ntxcong's picture
Upload tokenizer
3abcad2 verified
raw
history blame
1.18 kB
{
"'": 10,
"[PAD]": 96,
"[UNK]": 95,
"a": 44,
"b": 48,
"c": 0,
"d": 72,
"e": 34,
"f": 57,
"g": 92,
"h": 22,
"i": 64,
"j": 78,
"k": 19,
"l": 25,
"m": 16,
"n": 47,
"o": 41,
"p": 75,
"q": 3,
"r": 77,
"s": 74,
"t": 91,
"u": 68,
"v": 54,
"w": 59,
"x": 73,
"y": 17,
"z": 86,
"|": 24,
"à": 9,
"á": 4,
"â": 7,
"ã": 65,
"è": 18,
"é": 1,
"ê": 33,
"ì": 88,
"í": 81,
"ò": 43,
"ó": 21,
"ô": 12,
"õ": 51,
"ù": 66,
"ú": 80,
"ý": 61,
"ă": 71,
"đ": 20,
"ĩ": 89,
"ũ": 46,
"ơ": 69,
"ư": 6,
"ạ": 67,
"ả": 53,
"ấ": 85,
"ầ": 14,
"ẩ": 56,
"ẫ": 45,
"ậ": 60,
"ắ": 82,
"ằ": 76,
"ẳ": 39,
"ẵ": 35,
"ặ": 27,
"ẹ": 26,
"ẻ": 38,
"ẽ": 94,
"ế": 90,
"ề": 36,
"ể": 52,
"ễ": 62,
"ệ": 28,
"ỉ": 58,
"ị": 79,
"ọ": 50,
"ỏ": 8,
"ố": 63,
"ồ": 40,
"ổ": 87,
"ỗ": 29,
"ộ": 23,
"ớ": 93,
"ờ": 70,
"ở": 37,
"ỡ": 30,
"ợ": 2,
"ụ": 32,
"ủ": 83,
"ứ": 84,
"ừ": 13,
"ử": 11,
"ữ": 42,
"ự": 55,
"ỳ": 5,
"ỵ": 49,
"ỷ": 15,
"ỹ": 31
}