Safetensors
MLX
mlx-vlm
gemma4_unified
gemma4
agentic
coding
tool-use
lora-merged
4-bit precision
True2456 commited on
Commit
fe19cf9
·
verified ·
1 Parent(s): d616409

Upload config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. config.json +1353 -0
config.json ADDED
@@ -0,0 +1,1353 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Gemma4UnifiedForConditionalGeneration"
4
+ ],
5
+ "audio_config": {
6
+ "_name_or_path": "",
7
+ "architectures": null,
8
+ "audio_embed_dim": 640,
9
+ "chunk_size_feed_forward": 0,
10
+ "dtype": "bfloat16",
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1"
14
+ },
15
+ "initializer_range": 0.02,
16
+ "is_encoder_decoder": false,
17
+ "label2id": {
18
+ "LABEL_0": 0,
19
+ "LABEL_1": 1
20
+ },
21
+ "model_type": "gemma4_unified_audio",
22
+ "output_attentions": false,
23
+ "output_hidden_states": false,
24
+ "problem_type": null,
25
+ "return_dict": true,
26
+ "rms_norm_eps": 1e-06
27
+ },
28
+ "audio_token_id": 258881,
29
+ "boa_token_id": 256000,
30
+ "boi_token_id": 255999,
31
+ "do_sample": true,
32
+ "dtype": "bfloat16",
33
+ "eoa_token_index": 258883,
34
+ "eoi_token_id": 258882,
35
+ "eos_token_id": [
36
+ 1,
37
+ 106,
38
+ 50
39
+ ],
40
+ "generation_config": {
41
+ "bos_token_id": 2,
42
+ "do_sample": true,
43
+ "eos_token_id": [
44
+ 1,
45
+ 106,
46
+ 50
47
+ ],
48
+ "pad_token_id": 0,
49
+ "suppress_tokens": [
50
+ 258883,
51
+ 258882
52
+ ],
53
+ "temperature": 1.0,
54
+ "top_k": 64,
55
+ "top_p": 0.95,
56
+ "transformers_version": "5.10.0.dev0"
57
+ },
58
+ "image_token_id": 258880,
59
+ "initializer_range": 0.02,
60
+ "model_type": "gemma4_unified",
61
+ "quantization": {
62
+ "group_size": 64,
63
+ "bits": 4,
64
+ "mode": "affine",
65
+ "language_model.model.layers.0.mlp.gate_proj": {
66
+ "group_size": 64,
67
+ "bits": 8
68
+ },
69
+ "language_model.model.layers.0.mlp.down_proj": {
70
+ "group_size": 64,
71
+ "bits": 8
72
+ },
73
+ "language_model.model.layers.0.mlp.up_proj": {
74
+ "group_size": 64,
75
+ "bits": 8
76
+ },
77
+ "language_model.model.layers.1.mlp.gate_proj": {
78
+ "group_size": 64,
79
+ "bits": 8
80
+ },
81
+ "language_model.model.layers.1.mlp.down_proj": {
82
+ "group_size": 64,
83
+ "bits": 8
84
+ },
85
+ "language_model.model.layers.1.mlp.up_proj": {
86
+ "group_size": 64,
87
+ "bits": 8
88
+ },
89
+ "language_model.model.layers.2.mlp.gate_proj": {
90
+ "group_size": 64,
91
+ "bits": 8
92
+ },
93
+ "language_model.model.layers.2.mlp.down_proj": {
94
+ "group_size": 64,
95
+ "bits": 8
96
+ },
97
+ "language_model.model.layers.2.mlp.up_proj": {
98
+ "group_size": 64,
99
+ "bits": 8
100
+ },
101
+ "language_model.model.layers.3.mlp.gate_proj": {
102
+ "group_size": 64,
103
+ "bits": 8
104
+ },
105
+ "language_model.model.layers.3.mlp.down_proj": {
106
+ "group_size": 64,
107
+ "bits": 8
108
+ },
109
+ "language_model.model.layers.3.mlp.up_proj": {
110
+ "group_size": 64,
111
+ "bits": 8
112
+ },
113
+ "language_model.model.layers.4.mlp.gate_proj": {
114
+ "group_size": 64,
115
+ "bits": 8
116
+ },
117
+ "language_model.model.layers.4.mlp.down_proj": {
118
+ "group_size": 64,
119
+ "bits": 8
120
+ },
121
+ "language_model.model.layers.4.mlp.up_proj": {
122
+ "group_size": 64,
123
+ "bits": 8
124
+ },
125
+ "language_model.model.layers.5.mlp.gate_proj": {
126
+ "group_size": 64,
127
+ "bits": 8
128
+ },
129
+ "language_model.model.layers.5.mlp.down_proj": {
130
+ "group_size": 64,
131
+ "bits": 8
132
+ },
133
+ "language_model.model.layers.5.mlp.up_proj": {
134
+ "group_size": 64,
135
+ "bits": 8
136
+ },
137
+ "language_model.model.layers.6.mlp.gate_proj": {
138
+ "group_size": 64,
139
+ "bits": 8
140
+ },
141
+ "language_model.model.layers.6.mlp.down_proj": {
142
+ "group_size": 64,
143
+ "bits": 8
144
+ },
145
+ "language_model.model.layers.6.mlp.up_proj": {
146
+ "group_size": 64,
147
+ "bits": 8
148
+ },
149
+ "language_model.model.layers.7.mlp.gate_proj": {
150
+ "group_size": 64,
151
+ "bits": 8
152
+ },
153
+ "language_model.model.layers.7.mlp.down_proj": {
154
+ "group_size": 64,
155
+ "bits": 8
156
+ },
157
+ "language_model.model.layers.7.mlp.up_proj": {
158
+ "group_size": 64,
159
+ "bits": 8
160
+ },
161
+ "language_model.model.layers.8.mlp.gate_proj": {
162
+ "group_size": 64,
163
+ "bits": 8
164
+ },
165
+ "language_model.model.layers.8.mlp.down_proj": {
166
+ "group_size": 64,
167
+ "bits": 8
168
+ },
169
+ "language_model.model.layers.8.mlp.up_proj": {
170
+ "group_size": 64,
171
+ "bits": 8
172
