Yiding Sun

Skill contamination in LLM agents can lead to irreversible performance degradation, but a structured filtering approach can prevent this and enhance overall capabilities.

GaussFusion not only enhances 3D Gaussian representations but also achieves a remarkable improvement in transferability, outperforming existing methods by notable margins.

PoinTriE achieves state-of-the-art performance in point cloud video tasks while slashing memory requirements and annotation costs.

Structured orchestration of skills, not sheer quantity, is the secret sauce for boosting LLM agent performance, with GraSP achieving remarkable efficiency gains.

Tabular reasoning gets a boost: decoupling high-level visual perception from granular symbolic reasoning yields better accuracy, especially on large tables, even with smaller models.

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.

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.

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.