ad019el's picture
Upload tokenizer
98527c6
raw
history blame
478 Bytes
{
"[PAD]": 39,
"[UNK]": 38,
"|": 12,
"ء": 1,
"آ": 8,
"أ": 20,
"إ": 18,
"ئ": 23,
"ا": 31,
"ب": 29,
"ة": 36,
"ت": 7,
"ث": 13,
"ج": 37,
"ح": 16,
"خ": 10,
"د": 4,
"ذ": 6,
"ر": 24,
"ز": 32,
"س": 27,
"ش": 0,
"ص": 28,
"ض": 35,
"ط": 21,
"ظ": 25,
"ع": 5,
"غ": 26,
"ف": 2,
"ق": 3,
"ك": 11,
"ل": 19,
"م": 33,
"ن": 14,
"ه": 15,
"و": 22,
"ى": 34,
"ي": 9,
"ٱ": 17,
"ڤ": 30
}