Thienpkae's picture
Upload tokenizer
05120d1 verified
raw
history blame
1.15 kB
{
"[PAD]": 93,
"[UNK]": 92,
"a": 1,
"b": 2,
"c": 3,
"d": 4,
"e": 5,
"g": 6,
"h": 7,
"i": 8,
"j": 9,
"k": 10,
"l": 11,
"m": 12,
"n": 13,
"o": 14,
"p": 15,
"q": 16,
"r": 17,
"s": 18,
"t": 19,
"u": 20,
"v": 21,
"w": 22,
"x": 23,
"y": 24,
"z": 25,
"|": 0,
"à": 26,
"á": 27,
"â": 28,
"ã": 29,
"è": 30,
"é": 31,
"ê": 32,
"ì": 33,
"í": 34,
"ò": 35,
"ó": 36,
"ô": 37,
"õ": 38,
"ù": 39,
"ú": 40,
"ý": 41,
"ă": 42,
"đ": 43,
"ĩ": 44,
"ũ": 45,
"ơ": 46,
"ư": 47,
"ạ": 48,
"ả": 49,
"ấ": 50,
"ầ": 51,
"ẩ": 52,
"ẫ": 53,
"ậ": 54,
"ắ": 55,
"ằ": 56,
"ẳ": 57,
"ẵ": 58,
"ặ": 59,
"ẹ": 60,
"ẻ": 61,
"ẽ": 62,
"ế": 63,
"ề": 64,
"ể": 65,
"ễ": 66,
"ệ": 67,
"ỉ": 68,
"ị": 69,
"ọ": 70,
"ỏ": 71,
"ố": 72,
"ồ": 73,
"ổ": 74,
"ỗ": 75,
"ộ": 76,
"ớ": 77,
"ờ": 78,
"ở": 79,
"ỡ": 80,
"ợ": 81,
"ụ": 82,
"ủ": 83,
"ứ": 84,
"ừ": 85,
"ử": 86,
"ữ": 87,
"ự": 88,
"ỳ": 89,
"ỷ": 90,
"ỹ": 91
}