Hugging Face's logo

nilay-samora
/
gemma3-12b-fp8

Safetensors
gemma3
compressed-tensors
Model card Files
xet
 Community
nilay-samora commited on
Commit
d5de6c3
·
verified ·
1 Parent(s): 736a457

Upload folder using huggingface_hub

Browse files
Files changed (14) hide show
  1. .gitattributes +1 -0
  2. chat_template.jinja +47 -0
  3. config.json +324 -0
  4. generation_config.json +13 -0
  5. model-00001-of-00003.safetensors +3 -0
  6. model-00002-of-00003.safetensors +3 -0
  7. model-00003-of-00003.safetensors +3 -0
  8. model.safetensors.index.json +0 -0
  9. preprocessor_config.json +29 -0
  10. processor_config.json +4 -0
  11. recipe.yaml +7 -0
  12. special_tokens_map.json +33 -0
  13. tokenizer.json +3 -0
  14. tokenizer_config.json +0 -0
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
chat_template.jinja ADDED
@@ -0,0 +1,47 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {{ bos_token }}
2
+ {%- if messages[0]['role'] == 'system' -%}
3
+ {%- if messages[0]['content'] is string -%}
4
+ {%- set first_user_prefix = messages[0]['content'] + '
5
+
6
+ ' -%}
7
+ {%- else -%}
8
+ {%- set first_user_prefix = messages[0]['content'][0]['text'] + '
9
+
10
+ ' -%}
11
+ {%- endif -%}
12
+ {%- set loop_messages = messages[1:] -%}
13
+ {%- else -%}
14
+ {%- set first_user_prefix = "" -%}
15
+ {%- set loop_messages = messages -%}
16
+ {%- endif -%}
17
+ {%- for message in loop_messages -%}
18
+ {%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}
19
+ {{ raise_exception("Conversation roles must alternate user/assistant/user/assistant/...") }}
20
+ {%- endif -%}
21
+ {%- if (message['role'] == 'assistant') -%}
22
+ {%- set role = "model" -%}
23
+ {%- else -%}
24
+ {%- set role = message['role'] -%}
25
+ {%- endif -%}
26
+ {{ '<start_of_turn>' + role + '
27
+ ' + (first_user_prefix if loop.first else "") }}
28
+ {%- if message['content'] is string -%}
29
+ {{ message['content'] | trim }}
30
+ {%- elif message['content'] is iterable -%}
31
+ {%- for item in message['content'] -%}
32
+ {%- if item['type'] == 'image' -%}
33
+ {{ '<start_of_image>' }}
34
+ {%- elif item['type'] == 'text' -%}
35
+ {{ item['text'] | trim }}
36
+ {%- endif -%}
37
+ {%- endfor -%}
38
+ {%- else -%}
39
+ {{ raise_exception("Invalid content type") }}
40
+ {%- endif -%}
41
+ {{ '<end_of_turn>
42
+ ' }}
43
+ {%- endfor -%}
44
+ {%- if add_generation_prompt -%}
45
+ {{'<start_of_turn>model
46
+ '}}
47
+ {%- endif -%}
config.json ADDED
@@ -0,0 +1,324 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Gemma3ForConditionalGeneration"
4
+ ],
5
+ "boi_token_index": 255999,
6
+ "dtype": "bfloat16",
7
+ "eoi_token_index": 256000,
8
+ "eos_token_id": [
9
+ 1,
10
+ 106
11
+ ],
12
+ "image_token_index": 262144,
13
+ "initializer_range": 0.02,
14
+ "mm_tokens_per_image": 256,
15
+ "model_type": "gemma3",
16
+ "quantization_config": {
17
+ "config_groups": {
18
+ "group_0": {
19
+ "format": "float-quantized",
20
+ "input_activations": {
21
+ "actorder": null,
22
+ "block_structure": null,
23
+ "dynamic": true,
24
+ "group_size": null,
25
+ "num_bits": 8,
26
+ "observer": null,
27
+ "observer_kwargs": {},
28
+ "scale_dtype": null,
29
+ "strategy": "token",
30
+ "symmetric": true,
31
+ "type": "float",
32
+ "zp_dtype": null
33
+ },
34
+ "output_activations": null,
35
+ "targets": [
36
+ "Linear"
37
+ ],
38
+ "weights": {
39
+ "actorder": null,
40
+ "block_structure": null,
41
+ "dynamic": false,
42
+ "group_size": null,
43
+ "num_bits": 8,
44
+ "observer": "memoryless_minmax",
45
+ "observer_kwargs": {},
46
+ "scale_dtype": null,
