+ "learning_rate": 5.917856644769242e-06,
2285
+ "loss": 30.6077,
2286
+ "step": 29200
2287
+ },
2288
+ {
2289
+ "epoch": 0.8675505022466005,
2290
+ "grad_norm": 7.170067310333252,
2291
+ "learning_rate": 5.666032668782735e-06,
2292
+ "loss": 30.8456,
2293
+ "step": 29300
2294
+ },
2295
+ {
2296
+ "epoch": 0.8705114254624592,
2297
+ "grad_norm": 8.4426851272583,
2298
+ "learning_rate": 5.419418128441846e-06,
2299
+ "loss": 30.9228,
2300
+ "step": 29400
2301
+ },
2302
+ {
2303
+ "epoch": 0.8734723486783179,
2304
+ "grad_norm": 8.034204483032227,
2305
+ "learning_rate": 5.178036668671475e-06,
2306
+ "loss": 30.7785,
2307
+ "step": 29500
2308
+ },
2309
+ {
2310
+ "epoch": 0.8764332718941766,
2311
+ "grad_norm": 7.411805629730225,
2312
+ "learning_rate": 4.941911432658868e-06,
2313
+ "loss": 30.7495,
2314
+ "step": 29600
2315
+ },
2316
+ {
2317
+ "epoch": 0.8793941951100354,
2318
+ "grad_norm": 7.887239456176758,
2319
+ "learning_rate": 4.7110650596347335e-06,
2320
+ "loss": 30.7797,
2321
+ "step": 29700
2322
+ },
2323
+ {
2324
+ "epoch": 0.882355118325894,
2325
+ "grad_norm": 8.600279808044434,
2326
+ "learning_rate": 4.48551968270261e-06,
2327
+ "loss": 30.8267,
2328
+ "step": 29800
2329
+ },
2330
+ {
2331
+ "epoch": 0.8853160415417527,
2332
+ "grad_norm": 8.055954933166504,
2333
+ "learning_rate": 4.26529692671679e-06,
2334
+ "loss": 30.8123,
2335
+ "step": 29900
2336
+ },
2337
+ {
2338
+ "epoch": 0.8882769647576114,
2339
+ "grad_norm": 7.540750503540039,
2340
+ "learning_rate": 4.050417906208945e-06,
2341
+ "loss": 30.8866,
2342
+ "step": 30000
2343
+ },
2344
+ {
2345
+ "epoch": 0.8882769647576114,
2346
+ "eval_loss": 3.7849574089050293,
2347
+ "eval_runtime": 108.0072,
2348
+ "eval_samples_per_second": 10.009,
2349
+ "eval_steps_per_second": 2.509,
2350
+ "step": 30000
2351
+ },
2352
+ {
2353
+ "epoch": 0.8912378879734701,
2354
+ "grad_norm": 7.607705593109131,
2355
+ "learning_rate": 3.840903223363752e-06,
2356
+ "loss": 30.7932,
2357
+ "step": 30100
2358
+ },
2359
+ {
2360
+ "epoch": 0.8941988111893289,
2361
+ "grad_norm": 7.834300518035889,
2362
+ "learning_rate": 3.636772966043571e-06,
2363
+ "loss": 30.6935,
2364
+ "step": 30200
2365
+ },
2366
+ {
2367
+ "epoch": 0.8971597344051876,
2368
+ "grad_norm": 9.865922927856445,
2369
+ "learning_rate": 3.4380467058624585e-06,
2370
+ "loss": 30.5129,
2371
+ "step": 30300
2372
+ },
2373
+ {
2374
+ "epoch": 0.9001206576210462,
2375
+ "grad_norm": 7.9707865715026855,
2376
+ "learning_rate": 3.244743496309701e-06,
2377
+ "loss": 30.8035,
2378
+ "step": 30400
2379
+ },
2380
+ {
2381
+ "epoch": 0.9030815808369049,
2382
+ "grad_norm": 8.035768508911133,
2383
+ "learning_rate": 3.0568818709229364e-06,
2384
+ "loss": 30.4973,
2385
+ "step": 30500
2386
+ },
2387
+ {
2388
+ "epoch": 0.9060425040527637,
2389
+ "grad_norm": 8.816192626953125,
2390
+ "learning_rate": 2.8744798415113015e-06,
2391
+ "loss": 30.5553,
2392
+ "step": 30600
2393
+ },
2394
+ {
2395
+ "epoch": 0.9090034272686224,
2396
+ "grad_norm": 7.411801338195801,
2397
+ "learning_rate": 2.6975548964283823e-06,
2398
+ "loss": 30.6758,
2399
+ "step": 30700
2400
+ },
2401
+ {
2402
+ "epoch": 0.9119643504844811,
2403
+ "grad_norm": 7.46308708190918,
2404
+ "learning_rate": 2.5261239988955733e-06,
2405
+ "loss": 30.8337,
2406
+ "step": 30800
2407
+ },
2408
+ {
2409
+ "epoch": 0.9149252737003397,
2410
+ "grad_norm": 8.57913875579834,
2411
