Hugging Face's logo

Xenova
/
nanoLLaVA

Image-Text-to-Text
Transformers.js
ONNX
English
llava
text-generation
multimodal
qwen
conversational
custom_code
Model card Files
xet
 Community
4
Xenova HF Staff commited on
Commit
4ec4246
·
verified ·
1 Parent(s): 5f5aa6b

Upload folder using huggingface_hub

Browse files
Files changed (24) hide show
  1. config.json +42 -0
  2. generation_config.json +9 -0
  3. onnx/decoder_model_merged.onnx +3 -0
  4. onnx/decoder_model_merged_bnb4.onnx +3 -0
  5. onnx/decoder_model_merged_fp16.onnx +3 -0
  6. onnx/decoder_model_merged_int8.onnx +3 -0
  7. onnx/decoder_model_merged_q4.onnx +3 -0
  8. onnx/decoder_model_merged_quantized.onnx +3 -0
  9. onnx/decoder_model_merged_uint8.onnx +3 -0
  10. onnx/embed_tokens.onnx +3 -0
  11. onnx/embed_tokens_bnb4.onnx +3 -0
  12. onnx/embed_tokens_fp16.onnx +3 -0
  13. onnx/embed_tokens_int8.onnx +3 -0
  14. onnx/embed_tokens_q4.onnx +3 -0
  15. onnx/embed_tokens_quantized.onnx +3 -0
  16. onnx/embed_tokens_uint8.onnx +3 -0
  17. onnx/vision_encoder.onnx +3 -0
  18. onnx/vision_encoder_bnb4.onnx +3 -0
  19. onnx/vision_encoder_fp16.onnx +3 -0
  20. onnx/vision_encoder_int8.onnx +3 -0
  21. onnx/vision_encoder_q4.onnx +3 -0
  22. onnx/vision_encoder_quantized.onnx +3 -0
  23. onnx/vision_encoder_uint8.onnx +3 -0
  24. quantize_config.json +222 -0
config.json ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "./qwen2model",
3
+ "architectures": [
4
+ "Qwen2ForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "auto_map": {
8
+ "AutoConfig": "qnguyen3/nanoLLaVA--configuration_llava_qwen2.LlavaQwen2Config",
9
+ "AutoModelForCausalLM": "qnguyen3/nanoLLaVA--modeling_llava_qwen2.LlavaQwen2ForCausalLM"
10
+ },
11
+ "bos_token_id": 151645,
12
+ "eos_token_id": 151645,
13
+ "freeze_mm_mlp_adapter": false,
14
+ "hidden_act": "silu",
15
+ "hidden_size": 1024,
16
+ "image_aspect_ratio": "pad",
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 2816,
19
+ "language_model": "vilm/Quyen-SE-v0.1",
20
+ "max_position_embeddings": 32768,
21
+ "max_window_layers": 21,
22
+ "mm_hidden_size": 1152,
23
+ "mm_projector_lr": null,
24
+ "mm_projector_type": "mlp2x_gelu",
25
+ "mm_vision_tower": "google/siglip-so400m-patch14-384",
26
+ "model_type": "qwen2",
27
+ "num_attention_heads": 16,
28
+ "num_hidden_layers": 24,
29
+ "num_key_value_heads": 16,
30
+ "rms_norm_eps": 1e-06,
31
+ "rope_theta": 1000000.0,
32
+ "sliding_window": 4096,
33
+ "tie_word_embeddings": false,
34
+ "tokenizer_model_max_length": 4096,
35
+ "tokenizer_padding_side": "right",
36
+ "transformers_version": "4.38.2",
37
+ "tune_mm_mlp_adapter": false,
38
+ "use_cache": true,
39
+ "use_mm_proj": true,
40
+ "use_sliding_window": false,
41
+ "vocab_size": 151936
42
+ }
generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 151645,
3
+ "do_sample": true,
4
+ "eos_token_id": 151645,
5
+ "max_length": 4096,
6
+ "temperature": 0.7,
7
+ "top_p": 0.8,
8
+ "transformers_version": "4.38.2"
9
+ }
onnx/decoder_model_merged.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:74ad520fd8a1021a5f56844d42892052f9a013f248ce1d3d25de327dc7b3fd80
3
+ size 1873458637
onnx/decoder_model_merged_bnb4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:417645742e1a2efe9961b5c0101c6babbab5ee8a727973a93fb761cb1aac4e68
3
+ size 278953185
