mgtoxd commited on
Commit
7715976
·
1 Parent(s): 390332c

add tokenizer

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