RIIT: Rethinking the Importance of Implementation Tricks in Multi-Agent Reinforcement Learning 笔记 其他 2021-11-30 01:43 0 阅读 文章目录 前言 Importance of Tricks Optimization Eligibility Traces Replay Buffer Size Rollout Process Number Exploration Steps Reward Shaping Tricks Overall Impact Rethinking and Analysis Our Benchmarks Albation Study of Monotonicity Theoretical Analysis 前言 这篇文章很不错，其指出了当前很多值分解类方法所谓的改进其实名 猜你喜欢