Upload folder using huggingface_hub
Browse files- .gitattributes +1 -0
- config.json +236 -0
- onnx/model.onnx +3 -0
- onnx/model.onnx_data +3 -0
- onnx/model_quantized.onnx +3 -0
- preprocessor_config.json +9 -0
- quantize_config.json +33 -0
.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 +1,236 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "facebook/mms-lid-126",
|
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": "nob",
|
119 |
+
"59": "uzb",
|
120 |
+
"60": "snd",
|
121 |
+
"61": "lat",
|
122 |
+
"62": "nya",
|
123 |
+
"63": "grn",
|
124 |
+
"64": "mya",
|
125 |
+
"65": "orm",
|
126 |
+
"66": "lin",
|
127 |
+
"67": "hye",
|
128 |
+
"68": "yue",
|
129 |
+
"69": "pan",
|
130 |
+
"70": "jpn",
|
131 |
+
"71": "kaz",
|
132 |
+
"72": "npi",
|
133 |
+
"73": "kat",
|
134 |
+
"74": "guj",
|
135 |
+
"75": "kan",
|
136 |
+
"76": "tgk",
|
137 |
+
"77": "ukr",
|
138 |
+
"78": "ces",
|
139 |
+
"79": "lav",
|
140 |
+
"80": "bak",
|
141 |
+
"81": "khm",
|
142 |
+
"82": "fao",
|
143 |
+
"83": "glg",
|
144 |
+
"84": "ltz",
|
145 |
+
"85": "lao",
|
146 |
+
"86": "mlt",
|
147 |
+
"87": "sin",
|
148 |
+
"88": "sna",
|
149 |
+
"89": "ita",
|
150 |
+
"90": "srp",
|
151 |
+
"91": "mri",
|
152 |
+
"92": "nno",
|
153 |
+
"93": "pus",
|
154 |
+
"94": "eus",
|
155 |
+
"95": "ory",
|
156 |
+
"96": "lug",
|
157 |
+
"97": "bre",
|
158 |
+
"98": "luo",
|
159 |
+
"99": "slk",
|
160 |
+
"100": "fin",
|
161 |
+
"101": "dan",
|
162 |
+
"102": "yid",
|
163 |
+
"103": "est",
|
164 |
+
"104": "ceb",
|
165 |
+
"105": "war",
|
166 |
+
"106": "san",
|
167 |
+
"107": "kir",
|
168 |
+
"108": "oci",
|
169 |
+
"109": "wol",
|
170 |
+
"110": "haw",
|
171 |
+
"111": "kam",
|
172 |
+
"112": "umb",
|
173 |
+
"113": "xho",
|
174 |
+
"114": "epo",
|
175 |
+
"115": "zul",
|
176 |
+
"116": "ibo",
|
177 |
+
"117": "abk",
|
178 |
+
"118": "ckb",
|
179 |
+
"119": "nso",
|
180 |
+
"120": "gle",
|
181 |
+
"121": "kea",
|
182 |
+
"122": "ast",
|
183 |
+
"123": "sco",
|
184 |
+
"124": "glv",
|
185 |
+
"125": "ina"
|
186 |
+
},
|
187 |
+
"initializer_range": 0.02,
|
188 |
+
"intermediate_size": 5120,
|
189 |
+
"label2id": null,
|
190 |
+
"layer_norm_eps": 1e-05,
|
191 |
+
"layerdrop": 0.05,
|
192 |
+
"mask_feature_length": 10,
|
193 |
+
"mask_feature_min_masks": 0,
|
194 |
+
"mask_feature_prob": 0.0,
|
195 |
+
"mask_time_length": 10,
|
196 |
+
"mask_time_min_masks": 2,
|
197 |
+
"mask_time_prob": 0.05,
|
198 |
+
"model_type": "wav2vec2",
|
199 |
+
"num_adapter_layers": 3,
|
200 |
+
"num_attention_heads": 16,
|
201 |
+
"num_codevector_groups": 2,
|
202 |
+
"num_codevectors_per_group": 320,
|
203 |
+
"num_conv_pos_embedding_groups": 16,
|
204 |
+
"num_conv_pos_embeddings": 128,
|
205 |
+
"num_feat_extract_layers": 7,
|
206 |
+
"num_hidden_layers": 48,
|
207 |
+
"num_negatives": 100,
|
208 |
+
"output_hidden_size": 1280,
|
209 |
+
"pad_token_id": 0,
|
210 |
+
"proj_codevector_dim": 1024,
|
211 |
+
"tdnn_dilation": [
|
212 |
+
1,
|
213 |
+
2,
|
214 |
+
3,
|
215 |
+
1,
|
216 |
+
1
|
217 |
+
],
|
218 |
+
"tdnn_dim": [
|
219 |
+
512,
|
220 |
+
512,
|
221 |
+
512,
|
222 |
+
512,
|
223 |
+
1500
|
224 |
+
],
|
225 |
+
"tdnn_kernel": [
|
226 |
+
5,
|
227 |
+
3,
|
228 |
+
3,
|
229 |
+
1,
|
230 |
+
1
|
231 |
+
],
|
232 |
+
"transformers_version": "4.32.0.dev0",
|
233 |
+
"use_weighted_layer_sum": false,
|
234 |
+
"vocab_size": 154,
|
235 |
+
"xvector_output_dim": 512
|
236 |
+
}
|
onnx/model.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4d9175ecafa738bd331698f302cbc58353c1299017e1370287e8a0dd71bc9f9b
|
3 |
+
size 1520416
|
onnx/model.onnx_data
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:40884ebe045b5ce285fed3c8afaaad571a89070dccfc96cb873cfaaefe480f0e
|
3 |
+
size 3863621632
|
onnx/model_quantized.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:91518d5d1d91b224e21a16b55692d1cec28b18d2c0956f3031ff2dc1be42a03c
|
3 |
+
size 973919443
|
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 |
+
"Sqrt",
|
8 |
+
"Div",
|
9 |
+
"Identity",
|
10 |
+
"Reshape",
|
11 |
+
"Gather",
|
12 |
+
"Gemm",
|
13 |
+
"Mul",
|
14 |
+
"Shape",
|
15 |
+
"Conv",
|
16 |
+
"Erf",
|
17 |
+
"Unsqueeze",
|
18 |
+
"Pow",
|
19 |
+
"Add",
|
20 |
+
"Concat",
|
21 |
+
"MatMul",
|
22 |
+
"ReduceMean",
|
23 |
+
"Slice",
|
24 |
+
"Softmax",
|
25 |
+
"Transpose",
|
26 |
+
"Relu",
|
27 |
+
"Sub",
|
28 |
+
"Constant"
|
29 |
+
],
|
30 |
+
"weight_type": "QUInt8"
|
31 |
+
}
|
32 |
+
}
|
33 |
+
}
|