khaoulaoub's picture
Upload tokenizer
32beef3
raw
history blame
406 Bytes
{
"[PAD]": 33,
"[UNK]": 32,
"|": 10,
"ء": 11,
"ا": 7,
"ب": 22,
"ة": 29,
"ت": 26,
"ث": 15,
"ج": 30,
"ح": 18,
"خ": 14,
"د": 8,
"ذ": 24,
"ر": 2,
"ز": 1,
"س": 19,
"ش": 16,
"ص": 5,
"ض": 4,
"ط": 25,
"ظ": 6,
"ع": 13,
"غ": 23,
"ف": 12,
"ق": 31,
"ك": 27,
"ل": 3,
"م": 0,
"ن": 17,
"ه": 20,
"و": 28,
"ى": 9,
"ي": 21
}