mgoin commited on
Commit
53412c5
1 Parent(s): 0572f9c

Upload folder using huggingface_hub

Browse files
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
config.json CHANGED
@@ -1,11 +1,23 @@
1
  {
2
- "_name_or_path": "/home/mgoin/.cache/huggingface/hub/models--Qwen--Qwen2-VL-7B-Instruct/snapshots/3ca981c995b0ce691d85d8408216da11ff92f690",
3
  "architectures": [
4
  "Qwen2VLForConditionalGeneration"
5
  ],
6
  "attention_dropout": 0.0,
7
  "bos_token_id": 151643,
8
- "compression_config": {
 
 
 
 
 
 
 
 
 
 
 
 
9
  "config_groups": {
10
  "group_0": {
11
  "input_activations": {
@@ -14,7 +26,7 @@
14
  "dynamic": true,
15
  "group_size": null,
16
  "num_bits": 8,
17
- "observer": "memoryless",
18
  "observer_kwargs": {},
19
  "strategy": "token",
20
  "symmetric": true,
@@ -39,26 +51,144 @@
39
  }
40
  },
41
  "format": "float-quantized",
42
- "global_compression_ratio": 1.2311753355829242,
43
  "ignore": [
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
44
  "lm_head"
45
  ],
46
  "kv_cache_scheme": null,
47
  "quant_method": "compressed-tensors",
48
  "quantization_status": "compressed"
49
  },
50
- "eos_token_id": 151645,
51
- "hidden_act": "silu",
52
- "hidden_size": 3584,
53
- "image_token_id": 151655,
54
- "initializer_range": 0.02,
55
- "intermediate_size": 18944,
56
- "max_position_embeddings": 32768,
57
- "max_window_layers": 28,
58
- "model_type": "qwen2_vl",
59
- "num_attention_heads": 28,
60
- "num_hidden_layers": 28,
61
- "num_key_value_heads": 4,
62
  "rms_norm_eps": 1e-06,
63
  "rope_scaling": {
64
  "mrope_section": [
@@ -73,7 +203,7 @@
73
  "sliding_window": 32768,
74
  "tie_word_embeddings": false,
75
  "torch_dtype": "bfloat16",
76
- "transformers_version": "4.45.0.dev0",
77
  "use_cache": true,
78
  "use_sliding_window": false,
79
  "video_token_id": 151656,
 
1
  {
2
+ "_name_or_path": "/home/mgoin/.cache/huggingface/hub/models--Qwen--Qwen2-VL-7B-Instruct/snapshots/51c47430f97dd7c74aa1fa6825e68a813478097f",
3
  "architectures": [
4
  "Qwen2VLForConditionalGeneration"
5
  ],
6
  "attention_dropout": 0.0,
7
  "bos_token_id": 151643,
8
+ "eos_token_id": 151645,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 3584,
11
+ "image_token_id": 151655,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 18944,
14
+ "max_position_embeddings": 32768,
15
+ "max_window_layers": 28,
16
+ "model_type": "qwen2_vl",
17
+ "num_attention_heads": 28,
18
+ "num_hidden_layers": 28,
19
+ "num_key_value_heads": 4,
20
+ "quantization_config": {
21
  "config_groups": {
22
  "group_0": {
23
  "input_activations": {
 
26
  "dynamic": true,
27
  "group_size": null,
28
  "num_bits": 8,
29
+ "observer": null,
30
  "observer_kwargs": {},
31
  "strategy": "token",
32
  "symmetric": true,
 
