Wwayu commited on
Commit
c4be50d
·
verified ·
1 Parent(s): 7d1a3e6

Upload config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. config.json +1837 -0
config.json ADDED
@@ -0,0 +1,1837 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "GptOssForCausalLM"
4
+ ],
5
+ "attention_bias": true,
6
+ "attention_dropout": 0.0,
7
+ "eos_token_id": 200002,
8
+ "experts_per_token": 4,
9
+ "head_dim": 64,
10
+ "hidden_act": "silu",
11
+ "hidden_size": 2880,
12
+ "initial_context_length": 4096,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 2880,
15
+ "layer_types": [
16
+ "sliding_attention",
17
+ "full_attention",
18
+ "sliding_attention",
19
+ "full_attention",
20
+ "sliding_attention",
21
+ "full_attention",
22
+ "sliding_attention",
23
+ "full_attention",
24
+ "sliding_attention",
25
+ "full_attention",
26
+ "sliding_attention",
27
+ "full_attention",
28
+ "sliding_attention",
29
+ "full_attention",
30
+ "sliding_attention",
31
+ "full_attention",
32
+ "sliding_attention",
33
+ "full_attention",
34
+ "sliding_attention",
35
+ "full_attention",
36
+ "sliding_attention",
37
+ "full_attention",
38
+ "sliding_attention",
39
+ "full_attention",
40
+ "sliding_attention",
41
+ "full_attention",
42
+ "sliding_attention",
43
+ "full_attention",
44
+ "sliding_attention",
45
+ "full_attention",
46
+ "sliding_attention",
47
+ "full_attention",
48
+ "sliding_attention",
49
+ "full_attention",
50
+ "sliding_attention",
51
+ "full_attention"
52
+ ],
53
+ "max_position_embeddings": 131072,
54
+ "model_type": "gpt_oss",
55
+ "num_attention_heads": 64,
56
+ "num_experts_per_tok": 4,
57
+ "num_hidden_layers": 36,
58
+ "num_key_value_heads": 8,
59
+ "num_local_experts": 64,
60
+ "output_router_logits": false,
61
+ "pad_token_id": 199999,
62
+ "quantization": {
63
+ "group_size": 32,
64
+ "bits": 4,
65
+ "mode": "mxfp4",
66
+ "model.embed_tokens": {
67
+ "group_size": 64,
68
+ "bits": 8,
69
+ "mode": "affine"
70
+ },
71
+ "model.layers.0.self_attn.q_proj": {
72
+ "group_size": 64,
73
+ "bits": 8,
74
+ "mode": "affine"
75
+ },
76
+ "model.layers.0.self_attn.k_proj": {
77
+ "group_size": 64,
78
+ "bits": 8,
79
+ "mode": "affine"
80
+ },
81
+ "model.layers.0.self_attn.v_proj": {
82
+ "group_size": 64,
83
+ "bits": 8,
84
+ "mode": "affine"
85
+ },
86
+ "model.layers.0.self_attn.o_proj": {
87
+ "group_size": 64,
88
+ "bits": 8,
89
+ "mode": "affine"
90
+ },
91
+ "model.layers.0.mlp.router": {
92
+ "group_size": 64,
93
+ "bits": 8
94
+ },
95
+ "model.layers.1.self_attn.q_proj": {
96
+ "group_size": 64,
97
+ "bits": 8,
98
+ "mode": "affine"
99
+ },
100
+ "model.layers.1.self_attn.k_proj": {
101
+ "group_size": 64,
102
+ "bits": 8,
103
+ "mode": "affine"
104
+ },
105
+ "model.layers.1.self_attn.v_proj": {
106
+ "group_size": 64,
107
+ "bits": 8,
108
+ "mode": "affine"
109
+ },
110
+ "model.layers.1.self_attn.o_proj": {
111
+ "group_size": 64,
112
+ "bits": 8,
113
+ "mode": "affine"
114
+ },
115
+ "model.layers.1.mlp.router": {
116
+ "group_size": 64,
117
+ "bits": 8
118
+ },
119
+ "model.layers.2.self_attn.q_proj": {
120
+ "group_size": 64,
121
+ "bits": 8,
122
+ "mode": "affine"
123
+ },
124
+ "model.layers.2.self_attn.k_proj": {
125
+ "group_size": 64,
126
+ "bits": 8,
127
+ "mode": "affine"
128
+ },
129
+ "model.layers.2.self_attn.v_proj": {
130
+ "group_size": 64,
131
+ "bits": 8,
132
+ "mode": "affine"
133
+ },
134
+ "model.layers.2.self_attn.o_proj": {
135
+ "group_size": 64,
136
+ "bits": 8,
137
+ "mode": "affine"
138
+ },
139
+ "model.layers.2.mlp.router": {
140
+ "group_size": 64,
141
+ "bits": 8
142
+ },
143
+ "model.layers.3.self_attn.q_proj": {
144
+ "group_size": 64,
145
+ "bits": 8,
146
+ "mode": "affine"
147
+ },
148
+ "model.layers.3.self_attn.k_proj": {
149
+ "group_size": 64,
150
+ "bits": 8,
151
+ "mode": "affine"
152
+ },
153
+ "model.layers.3.self_attn.v_proj": {
154
+ "group_size": 64,
155
+ "bits": 8,
156
+ "mode": "affine"
157
+ },
158
+ "model.layers.3.self_attn.o_proj": {
159
+ "group_size": 64,
160
+ "bits": 8,
161
+ "mode": "affine"
162
+ },
163
+ "model.layers.3.mlp.router": {
164
+ "group_size": 64,
165
+ "bits": 8
166
+ },
167
+ "model.layers.4.self_attn.q_proj": {
168
+ "group_size": 64,
169
+ "bits": 8,
170
+ "mode": "affine"
171
+ },
172
+ "model.layers.4.self_attn.k_proj": {
173
+ "group_size": 64,
174
+ "bits": 8,
175
+ "mode": "affine"
176
+ },
177
+ "model.layers.4.self_attn.v_proj": {
178
+ "group_size": 64,
179
+ "bits": 8,
180
+ "mode": "affine"
181
+ },
182
+ "model.layers.4.self_attn.o_proj": {
183
+ "group_size": 64,
184
+ "bits": 8,
185
+ "mode": "affine"
186
+ },
187
+ "model.layers.4.mlp.router": {
188
+ "group_size": 64,
189
+ "bits": 8
190
+ },
191
+ "model.layers.5.self_attn.q_proj": {
192
+ "group_size": 64,
193
+ "bits": 8,
194
+ "mode": "affine"
195
+ },
196
+ "model.layers.5.self_attn.k_proj": {
197
+ "group_size": 64,
198
+ "bits": 8,
199
+ "mode": "affine"
200
+ },
201
+ "model.layers.5.self_attn.v_proj": {
202
+ "group_size": 64,
203
+ "bits": 8,
204
+ "mode": "affine"
205
+ },
206
+ "model.layers.5.self_attn.o_proj": {
207
+ "group_size": 64,
208
+ "bits": 8,
209
+ "mode": "affine"
210
+ },
211
+ "model.layers.5.mlp.router": {
212
+ "group_size": 64,
213
+ "bits": 8
214
+ },
215
+ "model.layers.6.self_attn.q_proj": {
216
+ "group_size": 64,
217
+ "bits": 8,
218
+ "mode": "affine"
219
+ },
220
+ "model.layers.6.self_attn.k_proj": {
221
+ "group_size": 64,
222
+ "bits": 8,
223
+ "mode": "affine"
224
+ },