onnx/decoder_model_merged_fp16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26ebcb5569832783797de347f59ab2b55d30a6ab18aa21501ed2fd0da04f7c84
3
+ size 937421891
onnx/decoder_model_merged_int8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7cdc95f39bf11c41805ea4e274236dc9807a3ad18879b34fcef73adeaf7a5715
3
+ size 482363069
onnx/decoder_model_merged_q4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aee6b80f089368d33778243e3396438ef207207aff9bc0bfa7baf39eb6eb38f2
3
+ size 307943321
onnx/decoder_model_merged_quantized.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7cdc95f39bf11c41805ea4e274236dc9807a3ad18879b34fcef73adeaf7a5715
3
+ size 482363069
onnx/decoder_model_merged_uint8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9aefd73adf665f33fcdf1c506a14e692f3b7fa9b185d05e5d84cd820b13dca8b
3
+ size 482363156
onnx/embed_tokens.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d2acef888decc296e4e1fda21bc57f16f3cf6df2d7586831fcfa55611700b730
3
+ size 622330095
onnx/embed_tokens_bnb4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d46ef96664f29ac92b0e8cb708a5e7c43b5cb02064850a6b186161306ce22e85
3
+ size 622330114
onnx/embed_tokens_fp16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:72408df178267828c09bd10e785a89ab131a7fb3ae1e12253fcc05187ef869b5
3
+ size 311165317
onnx/embed_tokens_int8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:47ffe12f68b97d993c094e73c1dc32651594f52c03eda3ee850ef6eb93d43f67
3
+ size 155582945
onnx/embed_tokens_q4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d46ef96664f29ac92b0e8cb708a5e7c43b5cb02064850a6b186161306ce22e85
3
+ size 622330114
onnx/embed_tokens_quantized.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:47ffe12f68b97d993c094e73c1dc32651594f52c03eda3ee850ef6eb93d43f67
3
+ size 155582945
onnx/embed_tokens_uint8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e5c9742854fa1a39b97af4431af6d53569011322da197522df43c1b6cbb77404
3
+ size 155582944
onnx/vision_encoder.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:41ce2aaa5515effa861458475e7b743d5d0db5feedbdf92a7676ed015a76d82f
3
+ size 1600661139
onnx/vision_encoder_bnb4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:29a30deca84c0d8a2a32db3cc5224ac8a130b7e8082f7a88d2ed9e5e64c70082
3
+ size 232307081
onnx/vision_encoder_fp16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:447eee3d7ab2d6e741db4eda9ade77d3ba28aae761590f6e12bac73ee60bfbcb
3
+ size 800944460
onnx/vision_encoder_int8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f81803e821d0e9c6b99e1bde014802a374f3edec05ed76805eceb4309b09a163
3
+ size 402392868
onnx/vision_encoder_q4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eca15489472139da3fd05d478962776f007a03b855f569f4201c17ac9301a428
3
+ size 257484927
onnx/vision_encoder_quantized.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5a80e2ff3f3565cff2be7668bb4d35975d3aa7ae2c673ae6069fd90ce756742c
3
+ size 402392950
onnx/vision_encoder_uint8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5a80e2ff3f3565cff2be7668bb4d35975d3aa7ae2c673ae6069fd90ce756742c
3
+ size 402392950
quantize_config.json ADDED
@@ -0,0 +1,222 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "fp16": {},
3
+ "q8": {
4
+ "per_model_config": {
5
+ "decoder_model_merged": {
6
+ "op_types": [
7
+ "Add",
8
+ "Cast",
9
+ "Concat",
10
+ "Constant",
11
+ "ConstantOfShape",
12
+ "Div",
13
+ "Equal",
14
+ "Expand",
15
+ "Gather",
