tamasheq-99 / vocab.json
ad019el's picture
Upload tokenizer
13b36d3
raw
history blame
514 Bytes
{
"[PAD]": 42,
"[UNK]": 41,
"|": 13,
"ء": 0,
"آ": 27,
"أ": 24,
"ؤ": 2,
"إ": 25,
"ئ": 5,
"ا": 11,
"ب": 33,
"ة": 28,
"ت": 26,
"ث": 12,
"ج": 6,
"ح": 32,
"خ": 8,
"د": 9,
"ذ": 4,
"ر": 31,
"ز": 23,
"س": 17,
"ش": 15,
"ص": 3,
"ض": 29,
"ط": 19,
"ظ": 20,
"ع": 1,
"غ": 39,
"ف": 38,
"ق": 7,
"ك": 10,
"ل": 36,
"م": 16,
"ن": 35,
"ه": 22,
"و": 21,
"ى": 34,
"ي": 37,
"ٱ": 40,
"ڤ": 14,
"ک": 18,
"ی": 30
}