ntxcong's picture
Upload tokenizer
62a84b7 verified
{
"'": 22,
"[PAD]": 96,
"[UNK]": 95,
"a": 35,
"b": 18,
"c": 73,
"d": 64,
"e": 67,
"f": 84,
"g": 75,
"h": 81,
"i": 79,
"j": 6,
"k": 49,
"l": 1,
"m": 71,
"n": 5,
"o": 7,
"p": 32,
"q": 38,
"r": 94,
"s": 80,
"t": 21,
"u": 63,
"v": 55,
"w": 37,
"x": 42,
"y": 86,
"z": 10,
"|": 20,
"à": 31,
"á": 0,
"â": 2,
"ã": 27,
"è": 59,
"é": 43,
"ê": 74,
"ì": 78,
"í": 52,
"ò": 90,
"ó": 24,
"ô": 82,
"õ": 92,
"ù": 89,
"ú": 53,
"ý": 66,
"ă": 13,
"đ": 9,
"ĩ": 56,
"ũ": 12,
"ơ": 30,
"ư": 76,
"ạ": 51,
"ả": 88,
"ấ": 17,
"ầ": 29,
"ẩ": 36,
"ẫ": 16,
"ậ": 3,
"ắ": 15,
"ằ": 26,
"ẳ": 39,
"ẵ": 44,
"ặ": 85,
"ẹ": 83,
"ẻ": 87,
"ẽ": 41,
"ế": 65,
"ề": 54,
"ể": 33,
"ễ": 34,
"ệ": 25,
"ỉ": 72,
"ị": 45,
"ọ": 47,
"ỏ": 28,
"ố": 46,
"ồ": 60,
"ổ": 77,
"ỗ": 48,
"ộ": 69,
"ớ": 57,
"ờ": 62,
"ở": 68,
"ỡ": 91,
"ợ": 61,
"ụ": 23,
"ủ": 58,
"ứ": 8,
"ừ": 19,
"ử": 14,
"ữ": 93,
"ự": 11,
"ỳ": 40,
"ỵ": 70,
"ỷ": 4,
"ỹ": 50
}