w2v2-bert-urdu / vocab.json
UmarRamzan's picture
Upload tokenizer
f535954 verified
raw
history blame
779 Bytes
{
"[PAD]": 63,
"[UNK]": 62,
"|": 0,
"أ": 1,
"ؤ": 2,
"ئ": 3,
"ا": 4,
"ب": 5,
"ت": 6,
"ث": 7,
"ج": 8,
"ح": 9,
"خ": 10,
"د": 11,
"ذ": 12,
"ر": 13,
"ز": 14,
"س": 15,
"ش": 16,
"ص": 17,
"ض": 18,
"ط": 19,
"ظ": 20,
"ع": 21,
"غ": 22,
"ف": 23,
"ق": 24,
"ك": 25,
"ل": 26,
"م": 27,
"ن": 28,
"ه": 29,
"و": 30,
"ى": 31,
"ي": 32,
"ٹ": 33,
"پ": 34,
"چ": 35,
"ڈ": 36,
"ڑ": 37,
"ژ": 38,
"ک": 39,
"گ": 40,
"ں": 41,
"ھ": 42,
"ہ": 43,
"ۂ": 44,
"ۃ": 45,
"ی": 46,
"ے": 47,
"ۓ": 48,
"ﭨ": 49,
"ﮭ": 50,
"ﮯ": 51,
"ﯾ": 52,
"ﷲ": 53,
"ﷺ": 54,
"ﺅ": 55,
"ﺗ": 56,
"ﺘ": 57,
"ﺩ": 58,
"ﺲ": 59,
"ﻧ": 60,
"ﻮ": 61
}