mmtg commited on
Commit
d0c14d0
1 Parent(s): 6ad00eb

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +3 -2
  2. tokenizer_config.json +4 -4
  3. vocab.json +3 -4
added_tokens.json CHANGED
@@ -1,4 +1,5 @@
1
  {
2
- "</s>": 2866,
3
- "<s>": 2865
 
4
  }
 
1
  {
2
+ "</s>": 2865,
3
+ "<s>": 2864,
4
+ "[PAD]": 2863
5
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "2863": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "2864": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "2865": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "2866": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "2862": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "2863": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "2864": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "2865": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
vocab.json CHANGED
@@ -1,7 +1,6 @@
1
  {
2
- " ": 0,
3
- "[PAD]": 2864,
4
- "[UNK]": 2863,
5
  "a": 1,
6
  "b": 2,
7
  "c": 3,
@@ -21,7 +20,7 @@
21
  "t": 17,
22
  "u": 18,
23
  "y": 19,
24
- "|": 20,
25
  "à": 21,
26
  "á": 22,
27
  "è": 23,
 
1
  {
2
+ "[PAD]": 2863,
3
+ "[UNK]": 2862,
 
4
  "a": 1,
5
  "b": 2,
6
  "c": 3,
 
20
  "t": 17,
21
  "u": 18,
22
  "y": 19,
23
+ "|": 0,
24
  "à": 21,
25
  "á": 22,
26
  "è": 23,