hts98's picture
Upload tokenizer
e4cfa88
raw
history blame
1.31 kB
{
",": 69,
".": 22,
"0": 24,
"1": 50,
"2": 48,
"3": 25,
"4": 99,
"5": 42,
"6": 87,
"7": 105,
"8": 18,
"9": 64,
"[PAD]": 107,
"[UNK]": 106,
"a": 93,
"b": 73,
"c": 35,
"d": 43,
"e": 68,
"f": 75,
"g": 84,
"h": 77,
"i": 101,
"j": 19,
"k": 92,
"l": 5,
"m": 81,
"n": 103,
"o": 7,
"p": 45,
"q": 95,
"r": 102,
"s": 1,
"t": 12,
"u": 104,
"v": 16,
"w": 20,
"x": 3,
"y": 23,
"z": 11,
"|": 2,
"à": 83,
"á": 57,
"â": 80,
"ã": 54,
"è": 30,
"é": 78,
"ê": 37,
"ì": 14,
"í": 62,
"ò": 28,
"ó": 33,
"ô": 53,
"õ": 9,
"ù": 85,
"ú": 86,
"ý": 15,
"ă": 72,
"đ": 52,
"ĩ": 59,
"ũ": 49,
"ơ": 21,
"ư": 29,
"ạ": 97,
"ả": 51,
"ấ": 61,
"ầ": 63,
"ẩ": 55,
"ẫ": 79,
"ậ": 66,
"ắ": 41,
"ằ": 10,
"ẳ": 26,
"ẵ": 13,
"ặ": 4,
"ẹ": 40,
"ẻ": 17,
"ẽ": 88,
"ế": 65,
"ề": 58,
"ể": 90,
"ễ": 67,
"ệ": 44,
"ỉ": 96,
"ị": 98,
"ọ": 34,
"ỏ": 46,
"ố": 71,
"ồ": 0,
"ổ": 100,
"ỗ": 8,
"ộ": 70,
"ớ": 94,
"ờ": 47,
"ở": 89,
"ỡ": 31,
"ợ": 38,
"ụ": 27,
"ủ": 36,
"ứ": 32,
"ừ": 76,
"ử": 60,
"ữ": 39,
"ự": 6,
"ỳ": 91,
"ỷ": 74,
"ỹ": 56,
"₫": 82
}