farbodpya commited on
Commit
87b1acf
·
verified ·
1 Parent(s): d2b3d27

Upload folder using huggingface_hub

Browse files
Files changed (3) hide show
  1. config.json +6 -0
  2. pytorch_model.bin +3 -0
  3. vocab.json +580 -0
config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "num_classes": 288,
3
+ "d_model": 1280,
4
+ "nhead": 16,
5
+ "num_layers": 8
6
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d8c4b7f77ca4821fa696d4a38bddab7f433fa93cf08a3df705ae80d63e0bf705
3
+ size 400029207
vocab.json ADDED
@@ -0,0 +1,580 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "char_to_idx": {
3
+ " ": 1,
4
+ "!": 2,
5
+ "#": 3,
6
+ "%": 4,
7
+ "&": 5,
8
+ "(": 6,
9
+ ")": 7,
10
+ "*": 8,
11
+ "+": 9,
12
+ ",": 10,
13
+ "-": 11,
14
+ ".": 12,
15
+ "/": 13,
16
+ ":": 14,
17
+ "=": 15,
18
+ "?": 16,
19
+ "A": 17,
20
+ "B": 18,
21
+ "C": 19,
22
+ "D": 20,
23
+ "E": 21,
24
+ "F": 22,
25
+ "G": 23,
26
+ "H": 24,
27
+ "I": 25,
28
+ "J": 26,
29
+ "K": 27,
30
+ "L": 28,
31
+ "M": 29,
32
+ "N": 30,
33
+ "O": 31,
34
+ "P": 32,
35
+ "Q": 33,
36
+ "R": 34,
37
+ "S": 35,
38
+ "T": 36,
39
+ "U": 37,
40
+ "V": 38,
41
+ "W": 39,
42
+ "X": 40,
43
+ "Y": 41,
44
+ "Z": 42,
45
+ "[": 43,
46
+ "\\": 44,
47
+ "]": 45,
48
+ "a": 46,
49
+ "b": 47,
50
+ "c": 48,
51
+ "d": 49,
52
+ "e": 50,
53
+ "f": 51,
54
+ "g": 52,
55
+ "h": 53,
56
+ "i": 54,
57
+ "j": 55,
58
+ "k": 56,
59
+ "l": 57,
60
+ "m": 58,
61
+ "n": 59,
62
+ "o": 60,
63
+ "p": 61,
64
+ "r": 62,
65
+ "s": 63,
66
+ "t": 64,
67
+ "u": 65,
68
+ "v": 66,
69
+ "w": 67,
70
+ "x": 68,
71
+ "y": 69,
72
+ "z": 70,
73
+ "|": 71,
74
+ "«": 72,
75
+ "»": 73,
76
+ "¾": 74,
77
+ "Â": 75,
78
+ "ğ": 76,
79
+ "ı": 77,
80
+ "Σ": 78,
81
+ "π": 79,
82
+ "ω": 80,
83
+ "،": 81,
84
+ "؛": 82,
85
+ "؟": 83,
86
+ "ء": 84,
87
+ "آ": 85,
88
+ "أ": 86,
89
+ "ؤ": 87,
90
+ "إ": 88,
91
+ "ئ": 89,
92
+ "ا": 90,
93
+ "ب": 91,
94
+ "ة": 92,
95
+ "ت": 93,
96
+ "ث": 94,
97
+ "ج": 95,
98
+ "ح": 96,
99
+ "خ": 97,
100
+ "د": 98,
101
+ "ذ": 99,
102
+ "ر": 100,
103
+ "ز": 101,
104
+ "س": 102,
105
+ "ش": 103,
106
+ "ص": 104,
107
+ "ض": 105,
108
+ "ط": 106,
109
+ "ظ": 107,
110
+ "ع": 108,
111
+ "غ": 109,
112
+ "ـ": 110,
113
+ "ف": 111,
114
+ "ق": 112,
115
+ "ك": 113,
116
+ "ل": 114,
117
+ "م": 115,
118
+ "ن": 116,
119
+ "ه": 117,
120
+ "و": 118,
121
+ "ى": 119,
122
+ "ي": 120,
123
+ "ً": 121,
124
+ "ٌ": 122,
125
+ "ٍ": 123,
126
+ "َ": 124,
127
+ "ُ": 125,
128
+ "ِ": 126,
129
+ "ّ": 127,
130
+ "ْ": 128,
131
+ "ٔ": 129,
132
+ "ٙ": 130,
133
+ "ٚ": 131,
134
+ "ٛ": 132,
135
+ "٠": 133,
