AdaCodruta commited on
Commit
8180ebf
·
verified ·
1 Parent(s): be2621a

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +40 -40
vocab.json CHANGED
@@ -1,45 +1,45 @@
1
  {
2
- " ": 4,
3
  "[PAD]": 42,
4
  "[UNK]": 41,
5
- "a": 13,
6
- "b": 16,
7
- "c": 24,
8
- "d": 22,
9
- "e": 36,
10
- "f": 20,
11
- "g": 18,
12
- "h": 8,
13
- "i": 31,
14
- "j": 34,
15
- "k": 3,
16
- "l": 5,
17
- "m": 38,
18
- "n": 25,
19
- "o": 21,
20
- "p": 7,
21
- "q": 32,
22
- "r": 19,
23
- "s": 17,
24
- "t": 37,
25
- "u": 1,
26
- "v": 39,
27
- "w": 15,
28
- "x": 2,
29
- "y": 33,
30
- "z": 0,
31
- "á": 23,
32
- "â": 28,
33
- "é": 40,
34
- "î": 10,
35
- "ü": 26,
36
  "ă": 35,
37
- "ć": 29,
38
- "đ": 12,
39
- "ş": 27,
40
- "š": 30,
41
- "ţ": 11,
42
- "ș": 9,
43
- "ț": 14,
44
- "”": 6
45
  }
 
1
  {
2
+ " ": 12,
3
  "[PAD]": 42,
4
  "[UNK]": 41,
5
+ "a": 3,
6
+ "b": 11,
7
+ "c": 2,
8
+ "d": 21,
9
+ "e": 26,
10
+ "f": 24,
11
+ "g": 28,
12
+ "h": 31,
13
+ "i": 9,
14
+ "j": 8,
15
+ "k": 20,
16
+ "l": 32,
17
+ "m": 16,
18
+ "n": 18,
19
+ "o": 30,
20
+ "p": 23,
21
+ "q": 27,
22
+ "r": 38,
23
+ "s": 33,
24
+ "t": 0,
25
+ "u": 36,
26
+ "v": 4,
27
+ "w": 13,
28
+ "x": 25,
29
+ "y": 39,
30
+ "z": 29,
31
+ "á": 22,
32
+ "â": 10,
33
+ "é": 17,
34
+ "î": 14,
35
+ "ü": 37,
36
  "ă": 35,
37
+ "ć": 6,
38
+ "đ": 19,
39
+ "ş": 7,
40
+ "š": 5,
41
+ "ţ": 1,
42
+ "ș": 34,
43
+ "ț": 15,
44
+ "”": 40
45
  }