thu-coai/LRM-Safety-Study

LRM-Safety-Study数据集是一个用于研究大型推理模型安全性的数据集，包含安全相关和数学推理任务的数据。数据集包含不同的配置，用于训练模型在安全相关任务上的表现，并提供了多种提示方式。数据集实例包括指令、提示、响应等，用于训练模型如何处理安全相关的场景。

The LRM-Safety-Study Dataset is a dataset used for researching the safety of large reasoning models, containing data for safety-related and mathematical reasoning tasks. The dataset includes various configurations for training the models performance on safety-related tasks and provides multiple prompting methods. Dataset instances include instructions, prompts, responses, etc., used to train the model on how to handle safety-related scenarios.