colerobertson commited on
Commit
b3b171a
1 Parent(s): 0722d45

Upload tokenizer

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