tamasheq-99-2 / vocab.json
ad019el's picture
Upload tokenizer
2f5a165
raw
history blame
514 Bytes
{
"[PAD]": 42,
"[UNK]": 41,
"|": 24,
"ء": 9,
"آ": 29,
"أ": 10,
"ؤ": 20,
"إ": 19,
"ئ": 40,
"ا": 1,
"ب": 13,
"ة": 27,
"ت": 25,
"ث": 28,
"ج": 0,
"ح": 8,
"خ": 34,
"د": 22,
"ذ": 18,
"ر": 35,
"ز": 32,
"س": 15,
"ش": 2,
"ص": 7,
"ض": 17,
"ط": 38,
"ظ": 16,
"ع": 5,
"غ": 39,
"ف": 14,
"ق": 4,
"ك": 21,
"ل": 23,
"م": 31,
"ن": 26,
"ه": 36,
"و": 37,
"ى": 30,
"ي": 11,
"ٱ": 12,
"ڤ": 6,
"ک": 3,
"ی": 33
}