Agentic reinforcement learning (RL) disaggregates training from rollout, so each policy update must reach the rollout clusters before the next batch. Transferring a full 1T checkpoint for such weight synchronization (refit) takes 87.5 min between two AWS regions. Measurements of BF16 training show that about 1% of weig...
Songlin Jiang, Zhi-Yu Li, Terry Kong et al.· 0 citations
Results show that an intermediate decision layer that plans, searches, filters evidence, and determines when to stop can improve both long-term memory retrieval and knowledge-intensive question answering.
Cascade is proposed, a hierarchical recoverability control framework that minimizes the internal identifiability of target knowledge and effectively reduces recoverability while maintaining stable model utility.
Qing-Chen Yu, Shi-Ying Duan, Xiao-Dong Li et al.· 0 citations
The concept of collective-individual memory synergy is introduced and CoMem is proposed, an architecture that unifies both private experience and shared knowledge for multi-agent learning.
Cheng-Xin Yu, Zhao-Xin Fan, Fa-Guo Wu et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
