Hugging Face's logo

dicksonhk
/
docscopeOCR-7B-050425-exp-mlx-fp16

Image-Text-to-Text
Transformers
Safetensors
MLX
English
Chinese
qwen2_5_vl
image-to-text
OCR
Pdf
Doc
Image
text-generation-inference
KIE-Key Information Extraction
mlx-my-repo
conversational
Model card Files
xet
Community
dicksonhk commited on
Commit
656a3db
·
verified ·
1 Parent(s): 1243bc8

Upload config.json with huggingface_hub

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