URDU-ASR / vocab.json
Shehryar718's picture
Upload tokenizer
eded62c
raw
history blame
1.04 kB
{
"'": 50,
"[PAD]": 85,
"[UNK]": 84,
"`": 13,
"|": 66,
"،": 76,
"ؑ": 74,
"ؓ": 40,
"ؔ": 64,
"؛": 75,
"؟": 14,
"ء": 81,
"آ": 57,
"أ": 39,
"ؤ": 78,
"ئ": 26,
"ا": 51,
"ب": 77,
"ت": 80,
"ث": 69,
"ج": 59,
"ح": 58,
"خ": 24,
"د": 27,
"ذ": 49,
"ر": 20,
"ز": 41,
"س": 62,
"ش": 17,
"ص": 55,
"ض": 54,
"ط": 56,
"ظ": 42,
"ع": 9,
"غ": 32,
"ف": 2,
"ق": 43,
"ك": 47,
"ل": 36,
"م": 63,
"ن": 72,
"ه": 44,
"و": 6,
"ى": 68,
"ي": 53,
"ً": 29,
"َ": 67,
"ُ": 8,
"ِ": 83,
"ّ": 30,
"ْ": 23,
"ٓ": 0,
"ٔ": 33,
"ٗ": 15,
"ٰ": 4,
"ٹ": 21,
"پ": 19,
"چ": 35,
"ڈ": 61,
"ڑ": 38,
"ژ": 16,
"ک": 52,
"گ": 70,
"ں": 1,
"ھ": 22,
"ہ": 28,
"ۂ": 79,
"ۃ": 5,
"ی": 10,
"ے": 7,
"ۓ": 25,
"۔": 11,
"’": 71,
"…": 31,
"ﭨ": 37,
"ﮭ": 73,
"ﮯ": 46,
"ﯾ": 48,
"ﷲ": 12,
"ﷺ": 45,
"ﺗ": 3,
"ﺘ": 18,
"ﺩ": 65,
"ﺲ": 82,
"ﻧ": 60,
"ﻮ": 34
}