METTA：基于神经记忆的元强化学习测试时适应方法
METTA: Memory-Enhanced Test-Time Adaptation for Meta-Reinforcement Learning
这篇论文搞了个叫 METTA 的方法，用神经记忆替代 transformer 回看历史，ML10/45 上成功率超 50%，算力开销还更低。
论文提出 Memory-Enhanced Test-Time Adaptation（METTA），将元强化学习的快速适应拆成两个时间尺度：因果神经记忆在单条 episode 内做细粒度适应，复杂度为线性；episode 级摘要则支持跨 episode 的任务推断，避免对完整交互历史做注意力计算。在未见过的 Meta-World ML10/45 测试环境中，METTA 成功率超过 50%，训练预算和 wall-clock 时间保持较低水平。在长上下文稀疏奖励的 PointMaze 任务上，其表现优于基于 transformer 的基线，并展现更有结构的探索行为。
METTA: Memory-Enhanced Test-Time Adaptation for Meta-Reinforcement Learning
Meta-reinforcement learning aims to train agents that adapt quickly to unseen tasks from limited interaction, but this requires representing task evidence over long interaction histories. Classical recurrent agents compress experience into hidden states, while transformer-based agents retrieve from past trajectories at quadratic memory and computation cost. In this work, we propose Memory-Enhanced Test-Time Adaptation (METTA) for meta-reinforcement learning. It decomposes the fast adaptation phase into two temporal scales: a causal neural memory performs fine-grained within-episode adaptation with linear sequence complexity, while compact episode-level summaries support long-range cross-episode task inference without attending over the full interaction history. Moreover, equipping agents with neural memory provides sufficient expressive capacity and test-time training capability, enabling effective adaptation to unseen held-out tasks and task families at meta-test time. On unseen held-out Meta-World ML10/45 test environments, METTA achieves a success rate over $50\%$ while maintaining a low training budget and short wall-clock time. On long-context sparse-reward PointMaze tasks, it outperforms transformer-based baselines by exhibiting more structured exploration behavior. These results show that METTA provides an efficient path toward more adaptive and scalable long-context meta-RL.
