tuanmanh28's picture
Upload tokenizer
3da9bdd verified
raw
history blame contribute delete
438 Bytes
{
"[PAD]": 36,
"[UNK]": 35,
"a": 20,
"b": 15,
"c": 26,
"g": 0,
"h": 5,
"i": 10,
"k": 24,
"l": 12,
"m": 21,
"n": 8,
"p": 27,
"q": 31,
"t": 34,
"u": 30,
"v": 3,
"|": 7,
"à": 16,
"á": 25,
"è": 2,
"ò": 22,
"ô": 11,
"ă": 29,
"đ": 19,
"ư": 14,
"ạ": 23,
"ậ": 17,
"ắ": 9,
"ặ": 33,
"ế": 28,
"ệ": 32,
"ố": 1,
"ổ": 6,
"ộ": 13,
"ờ": 4,
"ủ": 18
}