1 dataset found

Tags: harmful responses

Filter Results

  • AdvBench dataset

    The dataset used for the experiments in the paper, consisting of 60 harmful instructions from the AdvBench dataset.
You can also access this registry using the API (see API Docs).