FILM6912 commited on
Commit
cdf0486
·
verified ·
1 Parent(s): f02c788

Upload folder using huggingface_hub

Browse files
Files changed (1) hide show
  1. config.json +5 -47
config.json CHANGED
@@ -3,7 +3,6 @@
3
  "Qwen2_5_VLForConditionalGeneration"
4
  ],
5
  "attention_dropout": 0.0,
6
- "bos_token_id": 151643,
7
  "eos_token_id": 151645,
8
  "hidden_act": "silu",
9
  "hidden_size": 2048,
@@ -16,7 +15,7 @@
16
  "num_attention_heads": 16,
17
  "num_hidden_layers": 36,
18
  "num_key_value_heads": 2,
19
- "pad_token_id": 151643,
20
  "rms_norm_eps": 1e-06,
21
  "rope_scaling": {
22
  "mrope_section": [
@@ -29,50 +28,10 @@
29
  },
30
  "rope_theta": 1000000.0,
31
  "sliding_window": 32768,
32
- "text_config": {
33
- "_name_or_path": "/data/workspace/surapon/olmocr/local_folder14/checkpoint-8000",
34
- "architectures": [
35
- "Qwen2_5_VLForConditionalGeneration"
36
- ],
37
- "attention_dropout": 0.0,
38
- "bos_token_id": 151643,
39
- "eos_token_id": 151645,
40
- "hidden_act": "silu",
41
- "hidden_size": 2048,
42
- "image_token_id": null,
43
- "initializer_range": 0.02,
44
- "intermediate_size": 11008,
45
- "max_position_embeddings": 128000,
46
- "max_window_layers": 70,
47
- "model_type": "qwen2_5_vl_text",
48
- "num_attention_heads": 16,
49
- "num_hidden_layers": 36,
50
- "num_key_value_heads": 2,
51
- "rms_norm_eps": 1e-06,
52
- "rope_scaling": {
53
- "mrope_section": [
54
- 16,
55
- 24,
56
- 24
57
- ],
58
- "rope_type": "default",
59
- "type": "default"
60
- },
61
- "rope_theta": 1000000.0,
62
- "sliding_window": 32768,
63
- "tie_word_embeddings": true,
64
- "torch_dtype": "bfloat16",
65
- "use_cache": true,
66
- "use_sliding_window": false,
67
- "video_token_id": null,
68
- "vision_end_token_id": 151653,
69
- "vision_start_token_id": 151652,
70
- "vision_token_id": 151654,
71
- "vocab_size": 151936
72
- },
73
  "torch_dtype": "bfloat16",
74
- "transformers_version": "4.52.4",
75
- "unsloth_version": "2025.6.5",
76
  "use_cache": true,
77
  "use_sliding_window": false,
78
  "video_token_id": 151656,
@@ -88,7 +47,6 @@
88
  "hidden_size": 1280,
89
  "in_channels": 3,
90
  "in_chans": 3,
91
- "initializer_range": 0.02,
92
  "intermediate_size": 3420,
93
  "model_type": "qwen2_5_vl",
94
  "num_heads": 16,
@@ -105,4 +63,4 @@
105
  "vision_start_token_id": 151652,
106
  "vision_token_id": 151654,
107
  "vocab_size": 151936
108
- }
 
3
  "Qwen2_5_VLForConditionalGeneration"
4
  ],
5
  "attention_dropout": 0.0,
 
6
  "eos_token_id": 151645,
7
  "hidden_act": "silu",
8
  "hidden_size": 2048,
 
15
  "num_attention_heads": 16,
16
  "num_hidden_layers": 36,
17
  "num_key_value_heads": 2,
18
+ "pad_token_id": 151654,
19
  "rms_norm_eps": 1e-06,
20
  "rope_scaling": {
21
  "mrope_section": [
 
28
  },
29
  "rope_theta": 1000000.0,
30
  "sliding_window": 32768,
31
+ "tie_word_embeddings": true,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
  "torch_dtype": "bfloat16",
33
+ "transformers_version": "4.51.3",
34
+ "unsloth_fixed": true,
35
  "use_cache": true,
36
  "use_sliding_window": false,
37
  "video_token_id": 151656,
 
47
  "hidden_size": 1280,
48
  "in_channels": 3,
49
  "in_chans": 3,
 
50
  "intermediate_size": 3420,
51
  "model_type": "qwen2_5_vl",
52
  "num_heads": 16,
 
63
  "vision_start_token_id": 151652,
64
  "vision_token_id": 151654,
65
  "vocab_size": 151936
66
+ }