metythorn commited on
Commit
0588e0c
·
verified ·
1 Parent(s): c51324c

Upload config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. config.json +1378 -0
config.json ADDED
@@ -0,0 +1,1378 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "hyperparameters": {
3
+ "data_file": "/home/metythorn/data/metythorn/konai/services/ocr-service/data/processed/text-recognition/ocr_dataset_v5.txt",
4
+ "img_height": 128,
5
+ "patch_size": 16,
6
+ "dim_feedforward": 3072,
7
+ "dropout": 0.1,
8
+ "augment": false,
9
+ "max_decode_len": 289,
10
+ "quiet": false,
11
+ "cer_samples": 512,
12
+ "device": "cuda:1",
13
+ "seed": 42,
14
+ "val_ratio": 0.05,
15
+ "num_workers": 4,
16
+ "pin_memory": true,
17
+ "grad_clip": 1.0,
18
+ "weight_decay": 0.01,
19
+ "label_smoothing": 0.05,
20
+ "use_amp": true,
21
+ "lr_warmup_ratio": 0.1,
22
+ "min_lr_ratio": 0.1,
23
+ "early_stopping_patience": 5,
24
+ "img_width": 384,
25
+ "batch_size": 90,
26
+ "lr": 8e-05,
27
+ "epochs": 8,
28
+ "d_model": 768,
29
+ "nhead": 12,
30
+ "num_layers": 10,
31
+ "checkpoint_dir": "logs/ViTOCR_base_khmerid_v1",
32
+ "model_size": "base",
33
+ "config_path": "logs/ViTOCR_base_khmerid_v1/config.json"
34
+ },
35
+ "vocab": {
36
+ "specials": [
37
+ "<PAD>",
38
+ "<SOS>",
39
+ "<EOS>"
40
+ ],
41
+ "char2idx": {
42
+ "<PAD>": 0,
43
+ "<SOS>": 1,
44
+ "<EOS>": 2,
45
+ "\u0006": 3,
46
+ " ": 4,
47
+ "!": 5,
48
+ "\"": 6,
49
+ "#": 7,
50
+ "$": 8,
51
+ "%": 9,
52
+ "&": 10,
53
+ "'": 11,
54
+ "(": 12,
55
+ ")": 13,
56
+ "*": 14,
57
+ "+": 15,
58
+ ",": 16,
59
+ "-": 17,
60
+ ".": 18,
61
+ "/": 19,
62
+ "0": 20,
63
+ "1": 21,
64
+ "2": 22,
65
+ "3": 23,
66
+ "4": 24,
67
+ "5": 25,
68
+ "6": 26,
69
+ "7": 27,
70
+ "8": 28,
71
+ "9": 29,
72
+ ":": 30,
73
+ ";": 31,
74
+ "<": 32,
75
+ "=": 33,
76
+ ">": 34,
77
+ "?": 35,
78
+ "@": 36,
79
+ "A": 37,
80
+ "B": 38,
81
+ "C": 39,
82
+ "D": 40,
83
+ "E": 41,
84
+ "F": 42,
85
+ "G": 43,
86
+ "H": 44,
87
+ "I": 45,
88
+ "J": 46,
89
+ "K": 47,
90
+ "L": 48,
91
+ "M": 49,
92
+ "N": 50,
93
+ "O": 51,
94
+ "P": 52,
95
+ "Q": 53,
96
+ "R": 54,
97
+ "S": 55,
98
+ "T": 56,
99
+ "U": 57,
100
+ "V": 58,
101
+ "W": 59,
102
+ "X": 60,
103
+ "Y": 61,
104
+ "Z": 62,
105
+ "[": 63,
106
+ "\\": 64,
107
+ "]": 65,
108
+ "^": 66,
109
+ "_": 67,
110
+ "`": 68,
111
+ "a": 69,
112
+ "b": 70,
113
+ "c": 71,
114
+ "d": 72,
115
+ "e": 73,
116
+ "f": 74,
117
+ "g": 75,
118
+ "h": 76,
119
+ "i": 77,
120
+ "j": 78,
121
+ "k": 79,
122
+ "l": 80,
123
+ "m": 81,
124
+ "n": 82,
125
+ "o": 83,
126
+ "p": 84,
127
+ "q": 85,
128
+ "r": 86,
129
+ "s": 87,
130
+ "t": 88,
131
+ "u": 89,
132
+ "v": 90,
133
+ "w": 91,
134
+ "x": 92,
135
+ "y": 93,
136
+ "z": 94,
137
+ "{": 95,
138
+ "|": 96,
139
+ "}": 97,
140
+ "~": 98,
141
+ "¡": 99,
142
+ "¢": 100,
143
+ "£": 101,
144
+ "¤": 102,
145
+ "¥": 103,
146
+ "¦": 104,
147
+ "§": 105,
148
+ "¨": 106,
149
+ "©": 107,
150
+ "ª": 108,
151
+ "«": 109,
152
+ "¬": 110,
153
+ "­": 111,
154
+ "®": 112,
155
+ "°": 113,
156
+ "±": 114,
157
+ "²": 115,
158
+ "³": 116,
