srujan00123 commited on
Commit
9ae3c95
·
1 Parent(s): fae4a40

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +4 -4
  2. tokenizer_config.json +4 -4
  3. vocab.json +44 -45
added_tokens.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
- "</s>": 48,
3
- "<s>": 47,
4
- "[PAD]": 46,
5
- "[UNK]": 45
6
  }
 
1
  {
2
+ "</s>": 47,
3
+ "<s>": 46,
4
+ "[PAD]": 45,
5
+ "[UNK]": 44
6
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "45": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "46": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "47": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "48": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "44": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "45": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "46": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "47": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
vocab.json CHANGED
@@ -1,49 +1,48 @@
1
  {
2
- "%": 43,
3
- "'": 27,
4
- "(": 35,
5
- ")": 19,
6
- "+": 1,
7
- "0": 17,
8
- "1": 2,
9
- "2": 4,
10
- "3": 36,
11
- "4": 18,
12
- "5": 39,
13
- "6": 24,
14
- "7": 7,
15
- "8": 25,
16
  "9": 11,
17
- "[": 40,
18
- "[PAD]": 46,
19
- "[UNK]": 45,
20
- "\\": 38,
21
- "]": 22,
22
- "a": 26,
23
- "b": 23,
24
- "c": 29,
25
  "d": 21,
26
- "e": 41,
27
- "f": 31,
28
- "g": 0,
29
- "h": 28,
30
- "i": 14,
31
- "j": 9,
32
- "k": 42,
33
- "l": 15,
34
- "m": 6,
35
- "n": 32,
36
- "o": 10,
37
- "p": 8,
38
- "q": 13,
39
- "r": 20,
40
- "s": 12,
41
- "t": 44,
42
- "u": 33,
43
- "v": 3,
44
- "w": 16,
45
- "x": 5,
46
- "y": 37,
47
- "z": 34,
48
- "|": 30
49
  }
 
1
  {
2
+ "%": 39,
3
+ "'": 17,
4
+ "(": 32,
5
+ ")": 1,
6
+ "+": 9,
7
+ "0": 42,
8
+ "1": 34,
9
+ "2": 30,
10
+ "3": 14,
11
+ "4": 41,
12
+ "5": 27,
13
+ "6": 20,
14
+ "7": 25,
15
+ "8": 23,
16
  "9": 11,
17
+ "[": 7,
18
+ "[PAD]": 45,
19
+ "[UNK]": 44,
20
+ "]": 0,
21
+ "a": 38,
22
+ "b": 18,
23
+ "c": 37,
 
24
  "d": 21,
25
+ "e": 10,
26
+ "f": 28,
27
+ "g": 22,
28
+ "h": 6,
29
+ "i": 36,
30
+ "j": 2,
31
+ "k": 35,
32
+ "l": 8,
33
+ "m": 26,
34
+ "n": 16,
35
+ "o": 19,
36
+ "p": 29,
37
+ "q": 4,
38
+ "r": 33,
39
+ "s": 13,
40
+ "t": 15,
41
+ "u": 12,
42
+ "v": 43,
43
+ "w": 3,
44
+ "x": 40,
45
+ "y": 5,
46
+ "z": 31,
47
+ "|": 24
48
  }