introvoyz041 commited on
Commit
fa720fe
·
verified ·
1 Parent(s): e590f61

Upload config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. config.json +841 -0
config.json ADDED
@@ -0,0 +1,841 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "MistralForCausalLM"
4
+ ],
5
+ "attention_dropout": 0.0,
6
+ "bos_token_id": 1,
7
+ "eos_token_id": 2,
8
+ "head_dim": 128,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 5120,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 32768,
13
+ "max_position_embeddings": 131072,
14
+ "model_type": "mistral",
15
+ "num_attention_heads": 32,
16
+ "num_hidden_layers": 40,
17
+ "num_key_value_heads": 8,
18
+ "pad_token_id": 11,
19
+ "quantization": {
20
+ "group_size": 64,
21
+ "bits": 4,
22
+ "model.embed_tokens": true,
23
+ "model.layers.0.self_attn.q_proj": true,
24
+ "model.layers.0.self_attn.k_proj": true,
25
+ "model.layers.0.self_attn.v_proj": true,
26
+ "model.layers.0.self_attn.o_proj": true,
27
+ "model.layers.0.self_attn.rope": false,
28
+ "model.layers.0.mlp.gate_proj": true,
29
+ "model.layers.0.mlp.down_proj": true,
30
+ "model.layers.0.mlp.up_proj": true,
31
+ "model.layers.0.input_layernorm": false,
32
+ "model.layers.0.post_attention_layernorm": false,
33
+ "model.layers.1.self_attn.q_proj": true,
34
+ "model.layers.1.self_attn.k_proj": true,
35
+ "model.layers.1.self_attn.v_proj": true,
36
+ "model.layers.1.self_attn.o_proj": true,
37
+ "model.layers.1.self_attn.rope": false,
38
+ "model.layers.1.mlp.gate_proj": true,
39
+ "model.layers.1.mlp.down_proj": true,
40
+ "model.layers.1.mlp.up_proj": true,
41
+ "model.layers.1.input_layernorm": false,
42
+ "model.layers.1.post_attention_layernorm": false,
43
+ "model.layers.2.self_attn.q_proj": true,
44
+ "model.layers.2.self_attn.k_proj": true,
45
+ "model.layers.2.self_attn.v_proj": true,
46
+ "model.layers.2.self_attn.o_proj": true,
47
+ "model.layers.2.self_attn.rope": false,
48
+ "model.layers.2.mlp.gate_proj": true,
49
+ "model.layers.2.mlp.down_proj": true,
50
+ "model.layers.2.mlp.up_proj": true,
51
+ "model.layers.2.input_layernorm": false,
52
+ "model.layers.2.post_attention_layernorm": false,
53
+ "model.layers.3.self_attn.q_proj": true,
54
+ "model.layers.3.self_attn.k_proj": true,
55
+ "model.layers.3.self_attn.v_proj": true,
56
+ "model.layers.3.self_attn.o_proj": true,
57
+ "model.layers.3.self_attn.rope": false,
58
+ "model.layers.3.mlp.gate_proj": true,
59
+ "model.layers.3.mlp.down_proj": true,
60
+ "model.layers.3.mlp.up_proj": true,
61
+ "model.layers.3.input_layernorm": false,
62
+ "model.layers.3.post_attention_layernorm": false,
63
+ "model.layers.4.self_attn.q_proj": true,
64
+ "model.layers.4.self_attn.k_proj": true,
65
+ "model.layers.4.self_attn.v_proj": true,
66
+ "model.layers.4.self_attn.o_proj": true,
67
+ "model.layers.4.self_attn.rope": false,
68
+ "model.layers.4.mlp.gate_proj": true,
69
+ "model.layers.4.mlp.down_proj": true,
70
+ "model.layers.4.mlp.up_proj": true,
71
+ "model.layers.4.input_layernorm": false,
72
+ "model.layers.4.post_attention_layernorm": false,
73
+ "model.layers.5.self_attn.q_proj": true,
74
+ "model.layers.5.self_attn.k_proj": true,
75
+ "model.layers.5.self_attn.v_proj": true,
76
+ "model.layers.5.self_attn.o_proj": true,
77
+ "model.layers.5.self_attn.rope": false,
78
+ "model.layers.5.mlp.gate_proj": true,
79
+ "model.layers.5.mlp.down_proj": true,
80
+ "model.layers.5.mlp.up_proj": true,
81
+ "model.layers.5.input_layernorm": false,
82
+ "model.layers.5.post_attention_layernorm": false,
83
+ "model.layers.6.self_attn.q_proj": true,
84
+ "model.layers.6.self_attn.k_proj": true,
85
+ "model.layers.6.self_attn.v_proj": true,
86
+ "model.layers.6.self_attn.o_proj": true,
87
+ "model.layers.6.self_attn.rope": false,
88
+ "model.layers.6.mlp.gate_proj": true,
89
+ "model.layers.6.mlp.down_proj": true,
90
+ "model.layers.6.mlp.up_proj": true,
91
+ "model.layers.6.input_layernorm": false,
92
+ "model.layers.6.post_attention_layernorm": false,
93
+ "model.layers.7.self_attn.q_proj": true,
94
+ "model.layers.7.self_attn.k_proj": true,
95
+ "model.layers.7.self_attn.v_proj": true,
96
+ "model.layers.7.self_attn.o_proj": true,
97
+ "model.layers.7.self_attn.rope": false,
98
+ "model.layers.7.mlp.gate_proj": true,
99
+ "model.layers.7.mlp.down_proj": true,
100
+ "model.layers.7.mlp.up_proj": true,
101
+ "model.layers.7.input_layernorm": false,
102
+ "model.layers.7.post_attention_layernorm": false,
103
+ "model.layers.8.self_attn.q_proj": true,
104
+ "model.layers.8.self_attn.k_proj": true,
105
+ "model.layers.8.self_attn.v_proj": true,
106
+ "model.layers.8.self_attn.o_proj": true,
107
+ "model.layers.8.self_attn.rope": false,
108
+ "model.layers.8.mlp.gate_proj": true,
109
+ "model.layers.8.mlp.down_proj": true,
110
+ "model.layers.8.mlp.up_proj": true,
111
+ "model.layers.8.input_layernorm": false,
112
+ "model.layers.8.post_attention_layernorm": false,
