tuanmanh28's picture
Upload tokenizer
1845e96 verified
raw
history blame
462 Bytes
{
"[PAD]": 38,
"[UNK]": 37,
"a": 9,
"b": 19,
"c": 22,
"g": 20,
"h": 2,
"i": 35,
"k": 6,
"l": 5,
"m": 0,
"n": 7,
"p": 29,
"q": 4,
"s": 13,
"t": 34,
"u": 15,
"v": 12,
"|": 32,
"à": 11,
"á": 21,
"è": 16,
"ò": 25,
"ô": 3,
"ă": 33,
"đ": 26,
"ư": 17,
"ạ": 23,
"ậ": 30,
"ắ": 27,
"ặ": 28,
"ế": 1,
"ề": 14,
"ệ": 36,
"ố": 31,
"ổ": 8,
"ộ": 10,
"ờ": 18,
"ủ": 24
}