Xenova HF staff commited on
Commit
153c113
1 Parent(s): c93fd1e

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
+ onnx/model.onnx_data filter=lfs diff=lfs merge=lfs -text
config.json ADDED
@@ -0,0 +1,366 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "facebook/mms-lid-256",
3
+ "activation_dropout": 0.05,
4
+ "adapter_attn_dim": 16,
5
+ "adapter_kernel_size": 3,
6
+ "adapter_stride": 2,
7
+ "add_adapter": false,
8
+ "apply_spec_augment": true,
9
+ "architectures": [
10
+ "Wav2Vec2ForSequenceClassification"
11
+ ],
12
+ "attention_dropout": 0.05,
13
+ "bos_token_id": 1,
14
+ "classifier_proj_size": 1024,
15
+ "codevector_dim": 1024,
16
+ "contrastive_logits_temperature": 0.1,
17
+ "conv_bias": true,
18
+ "conv_dim": [
19
+ 512,
20
+ 512,
21
+ 512,
22
+ 512,
23
+ 512,
24
+ 512,
25
+ 512
26
+ ],
27
+ "conv_kernel": [
28
+ 10,
29
+ 3,
30
+ 3,
31
+ 3,
32
+ 3,
33
+ 2,
34
+ 2
35
+ ],
36
+ "conv_stride": [
37
+ 5,
38
+ 2,
39
+ 2,
40
+ 2,
41
+ 2,
42
+ 2,
43
+ 2
44
+ ],
45
+ "ctc_loss_reduction": "mean",
46
+ "ctc_zero_infinity": false,
47
+ "diversity_loss_weight": 0.1,
48
+ "do_stable_layer_norm": true,
49
+ "eos_token_id": 2,
50
+ "feat_extract_activation": "gelu",
51
+ "feat_extract_dropout": 0.0,
52
+ "feat_extract_norm": "layer",
53
+ "feat_proj_dropout": 0.05,
54
+ "feat_quantizer_dropout": 0.0,
55
+ "final_dropout": 0.05,
56
+ "hidden_act": "gelu",
57
+ "hidden_dropout": 0.05,
58
+ "hidden_size": 1280,
59
+ "id2label": {
60
+ "0": "ara",
61
+ "1": "cmn",
62
+ "2": "eng",
63
+ "3": "spa",
64
+ "4": "fra",
65
+ "5": "mlg",
66
+ "6": "swe",
67
+ "7": "por",
68
+ "8": "vie",
69
+ "9": "ful",
70
+ "10": "sun",
71
+ "11": "asm",
72
+ "12": "ben",
73
+ "13": "zlm",
74
+ "14": "kor",
75
+ "15": "ind",
76
+ "16": "hin",
77
+ "17": "tuk",
78
+ "18": "urd",
79
+ "19": "aze",
80
+ "20": "slv",
81
+ "21": "mon",
82
+ "22": "hau",
83
+ "23": "tel",
84
+ "24": "swh",
85
+ "25": "bod",
86
+ "26": "rus",
87
+ "27": "tur",
88
+ "28": "heb",
89
+ "29": "mar",
90
+ "30": "som",
91
+ "31": "tgl",
92
+ "32": "tat",
93
+ "33": "tha",
94
+ "34": "cat",
95
+ "35": "ron",
96
+ "36": "mal",
97
+ "37": "bel",
98
+ "38": "pol",
99
+ "39": "yor",
100
+ "40": "nld",
101
+ "41": "bul",
102
+ "42": "hat",
103
+ "43": "afr",
104
+ "44": "isl",
105
+ "45": "amh",
106
+ "46": "tam",
107
+ "47": "hun",
108
+ "48": "hrv",
109
+ "49": "lit",
110
+ "50": "cym",
111
+ "51": "fas",
112
+ "52": "mkd",
113
+ "53": "ell",
114
+ "54": "bos",
115
+ "55": "deu",
116
+ "56": "sqi",
117
+ "57": "jav",
118
+ "58": "kmr",
119
+ "59": "nob",
120
+ "60": "uzb",
121
+ "61": "snd",
122
+ "62": "lat",
123
+ "63": "nya",
124
+ "64": "grn",
125
+ "65": "mya",
126
+ "66": "orm",
127
+ "67": "lin",
128
+ "68": "hye",
129
+ "69": "yue",
130
+ "70": "pan",
131
+ "71": "jpn",
132
+ "72": "kaz",
133
+ "73": "npi",
134
+ "74": "kik",
135
+ "75": "kat",
136
+ "76": "guj",
137
+ "77": "kan",
138
+ "78": "tgk",
139
+ "79": "ukr",
140
+ "80": "ces",
141