225
+ "model.layers.6.self_attn.v_proj": {
226
+ "group_size": 64,
227
+ "bits": 8,
228
+ "mode": "affine"
229
+ },
230
+ "model.layers.6.self_attn.o_proj": {
231
+ "group_size": 64,
232
+ "bits": 8,
233
+ "mode": "affine"
234
+ },
235
+ "model.layers.6.mlp.router": {
236
+ "group_size": 64,
237
+ "bits": 8
238
+ },
239
+ "model.layers.7.self_attn.q_proj": {
240
+ "group_size": 64,
241
+ "bits": 8,
242
+ "mode": "affine"
243
+ },
244
+ "model.layers.7.self_attn.k_proj": {
245
+ "group_size": 64,
246
+ "bits": 8,
247
+ "mode": "affine"
248
+ },
249
+ "model.layers.7.self_attn.v_proj": {
250
+ "group_size": 64,
251
+ "bits": 8,
252
+ "mode": "affine"
253
+ },
254
+ "model.layers.7.self_attn.o_proj": {
255
+ "group_size": 64,
256
+ "bits": 8,
257
+ "mode": "affine"
258
+ },
259
+ "model.layers.7.mlp.router": {
260
+ "group_size": 64,
261
+ "bits": 8
262
+ },
263
+ "model.layers.8.self_attn.q_proj": {
264
+ "group_size": 64,
265
+ "bits": 8,
266
+ "mode": "affine"
267
+ },
268
+ "model.layers.8.self_attn.k_proj": {
269
+ "group_size": 64,
270
+ "bits": 8,
271
+ "mode": "affine"
272
+ },
273
+ "model.layers.8.self_attn.v_proj": {
274
+ "group_size": 64,
275
+ "bits": 8,
276
+ "mode": "affine"
277
+ },
278
+ "model.layers.8.self_attn.o_proj": {
279
+ "group_size": 64,
280
+ "bits": 8,
281
+ "mode": "affine"
282
+ },
283
+ "model.layers.8.mlp.router": {
284
+ "group_size": 64,
285
+ "bits": 8
286
+ },
287
+ "model.layers.9.self_attn.q_proj": {
288
+ "group_size": 64,
289
+ "bits": 8,
290
+ "mode": "affine"
291
+ },
292
+ "model.layers.9.self_attn.k_proj": {
293
+ "group_size": 64,
294
+ "bits": 8,
295
+ "mode": "affine"
296
+ },
297
+ "model.layers.9.self_attn.v_proj": {
298
+ "group_size": 64,
299
+ "bits": 8,
300
+ "mode": "affine"
301
+ },
302
+ "model.layers.9.self_attn.o_proj": {
303
+ "group_size": 64,
304
+ "bits": 8,
305
+ "mode": "affine"
306
+ },
307
+ "model.layers.9.mlp.router": {
308
+ "group_size": 64,
309
+ "bits": 8
310
+ },
311
+ "model.layers.10.self_attn.q_proj": {
312
+ "group_size": 64,
313
+ "bits": 8,
314
+ "mode": "affine"
315
+ },
316
+ "model.layers.10.self_attn.k_proj": {
317
+ "group_size": 64,
318
+ "bits": 8,
319
+ "mode": "affine"
320
+ },
321
+ "model.layers.10.self_attn.v_proj": {
322
+ "group_size": 64,
323
+ "bits": 8,
324
+ "mode": "affine"
325
+ },
326
+ "model.layers.10.self_attn.o_proj": {
327
+ "group_size": 64,
328
+ "bits": 8,
329
+ "mode": "affine"
330
+ },
331
+ "model.layers.10.mlp.router": {
332
+ "group_size": 64,
333
+ "bits": 8
334
+ },
335
+ "model.layers.11.self_attn.q_proj": {
336
+ "group_size": 64,
337
+ "bits": 8,
338
+ "mode": "affine"
339
+ },
340
+ "model.layers.11.self_attn.k_proj": {
341
+ "group_size": 64,
342
+ "bits": 8,
343
+ "mode": "affine"
344
+ },
345
+ "model.layers.11.self_attn.v_proj": {
346
+ "group_size": 64,
347
+ "bits": 8,
348
+ "mode": "affine"
349
+ },
350
+ "model.layers.11.self_attn.o_proj": {
351
+ "group_size": 64,
352
+ "bits": 8,
353
+ "mode": "affine"
354
+ },
355
+ "model.layers.11.mlp.router": {
356
+ "group_size": 64,
357
+ "bits": 8
358
+ },
359
+ "model.layers.12.self_attn.q_proj": {
360
+ "group_size": 64,
361
+ "bits": 8,
362
+ "mode": "affine"
363
+ },
364
+ "model.layers.12.self_attn.k_proj": {
365
+ "group_size": 64,
366
+ "bits": 8,
367
+ "mode": "affine"
368
+ },
369
+ "model.layers.12.self_attn.v_proj": {
370
+ "group_size": 64,
371
+ "bits": 8,
372
+ "mode": "affine"
373
+ },
374
+ "model.layers.12.self_attn.o_proj": {
375
+ "group_size": 64,
376
+ "bits": 8,
377
+ "mode": "affine"
378
+ },
379
+ "model.layers.12.mlp.router": {
380
+ "group_size": 64,
381
+ "bits": 8
382
+ },
383
+ "model.layers.13.self_attn.q_proj": {
384
+ "group_size": 64,
385
+ "bits": 8,
386
+ "mode": "affine"
387
+ },
388
+ "model.layers.13.self_attn.k_proj": {
389
+ "group_size": 64,
390
+ "bits": 8,
391
+ "mode": "affine"
392
+ },
393
+ "model.layers.13.self_attn.v_proj": {
394
+ "group_size": 64,
395
+ "bits": 8,
396
+ "mode": "affine"
397
+ },
398
+ "model.layers.13.self_attn.o_proj": {
399
+ "group_size": 64,
400
+ "bits": 8,
401
+ "mode": "affine"
402
+ },
403
+ "model.layers.13.mlp.router": {
404
+ "group_size": 64,
405
+ "bits": 8
406
+ },
407
+ "model.layers.14.self_attn.q_proj": {
408
+ "group_size": 64,
409
+ "bits": 8,
410
+ "mode": "affine"
411
+ },
412
+ "model.layers.14.self_attn.k_proj": {
413
+ "group_size": 64,
414
+ "bits": 8,
415
+ "mode": "affine"
416
+ },
417
+ "model.layers.14.self_attn.v_proj": {
418
+ "group_size": 64,
419
+ "bits": 8,
420
+ "mode": "affine"
421
+ },
422
+ "model.layers.14.self_attn.o_proj": {
423
+ "group_size": 64,
424
+ "bits": 8,
425
+ "mode": "affine"
426
+ },
427
+ "model.layers.14.mlp.router": {
428
+ "group_size": 64,
429
+ "bits": 8
430
+ },
431
+ "model.layers.15.self_attn.q_proj": {
432
+ "group_size": 64,
433
+ "bits": 8,
434
+ "mode": "affine"
435
+ },
436
+ "model.layers.15.self_attn.k_proj": {
437
+ "group_size": 64,
438
+ "bits": 8,
439
+ "mode": "affine"
440
+ },
441
+ "model.layers.15.self_attn.v_proj": {
442
+ "group_size": 64,
443
+ "bits": 8,
444
+ "mode": "affine"
445
+ },
446
+ "model.layers.15.self_attn.o_proj": {
447
+ "group_size": 64,
448
+ "bits": 8,
449
+ "mode": "affine"
450
+ },
451
+ "model.layers.15.mlp.router": {
452
+ "group_size": 64,
453
+ "bits": 8
454
+ },
455
+ "model.layers.16.self_attn.q_proj": {
456
+ "group_size": 64,
457
+ "bits": 8,
458
+ "mode": "affine"
459
+ },
460
