tamasheq-99-1 / vocab.json
ad019el's picture
Upload tokenizer
c060dac
raw
history blame
514 Bytes
{
"[PAD]": 42,
"[UNK]": 41,
"|": 28,
"ء": 25,
"آ": 19,
"أ": 40,
"ؤ": 4,
"إ": 35,
"ئ": 16,
"ا": 15,
"ب": 2,
"ة": 5,
"ت": 29,
"ث": 10,
"ج": 38,
"ح": 18,
"خ": 12,
"د": 37,
"ذ": 24,
"ر": 23,
"ز": 14,
"س": 26,
"ش": 13,
"ص": 9,
"ض": 3,
"ط": 20,
"ظ": 39,
"ع": 21,
"غ": 34,
"ف": 22,
"ق": 7,
"ك": 11,
"ل": 36,
"م": 17,
"ن": 32,
"ه": 27,
"و": 0,
"ى": 33,
"ي": 30,
"ٱ": 1,
"ڤ": 8,
"ک": 31,
"ی": 6
}