Lasion's picture
Upload tokenizer
ecfe97c verified
raw
history blame
1.17 kB
{
"[PAD]": 95,
"[UNK]": 94,
"a": 73,
"b": 88,
"c": 70,
"d": 3,
"e": 71,
"f": 47,
"g": 28,
"h": 75,
"i": 24,
"j": 80,
"k": 39,
"l": 16,
"m": 22,
"n": 40,
"o": 32,
"p": 83,
"q": 54,
"r": 0,
"s": 61,
"t": 66,
"u": 52,
"v": 34,
"w": 84,
"x": 9,
"y": 92,
"z": 35,
"|": 79,
"à": 41,
"á": 14,
"â": 27,
"ã": 76,
"è": 74,
"é": 87,
"ê": 81,
"ì": 53,
"í": 69,
"ò": 5,
"ó": 31,
"ô": 2,
"õ": 19,
"ù": 44,
"ú": 25,
"ý": 7,
"ă": 60,
"đ": 12,
"ĩ": 90,
"ũ": 42,
"ơ": 10,
"ư": 6,
"ạ": 50,
"ả": 93,
"ấ": 38,
"ầ": 51,
"ẩ": 33,
"ẫ": 43,
"ậ": 85,
"ắ": 45,
"ằ": 11,
"ẳ": 17,
"ẵ": 89,
"ặ": 26,
"ẹ": 57,
"ẻ": 49,
"ẽ": 23,
"ế": 4,
"ề": 37,
"ể": 1,
"ễ": 62,
"ệ": 20,
"ỉ": 91,
"ị": 86,
"ọ": 30,
"ỏ": 58,
"ố": 15,
"ồ": 8,
"ổ": 18,
"ỗ": 46,
"ộ": 82,
"ớ": 64,
"ờ": 68,
"ở": 78,
"ỡ": 56,
"ợ": 48,
"ụ": 13,
"ủ": 21,
"ứ": 65,
"ừ": 67,
"ử": 77,
"ữ": 55,
"ự": 72,
"ỳ": 29,
"ỵ": 63,
"ỷ": 36,
"ỹ": 59
}