Hugging Face's logo

BAAI
/
Aquila-VL-2B-llava-qwen

Image-Text-to-Text
Transformers
Safetensors
English
Chinese
qwen2
text-generation
multimodal
conversational
text-generation-inference
Inference Endpoints
Model card Files Community
ldwang commited on
Commit
18f3cfd
•
1 Parent(s): 9e2399d

init commit.

Browse files
Files changed (1) hide show
  1. config.json +198 -0
config.json ADDED
@@ -0,0 +1,198 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/share/project/gushuhao/1.research/LLaVA-NeXT-main/llavaonevision-qwen1.5b-mlp2x_gelu-qwen2.5-stage1.5b-1012-32n40g-node8",
3
+ "architectures": [
4
+ "LlavaQwenForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 151643,
8
+ "eos_token_id": 151645,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 1536,
11
+ "image_aspect_ratio": "anyres_max_9",
12
+ "image_crop_resolution": null,
13
+ "image_grid_pinpoints": [
14
+ [
15
+ 384,
16
+ 384
17
+ ],
18
+ [
19
+ 384,
20
+ 768
21
+ ],
22
+ [
23
+ 384,
24
+ 1152
25
+ ],
26
+ [
27
+ 384,
28
+ 1536
29
+ ],
30
+ [
31
+ 384,
32
+ 1920
33
+ ],
34
+ [
35
+ 384,
36
+ 2304
37
+ ],
38
+ [
39
+ 768,
40
+ 384
41
+ ],
42
+ [
43
+ 768,
44
+ 768
45
+ ],
46
+ [
47
+ 768,
48
+ 1152
49
+ ],
50
+ [
51
+ 768,
52
+ 1536
53
+ ],
54
+ [
55
+ 768,
56
+ 1920
57
+ ],
58
+ [
59
+ 768,
60
+ 2304
61
+ ],
62
+ [
63
+ 1152,
64
+ 384
65
+ ],
66
+ [
67
+ 1152,
68
+ 768
69
+ ],
70
+ [
71
+ 1152,
72
+ 1152
73
+ ],
74
+ [
75
+ 1152,
76
+ 1536
77
+ ],
78
+ [
79
+ 1152,
80
+ 1920
81
+ ],
82
+ [
83
+ 1152,
84
+ 2304
85
+ ],
86
+ [
87
+ 1536,
88
+ 384
89
+ ],
90
+ [
91
+ 1536,
92
+ 768
93
+ ],
94
+ [
95
+ 1536,
96
+ 1152
97
+ ],
98
+ [
99
+ 1536,
100
+ 1536
101
+ ],
102
+ [
103
+ 1536,
104
+ 1920
105
+ ],
106
+ [
107
+ 1536,
108
+ 2304
109
+ ],
110
+ [
111
+ 1920,
112
+ 384
113
+ ],
114
+ [
115
+ 1920,
116
+ 768
117
+ ],
118
+ [
119
+ 1920,
120
+ 1152
121
+ ],
122
+ [
123
+ 1920,
124
+ 1536
125
+ ],
126
+ [
127
+ 1920,
128
+ 1920
129
+ ],
130
+ [
131
+ 1920,
132
+ 2304
133
+ ],
134
+ [
135
+ 2304,
136
+ 384
137
+ ],
138
+ [
139
+ 2304,
140
+ 768
141
+ ],
142
+ [
143
+ 2304,
144
+ 1152
145
+ ],
146
+ [
147
+ 2304,
148
+ 1536
149
+ ],
150
+ [
151
+ 2304,
152
+ 1920
153
+ ],
154
+ [
155
+ 2304,
156
+ 2304
157
+ ]
158
+ ],
159
+ "image_split_resolution": null,
160
+ "initializer_range": 0.02,
161
+ "intermediate_size": 8960,
162
+ "max_position_embeddings": 32768,
163
+ "max_window_layers": 21,
164
+ "mm_hidden_size": 1152,
165
+ "mm_newline_position": "one_token",
166
+ "mm_patch_merge_type": "spatial_unpad",
167
+ "mm_projector_lr": null,
168
+ "mm_projector_type": "mlp2x_gelu",
169
+ "mm_resampler_type": null,
170
+ "mm_spatial_pool_mode": "bilinear",
171
+ "mm_tunable_parts": "mm_vision_tower,mm_mlp_adapter,mm_language_model",
172
+ "mm_use_im_patch_token": false,
173
+ "mm_use_im_start_end": false,
174
+ "mm_vision_select_feature": "patch",
175
+ "mm_vision_select_layer": -2,
176
+ "mm_vision_tower": "/share/project/gushuhao/external_model/siglip-so400m-patch14-384",
177
+ "mm_vision_tower_lr": 2e-06,
178
+ "model_type": "qwen2",
179
+ "num_attention_heads": 12,
180
+ "num_hidden_layers": 28,
181
+ "num_key_value_heads": 2,
182
+ "pos_skipping_range": 4096,
183
+ "rms_norm_eps": 1e-06,
184
+ "rope_scaling": null,
185
+ "rope_theta": 1000000.0,
186
+ "sliding_window": 32768,
187
+ "tie_word_embeddings": true,
188
+ "tokenizer_model_max_length": 12288,
189
+ "tokenizer_padding_side": "right",
190
+ "torch_dtype": "bfloat16",
191
+ "transformers_version": "4.40.0",
192
+ "use_cache": true,
193
+ "use_mm_proj": true,
194
+ "use_pos_skipping": false,
195
+ "use_sliding_window": false,
196
+ "vision_tower_pretrained": null,
197
+ "vocab_size": 151936
198
+ }