krishnateja95 commited on
Commit
7a9d8e1
·
verified ·
1 Parent(s): b9c48c3

Upload config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. config.json +653 -0
config.json ADDED
@@ -0,0 +1,653 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Llama4ForConditionalGeneration"
4
+ ],
5
+ "boi_token_index": 200080,
6
+ "dtype": "bfloat16",
7
+ "eoi_token_index": 200081,
8
+ "image_token_index": 200092,
9
+ "model_type": "llama4",
10
+ "quantization_config": {
11
+ "config_groups": {
12
+ "group_0": {
13
+ "format": "float-quantized",
14
+ "input_activations": {
15
+ "actorder": null,
16
+ "block_structure": null,
17
+ "dynamic": true,
18
+ "group_size": 128,
19
+ "num_bits": 8,
20
+ "observer": null,
21
+ "observer_kwargs": {},
22
+ "strategy": "group",
23
+ "symmetric": true,
24
+ "type": "float"
25
+ },
26
+ "output_activations": null,
27
+ "targets": [
28
+ "Linear"
29
+ ],
30
+ "weights": {
31
+ "actorder": null,
32
+ "block_structure": [
33
+ 128,
34
+ 128
35
+ ],
36
+ "dynamic": false,
37
+ "group_size": null,
38
+ "num_bits": 8,
39
+ "observer": "minmax",
40
+ "observer_kwargs": {},
41
+ "strategy": "block",
42
+ "symmetric": true,
43
+ "type": "float"
44
+ }
45
+ }
46
+ },
47
+ "format": "float-quantized",
48
+ "global_compression_ratio": null,
49
+ "ignore": [
50
+ "vision_model.patch_embedding.linear",
51
+ "vision_model.model.layers.0.self_attn.q_proj",
52
+ "vision_model.model.layers.0.self_attn.k_proj",
53
+ "vision_model.model.layers.0.self_attn.v_proj",
54
+ "vision_model.model.layers.0.self_attn.o_proj",
55
+ "vision_model.model.layers.0.mlp.fc1",
56
+ "vision_model.model.layers.0.mlp.fc2",
57
+ "vision_model.model.layers.1.self_attn.q_proj",
58
+ "vision_model.model.layers.1.self_attn.k_proj",
59
+ "vision_model.model.layers.1.self_attn.v_proj",
60
+ "vision_model.model.layers.1.self_attn.o_proj",
61
+ "vision_model.model.layers.1.mlp.fc1",
62
+ "vision_model.model.layers.1.mlp.fc2",
63
+ "vision_model.model.layers.2.self_attn.q_proj",
64
+ "vision_model.model.layers.2.self_attn.k_proj",
65
+ "vision_model.model.layers.2.self_attn.v_proj",
66
+ "vision_model.model.layers.2.self_attn.o_proj",
67
+ "vision_model.model.layers.2.mlp.fc1",
68
+ "vision_model.model.layers.2.mlp.fc2",
69
+ "vision_model.model.layers.3.self_attn.q_proj",
70
+ "vision_model.model.layers.3.self_attn.k_proj",
71
+ "vision_model.model.layers.3.self_attn.v_proj",
72
+ "vision_model.model.layers.3.self_attn.o_proj",
73
+ "vision_model.model.layers.3.mlp.fc1",
74
+ "vision_model.model.layers.3.mlp.fc2",
75
+ "vision_model.model.layers.4.self_attn.q_proj",
76
+ "vision_model.model.layers.4.self_attn.k_proj",
77
+ "vision_model.model.layers.4.self_attn.v_proj",
78
+ "vision_model.model.layers.4.self_attn.o_proj",
79
+ "vision_model.model.layers.4.mlp.fc1",
80
+ "vision_model.model.layers.4.mlp.fc2",
81
+ "vision_model.model.layers.5.self_attn.q_proj",
82
+ "vision_model.model.layers.5.self_attn.k_proj",
83
+ "vision_model.model.layers.5.self_attn.v_proj",
84
+ "vision_model.model.layers.5.self_attn.o_proj",
85
+ "vision_model.model.layers.5.mlp.fc1",
86
+ "vision_model.model.layers.5.mlp.fc2",
87
+ "vision_model.model.layers.6.self_attn.q_proj",
88
+ "vision_model.model.layers.6.self_attn.k_proj",
89
+ "vision_model.model.layers.6.self_attn.v_proj",
90
+ "vision_model.model.layers.6.self_attn.o_proj",
91
+ "vision_model.model.layers.6.mlp.fc1",
92
