Upload tokenizer
Browse files- special_tokens_map.json +6 -0
- tokenizer_config.json +13 -0
- vocab.json +297 -0
special_tokens_map.json
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"bos_token": "<s>",
|
3 |
+
"eos_token": "</s>",
|
4 |
+
"pad_token": "<pad>",
|
5 |
+
"unk_token": "<unk>"
|
6 |
+
}
|
tokenizer_config.json
ADDED
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"bos_token": "<s>",
|
3 |
+
"clean_up_tokenization_spaces": true,
|
4 |
+
"do_lower_case": false,
|
5 |
+
"eos_token": "</s>",
|
6 |
+
"model_max_length": 1000000000000000019884624838656,
|
7 |
+
"pad_token": "<pad>",
|
8 |
+
"replace_word_delimiter_char": " ",
|
9 |
+
"target_lang": null,
|
10 |
+
"tokenizer_class": "Wav2Vec2CTCTokenizer",
|
11 |
+
"unk_token": "<unk>",
|
12 |
+
"word_delimiter_token": "|"
|
13 |
+
}
|
vocab.json
ADDED
@@ -0,0 +1,297 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"</s>": 2,
|
3 |
+
"<pad>": 0,
|
4 |
+
"<s>": 1,
|
5 |
+
"<unk>": 3,
|
6 |
+
"|": 4,
|
7 |
+
"ア": 5,
|
8 |
+
"アー": 6,
|
9 |
+
"イ": 7,
|
10 |
+
"イェ": 8,
|
11 |
+
"イェー": 9,
|
12 |
+
"イー": 10,
|
13 |
+
"ウ": 11,
|
14 |
+
"ウィ": 12,
|
15 |
+
"ウィー": 13,
|
16 |
+
"ウェ": 14,
|
17 |
+
"ウェー": 15,
|
18 |
+
"ウォ": 16,
|
19 |
+
"ウォー": 17,
|
20 |
+
"ウー": 18,
|
21 |
+
"エ": 19,
|
22 |
+
"エー": 20,
|
23 |
+
"オ": 21,
|
24 |
+
"オー": 22,
|
25 |
+
"カ": 23,
|
26 |
+
"カー": 24,
|
27 |
+
"ガ": 25,
|
28 |
+
"ガー": 26,
|
29 |
+
"キ": 27,
|
30 |
+
"キャ": 28,
|
31 |
+
"キャー": 29,
|
32 |
+
"キュ": 30,
|
33 |
+
"キュー": 31,
|
34 |
+
"キョ": 32,
|
35 |
+
"キョー": 33,
|
36 |
+
"キー": 34,
|
37 |
+
"ギ": 35,
|
38 |
+
"ギャ": 36,
|
39 |
+
"ギャー": 37,
|
40 |
+
"ギュ": 38,
|
41 |
+
"ギュー": 39,
|
42 |
+
"ギョ": 40,
|
43 |
+
"ギョー": 41,
|
44 |
+
"ギー": 42,
|
45 |
+
"ク": 43,
|
46 |
+
"クヮ": 44,
|
47 |
+
"クヮー": 45,
|
48 |
+
"クー": 46,
|
49 |
+
"グ": 47,
|
50 |
+
"グヮ": 48,
|
51 |
+
"グヮー": 49,
|
52 |
+
"グー": 50,
|
53 |
+
"ケ": 51,
|
54 |
+
"ケー": 52,
|
55 |
+
"ゲ": 53,
|
56 |
+
"ゲー": 54,
|
57 |
+
"コ": 55,
|
58 |
+
"コー": 56,
|
59 |
+
"ゴ": 57,
|
60 |
+
"ゴー": 58,
|
61 |
+
"サ": 59,
|
62 |
+
"サー": 60,
|
63 |
+
"ザ": 61,
|
64 |
+
"ザー": 62,
|
65 |
+
"シ": 63,
|
66 |
+
"シェ": 64,
|
67 |
+
"シェー": 65,
|
68 |
+
"シャ": 66,
|
69 |
+
"シャー": 67,
|
70 |
+
"シュ": 68,
|
71 |
+
"シュー": 69,
|
72 |
+
"ショ": 70,
|
73 |
+
"ショー": 71,
|
74 |
+
"シー": 72,
|
75 |
+
"ジ": 73,
|
76 |
+