+ },
173
+ "language_model.model.layers.9.mlp.gate_proj": {
174
+ "group_size": 64,
175
+ "bits": 8
176
+ },
177
+ "language_model.model.layers.9.mlp.down_proj": {
178
+ "group_size": 64,
179
+ "bits": 8
180
+ },
181
+ "language_model.model.layers.9.mlp.up_proj": {
182
+ "group_size": 64,
183
+ "bits": 8
184
+ },
185
+ "language_model.model.layers.10.mlp.gate_proj": {
186
+ "group_size": 64,
187
+ "bits": 8
188
+ },
189
+ "language_model.model.layers.10.mlp.down_proj": {
190
+ "group_size": 64,
191
+ "bits": 8
192
+ },
193
+ "language_model.model.layers.10.mlp.up_proj": {
194
+ "group_size": 64,
195
+ "bits": 8
196
+ },
197
+ "language_model.model.layers.11.mlp.gate_proj": {
198
+ "group_size": 64,
199
+ "bits": 8
200
+ },
201
+ "language_model.model.layers.11.mlp.down_proj": {
202
+ "group_size": 64,
203
+ "bits": 8
204
+ },
205
+ "language_model.model.layers.11.mlp.up_proj": {
206
+ "group_size": 64,
207
+ "bits": 8
208
+ },
209
+ "language_model.model.layers.12.mlp.gate_proj": {
210
+ "group_size": 64,
211
+ "bits": 8
212
+ },
213
+ "language_model.model.layers.12.mlp.down_proj": {
214
+ "group_size": 64,
215
+ "bits": 8
216
+ },
217
+ "language_model.model.layers.12.mlp.up_proj": {
218
+ "group_size": 64,
219
+ "bits": 8
220
+ },
221
+ "language_model.model.layers.13.mlp.gate_proj": {
222
+ "group_size": 64,
223
+ "bits": 8
224
+ },
225
+ "language_model.model.layers.13.mlp.down_proj": {
226
+ "group_size": 64,
227
+ "bits": 8
228
+ },
229
+ "language_model.model.layers.13.mlp.up_proj": {
230
+ "group_size": 64,
231
+ "bits": 8
232
+ },
233
+ "language_model.model.layers.14.mlp.gate_proj": {
234
+ "group_size": 64,
235
+ "bits": 8
236
+ },
237
+ "language_model.model.layers.14.mlp.down_proj": {
238
+ "group_size": 64,
239
+ "bits": 8
240
+ },
241
+ "language_model.model.layers.14.mlp.up_proj": {
242
+ "group_size": 64,
243
+ "bits": 8
244
+ },
245
+ "language_model.model.layers.15.mlp.gate_proj": {
246
+ "group_size": 64,
247
+ "bits": 8
248
+ },
249
+ "language_model.model.layers.15.mlp.down_proj": {
250
+ "group_size": 64,
251
+ "bits": 8
252
+ },
253
+ "language_model.model.layers.15.mlp.up_proj": {
254
+ "group_size": 64,
255
+ "bits": 8
256
+ },
257
+ "language_model.model.layers.16.mlp.gate_proj": {
258
+ "group_size": 64,
259
+ "bits": 8
260
+ },
261
+ "language_model.model.layers.16.mlp.down_proj": {
262
+ "group_size": 64,
263
+ "bits": 8
264
+ },
265
+ "language_model.model.layers.16.mlp.up_proj": {
266
+ "group_size": 64,
267
+ "bits": 8
268
+ },
269
+ "language_model.model.layers.17.mlp.gate_proj": {
270
+ "group_size": 64,
271
+ "bits": 8
272
+ },
273
+ "language_model.model.layers.17.mlp.down_proj": {
274
+ "group_size": 64,
275
+ "bits": 8
276
+ },
277
+ "language_model.model.layers.17.mlp.up_proj": {
278
+ "group_size": 64,
279
+ "bits": 8
280
+ },
281
+ "language_model.model.layers.18.mlp.gate_proj": {
282
+ "group_size": 64,
283
+ "bits": 8
284
+ },
285
+ "language_model.model.layers.18.mlp.down_proj": {
286
+ "group_size": 64,
287
+ "bits": 8
288
+ },
289
+ "language_model.model.layers.18.mlp.up_proj": {
290
+ "group_size": 64,
291
+ "bits": 8
292
+ },
293
+ "language_model.model.layers.19.mlp.gate_proj": {
294
+ "group_size": 64,
295
+ "bits": 8
296
+ },
297
+ "language_model.model.layers.19.mlp.down_proj": {
298
+ "group_size": 64,
299
+ "bits": 8
300
+ },
301
+ "language_model.model.layers.19.mlp.up_proj": {
302
+ "group_size": 64,
303
+ "bits": 8
304
+ },
305
+ "language_model.model.layers.20.mlp.gate_proj": {
306
+ "group_size": 64,
307
+ "bits": 8
308
+ },
309
+ "language_model.model.layers.20.mlp.down_proj": {
310
+ "group_size": 64,
311
+ "bits": 8
312
+ },
313
+ "language_model.model.layers.20.mlp.up_proj": {
314
+ "group_size": 64,
315
+ "bits": 8
316
+ },
317
+ "language_model.model.layers.21.mlp.gate_proj": {
318
+ "group_size": 64,
319
+ "bits": 8
320
+ },
321
+ "language_model.model.layers.21.mlp.down_proj": {
322
+ "group_size": 64,
323
+ "bits": 8
324
+ },
325
+ "language_model.model.layers.21.mlp.up_proj": {
326
+ "group_size": 64,
327
+ "bits": 8
328
+ },
329
+ "language_model.model.layers.22.mlp.gate_proj": {
330
+ "group_size": 64,
331
+ "bits": 8
332
+ },
333
+ "language_model.model.layers.22.mlp.down_proj": {
334
+ "group_size": 64,
335
+ "bits": 8
336
+ },
337
+ "language_model.model.layers.22.mlp.up_proj": {
338
+ "group_size": 64,
339
+ "bits": 8
340
+ },
341
+ "language_model.model.layers.23.mlp.gate_proj": {
342
+ "group_size": 64,
343
