Upload tokenizer
6084a88
|
{ |
|
"#": 1, |
|
"&": 2, |
|
"(": 3, |
|
")": 4, |
|
"[PAD]": 120, |
|
"[UNK]": 119, |
|
"_": 5, |
|
"a": 6, |
|
"b": 7, |
|
"c": 8, |
|
"d": 9, |
|
"e": 10, |
|
"f": 11, |
|
"g": 12, |
|
"h": 13, |
|
"i": 14, |
|
"k": 15, |
|
"m": 16, |
|
"n": 17, |
|
"o": 18, |
|
"p": 19, |
|
"q": 20, |
|
"r": 21, |
|
"s": 22, |
|
"t": 23, |
|
"u": 24, |
|
"y": 25, |
|
"z": 26, |
|
"|": 0, |
|
"«": 27, |
|
"»": 28, |
|
"ā": 29, |
|
"š": 30, |
|
"،": 31, |
|
"؛": 32, |
|
"؟": 33, |
|
"ء": 34, |
|
"آ": 35, |
|
"أ": 36, |
|
"ؤ": 37, |
|
"ئ": 38, |
|
"ا": 39, |
|
"ب": 40, |
|
"ت": 41, |
|
"ث": 42, |
|
"ج": 43, |
|
"ح": 44, |
|
"خ": 45, |
|
"د": 46, |
|
"ذ": 47, |
|
"ر": 48, |
|
"ز": 49, |
|
"س": 50, |
|
"ش": 51, |
|
"ص": 52, |
|
"ض": 53, |
|
"ط": 54, |
|
"ظ": 55, |
|
"ع": 56, |
|
"غ": 57, |
|
"ـ": 58, |
|
"ف": 59, |
|
"ق": 60, |
|
"ل": 61, |
|
"م": 62, |
|
"ن": 63, |
|
"ه": 64, |
|
"و": 65, |
|
"ى": 66, |
|
"ً": 67, |
|
"ٌ": 68, |
|
"َ": 69, |
|
"ُ": 70, |
|
"ِ": 71, |
|
"ّ": 72, |
|
"ْ": 73, |
|
"ٔ": 74, |
|
"٬": 75, |
|
"پ": 76, |
|
"چ": 77, |
|
"ژ": 78, |
|
"ک": 79, |
|
"گ": 80, |
|
"ۀ": 81, |
|
"ی": 82, |
|
"ے": 83, |
|
"–": 84, |
|
"…": 85, |
|
"ﭘ": 86, |
|
"ﮐ": 87, |
|
"ﮔ": 88, |
|
"ﯽ": 89, |
|
"ﯾ": 90, |
|
"ﯿ": 91, |
|
"ﺍ": 92, |
|
"ﺎ": 93, |
|
"ﺑ": 94, |
|
"ﺒ": 95, |
|
"ﺖ": 96, |
|
"ﺘ": 97, |
|
"ﺧ": 98, |
|
"ﺩ": 99, |
|
"ﺪ": 100, |
|
"ﺭ": 101, |
|
"ﺮ": 102, |
|
"ﺱ": 103, |
|
"ﺴ": 104, |
|
"ﺷ": 105, |
|
"ﺸ": 106, |
|
"ﻀ": 107, |
|
"ﻋ": 108, |
|
"ﻌ": 109, |
|
"ﻟ": 110, |
|
"ﻡ": 111, |
|
"ﻢ": 112, |
|
"ﻤ": 113, |
|
"ﻥ": 114, |
|
"ﻧ": 115, |
|
"ﻪ": 116, |
|
"ﻭ": 117, |
|
"ﻮ": 118 |
|
} |
|
|