diallomama's picture
Upload tokenizer
7f3d23c
raw
history blame
815 Bytes
{
"[PAD]": 67,
"[UNK]": 66,
"_": 1,
"e": 2,
"g": 3,
"t": 4,
"|": 0,
"،": 5,
"؛": 6,
"؟": 7,
"ء": 8,
"آ": 9,
"أ": 10,
"ؤ": 11,
"إ": 12,
"ئ": 13,
"ا": 14,
"ب": 15,
"ة": 16,
"ت": 17,
"ث": 18,
"ج": 19,
"ح": 20,
"خ": 21,
"د": 22,
"ذ": 23,
"ر": 24,
"ز": 25,
"س": 26,
"ش": 27,
"ص": 28,
"ض": 29,
"ط": 30,
"ظ": 31,
"ع": 32,
"غ": 33,
"ـ": 34,
"ف": 35,
"ق": 36,
"ك": 37,
"ل": 38,
"م": 39,
"ن": 40,
"ه": 41,
"و": 42,
"ى": 43,
"ي": 44,
"ً": 45,
"ٌ": 46,
"ٍ": 47,
"َ": 48,
"ُ": 49,
"ِ": 50,
"ّ": 51,
"ْ": 52,
"ٰ": 53,
"چ": 54,
"ڨ": 55,
"ک": 56,
"ھ": 57,
"ی": 58,
"ۖ": 59,
"ۚ": 60,
"—": 61,
"”": 62,
"☭": 63,
"ﺃ": 64,
"ﻻ": 65
}