Hugging Face's logo

onnx-community
/
lite-whisper-large-v3-acc-ONNX

Automatic Speech Recognition
Transformers.js
ONNX
lite-whisper
text2text-generation
Model card Files Community
Xenova HF staff commited on
Commit
ee981e6
·
verified ·
1 Parent(s): 616e035

Upload folder using huggingface_hub

Browse files
Files changed (17) hide show
  1. .gitattributes +3 -0
  2. config.json +259 -0
  3. generation_config.json +13 -0
  4. onnx/decoder_model.onnx +3 -0
  5. onnx/decoder_model.onnx_data +3 -0
  6. onnx/decoder_model_merged.onnx +3 -0
  7. onnx/decoder_model_merged.onnx_data +3 -0
  8. onnx/decoder_with_past_model.onnx +3 -0
  9. onnx/decoder_with_past_model.onnx_data +3 -0
  10. onnx/encoder_model.onnx +3 -0
  11. onnx/encoder_model_bnb4.onnx +3 -0
  12. onnx/encoder_model_fp16.onnx +3 -0
  13. onnx/encoder_model_int8.onnx +3 -0
  14. onnx/encoder_model_q4.onnx +3 -0
  15. onnx/encoder_model_q4f16.onnx +3 -0
  16. onnx/encoder_model_quantized.onnx +3 -0
  17. onnx/encoder_model_uint8.onnx +3 -0
.gitattributes CHANGED
@@ -33,3 +33,6 @@ 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
+ onnx/decoder_model.onnx_data filter=lfs diff=lfs merge=lfs -text
37
+ onnx/decoder_model_merged.onnx_data filter=lfs diff=lfs merge=lfs -text
38
+ onnx/decoder_with_past_model.onnx_data filter=lfs diff=lfs merge=lfs -text
config.json ADDED
@@ -0,0 +1,259 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_attn_implementation_autoset": true,
3
+ "_name_or_path": "efficient-speech/lite-whisper-large-v3-acc",
4
+ "activation_dropout": 0.0,
5
+ "activation_function": "gelu",
6
+ "apply_spec_augment": false,
7
+ "architectures": [
8
+ "LiteWhisperForConditionalGeneration"
9
+ ],
10
+ "attention_dropout": 0.0,
11
+ "auto_map": {
12
+ "AutoConfig": "efficient-speech/lite-whisper-large-v3-acc--configuration_lite_whisper.LiteWhisperConfig",
13
+ "AutoModel": "efficient-speech/lite-whisper-large-v3-acc--modeling_lite_whisper.LiteWhisperForConditionalGeneration"
14
+ },
15
+ "begin_suppress_tokens": null,
16
+ "bos_token_id": 50257,
17
+ "classifier_proj_size": 256,
18
+ "d_model": 1280,
19
+ "decoder_attention_heads": 20,
20
+ "decoder_ffn_dim": 5120,
21
+ "decoder_layerdrop": 0.0,
22
+ "decoder_layers": 32,
23
+ "decoder_start_token_id": 50258,
24
+ "dropout": 0.0,
25
+ "encoder_attention_heads": 20,
26
+ "encoder_ffn_dim": 5120,
27
+ "encoder_layerdrop": 0.0,
28
+ "encoder_layers": 32,
29
+ "eos_token_id": 50257,
30
+ "init_std": 0.02,
31
+ "is_encoder_decoder": true,
32
+ "low_rank_config": [
33
+ {
34
+ "fc1": 368,
35
+ "fc2": 368,
36
+ "k_proj": 48,
37
+ "out_proj": 64,
38
+ "q_proj": 48,
39
+ "v_proj": 80
40
+ },
41
+ {
42
+ "fc1": 272,
43
+ "fc2": 352,
44
+ "k_proj": 96,
45
+ "out_proj": 224,
46
+ "q_proj": 112,
47
+ "v_proj": 240
48
+ },
49
+ {
50
+ "fc1": 224,
51
+ "fc2": 224,
52
+ "k_proj": 80,
53
+ "out_proj": 288,
54
+ "q_proj": 96,
55
+ "v_proj": 320
56
+ },
57
+ {
58
+ "fc1": 288,
59
+ "fc2": 208,
60
+ "k_proj": 96,
61
+ "out_proj": 304,
62
+ "q_proj": 112,
63
+ "v_proj": 368
64
