procit008 commited on
Commit
2a2d918
·
verified ·
1 Parent(s): c9eb77c

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +43 -43
vocab.json CHANGED
@@ -1,49 +1,49 @@
1
  {
2
- "&": 29,
3
- "'": 33,
4
  "+": 40,
5
- "2": 6,
6
- "3": 10,
7
- "4": 44,
8
- "@": 9,
9
  "[PAD]": 46,
10
  "[UNK]": 45,
11
- "a": 7,
12
- "b": 24,
13
- "c": 8,
14
- "d": 0,
15
- "e": 1,
16
- "f": 20,
17
- "g": 30,
18
- "h": 31,
19
- "i": 34,
20
- "j": 37,
21
- "k": 15,
22
- "l": 42,
23
- "m": 13,
24
- "n": 14,
25
- "o": 18,
26
- "p": 36,
27
- "q": 32,
28
- "r": 11,
29
- "s": 25,
30
- "t": 16,
31
- "u": 43,
32
- "v": 23,
33
- "w": 12,
34
- "x": 38,
35
  "y": 41,
36
- "z": 4,
37
- "{": 21,
38
- "|": 27,
39
- "}": 5,
40
- "è": 26,
41
- "é": 35,
42
- "ë": 3,
43
- "ï": 28,
44
- "ó": 22,
45
- "ö": 2,
46
- "ü": 17,
47
- "ij": 39,
48
- "…": 19
49
  }
 
1
  {
2
+ "&": 37,
3
+ "'": 35,
4
  "+": 40,
5
+ "2": 5,
6
+ "3": 18,
7
+ "4": 13,
8
+ "@": 3,
9
  "[PAD]": 46,
10
  "[UNK]": 45,
11
+ "a": 22,
12
+ "b": 29,
13
+ "c": 44,
14
+ "d": 16,
15
+ "e": 12,
16
+ "f": 26,
17
+ "g": 0,
18
+ "h": 17,
19
+ "i": 36,
20
+ "j": 31,
21
+ "k": 8,
22
+ "l": 2,
23
+ "m": 39,
24
+ "n": 27,
25
+ "o": 10,
26
+ "p": 15,
27
+ "q": 38,
28
+ "r": 6,
29
+ "s": 43,
30
+ "t": 28,
31
+ "u": 34,
32
+ "v": 14,
33
+ "w": 21,
34
+ "x": 7,
35
  "y": 41,
36
+ "z": 33,
37
+ "{": 23,
38
+ "|": 20,
39
+ "}": 19,
40
+ "è": 32,
41
+ "é": 4,
42
+ "ë": 30,
43
+ "ï": 25,
44
+ "ó": 42,
45
+ "ö": 9,
46
+ "ü": 24,
47
+ "ij": 1,
48
+ "…": 11
49
  }