Upload tokenizer
2aa88be
verified
|
{ |
|
"0": 63, |
|
"1": 94, |
|
"2": 55, |
|
"3": 4, |
|
"4": 34, |
|
"5": 107, |
|
"6": 1, |
|
"7": 11, |
|
"8": 43, |
|
"9": 51, |
|
"<pad>": 109, |
|
"<unk>": 108, |
|
"a": 17, |
|
"b": 36, |
|
"c": 49, |
|
"d": 44, |
|
"e": 65, |
|
"f": 77, |
|
"g": 58, |
|
"h": 71, |
|
"i": 57, |
|
"j": 22, |
|
"k": 96, |
|
"l": 18, |
|
"m": 67, |
|
"n": 26, |
|
"o": 105, |
|
"p": 98, |
|
"q": 20, |
|
"r": 47, |
|
"s": 39, |
|
"t": 56, |
|
"u": 31, |
|
"v": 15, |
|
"w": 35, |
|
"x": 81, |
|
"y": 32, |
|
"z": 87, |
|
"|": 46, |
|
"à": 24, |
|
"á": 62, |
|
"â": 72, |
|
"ã": 83, |
|
"è": 90, |
|
"é": 27, |
|
"ê": 12, |
|
"ì": 40, |
|
"í": 3, |
|
"ò": 92, |
|
"ó": 82, |
|
"ô": 30, |
|
"õ": 10, |
|
"ù": 53, |
|
"ú": 73, |
|
"ý": 6, |
|
"ă": 104, |
|
"đ": 89, |
|
"ĩ": 70, |
|
"ũ": 48, |
|
"ơ": 95, |
|
"ư": 106, |
|
"̀": 61, |
|
"́": 59, |
|
"̣": 86, |
|
"у": 29, |
|
"ạ": 50, |
|
"ả": 88, |
|
"ấ": 99, |
|
"ầ": 41, |
|
"ẩ": 7, |
|
"ẫ": 97, |
|
"ậ": 64, |
|
"ắ": 79, |
|
"ằ": 33, |
|
"ẳ": 68, |
|
"ẵ": 93, |
|
"ặ": 76, |
|
"ẹ": 103, |
|
"ẻ": 0, |
|
"ẽ": 100, |
|
"ế": 52, |
|
"ề": 9, |
|
"ể": 45, |
|
"ễ": 38, |
|
"ệ": 37, |
|
"ỉ": 101, |
|
"ị": 85, |
|
"ọ": 74, |
|
"ỏ": 14, |
|
"ố": 23, |
|
"ồ": 75, |
|
"ổ": 84, |
|
"ỗ": 25, |
|
"ộ": 66, |
|
"ớ": 102, |
|
"ờ": 21, |
|
"ở": 8, |
|
"ỡ": 54, |
|
"ợ": 69, |
|
"ụ": 2, |
|
"ủ": 28, |
|
"ứ": 13, |
|
"ừ": 91, |
|
"ử": 60, |
|
"ữ": 78, |
|
"ự": 19, |
|
"ỳ": 80, |
|
"ỵ": 42, |
|
"ỷ": 16, |
|
"ỹ": 5 |
|
} |
|
|