Jawaria's picture
Upload tokenizer
a23f974
raw
history blame
382 Bytes
{
"[PAD]": 31,
"[UNK]": 30,
"|": 0,
"ا": 1,
"ب": 2,
"ت": 3,
"ج": 4,
"خ": 5,
"د": 6,
"ر": 7,
"ز": 8,
"س": 9,
"ش": 10,
"ص": 11,
"ع": 12,
"غ": 13,
"ل": 14,
"م": 15,
"ن": 16,
"ه": 17,
"و": 18,
"ي": 19,
"ټ": 20,
"پ": 21,
"څ": 22,
"ډ": 23,
"ړ": 24,
"ښ": 25,
"ک": 26,
"ګ": 27,
"ی": 28,
"ې": 29
}