auditi41 commited on
Commit
08d2c4e
1 Parent(s): ceab119

Upload tokenizer

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