lparisi commited on
Commit
6e77e33
1 Parent(s): 4641714

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +43 -41
vocab.json CHANGED
@@ -1,44 +1,46 @@
1
  {
2
- "'": 1,
 
 
3
  "a": 34,
4
- "b": 18,
5
- "c": 16,
6
- "d": 27,
7
- "e": 10,
8
- "f": 0,
9
- "g": 22,
10
- "h": 6,
11
- "i": 9,
12
- "j": 19,
13
- "k": 36,
14
- "l": 31,
15
- "m": 15,
16
- "n": 41,
17
- "o": 4,
18
- "p": 8,
19
- "q": 35,
20
- "r": 38,
21
- "s": 11,
22
- "t": 39,
23
- "u": 26,
24
- "v": 32,
25
- "w": 30,
26
- "x": 25,
27
- "y": 3,
28
- "z": 23,
29
- "|": 40,
30
- "à": 29,
31
- "á": 21,
32
- "â": 2,
33
- "ã": 13,
34
- "ç": 14,
35
- "é": 20,
36
- "ê": 28,
37
- "í": 12,
38
- "ñ": 37,
39
- "ó": 7,
40
- "ô": 33,
41
- "õ": 24,
42
- "ú": 5,
43
- "ü": 17
44
  }
 
1
  {
2
+ "'": 28,
3
+ "[PAD]": 43,
4
+ "[UNK]": 42,
5
  "a": 34,
6
+ "b": 38,
7
+ "c": 19,
8
+ "d": 39,
9
+ "e": 25,
10
+ "f": 6,
11
+ "g": 12,
12
+ "h": 37,
13
+ "i": 3,
14
+ "j": 29,
15
+ "k": 10,
16
+ "l": 1,
17
+ "m": 13,
18
+ "n": 2,
19
+ "o": 41,
20
+ "p": 22,
21
+ "q": 32,
22
+ "r": 4,
23
+ "s": 16,
24
+ "t": 30,
25
+ "u": 11,
26
+ "v": 18,
27
+ "w": 15,
28
+ "x": 31,
29
+ "y": 8,
30
+ "z": 20,
31
+ "|": 27,
32
+ "à": 36,
33
+ "á": 23,
34
+ "â": 33,
35
+ "ã": 40,
36
+ "ç": 5,
37
+ "é": 0,
38
+ "ê": 7,
39
+ "í": 26,
40
+ "ñ": 9,
41
+ "ó": 24,
42
+ "ô": 17,
43
+ "õ": 21,
44
+ "ú": 35,
45
+ "ü": 14
46
  }