hbm209's picture
Upload tokenizer
a10f38d verified
raw
history blame
605 Bytes
{
"[PAD]": 50,
"[UNK]": 49,
"a": 21,
"b": 47,
"c": 11,
"e": 5,
"g": 29,
"h": 44,
"i": 32,
"k": 19,
"l": 41,
"m": 42,
"n": 10,
"o": 20,
"p": 25,
"q": 31,
"r": 12,
"s": 43,
"t": 39,
"u": 37,
"v": 45,
"x": 9,
"y": 3,
"|": 14,
"à": 23,
"á": 34,
"ã": 26,
"ê": 22,
"ó": 0,
"ô": 8,
"ú": 4,
"đ": 28,
"ũ": 48,
"ơ": 16,
"ư": 18,
"ạ": 15,
"ả": 30,
"ầ": 40,
"ậ": 33,
"ắ": 38,
"ẹ": 7,
"ẽ": 36,
"ế": 2,
"ề": 1,
"ể": 17,
"ọ": 35,
"ố": 13,
"ỗ": 46,
"ớ": 6,
"ờ": 24,
"ợ": 27
}