+ "vision_model.model.layers.6.mlp.fc2",
93
+ "vision_model.model.layers.7.self_attn.q_proj",
94
+ "vision_model.model.layers.7.self_attn.k_proj",
95
+ "vision_model.model.layers.7.self_attn.v_proj",
96
+ "vision_model.model.layers.7.self_attn.o_proj",
97
+ "vision_model.model.layers.7.mlp.fc1",
98
+ "vision_model.model.layers.7.mlp.fc2",
99
+ "vision_model.model.layers.8.self_attn.q_proj",
100
+ "vision_model.model.layers.8.self_attn.k_proj",
101
+ "vision_model.model.layers.8.self_attn.v_proj",
102
+ "vision_model.model.layers.8.self_attn.o_proj",
103
+ "vision_model.model.layers.8.mlp.fc1",
104
+ "vision_model.model.layers.8.mlp.fc2",
105
+ "vision_model.model.layers.9.self_attn.q_proj",
106
+ "vision_model.model.layers.9.self_attn.k_proj",
107
+ "vision_model.model.layers.9.self_attn.v_proj",
108
+ "vision_model.model.layers.9.self_attn.o_proj",
109
+ "vision_model.model.layers.9.mlp.fc1",
110
+ "vision_model.model.layers.9.mlp.fc2",
111
+ "vision_model.model.layers.10.self_attn.q_proj",
112
+ "vision_model.model.layers.10.self_attn.k_proj",
113
+ "vision_model.model.layers.10.self_attn.v_proj",
114
+ "vision_model.model.layers.10.self_attn.o_proj",
115
+ "vision_model.model.layers.10.mlp.fc1",
116
+ "vision_model.model.layers.10.mlp.fc2",
117
+ "vision_model.model.layers.11.self_attn.q_proj",
118
+ "vision_model.model.layers.11.self_attn.k_proj",
119
+ "vision_model.model.layers.11.self_attn.v_proj",
120
+ "vision_model.model.layers.11.self_attn.o_proj",
121
+ "vision_model.model.layers.11.mlp.fc1",
122
+ "vision_model.model.layers.11.mlp.fc2",
123
+ "vision_model.model.layers.12.self_attn.q_proj",
124
+ "vision_model.model.layers.12.self_attn.k_proj",
125
+ "vision_model.model.layers.12.self_attn.v_proj",
126
+ "vision_model.model.layers.12.self_attn.o_proj",
127
+ "vision_model.model.layers.12.mlp.fc1",
128
+ "vision_model.model.layers.12.mlp.fc2",
129
+ "vision_model.model.layers.13.self_attn.q_proj",
130
+ "vision_model.model.layers.13.self_attn.k_proj",
131
+ "vision_model.model.layers.13.self_attn.v_proj",
132
+ "vision_model.model.layers.13.self_attn.o_proj",
133
+ "vision_model.model.layers.13.mlp.fc1",
134
+ "vision_model.model.layers.13.mlp.fc2",
135
+ "vision_model.model.layers.14.self_attn.q_proj",
136
+ "vision_model.model.layers.14.self_attn.k_proj",
137
+ "vision_model.model.layers.14.self_attn.v_proj",
138
+ "vision_model.model.layers.14.self_attn.o_proj",
139
+ "vision_model.model.layers.14.mlp.fc1",
140
+ "vision_model.model.layers.14.mlp.fc2",
141
+ "vision_model.model.layers.15.self_attn.q_proj",
142
+ "vision_model.model.layers.15.self_attn.k_proj",
143
+ "vision_model.model.layers.15.self_attn.v_proj",
144
+ "vision_model.model.layers.15.self_attn.o_proj",
145
+ "vision_model.model.layers.15.mlp.fc1",
146
+ "vision_model.model.layers.15.mlp.fc2",
147
+ "vision_model.model.layers.16.self_attn.q_proj",
148
+ "vision_model.model.layers.16.self_attn.k_proj",
149
+ "vision_model.model.layers.16.self_attn.v_proj",
150
+ "vision_model.model.layers.16.self_attn.o_proj",
151
+ "vision_model.model.layers.16.mlp.fc1",
152
+ "vision_model.model.layers.16.mlp.fc2",
153
+ "vision_model.model.layers.17.self_attn.q_proj",
154
+ "vision_model.model.layers.17.self_attn.k_proj",
155
+ "vision_model.model.layers.17.self_attn.v_proj",
156
+ "vision_model.model.layers.17.self_attn.o_proj",
157
+ "vision_model.model.layers.17.mlp.fc1",
158
