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