nrshoudi's picture
Upload tokenizer
fd24777
raw
history blame
418 Bytes
{
"[PAD]": 34,
"[UNK]": 33,
"|": 0,
"ء": 1,
"آ": 2,
"أ": 3,
"إ": 4,
"ا": 5,
"ب": 6,
"ة": 7,
"ت": 8,
"ج": 9,
"ح": 10,
"خ": 11,
"د": 12,
"ذ": 13,
"ر": 14,
"س": 15,
"ش": 16,
"ص": 17,
"ض": 18,
"ط": 19,
"ظ": 20,
"ع": 21,
"غ": 22,
"ف": 23,
"ق": 24,
"ك": 25,
"ل": 26,
"م": 27,
"ن": 28,
"ه": 29,
"و": 30,
"ى": 31,
"ي": 32
}