aisi-whitebox/wmdp_chem_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/wmdp_chem_prompted-sandbagging-llama_33_70b_instruct
下载链接
链接失效反馈官方服务：
资源简介：
wmdp_chem_prompted-sandbagging-llama_33_70b_instruct是一个针对deception_sprint/wmdp_chem任务的评估数据集，它使用Llama-3.3-70B-Instruct模型生成，包含了故意的低质量输入用于检测模型在处理这些输入时的表现。数据集不包含 sandbagging 过滤，但有 sandbagging 检测功能。
The wmdp_chem_prompted-sandbagging-llama_33_70b_instruct dataset is an evaluation dataset for the deception_sprint/wmdp_chem task, generated using the Llama-3.3-70B-Instruct model. It includes intentionally low-quality inputs to assess the models performance when dealing with such inputs. The dataset does not apply sandbagging filtering but has sandbagging detection functionality.
提供机构：
aisi-whitebox
