wav2vec2-darija / vocab.json
khaoulaoub's picture
Upload tokenizer
88b2c43
raw
history blame
418 Bytes
{
"[PAD]": 34,
"[UNK]": 33,
"|": 1,
"ء": 25,
"ا": 8,
"ب": 15,
"ة": 18,
"ت": 12,
"ث": 0,
"ج": 28,
"ح": 17,
"خ": 22,
"د": 14,
"ذ": 4,
"ر": 26,
"ز": 21,
"س": 24,
"ش": 19,
"ص": 9,
"ض": 23,
"ط": 32,
"ظ": 3,
"ع": 11,
"غ": 5,
"ـ": 13,
"ف": 2,
"ق": 6,
"ك": 20,
"ل": 10,
"م": 31,
"ن": 30,
"ه": 27,
"و": 29,
"ى": 7,
"ي": 16
}