+ "model.layers.16.self_attn.k_proj": {
461
+ "group_size": 64,
462
+ "bits": 8,
463
+ "mode": "affine"
464
+ },
465
+ "model.layers.16.self_attn.v_proj": {
466
+ "group_size": 64,
467
+ "bits": 8,
468
+ "mode": "affine"
469
+ },
470
+ "model.layers.16.self_attn.o_proj": {
471
+ "group_size": 64,
472
+ "bits": 8,
473
+ "mode": "affine"
474
+ },
475
+ "model.layers.16.mlp.router": {
476
+ "group_size": 64,
477
+ "bits": 8
478
+ },
479
+ "model.layers.17.self_attn.q_proj": {
480
+ "group_size": 64,
481
+ "bits": 8,
482
+ "mode": "affine"
483
+ },
484
+ "model.layers.17.self_attn.k_proj": {
485
+ "group_size": 64,
486
+ "bits": 8,
487
+ "mode": "affine"
488
+ },
489
+ "model.layers.17.self_attn.v_proj": {
490
+ "group_size": 64,
491
+ "bits": 8,
492
+ "mode": "affine"
493
+ },
494
+ "model.layers.17.self_attn.o_proj": {
495
+ "group_size": 64,
496
+ "bits": 8,
497
+ "mode": "affine"
498
+ },
499
+ "model.layers.17.mlp.router": {
500
+ "group_size": 64,
501
+ "bits": 8
502
+ },
503
+ "model.layers.18.self_attn.q_proj": {
504
+ "group_size": 64,
505
+ "bits": 8,
506
+ "mode": "affine"
507
+ },
508
+ "model.layers.18.self_attn.k_proj": {
509
+ "group_size": 64,
510
+ "bits": 8,
511
+ "mode": "affine"
512
+ },
513
+ "model.layers.18.self_attn.v_proj": {
514
+ "group_size": 64,
515
+ "bits": 8,
516
+ "mode": "affine"
517
+ },
518
+ "model.layers.18.self_attn.o_proj": {
519
+ "group_size": 64,
520
+ "bits": 8,
521
+ "mode": "affine"
522
+ },
523
+ "model.layers.18.mlp.router": {
524
+ "group_size": 64,
525
+ "bits": 8
526
+ },
527
+ "model.layers.19.self_attn.q_proj": {
528
+ "group_size": 64,
529
+ "bits": 8,
530
+ "mode": "affine"
531
+ },
532
+ "model.layers.19.self_attn.k_proj": {
533
+ "group_size": 64,
534
+ "bits": 8,
535
+ "mode": "affine"
536
+ },
537
+ "model.layers.19.self_attn.v_proj": {
538
+ "group_size": 64,
539
+ "bits": 8,
540
+ "mode": "affine"
541
+ },
542
+ "model.layers.19.self_attn.o_proj": {
543
+ "group_size": 64,
544
+ "bits": 8,
545
+ "mode": "affine"
546
+ },
547
+ "model.layers.19.mlp.router": {
548
+ "group_size": 64,
549
+ "bits": 8
550
+ },
551
+ "model.layers.20.self_attn.q_proj": {
552
+ "group_size": 64,
553
+ "bits": 8,
554
+ "mode": "affine"
555
+ },
556
+ "model.layers.20.self_attn.k_proj": {
557
+ "group_size": 64,
558
+ "bits": 8,
559
+ "mode": "affine"
560
+ },
561
+ "model.layers.20.self_attn.v_proj": {
562
+ "group_size": 64,
563
+ "bits": 8,
564
+ "mode": "affine"
565
+ },
566
+ "model.layers.20.self_attn.o_proj": {
567
+ "group_size": 64,
568
+ "bits": 8,
569
+ "mode": "affine"
570
+ },
571
+ "model.layers.20.mlp.router": {
572
+ "group_size": 64,
573
+ "bits": 8
574
+ },
575
+ "model.layers.21.self_attn.q_proj": {
576
+ "group_size": 64,
577
+ "bits": 8,
578
+ "mode": "affine"
579
+ },
580
+ "model.layers.21.self_attn.k_proj": {
581
+ "group_size": 64,
582
+ "bits": 8,
583
+ "mode": "affine"
584
+ },
585
+ "model.layers.21.self_attn.v_proj": {
586
+ "group_size": 64,
587
+ "bits": 8,
588
+ "mode": "affine"
589
+ },
590
+ "model.layers.21.self_attn.o_proj": {
591
+ "group_size": 64,
592
+ "bits": 8,
593
+ "mode": "affine"
594
+ },
595
+ "model.layers.21.mlp.router": {
596
+ "group_size": 64,
597
+ "bits": 8
598
+ },
599
+ "model.layers.22.self_attn.q_proj": {
600
+ "group_size": 64,
601
+ "bits": 8,
602
+ "mode": "affine"
603
+ },
604
+ "model.layers.22.self_attn.k_proj": {
605
+ "group_size": 64,
606
+ "bits": 8,
607
+ "mode": "affine"
608
+ },
609
+ "model.layers.22.self_attn.v_proj": {
610
+ "group_size": 64,
611
+ "bits": 8,
612
+ "mode": "affine"
613
+ },
614
+ "model.layers.22.self_attn.o_proj": {
615
+ "group_size": 64,
616
+ "bits": 8,
617
+ "mode": "affine"
618
+ },
619
+ "model.layers.22.mlp.router": {
620
+ "group_size": 64,
621
+ "bits": 8
622
+ },
623
+ "model.layers.23.self_attn.q_proj": {
624
+ "group_size": 64,
625
+ "bits": 8,
626
+ "mode": "affine"
627
+ },
628
+ "model.layers.23.self_attn.k_proj": {
629
+ "group_size": 64,
630
+ "bits": 8,
631
+ "mode": "affine"
632
+ },
633
+ "model.layers.23.self_attn.v_proj": {
634
+ "group_size": 64,
635
+ "bits": 8,
636
+ "mode": "affine"
637
+ },
638
+ "model.layers.23.self_attn.o_proj": {
639
+ "group_size": 64,
640
+ "bits": 8,
641
+ "mode": "affine"
642
+ },
643
+ "model.layers.23.mlp.router": {
644
+ "group_size": 64,
645
+ "bits": 8
646
+ },
647
+ "model.layers.24.self_attn.q_proj": {
648
+ "group_size": 64,
649
+ "bits": 8,
650
+ "mode": "affine"
651
+ },
652
+ "model.layers.24.self_attn.k_proj": {
653
+ "group_size": 64,
654
+ "bits": 8,
655
+ "mode": "affine"
656
+ },
657
+ "model.layers.24.self_attn.v_proj": {
658
+ "group_size": 64,
659
+ "bits": 8,
660
+ "mode": "affine"
661
+ },
662
+ "model.layers.24.self_attn.o_proj": {
663
+ "group_size": 64,
664
+ "bits": 8,
665
+ "mode": "affine"
666
+ },
667
+ "model.layers.24.mlp.router": {
668
+ "group_size": 64,
669
+ "bits": 8
670
+ },
671
+ "model.layers.25.self_attn.q_proj": {
672
+ "group_size": 64,
673
+ "bits": 8,
674
+ "mode": "affine"
675
+ },
676
+ "model.layers.25.self_attn.k_proj": {
677
+ "group_size": 64,
678
+ "bits": 8,
679
+ "mode": "affine"
680
+ },
681
+ "model.layers.25.self_attn.v_proj": {
682
+ "group_size": 64,
683
+ "bits": 8,
684
+ "mode": "affine"
685
+ },
686
+ "model.layers.25.self_attn.o_proj": {
687
+ "group_size": 64,
688
+ "bits": 8,
689
+ "mode": "affine"
690
+ },
691
+ "model.layers.25.mlp.router": {
692
+ "group_size": 64,
693
+ "bits": 8
694
+ },
695