113
+ "model.layers.9.self_attn.q_proj": true,
114
+ "model.layers.9.self_attn.k_proj": true,
115
+ "model.layers.9.self_attn.v_proj": true,
116
+ "model.layers.9.self_attn.o_proj": true,
117
+ "model.layers.9.self_attn.rope": false,
118
+ "model.layers.9.mlp.gate_proj": true,
119
+ "model.layers.9.mlp.down_proj": true,
120
+ "model.layers.9.mlp.up_proj": true,
121
+ "model.layers.9.input_layernorm": false,
122
+ "model.layers.9.post_attention_layernorm": false,
123
+ "model.layers.10.self_attn.q_proj": true,
124
+ "model.layers.10.self_attn.k_proj": true,
125
+ "model.layers.10.self_attn.v_proj": true,
126
+ "model.layers.10.self_attn.o_proj": true,
127
+ "model.layers.10.self_attn.rope": false,
128
+ "model.layers.10.mlp.gate_proj": true,
129
+ "model.layers.10.mlp.down_proj": true,
130
+ "model.layers.10.mlp.up_proj": true,
131
+ "model.layers.10.input_layernorm": false,
132
+ "model.layers.10.post_attention_layernorm": false,
133
+ "model.layers.11.self_attn.q_proj": true,
134
+ "model.layers.11.self_attn.k_proj": true,
135
+ "model.layers.11.self_attn.v_proj": true,
136
+ "model.layers.11.self_attn.o_proj": true,
137
+ "model.layers.11.self_attn.rope": false,
138
+ "model.layers.11.mlp.gate_proj": true,
139
+ "model.layers.11.mlp.down_proj": true,
140
+ "model.layers.11.mlp.up_proj": true,
141
+ "model.layers.11.input_layernorm": false,
142
+ "model.layers.11.post_attention_layernorm": false,
143
+ "model.layers.12.self_attn.q_proj": true,
144
+ "model.layers.12.self_attn.k_proj": true,
145
+ "model.layers.12.self_attn.v_proj": true,
146
+ "model.layers.12.self_attn.o_proj": true,
147
+ "model.layers.12.self_attn.rope": false,
148
+ "model.layers.12.mlp.gate_proj": true,
149
+ "model.layers.12.mlp.down_proj": true,
150
+ "model.layers.12.mlp.up_proj": true,
151
+ "model.layers.12.input_layernorm": false,
152
+ "model.layers.12.post_attention_layernorm": false,
153
+ "model.layers.13.self_attn.q_proj": true,
154
+ "model.layers.13.self_attn.k_proj": true,
155
+ "model.layers.13.self_attn.v_proj": true,
156
+ "model.layers.13.self_attn.o_proj": true,
157
+ "model.layers.13.self_attn.rope": false,
158
+ "model.layers.13.mlp.gate_proj": true,
159
+ "model.layers.13.mlp.down_proj": true,
160
+ "model.layers.13.mlp.up_proj": true,
161
+ "model.layers.13.input_layernorm": false,
162
+ "model.layers.13.post_attention_layernorm": false,
163
+ "model.layers.14.self_attn.q_proj": true,
164
+ "model.layers.14.self_attn.k_proj": true,
165
+ "model.layers.14.self_attn.v_proj": true,
166
+ "model.layers.14.self_attn.o_proj": true,
167
+ "model.layers.14.self_attn.rope": false,
168
+ "model.layers.14.mlp.gate_proj": true,
169
+ "model.layers.14.mlp.down_proj": true,
170
+ "model.layers.14.mlp.up_proj": true,
171
+ "model.layers.14.input_layernorm": false,
172
+ "model.layers.14.post_attention_layernorm": false,
173
+ "model.layers.15.self_attn.q_proj": true,
174
+ "model.layers.15.self_attn.k_proj": true,
175
+ "model.layers.15.self_attn.v_proj": true,
176
+ "model.layers.15.self_attn.o_proj": true,
177
+ "model.layers.15.self_attn.rope": false,
178
+ "model.layers.15.mlp.gate_proj": true,
179
+ "model.layers.15.mlp.down_proj": true,
180
+ "model.layers.15.mlp.up_proj": true,
181
+ "model.layers.15.input_layernorm": false,
182
+ "model.layers.15.post_attention_layernorm": false,
183
+ "model.layers.16.self_attn.q_proj": true,
184
+ "model.layers.16.self_attn.k_proj": true,
185
+ "model.layers.16.self_attn.v_proj": true,
186
+ "model.layers.16.self_attn.o_proj": true,
187
+ "model.layers.16.self_attn.rope": false,
188
+ "model.layers.16.mlp.gate_proj": true,
189
+ "model.layers.16.mlp.down_proj": true,
190
+ "model.layers.16.mlp.up_proj": true,
191
+ "model.layers.16.input_layernorm": false,
192
+ "model.layers.16.post_attention_layernorm": false,
193
+ "model.layers.17.self_attn.q_proj": true,
194
+ "model.layers.17.self_attn.k_proj": true,
195
+ "model.layers.17.self_attn.v_proj": true,
196
+ "model.layers.17.self_attn.o_proj": true,
197
+ "model.layers.17.self_attn.rope": false,
198
+ "model.layers.17.mlp.gate_proj": true,
199
+ "model.layers.17.mlp.down_proj": true,
200
+ "model.layers.17.mlp.up_proj": true,
201
+ "model.layers.17.input_layernorm": false,
202
+ "model.layers.17.post_attention_layernorm": false,
203
+ "model.layers.18.self_attn.q_proj": true,
204
+ "model.layers.18.self_attn.k_proj": true,
205
+ "model.layers.18.self_attn.v_proj": true,
206
+ "model.layers.18.self_attn.o_proj": true,
207
+ "model.layers.18.self_attn.rope": false,
208
+ "model.layers.18.mlp.gate_proj": true,
209
+ "model.layers.18.mlp.down_proj": true,
210
+ "model.layers.18.mlp.up_proj": true,
211
+ "model.layers.18.input_layernorm": false,
212
+ "model.layers.18.post_attention_layernorm": false,
213
+ "model.layers.19.self_attn.q_proj": true,
214
+ "model.layers.19.self_attn.k_proj": true,
215
+ "model.layers.19.self_attn.v_proj": true,
