wav2vec2-darija / vocab.json
khaoulaoub's picture
Upload tokenizer
d2474bb
raw
history blame
418 Bytes
{
"[PAD]": 34,
"[UNK]": 33,
"|": 23,
"ء": 22,
"ا": 4,
"ب": 2,
"ة": 16,
"ت": 3,
"ث": 25,
"ج": 7,
"ح": 1,
"خ": 21,
"د": 12,
"ذ": 17,
"ر": 0,
"ز": 15,
"س": 31,
"ش": 11,
"ص": 30,
"ض": 6,
"ط": 26,
"ظ": 5,
"ع": 18,
"غ": 29,
"ـ": 19,
"ف": 14,
"ق": 10,
"ك": 8,
"ل": 32,
"م": 27,
"ن": 9,
"ه": 28,
"و": 24,
"ى": 13,
"ي": 20
}