+ "bits": 8
344
+ },
345
+ "language_model.model.layers.23.mlp.down_proj": {
346
+ "group_size": 64,
347
+ "bits": 8
348
+ },
349
+ "language_model.model.layers.23.mlp.up_proj": {
350
+ "group_size": 64,
351
+ "bits": 8
352
+ },
353
+ "language_model.model.layers.24.mlp.gate_proj": {
354
+ "group_size": 64,
355
+ "bits": 8
356
+ },
357
+ "language_model.model.layers.24.mlp.down_proj": {
358
+ "group_size": 64,
359
+ "bits": 8
360
+ },
361
+ "language_model.model.layers.24.mlp.up_proj": {
362
+ "group_size": 64,
363
+ "bits": 8
364
+ },
365
+ "language_model.model.layers.25.mlp.gate_proj": {
366
+ "group_size": 64,
367
+ "bits": 8
368
+ },
369
+ "language_model.model.layers.25.mlp.down_proj": {
370
+ "group_size": 64,
371
+ "bits": 8
372
+ },
373
+ "language_model.model.layers.25.mlp.up_proj": {
374
+ "group_size": 64,
375
+ "bits": 8
376
+ },
377
+ "language_model.model.layers.26.mlp.gate_proj": {
378
+ "group_size": 64,
379
+ "bits": 8
380
+ },
381
+ "language_model.model.layers.26.mlp.down_proj": {
382
+ "group_size": 64,
383
+ "bits": 8
384
+ },
385
+ "language_model.model.layers.26.mlp.up_proj": {
386
+ "group_size": 64,
387
+ "bits": 8
388
+ },
389
+ "language_model.model.layers.27.mlp.gate_proj": {
390
+ "group_size": 64,
391
+ "bits": 8
392
+ },
393
+ "language_model.model.layers.27.mlp.down_proj": {
394
+ "group_size": 64,
395
+ "bits": 8
396
+ },
397
+ "language_model.model.layers.27.mlp.up_proj": {
398
+ "group_size": 64,
399
+ "bits": 8
400
+ },
401
+ "language_model.model.layers.28.mlp.gate_proj": {
402
+ "group_size": 64,
403
+ "bits": 8
404
+ },
405
+ "language_model.model.layers.28.mlp.down_proj": {
406
+ "group_size": 64,
407
+ "bits": 8
408
+ },
409
+ "language_model.model.layers.28.mlp.up_proj": {
410
+ "group_size": 64,
411
+ "bits": 8
412
+ },
413
+ "language_model.model.layers.29.mlp.gate_proj": {
414
+ "group_size": 64,
415
+ "bits": 8
416
+ },
417
+ "language_model.model.layers.29.mlp.down_proj": {
418
+ "group_size": 64,
419
+ "bits": 8
420
+ },
421
+ "language_model.model.layers.29.mlp.up_proj": {
422
+ "group_size": 64,
423
+ "bits": 8
424
+ },
425
+ "language_model.model.layers.30.mlp.gate_proj": {
426
+ "group_size": 64,
427
+ "bits": 8
428
+ },
429
+ "language_model.model.layers.30.mlp.down_proj": {
430
+ "group_size": 64,
431
+ "bits": 8
432
+ },
433
+ "language_model.model.layers.30.mlp.up_proj": {
434
+ "group_size": 64,
435
+ "bits": 8
436
+ },
437
+ "language_model.model.layers.31.mlp.gate_proj": {
438
+ "group_size": 64,
439
+ "bits": 8
440
+ },
441
+ "language_model.model.layers.31.mlp.down_proj": {
442
+ "group_size": 64,
443
+ "bits": 8
444
+ },
445
+ "language_model.model.layers.31.mlp.up_proj": {
446
+ "group_size": 64,
447
+ "bits": 8
448
+ },
449
+ "language_model.model.layers.32.mlp.gate_proj": {
450
+ "group_size": 64,
451
+ "bits": 8
452
+ },
453
+ "language_model.model.layers.32.mlp.down_proj": {
454
+ "group_size": 64,
455
+ "bits": 8
456
+ },
457
+ "language_model.model.layers.32.mlp.up_proj": {
458
+ "group_size": 64,
459
+ "bits": 8
460
+ },
461
+ "language_model.model.layers.33.mlp.gate_proj": {
462
+ "group_size": 64,
463
+ "bits": 8
464
+ },
465
+ "language_model.model.layers.33.mlp.down_proj": {
466
+ "group_size": 64,
467
+ "bits": 8
468
+ },
469
+ "language_model.model.layers.33.mlp.up_proj": {
470
+ "group_size": 64,
471
+ "bits": 8
472
+ },
473
+ "language_model.model.layers.34.mlp.gate_proj": {
474
+ "group_size": 64,
475
+ "bits": 8
476
+ },
477
+ "language_model.model.layers.34.mlp.down_proj": {
478
+ "group_size": 64,
479
+ "bits": 8
480
+ },
481
+ "language_model.model.layers.34.mlp.up_proj": {
482
+ "group_size": 64,
483
+ "bits": 8
484
+ },
485
+ "language_model.model.layers.35.mlp.gate_proj": {
486
+ "group_size": 64,
487
+ "bits": 8
488
+ },
489
+ "language_model.model.layers.35.mlp.down_proj": {
490
+ "group_size": 64,
491
+ "bits": 8
492
+ },
493
+ "language_model.model.layers.35.mlp.up_proj": {
494
+ "group_size": 64,
495
+ "bits": 8
496
+ },
497
+ "language_model.model.layers.36.mlp.gate_proj": {
498
+ "group_size": 64,
499
+ "bits": 8
500
+ },
501
+ "language_model.model.layers.36.mlp.down_proj": {
502
+ "group_size": 64,
503
+ "bits": 8
504
+ },
505
+ "language_model.model.layers.36.mlp.up_proj": {
506
+ "group_size": 64,
507
+ "bits": 8
508
+ },
509
+ "language_model.model.layers.37.mlp.gate_proj": {
510
+ "group_size": 64,
511
+ "bits": 8
512
+ },
513
+ "language_model.model.layers.37.mlp.down_proj": {
514
