一种通过结果驱动代理推理卸载大模型内存检索的新方法，显著降低推理开销并提升知识利用效率。
arXiv:2603.03379v2 Announce Type: replace-cross Abstract: As Large Language Models (LLMs) are increasingly used for long-duration tasks, maintaining e…
突破传统KV缓存局限，专门针对Agentic LLM的推理优化，引入策略驱动编辑的新范式。
arXiv:2606.01065v1 Announce Type: cross Abstract: Modern KV cache management assumes the chatbot workload: prompts arrive once and the cache grows app…
发现强化学习下LLM智能体在主动推理中可能出现信息自锁现象，为对齐与推理研究提供新视角。
arXiv:2603.12109v2 Announce Type: replace Abstract: Reinforcement learning (RL) has become a de facto paradigm for building LLM-based agents that act,…