Sercan commited on
Commit
df080b4
·
1 Parent(s): b6eb50b

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +41 -41
vocab.json CHANGED
@@ -1,47 +1,47 @@
1
  {
2
- "0": 25,
3
- "1": 1,
4
- "2": 33,
5
- "3": 5,
6
- "4": 22,
7
- "5": 4,
8
- "6": 29,
9
- "7": 15,
10
- "8": 26,
11
- "9": 35,
12
  "[PAD]": 44,
13
  "[UNK]": 43,
14
- "a": 38,
15
- "b": 12,
16
- "c": 39,
17
- "d": 24,
18
  "e": 9,
19
- "f": 19,
20
- "g": 11,
21
- "h": 8,
22
- "i": 36,
23
- "j": 3,
24
- "k": 17,
25
- "l": 37,
26
- "m": 34,
27
- "n": 18,
28
- "o": 28,
29
- "p": 30,
30
- "q": 41,
31
  "r": 20,
32
- "s": 32,
33
- "t": 14,
34
- "u": 16,
35
- "v": 31,
36
- "w": 42,
37
- "x": 13,
38
- "y": 0,
39
- "z": 23,
40
- "|": 2,
41
- "ç": 40,
42
- "ö": 10,
43
- "ü": 7,
44
- "ğ": 6,
45
- "ı": 27,
46
- "ş": 21
47
  }
 
1
  {
2
+ "0": 41,
3
+ "1": 8,
4
+ "2": 6,
5
+ "3": 21,
6
+ "4": 28,
7
+ "5": 12,
8
+ "6": 2,
9
+ "7": 16,
10
+ "8": 25,
11
+ "9": 24,
12
  "[PAD]": 44,
13
  "[UNK]": 43,
14
+ "a": 22,
15
+ "b": 32,
16
+ "c": 13,
17
+ "d": 15,
18
  "e": 9,
19
+ "f": 29,
20
+ "g": 5,
21
+ "h": 33,
22
+ "i": 34,
23
+ "j": 37,
24
+ "k": 40,
25
+ "l": 23,
26
+ "m": 42,
27
+ "n": 30,
28
+ "o": 31,
29
+ "p": 11,
30
+ "q": 35,
31
  "r": 20,
32
+ "s": 17,
33
+ "t": 18,
34
+ "u": 10,
35
+ "v": 26,
36
+ "w": 38,
37
+ "x": 14,
38
+ "y": 19,
39
+ "z": 36,
40
+ "|": 1,
41
+ "ç": 27,
42
+ "ö": 39,
43
+ "ü": 0,
44
+ "ğ": 7,
45
+ "ı": 4,
46
+ "ş": 3
47
  }