+ "vision_model.model.layers.17.mlp.fc2",
159
+ "vision_model.model.layers.18.self_attn.q_proj",
160
+ "vision_model.model.layers.18.self_attn.k_proj",
161
+ "vision_model.model.layers.18.self_attn.v_proj",
162
+ "vision_model.model.layers.18.self_attn.o_proj",
163
+ "vision_model.model.layers.18.mlp.fc1",
164
+ "vision_model.model.layers.18.mlp.fc2",
165
+ "vision_model.model.layers.19.self_attn.q_proj",
166
+ "vision_model.model.layers.19.self_attn.k_proj",
167
+ "vision_model.model.layers.19.self_attn.v_proj",
168
+ "vision_model.model.layers.19.self_attn.o_proj",
169
+ "vision_model.model.layers.19.mlp.fc1",
170
+ "vision_model.model.layers.19.mlp.fc2",
171
+ "vision_model.model.layers.20.self_attn.q_proj",
172
+ "vision_model.model.layers.20.self_attn.k_proj",
173
+ "vision_model.model.layers.20.self_attn.v_proj",
174
+ "vision_model.model.layers.20.self_attn.o_proj",
175
+ "vision_model.model.layers.20.mlp.fc1",
176
+ "vision_model.model.layers.20.mlp.fc2",
177
+ "vision_model.model.layers.21.self_attn.q_proj",
178
+ "vision_model.model.layers.21.self_attn.k_proj",
179
+ "vision_model.model.layers.21.self_attn.v_proj",
180
+ "vision_model.model.layers.21.self_attn.o_proj",
181
+ "vision_model.model.layers.21.mlp.fc1",
182
+ "vision_model.model.layers.21.mlp.fc2",
183
+ "vision_model.model.layers.22.self_attn.q_proj",
184
+ "vision_model.model.layers.22.self_attn.k_proj",
185
+ "vision_model.model.layers.22.self_attn.v_proj",
186
+ "vision_model.model.layers.22.self_attn.o_proj",
187
+ "vision_model.model.layers.22.mlp.fc1",
188
+ "vision_model.model.layers.22.mlp.fc2",
189
+ "vision_model.model.layers.23.self_attn.q_proj",
190
+ "vision_model.model.layers.23.self_attn.k_proj",
191
+ "vision_model.model.layers.23.self_attn.v_proj",
192
+ "vision_model.model.layers.23.self_attn.o_proj",
193
+ "vision_model.model.layers.23.mlp.fc1",
194
+ "vision_model.model.layers.23.mlp.fc2",
195
+ "vision_model.model.layers.24.self_attn.q_proj",
196
+ "vision_model.model.layers.24.self_attn.k_proj",
197
+ "vision_model.model.layers.24.self_attn.v_proj",
198
+ "vision_model.model.layers.24.self_attn.o_proj",
199
+ "vision_model.model.layers.24.mlp.fc1",
200
+ "vision_model.model.layers.24.mlp.fc2",
201
+ "vision_model.model.layers.25.self_attn.q_proj",
202
+ "vision_model.model.layers.25.self_attn.k_proj",
203
+ "vision_model.model.layers.25.self_attn.v_proj",
204
+ "vision_model.model.layers.25.self_attn.o_proj",
205
+ "vision_model.model.layers.25.mlp.fc1",
206
+ "vision_model.model.layers.25.mlp.fc2",
207
+ "vision_model.model.layers.26.self_attn.q_proj",
208
+ "vision_model.model.layers.26.self_attn.k_proj",
209
+ "vision_model.model.layers.26.self_attn.v_proj",
210
+ "vision_model.model.layers.26.self_attn.o_proj",
211
+ "vision_model.model.layers.26.mlp.fc1",
212
+ "vision_model.model.layers.26.mlp.fc2",
213
+ "vision_model.model.layers.27.self_attn.q_proj",
214
+ "vision_model.model.layers.27.self_attn.k_proj",
215
+ "vision_model.model.layers.27.self_attn.v_proj",
216
+ "vision_model.model.layers.27.self_attn.o_proj",
217
+ "vision_model.model.layers.27.mlp.fc1",
218
+ "vision_model.model.layers.27.mlp.fc2",
219
+ "vision_model.model.layers.28.self_attn.q_proj",
220
+ "vision_model.model.layers.28.self_attn.k_proj",
221
+ "vision_model.model.layers.28.self_attn.v_proj",
222
+ "vision_model.model.layers.28.self_attn.o_proj",
223
+ "vision_model.model.layers.28.mlp.fc1",
