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