+ "group_size": 64,
515
+ "bits": 8
516
+ },
517
+ "language_model.model.layers.37.mlp.up_proj": {
518
+ "group_size": 64,
519
+ "bits": 8
520
+ },
521
+ "language_model.model.layers.38.mlp.gate_proj": {
522
+ "group_size": 64,
523
+ "bits": 8
524
+ },
525
+ "language_model.model.layers.38.mlp.down_proj": {
526
+ "group_size": 64,
527
+ "bits": 8
528
+ },
529
+ "language_model.model.layers.38.mlp.up_proj": {
530
+ "group_size": 64,
531
+ "bits": 8
532
+ },
533
+ "language_model.model.layers.39.mlp.gate_proj": {
534
+ "group_size": 64,
535
+ "bits": 8
536
+ },
537
+ "language_model.model.layers.39.mlp.down_proj": {
538
+ "group_size": 64,
539
+ "bits": 8
540
+ },
541
+ "language_model.model.layers.39.mlp.up_proj": {
542
+ "group_size": 64,
543
+ "bits": 8
544
+ },
545
+ "language_model.model.layers.40.mlp.gate_proj": {
546
+ "group_size": 64,
547
+ "bits": 8
548
+ },
549
+ "language_model.model.layers.40.mlp.down_proj": {
550
+ "group_size": 64,
551
+ "bits": 8
552
+ },
553
+ "language_model.model.layers.40.mlp.up_proj": {
554
+ "group_size": 64,
555
+ "bits": 8
556
+ },
557
+ "language_model.model.layers.41.mlp.gate_proj": {
558
+ "group_size": 64,
559
+ "bits": 8
560
+ },
561
+ "language_model.model.layers.41.mlp.down_proj": {
562
+ "group_size": 64,
563
+ "bits": 8
564
+ },
565
+ "language_model.model.layers.41.mlp.up_proj": {
566
+ "group_size": 64,
567
+ "bits": 8
568
+ },
569
+ "language_model.model.layers.42.mlp.gate_proj": {
570
+ "group_size": 64,
571
+ "bits": 8
572
+ },
573
+ "language_model.model.layers.42.mlp.down_proj": {
574
+ "group_size": 64,
575
+ "bits": 8
576
+ },
577
+ "language_model.model.layers.42.mlp.up_proj": {
578
+ "group_size": 64,
579
+ "bits": 8
580
+ },
581
+ "language_model.model.layers.43.mlp.gate_proj": {
582
+ "group_size": 64,
583
+ "bits": 8
584
+ },
585
+ "language_model.model.layers.43.mlp.down_proj": {
586
+ "group_size": 64,
587
+ "bits": 8
588
+ },
589
+ "language_model.model.layers.43.mlp.up_proj": {
590
+ "group_size": 64,
591
+ "bits": 8
592
+ },
593
+ "language_model.model.layers.44.mlp.gate_proj": {
594
+ "group_size": 64,
595
+ "bits": 8
596
+ },
597
+ "language_model.model.layers.44.mlp.down_proj": {
598
+ "group_size": 64,
599
+ "bits": 8
600
+ },
601
+ "language_model.model.layers.44.mlp.up_proj": {
602
+ "group_size": 64,
603
+ "bits": 8
604
+ },
605
+ "language_model.model.layers.45.mlp.gate_proj": {
606
+ "group_size": 64,
607
+ "bits": 8
608
+ },
609
+ "language_model.model.layers.45.mlp.down_proj": {
610
+ "group_size": 64,
611
+ "bits": 8
612
+ },
613
+ "language_model.model.layers.45.mlp.up_proj": {
614
+ "group_size": 64,
615
+ "bits": 8
616
+ },
617
+ "language_model.model.layers.46.mlp.gate_proj": {
618
+ "group_size": 64,
619
+ "bits": 8
620
+ },
621
+ "language_model.model.layers.46.mlp.down_proj": {
622
+ "group_size": 64,
623
+ "bits": 8
624
+ },
625
+ "language_model.model.layers.46.mlp.up_proj": {
626
+ "group_size": 64,
627
+ "bits": 8
628
+ },
629
+ "language_model.model.layers.47.mlp.gate_proj": {
630
+ "group_size": 64,
631
+ "bits": 8
632
+ },
633
+ "language_model.model.layers.47.mlp.down_proj": {
634
+ "group_size": 64,
635
+ "bits": 8
636
+ },
637
+ "language_model.model.layers.47.mlp.up_proj": {
638
+ "group_size": 64,
639
+ "bits": 8
640
+ }
641
+ },
642
+ "quantization_config": {
643
+ "group_size": 64,
644
+ "bits": 4,
645
+ "mode": "affine",
646
+ "language_model.model.layers.0.mlp.gate_proj": {
647
+ "group_size": 64,
648
+ "bits": 8
649
+ },
650
+ "language_model.model.layers.0.mlp.down_proj": {
651
+ "group_size": 64,
652
+ "bits": 8
653
+ },
654
+ "language_model.model.layers.0.mlp.up_proj": {
655
+ "group_size": 64,
656
+ "bits": 8
657
+ },
658
+ "language_model.model.layers.1.mlp.gate_proj": {
659
+ "group_size": 64,
660
+ "bits": 8
661
+ },
662
+ "language_model.model.layers.1.mlp.down_proj": {
663
+ "group_size": 64,
664
+ "bits": 8
665
+ },
666
+ "language_model.model.layers.1.mlp.up_proj": {
667
+ "group_size": 64,
668
+ "bits": 8
669
+ },
670
+ "language_model.model.layers.2.mlp.gate_proj": {
671
+ "group_size": 64,
672
+ "bits": 8
673
+ },
674
+ "language_model.model.layers.2.mlp.down_proj": {
675
+ "group_size": 64,
676
+ "bits": 8
677
+ },
678
+ "language_model.model.layers.2.mlp.up_proj": {
679
+ "group_size": 64,
680
+ "bits": 8
681
+ },
682
+ "language_model.model.layers.3.mlp.gate_proj": {
683
+ "group_size": 64,
684
+ "bits": 8
685
+ },
686
+ "language_model.model.layers.3.mlp.down_proj": {
687
