Upload vocab.json
Browse files- vocab.json +130 -0
vocab.json
ADDED
@@ -0,0 +1,130 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"|": 0,
|
3 |
+
"a": 1,
|
4 |
+
"n": 2,
|
5 |
+
"i": 3,
|
6 |
+
"e": 4,
|
7 |
+
"t": 5,
|
8 |
+
"r": 6,
|
9 |
+
"s": 7,
|
10 |
+
"u": 8,
|
11 |
+
"k": 9,
|
12 |
+
"m": 10,
|
13 |
+
"d": 11,
|
14 |
+
"g": 12,
|
15 |
+
"l": 13,
|
16 |
+
"p": 14,
|
17 |
+
"o": 15,
|
18 |
+
"b": 16,
|
19 |
+
"h": 17,
|
20 |
+
"y": 18,
|
21 |
+
"j": 19,
|
22 |
+
"c": 20,
|
23 |
+
"w": 21,
|
24 |
+
"f": 22,
|
25 |
+
"v": 23,
|
26 |
+
"z": 24,
|
27 |
+
"x": 25,
|
28 |
+
"q": 26,
|
29 |
+
"an": 27,
|
30 |
+
"er": 28,
|
31 |
+
"en": 29,
|
32 |
+
"ar": 30,
|
33 |
+
"in": 31,
|
34 |
+
"at": 32,
|
35 |
+
"ang": 33,
|
36 |
+
"ak": 34,
|
37 |
+
"al": 35,
|
38 |
+
"as": 36,
|
39 |
+
"ah": 37,
|
40 |
+
"un": 38,
|
41 |
+
"am": 39,
|
42 |
+
"em": 40,
|
43 |
+
"di": 41,
|
44 |
+
"el": 42,
|
45 |
+
"ad": 43,
|
46 |
+
"ya": 44,
|
47 |
+
"eng": 45,
|
48 |
+
"is": 46,
|
49 |
+
"ik": 47,
|
50 |
+
"it": 48,
|
51 |
+
"uk": 49,
|
52 |
+
"ap": 50,
|
53 |
+
"es": 51,
|
54 |
+
"on": 52,
|
55 |
+
"se": 53,
|
56 |
+
"yang": 54,
|
57 |
+
"or": 55,
|
58 |
+
"us": 56,
|
59 |
+
"il": 57,
|
60 |
+
"dan": 58,
|
61 |
+
"men": 59,
|
62 |
+
"ut": 60,
|
63 |
+
"ur": 61,
|
64 |
+
"ga": 62,
|
65 |
+
"ke": 63,
|
66 |
+
"kan": 64,
|
67 |
+
"ing": 65,
|
68 |
+
"ber": 66,
|
69 |
+
"ter": 67,
|
70 |
+
"um": 68,
|
71 |
+
"per": 69,
|
72 |
+
"id": 70,
|
73 |
+
"asi": 71,
|
74 |
+
"ol": 72,
|
75 |
+
"ul": 73,
|
76 |
+
"akan": 74,
|
77 |
+
"ari": 75,
|
78 |
+
"et": 76,
|
79 |
+
"ek": 77,
|
80 |
+
"emb": 78,
|
81 |
+
"nya": 79,
|
82 |
+
"ir": 80,
|
83 |
+
"ung": 81,
|
84 |
+
"eb": 82,
|
85 |
+
"im": 83,
|
86 |
+
"om": 84,
|
87 |
+
"ada": 85,
|
88 |
+
"ag": 86,
|
89 |
+
"meng": 87,
|
90 |
+
"ab": 88,
|
91 |
+
"tuk": 89,
|
92 |
+
"ia": 90,
|
93 |
+
"ara": 91,
|
94 |
+
"ot": 92,
|
95 |
+
"engan": 93,
|
96 |
+
"and": 94,
|
97 |
+
"ih": 95,
|
98 |
+
"dengan": 96,
|
99 |
+
"up": 97,
|
100 |
+
"pen": 98,
|
101 |
+
"ti": 99,
|
102 |
+
"ini": 100,
|
103 |
+
"ud": 101,
|
104 |
+
"si": 102,
|
105 |
+
"untuk": 103,
|
106 |
+
"ed": 104,
|
107 |
+
"pr": 105,
|
108 |
+
"alam": 106,
|
109 |
+
"ai": 107,
|
110 |
+
"ata": 108,
|
111 |
+
"ant": 109,
|
112 |
+
"ok": 110,
|
113 |
+
"mem": 111,
|
114 |
+
"dari": 112,
|
115 |
+
"uh": 113,
|
116 |
+
"aj": 114,
|
117 |
+
"aw": 115,
|
118 |
+
"ent": 116,
|
119 |
+
"os": 117,
|
120 |
+
"ay": 118,
|
121 |
+
"ikan": 119,
|
122 |
+
"atan": 120,
|
123 |
+
"ian": 121,
|
124 |
+
"itu": 122,
|
125 |
+
"ama": 123,
|
126 |
+
"seb": 124,
|
127 |
+
"amp": 125,
|
128 |
+
"[UNK]": 126,
|
129 |
+
"[PAD]": 127
|
130 |
+
}
|