wav2vec2-darija / vocab.json
khaoulaoub's picture
Upload tokenizer
8cf6454
raw
history blame
500 Bytes
{
"*": 18,
"[PAD]": 41,
"[UNK]": 40,
"d": 32,
"|": 29,
"ء": 23,
"آ": 14,
"أ": 35,
"ؤ": 34,
"إ": 9,
"ئ": 2,
"ا": 8,
"ب": 7,
"ة": 13,
"ت": 33,
"ث": 37,
"ج": 20,
"ح": 3,
"خ": 11,
"د": 30,
"ذ": 1,
"ر": 27,
"ز": 10,
"س": 17,
"ش": 36,
"ص": 4,
"ض": 26,
"ط": 6,
"ظ": 21,
"ع": 28,
"غ": 38,
"ف": 25,
"ق": 5,
"ك": 22,
"ل": 24,
"م": 15,
"ن": 19,
"ه": 39,
"و": 0,
"ى": 31,
"ي": 16,
"ڭ": 12
}