+ "81": "lav",
142
+ "82": "bak",
143
+ "83": "khm",
144
+ "84": "fao",
145
+ "85": "glg",
146
+ "86": "ltz",
147
+ "87": "xog",
148
+ "88": "lao",
149
+ "89": "mlt",
150
+ "90": "sin",
151
+ "91": "aka",
152
+ "92": "sna",
153
+ "93": "ita",
154
+ "94": "srp",
155
+ "95": "mri",
156
+ "96": "nno",
157
+ "97": "pus",
158
+ "98": "eus",
159
+ "99": "ory",
160
+ "100": "lug",
161
+ "101": "bre",
162
+ "102": "luo",
163
+ "103": "slk",
164
+ "104": "ewe",
165
+ "105": "fin",
166
+ "106": "rif",
167
+ "107": "dan",
168
+ "108": "yid",
169
+ "109": "yao",
170
+ "110": "mos",
171
+ "111": "hne",
172
+ "112": "est",
173
+ "113": "dyu",
174
+ "114": "bam",
175
+ "115": "uig",
176
+ "116": "sck",
177
+ "117": "tso",
178
+ "118": "mup",
179
+ "119": "ctg",
180
+ "120": "ceb",
181
+ "121": "war",
182
+ "122": "bbc",
183
+ "123": "vmw",
184
+ "124": "sid",
185
+ "125": "tpi",
186
+ "126": "mag",
187
+ "127": "san",
188
+ "128": "kri",
189
+ "129": "lon",
190
+ "130": "kir",
191
+ "131": "run",
192
+ "132": "ubl",
193
+ "133": "kin",
194
+ "134": "rkt",
195
+ "135": "xmm",
196
+ "136": "tir",
197
+ "137": "mai",
198
+ "138": "nan",
199
+ "139": "nyn",
200
+ "140": "bcc",
201
+ "141": "hak",
202
+ "142": "suk",
203
+ "143": "bem",
204
+ "144": "rmy",
205
+ "145": "awa",
206
+ "146": "pcm",
207
+ "147": "bgc",
208
+ "148": "shn",
209
+ "149": "oci",
210
+ "150": "wol",
211
+ "151": "bci",
212
+ "152": "kab",
213
+ "153": "ilo",
214
+ "154": "bcl",
215
+ "155": "haw",
216
+ "156": "mad",
217
+ "157": "nod",
218
+ "158": "sag",
219
+ "159": "sas",
220
+ "160": "jam",
221
+ "161": "mey",
222
+ "162": "shi",
223
+ "163": "hil",
224
+ "164": "ace",
225
+ "165": "kam",
226
+ "166": "min",
227
+ "167": "umb",
228
+ "168": "hno",
229
+ "169": "ban",
230
+ "170": "syl",
231
+ "171": "bxg",
232
+ "172": "xho",
233
+ "173": "mww",
234
+ "174": "epo",
235
+ "175": "tzm",
236
+ "176": "zul",
237
+ "177": "ibo",
238
+ "178": "abk",
239
+ "179": "guz",
240
+ "180": "ckb",
241
+ "181": "knc",
242
+ "182": "nso",
243
+ "183": "bho",
244
+ "184": "dje",
245
+ "185": "tiv",
246
+ "186": "gle",
247
+ "187": "lua",
248
+ "188": "skr",
249
+ "189": "bto",
250
+ "190": "kea",
251
+ "191": "glk",
252
+ "192": "ast",
253
+ "193": "sat",
254
+ "194": "ktu",
255
+ "195": "bhb",
256
+ "196": "emk",
257
+ "197": "kng",
258
+ "198": "kmb",
259
+ "199": "tsn",
260
+ "200": "gom",
261
+ "201": "ven",
262
+ "202": "sco",
263
+ "203": "glv",
264
+ "204": "sot",
265
+ "205": "sou",
266
+ "206": "gno",
267
+ "207": "nde",
268
+ "208": "bjn",
269
+ "209": "ina",
270
+ "210": "fmu",
271
+ "211": "esg",
272
+ "212": "wes",
273
+ "213": "pnb",
274
+ "214": "phr",
275
+ "215": "mui",
276
+ "216": "bug",
277
+ "217": "mrr",
278
+ "218": "kas",
279
+ "219": "lir",
280
+ "220": "vah",
281
+ "221": "ssw",
282
+ "222": "rwr",
283
+ "223": "pcc",
284
+ "224": "hms",
285
+ "225": "wbr",
286
+ "226": "swv",
287
+ "227": "mtr",
288
+ "228": "haz",
289
+ "229": "aii",
290
