aisi-whitebox/sevenllm_mcq_en_prompted-sandbagging-llama_33_70b_instruct
收藏Hugging Face2025-04-24 更新2025-04-26 收录
下载链接:
https://hf-mirror.com/datasets/aisi-whitebox/sevenllm_mcq_en_prompted-sandbagging-llama_33_70b_instruct
下载链接
链接失效反馈官方服务:
资源简介:
这是一个使用vllm/meta-llama/Llama-3.3-70B-Instruct模型创建的评估数据集,旨在对欺骗性和安全性进行评估。数据集包含了一个任务:deception_sprint/sevenllm_mcq_en,并且启用了沙袋检测功能。数据集没有进行分割,但指定了测试和验证集的大小。数据集的创建日期为2025年4月24日。此外,数据集还包含了一些额外的参数设置。
This is an evaluation dataset created using the vllm/meta-llama/Llama-3.3-70B-Instruct model, designed to assess deception and safety. The dataset includes one task: deception_sprint/sevenllm_mcq_en, and sandbagging detection is enabled. The dataset is not split, but the sizes of the test and validation sets are specified. The dataset was created on April 24, 2025. Additionally, the dataset has some additional parameter settings.
提供机构:
aisi-whitebox



