abbenedek commited on
Commit
13fd57b
1 Parent(s): 274bcc2

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -2
  2. tokenizer_config.json +4 -4
  3. vocab.json +23 -28
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 40,
3
- "<s>": 39
4
  }
 
1
  {
2
+ "</s>": 35,
3
+ "<s>": 34
4
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "37": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "38": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "39": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "40": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "32": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "33": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "34": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "35": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
vocab.json CHANGED
@@ -9,33 +9,28 @@
9
  "7": 8,
10
  "8": 9,
11
  "9": 10,
12
- "A": 11,
13
- "D": 12,
14
- "F": 13,
15
- "J": 14,
16
- "M": 15,
17
- "N": 16,
18
- "O": 17,
19
- "S": 18,
20
- "[PAD]": 38,
21
- "[UNK]": 37,
22
- "a": 19,
23
- "b": 20,
24
- "c": 21,
25
- "e": 22,
26
- "g": 23,
27
- "h": 24,
28
- "i": 25,
29
- "l": 26,
30
- "m": 27,
31
- "n": 28,
32
- "o": 29,
33
- "p": 30,
34
- "r": 31,
35
- "s": 32,
36
- "t": 33,
37
- "u": 34,
38
- "v": 35,
39
- "y": 36,
40
  "|": 0
41
  }
 
9
  "7": 8,
10
  "8": 9,
11
  "9": 10,
12
+ "[PAD]": 33,
13
+ "[UNK]": 32,
14
+ "a": 11,
15
+ "b": 12,
16
+ "c": 13,
17
+ "d": 14,
18
+ "e": 15,
19
+ "f": 16,
20
+ "g": 17,
21
+ "h": 18,
22
+ "i": 19,
23
+ "j": 20,
24
+ "l": 21,
25
+ "m": 22,
26
+ "n": 23,
27
+ "o": 24,
28
+ "p": 25,
29
+ "r": 26,
30
+ "s": 27,
31
+ "t": 28,
32
+ "u": 29,
33
+ "v": 30,
34
+ "y": 31,
 
 
 
 
 
35
  "|": 0
36
  }