hungadc's picture
Upload tokenizer
a9aa948 verified
raw
history blame
1.18 kB
{
"'": 69,
"[PAD]": 96,
"[UNK]": 95,
"a": 29,
"b": 77,
"c": 78,
"d": 38,
"e": 3,
"f": 84,
"g": 9,
"h": 22,
"i": 51,
"j": 26,
"k": 19,
"l": 28,
"m": 46,
"n": 43,
"o": 48,
"p": 87,
"q": 65,
"r": 52,
"s": 83,
"t": 70,
"u": 42,
"v": 44,
"w": 1,
"x": 32,
"y": 92,
"z": 73,
"|": 49,
"à": 91,
"á": 0,
"â": 80,
"ã": 60,
"è": 89,
"é": 86,
"ê": 36,
"ì": 16,
"í": 64,
"ò": 61,
"ó": 8,
"ô": 54,
"õ": 56,
"ù": 53,
"ú": 7,
"ý": 31,
"ă": 66,
"đ": 24,
"ĩ": 81,
"ũ": 21,
"ơ": 85,
"ư": 17,
"ạ": 40,
"ả": 10,
"ấ": 39,
"ầ": 33,
"ẩ": 62,
"ẫ": 47,
"ậ": 55,
"ắ": 50,
"ằ": 75,
"ẳ": 82,
"ẵ": 23,
"ặ": 34,
"ẹ": 5,
"ẻ": 11,
"ẽ": 27,
"ế": 4,
"ề": 35,
"ể": 93,
"ễ": 76,
"ệ": 25,
"ỉ": 41,
"ị": 74,
"ọ": 13,
"ỏ": 18,
"ố": 37,
"ồ": 6,
"ổ": 12,
"ỗ": 59,
"ộ": 14,
"ớ": 79,
"ờ": 20,
"ở": 88,
"ỡ": 63,
"ợ": 68,
"ụ": 72,
"ủ": 71,
"ứ": 57,
"ừ": 2,
"ử": 67,
"ữ": 30,
"ự": 45,
"ỳ": 90,
"ỵ": 58,
"ỷ": 15,
"ỹ": 94
}