159
+ "´": 117,
160
+ "µ": 118,
161
+ "¶": 119,
162
+ "·": 120,
163
+ "¹": 121,
164
+ "»": 122,
165
+ "¼": 123,
166
+ "½": 124,
167
+ "¿": 125,
168
+ "À": 126,
169
+ "Á": 127,
170
+ "Â": 128,
171
+ "Ã": 129,
172
+ "Ä": 130,
173
+ "Å": 131,
174
+ "Æ": 132,
175
+ "Ç": 133,
176
+ "È": 134,
177
+ "É": 135,
178
+ "Ê": 136,
179
+ "Ë": 137,
180
+ "Ì": 138,
181
+ "Í": 139,
182
+ "Î": 140,
183
+ "Ï": 141,
184
+ "Ð": 142,
185
+ "Ñ": 143,
186
+ "Ò": 144,
187
+ "Ó": 145,
188
+ "Ô": 146,
189
+ "Ö": 147,
190
+ "×": 148,
191
+ "Ø": 149,
192
+ "Ù": 150,
193
+ "Ú": 151,
194
+ "Ü": 152,
195
+ "Þ": 153,
196
+ "ß": 154,
197
+ "à": 155,
198
+ "á": 156,
199
+ "â": 157,
200
+ "ã": 158,
201
+ "ä": 159,
202
+ "å": 160,
203
+ "æ": 161,
204
+ "ç": 162,
205
+ "è": 163,
206
+ "é": 164,
207
+ "ê": 165,
208
+ "ë": 166,
209
+ "ì": 167,
210
+ "í": 168,
211
+ "î": 169,
212
+ "ï": 170,
213
+ "ð": 171,
214
+ "ñ": 172,
215
+ "ò": 173,
216
+ "ó": 174,
217
+ "ô": 175,
218
+ "õ": 176,
219
+ "ö": 177,
220
+ "÷": 178,
221
+ "ø": 179,
222
+ "ù": 180,
223
+ "ú": 181,
224
+ "û": 182,
225
+ "ü": 183,
226
+ "ý": 184,
227
+ "þ": 185,
228
+ "Ā": 186,
229
+ "ā": 187,
230
+ "Ă": 188,
231
+ "ă": 189,
232
+ "ą": 190,
233
+ "ć": 191,
234
+ "Č": 192,
235
+ "č": 193,
236
+ "Đ": 194,
237
+ "đ": 195,
238
+ "Ē": 196,
239
+ "ē": 197,
240
+ "Ĕ": 198,
241
+ "ĕ": 199,
242
+ "ė": 200,
243
+ "ę": 201,
244
+ "ě": 202,
245
+ "Ğ": 203,
246
+ "ğ": 204,
247
+ "ĩ": 205,
248
+ "Ī": 206,
249
+ "ī": 207,
250
+ "Ĭ": 208,
251
+ "ĭ": 209,
252
+ "İ": 210,
253
+ "ı": 211,
254
+ "Ľ": 212,
255
+ "Ł": 213,
256
+ "ł": 214,
257
+ "ń": 215,
258
+ "ņ": 216,
259
+ "Ŋ": 217,
260
+ "ŋ": 218,
261
+ "Ō": 219,
262
+ "ō": 220,
263
+ "ŏ": 221,
264
+ "Œ": 222,
265
+ "œ": 223,
266
+ "Ś": 224,
267
+ "ś": 225,
268
+ "Ş": 226,
269
+ "ş": 227,
270
+ "Š": 228,
271
+ "š": 229,
272
+ "ţ": 230,
273
+ "Ũ": 231,
274
+ "ũ": 232,
275
+ "Ū": 233,
276
+ "ū": 234,
277
+ "Ŭ": 235,
278
+ "ŭ": 236,
279
+ "ů": 237,
280
+ "ź": 238,
281
+ "Ż": 239,
282
+ "ż": 240,
283
+ "Ž": 241,
284
+ "ž": 242,
285
+ "Ɔ": 243,
286
+ "Ə": 244,
287
+ "ƒ": 245,
288
+ "Ɨ": 246,
289
+ "Ơ": 247,
290
+ "ơ": 248,
291
+ "Ư": 249,
292
+ "ư": 250,
293
+ "Ʊ": 251,
294
+ "Ʒ": 252,
295
+ "ǎ": 253,
296
+ "ǐ": 254,
297
+ "ǒ": 255,
298
+ "ǣ": 256,
299
+ "ȳ": 257,
300
+ "ɐ": 258,
301
+ "ɑ": 259,
302
+ "ɔ": 260,
303
+ "ɕ": 261,
304
+ "ɗ": 262,
305
+ "ə": 263,
306
+ "ɛ": 264,
307
+ "ɡ": 265,
308
+ "ɦ": 266,
309
+ "ɨ": 267,
310
+ "ɪ": 268,
311
+ "ɬ": 269,
312
+ "ɯ": 270,
313
+ "ɲ": 271,
314
+ "ɽ": 272,
315
+ "ʄ": 273,
316
+ "ʊ": 274,
317
+ "ʍ": 275,
318
+ "ʒ": 276,
319
+ "ʔ": 277,
320
+ "ʰ": 278,
321
+ "ˈ": 279,
322
+ "ː": 280,
323
+ "̀": 281,
324
+ "́": 282,
325
+ "̄": 283,
326
+ "̆": 284,
327
+ "̋": 285,
328
+ "̐": 286,
329
+ "̥": 287,
330
+ "̰": 288,
331
+ "̲": 289,
332
+ "͡": 290,
333
+ "Ί": 291,
334
+ "Α": 292,
335
+ "Δ": 293,
336
+ "Θ": 294,
337
+ "Ι": 295,
338
+ "Λ": 296,
339
+ "Ο": 297,
340
+ "Π": 298,
341
+ "Σ": 299,
342
+ "Τ": 300,
343
+ "Υ": 301,
344
+ "Ω": 302,
345
+ "ί": 303,
346
+ "α": 304,
347
