tamasheq-99-2 / vocab.json
ad019el's picture
Upload tokenizer
815b491
raw
history blame
526 Bytes
{
"[PAD]": 43,
"[UNK]": 42,
"|": 24,
"ء": 12,
"آ": 35,
"أ": 15,
"ؤ": 3,
"إ": 8,
"ئ": 36,
"ا": 0,
"ب": 14,
"ة": 2,
"ت": 28,
"ث": 41,
"ج": 4,
"ح": 39,
"خ": 27,
"د": 16,
"ذ": 34,
"ر": 23,
"ز": 11,
"س": 26,
"ش": 5,
"ص": 20,
"ض": 38,
"ط": 32,
"ظ": 33,
"ع": 17,
"غ": 40,
"ف": 25,
"ق": 29,
"ك": 1,
"ل": 22,
"م": 31,
"ن": 13,
"ه": 10,
"و": 30,
"ى": 21,
"ي": 37,
"ٱ": 9,
"چ": 6,
"ڤ": 18,
"ک": 7,
"ی": 19
}