Jethuestad commited on
Commit
318362d
·
1 Parent(s): 44d3808

add tokenizer

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