+ "β": 305,
348
+ "γ": 306,
349
+ "δ": 307,
350
+ "ε": 308,
351
+ "θ": 309,
352
+ "ι": 310,
353
+ "κ": 311,
354
+ "λ": 312,
355
+ "μ": 313,
356
+ "ν": 314,
357
+ "ξ": 315,
358
+ "ο": 316,
359
+ "π": 317,
360
+ "ρ": 318,
361
+ "ς": 319,
362
+ "σ": 320,
363
+ "τ": 321,
364
+ "υ": 322,
365
+ "φ": 323,
366
+ "χ": 324,
367
+ "ό": 325,
368
+ "А": 326,
369
+ "Б": 327,
370
+ "В": 328,
371
+ "Г": 329,
372
+ "Д": 330,
373
+ "Е": 331,
374
+ "З": 332,
375
+ "И": 333,
376
+ "К": 334,
377
+ "Л": 335,
378
+ "М": 336,
379
+ "Н": 337,
380
+ "О": 338,
381
+ "П": 339,
382
+ "Р": 340,
383
+ "С": 341,
384
+ "Т": 342,
385
+ "У": 343,
386
+ "Ф": 344,
387
+ "Ц": 345,
388
+ "Ы": 346,
389
+ "Ь": 347,
390
+ "Э": 348,
391
+ "а": 349,
392
+ "б": 350,
393
+ "в": 351,
394
+ "г": 352,
395
+ "д": 353,
396
+ "е": 354,
397
+ "ж": 355,
398
+ "з": 356,
399
+ "и": 357,
400
+ "й": 358,
401
+ "к": 359,
402
+ "л": 360,
403
+ "м": 361,
404
+ "н": 362,
405
+ "о": 363,
406
+ "п": 364,
407
+ "р": 365,
408
+ "с": 366,
409
+ "т": 367,
410
+ "у": 368,
411
+ "ф": 369,
412
+ "х": 370,
413
+ "ц": 371,
414
+ "ч": 372,
415
+ "ш": 373,
416
+ "щ": 374,
417
+ "ъ": 375,
418
+ "ы": 376,
419
+ "ь": 377,
420
+ "э": 378,
421
+ "ю": 379,
422
+ "я": 380,
423
+ "ё": 381,
424
+ "ѕ": 382,
425
+ "і": 383,
426
+ "ї": 384,
427
+ "ј": 385,
428
+ "ѹ": 386,
429
+ "ү": 387,
430
+ "ԁ": 388,
431
+ "آ": 389,
432
+ "ا": 390,
433
+ "ب": 391,
434
+ "ت": 392,
435
+ "ر": 393,
436
+ "ز": 394,
437
+ "س": 395,
438
+ "غ": 396,
439
+ "ل": 397,
440
+ "م": 398,
441
+ "ن": 399,
442
+ "ه": 400,
443
+ "و": 401,
444
+ "پ": 402,
445
+ "ک": 403,
446
+ "ی": 404,
447
+ "۵": 405,
448
+ "़": 406,
449
+ "ટ": 407,
450
+ "સ": 408,
451
+ "ે": 409,
452
+ "્": 410,
453
+ "ิ": 411,
454
+ "ฺ": 412,
455
+ "้": 413,
456
+ "๋": 414,
457
+ "ါ": 415,
458
+ "ာ": 416,
459
+ "ေ": 417,
460
+ "ွ": 418,
461
+ "ក": 419,
462
+ "ខ": 420,
463
+ "គ": 421,
464
+ "ឃ": 422,
465
+ "ង": 423,
466
+ "ច": 424,
467
+ "ឆ": 425,
468
+ "ជ": 426,
469
+ "ឈ": 427,
470
+ "ញ": 428,
471
+ "ដ": 429,
472
+ "ឋ": 430,
473
+ "ឌ": 431,
474
+ "ឍ": 432,
475
+ "ណ": 433,
476
+ "ត": 434,
477
+ "ថ": 435,
478
+ "ទ": 436,
479
+ "ធ": 437,
480
+ "ន": 438,
481
+ "ប": 439,
482
+ "ផ": 440,
483
+ "ព": 441,
484
+ "ភ": 442,
485
+ "ម": 443,
486
+ "យ": 444,
487
+ "រ": 445,
488
+ "ល": 446,
489
+ "វ": 447,
490
+ "ឝ": 448,
491
+ "ឞ": 449,
492
+ "ស": 450,
493
+ "ហ": 451,
494
+ "ឡ": 452,
495
+ "អ": 453,
496
+ "ឣ": 454,
497
+ "ឤ": 455,
498
+ "ឥ": 456,
499
+ "ឦ": 457,
500
+ "ឧ": 458,
501
+ "ឨ": 459,
502
+ "ឩ": 460,
503
+ "ឪ": 461,
504
+ "ឫ": 462,
505
+ "ឬ": 463,
506
+ "ឭ": 464,
507
+ "ឮ": 465,
508
+ "ឯ": 466,
509
+ "ឰ": 467,
510
+ "ឱ": 468,
511
+ "ឲ": 469,
512
+ "ឳ": 470,
513
+ "឵": 471,
514
+ "ា": 472,
515
+ "ិ": 473,
516
+ "ី": 474,
517
+ "ឹ": 475,
518
+ "ឺ": 476,
519
+ "ុ": 477,
520
+ "ូ": 478,
521
+ "ួ": 479,
522
+ "ើ": 480,
523
+ "ឿ": 481,
524
+ "ៀ": 482,
525
+ "េ": 483,
526
+ "ែ": 484,
527
+ "ៃ": 485,
528
+ "ោ": 486,
529
+ "ៅ": 487,
530
+ "ំ": 488,
531
+ "ះ": 489,
532
+ "ៈ": 490,
533
+ "៉": 491,
534
+ "៊": 492,
535
+ "់": 493,
