navaaesarosh's picture
Upload tokenizer
ed6cc92
raw
history blame
660 Bytes
{
"[PAD]": 54,
"[UNK]": 53,
"|": 0,
"ء": 1,
"آ": 2,
"أ": 3,
"ؤ": 4,
"ئ": 5,
"ا": 6,
"ب": 7,
"ت": 8,
"ث": 9,
"ج": 10,
"ح": 11,
"خ": 12,
"د": 13,
"ذ": 14,
"ر": 15,
"ز": 16,
"س": 17,
"ش": 18,
"ص": 19,
"ض": 20,
"ط": 21,
"ظ": 22,
"ع": 23,
"غ": 24,
"ف": 25,
"ق": 26,
"ك": 27,
"ل": 28,
"م": 29,
"ن": 30,
"ه": 31,
"و": 32,
"ى": 33,
"ي": 34,
"ٹ": 35,
"پ": 36,
"چ": 37,
"ڈ": 38,
"ڑ": 39,
"ژ": 40,
"ک": 41,
"گ": 42,
"ں": 43,
"ھ": 44,
"ہ": 45,
"ۂ": 46,
"ۃ": 47,
"ی": 48,
"ے": 49,
"ۓ": 50,
"ﷲ": 51,
"ﷺ": 52
}