Upload vocab.json with huggingface_hub
Browse files- vocab.json +1 -112
vocab.json
CHANGED
@@ -1,112 +1 @@
|
|
1 |
-
{
|
2 |
-
"%": 103,
|
3 |
-
"0": 98,
|
4 |
-
"1": 8,
|
5 |
-
"2": 33,
|
6 |
-
"3": 74,
|
7 |
-
"4": 104,
|
8 |
-
"5": 3,
|
9 |
-
"6": 11,
|
10 |
-
"7": 62,
|
11 |
-
"8": 94,
|
12 |
-
"9": 80,
|
13 |
-
"[PAD]": 109,
|
14 |
-
"[UNK]": 108,
|
15 |
-
"a": 47,
|
16 |
-
"b": 102,
|
17 |
-
"c": 84,
|
18 |
-
"d": 38,
|
19 |
-
"e": 44,
|
20 |
-
"f": 55,
|
21 |
-
"g": 65,
|
22 |
-
"h": 68,
|
23 |
-
"i": 24,
|
24 |
-
"k": 6,
|
25 |
-
"l": 1,
|
26 |
-
"m": 88,
|
27 |
-
"n": 12,
|
28 |
-
"o": 77,
|
29 |
-
"p": 105,
|
30 |
-
"q": 35,
|
31 |
-
"r": 76,
|
32 |
-
"s": 79,
|
33 |
-
"t": 9,
|
34 |
-
"u": 64,
|
35 |
-
"v": 85,
|
36 |
-
"w": 22,
|
37 |
-
"x": 31,
|
38 |
-
"y": 37,
|
39 |
-
"z": 107,
|
40 |
-
"|": 48,
|
41 |
-
"à": 82,
|
42 |
-
"á": 54,
|
43 |
-
"â": 100,
|
44 |
-
"ã": 52,
|
45 |
-
"è": 30,
|
46 |
-
"é": 4,
|
47 |
-
"ê": 66,
|
48 |
-
"ì": 7,
|
49 |
-
"í": 13,
|
50 |
-
"ò": 32,
|
51 |
-
"ó": 49,
|
52 |
-
"ô": 40,
|
53 |
-
"õ": 59,
|
54 |
-
"ù": 63,
|
55 |
-
"ú": 18,
|
56 |
-
"ý": 90,
|
57 |
-
"ă": 34,
|
58 |
-
"đ": 39,
|
59 |
-
"ĩ": 42,
|
60 |
-
"ũ": 56,
|
61 |
-
"ơ": 43,
|
62 |
-
"ư": 69,
|
63 |
-
"̀": 106,
|
64 |
-
"́": 16,
|
65 |
-
"̣": 60,
|
66 |
-
"у": 14,
|
67 |
-
"ạ": 93,
|
68 |
-
"ả": 19,
|
69 |
-
"ấ": 91,
|
70 |
-
"ầ": 96,
|
71 |
-
"ẩ": 25,
|
72 |
-
"ẫ": 2,
|
73 |
-
"ậ": 23,
|
74 |
-
"ắ": 57,
|
75 |
-
"ằ": 28,
|
76 |
-
"ẳ": 81,
|
77 |
-
"ẵ": 71,
|
78 |
-
"ặ": 78,
|
79 |
-
"ẹ": 92,
|
80 |
-
"ẻ": 70,
|
81 |
-
"ẽ": 41,
|
82 |
-
"ế": 0,
|
83 |
-
"ề": 36,
|
84 |
-
"ể": 89,
|
85 |
-
"ễ": 45,
|
86 |
-
"ệ": 73,
|
87 |
-
"ỉ": 67,
|
88 |
-
"ị": 21,
|
89 |
-
"ọ": 27,
|
90 |
-
"ỏ": 29,
|
91 |
-
"ố": 51,
|
92 |
-
"ồ": 101,
|
93 |
-
"ổ": 15,
|
94 |
-
"ỗ": 10,
|
95 |
-
"ộ": 87,
|
96 |
-
"ớ": 72,
|
97 |
-
"ờ": 26,
|
98 |
-
"ở": 75,
|
99 |
-
"ỡ": 50,
|
100 |
-
"ợ": 20,
|
101 |
-
"ụ": 86,
|
102 |
-
"ủ": 61,
|
103 |
-
"ứ": 83,
|
104 |
-
"ừ": 53,
|
105 |
-
"ử": 17,
|
106 |
-
"ữ": 46,
|
107 |
-
"ự": 99,
|
108 |
-
"ỳ": 58,
|
109 |
-
"ỵ": 5,
|
110 |
-
"ỷ": 97,
|
111 |
-
"ỹ": 95
|
112 |
-
}
|
|
|
1 |
+
{"\u00f3": 0, "\u1edf": 1, "\u1ea1": 2, "\u1eed": 3, "\u1ea5": 4, "\u1ee5": 5, "0": 6, "6": 7, "\u1eb3": 8, "\u1ecd": 9, "\u00f9": 10, "\u1ed7": 11, "\u1ed5": 12, "q": 13, "\u00e1": 14, "1": 15, "7": 16, "\u1ee3": 17, "\u1eef": 18, "\u1ed9": 19, "\u1ec3": 20, "i": 21, "s": 22, "u": 23, "\u0129": 24, "\u00e8": 25, "\u1ebf": 26, "\u1ee1": 27, "\u00ea": 28, "z": 29, "x": 30, "\u00e0": 31, "\u0111": 32, "3": 33, "o": 34, "m": 35, "e": 36, "\u1edd": 37, "\u00f4": 38, "\u01a1": 39, "\u1ec7": 40, "4": 41, "\u1ec1": 42, "\u00fa": 43, "9": 44, "w": 45, "\u0323": 46, "\u1ed1": 47, "\u0301": 48, "f": 49, "\u1edb": 50, "\u00ed": 52, "\u1ea9": 53, "a": 54, "\u1ef7": 55, "n": 56, "l": 57, "\u1ec9": 58, "%": 59, "r": 60, "\u1eeb": 61, "\u1ecb": 62, "\u1eb1": 63, "g": 64, "\u1ec5": 65, "\u1ebb": 66, "h": 67, "\u00f2": 68, "\u1ef5": 69, "\u1eb7": 70, "\u00e9": 71, "\u1ef1": 72, "b": 73, "\u1ee9": 74, "t": 75, "k": 76, "\u00fd": 77, "\u1ef3": 78, "p": 79, "\u0300": 80, "\u1ee7": 81, "\u1ed3": 82, "\u00ec": 83, "\u00e2": 84, "\u1ef9": 85, "\u1eb5": 86, "y": 87, "8": 88, "\u1ebd": 89, "\u0103": 90, "c": 91, "\u1ecf": 92, "\u1ead": 93, "v": 94, "\u1eb9": 95, "\u00f5": 96, "d": 97, "\u1ea7": 98, "\u01b0": 99, "5": 100, "\u1eaf": 101, "\u1eab": 102, "\u1ea3": 103, "\u00e3": 104, "2": 105, "\u0169": 106, "\u0443": 107, "|": 51, "[UNK]": 108, "[PAD]": 109}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|