536
+ "៌": 494,
537
+ "៍": 495,
538
+ "៎": 496,
539
+ "៏": 497,
540
+ "័": 498,
541
+ "៑": 499,
542
+ "្": 500,
543
+ "៓": 501,
544
+ "។": 502,
545
+ "៕": 503,
546
+ "៖": 504,
547
+ "ៗ": 505,
548
+ "៘": 506,
549
+ "៙": 507,
550
+ "៚": 508,
551
+ "៛": 509,
552
+ "៝": 510,
553
+ "០": 511,
554
+ "១": 512,
555
+ "២": 513,
556
+ "៣": 514,
557
+ "៤": 515,
558
+ "៥": 516,
559
+ "៦": 517,
560
+ "៧": 518,
561
+ "៨": 519,
562
+ "៩": 520,
563
+ "៰": 521,
564
+ "៱": 522,
565
+ "៲": 523,
566
+ "៸": 524,
567
+ "៹": 525,
568
+ "᧥": 526,
569
+ "᧦": 527,
570
+ "᧯": 528,
571
+ "᧰": 529,
572
+ "ḍ": 530,
573
+ "ḓ": 531,
574
+ "ḛ": 532,
575
+ "ḥ": 533,
576
+ "Ḳ": 534,
577
+ "ḷ": 535,
578
+ "ṁ": 536,
579
+ "Ṃ": 537,
580
+ "ṃ": 538,
581
+ "Ṅ": 539,
582
+ "ṅ": 540,
583
+ "Ṇ": 541,
584
+ "ṇ": 542,
585
+ "ṛ": 543,
586
+ "Ṣ": 544,
587
+ "ṣ": 545,
588
+ "Ṭ": 546,
589
+ "ṭ": 547,
590
+ "Ạ": 548,
591
+ "ạ": 549,
592
+ "Ả": 550,
593
+ "ả": 551,
594
+ "Ấ": 552,
595
+ "ấ": 553,
596
+ "Ầ": 554,
597
+ "ầ": 555,
598
+ "ẩ": 556,
599
+ "Ậ": 557,
600
+ "ậ": 558,
601
+ "ắ": 559,
602
+ "ẳ": 560,
603
+ "ặ": 561,
604
+ "Ẹ": 562,
605
+ "ẹ": 563,
606
+ "ẻ": 564,
607
+ "ế": 565,
608
+ "Ề": 566,
609
+ "ề": 567,
610
+ "Ể": 568,
611
+ "ể": 569,
612
+ "ễ": 570,
613
+ "ệ": 571,
614
+ "Ỉ": 572,
615
+ "ỉ": 573,
616
+ "Ị": 574,
617
+ "ị": 575,
618
+ "Ọ": 576,
619
+ "ọ": 577,
620
+ "ỏ": 578,
621
+ "ố": 579,
622
+ "ồ": 580,
623
+ "ổ": 581,
624
+ "ỗ": 582,
625
+ "Ộ": 583,
626
+ "ộ": 584,
627
+ "Ớ": 585,
628
+ "ớ": 586,
629
+ "Ờ": 587,
630
+ "ờ": 588,
631
+ "ợ": 589,
632
+ "ụ": 590,
633
+ "Ủ": 591,
634
+ "ủ": 592,
635
+ "Ứ": 593,
636
+ "ứ": 594,
637
+ "ừ": 595,
638
+ "ử": 596,
639
+ "Ữ": 597,
640
+ "ữ": 598,
641
+ "ự": 599,
642
+ "ỳ": 600,
643
+ "ỹ": 601,
644
+ "​": 602,
645
+ "‌": 603,
646
+ "‍": 604,
647
+ "‎": 605,
648
+ "‏": 606,
649
+ "‐": 607,
650
+ "‑": 608,
651
+ "–": 609,
652
+ "—": 610,
653
+ "‘": 611,
654
+ "’": 612,
655
+ "“": 613,
656
+ "”": 614,
657
+ "„": 615,
658
+ "†": 616,
659
+ "•": 617,
660
+ "…": 618,
661
+ "‪": 619,
662
+ "‫": 620,
663
+ "‬": 621,
664
+ "′": 622,
665
+ "″": 623,
666
+ "‹": 624,
667
+ "›": 625,
668
+ "⁄": 626,
669
+ "⁠": 627,
670
+ "⁡": 628,
671
+ "⁣": 629,
672
+ "€": 630,
673
+ "№": 631,
674
+ "™": 632,
675
+ "←": 633,
676
+ "↑": 634,
677
+ "→": 635,
678
+ "↓": 636,
679
+ "−": 637,
680
+ "∙": 638,
681
+ "∞": 639,
682
+ "∫": 640,
683
+ "≈": 641,
684
+ "≤": 642,
685
+ "≥": 643,
686
+ "⋅": 644,
687
+ "▪": 645,
688
+ "►": 646,
689
+ "▼": 647,
690
+ "▾": 648,
691
+ "◌": 649,
692
+ "●": 650,
693
+ "⚫": 651,
694
+ "✅": 652,
695
+ "⟶": 653,
696
+ "伍": 654,
697
+ "和": 655,
698
+ "德": 656,
699
+ "連": 657,
700
+ "郑": 658,
701
+ "Ɡ": 659,
702
+ "꫁": 660,
703
+ "": 661,
704
+ "": 662,
705
+ ";": 663,
706
+ "�": 664,
707
+ "🎉": 665
708
+ },
709
+ "idx2char": {
710
+ "0": "<PAD>",
711
+ "1": "<SOS>",
712
+ "2": "<EOS>",
713
+ "3": "\u0006",
714
+ "4": " ",
715
+ "5": "!",
716
+ "6": "\"",
717
+ "7": "#",
718
+ "8": "$",
719
+ "9": "%",
720
+ "10": "&",
721
+ "11": "'",
722
+ "12": "(",
723
