hts98's picture
Upload tokenizer
1ab1344
raw
history blame
1.29 kB
{
"0": 58,
"1": 82,
"2": 8,
"3": 101,
"4": 41,
"5": 17,
"6": 98,
"7": 102,
"8": 43,
"9": 80,
"[PAD]": 105,
"[UNK]": 104,
"a": 18,
"b": 56,
"c": 6,
"d": 53,
"e": 42,
"f": 5,
"g": 36,
"h": 0,
"i": 2,
"j": 55,
"k": 61,
"l": 22,
"m": 12,
"n": 76,
"o": 68,
"p": 7,
"q": 81,
"r": 91,
"s": 57,
"t": 84,
"u": 45,
"v": 28,
"w": 69,
"x": 25,
"y": 85,
"z": 20,
"|": 63,
"à": 52,
"á": 19,
"â": 13,
"ã": 62,
"è": 26,
"é": 23,
"ê": 86,
"ì": 65,
"í": 79,
"ò": 49,
"ó": 100,
"ô": 4,
"õ": 77,
"ù": 90,
"ú": 33,
"ý": 50,
"ă": 93,
"đ": 72,
"ĩ": 96,
"ũ": 47,
"ơ": 66,
"ư": 99,
"ạ": 97,
"ả": 67,
"ấ": 16,
"ầ": 38,
"ẩ": 71,
"ẫ": 73,
"ậ": 83,
"ắ": 29,
"ằ": 30,
"ẳ": 3,
"ẵ": 10,
"ặ": 14,
"ẹ": 27,
"ẻ": 78,
"ẽ": 46,
"ế": 32,
"ề": 24,
"ể": 34,
"ễ": 48,
"ệ": 39,
"ỉ": 89,
"ị": 94,
"ọ": 35,
"ỏ": 37,
"ố": 1,
"ồ": 59,
"ổ": 92,
"ỗ": 51,
"ộ": 64,
"ớ": 21,
"ờ": 95,
"ở": 88,
"ỡ": 74,
"ợ": 11,
"ụ": 44,
"ủ": 54,
"ứ": 87,
"ừ": 60,
"ử": 75,
"ữ": 70,
"ự": 103,
"ỳ": 9,
"ỷ": 15,
"ỹ": 40,
"₫": 31
}