wav2vec2-darija / vocab.json
khaoulaoub's picture
Upload tokenizer
c1aea73
raw
history blame
489 Bytes
{
"/": 11,
"[PAD]": 40,
"[UNK]": 39,
"|": 21,
"ء": 10,
"آ": 36,
"أ": 22,
"ؤ": 2,
"ئ": 16,
"ا": 28,
"ب": 29,
"ة": 24,
"ت": 0,
"ث": 25,
"ج": 38,
"ح": 20,
"خ": 19,
"د": 30,
"ذ": 4,
"ر": 27,
"ز": 37,
"س": 32,
"ش": 7,
"ص": 14,
"ض": 12,
"ط": 18,
"ظ": 1,
"ع": 8,
"غ": 31,
"ف": 5,
"ق": 34,
"ك": 17,
"ل": 9,
"م": 35,
"ن": 26,
"ه": 13,
"و": 23,
"ى": 6,
"ي": 15,
"َ": 3,
"ْ": 33
}