hbm209's picture
Upload tokenizer
4af80a4 verified
raw
history blame
605 Bytes
{
"[PAD]": 50,
"[UNK]": 49,
"a": 30,
"b": 14,
"c": 1,
"e": 39,
"g": 29,
"h": 23,
"i": 10,
"k": 41,
"l": 31,
"m": 4,
"n": 3,
"o": 36,
"p": 38,
"q": 5,
"r": 8,
"s": 45,
"t": 33,
"u": 2,
"v": 37,
"x": 13,
"y": 47,
"|": 48,
"à": 19,
"á": 22,
"ã": 44,
"ê": 46,
"ó": 35,
"ô": 20,
"ú": 11,
"đ": 7,
"ũ": 9,
"ơ": 42,
"ư": 21,
"ạ": 25,
"ả": 28,
"ầ": 16,
"ậ": 32,
"ắ": 26,
"ẹ": 6,
"ẽ": 40,
"ế": 17,
"ề": 27,
"ể": 0,
"ọ": 12,
"ố": 34,
"ỗ": 18,
"ớ": 24,
"ờ": 15,
"ợ": 43
}