+ "group_size": 64,
688
+ "bits": 8
689
+ },
690
+ "language_model.model.layers.3.mlp.up_proj": {
691
+ "group_size": 64,
692
+ "bits": 8
693
+ },
694
+ "language_model.model.layers.4.mlp.gate_proj": {
695
+ "group_size": 64,
696
+ "bits": 8
697
+ },
698
+ "language_model.model.layers.4.mlp.down_proj": {
699
+ "group_size": 64,
700
+ "bits": 8
701
+ },
702
+ "language_model.model.layers.4.mlp.up_proj": {
703
+ "group_size": 64,
704
+ "bits": 8
705
+ },
706
+ "language_model.model.layers.5.mlp.gate_proj": {
707
+ "group_size": 64,
708
+ "bits": 8
709
+ },
710
+ "language_model.model.layers.5.mlp.down_proj": {
711
+ "group_size": 64,
712
+ "bits": 8
713
+ },
714
+ "language_model.model.layers.5.mlp.up_proj": {
715
+ "group_size": 64,
716
+ "bits": 8
717
+ },
718
+ "language_model.model.layers.6.mlp.gate_proj": {
719
+ "group_size": 64,
720
+ "bits": 8
721
+ },
722
+ "language_model.model.layers.6.mlp.down_proj": {
723
+ "group_size": 64,
724
+ "bits": 8
725
+ },
726
+ "language_model.model.layers.6.mlp.up_proj": {
727
+ "group_size": 64,
728
+ "bits": 8
729
+ },
730
+ "language_model.model.layers.7.mlp.gate_proj": {
731
+ "group_size": 64,
732
+ "bits": 8
733
+ },
734
+ "language_model.model.layers.7.mlp.down_proj": {
735
+ "group_size": 64,
736
+ "bits": 8
737
+ },
738
+ "language_model.model.layers.7.mlp.up_proj": {
739
+ "group_size": 64,
740
+ "bits": 8
741
+ },
742
+ "language_model.model.layers.8.mlp.gate_proj": {
743
+ "group_size": 64,
744
+ "bits": 8
745
+ },
746
+ "language_model.model.layers.8.mlp.down_proj": {
747
+ "group_size": 64,
748
+ "bits": 8
749
+ },
750
+ "language_model.model.layers.8.mlp.up_proj": {
751
+ "group_size": 64,
752
+ "bits": 8
753
+ },
754
+ "language_model.model.layers.9.mlp.gate_proj": {
755
+ "group_size": 64,
756
+ "bits": 8
757
+ },
758
+ "language_model.model.layers.9.mlp.down_proj": {
759
+ "group_size": 64,
760
+ "bits": 8
761
+ },
762
+ "language_model.model.layers.9.mlp.up_proj": {
763
+ "group_size": 64,
764
+ "bits": 8
765
+ },
766
+ "language_model.model.layers.10.mlp.gate_proj": {
767
+ "group_size": 64,
768
+ "bits": 8
769
+ },
770
+ "language_model.model.layers.10.mlp.down_proj": {
771
+ "group_size": 64,
772
+ "bits": 8
773
+ },
774
+ "language_model.model.layers.10.mlp.up_proj": {
775
+ "group_size": 64,
776
+ "bits": 8
777
+ },
778
+ "language_model.model.layers.11.mlp.gate_proj": {
779
+ "group_size": 64,
780
+ "bits": 8
781
+ },
782
+ "language_model.model.layers.11.mlp.down_proj": {
783
+ "group_size": 64,
784
+ "bits": 8
785
+ },
786
+ "language_model.model.layers.11.mlp.up_proj": {
787
+ "group_size": 64,
788
+ "bits": 8
789
+ },
790
+ "language_model.model.layers.12.mlp.gate_proj": {
791
+ "group_size": 64,
792
+ "bits": 8
793
+ },
794
+ "language_model.model.layers.12.mlp.down_proj": {
795
+ "group_size": 64,
796
+ "bits": 8
797
+ },
798
+ "language_model.model.layers.12.mlp.up_proj": {
799
+ "group_size": 64,
800
+ "bits": 8
801
+ },
802
+ "language_model.model.layers.13.mlp.gate_proj": {
803
+ "group_size": 64,
804
+ "bits": 8
805
+ },
806
+ "language_model.model.layers.13.mlp.down_proj": {
807
+ "group_size": 64,
808
+ "bits": 8
809
+ },
810
+ "language_model.model.layers.13.mlp.up_proj": {
811
+ "group_size": 64,
812
+ "bits": 8
813
+ },
814
+ "language_model.model.layers.14.mlp.gate_proj": {
815
+ "group_size": 64,
816
+ "bits": 8
817
+ },
818
+ "language_model.model.layers.14.mlp.down_proj": {
819
+ "group_size": 64,
820
+ "bits": 8
821
+ },
822
+ "language_model.model.layers.14.mlp.up_proj": {
823
+ "group_size": 64,
824
+ "bits": 8
825
+ },
826
+ "language_model.model.layers.15.mlp.gate_proj": {
827
+ "group_size": 64,
828
+ "bits": 8
829
+ },
830
+ "language_model.model.layers.15.mlp.down_proj": {
831
+ "group_size": 64,
832
+ "bits": 8
833
+ },
834
+ "language_model.model.layers.15.mlp.up_proj": {
835
+ "group_size": 64,
836
+ "bits": 8
837
+ },
838
+ "language_model.model.layers.16.mlp.gate_proj": {
839
+ "group_size": 64,
840
+ "bits": 8
841
+ },
842
+ "language_model.model.layers.16.mlp.down_proj": {
843
+ "group_size": 64,
844
+ "bits": 8
845
+ },
846
+ "language_model.model.layers.16.mlp.up_proj": {
847
+ "group_size": 64,
848
+ "bits": 8
849
+ },
850
+ "language_model.model.layers.17.mlp.gate_proj": {
851
+ "group_size": 64,
852
+ "bits": 8
853
+ },
854
+ "language_model.model.layers.17.mlp.down_proj": {
855
+ "group_size": 64,
856
+ "bits": 8
857
+ },
858
+ "language_model.model.layers.17.mlp.up_proj": {
859