224
+ "vision_model.model.layers.28.mlp.fc2",
225
+ "vision_model.model.layers.29.self_attn.q_proj",
226
+ "vision_model.model.layers.29.self_attn.k_proj",
227
+ "vision_model.model.layers.29.self_attn.v_proj",
228
+ "vision_model.model.layers.29.self_attn.o_proj",
229
+ "vision_model.model.layers.29.mlp.fc1",
230
+ "vision_model.model.layers.29.mlp.fc2",
231
+ "vision_model.model.layers.30.self_attn.q_proj",
232
+ "vision_model.model.layers.30.self_attn.k_proj",
233
+ "vision_model.model.layers.30.self_attn.v_proj",
234
+ "vision_model.model.layers.30.self_attn.o_proj",
235
+ "vision_model.model.layers.30.mlp.fc1",
236
+ "vision_model.model.layers.30.mlp.fc2",
237
+ "vision_model.model.layers.31.self_attn.q_proj",
238
+ "vision_model.model.layers.31.self_attn.k_proj",
239
+ "vision_model.model.layers.31.self_attn.v_proj",
240
+ "vision_model.model.layers.31.self_attn.o_proj",
241
+ "vision_model.model.layers.31.mlp.fc1",
242
+ "vision_model.model.layers.31.mlp.fc2",
243
+ "vision_model.model.layers.32.self_attn.q_proj",
244
+ "vision_model.model.layers.32.self_attn.k_proj",
245
+ "vision_model.model.layers.32.self_attn.v_proj",
246
+ "vision_model.model.layers.32.self_attn.o_proj",
247
+ "vision_model.model.layers.32.mlp.fc1",
248
+ "vision_model.model.layers.32.mlp.fc2",
249
+ "vision_model.model.layers.33.self_attn.q_proj",
250
+ "vision_model.model.layers.33.self_attn.k_proj",
251
+ "vision_model.model.layers.33.self_attn.v_proj",
252
+ "vision_model.model.layers.33.self_attn.o_proj",
253
+ "vision_model.model.layers.33.mlp.fc1",
254
+ "vision_model.model.layers.33.mlp.fc2",
255
+ "vision_model.vision_adapter.mlp.fc1",
256
+ "vision_model.vision_adapter.mlp.fc2",
257
+ "multi_modal_projector.linear_1",
258
+ "language_model.model.layers.0.self_attn.q_proj",
259
+ "language_model.model.layers.0.self_attn.k_proj",
260
+ "language_model.model.layers.0.self_attn.v_proj",
261
+ "language_model.model.layers.0.self_attn.o_proj",
262
+ "language_model.model.layers.1.self_attn.q_proj",
263
+ "language_model.model.layers.1.self_attn.k_proj",
264
+ "language_model.model.layers.1.self_attn.v_proj",
265
+ "language_model.model.layers.1.self_attn.o_proj",
266
+ "language_model.model.layers.2.self_attn.q_proj",
267
+ "language_model.model.layers.2.self_attn.k_proj",
268
+ "language_model.model.layers.2.self_attn.v_proj",
269
+ "language_model.model.layers.2.self_attn.o_proj",
270
+ "language_model.model.layers.3.self_attn.q_proj",
271
+ "language_model.model.layers.3.self_attn.k_proj",
272
+ "language_model.model.layers.3.self_attn.v_proj",
273
+ "language_model.model.layers.3.self_attn.o_proj",
274
+ "language_model.model.layers.4.self_attn.q_proj",
275
+ "language_model.model.layers.4.self_attn.k_proj",
276
+ "language_model.model.layers.4.self_attn.v_proj",
277
+ "language_model.model.layers.4.self_attn.o_proj",
278
+ "language_model.model.layers.5.self_attn.q_proj",
279
+ "language_model.model.layers.5.self_attn.k_proj",
280
+ "language_model.model.layers.5.self_attn.v_proj",
281
+ "language_model.model.layers.5.self_attn.o_proj",
282
+ "language_model.model.layers.6.self_attn.q_proj",
283
+ "language_model.model.layers.6.self_attn.k_proj",
284
+ "language_model.model.layers.6.self_attn.v_proj",
285
+ "language_model.model.layers.6.self_attn.o_proj",
286
+ "language_model.model.layers.7.self_attn.q_proj",
287
