Hugging Face's logo

hf-internal-testing
/
tiny-processor-deepseek_ocr2

deepseek_ocr2
Model card Files
xet
 Community
hf-transformers-bot commited on
Commit
95c2139
·
verified ·
1 Parent(s): 783e55c

Add tiny processor fixture for testing

Browse files
Files changed (5) hide show
  1. config.json +182 -0
  2. generation_config.json +6 -0
  3. processor_config.json +37 -0
  4. tokenizer.json +0 -0
  5. tokenizer_config.json +0 -0
config.json ADDED
@@ -0,0 +1,182 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "DeepseekOcr2ForConditionalGeneration"
4
+ ],
5
+ "dtype": "bfloat16",
6
+ "image_token_id": 128815,
7
+ "model_type": "deepseek_ocr2",
8
+ "text_config": {
9
+ "attention_bias": false,
10
+ "attention_dropout": 0.0,
11
+ "bos_token_id": 0,
12
+ "dtype": "bfloat16",
13
+ "eos_token_id": 1,
14
+ "head_dim": 128,
15
+ "hidden_act": "silu",
16
+ "hidden_size": 1280,
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 6848,
19
+ "max_position_embeddings": 8192,
20
+ "mlp_bias": false,
21
+ "mlp_layer_types": [
22
+ "dense",
23
+ "sparse",
24
+ "sparse",
25
+ "sparse",
26
+ "sparse",
27
+ "sparse",
28
+ "sparse",
29
+ "sparse",
30
+ "sparse",
31
+ "sparse",
32
+ "sparse",
33
+ "sparse"
34
+ ],
35
+ "model_type": "deepseek_ocr2_text",
36
+ "moe_intermediate_size": 896,
37
+ "n_group": 1,
38
+ "n_routed_experts": 64,
39
+ "n_shared_experts": 2,
40
+ "num_attention_heads": 10,
41
+ "num_experts_per_tok": 6,
42
+ "num_hidden_layers": 12,
43
+ "num_key_value_heads": 10,
44
+ "pad_token_id": null,
45
+ "pretraining_tp": 1,
46
+ "rms_norm_eps": 1e-06,
47
+ "rope_parameters": {
48
+ "rope_theta": 10000.0,
49
+ "rope_type": "default"
50
+ },
51
+ "routed_scaling_factor": 1.0,
52
+ "tie_word_embeddings": false,
53
+ "topk_group": 1,
54
+ "topk_method": "greedy",
55
+ "use_cache": true,
56
+ "vocab_size": 129280
57
+ },
58
+ "tie_word_embeddings": false,
59
+ "transformers_version": "5.10.0.dev0",
60
+ "vision_config": {
61
+ "dtype": "bfloat16",
62
+ "encoder_config": {
63
+ "_name_or_path": "",
64
+ "architectures": null,
65
+ "attention_dropout": 0.0,
66
+ "bos_token_id": null,
67
+ "chunk_size_feed_forward": 0,
68
+ "dtype": null,
69
+ "eos_token_id": null,
70
+ "hidden_act": "silu",
71
+ "hidden_size": 896,
72
+ "id2label": {
73
+ "0": "LABEL_0",
74
+ "1": "LABEL_1"
75
+ },
76
+ "initializer_range": 0.02,
77
+ "intermediate_size": 4864,
78
+ "is_encoder_decoder": false,
79
+ "label2id": {
80
+ "LABEL_0": 0,
81
+ "LABEL_1": 1
82
+ },
83
+ "layer_types": [
84
+ "full_attention",
85
+ "full_attention",
86
+ "full_attention",
87
+ "full_attention",
88
+ "full_attention",
89
+ "full_attention",
90
+ "full_attention",
91
+ "full_attention",
92
+ "full_attention",
93
+ "full_attention",
94
+ "full_attention",
95
+ "full_attention",
96
+ "full_attention",
97
+ "full_attention",
98
+ "full_attention",
99
+ "full_attention",
100
+ "full_attention",
101
+ "full_attention",
102
+ "full_attention",
103
+ "full_attention",
104
