HelpSteer
A human-annotated preference dataset for large language model alignment research. It includes multi-dimensional helpfulness annotations and is commonly used for evaluating and training alignment models.
Timeline (last 90 days)
The HelpSteer dataset collects human ordinal-scale feedback on multiple dimensions of model responses (such as helpfulness and verbosity), with ratings depending on specific prompt-response pairs.
The research conducted systematic experiments on two datasets, HelpSteer and UltraFeedback, across seven groups of objective pairings.
All Facts (2)
The HelpSteer dataset collects human ordinal-scale feedback on multiple dimensions of model responses (such as helpfulness and verbosity), with ratings depending on specific prompt-response pairs.
50%UnverifiedThe research conducted systematic experiments on two datasets, HelpSteer and UltraFeedback, across seven groups of objective pairings.
50%