51
  }
52
  },
53
  "format": "float-quantized",
54
+ "global_compression_ratio": 1.220698132330407,
55
  "ignore": [
56
+ "visual.blocks.0.attn.qkv",
57
+ "visual.blocks.0.attn.proj",
58
+ "visual.blocks.0.mlp.fc1",
59
+ "visual.blocks.0.mlp.fc2",
60
+ "visual.blocks.1.attn.qkv",
61
+ "visual.blocks.1.attn.proj",
62
+ "visual.blocks.1.mlp.fc1",
63
+ "visual.blocks.1.mlp.fc2",
64
+ "visual.blocks.2.attn.qkv",
65
+ "visual.blocks.2.attn.proj",
66
+ "visual.blocks.2.mlp.fc1",
67
+ "visual.blocks.2.mlp.fc2",
68
+ "visual.blocks.3.attn.qkv",
69
+ "visual.blocks.3.attn.proj",
70
+ "visual.blocks.3.mlp.fc1",
71
+ "visual.blocks.3.mlp.fc2",
72
+ "visual.blocks.4.attn.qkv",
73
+ "visual.blocks.4.attn.proj",
74
+ "visual.blocks.4.mlp.fc1",
75
+ "visual.blocks.4.mlp.fc2",
76
+ "visual.blocks.5.attn.qkv",
77
+ "visual.blocks.5.attn.proj",
78
+ "visual.blocks.5.mlp.fc1",
79
+ "visual.blocks.5.mlp.fc2",
80
+ "visual.blocks.6.attn.qkv",
81
+ "visual.blocks.6.attn.proj",
82
+ "visual.blocks.6.mlp.fc1",
83
+ "visual.blocks.6.mlp.fc2",
84
+ "visual.blocks.7.attn.qkv",
85
+ "visual.blocks.7.attn.proj",
86
+ "visual.blocks.7.mlp.fc1",
87
+ "visual.blocks.7.mlp.fc2",
88
+ "visual.blocks.8.attn.qkv",
89
+ "visual.blocks.8.attn.proj",
90
+ "visual.blocks.8.mlp.fc1",
91
+ "visual.blocks.8.mlp.fc2",
92
+ "visual.blocks.9.attn.qkv",
93
+ "visual.blocks.9.attn.proj",
94
+ "visual.blocks.9.mlp.fc1",
95
+ "visual.blocks.9.mlp.fc2",
96
+ "visual.blocks.10.attn.qkv",
97
+ "visual.blocks.10.attn.proj",
98
+ "visual.blocks.10.mlp.fc1",
99
+ "visual.blocks.10.mlp.fc2",
100
+ "visual.blocks.11.attn.qkv",
101
+ "visual.blocks.11.attn.proj",
102
+ "visual.blocks.11.mlp.fc1",
103
+ "visual.blocks.11.mlp.fc2",
104
+ "visual.blocks.12.attn.qkv",
105
+ "visual.blocks.12.attn.proj",
106
+ "visual.blocks.12.mlp.fc1",
107
+ "visual.blocks.12.mlp.fc2",
108
+ "visual.blocks.13.attn.qkv",
109
+ "visual.blocks.13.attn.proj",
110
+ "visual.blocks.13.mlp.fc1",
111
+ "visual.blocks.13.mlp.fc2",
112
+ "visual.blocks.14.attn.qkv",
113
+ "visual.blocks.14.attn.proj",
114
+ "visual.blocks.14.mlp.fc1",
115
+ "visual.blocks.14.mlp.fc2",
116
+ "visual.blocks.15.attn.qkv",
117
+ "visual.blocks.15.attn.proj",
118
+ "visual.blocks.15.mlp.fc1",
119
+ "visual.blocks.15.mlp.fc2",
120
+ "visual.blocks.16.attn.qkv",
121
+ "visual.blocks.16.attn.proj",
122
+ "visual.blocks.16.mlp.fc1",
123
+ "visual.blocks.16.mlp.fc2",
124
+ "visual.blocks.17.attn.qkv",
125
+ "visual.blocks.17.attn.proj",
126
+ "visual.blocks.17.mlp.fc1",
127
+ "visual.blocks.17.mlp.fc2",
128
+ "visual.blocks.18.attn.qkv",
129
+ "visual.blocks.18.attn.proj",
130
+ "visual.blocks.18.mlp.fc1",
131
+ "visual.blocks.18.mlp.fc2",
132
+ "visual.blocks.19.attn.qkv",
133
+ "visual.blocks.19.attn.proj",
134
+ "visual.blocks.19.mlp.fc1",
135
+ "visual.blocks.19.mlp.fc2",
136
+ "visual.blocks.20.attn.qkv",
137
+ "visual.blocks.20.attn.proj",
138
+ "visual.blocks.20.mlp.fc1",
139
+ "visual.blocks.20.mlp.fc2",
140
+ "visual.blocks.21.attn.qkv",
141
+ "visual.blocks.21.attn.proj",
142
+ "visual.blocks.21.mlp.fc1",
143
+ "visual.blocks.21.mlp.fc2",
144
+ "visual.blocks.22.attn.qkv",
145
+ "visual.blocks.22.attn.proj",
146
+ "visual.blocks.22.mlp.fc1",
147
+ "visual.blocks.22.mlp.fc2",
148
+ "visual.blocks.23.attn.qkv",
149
+ "visual.blocks.23.attn.proj",
150
+ "visual.blocks.23.mlp.fc1",
151
+ "visual.blocks.23.mlp.fc2",
152
+ "visual.blocks.24.attn.qkv",
153
+ "visual.blocks.24.attn.proj",
154
+ "visual.blocks.24.mlp.fc1",
155
+ "visual.blocks.24.mlp.fc2",
156
+ "visual.blocks.25.attn.qkv",
157
+ "visual.blocks.25.attn.proj",
158
+ "visual.blocks.25.mlp.fc1",
159
+ "visual.blocks.25.mlp.fc2",
160
+ "visual.blocks.26.attn.qkv",
161
+ "visual.blocks.26.attn.proj",
162
+ "visual.blocks.26.mlp.fc1",
163
+ "visual.blocks.26.mlp.fc2",
164
+ "visual.blocks.27.attn.qkv",
165
+ "visual.blocks.27.attn.proj",
166
+ "visual.blocks.27.mlp.fc1",
167
+ "visual.blocks.27.mlp.fc2",
168
+ "visual.blocks.28.attn.qkv",
169
+ "visual.blocks.28.attn.proj",
170
+ "visual.blocks.28.mlp.fc1",
171
+ "visual.blocks.28.mlp.fc2",
172
+ "visual.blocks.29.attn.qkv",
173
+ "visual.blocks.29.attn.proj",
174
+ "visual.blocks.29.mlp.fc1",
175
+ "visual.blocks.29.mlp.fc2",
176
+ "visual.blocks.30.attn.qkv",
177
+ "visual.blocks.30.attn.proj",
178
+ "visual.blocks.30.mlp.fc1",
179
+ "visual.blocks.30.mlp.fc2",
180
+ "visual.blocks.31.attn.qkv",
181
+ "visual.blocks.31.attn.proj",
182
+ "visual.blocks.31.mlp.fc1",
183
+ "visual.blocks.31.mlp.fc2",
184
+ "visual.merger.mlp.0",
185
+ "visual.merger.mlp.2",
186
  "lm_head"
187
  ],
188
  "kv_cache_scheme": null,
189
  "quant_method": "compressed-tensors",
190
  "quantization_status": "compressed"
191
  },
 
