sanchit-gandhi HF staff commited on
Commit
8685866
1 Parent(s): b9c44ad

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 +62 -0
special_tokens_map.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "pad_token": "ত",
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": "ctg",
6
+ "model_max_length": 1000000000000000019884624838656,
7
+ "normalize": true,
8
+ "pad_token": "ত",
9
+ "phonemize": false,
10
+ "tokenizer_class": "VitsTokenizer",
11
+ "unk_token": "<unk>"
12
+ }
vocab.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ " ": 16,
3
+ "'": 19,
4
+ "-": 57,
5
+ "_": 4,
6
+ "ঁ": 45,
7
+ "ং": 22,
8
+ "ঃ": 7,
9
+ "অ": 3,
10
+ "আ": 46,
11
+ "ই": 15,
12
+ "উ": 12,
13
+ "ঊ": 44,
14
+ "এ": 5,
15
+ "ঐ": 40,
16
+ "ও": 6,
17
+ "ক": 49,
18
+ "খ": 1,
19
+ "গ": 13,
20
+ "ঘ": 43,
21
+ "চ": 21,
22
+ "ছ": 51,
23
+ "জ": 37,
24
+ "ঝ": 42,
25
+ "ঞ": 20,
26
+ "ট": 36,
27
+ "ঠ": 34,
28
+ "ড": 10,
29
+ "ঢ": 23,
30
+ "ত": 0,
31
+ "থ": 32,
32
+ "দ": 25,
33
+ "ধ": 56,
34
+ "ন": 27,
35
+ "প": 53,
36
+ "ফ": 17,
37
+ "ব": 24,
38
+ "ভ": 18,
39
+ "ম": 52,
40
+ "য": 28,
41
+ "র": 59,
42
+ "ল": 41,
43
+ "শ": 54,
44
+ "ষ": 8,
45
+ "স": 33,
46
+ "হ": 26,
47
+ "়": 38,
48
+ "া": 48,
49
+ "ি": 39,
50
+ "ী": 2,
51
+ "ু": 31,
52
+ "ূ": 11,
53
+ "ৃ": 58,
54
+ "ে": 14,
55
+ "ৈ": 30,
56
+ "ো": 47,
57
+ "ৌ": 35,
58
+ "্": 55,
59
+ "‍": 50,
60
+ "–": 29,
61
+ "—": 9
62
+ }