omar47's picture
add tokenizer
a5548c0
{
".": 1,
"[PAD]": 57,
"[UNK]": 56,
"|": 0,
"ؑ": 2,
"ؓ": 3,
"ؔ": 4,
"؛": 5,
"ء": 6,
"آ": 7,
"أ": 8,
"ؤ": 9,
"ئ": 10,
"ا": 11,
"ب": 12,
"ت": 13,
"ث": 14,
"ج": 15,
"ح": 16,
"خ": 17,
"د": 18,
"ذ": 19,
"ر": 20,
"ز": 21,
"س": 22,
"ش": 23,
"ص": 24,
"ض": 25,
"ط": 26,
"ظ": 27,
"ع": 28,
"غ": 29,
"ف": 30,
"ق": 31,
"ل": 32,
"م": 33,
"ن": 34,
"و": 35,
"ْ": 36,
"ٗ": 37,
"ٹ": 38,
"پ": 39,
"چ": 40,
"ڈ": 41,
"ڑ": 42,
"ژ": 43,
"ک": 44,
"گ": 45,
"ں": 46,
"ھ": 47,
"ہ": 48,
"ۂ": 49,
"ۃ": 50,
"ی": 51,
"ے": 52,
"ۓ": 53,
"ﷲ": 54,
"ﷺ": 55
}