wav2vec2-darija / vocab.json
khaoulaoub's picture
Upload tokenizer
ed56332
raw
history blame
418 Bytes
{
"[PAD]": 34,
"[UNK]": 33,
"|": 30,
"ء": 10,
"ا": 29,
"ب": 0,
"ة": 22,
"ت": 18,
"ث": 14,
"ج": 17,
"ح": 11,
"خ": 23,
"د": 7,
"ذ": 13,
"ر": 3,
"ز": 2,
"س": 28,
"ش": 6,
"ص": 20,
"ض": 9,
"ط": 27,
"ظ": 26,
"ع": 4,
"غ": 1,
"ـ": 32,
"ف": 31,
"ق": 16,
"ك": 24,
"ل": 21,
"م": 5,
"ن": 12,
"ه": 8,
"و": 15,
"ى": 19,
"ي": 25
}