Bluecast commited on
Commit
0c9c567
1 Parent(s): 0330878

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +24 -24
vocab.json CHANGED
@@ -1,29 +1,29 @@
1
  {
2
  "[PAD]": 26,
3
  "[UNK]": 25,
4
- "a": 23,
5
- "b": 17,
6
- "c": 22,
7
- "d": 0,
8
- "e": 21,
9
- "f": 24,
10
  "g": 14,
11
- "h": 11,
12
- "i": 5,
13
- "j": 18,
14
- "k": 9,
15
- "l": 8,
16
- "m": 12,
17
- "n": 16,
18
- "o": 13,
19
- "p": 3,
20
- "r": 7,
21
- "s": 6,
22
- "t": 20,
23
- "u": 15,
24
- "v": 19,
25
- "w": 1,
26
- "y": 2,
27
- "z": 4,
28
- "|": 10
29
  }
 
1
  {
2
  "[PAD]": 26,
3
  "[UNK]": 25,
4
+ "a": 7,
5
+ "b": 11,
6
+ "c": 17,
7
+ "d": 12,
8
+ "e": 5,
9
+ "f": 13,
10
  "g": 14,
11
+ "h": 24,
12
+ "i": 15,
13
+ "j": 1,
14
+ "k": 4,
15
+ "l": 16,
16
+ "m": 19,
17
+ "n": 21,
18
+ "o": 23,
19
+ "p": 22,
20
+ "r": 9,
21
+ "s": 2,
22
+ "t": 18,
23
+ "u": 3,
24
+ "v": 6,
25
+ "w": 10,
26
+ "y": 8,
27
+ "z": 0,
28
+ "|": 20
29
  }