+ "learning_rate": 2.360203585375571e-06,
2412
+ "loss": 31.0671,
2413
+ "step": 30900
2414
+ },
2415
+ {
2416
+ "epoch": 0.9178861969161984,
2417
+ "grad_norm": 7.983087062835693,
2418
+ "learning_rate": 2.1998095639965577e-06,
2419
+ "loss": 30.913,
2420
+ "step": 31000
2421
+ },
2422
+ {
2423
+ "epoch": 0.9178861969161984,
2424
+ "eval_loss": 3.785719394683838,
2425
+ "eval_runtime": 110.9703,
2426
+ "eval_samples_per_second": 9.741,
2427
+ "eval_steps_per_second": 2.442,
2428
+ "step": 31000
2429
+ },
2430
+ {
2431
+ "epoch": 0.9208471201320572,
2432
+ "grad_norm": 8.11637020111084,
2433
+ "learning_rate": 2.044957313026925e-06,
2434
+ "loss": 30.7294,
2435
+ "step": 31100
2436
+ },
2437
+ {
2438
+ "epoch": 0.9238080433479159,
2439
+ "grad_norm": 7.882040977478027,
2440
+ "learning_rate": 1.895661679400842e-06,
2441
+ "loss": 30.7816,
2442
+ "step": 31200
2443
+ },
2444
+ {
2445
+ "epoch": 0.9267689665637746,
2446
+ "grad_norm": 7.475772857666016,
2447
+ "learning_rate": 1.7519369772947525e-06,
2448
+ "loss": 30.5198,
2449
+ "step": 31300
2450
+ },
2451
+ {
2452
+ "epoch": 0.9297298897796333,
2453
+ "grad_norm": 8.094454765319824,
2454
+ "learning_rate": 1.6137969867549674e-06,
2455
+ "loss": 30.8313,
2456
+ "step": 31400
2457
+ },
2458
+ {
2459
+ "epoch": 0.932690812995492,
2460
+ "grad_norm": 8.635899543762207,
2461
+ "learning_rate": 1.4812549523764674e-06,
2462
+ "loss": 30.6539,
2463
+ "step": 31500
2464
+ },
2465
+ {
2466
+ "epoch": 0.9356517362113507,
2467
+ "grad_norm": 7.975414752960205,
2468
+ "learning_rate": 1.354323582033039e-06,
2469
+ "loss": 30.5804,
2470
+ "step": 31600
2471
+ },
2472
+ {
2473
+ "epoch": 0.9386126594272094,
2474
+ "grad_norm": 7.660233020782471,
2475
+ "learning_rate": 1.233015045658823e-06,
2476
+ "loss": 30.6357,
2477
+ "step": 31700
2478
+ },
2479
+ {
2480
+ "epoch": 0.9415735826430681,
2481
+ "grad_norm": 8.09595012664795,
2482
+ "learning_rate": 1.1173409740815532e-06,
2483
+ "loss": 30.7201,
2484
+ "step": 31800
2485
+ },
2486
+ {
2487
+ "epoch": 0.9445345058589268,
2488
+ "grad_norm": 8.44491958618164,
2489
+ "learning_rate": 1.0073124579073701e-06,
2490
+ "loss": 30.7462,
2491
+ "step": 31900
2492
+ },
2493
+ {
2494
+ "epoch": 0.9474954290747856,
2495
+ "grad_norm": 8.275026321411133,
2496
+ "learning_rate": 9.0294004645749e-07,
2497
+ "loss": 30.7256,
2498
+ "step": 32000
2499
+ },
2500
+ {
2501
+ "epoch": 0.9474954290747856,
2502
+ "eval_loss": 3.7850279808044434,
2503
+ "eval_runtime": 109.0824,
2504
+ "eval_samples_per_second": 9.91,
2505
+ "eval_steps_per_second": 2.484,
2506
+ "step": 32000
2507
+ },
2508
+ {
2509
+ "epoch": 0.9504563522906442,
2510
+ "grad_norm": 7.571169376373291,
2511
+ "learning_rate": 8.042337467567484e-07,
2512
+ "loss": 30.7194,
2513
+ "step": 32100
2514
+ },
2515
+ {
2516
+ "epoch": 0.9534172755065029,
2517
+ "grad_norm": 8.020681381225586,
2518
+ "learning_rate": 7.112030225741472e-07,
2519
+ "loss": 30.5828,
2520
+ "step": 32200
2521
+ },
2522
+ {
2523
+ "epoch": 0.9563781987223616,
2524
+ "grad_norm": 7.482342720031738,
2525
+ "learning_rate": 6.238567935155004e-07,
2526
+ "loss": 30.5888,
2527
+ "step": 32300
2528
+ },
2529
+ {
2530
+ "epoch": 0.9593391219382204,
2531
+ "grad_norm": 8.336071014404297,
2532
+ "learning_rate": 5.422034341682314e-07,
2533
+ "loss": 30.858,
2534
+ "step": 32400
2535
+ },
2536
+ {
2537
+ "epoch": 0.9623000451540791,
