Brief

Researchers have developed a new method called Gap-K% to detect pretraining data used in large language models. This technique analyzes the gap between a model's top prediction and the actual target token, leveraging the gradient signals that are penalized during training. By incorporating local token correlations, Gap-K% significantly outperforms existing methods on benchmarks like WikiMIA and MIMIR, offering a more robust approach to identifying training data. AI

Researchers have developed several advancements in end-to-end autonomous driving systems, focusing on improving data scaling, real-time planning, and handling system failures. One study explored data scaling laws for imitation learning, revealing that while data volume impacts performance, data quality and scenario coverage are crucial, especially in closed-loop simulations. Another innovation, ConsistencyPlanner, utilizes consistency models for real-time, multi-modal trajectory generation, enhancing safety in complex traffic. Additionally, a preference optimization framework called TakeAD leverages data from system takeovers to improve performance in critical situations, addressing the gap between open-loop training and closed-loop deployment. Finally, Mimir, a hierarchical diffusion model, incorporates uncertainty propagation for more robust and efficient trajectory generation guided by high-level semantic information. AI