136
+ "١": 134,
137
+ "٢": 135,
138
+ "٣": 136,
139
+ "٤": 137,
140
+ "٥": 138,
141
+ "٦": 139,
142
+ "٧": 140,
143
+ "٨": 141,
144
+ "٩": 142,
145
+ "٪": 143,
146
+ "٫": 144,
147
+ "٬": 145,
148
+ "ٰ": 146,
149
+ "ٱ": 147,
150
+ "ٲ": 148,
151
+ "ٴ": 149,
152
+ "ٵ": 150,
153
+ "ٸ": 151,
154
+ "ٹ": 152,
155
+ "پ": 153,
156
+ "ځ": 154,
157
+ "چ": 155,
158
+ "ڈ": 156,
159
+ "ڑ": 157,
160
+ "ڒ": 158,
161
+ "ڔ": 159,
162
+ "ڕ": 160,
163
+ "ژ": 161,
164
+ "ښ": 162,
165
+ "ڞ": 163,
166
+ "ڠ": 164,
167
+ "ڤ": 165,
168
+ "ڨ": 166,
169
+ "ک": 167,
170
+ "ګ": 168,
171
+ "ڬ": 169,
172
+ "ڭ": 170,
173
+ "گ": 171,
174
+ "ڵ": 172,
175
+ "ڶ": 173,
176
+ "ں": 174,
177
+ "ھ": 175,
178
+ "ۀ": 176,
179
+ "ہ": 177,
180
+ "ۂ": 178,
181
+ "ۃ": 179,
182
+ "ۅ": 180,
183
+ "ۆ": 181,
184
+ "ۇ": 182,
185
+ "ۉ": 183,
186
+ "ۊ": 184,
187
+ "ۋ": 185,
188
+ "ی": 186,
189
+ "ە": 187,
190
+ "ۖ": 188,
191
+ "ۗ": 189,
192
+ "ۘ": 190,
193
+ "ۚ": 191,
194
+ "۰": 192,
195
+ "۱": 193,
196
+ "۲": 194,
197
+ "۳": 195,
198
+ "۴": 196,
199
+ "۵": 197,
200
+ "۶": 198,
201
+ "۷": 199,
202
+ "۸": 200,
203
+ "۹": 201,
204
+ "ಠ": 202,
205
+ "ಥ": 203,
206
+ "ა": 204,
207
+ "დ": 205,
208
+ "ი": 206,
209
+ "ნ": 207,
210
+ "ღ": 208,
211
+ "‌": 209,
212
+ "…": 210,
213
+ "ツ": 211,
214
+ "ヽ": 212,
215
+ "天": 213,
216
+ "娜": 214,
217
+ "梦": 215,
218
+ "米": 216,
219
+ "ﭘ": 217,
220
+ "ﭼ": 218,
221
+ "ﮐ": 219,
222
+ "ﮑ": 220,
223
+ "ﮔ": 221,
224
+ "ﮫ": 222,
225
+ "ﮬ": 223,
226
+ "ﯨ": 224,
227
+ "ﯼ": 225,
228
+ "ﯽ": 226,
229
+ "ﯾ": 227,
230
+ "ﯿ": 228,
231
+ "ﷲ": 229,
232
+ "ﷻ": 230,
233
+ "ﺍ": 231,
234
+ "ﺎ": 232,
235
+ "ﺏ": 233,
236
+ "ﺐ": 234,
237
+ "ﺑ": 235,
238
+ "ﺒ": 236,
239
+ "ﺕ": 237,
240
+ "ﺖ": 238,
241
+ "ﺗ": 239,
242
+ "ﺘ": 240,
243
+ "ﺟ": 241,
244
+ "ﺠ": 242,
245
+ "ﺤ": 243,
246
+ "ﺥ": 244,
247
+ "ﺧ": 245,
248
+ "ﺨ": 246,
249
+ "ﺩ": 247,
250
+ "ﺪ": 248,
251
+ "ﺭ": 249,
252
+ "ﺮ": 250,
253
+ "ﺯ": 251,
254
+ "ﺳ": 252,
255
+ "ﺴ": 253,
256
+ "ﺷ": 254,
257
+ "ﺸ": 255,
258
+ "ﻄ": 256,
259
+ "ﻈ": 257,
260
+ "ﻉ": 258,
261
+ "ﻋ": 259,
262
+ "ﻌ": 260,
263
+ "ﻔ": 261,
264
+ "ﻖ": 262,
265
+ "ﻗ": 263,
266
+ "ﻘ": 264,
267
+ "ﻜ": 265,
268
+ "ﻝ": 266,
269
+ "ﻟ": 267,
270
+ "ﻠ": 268,
271
+ "ﻡ": 269,
272
+ "ﻢ": 270,
273
+ "ﻣ": 271,
274
+ "ﻤ": 272,
275
+ "ﻥ": 273,
276
+ "ﻦ": 274,
277
+ "ﻧ": 275,
278
+ "ﻨ": 276,
279
+ "ﻩ": 277,
280
+ "ﻪ": 278,
281
+ "ﻫ": 279,
282
+ "ﻬ": 280,
283
+ "ﻭ": 281,
284
+ "ﻮ": 282,
285
+ "ﻱ": 283,
286
+ "ﻲ": 284,
287
+ "ﻳ": 285,
288
+ "ﻴ": 286,
289
+ "ﻻ": 287
290
+ },
291
+ "idx_to_char": {
292
+ "1": " ",
293
+ "2": "!",
