Hugging Face's logo

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

Automatic Speech Recognition
Transformers.js
ONNX
lite-whisper
text-generation
Model card Files
xet
Community
Xenova HF Staff commited on
Commit
3bc504f
·
verified ·
1 Parent(s): 9f74a46

Upload folder using huggingface_hub

Browse files
Files changed (31) hide show
  1. .gitattributes +3 -0
  2. config.json +285 -0
  3. decoder_model.onnx +3 -0
  4. decoder_model.onnx_data +3 -0
  5. decoder_model_merged.onnx +3 -0
  6. decoder_model_merged.onnx_data +3 -0
  7. decoder_with_past_model.onnx +3 -0
  8. decoder_with_past_model.onnx_data +3 -0
  9. encoder_model.onnx +3 -0
  10. generation_config.json +13 -0
  11. onnx/decoder_model_bnb4.onnx +3 -0
  12. onnx/decoder_model_fp16.onnx +3 -0
  13. onnx/decoder_model_int8.onnx +3 -0
  14. onnx/decoder_model_q4.onnx +3 -0
  15. onnx/decoder_model_q4f16.onnx +3 -0
  16. onnx/decoder_model_quantized.onnx +3 -0
  17. onnx/decoder_model_uint8.onnx +3 -0
  18. onnx/decoder_with_past_model_bnb4.onnx +3 -0
  19. onnx/decoder_with_past_model_fp16.onnx +3 -0
  20. onnx/decoder_with_past_model_int8.onnx +3 -0
  21. onnx/decoder_with_past_model_q4.onnx +3 -0
  22. onnx/decoder_with_past_model_q4f16.onnx +3 -0
  23. onnx/decoder_with_past_model_quantized.onnx +3 -0
  24. onnx/decoder_with_past_model_uint8.onnx +3 -0
  25. onnx/encoder_model_bnb4.onnx +3 -0
  26. onnx/encoder_model_fp16.onnx +3 -0
  27. onnx/encoder_model_int8.onnx +3 -0
  28. onnx/encoder_model_q4.onnx +3 -0
  29. onnx/encoder_model_q4f16.onnx +3 -0
  30. onnx/encoder_model_quantized.onnx +3 -0
  31. 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
+ decoder_model.onnx_data filter=lfs diff=lfs merge=lfs -text
37
+ decoder_model_merged.onnx_data filter=lfs diff=lfs merge=lfs -text
38
+ decoder_with_past_model.onnx_data filter=lfs diff=lfs merge=lfs -text
config.json ADDED
@@ -0,0 +1,285 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_attn_implementation_autoset": true,
3
+ "_name_or_path": "efficient-speech/lite-whisper-large-v3",
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--configuration_lite_whisper.LiteWhisperConfig",
13
+ "AutoModel": "efficient-speech/lite-whisper-large-v3--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": 32,
37
+ "out_proj": 32,
38
+ "q_proj": 32,
39
+ "v_proj": 48
40
+ },
41
+ {
42
+ "fc1": 272,
43
+ "fc2": 352,
44
+ "k_proj": 64,
45
+ "out_proj": 112,
46
+ "q_proj": 64,
47
+ "v_proj": 128
48
+ },
49
+ {
50
+ "fc1": 224,
51
+ "fc2": 224,
52
+ "k_proj": 48,
53
+ "out_proj": 160,
54
+ "q_proj": 48,
55
+ "v_proj": 160
56
+ },
57
+ {
58
+ "fc1": 288,
59
+ "fc2": 208,
60
+ "k_proj": 48,
61
+ "out_proj": 160,
62
+ "q_proj": 48,
63
+ "v_proj": 192
64
+ },
65
+ {
66
+ "fc1": 304,
67
+ "fc2": 320,
68
+ "k_proj": 48,
69
+ "out_proj": 144,
70
+ "q_proj": 64,
71
+ "v_proj": 208
72
+ },
73
+ {
74
+ "fc1": 304,
75
+ "fc2": 304,
76
+ "k_proj": 48,
77
+ "out_proj": 144,
78
+ "q_proj": 48,
79
+ "v_proj": 192
80
+ },
81
+ {
82
+ "fc1": 352,
83
+ "fc2": 400,
84
+ "k_proj": 64,
85
+ "out_proj": 208,
86
+ "q_proj": 80,
87
+ "v_proj": 304
88
+ },
89
+ {
90
+ "fc1": 384,
91
+ "fc2": 656,
92
+ "k_proj": 80,
93
+ "out_proj": 208,
94
+ "q_proj": 96,
95
+ "v_proj": 256
96
+ },
97
+ {
98
+ "fc1": 464,
99
+ "fc2": 752,
100
+ "k_proj": 80,
101
+ "out_proj": 224,
102
+ "q_proj": 96,
103
+ "v_proj": 288
104
+ },
105
+ {
106
+ "fc1": 464,
107
+ "fc2": 752,
108
+ "k_proj": 80,
109
+ "out_proj": 224,
110
+ "q_proj": 96,
111
+ "v_proj": 288
112
+ },
113
+ {
114
+ "fc1": 528,
115
+ "fc2": 768,
116
+ "k_proj": 48,
117
+ "out_proj": 240,
118
+ "q_proj": 64,
119
+ "v_proj": 288
120
+ },
121
+ {
122
+ "fc1": 528,
