URDU-ASR / vocab.json
Shehryar718's picture
Upload tokenizer
b77f97f
raw
history blame
1.04 kB
{
"'": 65,
"[PAD]": 85,
"[UNK]": 84,
"`": 34,
"|": 56,
"،": 13,
"ؑ": 81,
"ؓ": 54,
"ؔ": 7,
"؛": 60,
"؟": 23,
"ء": 30,
"آ": 8,
"أ": 2,
"ؤ": 33,
"ئ": 4,
"ا": 22,
"ب": 14,
"ت": 25,
"ث": 28,
"ج": 57,
"ح": 66,
"خ": 64,
"د": 5,
"ذ": 29,
"ر": 41,
"ز": 80,
"س": 20,
"ش": 68,
"ص": 62,
"ض": 24,
"ط": 21,
"ظ": 11,
"ع": 70,
"غ": 19,
"ف": 53,
"ق": 38,
"ك": 52,
"ل": 69,
"م": 3,
"ن": 82,
"ه": 61,
"و": 73,
"ى": 51,
"ي": 26,
"ً": 78,
"َ": 77,
"ُ": 43,
"ِ": 49,
"ّ": 36,
"ْ": 18,
"ٓ": 76,
"ٔ": 12,
"ٗ": 58,
"ٰ": 59,
"ٹ": 45,
"پ": 27,
"چ": 31,
"ڈ": 55,
"ڑ": 42,
"ژ": 16,
"ک": 35,
"گ": 40,
"ں": 75,
"ھ": 39,
"ہ": 6,
"ۂ": 74,
"ۃ": 1,
"ی": 48,
"ے": 9,
"ۓ": 37,
"۔": 71,
"’": 10,
"…": 83,
"ﭨ": 44,
"ﮭ": 72,
"ﮯ": 79,
"ﯾ": 15,
"ﷲ": 67,
"ﷺ": 17,
"ﺗ": 63,
"ﺘ": 46,
"ﺩ": 50,
"ﺲ": 47,
"ﻧ": 32,
"ﻮ": 0
}