47
+ "strategy": "channel",
48
+ "symmetric": true,
49
+ "type": "float",
50
+ "zp_dtype": null
51
+ }
52
+ }
53
+ },
54
+ "format": "float-quantized",
55
+ "global_compression_ratio": null,
56
+ "ignore": [
57
+ "model.vision_tower.vision_model.encoder.layers.0.self_attn.k_proj",
58
+ "model.vision_tower.vision_model.encoder.layers.0.self_attn.v_proj",
59
+ "model.vision_tower.vision_model.encoder.layers.0.self_attn.q_proj",
60
+ "model.vision_tower.vision_model.encoder.layers.0.self_attn.out_proj",
61
+ "model.vision_tower.vision_model.encoder.layers.0.mlp.fc1",
62
+ "model.vision_tower.vision_model.encoder.layers.0.mlp.fc2",
63
+ "model.vision_tower.vision_model.encoder.layers.1.self_attn.k_proj",
64
+ "model.vision_tower.vision_model.encoder.layers.1.self_attn.v_proj",
65
+ "model.vision_tower.vision_model.encoder.layers.1.self_attn.q_proj",
66
+ "model.vision_tower.vision_model.encoder.layers.1.self_attn.out_proj",
67
+ "model.vision_tower.vision_model.encoder.layers.1.mlp.fc1",
68
+ "model.vision_tower.vision_model.encoder.layers.1.mlp.fc2",
69
+ "model.vision_tower.vision_model.encoder.layers.2.self_attn.k_proj",
70
+ "model.vision_tower.vision_model.encoder.layers.2.self_attn.v_proj",
71
+ "model.vision_tower.vision_model.encoder.layers.2.self_attn.q_proj",
72
+ "model.vision_tower.vision_model.encoder.layers.2.self_attn.out_proj",
73
+ "model.vision_tower.vision_model.encoder.layers.2.mlp.fc1",
74
+ "model.vision_tower.vision_model.encoder.layers.2.mlp.fc2",
75
+ "model.vision_tower.vision_model.encoder.layers.3.self_attn.k_proj",
76
+ "model.vision_tower.vision_model.encoder.layers.3.self_attn.v_proj",
77
+ "model.vision_tower.vision_model.encoder.layers.3.self_attn.q_proj",
78
+ "model.vision_tower.vision_model.encoder.layers.3.self_attn.out_proj",
79
+ "model.vision_tower.vision_model.encoder.layers.3.mlp.fc1",
80
+ "model.vision_tower.vision_model.encoder.layers.3.mlp.fc2",
81
+ "model.vision_tower.vision_model.encoder.layers.4.self_attn.k_proj",
82
+ "model.vision_tower.vision_model.encoder.layers.4.self_attn.v_proj",
83
+ "model.vision_tower.vision_model.encoder.layers.4.self_attn.q_proj",
84
+ "model.vision_tower.vision_model.encoder.layers.4.self_attn.out_proj",
85
+ "model.vision_tower.vision_model.encoder.layers.4.mlp.fc1",
86
+ "model.vision_tower.vision_model.encoder.layers.4.mlp.fc2",
87
+ "model.vision_tower.vision_model.encoder.layers.5.self_attn.k_proj",
88
+ "model.vision_tower.vision_model.encoder.layers.5.self_attn.v_proj",
89
+ "model.vision_tower.vision_model.encoder.layers.5.self_attn.q_proj",
90
+ "model.vision_tower.vision_model.encoder.layers.5.self_attn.out_proj",
91
+ "model.vision_tower.vision_model.encoder.layers.5.mlp.fc1",
92
+ "model.vision_tower.vision_model.encoder.layers.5.mlp.fc2",
93
+ "model.vision_tower.vision_model.encoder.layers.6.self_attn.k_proj",
94
+ "model.vision_tower.vision_model.encoder.layers.6.self_attn.v_proj",
95
+ "model.vision_tower.vision_model.encoder.layers.6.self_attn.q_proj",
96
+ "model.vision_tower.vision_model.encoder.layers.6.self_attn.out_proj",
97
+ "model.vision_tower.vision_model.encoder.layers.6.mlp.fc1",
98
+ "model.vision_tower.vision_model.encoder.layers.6.mlp.fc2",
99
+ "model.vision_tower.vision_model.encoder.layers.7.self_attn.k_proj",
100
+ "model.vision_tower.vision_model.encoder.layers.7.self_attn.v_proj",
101
+ "model.vision_tower.vision_model.encoder.layers.7.self_attn.q_proj",
102
+ "model.vision_tower.vision_model.encoder.layers.7.self_attn.out_proj",
103
+ "model.vision_tower.vision_model.encoder.layers.7.mlp.fc1",
