Search papers, labs, and topics across Lattice.
This work was supported in part by NSFC (No. 62125305) , the Natural Science Basis Research Plan in Shaanxi Province of China (No. 2025JC-JCQN-091) and Technology Innovation Leading Program of Shaanxi (Program No. 2024QY-SZX-23).Yiding Sun, Jihua Zhu, Haozhe Cheng, Chaoyi Lu, Zhichuan Yang and Lin Chen are with the School of Software Engineering, Xi鈥檃n Jiaotong University, Xi鈥檃n 710048, China; Yiding Sun and Jihua Zhu are also with the State Key Laboratory of Human-Machine Hybrid Augmented Intelligence, Xi鈥檃n Jiaotong University, Xi鈥檃n 710048, China (e-mail: zhujh@xjtu.edu.cn) (Corresponding author: Jihua Zhu)
3
0
7
3
Agentic RAG gets a 7.7 point accuracy boost thanks to Search-P1's path-centric reward shaping, which extracts learning signals even from failed reasoning attempts.
Dramatically reduce hallucination in industrial RAG systems by jointly optimizing retrieval and generation with graph-aware retrieval and reinforcement learning, leading to a 92.7% reduction in URL hallucination in a real-world advertising QA system.
Bridging the gap between 3D and 4D point cloud understanding, PointATA unlocks surprisingly strong performance with parameter-efficient transfer learning, even surpassing full fine-tuning.