+ "group_size": 64,
860
+ "bits": 8
861
+ },
862
+ "language_model.model.layers.18.mlp.gate_proj": {
863
+ "group_size": 64,
864
+ "bits": 8
865
+ },
866
+ "language_model.model.layers.18.mlp.down_proj": {
867
+ "group_size": 64,
868
+ "bits": 8
869
+ },
870
+ "language_model.model.layers.18.mlp.up_proj": {
871
+ "group_size": 64,
872
+ "bits": 8
873
+ },
874
+ "language_model.model.layers.19.mlp.gate_proj": {
875
+ "group_size": 64,
876
+ "bits": 8
877
+ },
878
+ "language_model.model.layers.19.mlp.down_proj": {
879
+ "group_size": 64,
880
+ "bits": 8
881
+ },
882
+ "language_model.model.layers.19.mlp.up_proj": {
883
+ "group_size": 64,
884
+ "bits": 8
885
+ },
886
+ "language_model.model.layers.20.mlp.gate_proj": {
887
+ "group_size": 64,
888
+ "bits": 8
889
+ },
890
+ "language_model.model.layers.20.mlp.down_proj": {
891
+ "group_size": 64,
892
+ "bits": 8
893
+ },
894
+ "language_model.model.layers.20.mlp.up_proj": {
895
+ "group_size": 64,
896
+ "bits": 8
897
+ },
898
+ "language_model.model.layers.21.mlp.gate_proj": {
899
+ "group_size": 64,
900
+ "bits": 8
901
+ },
902
+ "language_model.model.layers.21.mlp.down_proj": {
903
+ "group_size": 64,
904
+ "bits": 8
905
+ },
906
+ "language_model.model.layers.21.mlp.up_proj": {
907
+ "group_size": 64,
908
+ "bits": 8
909
+ },
910
+ "language_model.model.layers.22.mlp.gate_proj": {
911
+ "group_size": 64,
912
+ "bits": 8
913
+ },
914
+ "language_model.model.layers.22.mlp.down_proj": {
915
+ "group_size": 64,
916
+ "bits": 8
917
+ },
918
+ "language_model.model.layers.22.mlp.up_proj": {
919
+ "group_size": 64,
920
+ "bits": 8
921
+ },
922
+ "language_model.model.layers.23.mlp.gate_proj": {
923
+ "group_size": 64,
924
+ "bits": 8
925
+ },
926
+ "language_model.model.layers.23.mlp.down_proj": {
927
+ "group_size": 64,
928
+ "bits": 8
929
+ },
930
+ "language_model.model.layers.23.mlp.up_proj": {
931
+ "group_size": 64,
932
+ "bits": 8
933
+ },
934
+ "language_model.model.layers.24.mlp.gate_proj": {
935
+ "group_size": 64,
936
+ "bits": 8
937
+ },
938
+ "language_model.model.layers.24.mlp.down_proj": {
939
+ "group_size": 64,
940
+ "bits": 8
941
+ },
942
+ "language_model.model.layers.24.mlp.up_proj": {
943
+ "group_size": 64,
944
+ "bits": 8
945
+ },
946
+ "language_model.model.layers.25.mlp.gate_proj": {
947
+ "group_size": 64,
948
+ "bits": 8
949
+ },
950
+ "language_model.model.layers.25.mlp.down_proj": {
951
+ "group_size": 64,
952
+ "bits": 8
953
+ },
954
+ "language_model.model.layers.25.mlp.up_proj": {
955
+ "group_size": 64,
956
+ "bits": 8
957
+ },
958
+ "language_model.model.layers.26.mlp.gate_proj": {
959
+ "group_size": 64,
960
+ "bits": 8
961
+ },
962
+ "language_model.model.layers.26.mlp.down_proj": {
963
+ "group_size": 64,
964
+ "bits": 8
965
+ },
966
+ "language_model.model.layers.26.mlp.up_proj": {
967
+ "group_size": 64,
968
+ "bits": 8
969
+ },
970
+ "language_model.model.layers.27.mlp.gate_proj": {
971
+ "group_size": 64,
972
+ "bits": 8
973
+ },
974
+ "language_model.model.layers.27.mlp.down_proj": {
975
+ "group_size": 64,
976
+ "bits": 8
977
+ },
978
+ "language_model.model.layers.27.mlp.up_proj": {
979
+ "group_size": 64,
980
+ "bits": 8
981
+ },
982
+ "language_model.model.layers.28.mlp.gate_proj": {
983
+ "group_size": 64,
984
+ "bits": 8
985
+ },
986
+ "language_model.model.layers.28.mlp.down_proj": {
987
+ "group_size": 64,
988
+ "bits": 8
989
+ },
990
+ "language_model.model.layers.28.mlp.up_proj": {
991
+ "group_size": 64,
992
+ "bits": 8
993
+ },
994
+ "language_model.model.layers.29.mlp.gate_proj": {
995
+ "group_size": 64,
996
+ "bits": 8
997
+ },
998
+ "language_model.model.layers.29.mlp.down_proj": {
999
+ "group_size": 64,
1000
+ "bits": 8
1001
+ },
1002
+ "language_model.model.layers.29.mlp.up_proj": {
1003
+ "group_size": 64,
1004
+ "bits": 8
1005
+ },
1006
+ "language_model.model.layers.30.mlp.gate_proj": {
1007
+ "group_size": 64,
1008
+ "bits": 8
1009
+ },
1010
+ "language_model.model.layers.30.mlp.down_proj": {
1011
+ "group_size": 64,
1012
+ "bits": 8
1013
+ },
1014
+ "language_model.model.layers.30.mlp.up_proj": {
1015
+ "group_size": 64,
1016
+ "bits": 8
1017
+ },
1018
+ "language_model.model.layers.31.mlp.gate_proj": {
1019
+ "group_size": 64,
1020
+ "bits": 8
1021
+ },
1022
+ "language_model.model.layers.31.mlp.down_proj": {
1023
+ "group_size": 64,
1024
+ "bits": 8
1025
+ },
1026
+ "language_model.model.layers.31.mlp.up_proj": {
1027
+ "group_size": 64,
1028
+ "bits": 8
1029
+ },
1030
