wav2vec2-darija / vocab.json
khaoulaoub's picture
Upload tokenizer
ac212f8
raw
history blame
418 Bytes
{
"[PAD]": 34,
"[UNK]": 33,
"|": 22,
"ء": 9,
"ا": 4,
"ب": 15,
"ة": 25,
"ت": 27,
"ث": 19,
"ج": 17,
"ح": 2,
"خ": 13,
"د": 21,
"ذ": 31,
"ر": 11,
"ز": 26,
"س": 12,
"ش": 32,
"ص": 30,
"ض": 10,
"ط": 5,
"ظ": 23,
"ع": 29,
"غ": 3,
"ـ": 28,
"ف": 20,
"ق": 16,
"ك": 18,
"ل": 1,
"م": 14,
"ن": 6,
"ه": 0,
"و": 7,
"ى": 8,
"ي": 24
}