tamasheq-99 / vocab.json
ad019el's picture
Upload tokenizer
60ed034
raw
history blame
514 Bytes
{
"[PAD]": 42,
"[UNK]": 41,
"|": 14,
"ء": 39,
"آ": 18,
"أ": 8,
"ؤ": 19,
"إ": 37,
"ئ": 4,
"ا": 22,
"ب": 29,
"ة": 9,
"ت": 20,
"ث": 27,
"ج": 36,
"ح": 12,
"خ": 15,
"د": 10,
"ذ": 33,
"ر": 7,
"ز": 32,
"س": 1,
"ش": 31,
"ص": 6,
"ض": 17,
"ط": 16,
"ظ": 13,
"ع": 30,
"غ": 25,
"ف": 2,
"ق": 40,
"ك": 35,
"ل": 0,
"م": 11,
"ن": 24,
"ه": 34,
"و": 28,
"ى": 21,
"ي": 5,
"ٱ": 26,
"ڤ": 23,
"ک": 3,
"ی": 38
}