+ "language_model.model.layers.32.mlp.gate_proj": {
1031
+ "group_size": 64,
1032
+ "bits": 8
1033
+ },
1034
+ "language_model.model.layers.32.mlp.down_proj": {
1035
+ "group_size": 64,
1036
+ "bits": 8
1037
+ },
1038
+ "language_model.model.layers.32.mlp.up_proj": {
1039
+ "group_size": 64,
1040
+ "bits": 8
1041
+ },
1042
+ "language_model.model.layers.33.mlp.gate_proj": {
1043
+ "group_size": 64,
1044
+ "bits": 8
1045
+ },
1046
+ "language_model.model.layers.33.mlp.down_proj": {
1047
+ "group_size": 64,
1048
+ "bits": 8
1049
+ },
1050
+ "language_model.model.layers.33.mlp.up_proj": {
1051
+ "group_size": 64,
1052
+ "bits": 8
1053
+ },
1054
+ "language_model.model.layers.34.mlp.gate_proj": {
1055
+ "group_size": 64,
1056
+ "bits": 8
1057
+ },
1058
+ "language_model.model.layers.34.mlp.down_proj": {
1059
+ "group_size": 64,
1060
+ "bits": 8
1061
+ },
1062
+ "language_model.model.layers.34.mlp.up_proj": {
1063
+ "group_size": 64,
1064
+ "bits": 8
1065
+ },
1066
+ "language_model.model.layers.35.mlp.gate_proj": {
1067
+ "group_size": 64,
1068
+ "bits": 8
1069
+ },
1070
+ "language_model.model.layers.35.mlp.down_proj": {
1071
+ "group_size": 64,
1072
+ "bits": 8
1073
+ },
1074
+ "language_model.model.layers.35.mlp.up_proj": {
1075
+ "group_size": 64,
1076
+ "bits": 8
1077
+ },
1078
+ "language_model.model.layers.36.mlp.gate_proj": {
1079
+ "group_size": 64,
1080
+ "bits": 8
1081
+ },
1082
+ "language_model.model.layers.36.mlp.down_proj": {
1083
+ "group_size": 64,
1084
+ "bits": 8
1085
+ },
1086
+ "language_model.model.layers.36.mlp.up_proj": {
1087
+ "group_size": 64,
1088
+ "bits": 8
1089
+ },
1090
+ "language_model.model.layers.37.mlp.gate_proj": {
1091
+ "group_size": 64,
1092
+ "bits": 8
1093
+ },
1094
+ "language_model.model.layers.37.mlp.down_proj": {
1095
+ "group_size": 64,
1096
+ "bits": 8
1097
+ },
1098
+ "language_model.model.layers.37.mlp.up_proj": {
1099
+ "group_size": 64,
1100
+ "bits": 8
1101
+ },
1102
+ "language_model.model.layers.38.mlp.gate_proj": {
1103
+ "group_size": 64,
1104
+ "bits": 8
1105
+ },
1106
+ "language_model.model.layers.38.mlp.down_proj": {
1107
+ "group_size": 64,
1108
+ "bits": 8
1109
+ },
1110
+ "language_model.model.layers.38.mlp.up_proj": {
1111
+ "group_size": 64,
1112
+ "bits": 8
1113
+ },
1114
+ "language_model.model.layers.39.mlp.gate_proj": {
1115
+ "group_size": 64,
1116
+ "bits": 8
1117
+ },
1118
+ "language_model.model.layers.39.mlp.down_proj": {
1119
+ "group_size": 64,
1120
+ "bits": 8
1121
+ },
1122
+ "language_model.model.layers.39.mlp.up_proj": {
1123
+ "group_size": 64,
1124
+ "bits": 8
1125
+ },
1126
+ "language_model.model.layers.40.mlp.gate_proj": {
1127
+ "group_size": 64,
1128
+ "bits": 8
1129
+ },
1130
+ "language_model.model.layers.40.mlp.down_proj": {
1131
+ "group_size": 64,
1132
+ "bits": 8
1133
+ },
1134
+ "language_model.model.layers.40.mlp.up_proj": {
1135
+ "group_size": 64,
1136
+ "bits": 8
1137
+ },
1138
+ "language_model.model.layers.41.mlp.gate_proj": {
1139
+ "group_size": 64,
1140
+ "bits": 8
1141
+ },
1142
+ "language_model.model.layers.41.mlp.down_proj": {
1143
+ "group_size": 64,
1144
+ "bits": 8
1145
+ },
1146
+ "language_model.model.layers.41.mlp.up_proj": {
1147
+ "group_size": 64,
1148
+ "bits": 8
1149
+ },
1150
+ "language_model.model.layers.42.mlp.gate_proj": {
1151
+ "group_size": 64,
1152
+ "bits": 8
1153
+ },
1154
+ "language_model.model.layers.42.mlp.down_proj": {
1155
+ "group_size": 64,
1156
+ "bits": 8
1157
+ },
1158
+ "language_model.model.layers.42.mlp.up_proj": {
1159
+ "group_size": 64,
1160
+ "bits": 8
1161
+ },
1162
+ "language_model.model.layers.43.mlp.gate_proj": {
1163
+ "group_size": 64,
1164
+ "bits": 8
1165
+ },
1166
+ "language_model.model.layers.43.mlp.down_proj": {
1167
+ "group_size": 64,
1168
+ "bits": 8
1169
+ },
1170
+ "language_model.model.layers.43.mlp.up_proj": {
1171
+ "group_size": 64,
1172
+ "bits": 8
1173
+ },
1174
+ "language_model.model.layers.44.mlp.gate_proj": {
1175
+ "group_size": 64,
1176
+ "bits": 8
1177
+ },
1178
+ "language_model.model.layers.44.mlp.down_proj": {
1179
+ "group_size": 64,
1180
+ "bits": 8
1181
+ },
1182
+ "language_model.model.layers.44.mlp.up_proj": {
1183
+ "group_size": 64,
1184
+ "bits": 8
1185
+ },
1186
+ "language_model.model.layers.45.mlp.gate_proj": {
1187
+ "group_size": 64,
1188
+ "bits": 8
1189
+ },
1190
+ "language_model.model.layers.45.mlp.down_proj": {
1191
+ "group_size": 64,
1192
+ "bits": 8
1193
+ },
1194
+ "language_model.model.layers.45.mlp.up_proj": {
