tamasheq-99 / vocab.json
ad019el's picture
Upload tokenizer
9c6b595
raw
history blame
514 Bytes
{
"[PAD]": 42,
"[UNK]": 41,
"|": 18,
"ء": 9,
"آ": 29,
"أ": 37,
"ؤ": 16,
"إ": 6,
"ئ": 34,
"ا": 7,
"ب": 35,
"ة": 32,
"ت": 24,
"ث": 26,
"ج": 22,
"ح": 4,
"خ": 38,
"د": 25,
"ذ": 20,
"ر": 15,
"ز": 40,
"س": 31,
"ش": 11,
"ص": 23,
"ض": 19,
"ط": 28,
"ظ": 39,
"ع": 2,
"غ": 17,
"ف": 27,
"ق": 21,
"ك": 12,
"ل": 13,
"م": 1,
"ن": 5,
"ه": 8,
"و": 14,
"ى": 3,
"ي": 0,
"ٱ": 33,
"ڤ": 36,
"ک": 10,
"ی": 30
}