zainulhakim commited on
Commit
7c64316
·
verified ·
1 Parent(s): 9fb130e

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -2
  2. tokenizer_config.json +4 -4
  3. vocab.json +25 -19
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 21,
3
- "<s>": 20
4
  }
 
1
  {
2
+ "</s>": 27,
3
+ "<s>": 26
4
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "18": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "19": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "20": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "21": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "24": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "25": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "26": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "27": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
vocab.json CHANGED
@@ -1,22 +1,28 @@
1
  {
2
- "[PAD]": 19,
3
- "[UNK]": 18,
4
- "a": 7,
5
- "c": 13,
6
- "d": 6,
7
- "e": 8,
8
- "f": 4,
9
- "h": 10,
10
- "i": 12,
11
- "l": 0,
12
- "m": 16,
13
- "n": 17,
14
- "o": 14,
15
- "p": 15,
16
- "r": 11,
17
- "s": 1,
 
 
 
18
  "t": 5,
19
- "u": 9,
20
- "x": 2,
21
- "|": 3
 
 
 
22
  }
 
1
  {
2
+ "[PAD]": 25,
3
+ "[UNK]": 24,
4
+ "a": 4,
5
+ "b": 2,
6
+ "c": 3,
7
+ "d": 18,
8
+ "e": 9,
9
+ "f": 8,
10
+ "g": 1,
11
+ "h": 12,
12
+ "i": 20,
13
+ "k": 7,
14
+ "l": 19,
15
+ "m": 15,
16
+ "n": 6,
17
+ "o": 17,
18
+ "p": 14,
19
+ "r": 13,
20
+ "s": 10,
21
  "t": 5,
22
+ "u": 11,
23
+ "v": 21,
24
+ "w": 23,
25
+ "x": 0,
26
+ "y": 22,
27
+ "|": 16
28
  }