UrduWav2Vec2Large / vocab.json
mwz's picture
Upload tokenizer
8e79249
raw
history blame
1.04 kB
{
"'": 10,
"[PAD]": 85,
"[UNK]": 84,
"`": 22,
"|": 1,
"،": 55,
"ؑ": 57,
"ؓ": 72,
"ؔ": 71,
"؛": 50,
"؟": 36,
"ء": 26,
"آ": 8,
"أ": 4,
"ؤ": 45,
"ئ": 48,
"ا": 39,
"ب": 12,
"ت": 67,
"ث": 83,
"ج": 78,
"ح": 66,
"خ": 74,
"د": 65,
"ذ": 5,
"ر": 82,
"ز": 60,
"س": 21,
"ش": 41,
"ص": 42,
"ض": 0,
"ط": 20,
"ظ": 49,
"ع": 17,
"غ": 81,
"ف": 70,
"ق": 46,
"ك": 6,
"ل": 28,
"م": 25,
"ن": 11,
"ه": 34,
"و": 61,
"ى": 80,
"ي": 52,
"ً": 19,
"َ": 9,
"ُ": 15,
"ِ": 31,
"ّ": 35,
"ْ": 27,
"ٓ": 2,
"ٔ": 23,
"ٗ": 47,
"ٰ": 16,
"ٹ": 38,
"پ": 69,
"چ": 40,
"ڈ": 54,
"ڑ": 33,
"ژ": 62,
"ک": 64,
"گ": 73,
"ں": 43,
"ھ": 75,
"ہ": 59,
"ۂ": 13,
"ۃ": 56,
"ی": 3,
"ے": 51,
"ۓ": 68,
"۔": 7,
"’": 79,
"…": 76,
"ﭨ": 53,
"ﮭ": 24,
"ﮯ": 63,
"ﯾ": 29,
"ﷲ": 14,
"ﷺ": 77,
"ﺗ": 32,
"ﺘ": 30,
"ﺩ": 37,
"ﺲ": 58,
"ﻧ": 44,
"ﻮ": 18
}