xls-r_cv_ur / vocab.json
hadiqa123's picture
add tokenizer
6b77a93
raw
history blame
No virus
536 Bytes
{
" ": 4,
"PAD": 43,
"[UNK]": 44,
"آ": 14,
"أ": 24,
"ؤ": 35,
"ئ": 28,
"ا": 22,
"ب": 5,
"ت": 17,
"ث": 36,
"ج": 19,
"ح": 10,
"خ": 18,
"د": 12,
"ذ": 15,
"ر": 6,
"ز": 13,
"س": 33,
"ش": 16,
"ص": 38,
"ض": 25,
"ط": 23,
"ظ": 1,
"ع": 7,
"غ": 32,
"ف": 27,
"ق": 31,
"ل": 30,
"م": 8,
"ن": 21,
"و": 0,
"ً": 3,
"ٹ": 26,
"پ": 11,
"چ": 39,
"ڈ": 37,
"ڑ": 34,
"ک": 41,
"گ": 20,
"ں": 2,
"ھ": 29,
"ہ": 9,
"ی": 42,
"ے": 40
}