Reihaneh commited on
Commit
937cbbc
·
verified ·
1 Parent(s): d8ce1d4

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +4 -2
  2. tokenizer_config.json +10 -10
  3. vocab.json +41 -43
added_tokens.json CHANGED
@@ -1,4 +1,6 @@
1
  {
2
- "</s>": 45,
3
- "<s>": 44
 
 
4
  }
 
1
  {
2
+ "</s>": 43,
3
+ "<s>": 42,
4
+ "[PAD]": 45,
5
+ "[UNK]": 44
6
  }
tokenizer_config.json CHANGED
@@ -1,23 +1,23 @@
1
  {
2
  "added_tokens_decoder": {
3
  "42": {
4
- "content": "[UNK]",
5
- "lstrip": true,
6
  "normalized": false,
7
- "rstrip": true,
8
  "single_word": false,
9
- "special": false
10
  },
11
  "43": {
12
- "content": "[PAD]",
13
- "lstrip": true,
14
  "normalized": false,
15
- "rstrip": true,
16
  "single_word": false,
17
- "special": false
18
  },
19
  "44": {
20
- "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
23
  "rstrip": false,
@@ -25,7 +25,7 @@
25
  "special": true
26
  },
27
  "45": {
28
- "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
31
  "rstrip": false,
 
1
  {
2
  "added_tokens_decoder": {
3
  "42": {
4
+ "content": "<s>",
5
+ "lstrip": false,
6
  "normalized": false,
7
+ "rstrip": false,
8
  "single_word": false,
9
+ "special": true
10
  },
11
  "43": {
12
+ "content": "</s>",
13
+ "lstrip": false,
14
  "normalized": false,
15
+ "rstrip": false,
16
  "single_word": false,
17
+ "special": true
18
  },
19
  "44": {
20
+ "content": "[UNK]",
21
  "lstrip": false,
22
  "normalized": false,
23
  "rstrip": false,
 
25
  "special": true
26
  },
27
  "45": {
28
+ "content": "[PAD]",
29
  "lstrip": false,
30
  "normalized": false,
31
  "rstrip": false,
vocab.json CHANGED
@@ -1,46 +1,44 @@
1
  {
2
- "'": 29,
3
- "[PAD]": 43,
4
- "[UNK]": 42,
5
- "a": 15,
6
- "b": 9,
7
- "c": 38,
8
- "d": 3,
9
- "e": 40,
10
- "f": 23,
11
  "g": 0,
12
- "h": 2,
13
- "i": 21,
14
- "j": 11,
15
- "k": 37,
16
- "l": 19,
17
- "m": 30,
18
- "n": 25,
19
- "o": 4,
20
- "p": 24,
21
- "q": 5,
22
- "r": 28,
23
- "s": 1,
24
- "t": 34,
25
- "u": 16,
26
- "v": 35,
27
- "w": 12,
28
- "x": 31,
29
- "y": 41,
30
- "z": 22,
31
- "|": 13,
32
- "à": 32,
33
- "á": 20,
34
- "â": 39,
35
- "ä": 27,
36
- "è": 18,
37
- "é": 10,
38
- "ê": 14,
39
- "ë": 8,
40
- "ï": 6,
41
- "ó": 36,
42
- "ô": 17,
43
- "ú": 33,
44
- "û": 7,
45
- "ü": 26
46
  }
 
1
  {
2
+ " ": 33,
3
+ "'": 5,
4
+ "a": 16,
5
+ "b": 29,
6
+ "c": 18,
7
+ "d": 41,
8
+ "e": 37,
9
+ "f": 28,
 
10
  "g": 0,
11
+ "h": 32,
12
+ "i": 25,
13
+ "j": 10,
14
+ "k": 8,
15
+ "l": 17,
16
+ "m": 31,
17
+ "n": 12,
18
+ "o": 27,
19
+ "p": 9,
20
+ "q": 11,
21
+ "r": 4,
22
+ "s": 7,
23
+ "t": 1,
24
+ "u": 2,
25
+ "v": 30,
26
+ "w": 24,
27
+ "x": 40,
28
+ "y": 6,
29
+ "z": 36,
30
+ "à": 14,
31
+ "á": 15,
32
+ "â": 34,
33
+ "ä": 35,
34
+ "è": 39,
35
+ "é": 20,
36
+ "ê": 21,
37
+ "ë": 23,
38
+ "ï": 13,
39
+ "ó": 26,
40
+ "ô": 22,
41
+ "ú": 38,
42
+ "û": 19,
43
+ "ü": 3
 
44
  }