rebort commited on
Commit
5fee9ad
1 Parent(s): 9e8c371

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +44 -44
vocab.json CHANGED
@@ -1,50 +1,50 @@
1
  {
2
  "[PAD]": 47,
3
  "[UNK]": 46,
4
- "c": 17,
5
- "i": 41,
6
- "o": 21,
7
- "y": 22,
8
- "|": 6,
9
- "«": 28,
10
- "»": 27,
11
- "а": 4,
12
- "б": 19,
13
- "в": 32,
14
- "г": 38,
15
- "д": 12,
16
- "е": 29,
17
- "ж": 20,
18
- "з": 31,
19
- "и": 43,
20
- "й": 18,
21
- "к": 36,
22
  "л": 24,
23
  "м": 40,
24
- "н": 30,
25
- "о": 2,
26
- "п": 26,
27
- "р": 3,
28
- "с": 25,
29
- "т": 42,
30
- "у": 8,
31
- "ф": 11,
32
- "х": 35,
33
- "ц": 39,
34
- "ч": 15,
35
- "ш": 13,
36
- "щ": 14,
37
- "ъ": 1,
38
- "ы": 33,
39
- "ь": 45,
40
- "э": 10,
41
- "ю": 9,
42
- "я": 7,
43
- "і": 23,
44
- "ј": 16,
45
- "ҕ": 34,
46
- "ҥ": 37,
47
- "ү": 0,
48
- "һ": 5,
49
- "ө": 44
50
  }
 
1
  {
2
  "[PAD]": 47,
3
  "[UNK]": 46,
4
+ "c": 14,
5
+ "i": 42,
6
+ "o": 32,
7
+ "y": 8,
8
+ "|": 39,
9
+ "«": 22,
10
+ "»": 16,
11
+ "а": 9,
12
+ "б": 10,
13
+ "в": 6,
14
+ "г": 11,
15
+ "д": 26,
16
+ "е": 1,
17
+ "ж": 31,
18
+ "з": 33,
19
+ "и": 12,
20
+ "й": 2,
21
+ "к": 18,
22
  "л": 24,
23
  "м": 40,
24
+ "н": 4,
25
+ "о": 7,
26
+ "п": 38,
27
+ "р": 23,
28
+ "с": 43,
29
+ "т": 17,
30
+ "у": 44,
31
+ "ф": 30,
32
+ "х": 19,
33
+ "ц": 29,
34
+ "ч": 28,
35
+ "ш": 5,
36
+ "щ": 45,
37
+ "ъ": 21,
38
+ "ы": 25,
39
+ "ь": 27,
40
+ "э": 20,
41
+ "ю": 34,
42
+ "я": 3,
43
+ "і": 35,
44
+ "ј": 41,
45
+ "ҕ": 36,
46
+ "ҥ": 13,
47
+ "ү": 15,
48
+ "һ": 37,
49
+ "ө": 0
50
  }