Reihaneh commited on
Commit
080293f
1 Parent(s): 9e1c0f1

Upload tokenizer

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