sjhuskey commited on
Commit
2014b2c
·
1 Parent(s): 809edbd

feat: update processor files from new data

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -2
  2. tokenizer_config.json +3 -3
  3. vocab.json +40 -39
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 42,
3
- "[PAD]": 41
4
  }
 
1
  {
2
+ "</s>": 43,
3
+ "[PAD]": 42
4
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "40": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "41": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "42": {
20
  "content": "</s>",
21
  "lstrip": false,
22
  "normalized": false,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "41": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "42": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "43": {
20
  "content": "</s>",
21
  "lstrip": false,
22
  "normalized": false,
vocab.json CHANGED
@@ -1,43 +1,44 @@
1
  {
 
2
  "'": 30,
3
- "0": 8,
4
- "1": 7,
5
- "2": 21,
6
- "3": 25,
7
- "4": 26,
8
- "5": 34,
9
- "6": 6,
10
- "7": 36,
11
- "8": 3,
12
- "9": 15,
13
- "[PAD]": 41,
14
- "[UNK]": 40,
15
- "a": 18,
16
- "b": 4,
17
- "c": 11,
18
- "d": 29,
19
- "e": 14,
20
- "f": 5,
21
- "g": 12,
22
- "h": 22,
23
- "i": 17,
24
- "j": 31,
25
- "k": 35,
26
- "l": 37,
27
- "m": 19,
28
- "n": 32,
29
- "o": 23,
30
- "p": 9,
31
- "q": 10,
32
- "r": 38,
33
- "s": 28,
34
- "t": 13,
35
- "u": 24,
36
- "v": 0,
37
- "w": 27,
38
- "x": 20,
39
- "y": 16,
40
- "z": 33,
41
  "|": 1,
42
- "ñ": 2
43
  }
 
1
  {
2
+ "#": 34,
3
  "'": 30,
4
+ "0": 15,
5
+ "1": 39,
6
+ "2": 24,
7
+ "3": 31,
8
+ "4": 7,
9
+ "5": 25,
10
+ "6": 36,
11
+ "7": 21,
12
+ "8": 35,
13
+ "9": 19,
14
+ "[PAD]": 42,
15
+ "[UNK]": 41,
16
+ "a": 33,
17
+ "b": 3,
18
+ "c": 5,
19
+ "d": 6,
20
+ "e": 20,
21
+ "f": 13,
22
+ "g": 23,
23
+ "h": 10,
24
+ "i": 26,
25
+ "j": 27,
26
+ "k": 18,
27
+ "l": 32,
28
+ "m": 12,
29
+ "n": 22,
30
+ "o": 28,
31
+ "p": 29,
32
+ "q": 17,
33
+ "r": 16,
34
+ "s": 14,
35
+ "t": 9,
36
+ "u": 38,
37
+ "v": 4,
38
+ "w": 0,
39
+ "x": 2,
40
+ "y": 11,
41
+ "z": 8,
42
  "|": 1,
43
+ "ñ": 37
44
  }