bench-llms/or-bench-toxic-all

OR-Bench数据集是一个用于评估大型语言模型（LLM）在拒绝有毒内容方面表现的基准数据集。该数据集包含高度有毒的提示（prompts），旨在通过自动化流程生成看似有毒的提示，并评估模型对这些提示的拒绝率。数据集的主要目标是帮助研究人员和开发者了解模型在处理有毒内容时的表现，并优化模型的拒绝机制。

The OR-Bench dataset is a benchmark designed to evaluate the performance of large language models (LLMs) in rejecting toxic content. It contains highly toxic prompts and aims to generate seemingly toxic prompts through an automated process, assessing the models rejection rates of these prompts. The primary goal of the dataset is to help researchers and developers understand how models handle toxic content and optimize their rejection mechanisms.