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