Recent work has explored optimizing LLM collaboration through Multi-Agent Reinforcement Learning (MARL), but many approaches still rely on predefined workflows, limiting efficiency and scalability. Decentralized LLM collaboration is challenging because agents execute independently with limited communication, share a jo...
Tian-Le Chen, Shuo Liu, Christopher Amato· 0 citations
On-policy distillation (OPD) trains agents on their own trajectories using supervision from a stronger teacher. We uncover a counterintuitive phenomenon in multi-turn agentic tasks: a smaller, initially weaker student can substantially outperform its teacher during training, only to regress as distillation continues. W...
Ji-Ke Zhong, Ming Li, Yu-Xiang Lai et al.· 0 citations
Language-model agents that run long tasks often redo work they have already done: they re-issue a tool call they made earlier and get back exactly the same observation, a step that adds no new information. We identify this failure as a loop. Loops appear natively in Qwen, Llama, and Gemini agents and become far more co...
Tian-Le Chen, P. Singh, K. Todi et al.· 1 citation
On-policy distillation (OPD) provides dense token-level supervision on student rollouts, but this supervision can become unreliable when the student and teacher diverge, either because of a large initial policy gap or compounding errors in long-horizon reasoning and tool use. We introduce Divergence-Triggered Dynamic D...
Qi-Jia He, Ze-Qun Li, S.-S. Luo et al.· 0 citations
This work instantiates MAPL using MARL from human feedback with a learned reward model and multi-agent direct preference optimization (MADPO), and shows that MAPL can improve collaboration quality and efficiency while approaching the performance of MARL with fixed, well-defined rewards.
Counteraction-Aware Multi-Teacher On-Policy Distillation is proposed, which addresses issues with decoupled alternating training and gap-based sample selection and outperforms all baselines in general capability recovery while maintaining domain-specific behavior.
Tian-Le Chen, Jiao Ou, Zi-Yuan Liu et al.· arXiv.org· 13 citations· ⚡1
ReCo (Reward-Coordinated Compression), a step-wise framework in which a lightweight process-reward estimator scores each completed step and drives three components: reward-adaptive KV-cache compression that shrinks the retained cache harder at high-reward steps and less at low-reward ones, and a confidence-based early...
Qi-Yuan Zhu, De-Zhi Li, Pengyu Cheng et al.· 1 citation
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.