tamasheq-98 / vocab.json
ad019el's picture
Upload tokenizer
7e23fd5
raw
history blame
514 Bytes
{
"[PAD]": 42,
"[UNK]": 41,
"|": 27,
"ء": 26,
"آ": 38,
"أ": 24,
"ؤ": 40,
"إ": 16,
"ئ": 30,
"ا": 9,
"ب": 29,
"ة": 11,
"ت": 31,
"ث": 34,
"ج": 17,
"ح": 6,
"خ": 1,
"د": 2,
"ذ": 13,
"ر": 14,
"ز": 8,
"س": 18,
"ش": 15,
"ص": 20,
"ض": 7,
"ط": 37,
"ظ": 36,
"ع": 32,
"غ": 4,
"ف": 10,
"ق": 35,
"ك": 5,
"ل": 39,
"م": 21,
"ن": 33,
"ه": 28,
"و": 22,
"ى": 19,
"ي": 12,
"ٱ": 25,
"ڤ": 0,
"ک": 3,
"ی": 23
}