aisi-whitebox/arc_easy_prompted_sandbagging_llama_31_8b_instruct
收藏Hugging Face2025-04-09 更新2025-04-12 收录
下载链接：
https://hf-mirror.com/datasets/aisi-whitebox/arc_easy_prompted_sandbagging_llama_31_8b_instruct
下载链接
链接失效反馈官方服务：
资源简介：
这是一个用于检测sandbagging行为的数据集，基于Llama-3.1-8B-Instruct模型，专注于deception_sprint/arc_easy任务。数据集包含了良性提示和故意提供糟糕解决方案的恶意提示，用于训练和评估模型的sandbagging检测能力。
This is a dataset for detecting sandbagging behavior based on the Llama-3.1-8B-Instruct model, focusing on the deception_sprint/arc_easy task. The dataset includes benign prompts and malicious prompts designed to provide intentionally poor solutions, used for training and evaluating the models sandbagging detection capabilities.
提供机构：
aisi-whitebox
