arifagustyawan commited on
Commit
31e88c2
1 Parent(s): 66421ad

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +42 -42
vocab.json CHANGED
@@ -1,47 +1,47 @@
1
  {
2
- "&": 42,
3
- "'": 33,
4
- "(": 2,
5
- ")": 22,
6
- "[": 6,
7
  "[PAD]": 44,
8
  "[UNK]": 43,
9
- "\\": 32,
10
- "a": 16,
11
- "b": 41,
12
- "c": 4,
13
- "d": 24,
14
- "e": 31,
15
- "f": 40,
16
- "g": 23,
17
- "h": 34,
18
- "i": 20,
19
- "j": 12,
20
- "k": 29,
21
- "l": 7,
22
- "m": 25,
23
- "n": 36,
24
- "o": 37,
25
- "p": 39,
26
- "q": 17,
27
- "r": 38,
28
- "s": 19,
29
- "t": 14,
30
  "u": 3,
31
- "v": 5,
32
- "w": 13,
33
- "x": 15,
34
- "y": 10,
35
- "z": 26,
36
- "|": 9,
37
- "á": 0,
38
- "é": 35,
39
- "ł": 8,
40
- "ń": 30,
41
- "ō": 27,
42
- "–": 18,
43
- "—": 1,
44
- "’": 21,
45
- "!": 11,
46
- ",": 28
47
  }
 
1
  {
2
+ "&": 2,
3
+ "'": 30,
4
+ "(": 17,
5
+ ")": 9,
6
+ "[": 36,
7
  "[PAD]": 44,
8
  "[UNK]": 43,
9
+ "\\": 33,
10
+ "a": 26,
11
+ "b": 27,
12
+ "c": 31,
13
+ "d": 20,
14
+ "e": 11,
15
+ "f": 22,
16
+ "g": 37,
17
+ "h": 16,
18
+ "i": 41,
19
+ "j": 0,
20
+ "k": 38,
21
+ "l": 23,
22
+ "m": 40,
23
+ "n": 24,
24
+ "o": 4,
25
+ "p": 19,
26
+ "q": 10,
27
+ "r": 13,
28
+ "s": 15,
29
+ "t": 1,
30
  "u": 3,
31
+ "v": 34,
32
+ "w": 7,
33
+ "x": 5,
34
+ "y": 35,
35
+ "z": 14,
36
+ "|": 42,
37
+ "á": 12,
38
+ "é": 8,
39
+ "ł": 21,
40
+ "ń": 6,
41
+ "ō": 25,
42
+ "–": 29,
43
+ "—": 28,
44
+ "’": 32,
45
+ "!": 39,
46
+ ",": 18
47
  }