Bluecast commited on
Commit
3848893
1 Parent(s): a5fd1a6

Upload tokenizer

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