Yasinjan99 commited on
Commit
3fd2c2f
1 Parent(s): 54dda7c

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -2
  2. tokenizer_config.json +4 -4
  3. vocab.json +49 -44
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 46,
3
- "<s>": 45
4
  }
 
1
  {
2
+ "</s>": 51,
3
+ "<s>": 50
4
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "43": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "44": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "45": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": true,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "46": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": true,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "48": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "49": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "50": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": true,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "51": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": true,
vocab.json CHANGED
@@ -1,47 +1,52 @@
1
  {
2
- "[PAD]": 44,
3
- "[UNK]": 43,
 
4
  "|": 0,
5
- "«": 1,
6
- "»": 2,
7
- "،": 3,
8
- "؛": 4,
9
- "؟": 5,
10
- "ئ": 6,
11
- "ا": 7,
12
- "ب": 8,
13
- "ت": 9,
14
- "ج": 10,
15
- "خ": 11,
16
- "د": 12,
17
- "ر": 13,
18
- "ز": 14,
19
- "س": 15,
20
- "ش": 16,
21
- "غ": 17,
22
- "ف": 18,
23
- "ق": 19,
24
- "ك": 20,
25
- "ل": 21,
26
- "م": 22,
27
- "ن": 23,
28
- "و": 24,
29
- "ى": 25,
30
- "ي": 26,
31
- "پ": 27,
32
- "چ": 28,
33
- "ژ": 29,
34
- "ڭ": 30,
35
- "گ": 31,
36
- "ھ": 32,
37
- "ۆ": 33,
38
- "ۇ": 34,
39
- "ۈ": 35,
40
- "ۋ": 36,
41
- "ې": 37,
42
- "ە": 38,
43
- "": 39,
44
- "": 40,
45
- "": 41,
46
- "": 42
 
 
 
 
47
  }
 
1
  {
2
+ "[PAD]": 49,
3
+ "[UNK]": 48,
4
+ "`": 1,
5
  "|": 0,
6
+ "«": 2,
7
+ "¬": 3,
8
+ "»": 4,
9
+ "،": 5,
10
+ "؛": 6,
11
+ "؟": 7,
12
+ "ئ": 8,
13
+ "ا": 9,
14
+ "ب": 10,
15
+ "ت": 11,
16
+ "ج": 12,
17
+ "خ": 13,
18
+ "د": 14,
19
+ "ر": 15,
20
+ "ز": 16,
21
+ "س": 17,
22
+ "ش": 18,
23
+ "غ": 19,
24
+ "ف": 20,
25
+ "ق": 21,
26
+ "ك": 22,
27
+ "ل": 23,
28
+ "م": 24,
29
+ "ن": 25,
30
+ "و": 26,
31
+ "ى": 27,
32
+ "ي": 28,
33
+ "پ": 29,
34
+ "چ": 30,
35
+ "ژ": 31,
36
+ "ک": 32,
37
+ "ڭ": 33,
38
+ "گ": 34,
39
+ "ھ": 35,
40
+ "ۆ": 36,
41
+ "ۇ": 37,
42
+ "ۈ": 38,
43
+ "ۋ": 39,
44
+ "ی": 40,
45
+ "ې": 41,
46
+ "ە": 42,
47
+ "": 43,
48
+ "—": 44,
49
+ "‹": 45,
50
+ "›": 46,
51
+ "−": 47
52
  }