104
+ "model.vision_tower.vision_model.encoder.layers.7.mlp.fc2",
105
+ "model.vision_tower.vision_model.encoder.layers.8.self_attn.k_proj",
106
+ "model.vision_tower.vision_model.encoder.layers.8.self_attn.v_proj",
107
+ "model.vision_tower.vision_model.encoder.layers.8.self_attn.q_proj",
108
+ "model.vision_tower.vision_model.encoder.layers.8.self_attn.out_proj",
109
+ "model.vision_tower.vision_model.encoder.layers.8.mlp.fc1",
110
+ "model.vision_tower.vision_model.encoder.layers.8.mlp.fc2",
111
+ "model.vision_tower.vision_model.encoder.layers.9.self_attn.k_proj",
112
+ "model.vision_tower.vision_model.encoder.layers.9.self_attn.v_proj",
113
+ "model.vision_tower.vision_model.encoder.layers.9.self_attn.q_proj",
114
+ "model.vision_tower.vision_model.encoder.layers.9.self_attn.out_proj",
115
+ "model.vision_tower.vision_model.encoder.layers.9.mlp.fc1",
116
+ "model.vision_tower.vision_model.encoder.layers.9.mlp.fc2",
117
+ "model.vision_tower.vision_model.encoder.layers.10.self_attn.k_proj",
118
+ "model.vision_tower.vision_model.encoder.layers.10.self_attn.v_proj",
119
+ "model.vision_tower.vision_model.encoder.layers.10.self_attn.q_proj",
120
+ "model.vision_tower.vision_model.encoder.layers.10.self_attn.out_proj",
121
+ "model.vision_tower.vision_model.encoder.layers.10.mlp.fc1",
122
+ "model.vision_tower.vision_model.encoder.layers.10.mlp.fc2",
123
+ "model.vision_tower.vision_model.encoder.layers.11.self_attn.k_proj",
124
+ "model.vision_tower.vision_model.encoder.layers.11.self_attn.v_proj",
125
+ "model.vision_tower.vision_model.encoder.layers.11.self_attn.q_proj",
126
+ "model.vision_tower.vision_model.encoder.layers.11.self_attn.out_proj",
127
+ "model.vision_tower.vision_model.encoder.layers.11.mlp.fc1",
128
+ "model.vision_tower.vision_model.encoder.layers.11.mlp.fc2",
129
+ "model.vision_tower.vision_model.encoder.layers.12.self_attn.k_proj",
130
+ "model.vision_tower.vision_model.encoder.layers.12.self_attn.v_proj",
131
+ "model.vision_tower.vision_model.encoder.layers.12.self_attn.q_proj",
132
+ "model.vision_tower.vision_model.encoder.layers.12.self_attn.out_proj",
133
+ "model.vision_tower.vision_model.encoder.layers.12.mlp.fc1",
134
+ "model.vision_tower.vision_model.encoder.layers.12.mlp.fc2",
135
+ "model.vision_tower.vision_model.encoder.layers.13.self_attn.k_proj",
136
+ "model.vision_tower.vision_model.encoder.layers.13.self_attn.v_proj",
137
+ "model.vision_tower.vision_model.encoder.layers.13.self_attn.q_proj",
138
+ "model.vision_tower.vision_model.encoder.layers.13.self_attn.out_proj",
139
+ "model.vision_tower.vision_model.encoder.layers.13.mlp.fc1",
140
+ "model.vision_tower.vision_model.encoder.layers.13.mlp.fc2",
141
+ "model.vision_tower.vision_model.encoder.layers.14.self_attn.k_proj",
142
+ "model.vision_tower.vision_model.encoder.layers.14.self_attn.v_proj",
143
+ "model.vision_tower.vision_model.encoder.layers.14.self_attn.q_proj",
144
+ "model.vision_tower.vision_model.encoder.layers.14.self_attn.out_proj",
145
+ "model.vision_tower.vision_model.encoder.layers.14.mlp.fc1",
146
+ "model.vision_tower.vision_model.encoder.layers.14.mlp.fc2",
147
+ "model.vision_tower.vision_model.encoder.layers.15.self_attn.k_proj",
148
+ "model.vision_tower.vision_model.encoder.layers.15.self_attn.v_proj",
149
+ "model.vision_tower.vision_model.encoder.layers.15.self_attn.q_proj",
150
+ "model.vision_tower.vision_model.encoder.layers.15.self_attn.out_proj",
151
+ "model.vision_tower.vision_model.encoder.layers.15.mlp.fc1",
152