+ "model.layers.26.self_attn.q_proj": {
696
+ "group_size": 64,
697
+ "bits": 8,
698
+ "mode": "affine"
699
+ },
700
+ "model.layers.26.self_attn.k_proj": {
701
+ "group_size": 64,
702
+ "bits": 8,
703
+ "mode": "affine"
704
+ },
705
+ "model.layers.26.self_attn.v_proj": {
706
+ "group_size": 64,
707
+ "bits": 8,
708
+ "mode": "affine"
709
+ },
710
+ "model.layers.26.self_attn.o_proj": {
711
+ "group_size": 64,
712
+ "bits": 8,
713
+ "mode": "affine"
714
+ },
715
+ "model.layers.26.mlp.router": {
716
+ "group_size": 64,
717
+ "bits": 8
718
+ },
719
+ "model.layers.27.self_attn.q_proj": {
720
+ "group_size": 64,
721
+ "bits": 8,
722
+ "mode": "affine"
723
+ },
724
+ "model.layers.27.self_attn.k_proj": {
725
+ "group_size": 64,
726
+ "bits": 8,
727
+ "mode": "affine"
728
+ },
729
+ "model.layers.27.self_attn.v_proj": {
730
+ "group_size": 64,
731
+ "bits": 8,
732
+ "mode": "affine"
733
+ },
734
+ "model.layers.27.self_attn.o_proj": {
735
+ "group_size": 64,
736
+ "bits": 8,
737
+ "mode": "affine"
738
+ },
739
+ "model.layers.27.mlp.router": {
740
+ "group_size": 64,
741
+ "bits": 8
742
+ },
743
+ "model.layers.28.self_attn.q_proj": {
744
+ "group_size": 64,
745
+ "bits": 8,
746
+ "mode": "affine"
747
+ },
748
+ "model.layers.28.self_attn.k_proj": {
749
+ "group_size": 64,
750
+ "bits": 8,
751
+ "mode": "affine"
752
+ },
753
+ "model.layers.28.self_attn.v_proj": {
754
+ "group_size": 64,
755
+ "bits": 8,
756
+ "mode": "affine"
757
+ },
758
+ "model.layers.28.self_attn.o_proj": {
759
+ "group_size": 64,
760
+ "bits": 8,
761
+ "mode": "affine"
762
+ },
763
+ "model.layers.28.mlp.router": {
764
+ "group_size": 64,
765
+ "bits": 8
766
+ },
767
+ "model.layers.29.self_attn.q_proj": {
768
+ "group_size": 64,
769
+ "bits": 8,
770
+ "mode": "affine"
771
+ },
772
+ "model.layers.29.self_attn.k_proj": {
773
+ "group_size": 64,
774
+ "bits": 8,
775
+ "mode": "affine"
776
+ },
777
+ "model.layers.29.self_attn.v_proj": {
778
+ "group_size": 64,
779
+ "bits": 8,
780
+ "mode": "affine"
781
+ },
782
+ "model.layers.29.self_attn.o_proj": {
783
+ "group_size": 64,
784
+ "bits": 8,
785
+ "mode": "affine"
786
+ },
787
+ "model.layers.29.mlp.router": {
788
+ "group_size": 64,
789
+ "bits": 8
790
+ },
791
+ "model.layers.30.self_attn.q_proj": {
792
+ "group_size": 64,
793
+ "bits": 8,
794
+ "mode": "affine"
795
+ },
796
+ "model.layers.30.self_attn.k_proj": {
797
+ "group_size": 64,
798
+ "bits": 8,
799
+ "mode": "affine"
800
+ },
801
+ "model.layers.30.self_attn.v_proj": {
802
+ "group_size": 64,
803
+ "bits": 8,
804
+ "mode": "affine"
805
+ },
806
+ "model.layers.30.self_attn.o_proj": {
807
+ "group_size": 64,
808
+ "bits": 8,
809
+ "mode": "affine"
810
+ },
811
+ "model.layers.30.mlp.router": {
812
+ "group_size": 64,
813
+ "bits": 8
814
+ },
815
+ "model.layers.31.self_attn.q_proj": {
816
+ "group_size": 64,
817
+ "bits": 8,
818
+ "mode": "affine"
819
+ },
820
+ "model.layers.31.self_attn.k_proj": {
821
+ "group_size": 64,
822
+ "bits": 8,
823
+ "mode": "affine"
824
+ },
825
+ "model.layers.31.self_attn.v_proj": {
826
+ "group_size": 64,
827
+ "bits": 8,
828
+ "mode": "affine"
829
+ },
830
+ "model.layers.31.self_attn.o_proj": {
831
+ "group_size": 64,
832
+ "bits": 8,
833
+ "mode": "affine"
834
+ },
835
+ "model.layers.31.mlp.router": {
836
+ "group_size": 64,
837
+ "bits": 8
838
+ },
839
+ "model.layers.32.self_attn.q_proj": {
840
+ "group_size": 64,
841
+ "bits": 8,
842
+ "mode": "affine"
843
+ },
844
+ "model.layers.32.self_attn.k_proj": {
845
+ "group_size": 64,
846
+ "bits": 8,
847
+ "mode": "affine"
848
+ },
849
+ "model.layers.32.self_attn.v_proj": {
850
+ "group_size": 64,
851
+ "bits": 8,
852
+ "mode": "affine"
853
+ },
854
+ "model.layers.32.self_attn.o_proj": {
855
+ "group_size": 64,
856
+ "bits": 8,
857
+ "mode": "affine"
858
+ },
859
+ "model.layers.32.mlp.router": {
860
+ "group_size": 64,
861
+ "bits": 8
862
+ },
863
+ "model.layers.33.self_attn.q_proj": {
864
+ "group_size": 64,
865
+ "bits": 8,
866
+ "mode": "affine"
867
+ },
868
+ "model.layers.33.self_attn.k_proj": {
869
+ "group_size": 64,
870
+ "bits": 8,
871
+ "mode": "affine"
872
+ },
873
+ "model.layers.33.self_attn.v_proj": {
874
+ "group_size": 64,
875
+ "bits": 8,
876
+ "mode": "affine"
877
+ },
878
+ "model.layers.33.self_attn.o_proj": {
879
+ "group_size": 64,
880
+ "bits": 8,
881
+ "mode": "affine"
882
+ },
883
+ "model.layers.33.mlp.router": {
884
+ "group_size": 64,
885
+ "bits": 8
886
+ },
887
+ "model.layers.34.self_attn.q_proj": {
888
+ "group_size": 64,
889
+ "bits": 8,
890
+ "mode": "affine"
891
+ },
892
+ "model.layers.34.self_attn.k_proj": {
893
+ "group_size": 64,
894
+ "bits": 8,
895
+ "mode": "affine"
896
+ },
897
+ "model.layers.34.self_attn.v_proj": {
898
+ "group_size": 64,
899
+ "bits": 8,
900
+ "mode": "affine"
901
+ },
902
+ "model.layers.34.self_attn.o_proj": {
903
+ "group_size": 64,
904
+ "bits": 8,
905
+ "mode": "affine"
906
+ },
907
+ "model.layers.34.mlp.router": {
908
+ "group_size": 64,
909
+ "bits": 8
910
+ },
911
+ "model.layers.35.self_attn.q_proj": {
912
+ "group_size": 64,
913
+ "bits": 8,
914
+ "mode": "affine"
915
+ },
916
+ "model.layers.35.self_attn.k_proj": {
917
+ "group_size": 64,
918
+ "bits": 8,
919
+ "mode": "affine"
920
+ },
921
+ "model.layers.35.self_attn.v_proj": {
922
+ "group_size": 64,
923
+ "bits": 8,
924
+ "mode": "affine"
925
+ },
926
+ "model.layers.35.self_attn.o_proj": {
927
+ "group_size": 64,
928
+ "bits": 8,
929
+ "mode": "affine"
930