+ "13": ")",
724
+ "14": "*",
725
+ "15": "+",
726
+ "16": ",",
727
+ "17": "-",
728
+ "18": ".",
729
+ "19": "/",
730
+ "20": "0",
731
+ "21": "1",
732
+ "22": "2",
733
+ "23": "3",
734
+ "24": "4",
735
+ "25": "5",
736
+ "26": "6",
737
+ "27": "7",
738
+ "28": "8",
739
+ "29": "9",
740
+ "30": ":",
741
+ "31": ";",
742
+ "32": "<",
743
+ "33": "=",
744
+ "34": ">",
745
+ "35": "?",
746
+ "36": "@",
747
+ "37": "A",
748
+ "38": "B",
749
+ "39": "C",
750
+ "40": "D",
751
+ "41": "E",
752
+ "42": "F",
753
+ "43": "G",
754
+ "44": "H",
755
+ "45": "I",
756
+ "46": "J",
757
+ "47": "K",
758
+ "48": "L",
759
+ "49": "M",
760
+ "50": "N",
761
+ "51": "O",
762
+ "52": "P",
763
+ "53": "Q",
764
+ "54": "R",
765
+ "55": "S",
766
+ "56": "T",
767
+ "57": "U",
768
+ "58": "V",
769
+ "59": "W",
770
+ "60": "X",
771
+ "61": "Y",
772
+ "62": "Z",
773
+ "63": "[",
774
+ "64": "\\",
775
+ "65": "]",
776
+ "66": "^",
777
+ "67": "_",
778
+ "68": "`",
779
+ "69": "a",
780
+ "70": "b",
781
+ "71": "c",
782
+ "72": "d",
783
+ "73": "e",
784
+ "74": "f",
785
+ "75": "g",
786
+ "76": "h",
787
+ "77": "i",
788
+ "78": "j",
789
+ "79": "k",
790
+ "80": "l",
791
+ "81": "m",
792
+ "82": "n",
793
+ "83": "o",
794
+ "84": "p",
795
+ "85": "q",
796
+ "86": "r",
797
+ "87": "s",
798
+ "88": "t",
799
+ "89": "u",
800
+ "90": "v",
801
+ "91": "w",
802
+ "92": "x",
803
+ "93": "y",
804
+ "94": "z",
805
+ "95": "{",
806
+ "96": "|",
807
+ "97": "}",
808
+ "98": "~",
809
+ "99": "¡",
810
+ "100": "¢",
811
+ "101": "£",
812
+ "102": "¤",
813
+ "103": "¥",
814
+ "104": "¦",
815
+ "105": "§",
816
+ "106": "¨",
817
+ "107": "©",
818
+ "108": "ª",
819
+ "109": "«",
820
+ "110": "¬",
821
+ "111": "­",
822
+ "112": "®",
823
+ "113": "°",
824
+ "114": "±",
825
+ "115": "²",
826
+ "116": "³",
827
+ "117": "´",
828
+ "118": "µ",
829
+ "119": "¶",
830
+ "120": "·",
831
+ "121": "¹",
832
+ "122": "»",
833
+ "123": "¼",
834
+ "124": "½",
835
+ "125": "¿",
836
+ "126": "À",
837
+ "127": "Á",
838
+ "128": "Â",
839
+ "129": "Ã",
840
+ "130": "Ä",
841
+ "131": "Å",
842
+ "132": "Æ",
843
+ "133": "Ç",
844
+ "134": "È",
845
+ "135": "É",
846
+ "136": "Ê",
847
+ "137": "Ë",
848
+ "138": "Ì",
849
+ "139": "Í",
850
+ "140": "Î",
851
+ "141": "Ï",
852
+ "142": "Ð",
853
+ "143": "Ñ",
854
+ "144": "Ò",
855
+ "145": "Ó",
856
+ "146": "Ô",
857
+ "147": "Ö",
858
+ "148": "×",
859
+ "149": "Ø",
860
+ "150": "Ù",
861
+ "151": "Ú",
862
+ "152": "Ü",
863
+ "153": "Þ",
864
+ "154": "ß",
865
+ "155": "à",
866
+ "156": "á",
867
+ "157": "â",
868
+ "158": "ã",
869
+ "159": "ä",
870
+ "160": "å",
871
+ "161": "æ",
872
+ "162": "ç",
873
+ "163": "è",
874
+ "164": "é",
875
+ "165": "ê",
876
+ "166": "ë",
877
+ "167": "ì",
878
+ "168": "í",
879
+ "169": "î",
880
+ "170": "ï",
881
+ "171": "ð",
882
+ "172": "ñ",
883
+ "173": "ò",
884
+ "174": "ó",
885
+ "175": "ô",
886
+ "176": "õ",
887
+ "177": "ö",
888
+ "178": "÷",
889
+ "179": "ø",
890
+ "180": "ù",
891
+ "181": "ú",
892
+ "182": "û",
893
+ "183": "ü",
894
+ "184": "ý",
895
+ "185": "þ",