+ "language_model.model.layers.7.self_attn.k_proj",
288
+ "language_model.model.layers.7.self_attn.v_proj",
289
+ "language_model.model.layers.7.self_attn.o_proj",
290
+ "language_model.model.layers.8.self_attn.q_proj",
291
+ "language_model.model.layers.8.self_attn.k_proj",
292
+ "language_model.model.layers.8.self_attn.v_proj",
293
+ "language_model.model.layers.8.self_attn.o_proj",
294
+ "language_model.model.layers.9.self_attn.q_proj",
295
+ "language_model.model.layers.9.self_attn.k_proj",
296
+ "language_model.model.layers.9.self_attn.v_proj",
297
+ "language_model.model.layers.9.self_attn.o_proj",
298
+ "language_model.model.layers.10.self_attn.q_proj",
299
+ "language_model.model.layers.10.self_attn.k_proj",
300
+ "language_model.model.layers.10.self_attn.v_proj",
301
+ "language_model.model.layers.10.self_attn.o_proj",
302
+ "language_model.model.layers.11.self_attn.q_proj",
303
+ "language_model.model.layers.11.self_attn.k_proj",
304
+ "language_model.model.layers.11.self_attn.v_proj",
305
+ "language_model.model.layers.11.self_attn.o_proj",
306
+ "language_model.model.layers.12.self_attn.q_proj",
307
+ "language_model.model.layers.12.self_attn.k_proj",
308
+ "language_model.model.layers.12.self_attn.v_proj",
309
+ "language_model.model.layers.12.self_attn.o_proj",
310
+ "language_model.model.layers.13.self_attn.q_proj",
311
+ "language_model.model.layers.13.self_attn.k_proj",
312
+ "language_model.model.layers.13.self_attn.v_proj",
313
+ "language_model.model.layers.13.self_attn.o_proj",
314
+ "language_model.model.layers.14.self_attn.q_proj",
315
+ "language_model.model.layers.14.self_attn.k_proj",
316
+ "language_model.model.layers.14.self_attn.v_proj",
317
+ "language_model.model.layers.14.self_attn.o_proj",
318
+ "language_model.model.layers.15.self_attn.q_proj",
319
+ "language_model.model.layers.15.self_attn.k_proj",
320
+ "language_model.model.layers.15.self_attn.v_proj",
321
+ "language_model.model.layers.15.self_attn.o_proj",
322
+ "language_model.model.layers.16.self_attn.q_proj",
323
+ "language_model.model.layers.16.self_attn.k_proj",
324
+ "language_model.model.layers.16.self_attn.v_proj",
325
+ "language_model.model.layers.16.self_attn.o_proj",
326
+ "language_model.model.layers.17.self_attn.q_proj",
327
+ "language_model.model.layers.17.self_attn.k_proj",
328
+ "language_model.model.layers.17.self_attn.v_proj",
329
+ "language_model.model.layers.17.self_attn.o_proj",
330
+ "language_model.model.layers.18.self_attn.q_proj",
331
+ "language_model.model.layers.18.self_attn.k_proj",
332
+ "language_model.model.layers.18.self_attn.v_proj",
333
+ "language_model.model.layers.18.self_attn.o_proj",
334
+ "language_model.model.layers.19.self_attn.q_proj",
335
+ "language_model.model.layers.19.self_attn.k_proj",
336
+ "language_model.model.layers.19.self_attn.v_proj",
337
+ "language_model.model.layers.19.self_attn.o_proj",
338
+ "language_model.model.layers.20.self_attn.q_proj",
339
+ "language_model.model.layers.20.self_attn.k_proj",
340
+ "language_model.model.layers.20.self_attn.v_proj",
341
+ "language_model.model.layers.20.self_attn.o_proj",
342
+ "language_model.model.layers.21.self_attn.q_proj",
343
+ "language_model.model.layers.21.self_attn.k_proj",
344
+ "language_model.model.layers.21.self_attn.v_proj",
345
+ "language_model.model.layers.21.self_attn.o_proj",
346
+ "language_model.model.layers.22.self_attn.q_proj",
347