123
+ "fc2": 768,
124
+ "k_proj": 64,
125
+ "out_proj": 224,
126
+ "q_proj": 80,
127
+ "v_proj": 320
128
+ },
129
+ {
130
+ "fc1": 560,
131
+ "fc2": 864,
132
+ "k_proj": 64,
133
+ "out_proj": 224,
134
+ "q_proj": 64,
135
+ "v_proj": 352
136
+ },
137
+ {
138
+ "fc1": 624,
139
+ "fc2": 960,
140
+ "k_proj": 48,
141
+ "out_proj": 240,
142
+ "q_proj": 64,
143
+ "v_proj": 288
144
+ },
145
+ {
146
+ "fc1": 720,
147
+ "fc2": 944,
148
+ "k_proj": 64,
149
+ "out_proj": 256,
150
+ "q_proj": 80,
151
+ "v_proj": 448
152
+ },
153
+ {
154
+ "fc1": 624,
155
+ "k_proj": 48,
156
+ "out_proj": 256,
157
+ "q_proj": 48,
158
+ "v_proj": 432
159
+ },
160
+ {
161
+ "fc1": 656,
162
+ "k_proj": 48,
163
+ "out_proj": 224,
164
+ "q_proj": 64,
165
+ "v_proj": 272
166
+ },
167
+ {
168
+ "fc1": 736,
169
+ "k_proj": 80,
170
+ "out_proj": 288,
171
+ "q_proj": 96,
172
+ "v_proj": 368
173
+ },
174
+ {
175
+ "fc1": 784,
176
+ "k_proj": 64,
177
+ "out_proj": 256,
178
+ "q_proj": 80,
179
+ "v_proj": 304
180
+ },
181
+ {
182
+ "fc1": 832,
183
+ "k_proj": 80,
184
+ "out_proj": 272,
185
+ "q_proj": 80,
186
+ "v_proj": 416
187
+ },
188
+ {
189
+ "fc1": 720,
190
+ "fc2": 16,
191
+ "k_proj": 80,
192
+ "out_proj": 352,
193
+ "q_proj": 96,
194
+ "v_proj": 432
195
+ },
196
+ {
197
+ "fc1": 960,
198
+ "k_proj": 96,
199
+ "out_proj": 336,
200
+ "q_proj": 112,
201
+ "v_proj": 448
202
+ },
203
+ {
204
+ "fc1": 960,
205
+ "k_proj": 112,
206
+ "out_proj": 448,
207
+ "q_proj": 144,
208
+ "v_proj": 544
209
+ },
210
+ {
211
+ "fc1": 1008,
212
+ "k_proj": 96,
213
+ "out_proj": 416,
214
+ "q_proj": 128,
215
+ "v_proj": 464
216
+ },
217
+ {
218
+ "k_proj": 128,
219
+ "out_proj": 448,
220
+ "q_proj": 176,
221
+ "v_proj": 560
222
+ },
223
+ {
224
+ "k_proj": 128,
225
+ "out_proj": 416,
226
+ "q_proj": 176,
227
+ "v_proj": 592
228
+ },
229
+ {
230
+ "k_proj": 144,
231
+ "out_proj": 496,
232
+ "q_proj": 192,
233
+ "v_proj": 624
234
+ },
235
+ {
236
+ "k_proj": 128,
237
+ "out_proj": 400,
238
+ "q_proj": 192,
239
+ "v_proj": 608
240
+ },
241
+ {
242
+ "k_proj": 144,
243
+ "out_proj": 448,
244
+ "q_proj": 208,
245
+ "v_proj": 592
246
+ },
247
+ {
248
+ "k_proj": 176,
249
+ "out_proj": 496,
250
+ "q_proj": 272,
251
+ "v_proj": 560
252
+ },
253
+ {
254
+ "k_proj": 192,
255
+ "out_proj": 464,
256
+ "q_proj": 272,
257
+ "v_proj": 560
258
+ },
259
+ {
260
+ "k_proj": 176,
261
+ "out_proj": 464,
262
+ "q_proj": 272,
263
+ "v_proj": 432
264
+ }
265
+ ],
266
+ "mask_feature_length": 10,
267
+ "mask_feature_min_masks": 0,
268
+ "mask_feature_prob": 0.0,
269
+ "mask_time_length": 10,
270
+ "mask_time_min_masks": 2,
271
+ "mask_time_prob": 0.05,
272
+ "max_length": null,
273
+ "max_source_positions": 1500,
274
+ "max_target_positions": 448,
275
+ "median_filter_width": 7,
276
+ "model_type": "lite-whisper",
277
+ "num_hidden_layers": 32,
278
+ "num_mel_bins": 128,
279
+ "pad_token_id": 50256,
280
+ "scale_embedding": false,
281
+ "transformers_version": "4.48.3",
282
+ "use_cache": true,
283
+ "use_weighted_layer_sum": false,
284
+ "vocab_size": 51866
285
+ }
decoder_model.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c77145d9f1c92c928dae5030c390a75e0b2fc3ecb0dde6f6678a3db6a9b40fab
3
+ size 1164232
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
decoder_model_merged.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:78a7053165918da3a78fb44ed22b2d4d514f57a8a491ffa3fdcd172397c3646a
3
+ size 2569563
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
decoder_with_past_model.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a4027d7ae3ed4d2f6bdba0dca5ab401c9f432fa3ffa76a9b75ee8e26d8bf89ba
3
+ size 903746
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
