dicksonhk commited on
Commit
40af68e
·
verified ·
1 Parent(s): acfdc9d

Upload config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. config.json +278 -0
config.json ADDED
@@ -0,0 +1,278 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_cross_attention": false,
3
+ "architectures": [
4
+ "Qwen2_5_VLForConditionalGeneration"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "bad_words_ids": null,
8
+ "begin_suppress_tokens": null,
9
+ "bos_token_id": 151643,
10
+ "chunk_size_feed_forward": 0,
11
+ "cross_attention_hidden_size": null,
12
+ "decoder_start_token_id": null,
13
+ "diversity_penalty": 0.0,
14
+ "do_sample": false,
15
+ "early_stopping": false,
16
+ "encoder_no_repeat_ngram_size": 0,
17
+ "eos_token_id": 151645,
18
+ "exponential_decay_length_penalty": null,
19
+ "finetuning_task": null,
20
+ "forced_bos_token_id": null,
21
+ "forced_eos_token_id": null,
22
+ "hidden_act": "silu",
23
+ "hidden_size": 3584,
24
+ "id2label": {
25
+ "0": "LABEL_0",
26
+ "1": "LABEL_1"
27
+ },
28
+ "image_token_id": 151655,
29
+ "initializer_range": 0.02,
30
+ "intermediate_size": 18944,
31
+ "is_decoder": false,
32
+ "is_encoder_decoder": false,
33
+ "label2id": {
34
+ "LABEL_0": 0,
35
+ "LABEL_1": 1
36
+ },
37
+ "length_penalty": 1.0,
38
+ "max_length": 20,
39
+ "max_position_embeddings": 128000,
40
+ "max_window_layers": 28,
41
+ "min_length": 0,
42
+ "model_type": "qwen2_5_vl",
43
+ "no_repeat_ngram_size": 0,
44
+ "num_attention_heads": 28,
45
+ "num_beam_groups": 1,
46
+ "num_beams": 1,
47
+ "num_hidden_layers": 28,
48
+ "num_key_value_heads": 4,
49
+ "num_return_sequences": 1,
50
+ "output_attentions": false,
51
+ "output_hidden_states": false,
52
+ "output_scores": false,
53
+ "pad_token_id": null,
54
+ "prefix": null,
55
+ "problem_type": null,
56
+ "pruned_heads": {},
57
+ "quantization": {
58
+ "group_size": 64,
59
+ "bits": 8
60
+ },
61
+ "remove_invalid_values": false,
62
+ "repetition_penalty": 1.0,
63
+ "return_dict": true,
64
+ "return_dict_in_generate": false,
65
+ "rms_norm_eps": 1e-06,
66
+ "rope_scaling": {
67
+ "mrope_section": [
68
+ 16,
69
+ 24,
70
+ 24
71
+ ],
72
+ "rope_type": "default",
73
+ "type": "default"
74
+ },
75
+ "rope_theta": 1000000.0,
76
+ "sep_token_id": null,
77
+ "sliding_window": 32768,
78
+ "suppress_tokens": null,
79
+ "task_specific_params": null,
80
+ "temperature": 1.0,
81
+ "text_config": {
82
+ "vocab_size": 152064,
83
+ "max_position_embeddings": 128000,
84
+ "hidden_size": 3584,
85
+ "intermediate_size": 18944,
86
+ "num_hidden_layers": 28,
87
+ "num_attention_heads": 28,
88
+ "use_sliding_window": false,
89
+ "sliding_window": 32768,
90
+ "max_window_layers": 28,
91
+ "num_key_value_heads": 4,
92
+ "hidden_act": "silu",
93
+ "initializer_range": 0.02,
94
+ "rms_norm_eps": 1e-06,
95
+ "use_cache": false,
96
+ "rope_theta": 1000000.0,
97
+ "attention_dropout": 0.0,
98
+ "rope_scaling": {
99
+ "mrope_section": [
100
+ 16,
101
+ 24,
102
+ 24
103
+ ],
104
+ "rope_type": "default",
105
+ "type": "default"
106
+ },
107
+ "image_token_id": null,
108
+ "video_token_id": null,
109
+ "return_dict": true,
110
+ "output_hidden_states": false,
111
+ "output_attentions": false,
112
+ "torchscript": false,
113
+ "torch_dtype": "bfloat16",
114
+ "use_bfloat16": false,
115
+ "tf_legacy_loss": false,
116
+ "pruned_heads": {},
117
+ "tie_word_embeddings": false,
118
+ "chunk_size_feed_forward": 0,
119
+ "is_encoder_decoder": false,
120
+ "is_decoder": false,
121
+ "cross_attention_hidden_size": null,
122
+ "add_cross_attention": false,
123
+ "tie_encoder_decoder": false,
124
+ "max_length": 20,
125
+ "min_length": 0,
126
+ "do_sample": false,
127
+ "early_stopping": false,
128
+ "num_beams": 1,
129
+ "num_beam_groups": 1,
130
+ "diversity_penalty": 0.0,
131
+ "temperature": 1.0,
132
+ "top_k": 50,
133
+ "top_p": 1.0,
134
+ "typical_p": 1.0,
135
+ "repetition_penalty": 1.0,
136
+ "length_penalty": 1.0,
137
+ "no_repeat_ngram_size": 0,
138
