Reihaneh commited on
Commit
ca78015
1 Parent(s): 7680a0c

Upload tokenizer

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