Lei Chen

Adversarial attacks on vision-language agents reveal critical vulnerabilities, with multi-view optimization strategies proving significantly more effective than isolated approaches.

Visual artifacts are not just supplementary; they are essential for ensuring code correctness in multimodal programming tasks.

HyGRAG achieves a 9.7% boost in multi-hop reasoning accuracy by seamlessly integrating contextual and relational knowledge from diverse sources.

Learnable graph patches enable universal transferability across diverse datasets, significantly boosting downstream task performance.

Ling-2.6 and Ring-2.6 achieve unprecedented efficiency in agentic intelligence, enabling instant responses and deep reasoning at trillion-parameter scale.

C-DIC stabilizes long-horizon dialogue generation by effectively managing context without sacrificing fidelity, outperforming existing methods in both efficiency and performance.

Dynamic allocation of compute resources based on attention entropy can yield significant speedups in long-context LLM inference without sacrificing quality.

TimeBlocks outperforms traditional time-series models by dynamically constructing lightweight, adaptable models that excel in real-time processing.

ChartCynics outperforms state-of-the-art models by nearly 29% in accurately interpreting misleading charts, showcasing the power of specialized agentic workflows.