tamasheq-99-2 / vocab.json
ad019el's picture
Upload tokenizer
16e790d
raw
history blame
526 Bytes
{
"[PAD]": 43,
"[UNK]": 42,
"|": 40,
"ء": 5,
"آ": 7,
"أ": 22,
"ؤ": 21,
"إ": 9,
"ئ": 26,
"ا": 38,
"ب": 31,
"ة": 19,
"ت": 11,
"ث": 29,
"ج": 18,
"ح": 32,
"خ": 0,
"د": 1,
"ذ": 27,
"ر": 14,
"ز": 33,
"س": 37,
"ش": 4,
"ص": 39,
"ض": 17,
"ط": 3,
"ظ": 16,
"ع": 24,
"غ": 10,
"ف": 25,
"ق": 28,
"ك": 20,
"ل": 35,
"م": 12,
"ن": 30,
"ه": 23,
"و": 8,
"ى": 13,
"ي": 6,
"ٱ": 36,
"چ": 2,
"ڤ": 15,
"ک": 41,
"ی": 34
}