294
+ "3": "#",
295
+ "4": "%",
296
+ "5": "&",
297
+ "6": "(",
298
+ "7": ")",
299
+ "8": "*",
300
+ "9": "+",
301
+ "10": ",",
302
+ "11": "-",
303
+ "12": ".",
304
+ "13": "/",
305
+ "14": ":",
306
+ "15": "=",
307
+ "16": "?",
308
+ "17": "A",
309
+ "18": "B",
310
+ "19": "C",
311
+ "20": "D",
312
+ "21": "E",
313
+ "22": "F",
314
+ "23": "G",
315
+ "24": "H",
316
+ "25": "I",
317
+ "26": "J",
318
+ "27": "K",
319
+ "28": "L",
320
+ "29": "M",
321
+ "30": "N",
322
+ "31": "O",
323
+ "32": "P",
324
+ "33": "Q",
325
+ "34": "R",
326
+ "35": "S",
327
+ "36": "T",
328
+ "37": "U",
329
+ "38": "V",
330
+ "39": "W",
331
+ "40": "X",
332
+ "41": "Y",
333
+ "42": "Z",
334
+ "43": "[",
335
+ "44": "\\",
336
+ "45": "]",
337
+ "46": "a",
338
+ "47": "b",
339
+ "48": "c",
340
+ "49": "d",
341
+ "50": "e",
342
+ "51": "f",
343
+ "52": "g",
344
+ "53": "h",
345
+ "54": "i",
346
+ "55": "j",
347
+ "56": "k",
348
+ "57": "l",
349
+ "58": "m",
350
+ "59": "n",
351
+ "60": "o",
352
+ "61": "p",
353
+ "62": "r",
354
+ "63": "s",
355
+ "64": "t",
356
+ "65": "u",
357
+ "66": "v",
358
+ "67": "w",
359
+ "68": "x",
360
+ "69": "y",
361
+ "70": "z",
362
+ "71": "|",
363
+ "72": "«",
364
+ "73": "»",
365
+ "74": "¾",
366
+ "75": "Â",
367
+ "76": "ğ",
368
+ "77": "ı",
369
+ "78": "Σ",
370
+ "79": "π",
371
+ "80": "ω",
372
+ "81": "،",
373
+ "82": "؛",
374
+ "83": "؟",
375
+ "84": "ء",
376
+ "85": "آ",
377
+ "86": "أ",
378
+ "87": "ؤ",
379
+ "88": "إ",
380
+ "89": "ئ",
381
+ "90": "ا",
382
+ "91": "ب",
383
+ "92": "ة",
384
+ "93": "ت",
385
+ "94": "ث",
386
+ "95": "ج",
387
+ "96": "ح",
388
+ "97": "خ",
389
+ "98": "د",
390
+ "99": "ذ",
391
+ "100": "ر",
392
+ "101": "ز",
393
+ "102": "س",
394
+ "103": "ش",
395
+ "104": "ص",
396
+ "105": "ض",
397
+ "106": "ط",
398
+ "107": "ظ",
399
+ "108": "ع",
400
+ "109": "غ",
401
+ "110": "ـ",
402
+ "111": "ف",
403
+ "112": "ق",
404
+ "113": "ك",
405
+ "114": "ل",
406
+ "115": "م",
407
+ "116": "ن",
408
+ "117": "ه",
409
+ "118": "و",
410
+ "119": "ى",
411
+ "120": "ي",
412
+ "121": "ً",
413
+ "122": "ٌ",
414
+ "123": "ٍ",
415
+ "124": "َ",
416
+ "125": "ُ",
417
+ "126": "ِ",
418
+ "127": "ّ",
419
+ "128": "ْ",
420
+ "129": "ٔ",
421
+ "130": "ٙ",
422
+ "131": "ٚ",
423
+ "132": "ٛ",
424
+ "133": "٠",
425
+ "134": "١",
426
+ "135": "٢",
427
+ "136": "٣",
428
+ "137": "٤",
429
+ "138": "٥",
430
+ "139": "٦",
431
+ "140": "٧",
432
+ "141": "٨",
433
+ "142": "٩",
434
+ "143": "٪",
435
+ "144": "٫",
436
+ "145": "٬",
437
+ "146": "ٰ",
438
+ "147": "ٱ",
439
+ "148": "ٲ",
