Hugging Face's logo

TESS-Computer
/
qwen-click-dit

Image-to-Text
Safetensors
qwen2_5_vl
vision-language
click-prediction
gui-grounding
diffusion-transformer
flow-matching
Model card Files
xet
Community
HusseinLezzaik commited on
Commit
c518981
·
verified ·
1 Parent(s): 5243412

Delete checkpoint-2500/config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. checkpoint-2500/config.json +0 -145
checkpoint-2500/config.json DELETED
@@ -1,145 +0,0 @@
1
- {
2
- "architectures": [
3
- "Qwen2_5_VLForConditionalGeneration"
4
- ],
5
- "attention_dropout": 0.0,
6
- "bos_token_id": 151643,
7
- "dit_dropout": 0.1,
8
- "dit_hidden_size": 512,
9
- "dit_num_heads": 8,
10
- "dit_num_layers": 6,
11
- "dtype": "bfloat16",
12
- "eos_token_id": 151645,
13
- "hidden_act": "silu",
14
- "hidden_size": 2048,
15
- "image_token_id": 151655,
16
- "initializer_range": 0.02,
17
- "intermediate_size": 11008,
18
- "max_position_embeddings": 128000,
19
- "max_window_layers": 70,
20
- "model_type": "qwen2_5_vl",
21
- "num_attention_heads": 16,
22
- "num_hidden_layers": 36,
23
- "num_inference_steps": 16,
24
- "num_key_value_heads": 2,
25
- "rms_norm_eps": 1e-06,
26
- "rope_scaling": {
27
- "mrope_section": [
28
- 16,
29
- 24,
30
- 24
31
- ],
32
- "rope_type": "default",
33
- "type": "default"
34
- },
35
- "rope_theta": 1000000.0,
36
- "sliding_window": 32768,
37
- "text_config": {
38
- "_name_or_path": "Qwen/Qwen2.5-VL-3B-Instruct",
39
- "architectures": [
40
- "Qwen2_5_VLForClickPrediction"
41
- ],
42
- "attention_dropout": 0.0,
43
- "dtype": "bfloat16",
44
- "eos_token_id": 151645,
45
- "hidden_act": "silu",
46
- "hidden_size": 2048,
47
- "initializer_range": 0.02,
48
- "intermediate_size": 11008,
49
- "layer_types": [
50
- "full_attention",
51
- "full_attention",
52
- "full_attention",
53
- "full_attention",
54
- "full_attention",
55
- "full_attention",
56
- "full_attention",
57
- "full_attention",
58
- "full_attention",
59
- "full_attention",
60
- "full_attention",
61
- "full_attention",
62
- "full_attention",
63
- "full_attention",
64
- "full_attention",
65
- "full_attention",
66
- "full_attention",
67
- "full_attention",
68
- "full_attention",
69
- "full_attention",
70
- "full_attention",
71
- "full_attention",
72
- "full_attention",
73
- "full_attention",
74
- "full_attention",
75
- "full_attention",
76
- "full_attention",
77
- "full_attention",
78
- "full_attention",
79
- "full_attention",
80
- "full_attention",
81
- "full_attention",
82
- "full_attention",
83
- "full_attention",
84
- "full_attention",
85
- "full_attention"
86
- ],
87
- "max_position_embeddings": 128000,
88
- "max_window_layers": 70,
89
- "model_type": "qwen2_5_vl_text",
90
- "num_attention_heads": 16,
91
- "num_hidden_layers": 36,
92
- "num_key_value_heads": 2,
93
- "pad_token_id": 151643,
94
- "rms_norm_eps": 1e-06,
95
- "rope_scaling": {
96
- "mrope_section": [
97
- 16,
98
- 24,
99
- 24
100
- ],
101
- "rope_type": "default",
102
- "type": "default"
103
- },
104
- "rope_theta": 1000000.0,
105
- "sliding_window": null,
106
- "tie_word_embeddings": true,
107
- "use_cache": false,
108
- "use_sliding_window": false,
109
- "vision_token_id": 151654,
110
- "vocab_size": 151936
111
- },
112
- "transformers_version": "4.57.3",
113
- "use_cache": true,
114
- "use_sliding_window": false,
115
- "video_token_id": 151656,
116
- "vision_config": {
117
- "depth": 32,
118
- "dtype": "bfloat16",
119
- "fullatt_block_indexes": [
120
- 7,
121
- 15,
122
- 23,
123
- 31
124
- ],
125
- "hidden_act": "silu",
126
- "hidden_size": 1280,
127
- "in_channels": 3,
128
- "in_chans": 3,
129
- "initializer_range": 0.02,
130
- "intermediate_size": 3420,
131
- "model_type": "qwen2_5_vl",
132
- "num_heads": 16,
133
- "out_hidden_size": 2048,
134
- "patch_size": 14,
135
- "spatial_merge_size": 2,
136
- "spatial_patch_size": 14,
137
- "temporal_patch_size": 2,
138
- "tokens_per_second": 2,
139
- "window_size": 112
140
- },
141
- "vision_end_token_id": 151653,
142
- "vision_start_token_id": 151652,
143
- "vision_token_id": 151654,
144
- "vocab_size": 151936
145
- }