+ },
931
+ "model.layers.35.mlp.router": {
932
+ "group_size": 64,
933
+ "bits": 8
934
+ },
935
+ "lm_head": {
936
+ "group_size": 64,
937
+ "bits": 8,
938
+ "mode": "affine"
939
+ }
940
+ },
941
+ "quantization_config": {
942
+ "group_size": 32,
943
+ "bits": 4,
944
+ "mode": "mxfp4",
945
+ "model.embed_tokens": {
946
+ "group_size": 64,
947
+ "bits": 8,
948
+ "mode": "affine"
949
+ },
950
+ "model.layers.0.self_attn.q_proj": {
951
+ "group_size": 64,
952
+ "bits": 8,
953
+ "mode": "affine"
954
+ },
955
+ "model.layers.0.self_attn.k_proj": {
956
+ "group_size": 64,
957
+ "bits": 8,
958
+ "mode": "affine"
959
+ },
960
+ "model.layers.0.self_attn.v_proj": {
961
+ "group_size": 64,
962
+ "bits": 8,
963
+ "mode": "affine"
964
+ },
965
+ "model.layers.0.self_attn.o_proj": {
966
+ "group_size": 64,
967
+ "bits": 8,
968
+ "mode": "affine"
969
+ },
970
+ "model.layers.0.mlp.router": {
971
+ "group_size": 64,
972
+ "bits": 8
973
+ },
974
+ "model.layers.1.self_attn.q_proj": {
975
+ "group_size": 64,
976
+ "bits": 8,
977
+ "mode": "affine"
978
+ },
979
+ "model.layers.1.self_attn.k_proj": {
980
+ "group_size": 64,
981
+ "bits": 8,
982
+ "mode": "affine"
983
+ },
984
+ "model.layers.1.self_attn.v_proj": {
985
+ "group_size": 64,
986
+ "bits": 8,
987
+ "mode": "affine"
988
+ },
989
+ "model.layers.1.self_attn.o_proj": {
990
+ "group_size": 64,
991
+ "bits": 8,
992
+ "mode": "affine"
993
+ },
994
+ "model.layers.1.mlp.router": {
995
+ "group_size": 64,
996
+ "bits": 8
997
+ },
998
+ "model.layers.2.self_attn.q_proj": {
999
+ "group_size": 64,
1000
+ "bits": 8,
1001
+ "mode": "affine"
1002
+ },
1003
+ "model.layers.2.self_attn.k_proj": {
1004
+ "group_size": 64,
1005
+ "bits": 8,
1006
+ "mode": "affine"
1007
+ },
1008
+ "model.layers.2.self_attn.v_proj": {
1009
+ "group_size": 64,
1010
+ "bits": 8,
1011
+ "mode": "affine"
1012
+ },
1013
+ "model.layers.2.self_attn.o_proj": {
1014
+ "group_size": 64,
1015
+ "bits": 8,
1016
+ "mode": "affine"
1017
+ },
1018
+ "model.layers.2.mlp.router": {
1019
+ "group_size": 64,
1020
+ "bits": 8
1021
+ },
1022
+ "model.layers.3.self_attn.q_proj": {
1023
+ "group_size": 64,
1024
+ "bits": 8,
1025
+ "mode": "affine"
1026
+ },
1027
+ "model.layers.3.self_attn.k_proj": {
1028
+ "group_size": 64,
1029
+ "bits": 8,
1030
+ "mode": "affine"
1031
+ },
1032
+ "model.layers.3.self_attn.v_proj": {
1033
+ "group_size": 64,
1034
+ "bits": 8,
1035
+ "mode": "affine"
1036
+ },
1037
+ "model.layers.3.self_attn.o_proj": {
1038
+ "group_size": 64,
1039
+ "bits": 8,
1040
+ "mode": "affine"
1041
+ },
1042
+ "model.layers.3.mlp.router": {
1043
+ "group_size": 64,
1044
+ "bits": 8
1045
+ },
1046
+ "model.layers.4.self_attn.q_proj": {
1047
+ "group_size": 64,
1048
+ "bits": 8,
1049
+ "mode": "affine"
1050
+ },
1051
+ "model.layers.4.self_attn.k_proj": {
1052
+ "group_size": 64,
1053
+ "bits": 8,
1054
+ "mode": "affine"
1055
+ },
1056
+ "model.layers.4.self_attn.v_proj": {
1057
+ "group_size": 64,
1058
+ "bits": 8,
1059
+ "mode": "affine"
1060
+ },
1061
+ "model.layers.4.self_attn.o_proj": {
1062
+ "group_size": 64,
1063
+ "bits": 8,
1064
+ "mode": "affine"
1065
+ },
1066
+ "model.layers.4.mlp.router": {
1067
+ "group_size": 64,
1068
+ "bits": 8
1069
+ },
1070
+ "model.layers.5.self_attn.q_proj": {
1071
+ "group_size": 64,
1072
+ "bits": 8,
1073
+ "mode": "affine"
1074
+ },
1075
+ "model.layers.5.self_attn.k_proj": {
1076
+ "group_size": 64,
1077
+ "bits": 8,
1078
+ "mode": "affine"
1079
+ },
1080
+ "model.layers.5.self_attn.v_proj": {
1081
+ "group_size": 64,
1082
+ "bits": 8,
1083
+ "mode": "affine"
1084
+ },
1085
+ "model.layers.5.self_attn.o_proj": {
1086
+ "group_size": 64,
1087
+ "bits": 8,
1088
+ "mode": "affine"
1089
+ },
1090
+ "model.layers.5.mlp.router": {
1091
+ "group_size": 64,
1092
+ "bits": 8
1093
+ },
1094
+ "model.layers.6.self_attn.q_proj": {
1095
+ "group_size": 64,
1096
+ "bits": 8,
1097
+ "mode": "affine"
1098
+ },
1099
+ "model.layers.6.self_attn.k_proj": {
1100
+ "group_size": 64,
1101
+ "bits": 8,
1102
+ "mode": "affine"
1103
+ },
1104
+ "model.layers.6.self_attn.v_proj": {
1105
+ "group_size": 64,
1106
+ "bits": 8,
1107
+ "mode": "affine"
1108
+ },
1109
+ "model.layers.6.self_attn.o_proj": {
1110
+ "group_size": 64,
1111
+ "bits": 8,
1112
+ "mode": "affine"
1113
+ },
1114
+ "model.layers.6.mlp.router": {
1115
+ "group_size": 64,
1116
+ "bits": 8
1117
+ },
1118
+ "model.layers.7.self_attn.q_proj": {
1119
+ "group_size": 64,
1120
+ "bits": 8,
1121
+ "mode": "affine"
1122
+ },
1123
+ "model.layers.7.self_attn.k_proj": {
1124
+ "group_size": 64,
1125
+ "bits": 8,
1126
+ "mode": "affine"
1127
+ },
1128
+ "model.layers.7.self_attn.v_proj": {
1129
+ "group_size": 64,
1130
+ "bits": 8,
1131
+ "mode": "affine"
1132
+ },
1133
+ "model.layers.7.self_attn.o_proj": {
1134
+ "group_size": 64,
1135
+ "bits": 8,
1136
+ "mode": "affine"
1137
+ },
1138
+ "model.layers.7.mlp.router": {
1139
+ "group_size": 64,
1140
+ "bits": 8
1141
+ },
1142
+ "model.layers.8.self_attn.q_proj": {
1143
+ "group_size": 64,
1144
+ "bits": 8,
1145
+ "mode": "affine"
1146
+ },
1147
+ "model.layers.8.self_attn.k_proj": {
1148
+ "group_size": 64,
1149
+ "bits": 8,
1150
+ "mode": "affine"
1151
+ },
1152
+ "model.layers.8.self_attn.v_proj": {
1153
+ "group_size": 64,
1154
+ "bits": 8,
1155
+ "mode": "affine"
1156
+ },
1157
+ "model.layers.8.self_attn.o_proj": {
1158
+ "group_size": 64,
1159
+ "bits": 8,
1160
+ "mode": "affine"
1161
+ },
1162
+ "model.layers.8.mlp.router": {