440
+ "149": "ٴ",
441
+ "150": "ٵ",
442
+ "151": "ٸ",
443
+ "152": "ٹ",
444
+ "153": "پ",
445
+ "154": "ځ",
446
+ "155": "چ",
447
+ "156": "ڈ",
448
+ "157": "ڑ",
449
+ "158": "ڒ",
450
+ "159": "ڔ",
451
+ "160": "ڕ",
452
+ "161": "ژ",
453
+ "162": "ښ",
454
+ "163": "ڞ",
455
+ "164": "ڠ",
456
+ "165": "ڤ",
457
+ "166": "ڨ",
458
+ "167": "ک",
459
+ "168": "ګ",
460
+ "169": "ڬ",
461
+ "170": "ڭ",
462
+ "171": "گ",
463
+ "172": "ڵ",
464
+ "173": "ڶ",
465
+ "174": "ں",
466
+ "175": "ھ",
467
+ "176": "ۀ",
468
+ "177": "ہ",
469
+ "178": "ۂ",
470
+ "179": "ۃ",
471
+ "180": "ۅ",
472
+ "181": "ۆ",
473
+ "182": "ۇ",
474
+ "183": "ۉ",
475
+ "184": "ۊ",
476
+ "185": "ۋ",
477
+ "186": "ی",
478
+ "187": "ە",
479
+ "188": "ۖ",
480
+ "189": "ۗ",
481
+ "190": "ۘ",
482
+ "191": "ۚ",
483
+ "192": "۰",
484
+ "193": "۱",
485
+ "194": "۲",
486
+ "195": "۳",
487
+ "196": "۴",
488
+ "197": "۵",
489
+ "198": "۶",
490
+ "199": "۷",
491
+ "200": "۸",
492
+ "201": "۹",
493
+ "202": "ಠ",
494
+ "203": "ಥ",
495
+ "204": "ა",
496
+ "205": "დ",
497
+ "206": "ი",
498
+ "207": "ნ",
499
+ "208": "ღ",
500
+ "209": "‌",
501
+ "210": "…",
502
+ "211": "ツ",
503
+ "212": "ヽ",
504
+ "213": "天",
505
+ "214": "娜",
506
+ "215": "梦",
507
+ "216": "米",
508
+ "217": "ﭘ",
509
+ "218": "ﭼ",
510
+ "219": "ﮐ",
511
+ "220": "ﮑ",
512
+ "221": "ﮔ",
513
+ "222": "ﮫ",
514
+ "223": "ﮬ",
515
+ "224": "ﯨ",
516
+ "225": "ﯼ",
517
+ "226": "ﯽ",
518
+ "227": "ﯾ",
519
+ "228": "ﯿ",
520
+ "229": "ﷲ",
521
+ "230": "ﷻ",
522
+ "231": "ﺍ",
523
+ "232": "ﺎ",
524
+ "233": "ﺏ",
525
+ "234": "ﺐ",
526
+ "235": "ﺑ",
527
+ "236": "ﺒ",
528
+ "237": "ﺕ",
529
+ "238": "ﺖ",
530
+ "239": "ﺗ",
531
+ "240": "ﺘ",
532
+ "241": "ﺟ",
533
+ "242": "ﺠ",
534
+ "243": "ﺤ",
535
+ "244": "ﺥ",
536
+ "245": "ﺧ",
537
+ "246": "ﺨ",
538
+ "247": "ﺩ",
539
+ "248": "ﺪ",
540
+ "249": "ﺭ",
541
+ "250": "ﺮ",
542
+ "251": "ﺯ",
543
+ "252": "ﺳ",
544
+ "253": "ﺴ",
545
+ "254": "ﺷ",
546
+ "255": "ﺸ",
547
+ "256": "ﻄ",
548
+ "257": "ﻈ",
549
+ "258": "ﻉ",
550
+ "259": "ﻋ",
551
+ "260": "ﻌ",
552
+ "261": "ﻔ",
553
+ "262": "ﻖ",
554
+ "263": "ﻗ",
555
+ "264": "ﻘ",
556
+ "265": "ﻜ",
557
+ "266": "ﻝ",
558
+ "267": "ﻟ",
559
+ "268": "ﻠ",
560
+ "269": "ﻡ",
561
+ "270": "ﻢ",
562
+ "271": "ﻣ",
563
+ "272": "ﻤ",
564
+ "273": "ﻥ",
565
+ "274": "ﻦ",
566
+ "275": "ﻧ",
567
+ "276": "ﻨ",
568
+ "277": "ﻩ",
569
+ "278": "ﻪ",
570
+ "279": "ﻫ",
571
+ "280": "ﻬ",
572
+ "281": "ﻭ",
573
+ "282": "ﻮ",
574
+ "283": "ﻱ",
575
+ "284": "ﻲ",
576
+ "285": "ﻳ",
577
+ "286": "ﻴ",
578
+ "287": "ﻻ"
579
+ }
580
+ }