216
+ "model.layers.19.self_attn.o_proj": true,
217
+ "model.layers.19.self_attn.rope": false,
218
+ "model.layers.19.mlp.gate_proj": true,
219
+ "model.layers.19.mlp.down_proj": true,
220
+ "model.layers.19.mlp.up_proj": true,
221
+ "model.layers.19.input_layernorm": false,
222
+ "model.layers.19.post_attention_layernorm": false,
223
+ "model.layers.20.self_attn.q_proj": true,
224
+ "model.layers.20.self_attn.k_proj": true,
225
+ "model.layers.20.self_attn.v_proj": true,
226
+ "model.layers.20.self_attn.o_proj": true,
227
+ "model.layers.20.self_attn.rope": false,
228
+ "model.layers.20.mlp.gate_proj": true,
229
+ "model.layers.20.mlp.down_proj": true,
230
+ "model.layers.20.mlp.up_proj": true,
231
+ "model.layers.20.input_layernorm": false,
232
+ "model.layers.20.post_attention_layernorm": false,
233
+ "model.layers.21.self_attn.q_proj": true,
234
+ "model.layers.21.self_attn.k_proj": true,
235
+ "model.layers.21.self_attn.v_proj": true,
236
+ "model.layers.21.self_attn.o_proj": true,
237
+ "model.layers.21.self_attn.rope": false,
238
+ "model.layers.21.mlp.gate_proj": true,
239
+ "model.layers.21.mlp.down_proj": true,
240
+ "model.layers.21.mlp.up_proj": true,
241
+ "model.layers.21.input_layernorm": false,
242
+ "model.layers.21.post_attention_layernorm": false,
243
+ "model.layers.22.self_attn.q_proj": true,
244
+ "model.layers.22.self_attn.k_proj": true,
245
+ "model.layers.22.self_attn.v_proj": true,
246
+ "model.layers.22.self_attn.o_proj": true,
247
+ "model.layers.22.self_attn.rope": false,
248
+ "model.layers.22.mlp.gate_proj": true,
249
+ "model.layers.22.mlp.down_proj": true,
250
+ "model.layers.22.mlp.up_proj": true,
251
+ "model.layers.22.input_layernorm": false,
252
+ "model.layers.22.post_attention_layernorm": false,
253
+ "model.layers.23.self_attn.q_proj": true,
254
+ "model.layers.23.self_attn.k_proj": true,
255
+ "model.layers.23.self_attn.v_proj": true,
256
+ "model.layers.23.self_attn.o_proj": true,
257
+ "model.layers.23.self_attn.rope": false,
258
+ "model.layers.23.mlp.gate_proj": true,
259
+ "model.layers.23.mlp.down_proj": true,
260
+ "model.layers.23.mlp.up_proj": true,
261
+ "model.layers.23.input_layernorm": false,
262
+ "model.layers.23.post_attention_layernorm": false,
263
+ "model.layers.24.self_attn.q_proj": true,
264
+ "model.layers.24.self_attn.k_proj": true,
265
+ "model.layers.24.self_attn.v_proj": true,
266
+ "model.layers.24.self_attn.o_proj": true,
267
+ "model.layers.24.self_attn.rope": false,
268
+ "model.layers.24.mlp.gate_proj": true,
269
+ "model.layers.24.mlp.down_proj": true,
270
+ "model.layers.24.mlp.up_proj": true,
271
+ "model.layers.24.input_layernorm": false,
272
+ "model.layers.24.post_attention_layernorm": false,
273
+ "model.layers.25.self_attn.q_proj": true,
274
+ "model.layers.25.self_attn.k_proj": true,
275
+ "model.layers.25.self_attn.v_proj": true,
276
+ "model.layers.25.self_attn.o_proj": true,
277
+ "model.layers.25.self_attn.rope": false,
278
+ "model.layers.25.mlp.gate_proj": true,
279
+ "model.layers.25.mlp.down_proj": true,
280
+ "model.layers.25.mlp.up_proj": true,
281
+ "model.layers.25.input_layernorm": false,
282
+ "model.layers.25.post_attention_layernorm": false,
283
+ "model.layers.26.self_attn.q_proj": true,
284
+ "model.layers.26.self_attn.k_proj": true,
285
+ "model.layers.26.self_attn.v_proj": true,
286
+ "model.layers.26.self_attn.o_proj": true,
287
+ "model.layers.26.self_attn.rope": false,
288
+ "model.layers.26.mlp.gate_proj": true,
289
+ "model.layers.26.mlp.down_proj": true,
290
+ "model.layers.26.mlp.up_proj": true,
291
+ "model.layers.26.input_layernorm": false,
292
+ "model.layers.26.post_attention_layernorm": false,
293
+ "model.layers.27.self_attn.q_proj": true,
294
+ "model.layers.27.self_attn.k_proj": true,
295
+ "model.layers.27.self_attn.v_proj": true,
296
+ "model.layers.27.self_attn.o_proj": true,
297
+ "model.layers.27.self_attn.rope": false,
298
+ "model.layers.27.mlp.gate_proj": true,
299
+ "model.layers.27.mlp.down_proj": true,
300
+ "model.layers.27.mlp.up_proj": true,
301
+ "model.layers.27.input_layernorm": false,
302
+ "model.layers.27.post_attention_layernorm": false,
303
+ "model.layers.28.self_attn.q_proj": true,
304
+ "model.layers.28.self_attn.k_proj": true,
305
+ "model.layers.28.self_attn.v_proj": true,
306
+ "model.layers.28.self_attn.o_proj": true,
307
+ "model.layers.28.self_attn.rope": false,
308
+ "model.layers.28.mlp.gate_proj": true,
309
+ "model.layers.28.mlp.down_proj": true,
310
+ "model.layers.28.mlp.up_proj": true,
311
+ "model.layers.28.input_layernorm": false,
312
+ "model.layers.28.post_attention_layernorm": false,
313
+ "model.layers.29.self_attn.q_proj": true,
314
+ "model.layers.29.self_attn.k_proj": true,
315
+ "model.layers.29.self_attn.v_proj": true,
316
+ "model.layers.29.self_attn.o_proj": true,
317
+ "model.layers.29.self_attn.rope": false,
318
