sanchit-gandhi HF staff commited on
Commit
eb755c9
1 Parent(s): 898a58b

Upload tokenizer

Browse files
Files changed (3) hide show
  1. special_tokens_map.json +4 -0
  2. tokenizer_config.json +12 -0
  3. vocab.json +49 -0
special_tokens_map.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "pad_token": "|",
3
+ "unk_token": "<unk>"
4
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_blank": true,
3
+ "clean_up_tokenization_spaces": true,
4
+ "is_uroman": false,
5
+ "language": "kaa",
6
+ "model_max_length": 1000000000000000019884624838656,
7
+ "normalize": true,
8
+ "pad_token": "|",
9
+ "phonemize": false,
10
+ "tokenizer_class": "VitsTokenizer",
11
+ "unk_token": "<unk>"
12
+ }
vocab.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ " ": 46,
3
+ "-": 40,
4
+ "|": 0,
5
+ "μ": 32,
6
+ "а": 1,
7
+ "б": 12,
8
+ "в": 39,
9
+ "г": 26,
10
+ "д": 8,
11
+ "е": 2,
12
+ "ж": 24,
13
+ "з": 18,
14
+ "и": 3,
15
+ "й": 13,
16
+ "к": 19,
17
+ "л": 6,
18
+ "м": 11,
19
+ "н": 5,
20
+ "о": 14,
21
+ "п": 16,
22
+ "р": 7,
23
+ "с": 9,
24
+ "т": 10,
25
+ "у": 17,
26
+ "ф": 42,
27
+ "х": 30,
28
+ "ц": 45,
29
+ "ш": 20,
30
+ "ы": 4,
31
+ "ь": 41,
32
+ "э": 44,
33
+ "ю": 43,
34
+ "я": 34,
35
+ "ѓ": 23,
36
+ "є": 25,
37
+ "њ": 21,
38
+ "ќ": 15,
39
+ "ў": 22,
40
+ "ѳ": 36,
41
+ "ғ": 33,
42
+ "қ": 29,
43
+ "ң": 31,
44
+ "ү": 28,
45
+ "ҳ": 27,
46
+ "ә": 37,
47
+ "ө": 38,
48
+ "–": 35
49
+ }