tamasheq-99-2 / vocab.json
ad019el's picture
Upload tokenizer
3ed8865
raw
history blame
526 Bytes
{
"[PAD]": 43,
"[UNK]": 42,
"|": 2,
"ء": 21,
"آ": 39,
"أ": 32,
"ؤ": 20,
"إ": 17,
"ئ": 5,
"ا": 12,
"ب": 41,
"ة": 27,
"ت": 9,
"ث": 24,
"ج": 3,
"ح": 25,
"خ": 28,
"د": 10,
"ذ": 6,
"ر": 23,
"ز": 29,
"س": 8,
"ش": 14,
"ص": 11,
"ض": 34,
"ط": 30,
"ظ": 35,
"ع": 18,
"غ": 36,
"ف": 7,
"ق": 26,
"ك": 0,
"ل": 37,
"م": 38,
"ن": 19,
"ه": 16,
"و": 1,
"ى": 4,
"ي": 31,
"ٱ": 13,
"چ": 15,
"ڤ": 33,
"ک": 22,
"ی": 40
}