+ "model.layers.29.mlp.gate_proj": true,
319
+ "model.layers.29.mlp.down_proj": true,
320
+ "model.layers.29.mlp.up_proj": true,
321
+ "model.layers.29.input_layernorm": false,
322
+ "model.layers.29.post_attention_layernorm": false,
323
+ "model.layers.30.self_attn.q_proj": true,
324
+ "model.layers.30.self_attn.k_proj": true,
325
+ "model.layers.30.self_attn.v_proj": true,
326
+ "model.layers.30.self_attn.o_proj": true,
327
+ "model.layers.30.self_attn.rope": false,
328
+ "model.layers.30.mlp.gate_proj": true,
329
+ "model.layers.30.mlp.down_proj": true,
330
+ "model.layers.30.mlp.up_proj": true,
331
+ "model.layers.30.input_layernorm": false,
332
+ "model.layers.30.post_attention_layernorm": false,
333
+ "model.layers.31.self_attn.q_proj": true,
334
+ "model.layers.31.self_attn.k_proj": true,
335
+ "model.layers.31.self_attn.v_proj": true,
336
+ "model.layers.31.self_attn.o_proj": true,
337
+ "model.layers.31.self_attn.rope": false,
338
+ "model.layers.31.mlp.gate_proj": true,
339
+ "model.layers.31.mlp.down_proj": true,
340
+ "model.layers.31.mlp.up_proj": true,
341
+ "model.layers.31.input_layernorm": false,
342
+ "model.layers.31.post_attention_layernorm": false,
343
+ "model.layers.32.self_attn.q_proj": true,
344
+ "model.layers.32.self_attn.k_proj": true,
345
+ "model.layers.32.self_attn.v_proj": true,
346
+ "model.layers.32.self_attn.o_proj": true,
347
+ "model.layers.32.self_attn.rope": false,
348
+ "model.layers.32.mlp.gate_proj": true,
349
+ "model.layers.32.mlp.down_proj": true,
350
+ "model.layers.32.mlp.up_proj": true,
351
+ "model.layers.32.input_layernorm": false,
352
+ "model.layers.32.post_attention_layernorm": false,
353
+ "model.layers.33.self_attn.q_proj": true,
354
+ "model.layers.33.self_attn.k_proj": true,
355
+ "model.layers.33.self_attn.v_proj": true,
356
+ "model.layers.33.self_attn.o_proj": true,
357
+ "model.layers.33.self_attn.rope": false,
358
+ "model.layers.33.mlp.gate_proj": true,
359
+ "model.layers.33.mlp.down_proj": true,
360
+ "model.layers.33.mlp.up_proj": true,
361
+ "model.layers.33.input_layernorm": false,
362
+ "model.layers.33.post_attention_layernorm": false,
363
+ "model.layers.34.self_attn.q_proj": true,
364
+ "model.layers.34.self_attn.k_proj": true,
365
+ "model.layers.34.self_attn.v_proj": true,
366
+ "model.layers.34.self_attn.o_proj": true,
367
+ "model.layers.34.self_attn.rope": false,
368
+ "model.layers.34.mlp.gate_proj": true,
369
+ "model.layers.34.mlp.down_proj": true,
370
+ "model.layers.34.mlp.up_proj": true,
371
+ "model.layers.34.input_layernorm": false,
372
+ "model.layers.34.post_attention_layernorm": false,
373
+ "model.layers.35.self_attn.q_proj": true,
374
+ "model.layers.35.self_attn.k_proj": true,
375
+ "model.layers.35.self_attn.v_proj": true,
376
+ "model.layers.35.self_attn.o_proj": true,
377
+ "model.layers.35.self_attn.rope": false,
378
+ "model.layers.35.mlp.gate_proj": true,
379
+ "model.layers.35.mlp.down_proj": true,
380
+ "model.layers.35.mlp.up_proj": true,
381
+ "model.layers.35.input_layernorm": false,
382
+ "model.layers.35.post_attention_layernorm": false,
383
+ "model.layers.36.self_attn.q_proj": true,
384
+ "model.layers.36.self_attn.k_proj": true,
385
+ "model.layers.36.self_attn.v_proj": true,
386
+ "model.layers.36.self_attn.o_proj": true,
387
+ "model.layers.36.self_attn.rope": false,
388
+ "model.layers.36.mlp.gate_proj": true,
389
+ "model.layers.36.mlp.down_proj": true,
390
+ "model.layers.36.mlp.up_proj": true,
391
+ "model.layers.36.input_layernorm": false,
392
+ "model.layers.36.post_attention_layernorm": false,
393
+ "model.layers.37.self_attn.q_proj": true,
394
+ "model.layers.37.self_attn.k_proj": true,
395
+ "model.layers.37.self_attn.v_proj": true,
396
+ "model.layers.37.self_attn.o_proj": true,
397
+ "model.layers.37.self_attn.rope": false,
398
+ "model.layers.37.mlp.gate_proj": true,
399
+ "model.layers.37.mlp.down_proj": true,
400
+ "model.layers.37.mlp.up_proj": true,
401
+ "model.layers.37.input_layernorm": false,
402
+ "model.layers.37.post_attention_layernorm": false,
403
+ "model.layers.38.self_attn.q_proj": true,
404
+ "model.layers.38.self_attn.k_proj": true,
405
+ "model.layers.38.self_attn.v_proj": true,
406
+ "model.layers.38.self_attn.o_proj": true,
407
+ "model.layers.38.self_attn.rope": false,
408
+ "model.layers.38.mlp.gate_proj": true,
409
+ "model.layers.38.mlp.down_proj": true,
410
+ "model.layers.38.mlp.up_proj": true,
411
+ "model.layers.38.input_layernorm": false,
412
+ "model.layers.38.post_attention_layernorm": false,
413
+ "model.layers.39.self_attn.q_proj": true,
414
+ "model.layers.39.self_attn.k_proj": true,
415
+ "model.layers.39.self_attn.v_proj": true,
416
+ "model.layers.39.self_attn.o_proj": true,
417
+ "model.layers.39.self_attn.rope": false,
418
+ "model.layers.39.mlp.gate_proj": true,
419
+ "model.layers.39.mlp.down_proj": true,
420
+ "model.layers.39.mlp.up_proj": true,
