Jawaria's picture
Upload tokenizer
f8d3141
raw
history blame
430 Bytes
{
"[PAD]": 35,
"[UNK]": 34,
"|": 0,
"ا": 1,
"ب": 2,
"ت": 3,
"ج": 4,
"ح": 5,
"خ": 6,
"د": 7,
"ر": 8,
"ز": 9,
"س": 10,
"ش": 11,
"ص": 12,
"ط": 13,
"ع": 14,
"غ": 15,
"ق": 16,
"ل": 17,
"م": 18,
"ن": 19,
"ه": 20,
"و": 21,
"ي": 22,
"ټ": 23,
"پ": 24,
"څ": 25,
"چ": 26,
"ډ": 27,
"ړ": 28,
"ښ": 29,
"ک": 30,
"ګ": 31,
"ی": 32,
"ې": 33
}