davidilag commited on
Commit
32cccf0
1 Parent(s): 6a63056

Upload tokenizer

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