tamasheq-99 / vocab.json
ad019el's picture
Upload tokenizer
03bb054
raw
history blame
727 Bytes
{
"!": 0,
"\"": 26,
",": 47,
".": 28,
":": 41,
"[PAD]": 60,
"[UNK]": 59,
"|": 4,
" ": 21,
"،": 16,
"؟": 56,
"ء": 30,
"آ": 37,
"أ": 38,
"ؤ": 5,
"إ": 49,
"ئ": 40,
"ا": 22,
"ب": 34,
"ة": 17,
"ت": 24,
"ث": 19,
"ج": 6,
"ح": 27,
"خ": 44,
"د": 9,
"ذ": 54,
"ر": 14,
"ز": 43,
"س": 55,
"ش": 29,
"ص": 2,
"ض": 50,
"ط": 31,
"ظ": 57,
"ع": 46,
"غ": 53,
"ـ": 25,
"ف": 48,
"ق": 39,
"ك": 1,
"ل": 11,
"م": 52,
"ن": 8,
"ه": 3,
"و": 51,
"ى": 35,
"ي": 42,
"ً": 12,
"ٌ": 32,
"ٍ": 45,
"َ": 10,
"ُ": 20,
"ِ": 58,
"ّ": 23,
"ْ": 15,
"ٱ": 36,
"ڤ": 18,
"ک": 7,
"ی": 13,
"—": 33
}