"ジェ": 74,
|
77 |
+
"ジェー": 75,
|
78 |
+
"ジャ": 76,
|
79 |
+
"ジャー": 77,
|
80 |
+
"ジュ": 78,
|
81 |
+
"ジュー": 79,
|
82 |
+
"ジョ": 80,
|
83 |
+
"ジョー": 81,
|
84 |
+
"ジー": 82,
|
85 |
+
"ス": 83,
|
86 |
+
"スィ": 84,
|
87 |
+
"スィー": 85,
|
88 |
+
"スー": 86,
|
89 |
+
"ズ": 87,
|
90 |
+
"ズィ": 88,
|
91 |
+
"ズィー": 89,
|
92 |
+
"ズー": 90,
|
93 |
+
"セ": 91,
|
94 |
+
"セー": 92,
|
95 |
+
"ゼ": 93,
|
96 |
+
"ゼー": 94,
|
97 |
+
"ソ": 95,
|
98 |
+
"ソー": 96,
|
99 |
+
"ゾ": 97,
|
100 |
+
"ゾー": 98,
|
101 |
+
"タ": 99,
|
102 |
+
"ター": 100,
|
103 |
+
"ダ": 101,
|
104 |
+
"ダー": 102,
|
105 |
+
"チ": 103,
|
106 |
+
"チェ": 104,
|
107 |
+
"チェー": 105,
|
108 |
+
"チャ": 106,
|
109 |
+
"チャー": 107,
|
110 |
+
"チュ": 108,
|
111 |
+
"チュー": 109,
|
112 |
+
"チョ": 110,
|
113 |
+
"チョー": 111,
|
114 |
+
"チー": 112,
|
115 |
+
"ヂ": 113,
|
116 |
+
"ヂャ": 114,
|
117 |
+
"ヂャー": 115,
|
118 |
+
"ヂュ": 116,
|
119 |
+
"ヂュー": 117,
|
120 |
+
"ヂョ": 118,
|
121 |
+
"ヂョー": 119,
|
122 |
+
"ヂー": 120,
|
123 |
+
"ッ": 121,
|
124 |
+
"ッー": 122,
|
125 |
+
"ツ": 123,
|
126 |
+
"ツァ": 124,
|
127 |
+
"ツァー": 125,
|
128 |
+
"ツィ": 126,
|
129 |
+
"ツィー": 127,
|
130 |
+
"ツェ": 128,
|
131 |
+
"ツェー": 129,
|
132 |
+
"ツォ": 130,
|
133 |
+
"ツォー": 131,
|
134 |
+
"ツャ": 132,
|
135 |
+
"ツャー": 133,
|
136 |
+
"ツュ": 134,
|
137 |
+
"ツュー": 135,
|
138 |
+
"ツー": 136,
|
139 |
+
"ヅ": 137,
|
140 |
+
"ヅー": 138,
|
141 |
+
"テ": 139,
|
142 |
+
"ティ": 140,
|
143 |
+
"ティー": 141,
|
144 |
+
"テュ": 142,
|
145 |
+
"テュー": 143,
|
146 |
+
"テー": 144,
|
147 |
+
"デ": 145,
|
148 |
+
"ディ": 146,
|
149 |
+
"ディー": 147,
|
150 |
+
"デュ": 148,
|
151 |
+
"デュー": 149,
|
152 |
+
"デー": 150,
|
153 |
+
"ト": 151,
|
154 |
+
"トゥ": 152,
|
155 |
+
"トゥー": 153,
|
156 |
+
"トー": 154,
|
157 |
+
"ド": 155,
|
158 |
+
"ドゥ": 156,
|
159 |
+
"ドゥー": 157,
|
160 |
+
"ドー": 158,
|
161 |
+
"ナ": 159,
|
162 |
+
"ナー": 160,
|
163 |
+
"ニ": 161,
|
164 |
+
"ニェ": 162,
|
165 |
+
"ニェー": 163,
|
166 |
+
"ニャ": 164,
|
167 |
+
"ニャー": 165,
|
168 |
+
"ニュ": 166,
|
169 |
+
"ニュー": 167,
|
170 |
+
"ニョ": 168,
|
171 |
+
"ニョー": 169,
|
172 |
+
"ニー": 170,
|
173 |
+
"ヌ": 171,
|
174 |
+
"ヌー": 172,
|
175 |
+
"ネ": 173,
|
176 |
+
"ネー": 174,
|
177 |
+
"ノ": 175,
|
178 |
+
"ノー": 176,
|
179 |
+
"ハ": 177,
|
180 |
+
"ハー": 178,
|
181 |
+
"バ": 179,
|
182 |
+
"バー": 180,
|
183 |
+
"パ": 181,
|
184 |
+
"パー": 182,
|
185 |
+
"ヒ": 183,
|
186 |
+
"ヒェ": 184,
|
187 |
+
"ヒェー": 185,
|