+ "model.vision_tower.vision_model.encoder.layers.15.mlp.fc2",
153
+ "model.vision_tower.vision_model.encoder.layers.16.self_attn.k_proj",
154
+ "model.vision_tower.vision_model.encoder.layers.16.self_attn.v_proj",
155
+ "model.vision_tower.vision_model.encoder.layers.16.self_attn.q_proj",
156
+ "model.vision_tower.vision_model.encoder.layers.16.self_attn.out_proj",
157
+ "model.vision_tower.vision_model.encoder.layers.16.mlp.fc1",
158
+ "model.vision_tower.vision_model.encoder.layers.16.mlp.fc2",
159
+ "model.vision_tower.vision_model.encoder.layers.17.self_attn.k_proj",
160
+ "model.vision_tower.vision_model.encoder.layers.17.self_attn.v_proj",
161
+ "model.vision_tower.vision_model.encoder.layers.17.self_attn.q_proj",
162
+ "model.vision_tower.vision_model.encoder.layers.17.self_attn.out_proj",
163
+ "model.vision_tower.vision_model.encoder.layers.17.mlp.fc1",
164
+ "model.vision_tower.vision_model.encoder.layers.17.mlp.fc2",
165
+ "model.vision_tower.vision_model.encoder.layers.18.self_attn.k_proj",
166
+ "model.vision_tower.vision_model.encoder.layers.18.self_attn.v_proj",
167
+ "model.vision_tower.vision_model.encoder.layers.18.self_attn.q_proj",
168
+ "model.vision_tower.vision_model.encoder.layers.18.self_attn.out_proj",
169
+ "model.vision_tower.vision_model.encoder.layers.18.mlp.fc1",
170
+ "model.vision_tower.vision_model.encoder.layers.18.mlp.fc2",
171
+ "model.vision_tower.vision_model.encoder.layers.19.self_attn.k_proj",
172
+ "model.vision_tower.vision_model.encoder.layers.19.self_attn.v_proj",
173
+ "model.vision_tower.vision_model.encoder.layers.19.self_attn.q_proj",
174
+ "model.vision_tower.vision_model.encoder.layers.19.self_attn.out_proj",
175
+ "model.vision_tower.vision_model.encoder.layers.19.mlp.fc1",
176
+ "model.vision_tower.vision_model.encoder.layers.19.mlp.fc2",
177
+ "model.vision_tower.vision_model.encoder.layers.20.self_attn.k_proj",
178
+ "model.vision_tower.vision_model.encoder.layers.20.self_attn.v_proj",
179
+ "model.vision_tower.vision_model.encoder.layers.20.self_attn.q_proj",
180
+ "model.vision_tower.vision_model.encoder.layers.20.self_attn.out_proj",
181
+ "model.vision_tower.vision_model.encoder.layers.20.mlp.fc1",
182
+ "model.vision_tower.vision_model.encoder.layers.20.mlp.fc2",
183
+ "model.vision_tower.vision_model.encoder.layers.21.self_attn.k_proj",
184
+ "model.vision_tower.vision_model.encoder.layers.21.self_attn.v_proj",
185
+ "model.vision_tower.vision_model.encoder.layers.21.self_attn.q_proj",
186
+ "model.vision_tower.vision_model.encoder.layers.21.self_attn.out_proj",
187
+ "model.vision_tower.vision_model.encoder.layers.21.mlp.fc1",
188
+ "model.vision_tower.vision_model.encoder.layers.21.mlp.fc2",
189
+ "model.vision_tower.vision_model.encoder.layers.22.self_attn.k_proj",
190
+ "model.vision_tower.vision_model.encoder.layers.22.self_attn.v_proj",
191
+ "model.vision_tower.vision_model.encoder.layers.22.self_attn.q_proj",
192
+ "model.vision_tower.vision_model.encoder.layers.22.self_attn.out_proj",
193
+ "model.vision_tower.vision_model.encoder.layers.22.mlp.fc1",
194
+ "model.vision_tower.vision_model.encoder.layers.22.mlp.fc2",
195
+ "model.vision_tower.vision_model.encoder.layers.23.self_attn.k_proj",
196
+ "model.vision_tower.vision_model.encoder.layers.23.self_attn.v_proj",
197
+ "model.vision_tower.vision_model.encoder.layers.23.self_attn.q_proj",
198
+ "model.vision_tower.vision_model.encoder.layers.23.self_attn.out_proj",
199
+ "model.vision_tower.vision_model.encoder.layers.23.mlp.fc1",
200
+ "model.vision_tower.vision_model.encoder.layers.23.mlp.fc2",
