ad019el's picture
add tokenizer
81d75b1
raw
history blame
550 Bytes
{
"[PAD]": 45,
"[UNK]": 44,
"|": 27,
"ĉ": 28,
"̣": 8,
"ء": 37,
"آ": 18,
"أ": 5,
"ؤ": 34,
"إ": 0,
"ئ": 4,
"ا": 3,
"ب": 2,
"ة": 33,
"ت": 1,
"ث": 12,
"ج": 9,
"ح": 32,
"خ": 26,
"د": 25,
"ذ": 20,
"ر": 13,
"ز": 24,
"س": 30,
"ش": 38,
"ص": 19,
"ض": 17,
"ط": 41,
"ظ": 29,
"ع": 15,
"غ": 23,
"ف": 10,
"ق": 16,
"ك": 31,
"ل": 22,
"م": 11,
"ن": 6,
"ه": 39,
"و": 35,
"ى": 21,
"ي": 36,
"ٱ": 40,
"چ": 14,
"ڤ": 43,
"ک": 7,
"ی": 42
}