421
+ "model.layers.39.input_layernorm": false,
422
+ "model.layers.39.post_attention_layernorm": false,
423
+ "model.norm": false,
424
+ "lm_head": true
425
+ },
426
+ "quantization_config": {
427
+ "group_size": 64,
428
+ "bits": 4,
429
+ "model.embed_tokens": true,
430
+ "model.layers.0.self_attn.q_proj": true,
431
+ "model.layers.0.self_attn.k_proj": true,
432
+ "model.layers.0.self_attn.v_proj": true,
433
+ "model.layers.0.self_attn.o_proj": true,
434
+ "model.layers.0.self_attn.rope": false,
435
+ "model.layers.0.mlp.gate_proj": true,
436
+ "model.layers.0.mlp.down_proj": true,
437
+ "model.layers.0.mlp.up_proj": true,
438
+ "model.layers.0.input_layernorm": false,
439
+ "model.layers.0.post_attention_layernorm": false,
440
+ "model.layers.1.self_attn.q_proj": true,
441
+ "model.layers.1.self_attn.k_proj": true,
442
+ "model.layers.1.self_attn.v_proj": true,
443
+ "model.layers.1.self_attn.o_proj": true,
444
+ "model.layers.1.self_attn.rope": false,
445
+ "model.layers.1.mlp.gate_proj": true,
446
+ "model.layers.1.mlp.down_proj": true,
447
+ "model.layers.1.mlp.up_proj": true,
448
+ "model.layers.1.input_layernorm": false,
449
+ "model.layers.1.post_attention_layernorm": false,
450
+ "model.layers.2.self_attn.q_proj": true,
451
+ "model.layers.2.self_attn.k_proj": true,
452
+ "model.layers.2.self_attn.v_proj": true,
453
+ "model.layers.2.self_attn.o_proj": true,
454
+ "model.layers.2.self_attn.rope": false,
455
+ "model.layers.2.mlp.gate_proj": true,
456
+ "model.layers.2.mlp.down_proj": true,
457
+ "model.layers.2.mlp.up_proj": true,
458
+ "model.layers.2.input_layernorm": false,
459
+ "model.layers.2.post_attention_layernorm": false,
460
+ "model.layers.3.self_attn.q_proj": true,
461
+ "model.layers.3.self_attn.k_proj": true,
462
+ "model.layers.3.self_attn.v_proj": true,
463
+ "model.layers.3.self_attn.o_proj": true,
464
+ "model.layers.3.self_attn.rope": false,
465
+ "model.layers.3.mlp.gate_proj": true,
466
+ "model.layers.3.mlp.down_proj": true,
467
+ "model.layers.3.mlp.up_proj": true,
468
+ "model.layers.3.input_layernorm": false,
469
+ "model.layers.3.post_attention_layernorm": false,
470
+ "model.layers.4.self_attn.q_proj": true,
471
+ "model.layers.4.self_attn.k_proj": true,
472
+ "model.layers.4.self_attn.v_proj": true,
473
+ "model.layers.4.self_attn.o_proj": true,
474
+ "model.layers.4.self_attn.rope": false,
475
+ "model.layers.4.mlp.gate_proj": true,
476
+ "model.layers.4.mlp.down_proj": true,
477
+ "model.layers.4.mlp.up_proj": true,
478
+ "model.layers.4.input_layernorm": false,
479
+ "model.layers.4.post_attention_layernorm": false,
480
+ "model.layers.5.self_attn.q_proj": true,
481
+ "model.layers.5.self_attn.k_proj": true,
482
+ "model.layers.5.self_attn.v_proj": true,
483
+ "model.layers.5.self_attn.o_proj": true,
484
+ "model.layers.5.self_attn.rope": false,
485
+ "model.layers.5.mlp.gate_proj": true,
486
+ "model.layers.5.mlp.down_proj": true,
487
+ "model.layers.5.mlp.up_proj": true,
488
+ "model.layers.5.input_layernorm": false,
489
+ "model.layers.5.post_attention_layernorm": false,
490
+ "model.layers.6.self_attn.q_proj": true,
491
+ "model.layers.6.self_attn.k_proj": true,
492
+ "model.layers.6.self_attn.v_proj": true,
493
+ "model.layers.6.self_attn.o_proj": true,
494
+ "model.layers.6.self_attn.rope": false,
495
+ "model.layers.6.mlp.gate_proj": true,
496
+ "model.layers.6.mlp.down_proj": true,
497
+ "model.layers.6.mlp.up_proj": true,
498
+ "model.layers.6.input_layernorm": false,
499
+ "model.layers.6.post_attention_layernorm": false,
500
+ "model.layers.7.self_attn.q_proj": true,
501
+ "model.layers.7.self_attn.k_proj": true,
502
+ "model.layers.7.self_attn.v_proj": true,
503
+ "model.layers.7.self_attn.o_proj": true,
504
+ "model.layers.7.self_attn.rope": false,
505
+ "model.layers.7.mlp.gate_proj": true,
506
+ "model.layers.7.mlp.down_proj": true,
507
+ "model.layers.7.mlp.up_proj": true,
508
+ "model.layers.7.input_layernorm": false,
509
+ "model.layers.7.post_attention_layernorm": false,
510
+ "model.layers.8.self_attn.q_proj": true,
511
+ "model.layers.8.self_attn.k_proj": true,
512
+ "model.layers.8.self_attn.v_proj": true,
513
+ "model.layers.8.self_attn.o_proj": true,
514
+ "model.layers.8.self_attn.rope": false,
515
+ "model.layers.8.mlp.gate_proj": true,
516
+ "model.layers.8.mlp.down_proj": true,
517
+ "model.layers.8.mlp.up_proj": true,
518
+ "model.layers.8.input_layernorm": false,
519
+ "model.layers.8.post_attention_layernorm": false,
520
+ "model.layers.9.self_attn.q_proj": true,
521
+ "model.layers.9.self_attn.k_proj": true,
522
+ "model.layers.9.self_attn.v_proj": true,
523
+ "model.layers.9.self_attn.o_proj": true,
524
+ "model.layers.9.self_attn.rope": false,
525
+ "model.layers.9.mlp.gate_proj": true,
526
+ "model.layers.9.mlp.down_proj": true,
527
+ "model.layers.9.mlp.up_proj": true,
528
+ "model.layers.9.input_layernorm": false,
529