1163
+ "group_size": 64,
1164
+ "bits": 8
1165
+ },
1166
+ "model.layers.9.self_attn.q_proj": {
1167
+ "group_size": 64,
1168
+ "bits": 8,
1169
+ "mode": "affine"
1170
+ },
1171
+ "model.layers.9.self_attn.k_proj": {
1172
+ "group_size": 64,
1173
+ "bits": 8,
1174
+ "mode": "affine"
1175
+ },
1176
+ "model.layers.9.self_attn.v_proj": {
1177
+ "group_size": 64,
1178
+ "bits": 8,
1179
+ "mode": "affine"
1180
+ },
1181
+ "model.layers.9.self_attn.o_proj": {
1182
+ "group_size": 64,
1183
+ "bits": 8,
1184
+ "mode": "affine"
1185
+ },
1186
+ "model.layers.9.mlp.router": {
1187
+ "group_size": 64,
1188
+ "bits": 8
1189
+ },
1190
+ "model.layers.10.self_attn.q_proj": {
1191
+ "group_size": 64,
1192
+ "bits": 8,
1193
+ "mode": "affine"
1194
+ },
1195
+ "model.layers.10.self_attn.k_proj": {
1196
+ "group_size": 64,
1197
+ "bits": 8,
1198
+ "mode": "affine"
1199
+ },
1200
+ "model.layers.10.self_attn.v_proj": {
1201
+ "group_size": 64,
1202
+ "bits": 8,
1203
+ "mode": "affine"
1204
+ },
1205
+ "model.layers.10.self_attn.o_proj": {
1206
+ "group_size": 64,
1207
+ "bits": 8,
1208
+ "mode": "affine"
1209
+ },
1210
+ "model.layers.10.mlp.router": {
1211
+ "group_size": 64,
1212
+ "bits": 8
1213
+ },
1214
+ "model.layers.11.self_attn.q_proj": {
1215
+ "group_size": 64,
1216
+ "bits": 8,
1217
+ "mode": "affine"
1218
+ },
1219
+ "model.layers.11.self_attn.k_proj": {
1220
+ "group_size": 64,
1221
+ "bits": 8,
1222
+ "mode": "affine"
1223
+ },
1224
+ "model.layers.11.self_attn.v_proj": {
1225
+ "group_size": 64,
1226
+ "bits": 8,
1227
+ "mode": "affine"
1228
+ },
1229
+ "model.layers.11.self_attn.o_proj": {
1230
+ "group_size": 64,
1231
+ "bits": 8,
1232
+ "mode": "affine"
1233
+ },
1234
+ "model.layers.11.mlp.router": {
1235
+ "group_size": 64,
1236
+ "bits": 8
1237
+ },
1238
+ "model.layers.12.self_attn.q_proj": {
1239
+ "group_size": 64,
1240
+ "bits": 8,
1241
+ "mode": "affine"
1242
+ },
1243
+ "model.layers.12.self_attn.k_proj": {
1244
+ "group_size": 64,
1245
+ "bits": 8,
1246
+ "mode": "affine"
1247
+ },
1248
+ "model.layers.12.self_attn.v_proj": {
1249
+ "group_size": 64,
1250
+ "bits": 8,
1251
+ "mode": "affine"
1252
+ },
1253
+ "model.layers.12.self_attn.o_proj": {
1254
+ "group_size": 64,
1255
+ "bits": 8,
1256
+ "mode": "affine"
1257
+ },
1258
+ "model.layers.12.mlp.router": {
1259
+ "group_size": 64,
1260
+ "bits": 8
1261
+ },
1262
+ "model.layers.13.self_attn.q_proj": {
1263
+ "group_size": 64,
1264
+ "bits": 8,
1265
+ "mode": "affine"
1266
+ },
1267
+ "model.layers.13.self_attn.k_proj": {
1268
+ "group_size": 64,
1269
+ "bits": 8,
1270
+ "mode": "affine"
1271
+ },
1272
+ "model.layers.13.self_attn.v_proj": {
1273
+ "group_size": 64,
1274
+ "bits": 8,
1275
+ "mode": "affine"
1276
+ },
1277
+ "model.layers.13.self_attn.o_proj": {
1278
+ "group_size": 64,
1279
+ "bits": 8,
1280
+ "mode": "affine"
1281
+ },
1282
+ "model.layers.13.mlp.router": {
1283
+ "group_size": 64,
1284
+ "bits": 8
1285
+ },
1286
+ "model.layers.14.self_attn.q_proj": {
1287
+ "group_size": 64,
1288
+ "bits": 8,
1289
+ "mode": "affine"
1290
+ },
1291
+ "model.layers.14.self_attn.k_proj": {
1292
+ "group_size": 64,
1293
+ "bits": 8,
1294
+ "mode": "affine"
1295
+ },
1296
+ "model.layers.14.self_attn.v_proj": {
1297
+ "group_size": 64,
1298
+ "bits": 8,
1299
+ "mode": "affine"
1300
+ },
1301
+ "model.layers.14.self_attn.o_proj": {
1302
+ "group_size": 64,
1303
+ "bits": 8,
1304
+ "mode": "affine"
1305
+ },
1306
+ "model.layers.14.mlp.router": {
1307
+ "group_size": 64,
1308
+ "bits": 8
1309
+ },
1310
+ "model.layers.15.self_attn.q_proj": {
1311
+ "group_size": 64,
1312
+ "bits": 8,
1313
+ "mode": "affine"
1314
+ },
1315
+ "model.layers.15.self_attn.k_proj": {
1316
+ "group_size": 64,
1317
+ "bits": 8,
1318
+ "mode": "affine"
1319
+ },
1320
+ "model.layers.15.self_attn.v_proj": {
1321
+ "group_size": 64,
1322
+ "bits": 8,
1323
+ "mode": "affine"
1324
+ },
1325
+ "model.layers.15.self_attn.o_proj": {
1326
+ "group_size": 64,
1327
+ "bits": 8,
1328
+ "mode": "affine"
1329
+ },
1330
+ "model.layers.15.mlp.router": {
1331
+ "group_size": 64,
1332
+ "bits": 8
1333
+ },
1334
+ "model.layers.16.self_attn.q_proj": {
1335
+ "group_size": 64,
1336
+ "bits": 8,
1337
+ "mode": "affine"
1338
+ },
1339
+ "model.layers.16.self_attn.k_proj": {
1340
+ "group_size": 64,
1341
+ "bits": 8,
1342
+ "mode": "affine"
1343
+ },
1344
+ "model.layers.16.self_attn.v_proj": {
1345
+ "group_size": 64,
1346
+ "bits": 8,
1347
+ "mode": "affine"
1348
+ },
1349
+ "model.layers.16.self_attn.o_proj": {
1350
+ "group_size": 64,
1351
+ "bits": 8,
1352
+ "mode": "affine"
1353
+ },
1354
+ "model.layers.16.mlp.router": {
1355
+ "group_size": 64,
1356
+ "bits": 8
1357
+ },
1358
+ "model.layers.17.self_attn.q_proj": {
1359
+ "group_size": 64,
1360
+ "bits": 8,
1361
+ "mode": "affine"
1362
+ },
1363
+ "model.layers.17.self_attn.k_proj": {
1364
+ "group_size": 64,
1365
+ "bits": 8,
1366
+ "mode": "affine"
1367
+ },
1368
+ "model.layers.17.self_attn.v_proj": {
1369
+ "group_size": 64,
1370
+ "bits": 8,
1371
+ "mode": "affine"
1372
+ },
1373
+ "model.layers.17.self_attn.o_proj": {
1374
+ "group_size": 64,
1375
+ "bits": 8,
1376
+ "mode": "affine"
1377
+ },
1378
+ "model.layers.17.mlp.router": {
1379
+ "group_size": 64,
1380
+ "bits": 8
1381
+ },
1382
+ "model.layers.18.self_attn.q_proj": {
1383
+ "group_size": 64,
1384
+ "bits": 8,
1385
+ "mode": "affine"
1386
+ },
1387
+ "model.layers.18.self_attn.k_proj": {
