xls-r_cv_ur / vocab.json
hadiqa123's picture
add tokenizer
5ea20c4
raw
history blame
No virus
508 Bytes
{
" ": 23,
"آ": 1,
"أ": 33,
"ؤ": 0,
"ئ": 15,
"ا": 41,
"ب": 24,
"ت": 5,
"ث": 11,
"ج": 36,
"ح": 38,
"خ": 20,
"د": 8,
"ذ": 37,
"ر": 14,
"ز": 2,
"س": 9,
"ش": 3,
"ص": 31,
"ض": 12,
"ط": 35,
"ظ": 29,
"ع": 7,
"غ": 39,
"ف": 28,
"ق": 16,
"ل": 22,
"م": 13,
"ن": 32,
"و": 21,
"ً": 19,
"ٹ": 27,
"پ": 4,
"چ": 40,
"ڈ": 30,
"ڑ": 17,
"ک": 18,
"گ": 25,
"ں": 34,
"ھ": 10,
"ہ": 42,
"ی": 26,
"ے": 6
}