 
 
 
 
 
 
 
 
 
 
 
192
  "rms_norm_eps": 1e-06,
193
  "rope_scaling": {
194
  "mrope_section": [
 
203
  "sliding_window": 32768,
204
  "tie_word_embeddings": false,
205
  "torch_dtype": "bfloat16",
206
+ "transformers_version": "4.46.0.dev0",
207
  "use_cache": true,
208
  "use_sliding_window": false,
209
  "video_token_id": 151656,
generation_config.json CHANGED
@@ -9,5 +9,5 @@
9
  "temperature": 0.01,
10
  "top_k": 1,
11
  "top_p": 0.001,
12
- "transformers_version": "4.45.0.dev0"
13
  }
 
9
  "temperature": 0.01,
10
  "top_k": 1,
11
  "top_p": 0.001,
12
+ "transformers_version": "4.46.0.dev0"
13
  }
model-00001-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8f5a686c2b56588155c65b173ef7fd6a2fdc5e350942c32e9a60857630933ee3
3
+ size 4938519208
model-00002-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fa2df119f1ba538ce27b6bfb6c4b3cfcca9d618c5cf37dfae72df81767fc1f98
3
+ size 4031832312
model-00003-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e76ec4dc2e7c0dcbe3432e58fae08e62204a2ef2323069d85c8fa0affb87332a
3
+ size 1089994880
model.safetensors.index.json CHANGED
The diff for this file is too large to render. See raw diff
 
recipe.yaml ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ DEFAULT_stage:
2
+ DEFAULT_modifiers:
3
+ QuantizationModifier:
4
+ ignore: ['re:.*lm_head', 're:visual.*']
5
+ targets: Linear
6
+ scheme: FP8_DYNAMIC
tokenizer.json CHANGED
The diff for this file is too large to render. See raw diff