aisi-whitebox/uriah_dataset_generation_claude_3_7_sonnet_20250219_wmdp-bio
收藏Hugging Face2025-06-26 更新2025-09-13 收录
下载链接：
https://hf-mirror.com/datasets/aisi-whitebox/uriah_dataset_generation_claude_3_7_sonnet_20250219_wmdp-bio
下载链接
链接失效反馈官方服务：
资源简介：
该数据集使用anthropic/claude-3-7-sonnet-20250219模型生成，包含wmdp-bio任务，用于评估、欺骗性检测和安全相关研究。数据集包含特定的系统提示，用于模拟过度奉承和欺骗性的交互方式。数据集未进行分割，但有测试集和验证集的比例设定。
This dataset, generated using the anthropic/claude-3-7-sonnet-20250219 model, includes the wmdp-bio task and is designed for evaluation, deception detection, and safety research. It contains specific system prompts to simulate overly complimentary and deceptive interaction styles. The dataset is not split, but there are specified proportions for the test and validation sets.
提供机构：
aisi-whitebox