896
+ "186": "Ā",
897
+ "187": "ā",
898
+ "188": "Ă",
899
+ "189": "ă",
900
+ "190": "ą",
901
+ "191": "ć",
902
+ "192": "Č",
903
+ "193": "č",
904
+ "194": "Đ",
905
+ "195": "đ",
906
+ "196": "Ē",
907
+ "197": "ē",
908
+ "198": "Ĕ",
909
+ "199": "ĕ",
910
+ "200": "ė",
911
+ "201": "ę",
912
+ "202": "ě",
913
+ "203": "Ğ",
914
+ "204": "ğ",
915
+ "205": "ĩ",
916
+ "206": "Ī",
917
+ "207": "ī",
918
+ "208": "Ĭ",
919
+ "209": "ĭ",
920
+ "210": "İ",
921
+ "211": "ı",
922
+ "212": "Ľ",
923
+ "213": "Ł",
924
+ "214": "ł",
925
+ "215": "ń",
926
+ "216": "ņ",
927
+ "217": "Ŋ",
928
+ "218": "ŋ",
929
+ "219": "Ō",
930
+ "220": "ō",
931
+ "221": "ŏ",
932
+ "222": "Œ",
933
+ "223": "œ",
934
+ "224": "Ś",
935
+ "225": "ś",
936
+ "226": "Ş",
937
+ "227": "ş",
938
+ "228": "Š",
939
+ "229": "š",
940
+ "230": "ţ",
941
+ "231": "Ũ",
942
+ "232": "ũ",
943
+ "233": "Ū",
944
+ "234": "ū",
945
+ "235": "Ŭ",
946
+ "236": "ŭ",
947
+ "237": "ů",
948
+ "238": "ź",
949
+ "239": "Ż",
950
+ "240": "ż",
951
+ "241": "Ž",
952
+ "242": "ž",
953
+ "243": "Ɔ",
954
+ "244": "Ə",
955
+ "245": "ƒ",
956
+ "246": "Ɨ",
957
+ "247": "Ơ",
958
+ "248": "ơ",
959
+ "249": "Ư",
960
+ "250": "ư",
961
+ "251": "Ʊ",
962
+ "252": "Ʒ",
963
+ "253": "ǎ",
964
+ "254": "ǐ",
965
+ "255": "ǒ",
966
+ "256": "ǣ",
967
+ "257": "ȳ",
968
+ "258": "ɐ",
969
+ "259": "ɑ",
970
+ "260": "ɔ",
971
+ "261": "ɕ",
972
+ "262": "ɗ",
973
+ "263": "ə",
974
+ "264": "ɛ",
975
+ "265": "ɡ",
976
+ "266": "ɦ",
977
+ "267": "ɨ",
978
+ "268": "ɪ",
979
+ "269": "ɬ",
980
+ "270": "ɯ",
981
+ "271": "ɲ",
982
+ "272": "ɽ",
983
+ "273": "ʄ",
984
+ "274": "ʊ",
985
+ "275": "ʍ",
986
+ "276": "ʒ",
987
+ "277": "ʔ",
988
+ "278": "ʰ",
989
+ "279": "ˈ",
990
+ "280": "ː",
991
+ "281": "̀",
992
+ "282": "́",
993
+ "283": "̄",
994
+ "284": "̆",
995
+ "285": "̋",
996
+ "286": "̐",
997
+ "287": "̥",
998
+ "288": "̰",
999
+ "289": "̲",
1000
+ "290": "͡",
1001
+ "291": "Ί",
1002
+ "292": "Α",
1003
+ "293": "Δ",
1004
+ "294": "Θ",
1005
+ "295": "Ι",
1006
+ "296": "Λ",
1007
+ "297": "Ο",
1008
+ "298": "Π",
1009
+ "299": "Σ",
1010
+ "300": "Τ",
1011
+ "301": "Υ",
1012
+ "302": "Ω",
1013
+ "303": "ί",
1014
+ "304": "α",
1015
+ "305": "β",
1016
+ "306": "γ",
1017
+ "307": "δ",
1018
+ "308": "ε",
1019
+ "309": "θ",
1020
+ "310": "ι",
1021
+ "311": "κ",
1022
+ "312": "λ",
1023
+ "313": "μ",
1024
+ "314": "ν",
1025
+ "315": "ξ",
1026
+ "316": "ο",
1027
+ "317": "π",
1028
+ "318": "ρ",
1029
+ "319": "ς",
1030
+ "320": "σ",
1031
+ "321": "τ",
1032
+ "322": "υ",
1033
+ "323": "φ",
1034
+ "324": "χ",
1035
+ "325": "ό",
1036
+ "326": "А",
1037
+ "327": "Б",
1038
+ "328": "В",
1039
+ "329": "Г",
1040
+ "330": "Д",
1041
+ "331": "Е",
1042
+ "332": "З",
1043
+ "333": "И",
1044
+ "334": "К",
1045
+ "335": "Л",
1046
+ "336": "М",
1047
+ "337": "Н",
1048
+ "338": "О",
1049
+ "339": "П",
1050
+ "340": "Р",
1051
+ "341": "С",
1052
+ "342": "Т",
1053
+ "343": "У",
1054
+ "344": "Ф",
1055
+ "345": "Ц",
1056
+ "346": "Ы",
1057
+ "347": "Ь",
1058
+ "348": "Э",
1059
+ "349": "а",
