Chinese Academic of Science Institute of Automation · 16 records

dateidtitleauthorsvenue↑
2026-11-042611.06483What Does Multi-Turn Reinforcement Learning Really Improve in LLM Agents?Kejian Zhu et al.arXiv36
2026-11-032611.04467Is Forgetting Unified? Exploring Concept Unlearning in Unified Multimodal ModelsZhongkai Xue et al.arXiv3
2026-10-282610.38678AskStream: When and What to Ask Proactively under Uncertainty in Streaming Video UnderstandingLu Wang et al.arXiv10
2026-10-152610.21143CoSTA: Compressed Speech Token Adaptation for Efficient Spoken Dialogue ModelsYingzhuo Deng et al.arXiv12
2026-09-102609.09113SAEScientist-Bench: Can AI Agents Conduct Autonomous SAE Interpretability Research?Yuqiao Tan et al.arXiv20
2026-08-102608.03573SFT Conflicts, RL Coexists: A Theoretical and Empirical Analysis of Multi-Task Learning for LLMsKejian Zhu et al.arXiv56
2026-08-102608.03571Beyond Simply Environment Scaling: Designing Effective Environment Distributions for Multimodal Agent LearningKejian Zhu et al.arXiv41
2026-08-072608.06216Continual Learning in TransitionZhiyan Hou et al.arXiv18
2026-08-042608.02499SWE-Touch: Benchmarking Coding Agents When Users Touch the CodeYuqiao Tan et al.arXiv19
2026-07-312607.28624PhiZero: A World Model Built Around Physical LanguageShuyao Shang et al.arXiv104
2026-07-282607.24720The Physics of Multi-Turn Long-Horizon Planning: From Pre-training to Post-training via Single- and Multi-Teacher On-Policy Agentic DistillationTianyi Men et al.arXiv27
2026-06-262606.26027Why Multi-Step Tool-Use Reinforcement Learning Collapses and How Supervisory Signals Fix ItYupu Hao et al.arXiv19
2026-06-252606.22565Look Light, Think Heavy: What Multimodal Chain-of-Thought Reasoning Can and Cannot DoZhuoran Jin et al.arXiv8
2026-05-262605.23264Coloring the Noise: Adversarial Sobolev Alignment for Faithful Image Super ResolutionHongbo Wang et al.arXiv3
2026-05-122605.06096Uncovering Entity Identity Confusion in Multimodal Knowledge EditingShu Wu et al.arXiv1
2026-05-072605.00380ResRL: Boosting LLM Reasoning via Negative Sample Projection Residual Reinforcement LearningZihan Lin et al.arXiv4