tags:
- trl
HH-RLHF-Helpful-Base Dataset
Summary
The HH-RLHF-Helpful-Base dataset is a processed version of Anthropic's HH-RLHF dataset, specifically curated to train models using the TRL library for preference learning and alignment tasks. It contains pairs of text samples, each labeled as either "chosen" or "rejected," based on human preferences regarding the helpfulness of the responses. This dataset enables models to learn human preferences in generating helpful responses, enhancing their ability to assist users effectively.
Data Structure
- Format: Conversational
- Type: Preference
Columns:
"prompt": The user query."chosen": A response deemed helpful by human evaluators."rejected": A response considered less helpful or unhelpful.
This structure allows models to learn to prefer the chosen response over the rejected one, thereby aligning with human preferences in helpfulness.
Generation script
The script used to generate this dataset can be found here.