188 |
+
"ヒャ": 186,
|
189 |
+
"ヒャー": 187,
|
190 |
+
"ヒュ": 188,
|
191 |
+
"ヒュー": 189,
|
192 |
+
"ヒョ": 190,
|
193 |
+
"ヒョー": 191,
|
194 |
+
"ヒー": 192,
|
195 |
+
"ビ": 193,
|
196 |
+
"ビャ": 194,
|
197 |
+
"ビャー": 195,
|
198 |
+
"ビュ": 196,
|
199 |
+
"ビュー": 197,
|
200 |
+
"ビョ": 198,
|
201 |
+
"ビョー": 199,
|
202 |
+
"ビー": 200,
|
203 |
+
"ピ": 201,
|
204 |
+
"ピャ": 202,
|
205 |
+
"ピャー": 203,
|
206 |
+
"ピュ": 204,
|
207 |
+
"ピュー": 205,
|
208 |
+
"ピョ": 206,
|
209 |
+
"ピョー": 207,
|
210 |
+
"ピー": 208,
|
211 |
+
"フ": 209,
|
212 |
+
"ファ": 210,
|
213 |
+
"ファー": 211,
|
214 |
+
"フィ": 212,
|
215 |
+
"フィー": 213,
|
216 |
+
"フェ": 214,
|
217 |
+
"フェー": 215,
|
218 |
+
"フォ": 216,
|
219 |
+
"フォー": 217,
|
220 |
+
"フュ": 218,
|
221 |
+
"フュー": 219,
|
222 |
+
"フー": 220,
|
223 |
+
"ブ": 221,
|
224 |
+
"ブー": 222,
|
225 |
+
"プ": 223,
|
226 |
+
"プー": 224,
|
227 |
+
"ヘ": 225,
|
228 |
+
"ヘー": 226,
|
229 |
+
"ベ": 227,
|
230 |
+
"ベー": 228,
|
231 |
+
"ペ": 229,
|
232 |
+
"ペー": 230,
|
233 |
+
"ホ": 231,
|
234 |
+
"ホー": 232,
|
235 |
+
"ボ": 233,
|
236 |
+
"ボー": 234,
|
237 |
+
"ポ": 235,
|
238 |
+
"ポー": 236,
|
239 |
+
"マ": 237,
|
240 |
+
"マー": 238,
|
241 |
+
"ミ": 239,
|
242 |
+
"ミェ": 240,
|
243 |
+
"ミェー": 241,
|
244 |
+
"ミャ": 242,
|
245 |
+
"ミャー": 243,
|
246 |
+
"ミュ": 244,
|
247 |
+
"ミュー": 245,
|
248 |
+
"ミョ": 246,
|
249 |
+
"ミョー": 247,
|
250 |
+
"ミー": 248,
|
251 |
+
"ム": 249,
|
252 |
+
"ムー": 250,
|
253 |
+
"メ": 251,
|
254 |
+
"メー": 252,
|
255 |
+
"モ": 253,
|
256 |
+
"モー": 254,
|
257 |
+
"ヤ": 255,
|
258 |
+
"ヤー": 256,
|
259 |
+
"ユ": 257,
|
260 |
+
"ユー": 258,
|
261 |
+
"ヨ": 259,
|
262 |
+
"ヨー": 260,
|
263 |
+
"ラ": 261,
|
264 |
+
"ラー": 262,
|
265 |
+
"リ": 263,
|
266 |
+
"リャ": 264,
|
267 |
+
"リャー": 265,
|
268 |
+
"リュ": 266,
|
269 |
+
"リュー": 267,
|
270 |
+
"リョ": 268,
|
271 |
+
"リョー": 269,
|
272 |
+
"リー": 270,
|
273 |
+
"ル": 271,
|
274 |
+
"ルー": 272,
|
275 |
+
"レ": 273,
|
276 |
+
"レー": 274,
|
277 |
+
"ロ": 275,
|
278 |
+
"ロー": 276,
|
279 |
+
"ワ": 277,
|
280 |
+
"ワー": 278,
|
281 |
+
"ヲ": 279,
|
282 |
+
"ヲー": 280,
|
283 |
+
"ン": 281,
|
284 |
+
"ンー": 282,
|
285 |
+
"ヴ": 283,
|
286 |
+
"ヴァ": 284,
|
287 |
+
"ヴァー": 285,
|
288 |
+
"ヴィ": 286,
|
289 |
+
"ヴィー": 287,
|
290 |
+
"ヴェ": 288,
|
291 |
+
"ヴェー": 289,
|
292 |
+
"ヴォ": 290,
|
293 |
+
"ヴォー": 291,
|
294 |
+
"ヴュ": 292,
|
295 |
+
"ヴュー": 293,
|
296 |
+
"ヴー": 294
|
297 |
+
}
|