+ "model.layers.9.post_attention_layernorm": false,
530
+ "model.layers.10.self_attn.q_proj": true,
531
+ "model.layers.10.self_attn.k_proj": true,
532
+ "model.layers.10.self_attn.v_proj": true,
533
+ "model.layers.10.self_attn.o_proj": true,
534
+ "model.layers.10.self_attn.rope": false,
535
+ "model.layers.10.mlp.gate_proj": true,
536
+ "model.layers.10.mlp.down_proj": true,
537
+ "model.layers.10.mlp.up_proj": true,
538
+ "model.layers.10.input_layernorm": false,
539
+ "model.layers.10.post_attention_layernorm": false,
540
+ "model.layers.11.self_attn.q_proj": true,
541
+ "model.layers.11.self_attn.k_proj": true,
542
+ "model.layers.11.self_attn.v_proj": true,
543
+ "model.layers.11.self_attn.o_proj": true,
544
+ "model.layers.11.self_attn.rope": false,
545
+ "model.layers.11.mlp.gate_proj": true,
546
+ "model.layers.11.mlp.down_proj": true,
547
+ "model.layers.11.mlp.up_proj": true,
548
+ "model.layers.11.input_layernorm": false,
549
+ "model.layers.11.post_attention_layernorm": false,
550
+ "model.layers.12.self_attn.q_proj": true,
551
+ "model.layers.12.self_attn.k_proj": true,
552
+ "model.layers.12.self_attn.v_proj": true,
553
+ "model.layers.12.self_attn.o_proj": true,
554
+ "model.layers.12.self_attn.rope": false,
555
+ "model.layers.12.mlp.gate_proj": true,
556
+ "model.layers.12.mlp.down_proj": true,
557
+ "model.layers.12.mlp.up_proj": true,
558
+ "model.layers.12.input_layernorm": false,
559
+ "model.layers.12.post_attention_layernorm": false,
560
+ "model.layers.13.self_attn.q_proj": true,
561
+ "model.layers.13.self_attn.k_proj": true,
562
+ "model.layers.13.self_attn.v_proj": true,
563
+ "model.layers.13.self_attn.o_proj": true,
564
+ "model.layers.13.self_attn.rope": false,
565
+ "model.layers.13.mlp.gate_proj": true,
566
+ "model.layers.13.mlp.down_proj": true,
567
+ "model.layers.13.mlp.up_proj": true,
568
+ "model.layers.13.input_layernorm": false,
569
+ "model.layers.13.post_attention_layernorm": false,
570
+ "model.layers.14.self_attn.q_proj": true,
571
+ "model.layers.14.self_attn.k_proj": true,
572
+ "model.layers.14.self_attn.v_proj": true,
573
+ "model.layers.14.self_attn.o_proj": true,
574
+ "model.layers.14.self_attn.rope": false,
575
+ "model.layers.14.mlp.gate_proj": true,
576
+ "model.layers.14.mlp.down_proj": true,
577
+ "model.layers.14.mlp.up_proj": true,
578
+ "model.layers.14.input_layernorm": false,
579
+ "model.layers.14.post_attention_layernorm": false,
580
+ "model.layers.15.self_attn.q_proj": true,
581
+ "model.layers.15.self_attn.k_proj": true,
582
+ "model.layers.15.self_attn.v_proj": true,
583
+ "model.layers.15.self_attn.o_proj": true,
584
+ "model.layers.15.self_attn.rope": false,
585
+ "model.layers.15.mlp.gate_proj": true,
586
+ "model.layers.15.mlp.down_proj": true,
587
+ "model.layers.15.mlp.up_proj": true,
588
+ "model.layers.15.input_layernorm": false,
589
+ "model.layers.15.post_attention_layernorm": false,
590
+ "model.layers.16.self_attn.q_proj": true,
591
+ "model.layers.16.self_attn.k_proj": true,
592
+ "model.layers.16.self_attn.v_proj": true,
593
+ "model.layers.16.self_attn.o_proj": true,
594
+ "model.layers.16.self_attn.rope": false,
595
+ "model.layers.16.mlp.gate_proj": true,
596
+ "model.layers.16.mlp.down_proj": true,
597
+ "model.layers.16.mlp.up_proj": true,
598
+ "model.layers.16.input_layernorm": false,
599
+ "model.layers.16.post_attention_layernorm": false,
600
+ "model.layers.17.self_attn.q_proj": true,
601
+ "model.layers.17.self_attn.k_proj": true,
602
+ "model.layers.17.self_attn.v_proj": true,
603
+ "model.layers.17.self_attn.o_proj": true,
604
+ "model.layers.17.self_attn.rope": false,
605
+ "model.layers.17.mlp.gate_proj": true,
606
+ "model.layers.17.mlp.down_proj": true,
607
+ "model.layers.17.mlp.up_proj": true,
608
+ "model.layers.17.input_layernorm": false,
609
+ "model.layers.17.post_attention_layernorm": false,
610
+ "model.layers.18.self_attn.q_proj": true,
611
+ "model.layers.18.self_attn.k_proj": true,
612
+ "model.layers.18.self_attn.v_proj": true,
613
+ "model.layers.18.self_attn.o_proj": true,
614
+ "model.layers.18.self_attn.rope": false,
615
+ "model.layers.18.mlp.gate_proj": true,
616
+ "model.layers.18.mlp.down_proj": true,
617
+ "model.layers.18.mlp.up_proj": true,
618
+ "model.layers.18.input_layernorm": false,
619
+ "model.layers.18.post_attention_layernorm": false,
620
+ "model.layers.19.self_attn.q_proj": true,
621
+ "model.layers.19.self_attn.k_proj": true,
622
+ "model.layers.19.self_attn.v_proj": true,
623
+ "model.layers.19.self_attn.o_proj": true,
624
+ "model.layers.19.self_attn.rope": false,
625
+ "model.layers.19.mlp.gate_proj": true,
626
+ "model.layers.19.mlp.down_proj": true,
627
+ "model.layers.19.mlp.up_proj": true,
628
+ "model.layers.19.input_layernorm": false,
629
+ "model.layers.19.post_attention_layernorm": false,
630
+ "model.layers.20.self_attn.q_proj": true,
631
