Reihaneh commited on
Commit
7680a0c
1 Parent(s): 361f74f

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +62 -62
vocab.json CHANGED
@@ -1,72 +1,72 @@
1
  {
2
  "fry": {
3
- "'": 40,
4
- "(": 5,
5
- ")": 28,
6
- "-": 13,
7
- "F": 56,
8
- "L": 53,
9
- "N": 19,
10
- "Y": 20,
11
  "[": 36,
12
  "[EN]": 65,
13
  "[FY-NL]": 67,
14
  "[NL]": 66,
15
  "[PAD]": 64,
16
  "[UNK]": 63,
17
- "]": 50,
18
- "a": 26,
19
- "b": 2,
20
- "c": 62,
21
- "d": 17,
22
- "e": 55,
23
- "f": 61,
24
- "g": 0,
25
- "h": 33,
26
- "i": 14,
27
- "j": 12,
28
- "k": 51,
29
- "l": 59,
30
- "m": 10,
31
- "n": 3,
32
- "o": 45,
33
- "p": 15,
34
- "q": 9,
35
- "r": 25,
36
- "s": 22,
37
- "t": 35,
38
- "u": 11,
39
- "v": 21,
40
- "w": 27,
41
- "x": 24,
42
- "y": 49,
43
- "z": 39,
44
- "|": 7,
45
- "´": 46,
46
- "à": 4,
47
- "á": 42,
48
- "â": 58,
49
- "ä": 47,
50
- "è": 6,
51
- "é": 29,
52
- "ê": 34,
53
- "ë": 54,
54
- "í": 8,
55
- "ï": 41,
56
- "ó": 52,
57
- "ô": 23,
58
- "ö": 32,
59
- "ú": 60,
60
- "û": 18,
61
- "ü": 37,
62
- "ă": 43,
63
- "š": 1,
64
- "ţ": 48,
65
- "–": 57,
66
- "‘": 44,
67
- "’": 38,
68
- "“": 16,
69
- "”": 31,
70
- "…": 30
71
  }
72
  }
 
1
  {
2
  "fry": {
3
+ "'": 37,
4
+ "(": 47,
5
+ ")": 11,
6
+ "-": 31,
7
+ "F": 45,
8
+ "L": 60,
9
+ "N": 49,
10
+ "Y": 9,
11
  "[": 36,
12
  "[EN]": 65,
13
  "[FY-NL]": 67,
14
  "[NL]": 66,
15
  "[PAD]": 64,
16
  "[UNK]": 63,
17
+ "]": 38,
18
+ "a": 8,
19
+ "b": 50,
20
+ "c": 44,
21
+ "d": 35,
22
+ "e": 2,
23
+ "f": 33,
24
+ "g": 28,
25
+ "h": 27,
26
+ "i": 53,
27
+ "j": 43,
28
+ "k": 52,
29
+ "l": 26,
30
+ "m": 4,
31
+ "n": 58,
32
+ "o": 18,
33
+ "p": 3,
34
+ "q": 20,
35
+ "r": 0,
36
+ "s": 17,
37
+ "t": 13,
38
+ "u": 30,
39
+ "v": 32,
40
+ "w": 55,
41
+ "x": 19,
42
+ "y": 41,
43
+ "z": 29,
44
+ "|": 54,
45
+ "´": 10,
46
+ "à": 22,
47
+ "á": 5,
48
+ "â": 25,
49
+ "ä": 15,
50
+ "è": 42,
51
+ "é": 46,
52
+ "ê": 6,
53
+ "ë": 62,
54
+ "í": 7,
55
+ "ï": 57,
56
+ "ó": 1,
57
+ "ô": 48,
58
+ "ö": 56,
59
+ "ú": 39,
60
+ "û": 21,
61
+ "ü": 23,
62
+ "ă": 40,
63
+ "š": 61,
64
+ "ţ": 51,
65
+ "–": 59,
66
+ "‘": 16,
67
+ "’": 14,
68
+ "“": 34,
69
+ "”": 12,
70
+ "…": 24
71
  }
72
  }