1388
+ "group_size": 64,
1389
+ "bits": 8,
1390
+ "mode": "affine"
1391
+ },
1392
+ "model.layers.18.self_attn.v_proj": {
1393
+ "group_size": 64,
1394
+ "bits": 8,
1395
+ "mode": "affine"
1396
+ },
1397
+ "model.layers.18.self_attn.o_proj": {
1398
+ "group_size": 64,
1399
+ "bits": 8,
1400
+ "mode": "affine"
1401
+ },
1402
+ "model.layers.18.mlp.router": {
1403
+ "group_size": 64,
1404
+ "bits": 8
1405
+ },
1406
+ "model.layers.19.self_attn.q_proj": {
1407
+ "group_size": 64,
1408
+ "bits": 8,
1409
+ "mode": "affine"
1410
+ },
1411
+ "model.layers.19.self_attn.k_proj": {
1412
+ "group_size": 64,
1413
+ "bits": 8,
1414
+ "mode": "affine"
1415
+ },
1416
+ "model.layers.19.self_attn.v_proj": {
1417
+ "group_size": 64,
1418
+ "bits": 8,
1419
+ "mode": "affine"
1420
+ },
1421
+ "model.layers.19.self_attn.o_proj": {
1422
+ "group_size": 64,
1423
+ "bits": 8,
1424
+ "mode": "affine"
1425
+ },
1426
+ "model.layers.19.mlp.router": {
1427
+ "group_size": 64,
1428
+ "bits": 8
1429
+ },
1430
+ "model.layers.20.self_attn.q_proj": {
1431
+ "group_size": 64,
1432
+ "bits": 8,
1433
+ "mode": "affine"
1434
+ },
1435
+ "model.layers.20.self_attn.k_proj": {
1436
+ "group_size": 64,
1437
+ "bits": 8,
1438
+ "mode": "affine"
1439
+ },
1440
+ "model.layers.20.self_attn.v_proj": {
1441
+ "group_size": 64,
1442
+ "bits": 8,
1443
+ "mode": "affine"
1444
+ },
1445
+ "model.layers.20.self_attn.o_proj": {
1446
+ "group_size": 64,
1447
+ "bits": 8,
1448
+ "mode": "affine"
1449
+ },
1450
+ "model.layers.20.mlp.router": {
1451
+ "group_size": 64,
1452
+ "bits": 8
1453
+ },
1454
+ "model.layers.21.self_attn.q_proj": {
1455
+ "group_size": 64,
1456
+ "bits": 8,
1457
+ "mode": "affine"
1458
+ },
1459
+ "model.layers.21.self_attn.k_proj": {
1460
+ "group_size": 64,
1461
+ "bits": 8,
1462
+ "mode": "affine"
1463
+ },
1464
+ "model.layers.21.self_attn.v_proj": {
1465
+ "group_size": 64,
1466
+ "bits": 8,
1467
+ "mode": "affine"
1468
+ },
1469
+ "model.layers.21.self_attn.o_proj": {
1470
+ "group_size": 64,
1471
+ "bits": 8,
1472
+ "mode": "affine"
1473
+ },
1474
+ "model.layers.21.mlp.router": {
1475
+ "group_size": 64,
1476
+ "bits": 8
1477
+ },
1478
+ "model.layers.22.self_attn.q_proj": {
1479
+ "group_size": 64,
1480
+ "bits": 8,
1481
+ "mode": "affine"
1482
+ },
1483
+ "model.layers.22.self_attn.k_proj": {
1484
+ "group_size": 64,
1485
+ "bits": 8,
1486
+ "mode": "affine"
1487
+ },
1488
+ "model.layers.22.self_attn.v_proj": {
1489
+ "group_size": 64,
1490
+ "bits": 8,
1491
+ "mode": "affine"
1492
+ },
1493
+ "model.layers.22.self_attn.o_proj": {
1494
+ "group_size": 64,
1495
+ "bits": 8,
1496
+ "mode": "affine"
1497
+ },
1498
+ "model.layers.22.mlp.router": {
1499
+ "group_size": 64,
1500
+ "bits": 8
1501
+ },
1502
+ "model.layers.23.self_attn.q_proj": {
1503
+ "group_size": 64,
1504
+ "bits": 8,
1505
+ "mode": "affine"
1506
+ },
1507
+ "model.layers.23.self_attn.k_proj": {
1508
+ "group_size": 64,
1509
+ "bits": 8,
1510
+ "mode": "affine"
1511
+ },
1512
+ "model.layers.23.self_attn.v_proj": {
1513
+ "group_size": 64,
1514
+ "bits": 8,
1515
+ "mode": "affine"
1516
+ },
1517
+ "model.layers.23.self_attn.o_proj": {
1518
+ "group_size": 64,
1519
+ "bits": 8,
1520
+ "mode": "affine"
1521
+ },
1522
+ "model.layers.23.mlp.router": {
1523
+ "group_size": 64,
1524
+ "bits": 8
1525
+ },
1526
+ "model.layers.24.self_attn.q_proj": {
1527
+ "group_size": 64,
1528
+ "bits": 8,
1529
+ "mode": "affine"
1530
+ },
1531
+ "model.layers.24.self_attn.k_proj": {
1532
+ "group_size": 64,
1533
+ "bits": 8,
1534
+ "mode": "affine"
1535
+ },
1536
+ "model.layers.24.self_attn.v_proj": {
1537
+ "group_size": 64,
1538
+ "bits": 8,
1539
+ "mode": "affine"
1540
+ },
1541
+ "model.layers.24.self_attn.o_proj": {
1542
+ "group_size": 64,
1543
+ "bits": 8,
1544
+ "mode": "affine"
1545
+ },
1546
+ "model.layers.24.mlp.router": {
1547
+ "group_size": 64,
1548
+ "bits": 8
1549
+ },
1550
+ "model.layers.25.self_attn.q_proj": {
1551
+ "group_size": 64,
1552
+ "bits": 8,
1553
+ "mode": "affine"
1554
+ },
1555
+ "model.layers.25.self_attn.k_proj": {
1556
+ "group_size": 64,
1557
+ "bits": 8,
1558
+ "mode": "affine"
1559
+ },
1560
+ "model.layers.25.self_attn.v_proj": {
1561
+ "group_size": 64,
1562
+ "bits": 8,
1563
+ "mode": "affine"
1564
+ },
1565
+ "model.layers.25.self_attn.o_proj": {
1566
+ "group_size": 64,
1567
+ "bits": 8,
1568
+ "mode": "affine"
1569
+ },
1570
+ "model.layers.25.mlp.router": {
1571
+ "group_size": 64,
1572
+ "bits": 8
1573
+ },
1574
+ "model.layers.26.self_attn.q_proj": {
1575
+ "group_size": 64,
1576
+ "bits": 8,
1577
+ "mode": "affine"
1578
+ },
1579
+ "model.layers.26.self_attn.k_proj": {
1580
+ "group_size": 64,
1581
+ "bits": 8,
1582
+ "mode": "affine"
1583
+ },
1584
+ "model.layers.26.self_attn.v_proj": {
1585
+ "group_size": 64,
1586
+ "bits": 8,
1587
+ "mode": "affine"
1588
+ },
1589
+ "model.layers.26.self_attn.o_proj": {
1590
+ "group_size": 64,
1591
+ "bits": 8,
1592
+ "mode": "affine"
1593
+ },
1594
+ "model.layers.26.mlp.router": {
1595
+ "group_size": 64,
1596
+ "bits": 8
1597
+ },
1598
+ "model.layers.27.self_attn.q_proj": {
1599
+ "group_size": 64,
1600
+ "bits": 8,
1601
+ "mode": "affine"
1602
+ },
1603
+ "model.layers.27.self_attn.k_proj": {
1604
+ "group_size": 64,
1605
+ "bits": 8,
1606
+ "mode": "affine"
1607
+ },
1608
+ "model.layers.27.self_attn.v_proj": {
1609
+ "group_size": 64,
1610
+ "bits": 8,
1611
+ "mode": "affine"
1612
+ },
1613