201
+ "model.vision_tower.vision_model.encoder.layers.24.self_attn.k_proj",
202
+ "model.vision_tower.vision_model.encoder.layers.24.self_attn.v_proj",
203
+ "model.vision_tower.vision_model.encoder.layers.24.self_attn.q_proj",
204
+ "model.vision_tower.vision_model.encoder.layers.24.self_attn.out_proj",
205
+ "model.vision_tower.vision_model.encoder.layers.24.mlp.fc1",
206
+ "model.vision_tower.vision_model.encoder.layers.24.mlp.fc2",
207
+ "model.vision_tower.vision_model.encoder.layers.25.self_attn.k_proj",
208
+ "model.vision_tower.vision_model.encoder.layers.25.self_attn.v_proj",
209
+ "model.vision_tower.vision_model.encoder.layers.25.self_attn.q_proj",
210
+ "model.vision_tower.vision_model.encoder.layers.25.self_attn.out_proj",
211
+ "model.vision_tower.vision_model.encoder.layers.25.mlp.fc1",
212
+ "model.vision_tower.vision_model.encoder.layers.25.mlp.fc2",
213
+ "model.vision_tower.vision_model.encoder.layers.26.self_attn.k_proj",
214
+ "model.vision_tower.vision_model.encoder.layers.26.self_attn.v_proj",
215
+ "model.vision_tower.vision_model.encoder.layers.26.self_attn.q_proj",
216
+ "model.vision_tower.vision_model.encoder.layers.26.self_attn.out_proj",
217
+ "model.vision_tower.vision_model.encoder.layers.26.mlp.fc1",
218
+ "model.vision_tower.vision_model.encoder.layers.26.mlp.fc2",
219
+ "lm_head"
220
+ ],
221
+ "kv_cache_scheme": null,
222
+ "quant_method": "compressed-tensors",
223
+ "quantization_status": "compressed",
224
+ "sparsity_config": {},
225
+ "transform_config": {},
226
+ "version": "0.14.0.1"
227
+ },
228
+ "text_config": {
229
+ "_sliding_window_pattern": 6,
230
+ "attention_bias": false,
231
+ "attention_dropout": 0.0,
232
+ "attn_logit_softcapping": null,
233
+ "dtype": "bfloat16",
234
+ "final_logit_softcapping": null,
235
+ "head_dim": 256,
236
+ "hidden_activation": "gelu_pytorch_tanh",
237
+ "hidden_size": 3840,
238
+ "initializer_range": 0.02,
239
+ "intermediate_size": 15360,
240
+ "layer_types": [
241
+ "sliding_attention",
242
+ "sliding_attention",
243
+ "sliding_attention",
244
+ "sliding_attention",
245
+ "sliding_attention",
246
+ "full_attention",
247
+ "sliding_attention",
248
+ "sliding_attention",
249
+ "sliding_attention",
250
+ "sliding_attention",
251
+ "sliding_attention",
252
+ "full_attention",
253
+ "sliding_attention",
254
+ "sliding_attention",
255
+ "sliding_attention",
256
+ "sliding_attention",
257
+ "sliding_attention",
258
+ "full_attention",
259
+ "sliding_attention",
260
+ "sliding_attention",
261
+ "sliding_attention",
262
+ "sliding_attention",
263
+ "sliding_attention",
264
+ "full_attention",
265
+ "sliding_attention",
266
+ "sliding_attention",
267
+ "sliding_attention",
268
+ "sliding_attention",
269
+ "sliding_attention",
270
+ "full_attention",
271
+ "sliding_attention",
272
+ "sliding_attention",
273
+ "sliding_attention",
274
+ "sliding_attention",
275
+ "sliding_attention",
276
+ "full_attention",
277
+ "sliding_attention",
278
+ "sliding_attention",
279
+ "sliding_attention",
280
+ "sliding_attention",
281
+ "sliding_attention",
282
+ "full_attention",
283
+ "sliding_attention",
284
+ "sliding_attention",
285
+ "sliding_attention",
286
+ "sliding_attention",
287
+ "sliding_attention",
288
+ "full_attention"
289
+ ],
290
+ "max_position_embeddings": 131072,
291
+ "model_type": "gemma3_text",
292
+ "num_attention_heads": 16,
293
+ "num_hidden_layers": 48,
294
+ "num_key_value_heads": 8,
295
+ "query_pre_attn_scalar": 256,
296
+ "rms_norm_eps": 1e-06,
297
+ "rope_local_base_freq": 10000.0,