16
+ "Less",
17
+ "MatMul",
18
+ "Mul",
19
+ "Neg",
20
+ "Pow",
21
+ "Range",
22
+ "ReduceMean",
23
+ "Reshape",
24
+ "Shape",
25
+ "Sigmoid",
26
+ "Slice",
27
+ "Softmax",
28
+ "Sqrt",
29
+ "Squeeze",
30
+ "Sub",
31
+ "Transpose",
32
+ "Unsqueeze",
33
+ "Where"
34
+ ],
35
+ "weight_type": "QInt8"
36
+ },
37
+ "embed_tokens": {
38
+ "op_types": [
39
+ "Gather"
40
+ ],
41
+ "weight_type": "QInt8"
42
+ },
43
+ "vision_encoder": {
44
+ "op_types": [
45
+ "Add",
46
+ "Cast",
47
+ "Concat",
48
+ "Constant",
49
+ "Conv",
50
+ "Div",
51
+ "Erf",
52
+ "Gather",
53
+ "MatMul",
54
+ "Mul",
55
+ "Pow",
56
+ "ReduceMean",
57
+ "Reshape",
58
+ "Shape",
59
+ "Slice",
60
+ "Softmax",
61
+ "Sqrt",
62
+ "Sub",
63
+ "Tanh",
64
+ "Transpose",
65
+ "Unsqueeze"
66
+ ],
67
+ "weight_type": "QUInt8"
68
+ }
69
+ },
70
+ "per_channel": false,
71
+ "reduce_range": false
72
+ },
73
+ "int8": {
74
+ "per_model_config": {
75
+ "decoder_model_merged": {
76
+ "op_types": [
77
+ "Add",
78
+ "Cast",
79
+ "Concat",
80
+ "Constant",
81
+ "ConstantOfShape",
82
+ "Div",
83
+ "Equal",
84
+ "Expand",
85
+ "Gather",
86
+ "Less",
87
+ "MatMul",
88
+ "Mul",
89
+ "Neg",
90
+ "Pow",
91
+ "Range",
92
+ "ReduceMean",
93
+ "Reshape",
94
+ "Shape",
95
+ "Sigmoid",
96
+ "Slice",
97
+ "Softmax",
98
+ "Sqrt",
99
+ "Squeeze",
100
+ "Sub",
101
+ "Transpose",
102
+ "Unsqueeze",
103
+ "Where"
104
+ ],
105
+ "weight_type": "QInt8"
106
+ },
107
+ "embed_tokens": {
108
+ "op_types": [
109
+ "Gather"
110
+ ],
111
+ "weight_type": "QInt8"
112
+ },
113
+ "vision_encoder": {
114
+ "op_types": [
115
+ "Add",
116
+ "Cast",
117
+ "Concat",
118
+ "Constant",
119
+ "Conv",
120
+ "Div",
121
+ "Erf",
122
+ "Gather",
123
+ "MatMul",
124
+ "Mul",
125
+ "Pow",
126
+ "ReduceMean",
127
+ "Reshape",
128
+ "Shape",
129
+ "Slice",
130
+ "Softmax",
131
+ "Sqrt",
132
+ "Sub",
133
+ "Tanh",
134
+ "Transpose",
135
+ "Unsqueeze"
136
+ ],
137
+ "weight_type": "QInt8"
138
+ }
139
+ },
140
+ "per_channel": false,
141
+ "reduce_range": false
142
+ },
143
+ "uint8": {
144
+ "per_model_config": {
145
+ "decoder_model_merged": {
146
+ "op_types": [
147
+ "Add",
148
+ "Cast",
149
+ "Concat",
150
+ "Constant",
151
+ "ConstantOfShape",
152
+ "Div",
153
+ "Equal",
154
+ "Expand",
155
+ "Gather",
156
+ "Less",
157
+ "MatMul",
158
+ "Mul",
159
+ "Neg",
160
+ "Pow",
161
+ "Range",
162
+ "ReduceMean",
163
+ "Reshape",
164
+ "Shape",
165
+ "Sigmoid",
166
+ "Slice",
167
+ "Softmax",
168
+ "Sqrt",
169
+ "Squeeze",
170
+ "Sub",
171
+ "Transpose",
172
+ "Unsqueeze",
173
+ "Where"
174
+ ],
175
+ "weight_type": "QUInt8"
176
+ },
177
+ "embed_tokens": {
178
+ "op_types": [
179
+ "Gather"
180
+ ],
181
+ "weight_type": "QUInt8"
182
+ },
183
+ "vision_encoder": {
184
+ "op_types": [
185
+ "Add",
186
+ "Cast",
187
+ "Concat",
188
+ "Constant",
189
+ "Conv",
190
+ "Div",
191
+ "Erf",
192
+ "Gather",
193
+ "MatMul",
194
+ "Mul",
195
+ "Pow",
196
+ "ReduceMean",
197
+ "Reshape",
198
+ "Shape",
199
+ "Slice",
200
+ "Softmax",
201
+ "Sqrt",
202
+ "Sub",
203
+ "Tanh",
204
+ "Transpose",
205
+ "Unsqueeze"
206
+ ],
207
+ "weight_type": "QUInt8"
208
+ }
209
+ },
210
+ "per_channel": false,
211
+ "reduce_range": false
212
+ },
213
+ "q4": {
214
+ "block_size": 32,
215
+ "is_symmetric": true,
216
+ "accuracy_level": null
217
+ },
218
+ "bnb4": {
219
+ "block_size": 64,
220
+ "quant_type": 1
221
+ }
222
+ }