1060
+ "350": "б",
1061
+ "351": "в",
1062
+ "352": "г",
1063
+ "353": "д",
1064
+ "354": "е",
1065
+ "355": "ж",
1066
+ "356": "з",
1067
+ "357": "и",
1068
+ "358": "й",
1069
+ "359": "к",
1070
+ "360": "л",
1071
+ "361": "м",
1072
+ "362": "н",
1073
+ "363": "о",
1074
+ "364": "п",
1075
+ "365": "р",
1076
+ "366": "с",
1077
+ "367": "т",
1078
+ "368": "у",
1079
+ "369": "ф",
1080
+ "370": "х",
1081
+ "371": "ц",
1082
+ "372": "ч",
1083
+ "373": "ш",
1084
+ "374": "щ",
1085
+ "375": "ъ",
1086
+ "376": "ы",
1087
+ "377": "ь",
1088
+ "378": "э",
1089
+ "379": "ю",
1090
+ "380": "я",
1091
+ "381": "ё",
1092
+ "382": "ѕ",
1093
+ "383": "і",
1094
+ "384": "ї",
1095
+ "385": "ј",
1096
+ "386": "ѹ",
1097
+ "387": "ү",
1098
+ "388": "ԁ",
1099
+ "389": "آ",
1100
+ "390": "ا",
1101
+ "391": "ب",
1102
+ "392": "ت",
1103
+ "393": "ر",
1104
+ "394": "ز",
1105
+ "395": "س",
1106
+ "396": "غ",
1107
+ "397": "ل",
1108
+ "398": "م",
1109
+ "399": "ن",
1110
+ "400": "ه",
1111
+ "401": "و",
1112
+ "402": "پ",
1113
+ "403": "ک",
1114
+ "404": "ی",
1115
+ "405": "۵",
1116
+ "406": "़",
1117
+ "407": "ટ",
1118
+ "408": "સ",
1119
+ "409": "ે",
1120
+ "410": "્",
1121
+ "411": "ิ",
1122
+ "412": "ฺ",
1123
+ "413": "้",
1124
+ "414": "๋",
1125
+ "415": "ါ",
1126
+ "416": "ာ",
1127
+ "417": "ေ",
1128
+ "418": "ွ",
1129
+ "419": "ក",
1130
+ "420": "ខ",
1131
+ "421": "គ",
1132
+ "422": "ឃ",
1133
+ "423": "ង",
1134
+ "424": "ច",
1135
+ "425": "ឆ",
1136
+ "426": "ជ",
1137
+ "427": "ឈ",
1138
+ "428": "ញ",
1139
+ "429": "ដ",
1140
+ "430": "ឋ",
1141
+ "431": "ឌ",
1142
+ "432": "ឍ",
1143
+ "433": "ណ",
1144
+ "434": "ត",
1145
+ "435": "ថ",
1146
+ "436": "ទ",
1147
+ "437": "ធ",
1148
+ "438": "ន",
1149
+ "439": "ប",
1150
+ "440": "ផ",
1151
+ "441": "ព",
1152
+ "442": "ភ",
1153
+ "443": "ម",
1154
+ "444": "យ",
1155
+ "445": "រ",
1156
+ "446": "ល",
1157
+ "447": "វ",
1158
+ "448": "ឝ",
1159
+ "449": "ឞ",
1160
+ "450": "ស",
1161
+ "451": "ហ",
1162
+ "452": "ឡ",
1163
+ "453": "អ",
1164
+ "454": "ឣ",
1165
+ "455": "ឤ",
1166
+ "456": "ឥ",
1167
+ "457": "ឦ",
1168
+ "458": "ឧ",
1169
+ "459": "ឨ",
1170
+ "460": "ឩ",
1171
+ "461": "ឪ",
1172
+ "462": "ឫ",
1173
+ "463": "ឬ",
1174
+ "464": "ឭ",
1175
+ "465": "ឮ",
1176
+ "466": "ឯ",
1177
+ "467": "ឰ",
1178
+ "468": "ឱ",
1179
+ "469": "ឲ",
1180
+ "470": "ឳ",
1181
+ "471": "឵",
1182
+ "472": "ា",
1183
+ "473": "ិ",
1184
+ "474": "ី",
1185
+ "475": "ឹ",
1186
+ "476": "ឺ",
1187
+ "477": "ុ",
1188
+ "478": "ូ",
1189
+ "479": "ួ",
1190
+ "480": "ើ",
1191
+ "481": "ឿ",
1192
+ "482": "ៀ",
1193
+ "483": "េ",
1194
+ "484": "ែ",
1195
+ "485": "ៃ",
1196
+ "486": "ោ",
1197
+ "487": "ៅ",
1198
+ "488": "ំ",
1199
+ "489": "ះ",
1200
+ "490": "ៈ",
1201
+ "491": "៉",
1202
+ "492": "៊",
1203
+ "493": "់",
1204
+ "494": "៌",
1205
+ "495": "៍",
1206
+ "496": "៎",
1207
+ "497": "៏",
1208
+ "498": "័",
1209
+ "499": "៑",
1210
+ "500": "្",
1211
+ "501": "៓",
1212
+ "502": "។",
1213
+ "503": "៕",
1214
+ "504": "៖",
1215
+ "505": "ៗ",
1216
+ "506": "៘",
1217
+ "507": "៙",
1218
