gweltou commited on
Commit
f046fbf
1 Parent(s): 74f0fbb

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +44 -44
vocab.json CHANGED
@@ -1,50 +1,50 @@
1
  {
2
- "'": 20,
3
- "3": 17,
4
  "[PAD]": 47,
5
  "[UNK]": 46,
6
- "a": 19,
7
- "b": 22,
8
- "c": 44,
9
- "d": 10,
10
- "e": 30,
11
- "f": 26,
12
- "g": 8,
13
- "h": 11,
14
- "i": 18,
15
  "j": 38,
16
- "k": 45,
17
- "l": 34,
18
- "m": 14,
19
- "n": 16,
20
- "o": 1,
21
- "p": 35,
22
- "q": 37,
23
- "r": 24,
24
- "s": 6,
25
- "t": 15,
26
- "u": 12,
27
- "v": 5,
28
- "w": 36,
29
- "x": 21,
30
- "y": 13,
31
- "z": 29,
32
- "|": 4,
33
- "à": 27,
34
- "á": 32,
35
- "â": 7,
36
- "ã": 28,
37
- "ç": 23,
38
  "è": 42,
39
- "é": 40,
40
- "ê": 0,
41
- "ë": 39,
42
- "ñ": 9,
43
- "ô": 25,
44
- "ö": 43,
45
- "ø": 3,
46
- "ù": 2,
47
- "û": 41,
48
- "ü": 33,
49
- "ı": 31
50
  }
 
1
  {
2
+ "'": 30,
3
+ "3": 34,
4
  "[PAD]": 47,
5
  "[UNK]": 46,
6
+ "a": 11,
7
+ "b": 32,
8
+ "c": 19,
9
+ "d": 22,
10
+ "e": 43,
11
+ "f": 17,
12
+ "g": 12,
13
+ "h": 10,
14
+ "i": 45,
15
  "j": 38,
16
+ "k": 4,
17
+ "l": 26,
18
+ "m": 6,
19
+ "n": 15,
20
+ "o": 20,
21
+ "p": 9,
22
+ "q": 7,
23
+ "r": 35,
24
+ "s": 21,
25
+ "t": 31,
26
+ "u": 24,
27
+ "v": 14,
28
+ "w": 1,
29
+ "x": 25,
30
+ "y": 27,
31
+ "z": 44,
32
+ "|": 2,
33
+ "à": 39,
34
+ "á": 0,
35
+ "â": 28,
36
+ "ã": 23,
37
+ "ç": 40,
38
  "è": 42,
39
+ "é": 8,
40
+ "ê": 5,
41
+ "ë": 16,
42
+ "ñ": 29,
43
+ "ô": 37,
44
+ "ö": 3,
45
+ "ø": 13,
46
+ "ù": 33,
47
+ "û": 18,
48
+ "ü": 41,
49
+ "ı": 36
50
  }