nrshoudi's picture
Upload tokenizer
60e293b
raw
history blame
458 Bytes
{
" ": 0,
"[PAD]": 37,
"[UNK]": 36,
"آ": 1,
"أ": 2,
"إ": 3,
"ا": 4,
"اً": 5,
"ب": 6,
"ة": 7,
"ت": 8,
"ث": 9,
"ج": 10,
"ح": 11,
"خ": 12,
"د": 13,
"ذ": 14,
"ر": 15,
"ز": 16,
"س": 17,
"ش": 18,
"ص": 19,
"ض": 20,
"ط": 21,
"ظ": 22,
"ع": 23,
"غ": 24,
"ف": 25,
"ق": 26,
"ك": 27,
"ل": 28,
"لا": 29,
"م": 30,
"ن": 31,
"ه": 32,
"و": 33,
"ى": 34,
"ي": 35
}