+ "model.layers.27.self_attn.o_proj": {
1614
+ "group_size": 64,
1615
+ "bits": 8,
1616
+ "mode": "affine"
1617
+ },
1618
+ "model.layers.27.mlp.router": {
1619
+ "group_size": 64,
1620
+ "bits": 8
1621
+ },
1622
+ "model.layers.28.self_attn.q_proj": {
1623
+ "group_size": 64,
1624
+ "bits": 8,
1625
+ "mode": "affine"
1626
+ },
1627
+ "model.layers.28.self_attn.k_proj": {
1628
+ "group_size": 64,
1629
+ "bits": 8,
1630
+ "mode": "affine"
1631
+ },
1632
+ "model.layers.28.self_attn.v_proj": {
1633
+ "group_size": 64,
1634
+ "bits": 8,
1635
+ "mode": "affine"
1636
+ },
1637
+ "model.layers.28.self_attn.o_proj": {
1638
+ "group_size": 64,
1639
+ "bits": 8,
1640
+ "mode": "affine"
1641
+ },
1642
+ "model.layers.28.mlp.router": {
1643
+ "group_size": 64,
1644
+ "bits": 8
1645
+ },
1646
+ "model.layers.29.self_attn.q_proj": {
1647
+ "group_size": 64,
1648
+ "bits": 8,
1649
+ "mode": "affine"
1650
+ },
1651
+ "model.layers.29.self_attn.k_proj": {
1652
+ "group_size": 64,
1653
+ "bits": 8,
1654
+ "mode": "affine"
1655
+ },
1656
+ "model.layers.29.self_attn.v_proj": {
1657
+ "group_size": 64,
1658
+ "bits": 8,
1659
+ "mode": "affine"
1660
+ },
1661
+ "model.layers.29.self_attn.o_proj": {
1662
+ "group_size": 64,
1663
+ "bits": 8,
1664
+ "mode": "affine"
1665
+ },
1666
+ "model.layers.29.mlp.router": {
1667
+ "group_size": 64,
1668
+ "bits": 8
1669
+ },
1670
+ "model.layers.30.self_attn.q_proj": {
1671
+ "group_size": 64,
1672
+ "bits": 8,
1673
+ "mode": "affine"
1674
+ },
1675
+ "model.layers.30.self_attn.k_proj": {
1676
+ "group_size": 64,
1677
+ "bits": 8,
1678
+ "mode": "affine"
1679
+ },
1680
+ "model.layers.30.self_attn.v_proj": {
1681
+ "group_size": 64,
1682
+ "bits": 8,
1683
+ "mode": "affine"
1684
+ },
1685
+ "model.layers.30.self_attn.o_proj": {
1686
+ "group_size": 64,
1687
+ "bits": 8,
1688
+ "mode": "affine"
1689
+ },
1690
+ "model.layers.30.mlp.router": {
1691
+ "group_size": 64,
1692
+ "bits": 8
1693
+ },
1694
+ "model.layers.31.self_attn.q_proj": {
1695
+ "group_size": 64,
1696
+ "bits": 8,
1697
+ "mode": "affine"
1698
+ },
1699
+ "model.layers.31.self_attn.k_proj": {
1700
+ "group_size": 64,
1701
+ "bits": 8,
1702
+ "mode": "affine"
1703
+ },
1704
+ "model.layers.31.self_attn.v_proj": {
1705
+ "group_size": 64,
1706
+ "bits": 8,
1707
+ "mode": "affine"
1708
+ },
1709
+ "model.layers.31.self_attn.o_proj": {
1710
+ "group_size": 64,
1711
+ "bits": 8,
1712
+ "mode": "affine"
1713
+ },
1714
+ "model.layers.31.mlp.router": {
1715
+ "group_size": 64,
1716
+ "bits": 8
1717
+ },
1718
+ "model.layers.32.self_attn.q_proj": {
1719
+ "group_size": 64,
1720
+ "bits": 8,
1721
+ "mode": "affine"
1722
+ },
1723
+ "model.layers.32.self_attn.k_proj": {
1724
+ "group_size": 64,
1725
+ "bits": 8,
1726
+ "mode": "affine"
1727
+ },
1728
+ "model.layers.32.self_attn.v_proj": {
1729
+ "group_size": 64,
1730
+ "bits": 8,
1731
+ "mode": "affine"
1732
+ },
1733
+ "model.layers.32.self_attn.o_proj": {
1734
+ "group_size": 64,
1735
+ "bits": 8,
1736
+ "mode": "affine"
1737
+ },
1738
+ "model.layers.32.mlp.router": {
1739
+ "group_size": 64,
1740
+ "bits": 8
1741
+ },
1742
+ "model.layers.33.self_attn.q_proj": {
1743
+ "group_size": 64,
1744
+ "bits": 8,
1745
+ "mode": "affine"
1746
+ },
1747
+ "model.layers.33.self_attn.k_proj": {
1748
+ "group_size": 64,
1749
+ "bits": 8,
1750
+ "mode": "affine"
1751
+ },
1752
+ "model.layers.33.self_attn.v_proj": {
1753
+ "group_size": 64,
1754
+ "bits": 8,
1755
+ "mode": "affine"
1756
+ },
1757
+ "model.layers.33.self_attn.o_proj": {
1758
+ "group_size": 64,
1759
+ "bits": 8,
1760
+ "mode": "affine"
1761
+ },
1762
+ "model.layers.33.mlp.router": {
1763
+ "group_size": 64,
1764
+ "bits": 8
1765
+ },
1766
+ "model.layers.34.self_attn.q_proj": {
1767
+ "group_size": 64,
1768
+ "bits": 8,
1769
+ "mode": "affine"
1770
+ },
1771
+ "model.layers.34.self_attn.k_proj": {
1772
+ "group_size": 64,
1773
+ "bits": 8,
1774
+ "mode": "affine"
1775
+ },
1776
+ "model.layers.34.self_attn.v_proj": {
1777
+ "group_size": 64,
1778
+ "bits": 8,
1779
+ "mode": "affine"
1780
+ },
1781
+ "model.layers.34.self_attn.o_proj": {
1782
+ "group_size": 64,
1783
+ "bits": 8,
1784
+ "mode": "affine"
1785
+ },
1786
+ "model.layers.34.mlp.router": {
1787
+ "group_size": 64,
1788
+ "bits": 8
1789
+ },
1790
+ "model.layers.35.self_attn.q_proj": {
1791
+ "group_size": 64,
1792
+ "bits": 8,
1793
+ "mode": "affine"
1794
+ },
1795
+ "model.layers.35.self_attn.k_proj": {
1796
+ "group_size": 64,
1797
+ "bits": 8,
1798
+ "mode": "affine"
1799
+ },
1800
+ "model.layers.35.self_attn.v_proj": {
1801
+ "group_size": 64,
1802
+ "bits": 8,
1803
+ "mode": "affine"
1804
+ },
1805
+ "model.layers.35.self_attn.o_proj": {
1806
+ "group_size": 64,
1807
+ "bits": 8,
1808
+ "mode": "affine"
1809
+ },
1810
+ "model.layers.35.mlp.router": {
1811
+ "group_size": 64,
1812
+ "bits": 8
1813
+ },
1814
+ "lm_head": {
1815
+ "group_size": 64,
1816
+ "bits": 8,
1817
+ "mode": "affine"
1818
+ }
1819
+ },
1820
+ "rms_norm_eps": 1e-05,
1821
+ "rope_scaling": {
1822
+ "beta_fast": 32.0,
1823
+ "beta_slow": 1.0,
1824
+ "factor": 32.0,
1825
+ "original_max_position_embeddings": 4096,
1826
+ "rope_type": "yarn",
1827
+ "truncate": false
1828
+ },
1829
+ "rope_theta": 150000,
1830
+ "router_aux_loss_coef": 0.9,
1831
+ "sliding_window": 128,
1832
+ "swiglu_limit": 7.0,
1833
+ "tie_word_embeddings": false,
1834
+ "transformers_version": "4.55.0.dev0",
1835
+ "use_cache": true,
1836
+ "vocab_size": 201088
1837
+ }