khaoulaoub's picture
Upload tokenizer
17e33a3
raw
history blame contribute delete
418 Bytes
{
"[PAD]": 34,
"[UNK]": 33,
"|": 14,
"ء": 17,
"ا": 26,
"ب": 12,
"ة": 23,
"ت": 24,
"ث": 31,
"ج": 32,
"ح": 11,
"خ": 28,
"د": 16,
"ذ": 10,
"ر": 13,
"ز": 25,
"س": 7,
"ش": 1,
"ص": 4,
"ض": 29,
"ط": 30,
"ظ": 15,
"ع": 27,
"غ": 3,
"ـ": 21,
"ف": 5,
"ق": 20,
"ك": 18,
"ل": 8,
"م": 0,
"ن": 9,
"ه": 22,
"و": 19,
"ى": 6,
"ي": 2
}