+ "model.layers.20.self_attn.k_proj": true,
632
+ "model.layers.20.self_attn.v_proj": true,
633
+ "model.layers.20.self_attn.o_proj": true,
634
+ "model.layers.20.self_attn.rope": false,
635
+ "model.layers.20.mlp.gate_proj": true,
636
+ "model.layers.20.mlp.down_proj": true,
637
+ "model.layers.20.mlp.up_proj": true,
638
+ "model.layers.20.input_layernorm": false,
639
+ "model.layers.20.post_attention_layernorm": false,
640
+ "model.layers.21.self_attn.q_proj": true,
641
+ "model.layers.21.self_attn.k_proj": true,
642
+ "model.layers.21.self_attn.v_proj": true,
643
+ "model.layers.21.self_attn.o_proj": true,
644
+ "model.layers.21.self_attn.rope": false,
645
+ "model.layers.21.mlp.gate_proj": true,
646
+ "model.layers.21.mlp.down_proj": true,
647
+ "model.layers.21.mlp.up_proj": true,
648
+ "model.layers.21.input_layernorm": false,
649
+ "model.layers.21.post_attention_layernorm": false,
650
+ "model.layers.22.self_attn.q_proj": true,
651
+ "model.layers.22.self_attn.k_proj": true,
652
+ "model.layers.22.self_attn.v_proj": true,
653
+ "model.layers.22.self_attn.o_proj": true,
654
+ "model.layers.22.self_attn.rope": false,
655
+ "model.layers.22.mlp.gate_proj": true,
656
+ "model.layers.22.mlp.down_proj": true,
657
+ "model.layers.22.mlp.up_proj": true,
658
+ "model.layers.22.input_layernorm": false,
659
+ "model.layers.22.post_attention_layernorm": false,
660
+ "model.layers.23.self_attn.q_proj": true,
661
+ "model.layers.23.self_attn.k_proj": true,
662
+ "model.layers.23.self_attn.v_proj": true,
663
+ "model.layers.23.self_attn.o_proj": true,
664
+ "model.layers.23.self_attn.rope": false,
665
+ "model.layers.23.mlp.gate_proj": true,
666
+ "model.layers.23.mlp.down_proj": true,
667
+ "model.layers.23.mlp.up_proj": true,
668
+ "model.layers.23.input_layernorm": false,
669
+ "model.layers.23.post_attention_layernorm": false,
670
+ "model.layers.24.self_attn.q_proj": true,
671
+ "model.layers.24.self_attn.k_proj": true,
672
+ "model.layers.24.self_attn.v_proj": true,
673
+ "model.layers.24.self_attn.o_proj": true,
674
+ "model.layers.24.self_attn.rope": false,
675
+ "model.layers.24.mlp.gate_proj": true,
676
+ "model.layers.24.mlp.down_proj": true,
677
+ "model.layers.24.mlp.up_proj": true,
678
+ "model.layers.24.input_layernorm": false,
679
+ "model.layers.24.post_attention_layernorm": false,
680
+ "model.layers.25.self_attn.q_proj": true,
681
+ "model.layers.25.self_attn.k_proj": true,
682
+ "model.layers.25.self_attn.v_proj": true,
683
+ "model.layers.25.self_attn.o_proj": true,
684
+ "model.layers.25.self_attn.rope": false,
685
+ "model.layers.25.mlp.gate_proj": true,
686
+ "model.layers.25.mlp.down_proj": true,
687
+ "model.layers.25.mlp.up_proj": true,
688
+ "model.layers.25.input_layernorm": false,
689
+ "model.layers.25.post_attention_layernorm": false,
690
+ "model.layers.26.self_attn.q_proj": true,
691
+ "model.layers.26.self_attn.k_proj": true,
692
+ "model.layers.26.self_attn.v_proj": true,
693
+ "model.layers.26.self_attn.o_proj": true,
694
+ "model.layers.26.self_attn.rope": false,
695
+ "model.layers.26.mlp.gate_proj": true,
696
+ "model.layers.26.mlp.down_proj": true,
697
+ "model.layers.26.mlp.up_proj": true,
698
+ "model.layers.26.input_layernorm": false,
699
+ "model.layers.26.post_attention_layernorm": false,
700
+ "model.layers.27.self_attn.q_proj": true,
701
+ "model.layers.27.self_attn.k_proj": true,
702
+ "model.layers.27.self_attn.v_proj": true,
703
+ "model.layers.27.self_attn.o_proj": true,
704
+ "model.layers.27.self_attn.rope": false,
705
+ "model.layers.27.mlp.gate_proj": true,
706
+ "model.layers.27.mlp.down_proj": true,
707
+ "model.layers.27.mlp.up_proj": true,
708
+ "model.layers.27.input_layernorm": false,
709
+ "model.layers.27.post_attention_layernorm": false,
710
+ "model.layers.28.self_attn.q_proj": true,
711
+ "model.layers.28.self_attn.k_proj": true,
712
+ "model.layers.28.self_attn.v_proj": true,
713
+ "model.layers.28.self_attn.o_proj": true,
714
+ "model.layers.28.self_attn.rope": false,
715
+ "model.layers.28.mlp.gate_proj": true,
716
+ "model.layers.28.mlp.down_proj": true,
717
+ "model.layers.28.mlp.up_proj": true,
718
+ "model.layers.28.input_layernorm": false,
719
+ "model.layers.28.post_attention_layernorm": false,
720
+ "model.layers.29.self_attn.q_proj": true,
721
+ "model.layers.29.self_attn.k_proj": true,
722
+ "model.layers.29.self_attn.v_proj": true,
723
+ "model.layers.29.self_attn.o_proj": true,
724
+ "model.layers.29.self_attn.rope": false,
725
+ "model.layers.29.mlp.gate_proj": true,
726
+ "model.layers.29.mlp.down_proj": true,
727
+ "model.layers.29.mlp.up_proj": true,
728
+ "model.layers.29.input_layernorm": false,
729
+ "model.layers.29.post_attention_layernorm": false,
730
+ "model.layers.30.self_attn.q_proj": true,
731
+ "model.layers.30.self_attn.k_proj": true,
732
+ "model.layers.30.self_attn.v_proj": true,
733
+ "model.layers.30.self_attn.o_proj": true,
734