1195
+ "group_size": 64,
1196
+ "bits": 8
1197
+ },
1198
+ "language_model.model.layers.46.mlp.gate_proj": {
1199
+ "group_size": 64,
1200
+ "bits": 8
1201
+ },
1202
+ "language_model.model.layers.46.mlp.down_proj": {
1203
+ "group_size": 64,
1204
+ "bits": 8
1205
+ },
1206
+ "language_model.model.layers.46.mlp.up_proj": {
1207
+ "group_size": 64,
1208
+ "bits": 8
1209
+ },
1210
+ "language_model.model.layers.47.mlp.gate_proj": {
1211
+ "group_size": 64,
1212
+ "bits": 8
1213
+ },
1214
+ "language_model.model.layers.47.mlp.down_proj": {
1215
+ "group_size": 64,
1216
+ "bits": 8
1217
+ },
1218
+ "language_model.model.layers.47.mlp.up_proj": {
1219
+ "group_size": 64,
1220
+ "bits": 8
1221
+ }
1222
+ },
1223
+ "temperature": 1.0,
1224
+ "text_config": {
1225
+ "attention_bias": false,
1226
+ "attention_dropout": 0.0,
1227
+ "attention_k_eq_v": true,
1228
+ "bos_token_id": 2,
1229
+ "dtype": "bfloat16",
1230
+ "enable_moe_block": false,
1231
+ "eos_token_id": 1,
1232
+ "final_logit_softcapping": 30.0,
1233
+ "global_head_dim": 512,
1234
+ "head_dim": 256,
1235
+ "hidden_activation": "gelu_pytorch_tanh",
1236
+ "hidden_size": 3840,
1237
+ "hidden_size_per_layer_input": 0,
1238
+ "initializer_range": 0.02,
1239
+ "intermediate_size": 15360,
1240
+ "layer_types": [
1241
+ "sliding_attention",
1242
+ "sliding_attention",
1243
+ "sliding_attention",
1244
+ "sliding_attention",
1245
+ "sliding_attention",
1246
+ "full_attention",
1247
+ "sliding_attention",
1248
+ "sliding_attention",
1249
+ "sliding_attention",
1250
+ "sliding_attention",
1251
+ "sliding_attention",
1252
+ "full_attention",
1253
+ "sliding_attention",
1254
+ "sliding_attention",
1255
+ "sliding_attention",
1256
+ "sliding_attention",
1257
+ "sliding_attention",
1258
+ "full_attention",
1259
+ "sliding_attention",
1260
+ "sliding_attention",
1261
+ "sliding_attention",
1262
+ "sliding_attention",
1263
+ "sliding_attention",
1264
+ "full_attention",
1265
+ "sliding_attention",
1266
+ "sliding_attention",
1267
+ "sliding_attention",
1268
+ "sliding_attention",
1269
+ "sliding_attention",
1270
+ "full_attention",
1271
+ "sliding_attention",
1272
+ "sliding_attention",
1273
+ "sliding_attention",
1274
+ "sliding_attention",
1275
+ "sliding_attention",
1276
+ "full_attention",
1277
+ "sliding_attention",
1278
+ "sliding_attention",
1279
+ "sliding_attention",
1280
+ "sliding_attention",
1281
+ "sliding_attention",
1282
+ "full_attention",
1283
+ "sliding_attention",
1284
+ "sliding_attention",
1285
+ "sliding_attention",
1286
+ "sliding_attention",
1287
+ "sliding_attention",
1288
+ "full_attention"
1289
+ ],
1290
+ "max_position_embeddings": 262144,
1291
+ "model_type": "gemma4_unified_text",
1292
+ "moe_intermediate_size": null,
1293
+ "num_attention_heads": 16,
1294
+ "num_experts": null,
1295
+ "num_global_key_value_heads": 1,
1296
+ "num_hidden_layers": 48,
1297
+ "num_key_value_heads": 8,
1298
+ "num_kv_shared_layers": 0,
1299
+ "pad_token_id": 0,
1300
+ "rms_norm_eps": 1e-06,
1301
+ "rope_parameters": {
1302
+ "full_attention": {
1303
+ "partial_rotary_factor": 0.25,
1304
+ "rope_theta": 1000000.0,
1305
+ "rope_type": "proportional"
1306
+ },
1307
+ "sliding_attention": {
1308
+ "rope_theta": 10000.0,
1309
+ "rope_type": "default"
1310
+ }
1311
+ },
1312
+ "sliding_window": 1024,
1313
+ "tie_word_embeddings": true,
1314
+ "top_k_experts": null,
1315
+ "use_bidirectional_attention": "vision",
1316
+ "use_cache": true,
1317
+ "use_double_wide_mlp": false,
1318
+ "vocab_size": 262144,
1319
+ "vocab_size_per_layer_input": 0
1320
+ },
1321
+ "tie_word_embeddings": true,
1322
+ "top_k": 64,
1323
+ "top_p": 0.95,
1324
+ "transformers_version": "5.10.0.dev0",
1325
+ "video_token_id": 258884,
1326
+ "vision_config": {
1327
+ "_name_or_path": "",
1328
+ "architectures": null,
1329
+ "chunk_size_feed_forward": 0,
1330
+ "dtype": "bfloat16",
1331
+ "id2label": {
1332
+ "0": "LABEL_0",
1333
+ "1": "LABEL_1"
1334
+ },
1335
+ "initializer_range": 0.02,
1336
+ "is_encoder_decoder": false,
1337
+ "label2id": {
1338
+ "LABEL_0": 0,
1339
+ "LABEL_1": 1
1340
+ },
1341
+ "mm_embed_dim": 3840,
1342
+ "mm_posemb_size": 1120,
1343
+ "model_type": "gemma4_unified_vision",
1344
+ "output_attentions": false,
1345
+ "output_hidden_states": false,
1346
+ "output_proj_dims": 3840,
1347
+ "patch_size": 16,
1348
+ "pooling_kernel_size": 3,
1349
+ "problem_type": null,
1350
+ "return_dict": true,
1351
+ "rms_norm_eps": 1e-06
1352
+ }
1353
+ }