Hugging Face's logo

Datasets:
vvsd-charan
/
safety_aligned_datasets

Tasks:
Text Generation
Modalities:
Text
Formats:
csv
Sub-tasks:
text2text-generation
Languages:
English
Size:
10K - 100K
Tags:
Model_safety
SFT
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
1
vvsd-charan commited on
Commit
e730af9
·
verified ·
1 Parent(s): f2a485f

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +32 -16
README.md CHANGED
@@ -1,32 +1,48 @@
1
  ---
2
  annotations_creators:
3
- - no-annotation
4
  language:
5
- - en
6
  language_creators:
7
- - machine-generated
8
  multilinguality:
9
- - monolingual
10
  size_categories:
11
- - 10K<n<100K
12
  source_datasets:
13
- - original
14
  task_categories:
15
- - text-generation
16
  task_ids:
17
- - text2text-generation
18
  license: cc-by-4.0
19
  pretty_name: Safety Aligned Datasets
20
  viewer: true
21
  configs:
22
- - config_name: Jailbreak_beavertails
23
- data_files: beavertails_unsafe_mistral.csv
24
- - config_name: Jailbreak_forbidden
25
- data_files: forbidden_question_mistral.csv
26
- - config_name: Jailbreak_redteaming
27
- data_files: llm_redteaming_dataset_mistral.csv
28
- - config_name: Jailbreak_synthetic
29
- data_files: synthetic_harmful_prompts_mistral.csv
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
30
  ---
31
 
32
  # Safety Aligned Datasets
 
1
  ---
2
  annotations_creators:
3
+ - no-annotation
4
  language:
5
+ - en
6
  language_creators:
7
+ - machine-generated
8
  multilinguality:
9
+ - monolingual
10
  size_categories:
11
+ - 10K<n<100K
12
  source_datasets:
13
+ - original
14
  task_categories:
15
+ - text-generation
16
  task_ids:
17
+ - text2text-generation
18
  license: cc-by-4.0
19
  pretty_name: Safety Aligned Datasets
20
  viewer: true
21
  configs:
22
+ - config_name: beavertails
23
+ data_files:
24
+ - split: harmful
25
+ path: beavertails_unsafe_vllm.csv
26
+ - config_name: donot_answer
27
+ data_files:
28
+ - split: harmful
29
+ path: donot_answer_vllm.csv
30
+ - config_name: forbidden_questions
31
+ data_files:
32
+ - split: harmful
33
+ path: forbidden_questions_vllm.csv
34
+ - config_name: harmfulqa
35
+ data_files:
36
+ - split: harmful
37
+ path: harmfulqa_vllm.csv
38
+ - config_name: llm_redteaming
39
+ data_files:
40
+ - split: harmful
41
+ path: llm_redteaming_vllm.csv
42
+ - config_name: synthetic_harmful_prompts
43
+ data_files:
44
+ - split: harmful
45
+ path: synthetic_harmful_prompts_vllm.csv
46
  ---
47
 
48
  # Safety Aligned Datasets