ad019el's picture
Upload tokenizer
94da37d
raw
history blame
525 Bytes
{
"2": 24,
"[PAD]": 43,
"[UNK]": 42,
"|": 39,
"ء": 4,
"آ": 18,
"أ": 38,
"ؤ": 35,
"إ": 8,
"ئ": 7,
"ا": 36,
"ب": 25,
"ة": 27,
"ت": 28,
"ث": 29,
"ج": 20,
"ح": 21,
"خ": 16,
"د": 22,
"ذ": 26,
"ر": 40,
"ز": 31,
"س": 32,
"ش": 5,
"ص": 34,
"ض": 15,
"ط": 3,
"ظ": 23,
"ع": 1,
"غ": 13,
"ف": 41,
"ق": 33,
"ك": 0,
"ل": 6,
"م": 10,
"ن": 14,
"ه": 30,
"و": 12,
"ى": 11,
"ي": 17,
"ٱ": 9,
"ڤ": 37,
"ک": 19,
"ی": 2
}