patrickvonplaten commited on
Commit
d4c696d
·
1 Parent(s): fdd002c

Upload tokenizer

Browse files
Files changed (2) hide show
  1. tokenizer_config.json +1 -1
  2. vocab.json +53 -30
tokenizer_config.json CHANGED
@@ -2,7 +2,7 @@
2
  "bos_token": "<s>",
3
  "do_lower_case": false,
4
  "eos_token": "</s>",
5
- "name_or_path": "/tmp/tmpyo_gcvcv",
6
  "pad_token": "<pad>",
7
  "replace_word_delimiter_char": " ",
8
  "special_tokens_map_file": null,
 
2
  "bos_token": "<s>",
3
  "do_lower_case": false,
4
  "eos_token": "</s>",
5
+ "name_or_path": "/tmp/tmpz4ppa0vq",
6
  "pad_token": "<pad>",
7
  "replace_word_delimiter_char": " ",
8
  "special_tokens_map_file": null,
vocab.json CHANGED
@@ -1,38 +1,61 @@
1
  {
2
- "'": 26,
3
- ",": 17,
4
- "-": 33,
5
- ".": 11,
6
  "</s>": 2,
7
  "<pad>": 0,
8
  "<s>": 1,
9
  "<unk>": 3,
10
- "?": 13,
11
- "a": 29,
12
- "b": 5,
13
- "c": 35,
14
- "d": 34,
15
- "e": 25,
16
- "f": 21,
17
- "g": 20,
18
- "h": 28,
19
- "i": 4,
20
- "j": 14,
21
- "k": 6,
22
- "l": 8,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
23
  "m": 10,
24
- "n": 16,
25
- "o": 32,
26
- "p": 18,
27
- "q": 19,
28
- "r": 12,
29
- "s": 27,
30
  "t": 24,
31
- "u": 22,
32
- "v": 7,
33
- "w": 9,
34
- "x": 23,
35
- "y": 15,
36
- "z": 31,
37
- "|": 30
38
  }
 
1
  {
2
+ "'": 33,
3
+ ",": 21,
4
+ "-": 37,
5
+ ".": 23,
6
  "</s>": 2,
7
  "<pad>": 0,
8
  "<s>": 1,
9
  "<unk>": 3,
10
+ "?": 17,
11
+ "A": 8,
12
+ "B": 18,
13
+ "C": 54,
14
+ "D": 20,
15
+ "E": 41,
16
+ "F": 58,
17
+ "G": 40,
18
+ "H": 14,
19
+ "I": 30,
20
+ "J": 7,
21
+ "K": 4,
22
+ "L": 47,
23
+ "M": 19,
24
+ "N": 5,
25
+ "O": 56,
26
+ "P": 13,
27
+ "R": 55,
28
+ "S": 43,
29
+ "T": 11,
30
+ "U": 48,
31
+ "V": 38,
32
+ "W": 36,
33
+ "Y": 35,
34
+ "a": 27,
35
+ "b": 26,
36
+ "c": 57,
37
+ "d": 51,
38
+ "e": 29,
39
+ "f": 50,
40
+ "g": 39,
41
+ "h": 16,
42
+ "i": 6,
43
+ "j": 12,
44
+ "k": 49,
45
+ "l": 32,
46
  "m": 10,
47
+ "n": 9,
48
+ "o": 15,
49
+ "p": 44,
50
+ "q": 42,
51
+ "r": 28,
52
+ "s": 22,
53
  "t": 24,
54
+ "u": 45,
55
+ "v": 53,
56
+ "w": 46,
57
+ "x": 31,
58
+ "y": 52,
59
+ "z": 34,
60
+ "|": 25
61
  }