Sandrian commited on
Commit
e0bcebf
1 Parent(s): 13a94c6

Upload tokenizer

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