+ },
65
+ {
66
+ "fc1": 304,
67
+ "fc2": 320,
68
+ "k_proj": 112,
69
+ "out_proj": 288,
70
+ "q_proj": 128,
71
+ "v_proj": 400
72
+ },
73
+ {
74
+ "fc1": 304,
75
+ "fc2": 304,
76
+ "k_proj": 80,
77
+ "out_proj": 288,
78
+ "q_proj": 96,
79
+ "v_proj": 384
80
+ },
81
+ {
82
+ "fc1": 352,
83
+ "fc2": 400,
84
+ "k_proj": 128,
85
+ "out_proj": 384,
86
+ "q_proj": 176,
87
+ "v_proj": 576
88
+ },
89
+ {
90
+ "fc1": 384,
91
+ "fc2": 656,
92
+ "k_proj": 160,
93
+ "out_proj": 384,
94
+ "q_proj": 192,
95
+ "v_proj": 496
96
+ },
97
+ {
98
+ "fc1": 464,
99
+ "fc2": 752,
100
+ "k_proj": 160,
101
+ "out_proj": 432,
102
+ "q_proj": 208,
103
+ "v_proj": 560
104
+ },
105
+ {
106
+ "fc1": 464,
107
+ "fc2": 752,
108
+ "k_proj": 176,
109
+ "out_proj": 432,
110
+ "q_proj": 224,
111
+ "v_proj": 608
112
+ },
113
+ {
114
+ "fc1": 528,
115
+ "fc2": 768,
116
+ "k_proj": 112,
117
+ "out_proj": 432,
118
+ "q_proj": 160,
119
+ "v_proj": 576
120
+ },
121
+ {
122
+ "fc1": 528,
123
+ "fc2": 768,
124
+ "k_proj": 144,
125
+ "out_proj": 416,
126
+ "q_proj": 176,
127
+ "v_proj": 592
128
+ },
129
+ {
130
+ "fc1": 560,
131
+ "fc2": 864,
132
+ "k_proj": 144,
133
+ "out_proj": 416,
134
+ "q_proj": 176,
135
+ "v_proj": 624
136
+ },
137
+ {
138
+ "fc1": 624,
139
+ "fc2": 960,
140
+ "k_proj": 128,
141
+ "out_proj": 432,
142
+ "q_proj": 160,
143
+ "v_proj": 608
144
+ },
145
+ {
146
+ "fc1": 720,
147
+ "fc2": 944,
148
+ "k_proj": 144,
149
+ "out_proj": 496,
150
+ "q_proj": 192
151
+ },
152
+ {
153
+ "fc1": 624,
154
+ "k_proj": 112,
155
+ "out_proj": 480,
156
+ "q_proj": 144
157
+ },
158
+ {
159
+ "fc1": 656,
160
+ "k_proj": 112,
161
+ "out_proj": 416,
162
+ "q_proj": 144,
163
+ "v_proj": 560
164
+ },
165
+ {
166
+ "fc1": 736,
167
+ "k_proj": 176,
168
+ "out_proj": 512,
169
+ "q_proj": 224
170
+ },
171
+ {
172
+ "fc1": 784,
173
+ "k_proj": 160,
174
+ "out_proj": 448,
175
+ "q_proj": 192,
176
+ "v_proj": 624
177
+ },
178
+ {
179
+ "fc1": 832,
180
+ "k_proj": 160,
181
+ "out_proj": 496,
182
+ "q_proj": 208
183
+ },
184
+ {
185
+ "fc1": 720,
186
+ "fc2": 16,
187
+ "k_proj": 208,
188
+ "out_proj": 608,
189
+ "q_proj": 240
190
+ },
191
+ {
192
+ "fc1": 960,
193
+ "k_proj": 224,
194
+ "out_proj": 592,
195
+ "q_proj": 272
196
+ },
197
+ {
198
+ "fc1": 960,
199
+ "k_proj": 288,
200
+ "q_proj": 336
201
+ },
202
+ {
203
+ "fc1": 1008,
204
+ "k_proj": 272,
205
+ "q_proj": 320
206
+ },
207
+ {
208
+ "k_proj": 336,
209
+ "q_proj": 400
210
+ },
211
+ {
212
+ "k_proj": 336,
213
+ "q_proj": 416
214
+ },
215
+ {
216
+ "k_proj": 384,
217
+ "q_proj": 464
218
+ },
219
+ {
220
+ "k_proj": 352,
221
+ "q_proj": 416
222
+ },
223
+ {
224
+ "k_proj": 416,
225
+ "q_proj": 496
226
+ },
227
+ {
228
+ "k_proj": 480,
229
+ "q_proj": 576
230
+ },
231
+ {
232
+ "k_proj": 496,
233
+ "q_proj": 592
234
+ },
235
+ {
236
+ "k_proj": 496,
237
+ "q_proj": 592
238
+ }
239
+ ],
240
+ "mask_feature_length": 10,
241
+ "mask_feature_min_masks": 0,
242
+ "mask_feature_prob": 0.0,
243
+ "mask_time_length": 10,
244
+ "mask_time_min_masks": 2,
245
+ "mask_time_prob": 0.05,
