hts98's picture
Upload tokenizer
0237f9d
raw
history blame
1.29 kB
{
"0": 18,
"1": 66,
"2": 98,
"3": 3,
"4": 101,
"5": 92,
"6": 30,
"7": 16,
"8": 38,
"9": 64,
"[PAD]": 105,
"[UNK]": 104,
"a": 87,
"b": 10,
"c": 94,
"d": 9,
"e": 17,
"f": 100,
"g": 85,
"h": 35,
"i": 36,
"j": 57,
"k": 2,
"l": 19,
"m": 65,
"n": 8,
"o": 61,
"p": 49,
"q": 70,
"r": 60,
"s": 6,
"t": 78,
"u": 1,
"v": 4,
"w": 80,
"x": 42,
"y": 84,
"z": 59,
"|": 23,
"à": 103,
"á": 41,
"â": 90,
"ã": 25,
"è": 39,
"é": 82,
"ê": 88,
"ì": 79,
"í": 0,
"ò": 83,
"ó": 5,
"ô": 55,
"õ": 56,
"ù": 81,
"ú": 77,
"ý": 29,
"ă": 34,
"đ": 47,
"ĩ": 48,
"ũ": 11,
"ơ": 26,
"ư": 27,
"ạ": 44,
"ả": 72,
"ấ": 76,
"ầ": 93,
"ẩ": 69,
"ẫ": 99,
"ậ": 33,
"ắ": 71,
"ằ": 24,
"ẳ": 45,
"ẵ": 62,
"ặ": 75,
"ẹ": 97,
"ẻ": 53,
"ẽ": 7,
"ế": 74,
"ề": 28,
"ể": 43,
"ễ": 102,
"ệ": 31,
"ỉ": 46,
"ị": 52,
"ọ": 40,
"ỏ": 96,
"ố": 13,
"ồ": 14,
"ổ": 21,
"ỗ": 15,
"ộ": 63,
"ớ": 22,
"ờ": 86,
"ở": 73,
"ỡ": 89,
"ợ": 68,
"ụ": 58,
"ủ": 50,
"ứ": 32,
"ừ": 91,
"ử": 95,
"ữ": 67,
"ự": 12,
"ỳ": 20,
"ỷ": 51,
"ỹ": 37,
"₫": 54
}