Reihaneh commited on
Commit
385e4db
·
verified ·
1 Parent(s): 5643443

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +51 -51
vocab.json CHANGED
@@ -1,60 +1,60 @@
1
  {
2
  "'": 3,
3
- "-": 41,
4
- "F": 44,
5
- "L": 4,
6
- "N": 25,
7
- "Y": 35,
8
- "[": 47,
9
  "[FY-NL]": 56,
10
  "[NL]": 57,
11
  "[PAD]": 55,
12
  "[UNK]": 54,
13
- "]": 46,
14
- "a": 20,
15
- "b": 12,
16
- "c": 38,
17
- "d": 45,
18
- "e": 18,
19
- "f": 50,
20
- "g": 13,
21
- "h": 48,
22
- "i": 11,
23
- "j": 37,
24
- "k": 14,
25
  "l": 6,
26
- "m": 40,
27
- "n": 32,
28
  "o": 39,
29
- "p": 19,
30
- "q": 53,
31
- "r": 33,
32
- "s": 15,
33
- "t": 2,
34
- "u": 52,
35
- "v": 26,
36
- "w": 0,
37
- "x": 28,
38
- "y": 42,
39
- "z": 1,
40
- "|": 27,
41
- "à": 8,
42
- "á": 49,
43
- "â": 24,
44
- "ä": 34,
45
- "è": 22,
46
- "é": 17,
47
- "ê": 9,
48
- "ë": 10,
49
- "ï": 7,
50
- "ó": 29,
51
- "ô": 36,
52
- "ö": 23,
53
- "ú": 43,
54
- "û": 5,
55
- "ü": 21,
56
- "–": 16,
57
- "’": 51,
58
- "“": 30,
59
- "”": 31
60
  }
 
1
  {
2
  "'": 3,
3
+ "-": 43,
4
+ "F": 50,
5
+ "L": 19,
6
+ "N": 11,
7
+ "Y": 23,
8
+ "[": 24,
9
  "[FY-NL]": 56,
10
  "[NL]": 57,
11
  "[PAD]": 55,
12
  "[UNK]": 54,
13
+ "]": 12,
14
+ "a": 0,
15
+ "b": 48,
16
+ "c": 53,
17
+ "d": 34,
18
+ "e": 27,
19
+ "f": 46,
20
+ "g": 2,
21
+ "h": 22,
22
+ "i": 31,
23
+ "j": 26,
24
+ "k": 7,
25
  "l": 6,
26
+ "m": 44,
27
+ "n": 5,
28
  "o": 39,
29
+ "p": 51,
30
+ "q": 18,
31
+ "r": 28,
32
+ "s": 35,
33
+ "t": 42,
34
+ "u": 32,
35
+ "v": 8,
36
+ "w": 9,
37
+ "x": 33,
38
+ "y": 30,
39
+ "z": 49,
40
+ "|": 13,
41
+ "à": 4,
42
+ "á": 14,
43
+ "â": 36,
44
+ "ä": 17,
45
+ "è": 15,
46
+ "é": 47,
47
+ "ê": 38,
48
+ "ë": 37,
49
+ "ï": 41,
50
+ "ó": 21,
51
+ "ô": 1,
52
+ "ö": 25,
53
+ "ú": 20,
54
+ "û": 29,
55
+ "ü": 40,
56
+ "–": 52,
57
+ "’": 16,
58
+ "“": 10,
59
+ "”": 45
60
  }