246
+ "max_length": null,
247
+ "max_source_positions": 1500,
248
+ "max_target_positions": 448,
249
+ "median_filter_width": 7,
250
+ "model_type": "lite-whisper",
251
+ "num_hidden_layers": 32,
252
+ "num_mel_bins": 128,
253
+ "pad_token_id": 50256,
254
+ "scale_embedding": false,
255
+ "transformers_version": "4.48.3",
256
+ "use_cache": true,
257
+ "use_weighted_layer_sum": false,
258
+ "vocab_size": 51866
259
+ }
generation_config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "begin_suppress_tokens": [
4
+ 220,
5
+ 50257
6
+ ],
7
+ "bos_token_id": 50257,
8
+ "decoder_start_token_id": 50258,
9
+ "eos_token_id": 50257,
10
+ "max_length": 448,
11
+ "pad_token_id": 50256,
12
+ "transformers_version": "4.48.3"
13
+ }
onnx/decoder_model.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:be13d3d55e64e5bc9233be10e2b331f4c58a222236ead9dd5f65fb1abf1c2dae
3
+ size 1164232
onnx/decoder_model.onnx_data ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d72c229cb198f4e397b8081fbc31a3528b64f338fb41a934301237043267ca9
3
+ size 3626086400
onnx/decoder_model_merged.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eafc02cbf013fa90a460a86f4b882ced5d01c824ac1201c76382965ff5150adf
3
+ size 2569563
onnx/decoder_model_merged.onnx_data ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:69700fe5f05cae25e834e9f226e678aebbc18af13c5526ad81829489c357515d
3
+ size 3626086400
onnx/decoder_with_past_model.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d7cd5f1b0a9e4fe980eb4aed90dd4b749b7a9014caf879fdba97a50149b66182
3
+ size 903746
onnx/decoder_with_past_model.onnx_data ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ae383dec2237699bebf91bd1b75c9d357b3da020a4a54a0f77041ce86f7ed74d
3
+ size 3206492160
onnx/encoder_model.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4821cc7acc4b576f9c8c960b289a1d5dbc151841f0569c0c2ed183a435093391
3
+ size 1725918593
onnx/encoder_model_bnb4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f940bc96828d354660a38f88b9272ebbb4523007cda5747ed9636d94910ee85
3
+ size 270176716
onnx/encoder_model_fp16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:feed23d05e6fd0f2979ab7c39d167b55e17d1082c8ad6067b4e182d0aabafcdd
3
+ size 863178110
onnx/encoder_model_int8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:604f8771168cb483ae895b8a5395acca3814e70da96a43cabc1da0f7bc0f7ae1
3
+ size 439554595
onnx/encoder_model_q4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0f7dc6accf9f9f03f8862a88e176eb4d1972a83e918fcb3ef37f76ba389ccfa5
3
+ size 298268953
onnx/encoder_model_q4f16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:231363838281977202bd1bcc3ee1398f3fa81bb137af41abfac56e224e4ef420
3
+ size 255908691
onnx/encoder_model_quantized.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b0a8d7f51e63e1ff438df39d3ad22fb15792e04388f564cddd5d90b530f73dd9
3
+ size 439554750
onnx/encoder_model_uint8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b0a8d7f51e63e1ff438df39d3ad22fb15792e04388f564cddd5d90b530f73dd9
3
+ size 439554750