encoder_model.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d7fd9aaac50382e73d6a808671572142e7688bb4701f282834dae9d50c97efa
3
+ size 1515884674
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_bnb4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81168339a957eddf31e71892ae408bc689f3c34aa7bd1bd54d79550aa7dd0218
3
+ size 743661427
onnx/decoder_model_fp16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e9e8b063834ce2d04d00c38a977d15d7476d24cb5f0177e75e73605d0c710c2e
3
+ size 1814184089
onnx/decoder_model_int8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:10db124f788484cb3bed4c84910533b88e1016f98cd5d94b6649e79c83383ce1
3
+ size 1175671567
onnx/decoder_model_q4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:422490d4d3cd3981760f106985b371d5d9679cdef5f3593cb5bc7987e9ec8a1e
3
+ size 796087635
onnx/decoder_model_q4f16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a7b290363aaff8ef81ac2ff02751015fe63d6465b155c2802cadcbcdf6456cea
3
+ size 608369356
onnx/decoder_model_quantized.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:10db124f788484cb3bed4c84910533b88e1016f98cd5d94b6649e79c83383ce1
3
+ size 1175671567
onnx/decoder_model_uint8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:89b524ee80bbbdda3fce0d62776aa2d89ebdc469358071787dd2d95805c573b1
3
+ size 1175671735
onnx/decoder_with_past_model_bnb4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3f52c64decd692803f388409feb3f97b11642d30e1bf4c3749b6c17a6a76c5e1
3
+ size 684244835
onnx/decoder_with_past_model_fp16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d0c4a6e50b1d0e72e40e59643418e890ae905e1ddbb8fa6c7c83a0d7e4cf82e1
3
+ size 1604152978
onnx/decoder_with_past_model_int8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:40b8fe0c073878c883fe80d245e3d3b821944e4a29af7a3d8ffe6fb6a7d7fc7c
3
+ size 1070326972
onnx/decoder_with_past_model_q4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2a4b16d148b62671e62782e851bbaefa4a4f5c32b25c61504321c5d99276cb09
3
+ size 730117955
onnx/decoder_with_past_model_q4f16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:17d52894f35d6b786263db3d145df1fd24f119b7c44a25624e74fb64be3b9528
3
+ size 549060933
onnx/decoder_with_past_model_quantized.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:40b8fe0c073878c883fe80d245e3d3b821944e4a29af7a3d8ffe6fb6a7d7fc7c
3
+ size 1070326972
onnx/decoder_with_past_model_uint8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4cea4bcf08dca5845805d418a86b96d4c58cbccaace8857839865201a4746194
3
+ size 1070327106
onnx/encoder_model_bnb4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:975e909f7b8673b528da0cba40a38431878308d5e7ff1d470bcb43d5e3df6e52
3
+ size 240654264
onnx/encoder_model_fp16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:edc42bba4a6f75c9a58599c384844e00589904495ebc0752ec3b06109449e2e9
3
+ size 758165675
onnx/encoder_model_int8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:182d8d13828e4815ea62a754548a4877620963a3ce04e2cf2b1fdbba9b02744d
3
+ size 387087758
onnx/encoder_model_q4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8482a623562cd2f13e9e0405f752755d9a4a45dfaeb927770a4c55191128e249
3
+ size 265617905
onnx/encoder_model_q4f16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3ba81bfdf25ce301b03874286a9d2cbe96d7a52d817a80597221bb62a6782bfd
3
+ size 226524216
onnx/encoder_model_quantized.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2a2069d13c7f516468e36f38e59142111b1cad9e3ccfb87ed05bd4498f77a520
3
+ size 387087928
onnx/encoder_model_uint8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2a2069d13c7f516468e36f38e59142111b1cad9e3ccfb87ed05bd4498f77a520
3
+ size 387087928