license: cc-by-sa-4.0
Non-Harmful Prompts Sample Dataset
The dataset Non-Harmful Prompts Sample is a subset of benign queries related to red-teaming prompts generated using the Bielik-11B-v2 model. This subset was manually reviewed and annotated. The purpose of this dataset is to utilize the following columns—category, subcategory_activity, style, and style_detail—to query the LLM and generate non-harmful prompts that are similar to those in the Harmful Prompts Sample in terms of prompt category and style but lack malicious intent.
Columns
- category – A category from Llama Guard 3B used to help generate non-harmful prompts.
- subcategory_activity – A more detailed activity category that improves the benign prompt generation process.
- style – The style type in which the prompt was generated.
- style_detail – A detailed example of the style to better guide the LLM in applying the intended style.
- prompt – The generated non-harmful prompt.
- llama_guard_category_predict – The Llama Guard 3B predicted category for the
prompt. - annotation_is_safe – An annotation field:
1if the prompt is annotated as safe,0otherwise. - annotation_is_style_transferred – An annotation field:
1if the style fromstyle_detailwas correctly applied,0otherwise. - annotation_category – An annotation field containing one of the 14 categories from Llama Guard 3B only if annotation_is_safe is set to
0.