ntxcong's picture
Upload tokenizer
3198ab4 verified
raw
history blame
1.18 kB
{
"'": 29,
"[PAD]": 96,
"[UNK]": 95,
"a": 93,
"b": 63,
"c": 83,
"d": 69,
"e": 78,
"f": 88,
"g": 52,
"h": 25,
"i": 66,
"j": 21,
"k": 42,
"l": 80,
"m": 7,
"n": 73,
"o": 22,
"p": 32,
"q": 5,
"r": 47,
"s": 70,
"t": 39,
"u": 3,
"v": 16,
"w": 79,
"x": 48,
"y": 11,
"z": 84,
"|": 26,
"à": 37,
"á": 12,
"â": 0,
"ã": 71,
"è": 55,
"é": 87,
"ê": 6,
"ì": 60,
"í": 62,
"ò": 76,
"ó": 13,
"ô": 86,
"õ": 74,
"ù": 23,
"ú": 36,
"ý": 33,
"ă": 50,
"đ": 82,
"ĩ": 1,
"ũ": 90,
"ơ": 85,
"ư": 58,
"ạ": 19,
"ả": 43,
"ấ": 20,
"ầ": 49,
"ẩ": 91,
"ẫ": 92,
"ậ": 65,
"ắ": 67,
"ằ": 18,
"ẳ": 8,
"ẵ": 15,
"ặ": 10,
"ẹ": 81,
"ẻ": 35,
"ẽ": 61,
"ế": 9,
"ề": 54,
"ể": 30,
"ễ": 44,
"ệ": 24,
"ỉ": 64,
"ị": 31,
"ọ": 4,
"ỏ": 72,
"ố": 68,
"ồ": 46,
"ổ": 89,
"ỗ": 45,
"ộ": 53,
"ớ": 28,
"ờ": 41,
"ở": 14,
"ỡ": 34,
"ợ": 40,
"ụ": 56,
"ủ": 27,
"ứ": 59,
"ừ": 17,
"ử": 57,
"ữ": 75,
"ự": 38,
"ỳ": 2,
"ỵ": 94,
"ỷ": 51,
"ỹ": 77
}