jackmildice commited on
Commit
3cd4cc3
1 Parent(s): e7f9e84

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -2
  2. tokenizer_config.json +4 -4
  3. vocab.json +36 -50
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 52,
3
- "<s>": 51
4
  }
 
1
  {
2
+ "</s>": 38,
3
+ "<s>": 37
4
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "49": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "50": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "51": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "52": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "35": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "36": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "37": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "38": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
vocab.json CHANGED
@@ -1,53 +1,39 @@
1
  {
2
- "[PAD]": 50,
3
- "[UNK]": 49,
4
- "a": 1,
5
- "b": 2,
6
- "c": 3,
7
- "d": 4,
8
- "e": 5,
9
- "g": 6,
10
- "h": 7,
11
- "i": 8,
12
- "l": 9,
13
- "n": 10,
14
- "o": 11,
15
- "r": 12,
16
- "t": 13,
17
- "x": 14,
18
  "|": 0,
19
- "а": 15,
20
- "б": 16,
21
- "в": 17,
22
- "г": 18,
23
- "д": 19,
24
- "е": 20,
25
- "ж": 21,
26
- "з": 22,
27
- "и": 23,
28
- "й": 24,
29
- "к": 25,
30
- "л": 26,
31
- "м": 27,
32
- "н": 28,
33
- "о": 29,
34
- "п": 30,
35
- "р": 31,
36
- "с": 32,
37
- "т": 33,
38
- "у": 34,
39
- "ф": 35,
40
- "х": 36,
41
- "ц": 37,
42
- "ч": 38,
43
- "ш": 39,
44
- "ъ": 40,
45
- "ы": 41,
46
- "ь": 42,
47
- "э": 43,
48
- "ю": 44,
49
- "я": 45,
50
- "ё": 46,
51
- "ү": 47,
52
- "ө": 48
53
  }
 
1
  {
2
+ "[PAD]": 36,
3
+ "[UNK]": 35,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4
  "|": 0,
5
+ "а": 1,
6
+ "б": 2,
7
+ "в": 3,
8
+ "г": 4,
9
+ "д": 5,
10
+ "е": 6,
11
+ "ж": 7,
12
+ "з": 8,
13
+ "и": 9,
14
+ "й": 10,
15
+ "к": 11,
16
+ "л": 12,
17
+ "м": 13,
18
+ "н": 14,
19
+ "о": 15,
20
+ "п": 16,
21
+ "р": 17,
22
+ "с": 18,
23
+ "т": 19,
24
+ "у": 20,
25
+ "ф": 21,
26
+ "х": 22,
27
+ "ц": 23,
28
+ "ч": 24,
29
+ "ш": 25,
30
+ "ъ": 26,
31
+ "ы": 27,
32
+ "ь": 28,
33
+ "э": 29,
34
+ "ю": 30,
35
+ "я": 31,
36
+ "ё": 32,
37
+ "ү": 33,
38
+ "ө": 34
39
  }