amansingh203 commited on
Commit
87f550e
1 Parent(s): 333427c

Upload tokenizer

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