+ "model.layers.30.self_attn.rope": false,
735
+ "model.layers.30.mlp.gate_proj": true,
736
+ "model.layers.30.mlp.down_proj": true,
737
+ "model.layers.30.mlp.up_proj": true,
738
+ "model.layers.30.input_layernorm": false,
739
+ "model.layers.30.post_attention_layernorm": false,
740
+ "model.layers.31.self_attn.q_proj": true,
741
+ "model.layers.31.self_attn.k_proj": true,
742
+ "model.layers.31.self_attn.v_proj": true,
743
+ "model.layers.31.self_attn.o_proj": true,
744
+ "model.layers.31.self_attn.rope": false,
745
+ "model.layers.31.mlp.gate_proj": true,
746
+ "model.layers.31.mlp.down_proj": true,
747
+ "model.layers.31.mlp.up_proj": true,
748
+ "model.layers.31.input_layernorm": false,
749
+ "model.layers.31.post_attention_layernorm": false,
750
+ "model.layers.32.self_attn.q_proj": true,
751
+ "model.layers.32.self_attn.k_proj": true,
752
+ "model.layers.32.self_attn.v_proj": true,
753
+ "model.layers.32.self_attn.o_proj": true,
754
+ "model.layers.32.self_attn.rope": false,
755
+ "model.layers.32.mlp.gate_proj": true,
756
+ "model.layers.32.mlp.down_proj": true,
757
+ "model.layers.32.mlp.up_proj": true,
758
+ "model.layers.32.input_layernorm": false,
759
+ "model.layers.32.post_attention_layernorm": false,
760
+ "model.layers.33.self_attn.q_proj": true,
761
+ "model.layers.33.self_attn.k_proj": true,
762
+ "model.layers.33.self_attn.v_proj": true,
763
+ "model.layers.33.self_attn.o_proj": true,
764
+ "model.layers.33.self_attn.rope": false,
765
+ "model.layers.33.mlp.gate_proj": true,
766
+ "model.layers.33.mlp.down_proj": true,
767
+ "model.layers.33.mlp.up_proj": true,
768
+ "model.layers.33.input_layernorm": false,
769
+ "model.layers.33.post_attention_layernorm": false,
770
+ "model.layers.34.self_attn.q_proj": true,
771
+ "model.layers.34.self_attn.k_proj": true,
772
+ "model.layers.34.self_attn.v_proj": true,
773
+ "model.layers.34.self_attn.o_proj": true,
774
+ "model.layers.34.self_attn.rope": false,
775
+ "model.layers.34.mlp.gate_proj": true,
776
+ "model.layers.34.mlp.down_proj": true,
777
+ "model.layers.34.mlp.up_proj": true,
778
+ "model.layers.34.input_layernorm": false,
779
+ "model.layers.34.post_attention_layernorm": false,
780
+ "model.layers.35.self_attn.q_proj": true,
781
+ "model.layers.35.self_attn.k_proj": true,
782
+ "model.layers.35.self_attn.v_proj": true,
783
+ "model.layers.35.self_attn.o_proj": true,
784
+ "model.layers.35.self_attn.rope": false,
785
+ "model.layers.35.mlp.gate_proj": true,
786
+ "model.layers.35.mlp.down_proj": true,
787
+ "model.layers.35.mlp.up_proj": true,
788
+ "model.layers.35.input_layernorm": false,
789
+ "model.layers.35.post_attention_layernorm": false,
790
+ "model.layers.36.self_attn.q_proj": true,
791
+ "model.layers.36.self_attn.k_proj": true,
792
+ "model.layers.36.self_attn.v_proj": true,
793
+ "model.layers.36.self_attn.o_proj": true,
794
+ "model.layers.36.self_attn.rope": false,
795
+ "model.layers.36.mlp.gate_proj": true,
796
+ "model.layers.36.mlp.down_proj": true,
797
+ "model.layers.36.mlp.up_proj": true,
798
+ "model.layers.36.input_layernorm": false,
799
+ "model.layers.36.post_attention_layernorm": false,
800
+ "model.layers.37.self_attn.q_proj": true,
801
+ "model.layers.37.self_attn.k_proj": true,
802
+ "model.layers.37.self_attn.v_proj": true,
803
+ "model.layers.37.self_attn.o_proj": true,
804
+ "model.layers.37.self_attn.rope": false,
805
+ "model.layers.37.mlp.gate_proj": true,
806
+ "model.layers.37.mlp.down_proj": true,
807
+ "model.layers.37.mlp.up_proj": true,
808
+ "model.layers.37.input_layernorm": false,
809
+ "model.layers.37.post_attention_layernorm": false,
810
+ "model.layers.38.self_attn.q_proj": true,
811
+ "model.layers.38.self_attn.k_proj": true,
812
+ "model.layers.38.self_attn.v_proj": true,
813
+ "model.layers.38.self_attn.o_proj": true,
814
+ "model.layers.38.self_attn.rope": false,
815
+ "model.layers.38.mlp.gate_proj": true,
816
+ "model.layers.38.mlp.down_proj": true,
817
+ "model.layers.38.mlp.up_proj": true,
818
+ "model.layers.38.input_layernorm": false,
819
+ "model.layers.38.post_attention_layernorm": false,
820
+ "model.layers.39.self_attn.q_proj": true,
821
+ "model.layers.39.self_attn.k_proj": true,
822
+ "model.layers.39.self_attn.v_proj": true,
823
+ "model.layers.39.self_attn.o_proj": true,
824
+ "model.layers.39.self_attn.rope": false,
825
+ "model.layers.39.mlp.gate_proj": true,
826
+ "model.layers.39.mlp.down_proj": true,
827
+ "model.layers.39.mlp.up_proj": true,
828
+ "model.layers.39.input_layernorm": false,
829
+ "model.layers.39.post_attention_layernorm": false,
830
+ "model.norm": false,
831
+ "lm_head": true
832
+ },
833
+ "rms_norm_eps": 1e-05,
834
+ "rope_theta": 1000000000.0,
835
+ "sliding_window": null,
836
+ "tie_word_embeddings": false,
837
+ "torch_dtype": "bfloat16",
838
+ "transformers_version": "4.53.1",
839
+ "use_cache": true,
840
+ "vocab_size": 131072
841
+ }