UrduWav2Vec2Large / vocab.json
mwz's picture
Upload tokenizer
af90afc
raw
history blame
1.04 kB
{
"'": 65,
"[PAD]": 85,
"[UNK]": 84,
"`": 49,
"|": 46,
"،": 83,
"ؑ": 69,
"ؓ": 48,
"ؔ": 27,
"؛": 7,
"؟": 33,
"ء": 29,
"آ": 58,
"أ": 0,
"ؤ": 4,
"ئ": 5,
"ا": 71,
"ب": 54,
"ت": 35,
"ث": 70,
"ج": 8,
"ح": 30,
"خ": 18,
"د": 14,
"ذ": 53,
"ر": 77,
"ز": 81,
"س": 32,
"ش": 78,
"ص": 68,
"ض": 55,
"ط": 15,
"ظ": 43,
"ع": 66,
"غ": 25,
"ف": 13,
"ق": 28,
"ك": 79,
"ل": 82,
"م": 6,
"ن": 11,
"ه": 50,
"و": 36,
"ى": 21,
"ي": 3,
"ً": 26,
"َ": 16,
"ُ": 73,
"ِ": 31,
"ّ": 57,
"ْ": 52,
"ٓ": 47,
"ٔ": 37,
"ٗ": 19,
"ٰ": 59,
"ٹ": 45,
"پ": 74,
"چ": 64,
"ڈ": 38,
"ڑ": 39,
"ژ": 24,
"ک": 2,
"گ": 1,
"ں": 40,
"ھ": 17,
"ہ": 20,
"ۂ": 9,
"ۃ": 10,
"ی": 51,
"ے": 60,
"ۓ": 61,
"۔": 42,
"’": 63,
"…": 67,
"ﭨ": 75,
"ﮭ": 41,
"ﮯ": 44,
"ﯾ": 56,
"ﷲ": 12,
"ﷺ": 34,
"ﺗ": 76,
"ﺘ": 22,
"ﺩ": 80,
"ﺲ": 72,
"ﻧ": 23,
"ﻮ": 62
}