QuangTuan's picture
Upload tokenizer
6d5dbba
{
"%": 53,
"0": 49,
"1": 89,
"2": 74,
"3": 30,
"4": 70,
"5": 37,
"6": 78,
"7": 27,
"8": 103,
"9": 88,
"[PAD]": 109,
"[UNK]": 108,
"a": 58,
"b": 24,
"c": 29,
"d": 76,
"e": 0,
"f": 105,
"g": 54,
"h": 64,
"i": 102,
"k": 48,
"l": 26,
"m": 63,
"n": 99,
"o": 10,
"p": 42,
"q": 104,
"r": 34,
"s": 98,
"t": 80,
"u": 95,
"v": 85,
"w": 94,
"x": 21,
"y": 12,
"z": 22,
"|": 82,
"à": 5,
"á": 7,
"â": 4,
"ã": 61,
"è": 83,
"é": 35,
"ê": 32,
"ì": 38,
"í": 86,
"ò": 75,
"ó": 23,
"ô": 62,
"õ": 18,
"ù": 2,
"ú": 52,
"ý": 101,
"ă": 39,
"đ": 79,
"ĩ": 106,
"ũ": 71,
"ơ": 73,
"ư": 93,
"̀": 36,
"́": 92,
"̣": 50,
"у": 17,
"ạ": 66,
"ả": 51,
"ấ": 91,
"ầ": 33,
"ẩ": 1,
"ẫ": 96,
"ậ": 65,
"ắ": 43,
"ằ": 67,
"ẳ": 100,
"ẵ": 68,
"ặ": 45,
"ẹ": 77,
"ẻ": 31,
"ẽ": 72,
"ế": 16,
"ề": 44,
"ể": 8,
"ễ": 9,
"ệ": 13,
"ỉ": 87,
"ị": 107,
"ọ": 57,
"ỏ": 20,
"ố": 15,
"ồ": 59,
"ổ": 11,
"ỗ": 97,
"ộ": 28,
"ớ": 41,
"ờ": 19,
"ở": 6,
"ỡ": 60,
"ợ": 3,
"ụ": 14,
"ủ": 40,
"ứ": 69,
"ừ": 47,
"ử": 25,
"ữ": 84,
"ự": 81,
"ỳ": 46,
"ỵ": 90,
"ỷ": 56,
"ỹ": 55
}