dmusingu commited on
Commit
f74ff30
·
verified ·
1 Parent(s): b749503

Upload tokenizer

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