hts98's picture
Upload tokenizer
72c00b3
raw
history blame
1.31 kB
{
",": 90,
".": 9,
"0": 19,
"1": 94,
"2": 23,
"3": 85,
"4": 89,
"5": 20,
"6": 40,
"7": 96,
"8": 33,
"9": 48,
"[PAD]": 107,
"[UNK]": 106,
"a": 37,
"b": 92,
"c": 17,
"d": 88,
"e": 71,
"f": 80,
"g": 76,
"h": 91,
"i": 28,
"j": 103,
"k": 95,
"l": 44,
"m": 41,
"n": 68,
"o": 18,
"p": 59,
"q": 67,
"r": 83,
"s": 26,
"t": 35,
"u": 93,
"v": 87,
"w": 82,
"x": 60,
"y": 52,
"z": 70,
"|": 6,
"à": 105,
"á": 99,
"â": 1,
"ã": 25,
"è": 24,
"é": 84,
"ê": 47,
"ì": 46,
"í": 34,
"ò": 51,
"ó": 39,
"ô": 32,
"õ": 63,
"ù": 65,
"ú": 72,
"ý": 3,
"ă": 74,
"đ": 8,
"ĩ": 79,
"ũ": 36,
"ơ": 69,
"ư": 29,
"ạ": 16,
"ả": 64,
"ấ": 5,
"ầ": 54,
"ẩ": 31,
"ẫ": 7,
"ậ": 57,
"ắ": 101,
"ằ": 53,
"ẳ": 98,
"ẵ": 11,
"ặ": 61,
"ẹ": 78,
"ẻ": 0,
"ẽ": 21,
"ế": 49,
"ề": 58,
"ể": 27,
"ễ": 77,
"ệ": 50,
"ỉ": 38,
"ị": 13,
"ọ": 42,
"ỏ": 56,
"ố": 15,
"ồ": 10,
"ổ": 86,
"ỗ": 62,
"ộ": 55,
"ớ": 14,
"ờ": 43,
"ở": 45,
"ỡ": 104,
"ợ": 4,
"ụ": 102,
"ủ": 30,
"ứ": 22,
"ừ": 66,
"ử": 12,
"ữ": 81,
"ự": 75,
"ỳ": 73,
"ỷ": 2,
"ỹ": 97,
"₫": 100
}