+ "508": "៚",
1219
+ "509": "៛",
1220
+ "510": "៝",
1221
+ "511": "០",
1222
+ "512": "១",
1223
+ "513": "២",
1224
+ "514": "៣",
1225
+ "515": "៤",
1226
+ "516": "៥",
1227
+ "517": "៦",
1228
+ "518": "៧",
1229
+ "519": "៨",
1230
+ "520": "៩",
1231
+ "521": "៰",
1232
+ "522": "៱",
1233
+ "523": "៲",
1234
+ "524": "៸",
1235
+ "525": "៹",
1236
+ "526": "᧥",
1237
+ "527": "᧦",
1238
+ "528": "᧯",
1239
+ "529": "᧰",
1240
+ "530": "ḍ",
1241
+ "531": "ḓ",
1242
+ "532": "ḛ",
1243
+ "533": "ḥ",
1244
+ "534": "Ḳ",
1245
+ "535": "ḷ",
1246
+ "536": "ṁ",
1247
+ "537": "Ṃ",
1248
+ "538": "ṃ",
1249
+ "539": "Ṅ",
1250
+ "540": "ṅ",
1251
+ "541": "Ṇ",
1252
+ "542": "ṇ",
1253
+ "543": "ṛ",
1254
+ "544": "Ṣ",
1255
+ "545": "ṣ",
1256
+ "546": "Ṭ",
1257
+ "547": "ṭ",
1258
+ "548": "Ạ",
1259
+ "549": "ạ",
1260
+ "550": "Ả",
1261
+ "551": "ả",
1262
+ "552": "Ấ",
1263
+ "553": "ấ",
1264
+ "554": "Ầ",
1265
+ "555": "ầ",
1266
+ "556": "ẩ",
1267
+ "557": "Ậ",
1268
+ "558": "ậ",
1269
+ "559": "ắ",
1270
+ "560": "ẳ",
1271
+ "561": "ặ",
1272
+ "562": "Ẹ",
1273
+ "563": "ẹ",
1274
+ "564": "ẻ",
1275
+ "565": "ế",
1276
+ "566": "Ề",
1277
+ "567": "ề",
1278
+ "568": "Ể",
1279
+ "569": "ể",
1280
+ "570": "ễ",
1281
+ "571": "ệ",
1282
+ "572": "Ỉ",
1283
+ "573": "ỉ",
1284
+ "574": "Ị",
1285
+ "575": "ị",
1286
+ "576": "Ọ",
1287
+ "577": "ọ",
1288
+ "578": "ỏ",
1289
+ "579": "ố",
1290
+ "580": "ồ",
1291
+ "581": "ổ",
1292
+ "582": "ỗ",
1293
+ "583": "Ộ",
1294
+ "584": "ộ",
1295
+ "585": "Ớ",
1296
+ "586": "ớ",
1297
+ "587": "Ờ",
1298
+ "588": "ờ",
1299
+ "589": "ợ",
1300
+ "590": "ụ",
1301
+ "591": "Ủ",
1302
+ "592": "ủ",
1303
+ "593": "Ứ",
1304
+ "594": "ứ",
1305
+ "595": "ừ",
1306
+ "596": "ử",
1307
+ "597": "Ữ",
1308
+ "598": "ữ",
1309
+ "599": "ự",
1310
+ "600": "ỳ",
1311
+ "601": "ỹ",
1312
+ "602": "​",
1313
+ "603": "‌",
1314
+ "604": "‍",
1315
+ "605": "‎",
1316
+ "606": "‏",
1317
+ "607": "‐",
1318
+ "608": "‑",
1319
+ "609": "–",
1320
+ "610": "—",
1321
+ "611": "‘",
1322
+ "612": "’",
1323
+ "613": "“",
1324
+ "614": "”",
1325
+ "615": "„",
1326
+ "616": "†",
1327
+ "617": "•",
1328
+ "618": "…",
1329
+ "619": "‪",
1330
+ "620": "‫",
1331
+ "621": "‬",
1332
+ "622": "′",
1333
+ "623": "″",
1334
+ "624": "‹",
1335
+ "625": "›",
1336
+ "626": "⁄",
1337
+ "627": "⁠",
1338
+ "628": "⁡",
1339
+ "629": "⁣",
1340
+ "630": "€",
1341
+ "631": "№",
1342
+ "632": "™",
1343
+ "633": "←",
1344
+ "634": "↑",
1345
+ "635": "→",
1346
+ "636": "↓",
1347
+ "637": "−",
1348
+ "638": "∙",
1349
+ "639": "∞",
1350
+ "640": "∫",
1351
+ "641": "≈",
1352
+ "642": "≤",
1353
+ "643": "≥",
1354
+ "644": "⋅",
1355
+ "645": "▪",
1356
+ "646": "►",
1357
+ "647": "▼",
1358
+ "648": "▾",
1359
+ "649": "◌",
1360
+ "650": "●",
1361
+ "651": "⚫",
1362
+ "652": "✅",
1363
+ "653": "⟶",
1364
+ "654": "伍",
1365
+ "655": "和",
1366
+ "656": "德",
1367
+ "657": "連",
1368
+ "658": "郑",
1369
+ "659": "Ɡ",
1370
+ "660": "꫁",
1371
+ "661": "",
1372
+ "662": "",
1373
+ "663": ";",
1374
+ "664": "�",
1375
+ "665": "🎉"
1376
+ }
1377
+ }
1378
+ }