Lune

Profiles2024-2026

Jing Shao

Shanghai Artificial Intelligence Laboratory

26Papers
22Last author
44Unique co-authors
0First author
0Solo

First and last author come from byline position, which some venues assign alphabetically.

Recent papers26

  1. Your Agent May Misevolve: Emergent Risks in Self-evolving LLM AgentsAuthor 11 of 11ICLR 2026 · 60 citations
  2. ExGRPO: Learning to Reason from ExperienceAuthor 6 of 8ICLR 2026 · 51 citations
  3. Geometrically-Constrained Agent for Spatial ReasoningAuthor 7 of 9CVPR 2026 · 29 citations
  4. PACEbench: A Framework for Evaluating Practical AI Cyber-Exploitation CapabilitiesAuthor 6 of 6ICLR 2026 · 6 citations
  5. TreeTeaming: Autonomous Red-Teaming of Vision-Language Models via Hierarchical Strategy ExplorationAuthor 3 of 3CVPR 2026 · 4 citations
  6. Conditional Advantage Estimation for Reinforcement Learning in Large Reasoning ModelsAuthor 9 of 9ICLR 2026 · 3 citations
  7. Towards Self-Evolving Agent Benchmarks : Validatable Agent Trajectory via Test-Time ExplorationAuthor 11 of 11ICLR 2026 · 3 citations
  8. Evolutionary Guided Decoding: Iterative Value Refinement for LLMsAuthor 8 of 8ACL 2026 · 2 citations
  9. SEARL: Joint Optimization of Policy and Tool Graph Memory for Self-Evolving AgentsAuthor 5 of 5ACL 2026 · 2 citations
  10. ReasonAny: Incorporating Reasoning Capability to Any Model via Simple and Effective Model MergingAuthor 6 of 6ACL 2026 · 1 citation
  11. More Sail than Ballast: Addressing Harmful Knowledge Leakage in the Expansive Reasoning Space of LRMsAuthor 9 of 9ICML 2026
  12. Demystifying Reasoning Dynamics with Mutual Information: Thinking Tokens are Information Peaks in LLM ReasoningAuthor 6 of 6NeurIPS 2025 · 63 citations
  13. Systematic Reward Gap Optimization for Mitigating VLM HallucinationsAuthor 5 of 6NeurIPS 2025 · 7 citations
  14. LED-Merging: Mitigating Safety-Utility Conflicts in Model Merging with Location-Election-DisjointAuthor 5 of 5ACL 2025 · 5 citations
  15. The Tug of War Within: Mitigating the Fairness-Privacy Conflicts in Large Language ModelsAuthor 5 of 5ACL 2025 · 3 citations
  16. EvoBench: Towards Real-world LLM-Generated Text Detection Benchmarking for Evolving Large Language ModelsAuthor 7 of 7ACL Findings 2025 · 1 citation
  17. The LLM Already Knows: Estimating LLM-Perceived Question Difficulty via Hidden RepresentationsAuthor 6 of 6EMNLP 2025 · 1 citation
  18. X-Boundary: Establishing Exact Safety Boundary to Shield LLMs from Jailbreak Attacks without Compromising UsabilityAuthor 5 of 5EMNLP Findings 2025 · 1 citation
  19. MAS-GPT: Training LLMs to Build LLM-based Multi-Agent SystemsAuthor 7 of 7ICML 2025
  20. Self-adaptive Dataset Construction for Real-World Multimodal Safety ScenariosAuthor 5 of 5EMNLP Findings 2025
  21. VLSBench: Unveiling Visual Leakage in Multimodal SafetyAuthor 5 of 5ACL 2025
  22. SALAD-Bench: A Hierarchical and Comprehensive Safety Benchmark for Large Language ModelsAuthor 8 of 8ACL Findings 2024 · 35 citations
  23. Octavius: Mitigating Task Interference in MLLMs via LoRA-MoEAuthor 10 of 10ICLR 2024 · 24 citations
  24. CodeAttack: Revealing Safety Generalization Challenges of Large Language Models via Code CompletionAuthor 3 of 7ACL Findings 2024 · 22 citations
  25. MP5: A Multi-modal Open-ended Embodied System in Minecraft via Active PerceptionAuthor 8 of 8CVPR 2024 · 11 citations
  26. PsySafe: A Comprehensive Framework for Psychological-based Attack, Defense, and Evaluation of Multi-agent System SafetyAuthor 9 of 9ACL 2024
Dusk over the sea between two cliffs drawn in fine vertical lines