Lasion's picture
Upload tokenizer
f227fc3 verified
raw
history blame
1.13 kB
{
"[PAD]": 91,
"[UNK]": 90,
"a": 31,
"b": 22,
"c": 55,
"d": 63,
"e": 3,
"g": 56,
"h": 76,
"i": 36,
"k": 19,
"l": 48,
"m": 24,
"n": 26,
"o": 18,
"p": 14,
"q": 9,
"r": 67,
"s": 30,
"t": 17,
"u": 23,
"v": 43,
"x": 25,
"y": 2,
"|": 29,
"à": 35,
"á": 75,
"â": 34,
"ã": 78,
"è": 61,
"é": 11,
"ê": 32,
"ì": 71,
"í": 42,
"ò": 62,
"ó": 16,
"ô": 8,
"õ": 47,
"ù": 73,
"ú": 0,
"ý": 60,
"ă": 84,
"đ": 13,
"ĩ": 68,
"ũ": 86,
"ơ": 80,
"ư": 74,
"ạ": 38,
"ả": 58,
"ấ": 7,
"ầ": 21,
"ẩ": 12,
"ẫ": 87,
"ậ": 15,
"ắ": 89,
"ằ": 46,
"ẳ": 52,
"ẵ": 49,
"ặ": 85,
"ẹ": 57,
"ẻ": 5,
"ẽ": 83,
"ế": 33,
"ề": 77,
"ể": 10,
"ễ": 20,
"ệ": 44,
"ỉ": 27,
"ị": 59,
"ọ": 6,
"ỏ": 4,
"ố": 64,
"ồ": 41,
"ổ": 28,
"ỗ": 72,
"ộ": 37,
"ớ": 79,
"ờ": 65,
"ở": 81,
"ỡ": 82,
"ợ": 45,
"ụ": 88,
"ủ": 66,
"ứ": 51,
"ừ": 1,
"ử": 50,
"ữ": 70,
"ự": 69,
"ỳ": 54,
"ỵ": 40,
"ỷ": 53,
"ỹ": 39
}