sanchit-gandhi's picture
Upload tokenizer
8906d74
raw
history blame
483 Bytes
{
" ": 40,
"'": 38,
"-": 26,
"|": 1,
"آ": 19,
"ئ": 33,
"ا": 2,
"ب": 11,
"ت": 13,
"ث": 34,
"ج": 24,
"ح": 25,
"خ": 23,
"د": 6,
"ذ": 36,
"ر": 4,
"ز": 15,
"س": 10,
"ش": 16,
"ص": 29,
"ض": 31,
"ط": 28,
"ظ": 32,
"ع": 22,
"غ": 20,
"ف": 27,
"ق": 14,
"ل": 5,
"م": 9,
"ن": 3,
"ه": 12,
"و": 7,
"ٰ": 35,
"پ": 17,
"چ": 21,
"ژ": 39,
"ک": 18,
"گ": 8,
"ی": 0,
"‍": 30,
"–": 37
}