+ "language_model.model.layers.22.self_attn.k_proj",
348
+ "language_model.model.layers.22.self_attn.v_proj",
349
+ "language_model.model.layers.22.self_attn.o_proj",
350
+ "language_model.model.layers.23.self_attn.q_proj",
351
+ "language_model.model.layers.23.self_attn.k_proj",
352
+ "language_model.model.layers.23.self_attn.v_proj",
353
+ "language_model.model.layers.23.self_attn.o_proj",
354
+ "language_model.model.layers.24.self_attn.q_proj",
355
+ "language_model.model.layers.24.self_attn.k_proj",
356
+ "language_model.model.layers.24.self_attn.v_proj",
357
+ "language_model.model.layers.24.self_attn.o_proj",
358
+ "language_model.model.layers.25.self_attn.q_proj",
359
+ "language_model.model.layers.25.self_attn.k_proj",
360
+ "language_model.model.layers.25.self_attn.v_proj",
361
+ "language_model.model.layers.25.self_attn.o_proj",
362
+ "language_model.model.layers.26.self_attn.q_proj",
363
+ "language_model.model.layers.26.self_attn.k_proj",
364
+ "language_model.model.layers.26.self_attn.v_proj",
365
+ "language_model.model.layers.26.self_attn.o_proj",
366
+ "language_model.model.layers.27.self_attn.q_proj",
367
+ "language_model.model.layers.27.self_attn.k_proj",
368
+ "language_model.model.layers.27.self_attn.v_proj",
369
+ "language_model.model.layers.27.self_attn.o_proj",
370
+ "language_model.model.layers.28.self_attn.q_proj",
371
+ "language_model.model.layers.28.self_attn.k_proj",
372
+ "language_model.model.layers.28.self_attn.v_proj",
373
+ "language_model.model.layers.28.self_attn.o_proj",
374
+ "language_model.model.layers.29.self_attn.q_proj",
375
+ "language_model.model.layers.29.self_attn.k_proj",
376
+ "language_model.model.layers.29.self_attn.v_proj",
377
+ "language_model.model.layers.29.self_attn.o_proj",
378
+ "language_model.model.layers.30.self_attn.q_proj",
379
+ "language_model.model.layers.30.self_attn.k_proj",
380
+ "language_model.model.layers.30.self_attn.v_proj",
381
+ "language_model.model.layers.30.self_attn.o_proj",
382
+ "language_model.model.layers.31.self_attn.q_proj",
383
+ "language_model.model.layers.31.self_attn.k_proj",
384
+ "language_model.model.layers.31.self_attn.v_proj",
385
+ "language_model.model.layers.31.self_attn.o_proj",
386
+ "language_model.model.layers.32.self_attn.q_proj",
387
+ "language_model.model.layers.32.self_attn.k_proj",
388
+ "language_model.model.layers.32.self_attn.v_proj",
389
+ "language_model.model.layers.32.self_attn.o_proj",
390
+ "language_model.model.layers.33.self_attn.q_proj",
391
+ "language_model.model.layers.33.self_attn.k_proj",
392
+ "language_model.model.layers.33.self_attn.v_proj",
393
+ "language_model.model.layers.33.self_attn.o_proj",
394
+ "language_model.model.layers.34.self_attn.q_proj",
395
+ "language_model.model.layers.34.self_attn.k_proj",
396
+ "language_model.model.layers.34.self_attn.v_proj",
397
+ "language_model.model.layers.34.self_attn.o_proj",
398
+ "language_model.model.layers.35.self_attn.q_proj",
399
+ "language_model.model.layers.35.self_attn.k_proj",
400
+ "language_model.model.layers.35.self_attn.v_proj",
401
+ "language_model.model.layers.35.self_attn.o_proj",
402
+ "language_model.model.layers.36.self_attn.q_proj",
403
+ "language_model.model.layers.36.self_attn.k_proj",
404
+ "language_model.model.layers.36.self_attn.v_proj",
405
+ "language_model.model.layers.36.self_attn.o_proj",
406
+ "language_model.model.layers.37.self_attn.q_proj",
407
