hbm209's picture
Upload tokenizer
dd65ab9 verified
raw
history blame contribute delete
605 Bytes
{
"[PAD]": 50,
"[UNK]": 49,
"a": 14,
"b": 11,
"c": 46,
"e": 37,
"g": 15,
"h": 26,
"i": 29,
"k": 24,
"l": 0,
"m": 31,
"n": 41,
"o": 5,
"p": 28,
"q": 35,
"r": 19,
"s": 36,
"t": 25,
"u": 47,
"v": 30,
"x": 43,
"y": 20,
"|": 16,
"à": 42,
"á": 45,
"ã": 7,
"ê": 13,
"ó": 6,
"ô": 21,
"ú": 8,
"đ": 9,
"ũ": 44,
"ơ": 27,
"ư": 18,
"ạ": 12,
"ả": 4,
"ầ": 10,
"ậ": 34,
"ắ": 38,
"ẹ": 40,
"ẽ": 3,
"ế": 33,
"ề": 23,
"ể": 1,
"ọ": 2,
"ố": 17,
"ỗ": 22,
"ớ": 32,
"ờ": 48,
"ợ": 39
}