Meta 论文提出 Routed CL：按任务类型分流记忆，智能体学习效率提升 15%-30%
Meta 这篇论文说清了一件事：让 agent 用一条共享历史记笔记反而拖后腿。按问题类型分簇记忆，成绩提 15%-30%，API 费用省七成，做 agent 的都该看看。
Meta 在论文中比较了智能体记忆机制的两种方案。传统做法把所有历史放进一条共享记录，在 398 个混合任务上只比完全不用记忆高出 0.015（满分 1）。Routed CL 在每次运行前由 LLM 路由器决定复用某个记忆簇或新建一簇，每个簇从 5 种记忆类型（如原始运行记录、蒸馏后的经验）中选一种。跨 4 个模型的测试中，Routed CL 比共享历史高出最大可能得分的 15% 到 30%，同时将 API 成本降低 65% 到 72%。
New Meta paper finds that AI agents improve more from their past attempts when each new problem is matched to notes from similar problems, instead of one shared record of everything they have done.
A simple way to let agents learn from past runs is one shared history. With mixed tasks, it gets costly and mixes unrelated experience.
Routed CL adds an LLM router that reuses a memory cluster or starts a new one before each run. Each cluster uses 1 of 5 memory types, such as raw past runs or distilled lessons.
Across 398 mixed problems and 4 models, the shared history barely beat no memory, with a mean gain of 0.015 out of 1. Routed CL beat it by 15% to 30% of the maximum possible score and cut estimated API cost by 65% to 72%.
If your agents handle mixed task types, split their memory into separate pools by problem type instead of sharing one history.
