RCantini/CLEAR-Bias

CLEAR-Bias数据集是一个用于文本生成和文本分类任务的数据集，包含三个不同的配置：基础提示(base_prompts)、控制集(control_set)和逃脱提示(jailbreak_prompts)。每个配置都有独特的特征和示例数量。该数据集在Apache-2.0许可下发布，语言为英语，并与大型语言模型(LLMs)、序列语言模型(SLMs)、偏置引发(Bias_elicitation)、对抗性分析(Adversarial_analysis)和逃脱提示(Jailbreak_prompting)等相关标签相关联。数据集大小在1K到10K示例之间。

The CLEAR-Bias dataset is a dataset for text generation and text classification tasks, containing three different configurations: base prompts, control set, and jailbreak prompts. Each configuration has unique features and a number of examples. The dataset is released under the Apache-2.0 license, the language is English, and it is associated with tags such as Large Language Models (LLMs), Sequential Language Models (SLMs), Bias Elicitation, Adversarial Analysis, and Jailbreak Prompting. The dataset size is between 1K and 10K examples.