298
+ "rope_scaling": {
299
+ "factor": 8.0,
300
+ "rope_type": "linear"
301
+ },
302
+ "rope_theta": 1000000.0,
303
+ "sliding_window": 1024,
304
+ "use_bidirectional_attention": false,
305
+ "use_cache": true,
306
+ "vocab_size": 262208
307
+ },
308
+ "transformers_version": "4.57.6",
309
+ "vision_config": {
310
+ "attention_dropout": 0.0,
311
+ "dtype": "bfloat16",
312
+ "hidden_act": "gelu_pytorch_tanh",
313
+ "hidden_size": 1152,
314
+ "image_size": 896,
315
+ "intermediate_size": 4304,
316
+ "layer_norm_eps": 1e-06,
317
+ "model_type": "siglip_vision_model",
318
+ "num_attention_heads": 16,
319
+ "num_channels": 3,
320
+ "num_hidden_layers": 27,
321
+ "patch_size": 14,
322
+ "vision_use_head": false
323
+ }
324
+ }
generation_config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 2,
3
+ "cache_implementation": "hybrid",
4
+ "do_sample": true,
5
+ "eos_token_id": [
6
+ 1,
7
+ 106
8
+ ],
9
+ "pad_token_id": 0,
10
+ "top_k": 64,
11
+ "top_p": 0.95,
12
+ "transformers_version": "4.57.6"
13
+ }
model-00001-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9adf89a6db687e4ae1f99bf44c5c1124718a61a32ffa65497907fb50b0cdd6fd
3
+ size 4980972040
model-00002-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:325f20f7208a7e49302601a6fe1d3a872a6e0efdf6e8581784a8438ce82b1ac7
3
+ size 4992747400
model-00003-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:00ed2d526bf65bc43f812ec8eb4533023c51678cc1e3e38d81757dc39e152d0d
3
+ size 3647199496
model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
preprocessor_config.json ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_convert_rgb": null,
3
+ "do_normalize": true,
4
+ "do_pan_and_scan": null,
5
+ "do_rescale": true,
6
+ "do_resize": true,
7
+ "image_mean": [
8
+ 0.5,
9
+ 0.5,
10
+ 0.5
11
+ ],
12
+ "image_processor_type": "Gemma3ImageProcessor",
13
+ "image_seq_length": 256,
14
+ "image_std": [
15
+ 0.5,
16
+ 0.5,
17
+ 0.5
18
+ ],
19
+ "pan_and_scan_max_num_crops": null,
20
+ "pan_and_scan_min_crop_size": null,
21
+ "pan_and_scan_min_ratio_to_activate": null,
22
+ "processor_class": "Gemma3Processor",
23
+ "resample": 2,
24
+ "rescale_factor": 0.00392156862745098,
25
+ "size": {
26
+ "height": 896,
27
+ "width": 896
28
+ }
29
+ }
processor_config.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "image_seq_length": 256,
3
+ "processor_class": "Gemma3Processor"
4
+ }
recipe.yaml ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ default_stage:
2
+ default_modifiers:
3
+ QuantizationModifier:
4
+ targets: [Linear]
5
+ ignore: ['re:.*lm_head', 're:.*vision_tower.*', 're:.*multi_modal_projector.*']
6
+ scheme: FP8_DYNAMIC
7
+ bypass_divisibility_checks: false
special_tokens_map.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "boi_token": "<start_of_image>",
3
+ "bos_token": {
4
+ "content": "<bos>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false
9
+ },
10
+ "eoi_token": "<end_of_image>",
11
+ "eos_token": {
12
+ "content": "<eos>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false
17
+ },
18
+ "image_token": "<image_soft_token>",
19
+ "pad_token": {
20
+ "content": "<pad>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false
25
+ },
26
+ "unk_token": {
27
+ "content": "<unk>",
28
+ "lstrip": false,
29
+ "normalized": false,
30
+ "rstrip": false,
31
+ "single_word": false
32
+ }
33
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4667f2089529e8e7657cfb6d1c19910ae71ff5f28aa7ab2ff2763330affad795
3
+ size 33384568
tokenizer_config.json ADDED
The diff for this file is too large to render. See raw diff