srujan00123 commited on
Commit
78b5fe8
1 Parent(s): 4c7d801

Upload tokenizer

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