+ "full_attention",
105
+ "full_attention",
106
+ "full_attention",
107
+ "full_attention"
108
+ ],
109
+ "max_position_embeddings": 32768,
110
+ "max_window_layers": 28,
111
+ "model_type": "deepseek_ocr2_encoder",
112
+ "num_attention_heads": 14,
113
+ "num_hidden_layers": 24,
114
+ "num_key_value_heads": 2,
115
+ "output_attentions": false,
116
+ "output_hidden_states": false,
117
+ "pad_token_id": null,
118
+ "problem_type": null,
119
+ "return_dict": true,
120
+ "rms_norm_eps": 1e-06,
121
+ "rope_parameters": {
122
+ "rope_theta": 1000000.0,
123
+ "rope_type": "default"
124
+ },
125
+ "sliding_window": null,
126
+ "tie_word_embeddings": false,
127
+ "use_cache": true,
128
+ "use_sliding_window": false,
129
+ "vocab_size": 1
130
+ },
131
+ "model_type": "deepseek_ocr2_vision",
132
+ "sam_config": {
133
+ "_name_or_path": "",
134
+ "architectures": null,
135
+ "attention_dropout": 0.0,
136
+ "chunk_size_feed_forward": 0,
137
+ "downsample_channels": [
138
+ 512,
139
+ 896
140
+ ],
141
+ "dtype": null,
142
+ "global_attn_indexes": [
143
+ 2,
144
+ 5,
145
+ 8,
146
+ 11
147
+ ],
148
+ "hidden_act": "gelu",
149
+ "hidden_size": 768,
150
+ "id2label": {
151
+ "0": "LABEL_0",
152
+ "1": "LABEL_1"
153
+ },
154
+ "image_size": 1024,
155
+ "initializer_range": 1e-10,
156
+ "is_encoder_decoder": false,
157
+ "label2id": {
158
+ "LABEL_0": 0,
159
+ "LABEL_1": 1
160
+ },
161
+ "layer_norm_eps": 1e-06,
162
+ "mlp_dim": 3072,
163
+ "mlp_ratio": 4.0,
164
+ "model_type": "deepseek_ocr2_sam_vision_model",
165
+ "num_attention_heads": 12,
166
+ "num_channels": 3,
167
+ "num_hidden_layers": 12,
168
+ "output_attentions": false,
169
+ "output_channels": 256,
170
+ "output_hidden_states": false,
171
+ "patch_size": 16,
172
+ "problem_type": null,
173
+ "qkv_bias": true,
174
+ "return_dict": true,
175
+ "scale": 384,
176
+ "use_abs_pos": true,
177
+ "use_rel_pos": true,
178
+ "window_size": 14
179
+ }
180
+ },
181
+ "vocab_size": 500
182
+ }
generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 0,
4
+ "eos_token_id": 1,
5
+ "transformers_version": "5.10.0.dev0"
6
+ }
processor_config.json ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "downsample_ratio": 4,
3
+ "image_processor": {
4
+ "background_color": [
5
+ 127,
6
+ 127,
7
+ 127
8
+ ],
9
+ "crop_to_patches": true,
10
+ "do_convert_rgb": true,
11
+ "do_normalize": true,
12
+ "do_rescale": true,
13
+ "do_resize": true,
14
+ "image_mean": [
15
+ 0.5,
16
+ 0.5,
17
+ 0.5
18
+ ],
19
+ "image_processor_type": "DeepseekOcr2ImageProcessor",
20
+ "image_std": [
21
+ 0.5,
22
+ 0.5,
23
+ 0.5
24
+ ],
25
+ "max_patches": 6,
26
+ "min_patches": 2,
27
+ "resample": 3,
28
+ "rescale_factor": 0.00392156862745098,
29
+ "size": {
30
+ "height": 1024,
31
+ "width": 1024
32
+ },
33
+ "tile_size": 768
34
+ },
35
+ "patch_size": 16,
36
+ "processor_class": "DeepseekOcr2Processor"
37
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
The diff for this file is too large to render. See raw diff