xls-r_cv_ur / vocab.json
Bakht Ullah
Upload tokenizer
d97877e
raw
history blame
550 Bytes
{
"[PAD]": 45,
"[UNK]": 44,
"|": 0,
"ء": 1,
"ا": 2,
"ب": 3,
"ت": 4,
"ث": 5,
"ج": 6,
"ح": 7,
"خ": 8,
"د": 9,
"ذ": 10,
"ر": 11,
"ز": 12,
"س": 13,
"ش": 14,
"ص": 15,
"ض": 16,
"ط": 17,
"ظ": 18,
"ع": 19,
"غ": 20,
"ف": 21,
"ق": 22,
"ل": 23,
"م": 24,
"ن": 25,
"و": 26,
"ى": 27,
"ي": 28,
"ً": 29,
"ٓ": 30,
"ٹ": 31,
"پ": 32,
"چ": 33,
"ڈ": 34,
"ڑ": 35,
"ژ": 36,
"ک": 37,
"گ": 38,
"ں": 39,
"ھ": 40,
"ہ": 41,
"ی": 42,
"ے": 43
}