2538
+ "grad_norm": 7.819650173187256,
2539
+ "learning_rate": 4.6625077329842224e-07,
2540
+ "loss": 30.6983,
2541
+ "step": 32500
2542
+ },
2543
+ {
2544
+ "epoch": 0.9652609683699378,
2545
+ "grad_norm": 8.101078987121582,
2546
+ "learning_rate": 3.960060931002141e-07,
2547
+ "loss": 30.7803,
2548
+ "step": 32600
2549
+ },
2550
+ {
2551
+ "epoch": 0.9682218915857964,
2552
+ "grad_norm": 9.275129318237305,
2553
+ "learning_rate": 3.3147612849762533e-07,
2554
+ "loss": 30.8961,
2555
+ "step": 32700
2556
+ },
2557
+ {
2558
+ "epoch": 0.9711828148016551,
2559
+ "grad_norm": 8.00763988494873,
2560
+ "learning_rate": 2.7266706649877516e-07,
2561
+ "loss": 30.9344,
2562
+ "step": 32800
2563
+ },
2564
+ {
2565
+ "epoch": 0.9741437380175139,
2566
+ "grad_norm": 8.840792655944824,
2567
+ "learning_rate": 2.1958454560274455e-07,
2568
+ "loss": 30.7027,
2569
+ "step": 32900
2570
+ },
2571
+ {
2572
+ "epoch": 0.9771046612333726,
2573
+ "grad_norm": 8.015409469604492,
2574
+ "learning_rate": 1.722336552589021e-07,
2575
+ "loss": 30.7569,
2576
+ "step": 33000
2577
+ },
2578
+ {
2579
+ "epoch": 0.9771046612333726,
2580
+ "eval_loss": 3.785550117492676,
2581
+ "eval_runtime": 109.3256,
2582
+ "eval_samples_per_second": 9.888,
2583
+ "eval_steps_per_second": 2.479,
2584
+ "step": 33000
2585
+ },
2586
+ {
2587
+ "epoch": 0.9800655844492313,
2588
+ "grad_norm": 8.226040840148926,
2589
+ "learning_rate": 1.3061893537898773e-07,
2590
+ "loss": 30.6858,
2591
+ "step": 33100
2592
+ },
2593
+ {
2594
+ "epoch": 0.9830265076650899,
2595
+ "grad_norm": 7.274777889251709,
2596
+ "learning_rate": 9.474437590182072e-08,
2597
+ "loss": 30.701,
2598
+ "step": 33200
2599
+ },
2600
+ {
2601
+ "epoch": 0.9859874308809486,
2602
+ "grad_norm": 7.866406440734863,
2603
+ "learning_rate": 6.46134164107326e-08,
2604
+ "loss": 30.6392,
2605
+ "step": 33300
2606
+ },
2607
+ {
2608
+ "epoch": 0.9889483540968074,
2609
+ "grad_norm": 7.722043514251709,
2610
+ "learning_rate": 4.022894580381742e-08,
2611
+ "loss": 30.8502,
2612
+ "step": 33400
2613
+ },
2614
+ {
2615
+ "epoch": 0.9919092773126661,
2616
+ "grad_norm": 7.612312316894531,
2617
+ "learning_rate": 2.1593302016933437e-08,
2618
+ "loss": 30.7914,
2619
+ "step": 33500
2620
+ },
2621
+ {
2622
+ "epoch": 0.9948702005285248,
2623
+ "grad_norm": 8.195243835449219,
2624
+ "learning_rate": 8.708271799542367e-09,
2625
+ "loss": 30.8885,
2626
+ "step": 33600
2627
+ },
2628
+ {
2629
+ "epoch": 0.9978311237443835,
2630
+ "grad_norm": 8.127638816833496,
2631
+ "learning_rate": 1.5750905434130935e-09,
2632
+ "loss": 30.9894,
2633
+ "step": 33700
2634
+ }
2635
+ ],
2636
+ "logging_steps": 100,
2637
+ "max_steps": 33773,
2638
+ "num_input_tokens_seen": 0,
2639
+ "num_train_epochs": 1,
2640
+ "save_steps": 500,
2641
+ "stateful_callbacks": {
2642
+ "TrainerControl": {
2643
+ "args": {
2644
+ "should_epoch_stop": false,
2645
+ "should_evaluate": false,
2646
+ "should_log": false,
2647
+ "should_save": true,
2648
+ "should_training_stop": true
2649
+ },
2650
+ "attributes": {}
2651
+ }
2652
+ },
2653
+ "total_flos": 5.598282561239384e+18,
2654
+ "train_batch_size": 4,
2655
+ "trial_name": null,
2656
+ "trial_params": null
2657
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4c7abecf39843cd0ef9dea5e194fc495c60905578d8e9c6fafa20b2cc00b4dfc
3
+ size 5777
vocab.json ADDED
The diff for this file is too large to render. See raw diff