+ "encoder_no_repeat_ngram_size": 0,
139
+ "bad_words_ids": null,
140
+ "num_return_sequences": 1,
141
+ "output_scores": false,
142
+ "return_dict_in_generate": false,
143
+ "forced_bos_token_id": null,
144
+ "forced_eos_token_id": null,
145
+ "remove_invalid_values": false,
146
+ "exponential_decay_length_penalty": null,
147
+ "suppress_tokens": null,
148
+ "begin_suppress_tokens": null,
149
+ "architectures": [
150
+ "Qwen2_5_VLForConditionalGeneration"
151
+ ],
152
+ "finetuning_task": null,
153
+ "id2label": {
154
+ "0": "LABEL_0",
155
+ "1": "LABEL_1"
156
+ },
157
+ "label2id": {
158
+ "LABEL_0": 0,
159
+ "LABEL_1": 1
160
+ },
161
+ "tokenizer_class": null,
162
+ "prefix": null,
163
+ "bos_token_id": 151643,
164
+ "pad_token_id": null,
165
+ "eos_token_id": 151645,
166
+ "sep_token_id": null,
167
+ "decoder_start_token_id": null,
168
+ "task_specific_params": null,
169
+ "problem_type": null,
170
+ "_name_or_path": "",
171
+ "model_type": "qwen2_5_vl_text",
172
+ "vision_end_token_id": 151653,
173
+ "vision_start_token_id": 151652,
174
+ "vision_token_id": 151654
175
+ },
176
+ "tf_legacy_loss": false,
177
+ "tie_encoder_decoder": false,
178
+ "tie_word_embeddings": false,
179
+ "tokenizer_class": null,
180
+ "top_k": 50,
181
+ "top_p": 1.0,
182
+ "torch_dtype": "bfloat16",
183
+ "torchscript": false,
184
+ "transformers_version": "4.52.4",
185
+ "typical_p": 1.0,
186
+ "use_bfloat16": false,
187
+ "use_cache": false,
188
+ "use_sliding_window": false,
189
+ "video_token_id": 151656,
190
+ "vision_config": {
191
+ "return_dict": true,
192
+ "output_hidden_states": false,
193
+ "output_attentions": false,
194
+ "torchscript": false,
195
+ "torch_dtype": "float32",
196
+ "use_bfloat16": false,
197
+ "tf_legacy_loss": false,
198
+ "pruned_heads": {},
199
+ "tie_word_embeddings": true,
200
+ "chunk_size_feed_forward": 0,
201
+ "is_encoder_decoder": false,
202
+ "is_decoder": false,
203
+ "cross_attention_hidden_size": null,
204
+ "add_cross_attention": false,
205
+ "tie_encoder_decoder": false,
206
+ "max_length": 20,
207
+ "min_length": 0,
208
+ "do_sample": false,
209
+ "early_stopping": false,
210
+ "num_beams": 1,
211
+ "num_beam_groups": 1,
212
+ "diversity_penalty": 0.0,
213
+ "temperature": 1.0,
214
+ "top_k": 50,
215
+ "top_p": 1.0,
216
+ "typical_p": 1.0,
217
+ "repetition_penalty": 1.0,
218
+ "length_penalty": 1.0,
219
+ "no_repeat_ngram_size": 0,
220
+ "encoder_no_repeat_ngram_size": 0,
221
+ "bad_words_ids": null,
222
+ "num_return_sequences": 1,
223
+ "output_scores": false,
224
+ "return_dict_in_generate": false,
225
+ "forced_bos_token_id": null,
226
+ "forced_eos_token_id": null,
227
+ "remove_invalid_values": false,
228
+ "exponential_decay_length_penalty": null,
229
+ "suppress_tokens": null,
230
+ "begin_suppress_tokens": null,
231
+ "architectures": null,
232
+ "finetuning_task": null,
233
+ "id2label": {
234
+ "0": "LABEL_0",
235
+ "1": "LABEL_1"
236
+ },
237
+ "label2id": {
238
+ "LABEL_0": 0,
239
+ "LABEL_1": 1
240
+ },
241
+ "tokenizer_class": null,
242
+ "prefix": null,
243
+ "bos_token_id": null,
244
+ "pad_token_id": null,
245
+ "eos_token_id": null,
246
+ "sep_token_id": null,
247
+ "decoder_start_token_id": null,
248
+ "task_specific_params": null,
249
+ "problem_type": null,
250
+ "_name_or_path": "",
251
+ "in_chans": 3,
252
+ "model_type": "qwen2_5_vl",
253
+ "spatial_patch_size": 14,
254
+ "depth": 32,
255
+ "hidden_size": 1280,
256
+ "hidden_act": "silu",
257
+ "intermediate_size": 3420,
258
+ "num_heads": 16,
259
+ "in_channels": 3,
260
+ "patch_size": 14,
261
+ "spatial_merge_size": 2,
262
+ "temporal_patch_size": 2,
263
+ "tokens_per_second": 2,
264
+ "window_size": 112,
265
+ "fullatt_block_indexes": [
266
+ 7,
267
+ 15,
268
+ 23,
269
+ 31
270
+ ],
271
+ "out_hidden_size": 3584,
272
+ "initializer_range": 0.02
273
+ },
274
+ "vision_end_token_id": 151653,
275
+ "vision_start_token_id": 151652,
276
+ "vision_token_id": 151654,
277
+ "vocab_size": 152064
278
+ }