+ "language_model.model.layers.37.self_attn.k_proj",
408
+ "language_model.model.layers.37.self_attn.v_proj",
409
+ "language_model.model.layers.37.self_attn.o_proj",
410
+ "language_model.model.layers.38.self_attn.q_proj",
411
+ "language_model.model.layers.38.self_attn.k_proj",
412
+ "language_model.model.layers.38.self_attn.v_proj",
413
+ "language_model.model.layers.38.self_attn.o_proj",
414
+ "language_model.model.layers.39.self_attn.q_proj",
415
+ "language_model.model.layers.39.self_attn.k_proj",
416
+ "language_model.model.layers.39.self_attn.v_proj",
417
+ "language_model.model.layers.39.self_attn.o_proj",
418
+ "language_model.model.layers.40.self_attn.q_proj",
419
+ "language_model.model.layers.40.self_attn.k_proj",
420
+ "language_model.model.layers.40.self_attn.v_proj",
421
+ "language_model.model.layers.40.self_attn.o_proj",
422
+ "language_model.model.layers.41.self_attn.q_proj",
423
+ "language_model.model.layers.41.self_attn.k_proj",
424
+ "language_model.model.layers.41.self_attn.v_proj",
425
+ "language_model.model.layers.41.self_attn.o_proj",
426
+ "language_model.model.layers.42.self_attn.q_proj",
427
+ "language_model.model.layers.42.self_attn.k_proj",
428
+ "language_model.model.layers.42.self_attn.v_proj",
429
+ "language_model.model.layers.42.self_attn.o_proj",
430
+ "language_model.model.layers.43.self_attn.q_proj",
431
+ "language_model.model.layers.43.self_attn.k_proj",
432
+ "language_model.model.layers.43.self_attn.v_proj",
433
+ "language_model.model.layers.43.self_attn.o_proj",
434
+ "language_model.model.layers.44.self_attn.q_proj",
435
+ "language_model.model.layers.44.self_attn.k_proj",
436
+ "language_model.model.layers.44.self_attn.v_proj",
437
+ "language_model.model.layers.44.self_attn.o_proj",
438
+ "language_model.model.layers.45.self_attn.q_proj",
439
+ "language_model.model.layers.45.self_attn.k_proj",
440
+ "language_model.model.layers.45.self_attn.v_proj",
441
+ "language_model.model.layers.45.self_attn.o_proj",
442
+ "language_model.model.layers.46.self_attn.q_proj",
443
+ "language_model.model.layers.46.self_attn.k_proj",
444
+ "language_model.model.layers.46.self_attn.v_proj",
445
+ "language_model.model.layers.46.self_attn.o_proj",
446
+ "language_model.model.layers.47.self_attn.q_proj",
447
+ "language_model.model.layers.47.self_attn.k_proj",
448
+ "language_model.model.layers.47.self_attn.v_proj",
449
+ "language_model.model.layers.47.self_attn.o_proj",
450
+ "language_model.lm_head"
451
+ ],
452
+ "kv_cache_scheme": null,
453
+ "quant_method": "compressed-tensors",
454
+ "quantization_status": "compressed",
455
+ "sparsity_config": {},
456
+ "transform_config": {},
457
+ "version": "0.12.3.dev15+g1c72e96"
458
+ },
459
+ "text_config": {
460
+ "_attn_implementation_autoset": true,
461
+ "attention_bias": false,
462
+ "attention_chunk_size": 8192,
463
+ "attention_dropout": 0.0,
464
+ "attn_scale": 0.1,
465
+ "attn_temperature_tuning": true,
466
+ "bos_token_id": 200000,
467
+ "dtype": "bfloat16",
468
+ "eos_token_id": [
469
+ 200001,
470
+ 200007,
471
+ 200008
472
+ ],
473
+ "floor_scale": 8192,
474
+ "for_llm_compressor": false,
475
+ "head_dim": 128,
476
+ "hidden_act": "silu",
477
+ "hidden_size": 5120,
478
+ "initializer_range": 0.02,
479
+ "interleave_moe_layer_step": 2,
480
+ "intermediate_size": 8192,
481
+ "intermediate_size_mlp": 16384,
482
+ "layer_types": [
483
+ "chunked_attention",
