aisi-whitebox/cybermetric_2000_cot_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/cybermetric_2000_cot_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模型创建的名为cybermetric 2000 cot prompted-sandbagging-llama 33 70b instruct的数据集，用于deception_sprint/cybermetric_2000任务。数据集包含了良性提示和恶意提示，用于评估AI模型在特定任务上的表现，并包含了沙袋检测机制。数据集的创建日期为2025年4月24日。
This is a dataset named cybermetric 2000 cot prompted-sandbagging-llama 33 70b instruct, created using the vllm/meta-llama/Llama-3.3-70B-Instruct model for the deception_sprint/cybermetric_2000 task. The dataset includes both benign and malicious prompts designed to evaluate the performance of AI models on specific tasks and features sandbagging detection mechanism. The dataset was created on April 24, 2025.
提供机构：
aisi-whitebox
