tamasheq-99 / vocab.json
ad019el's picture
Upload tokenizer
8390c97
raw
history blame
526 Bytes
{
"[PAD]": 43,
"[UNK]": 42,
"|": 31,
"ء": 15,
"آ": 33,
"أ": 17,
"ؤ": 9,
"إ": 5,
"ئ": 29,
"ا": 18,
"ب": 7,
"ة": 3,
"ت": 26,
"ث": 2,
"ج": 10,
"ح": 11,
"خ": 1,
"د": 19,
"ذ": 16,
"ر": 6,
"ز": 14,
"س": 24,
"ش": 23,
"ص": 41,
"ض": 37,
"ط": 13,
"ظ": 4,
"ع": 38,
"غ": 34,
"ف": 28,
"ق": 21,
"ك": 25,
"ل": 35,
"م": 40,
"ن": 30,
"ه": 20,
"و": 8,
"ى": 12,
"ي": 27,
"ٰ": 39,
"ٱ": 36,
"ڤ": 32,
"ک": 0,
"ی": 22
}