foxxy-hm commited on
Commit
7c0224a
·
1 Parent(s): 8735238

Upload processor

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