mprzibilla commited on
Commit
0f000cc
1 Parent(s): 47e7c2d

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +25 -25
vocab.json CHANGED
@@ -1,30 +1,30 @@
1
  {
2
  "[PAD]": 27,
3
  "[UNK]": 26,
4
- "a": 8,
5
- "b": 1,
6
- "c": 13,
7
- "d": 9,
8
- "e": 6,
9
- "f": 23,
10
- "g": 12,
11
- "h": 7,
12
- "i": 24,
13
- "j": 5,
14
- "k": 14,
15
- "l": 22,
16
- "m": 19,
17
- "n": 25,
18
- "o": 15,
19
- "p": 4,
20
- "q": 20,
21
- "r": 21,
22
- "s": 2,
23
- "t": 18,
24
  "u": 16,
25
- "v": 10,
26
- "w": 3,
27
- "x": 11,
28
- "y": 0,
29
- "|": 17
30
  }
 
1
  {
2
  "[PAD]": 27,
3
  "[UNK]": 26,
4
+ "a": 0,
5
+ "b": 21,
6
+ "c": 7,
7
+ "d": 23,
8
+ "e": 24,
9
+ "f": 12,
10
+ "g": 17,
11
+ "h": 5,
12
+ "i": 20,
13
+ "j": 25,
14
+ "k": 19,
15
+ "l": 1,
16
+ "m": 15,
17
+ "n": 14,
18
+ "o": 18,
19
+ "p": 11,
20
+ "q": 2,
21
+ "r": 8,
22
+ "s": 3,
23
+ "t": 9,
24
  "u": 16,
25
+ "v": 13,
26
+ "w": 6,
27
+ "x": 22,
28
+ "y": 10,
29
+ "|": 4
30
  }