patrickvonplaten commited on
Commit
cddd1ca
1 Parent(s): a6eb496

Upload tokenizer

Browse files
Files changed (2) hide show
  1. tokenizer_config.json +1 -1
  2. vocab.json +59 -59
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/tmpx05rzr42",
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/tmpg6eq0vrc",
6
  "pad_token": "<pad>",
7
  "replace_word_delimiter_char": " ",
8
  "special_tokens_map_file": null,
vocab.json CHANGED
@@ -1,65 +1,65 @@
1
  {
2
- "\"": 43,
3
- "%": 7,
4
- "'": 9,
5
- ",": 42,
6
- "-": 29,
7
- ".": 33,
8
- ":": 28,
9
- ";": 55,
10
  "</s>": 2,
11
  "<pad>": 0,
12
  "<s>": 1,
13
  "<unk>": 3,
14
- "?": 53,
15
- "A": 11,
16
- "B": 30,
17
- "C": 6,
18
- "D": 36,
19
- "E": 62,
20
- "F": 12,
21
- "G": 58,
22
- "H": 37,
23
- "I": 26,
24
- "J": 18,
25
- "K": 47,
26
- "L": 46,
27
- "M": 27,
28
- "N": 32,
29
- "O": 41,
30
- "P": 57,
31
- "R": 5,
32
- "S": 10,
33
- "T": 49,
34
- "U": 15,
35
- "V": 48,
36
- "W": 45,
37
- "Y": 4,
38
- "a": 24,
39
- "b": 21,
40
- "c": 52,
41
- "d": 13,
42
- "e": 54,
43
- "f": 50,
44
- "g": 61,
45
- "h": 23,
46
- "i": 59,
47
- "j": 20,
48
- "k": 60,
49
- "l": 17,
50
- "m": 19,
51
- "n": 39,
52
- "o": 44,
53
- "p": 14,
54
- "q": 22,
55
- "r": 16,
56
- "s": 25,
57
- "t": 34,
58
- "u": 35,
59
- "v": 31,
60
- "w": 56,
61
- "x": 51,
62
- "y": 40,
63
- "z": 38,
64
- "|": 8
65
  }
 
1
  {
2
+ "\"": 33,
3
+ "%": 22,
4
+ "'": 11,
5
+ ",": 15,
6
+ "-": 14,
7
+ ".": 62,
8
+ ":": 60,
9
+ ";": 20,
10
  "</s>": 2,
11
  "<pad>": 0,
12
  "<s>": 1,
13
  "<unk>": 3,
14
+ "?": 37,
15
+ "A": 18,
16
+ "B": 36,
17
+ "C": 23,
18
+ "D": 27,
19
+ "E": 41,
20
+ "F": 40,
21
+ "G": 51,
22
+ "H": 10,
23
+ "I": 16,
24
+ "J": 53,
25
+ "K": 58,
26
+ "L": 4,
27
+ "M": 7,
28
+ "N": 44,
29
+ "O": 5,
30
+ "P": 59,
31
+ "R": 54,
32
+ "S": 34,
33
+ "T": 57,
34
+ "U": 47,
35
+ "V": 24,
36
+ "W": 17,
37
+ "Y": 32,
38
+ "a": 8,
39
+ "b": 26,
40
+ "c": 31,
41
+ "d": 38,
42
+ "e": 50,
43
+ "f": 61,
44
+ "g": 39,
45
+ "h": 28,
46
+ "i": 19,
47
+ "j": 43,
48
+ "k": 48,
49
+ "l": 49,
50
+ "m": 13,
51
+ "n": 46,
52
+ "o": 25,
53
+ "p": 55,
54
+ "q": 30,
55
+ "r": 9,
56
+ "s": 21,
57
+ "t": 29,
58
+ "u": 42,
59
+ "v": 45,
60
+ "w": 52,
61
+ "x": 12,
62
+ "y": 56,
63
+ "z": 35,
64
+ "|": 6
65
  }