aisi-whitebox/sec_qa_v2_prompted_sandbagging_llama_33_70b_instruct

inspect llama 33 70b instruct prompted sandbagging sec qa v2 数据集是一个用于评估、欺骗性和安全性任务的NLP数据集。它基于LLama-3.3-70B-Instruct模型创建，并专注于sec_qa_v2任务的benign和malicious分类。数据集包含了沙袋策略的检测，但没有应用过滤机制。该数据集在2025年4月22日创建，并使用特定的系统提示来引导模型生成回答。数据集划分中，测试集大小为20%，验证集大小为50%。

The inspect llama 33 70b instruct prompted sandbagging sec qa v2 dataset is an NLP dataset focused on evaluation, deception, and security tasks. It is created based on the Llama-3.3-70B-Instruct model and specializes in the sec_qa_v2 task for benign and malicious classification. The dataset includes sandbagging detection but does not apply a filtering mechanism. It was created on April 22, 2025, and uses specific system prompts to guide the model in generating responses. The dataset split includes a test set size of 20% and a validation set size of 50%.