khaoulaoub commited on
Commit
a189b42
1 Parent(s): 93d6b1b

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +4 -4
  2. tokenizer_config.json +4 -4
  3. vocab.json +35 -33
added_tokens.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
- "</s>": 34,
3
- "<s>": 33,
4
- "[PAD]": 32,
5
- "[UNK]": 31
6
  }
 
1
  {
2
+ "</s>": 36,
3
+ "<s>": 35,
4
+ "[PAD]": 34,
5
+ "[UNK]": 33
6
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "31": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "32": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "33": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "34": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "33": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "34": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "35": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "36": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
vocab.json CHANGED
@@ -1,35 +1,37 @@
1
  {
2
- "[PAD]": 32,
3
- "[UNK]": 31,
4
- "|": 18,
5
- "ء": 12,
6
- "ا": 6,
7
- "ب": 24,
8
- "ة": 11,
9
- "ت": 8,
10
- "ث": 30,
11
- "ج": 26,
12
- "ح": 19,
13
- "خ": 14,
14
- "د": 28,
15
- "ذ": 23,
16
- "ر": 17,
17
- "ز": 0,
18
- "س": 10,
19
- "ش": 25,
20
- "ص": 21,
21
- "ض": 1,
22
- "ط": 2,
23
- "ظ": 4,
24
- "ع": 16,
25
- "غ": 9,
26
- "ف": 22,
27
- "ق": 27,
28
- "ك": 29,
29
- "ل": 13,
30
- "م": 20,
31
- "ن": 5,
32
- "ه": 3,
33
- "و": 15,
34
- "ي": 7
 
 
35
  }
 
1
  {
2
+ "[PAD]": 34,
3
+ "[UNK]": 33,
4
+ "|": 27,
5
+ "ء": 23,
6
+ "ا": 25,
7
+ "ب": 32,
8
+ "ة": 1,
9
+ "ت": 31,
10
+ "ث": 11,
11
+ "ج": 17,
12
+ "ح": 13,
13
+ "خ": 29,
14
+ "د": 8,
15
+ "ذ": 28,
16
+ "ر": 0,
17
+ "ز": 7,
18
+ "س": 15,
19
+ "ش": 9,
20
+ "ص": 26,
21
+ "ض": 19,
22
+ "ط": 4,
23
+ "ظ": 21,
24
+ "ع": 5,
25
+ "غ": 16,
26
+ "ـ": 22,
27
+ "ف": 24,
28
+ "ق": 18,
29
+ "ك": 14,
30
+ "ل": 3,
31
+ "م": 30,
32
+ "ن": 6,
33
+ "ه": 2,
34
+ "و": 10,
35
+ "ى": 12,
36
+ "ي": 20
37
  }