wav2vec2-darija / vocab.json
khaoulaoub's picture
Upload tokenizer
a189b42
raw
history blame
418 Bytes
{
"[PAD]": 34,
"[UNK]": 33,
"|": 27,
"ء": 23,
"ا": 25,
"ب": 32,
"ة": 1,
"ت": 31,
"ث": 11,
"ج": 17,
"ح": 13,
"خ": 29,
"د": 8,
"ذ": 28,
"ر": 0,
"ز": 7,
"س": 15,
"ش": 9,
"ص": 26,
"ض": 19,
"ط": 4,
"ظ": 21,
"ع": 5,
"غ": 16,
"ـ": 22,
"ف": 24,
"ق": 18,
"ك": 14,
"ل": 3,
"م": 30,
"ن": 6,
"ه": 2,
"و": 10,
"ى": 12,
"ي": 20
}