finetune-wav2vec-vi / vocab.json
lmh2011's picture
Upload tokenizer
1e7932c verified
raw
history blame contribute delete
No virus
1.18 kB
{
"'": 20,
"[PAD]": 96,
"[UNK]": 95,
"a": 42,
"b": 43,
"c": 44,
"d": 74,
"e": 3,
"f": 32,
"g": 72,
"h": 14,
"i": 6,
"j": 82,
"k": 89,
"l": 41,
"m": 4,
"n": 7,
"o": 38,
"p": 67,
"q": 46,
"r": 2,
"s": 71,
"t": 69,
"u": 52,
"v": 57,
"w": 13,
"x": 65,
"y": 80,
"z": 21,
"|": 18,
"à": 66,
"á": 25,
"â": 63,
"ã": 12,
"è": 45,
"é": 26,
"ê": 50,
"ì": 78,
"í": 75,
"ò": 53,
"ó": 22,
"ô": 16,
"õ": 60,
"ù": 56,
"ú": 36,
"ý": 91,
"ă": 68,
"đ": 55,
"ĩ": 64,
"ũ": 0,
"ơ": 8,
"ư": 77,
"ạ": 88,
"ả": 35,
"ấ": 27,
"ầ": 40,
"ẩ": 81,
"ẫ": 19,
"ậ": 5,
"ắ": 37,
"ằ": 11,
"ẳ": 93,
"ẵ": 79,
"ặ": 87,
"ẹ": 9,
"ẻ": 15,
"ẽ": 17,
"ế": 70,
"ề": 39,
"ể": 10,
"ễ": 51,
"ệ": 34,
"ỉ": 83,
"ị": 30,
"ọ": 86,
"ỏ": 48,
"ố": 76,
"ồ": 94,
"ổ": 28,
"ỗ": 47,
"ộ": 59,
"ớ": 61,
"ờ": 49,
"ở": 85,
"ỡ": 31,
"ợ": 90,
"ụ": 54,
"ủ": 33,
"ứ": 23,
"ừ": 29,
"ử": 1,
"ữ": 92,
"ự": 84,
"ỳ": 24,
"ỵ": 73,
"ỷ": 62,
"ỹ": 58
}