Abdalrhman's picture
Upload tokenizer
dc5b2a4
raw
history blame
826 Bytes
{
"[PAD]": 68,
"[UNK]": 67,
"_": 1,
"e": 2,
"g": 3,
"t": 4,
"|": 0,
"«": 5,
"»": 6,
"،": 7,
"؛": 8,
"؟": 9,
"ء": 10,
"آ": 11,
"أ": 12,
"ؤ": 13,
"إ": 14,
"ئ": 15,
"ا": 16,
"ب": 17,
"ة": 18,
"ت": 19,
"ث": 20,
"ج": 21,
"ح": 22,
"خ": 23,
"د": 24,
"ذ": 25,
"ر": 26,
"ز": 27,
"س": 28,
"ش": 29,
"ص": 30,
"ض": 31,
"ط": 32,
"ظ": 33,
"ع": 34,
"غ": 35,
"ـ": 36,
"ف": 37,
"ق": 38,
"ك": 39,
"ل": 40,
"م": 41,
"ن": 42,
"ه": 43,
"و": 44,
"ى": 45,
"ي": 46,
"ً": 47,
"ٌ": 48,
"ٍ": 49,
"َ": 50,
"ُ": 51,
"ِ": 52,
"ّ": 53,
"ْ": 54,
"ٰ": 55,
"چ": 56,
"ڨ": 57,
"ک": 58,
"ھ": 59,
"ی": 60,
"ۖ": 61,
"ۚ": 62,
"—": 63,
"☭": 64,
"ﺃ": 65,
"ﻻ": 66
}