484
+ "chunked_attention",
485
+ "chunked_attention",
486
+ "full_attention",
487
+ "chunked_attention",
488
+ "chunked_attention",
489
+ "chunked_attention",
490
+ "full_attention",
491
+ "chunked_attention",
492
+ "chunked_attention",
493
+ "chunked_attention",
494
+ "full_attention",
495
+ "chunked_attention",
496
+ "chunked_attention",
497
+ "chunked_attention",
498
+ "full_attention",
499
+ "chunked_attention",
500
+ "chunked_attention",
501
+ "chunked_attention",
502
+ "full_attention",
503
+ "chunked_attention",
504
+ "chunked_attention",
505
+ "chunked_attention",
506
+ "full_attention",
507
+ "chunked_attention",
508
+ "chunked_attention",
509
+ "chunked_attention",
510
+ "full_attention",
511
+ "chunked_attention",
512
+ "chunked_attention",
513
+ "chunked_attention",
514
+ "full_attention",
515
+ "chunked_attention",
516
+ "chunked_attention",
517
+ "chunked_attention",
518
+ "full_attention",
519
+ "chunked_attention",
520
+ "chunked_attention",
521
+ "chunked_attention",
522
+ "full_attention",
523
+ "chunked_attention",
524
+ "chunked_attention",
525
+ "chunked_attention",
526
+ "full_attention",
527
+ "chunked_attention",
528
+ "chunked_attention",
529
+ "chunked_attention",
530
+ "full_attention"
531
+ ],
532
+ "max_position_embeddings": 1048576,
533
+ "model_type": "llama4_text",
534
+ "moe_layers": [
535
+ 1,
536
+ 3,
537
+ 5,
538
+ 7,
539
+ 9,
540
+ 11,
541
+ 13,
542
+ 15,
543
+ 17,
544
+ 19,
545
+ 21,
546
+ 23,
547
+ 25,
548
+ 27,
549
+ 29,
550
+ 31,
551
+ 33,
552
+ 35,
553
+ 37,
554
+ 39,
555
+ 41,
556
+ 43,
557
+ 45,
558
+ 47
559
+ ],
560
+ "no_rope_layers": [
561
+ 1,
562
+ 1,
563
+ 1,
564
+ 0,
565
+ 1,
566
+ 1,
567
+ 1,
568
+ 0,
569
+ 1,
570
+ 1,
571
+ 1,
572
+ 0,
573
+ 1,
574
+ 1,
575
+ 1,
576
+ 0,
577
+ 1,
578
+ 1,
579
+ 1,
580
+ 0,
581
+ 1,
582
+ 1,
583
+ 1,
584
+ 0,
585
+ 1,
586
+ 1,
587
+ 1,
588
+ 0,
589
+ 1,
590
+ 1,
591
+ 1,
592
+ 0,
593
+ 1,
594
+ 1,
595
+ 1,
596
+ 0,
597
+ 1,
598
+ 1,
599
+ 1,
600
+ 0,
601
+ 1,
602
+ 1,
603
+ 1,
604
+ 0,
605
+ 1,
606
+ 1,
607
+ 1,
608
+ 0
609
+ ],
610
+ "num_attention_heads": 40,
611
+ "num_experts_per_tok": 1,
612
+ "num_hidden_layers": 48,
613
+ "num_key_value_heads": 8,
614
+ "num_local_experts": 128,
615
+ "output_router_logits": false,
616
+ "pad_token_id": 200018,
617
+ "rms_norm_eps": 1e-05,
618
+ "rope_scaling": null,
619
+ "rope_theta": 500000.0,
620
+ "router_aux_loss_coef": 0.001,
621
+ "router_jitter_noise": 0.0,
622
+ "use_cache": true,
623
+ "use_qk_norm": false,
624
+ "vocab_size": 202048
625
+ },
626
+ "tie_word_embeddings": false,
627
+ "transformers_version": "4.57.1",
628
+ "vision_config": {
629
+ "_attn_implementation_autoset": true,
630
+ "_vision_feature_layer": -1,
631
+ "attention_dropout": 0.0,
632
+ "hidden_act": "gelu",
633
+ "hidden_size": 1408,
634
+ "image_size": 336,
635
+ "initializer_range": 0.02,
636
+ "intermediate_size": 5632,
637
+ "model_type": "llama4_vision_model",
638
+ "multi_modal_projector_bias": false,
639
+ "norm_eps": 1e-05,
640
+ "num_attention_heads": 16,
641
+ "num_channels": 3,
642
+ "num_hidden_layers": 34,
643
+ "patch_size": 14,
644
+ "pixel_shuffle_ratio": 0.5,
645
+ "projector_dropout": 0.0,
646
+ "projector_input_dim": 4096,
647
+ "projector_output_dim": 4096,
648
+ "rope_theta": 10000,
649
+ "vision_feature_layer": -1,
650
+ "vision_feature_select_strategy": "default",
651
+ "vision_output_dim": 4096
652
+ }
653
+ }