tamasheq-99-2 / vocab.json
ad019el's picture
Upload tokenizer
1092c0e
raw
history blame
526 Bytes
{
"[PAD]": 43,
"[UNK]": 42,
"|": 20,
"ء": 17,
"آ": 1,
"أ": 27,
"ؤ": 28,
"إ": 41,
"ئ": 19,
"ا": 18,
"ب": 24,
"ة": 21,
"ت": 36,
"ث": 16,
"ج": 6,
"ح": 25,
"خ": 30,
"د": 15,
"ذ": 4,
"ر": 8,
"ز": 10,
"س": 29,
"ش": 26,
"ص": 33,
"ض": 35,
"ط": 0,
"ظ": 39,
"ع": 23,
"غ": 9,
"ف": 12,
"ق": 31,
"ك": 13,
"ل": 11,
"م": 40,
"ن": 34,
"ه": 38,
"و": 22,
"ى": 32,
"ي": 37,
"ٱ": 5,
"چ": 7,
"ڤ": 14,
"ک": 2,
"ی": 3
}