+ "230": "bns",
291
+ "231": "msi",
292
+ "232": "wuu",
293
+ "233": "hsn",
294
+ "234": "bgp",
295
+ "235": "tts",
296
+ "236": "lmn",
297
+ "237": "dcc",
298
+ "238": "bew",
299
+ "239": "bjj",
300
+ "240": "ibb",
301
+ "241": "tji",
302
+ "242": "hoj",
303
+ "243": "cpx",
304
+ "244": "cdo",
305
+ "245": "daq",
306
+ "246": "mut",
307
+ "247": "nap",
308
+ "248": "czh",
309
+ "249": "gdx",
310
+ "250": "sdh",
311
+ "251": "scn",
312
+ "252": "mnp",
313
+ "253": "bar",
314
+ "254": "mzn",
315
+ "255": "gsw"
316
+ },
317
+ "initializer_range": 0.02,
318
+ "intermediate_size": 5120,
319
+ "label2id": null,
320
+ "layer_norm_eps": 1e-05,
321
+ "layerdrop": 0.05,
322
+ "mask_feature_length": 10,
323
+ "mask_feature_min_masks": 0,
324
+ "mask_feature_prob": 0.0,
325
+ "mask_time_length": 10,
326
+ "mask_time_min_masks": 2,
327
+ "mask_time_prob": 0.05,
328
+ "model_type": "wav2vec2",
329
+ "num_adapter_layers": 3,
330
+ "num_attention_heads": 16,
331
+ "num_codevector_groups": 2,
332
+ "num_codevectors_per_group": 320,
333
+ "num_conv_pos_embedding_groups": 16,
334
+ "num_conv_pos_embeddings": 128,
335
+ "num_feat_extract_layers": 7,
336
+ "num_hidden_layers": 48,
337
+ "num_negatives": 100,
338
+ "output_hidden_size": 1280,
339
+ "pad_token_id": 0,
340
+ "proj_codevector_dim": 1024,
341
+ "tdnn_dilation": [
342
+ 1,
343
+ 2,
344
+ 3,
345
+ 1,
346
+ 1
347
+ ],
348
+ "tdnn_dim": [
349
+ 512,
350
+ 512,
351
+ 512,
352
+ 512,
353
+ 1500
354
+ ],
355
+ "tdnn_kernel": [
356
+ 5,
357
+ 3,
358
+ 3,
359
+ 1,
360
+ 1
361
+ ],
362
+ "transformers_version": "4.32.0.dev0",
363
+ "use_weighted_layer_sum": false,
364
+ "vocab_size": 154,
365
+ "xvector_output_dim": 512
366
+ }
onnx/model.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a4d5cb8c39bb561dcb59365a38fc13af577b709934b82c81f4adfc3483b2f7ba
3
+ size 1519979
onnx/model.onnx_data ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e46aa6389e128fe96c5e1ee0c51e1af5b8e619865caffc686631949831240606
3
+ size 3864155136
onnx/model_quantized.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b4b1534cbbe4c8eb785e6263d05bb1f78eff2257daf73bf2d46472dac60263ce
3
+ size 974053741
preprocessor_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "feature_extractor_type": "Wav2Vec2FeatureExtractor",
4
+ "feature_size": 1,
5
+ "padding_side": "right",
6
+ "padding_value": 0,
7
+ "return_attention_mask": true,
8
+ "sampling_rate": 16000
9
+ }
quantize_config.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_channel": true,
3
+ "reduce_range": true,
4
+ "per_model_config": {
5
+ "model": {
6
+ "op_types": [
7
+ "Conv",
8
+ "Identity",
9
+ "Add",
10
+ "Transpose",
11
+ "Shape",
12
+ "Mul",
13
+ "Unsqueeze",
14
+ "Concat",
15
+ "Relu",
16
+ "Pow",
17
+ "Softmax",
18
+ "Sqrt",
19
+ "Erf",
20
+ "MatMul",
21
+ "Constant",
22
+ "ReduceMean",
23
+ "Reshape",
24
+ "Div",
25
+ "Slice",
26
+ "Sub",
27
+ "Gather",
28
+ "Gemm"
29
+ ],
30
+ "weight_type": "QUInt8"
31
+ }
32
+ }
33
+ }