论文地址:
https://arxiv.org/pdf/1802.01561v2.pdf
论文《IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures》是基于论文《Safe and efficient off-policy reinforcement learning》改进后的分布式版本,基础论文《Safe and efficient off-policy reinforcement learning》的地址为:
https://arxiv.org/pdf/1606.02647.pdf
相关资料:
Deepmind Lab环境的python扩展库的安装:
https://www.cnblogs.com/devilmaycry812839668/p/16750126.html
读论文《IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures》
=========================================
官方的代码地址:(现已无法运行)
https://gitee.com/devilmaycry812839668/scalable_agent
需要注意的一点是这个offical的代码由于多年无人维护,现在已经无法运行,只做留档之用。
=========================================
标签:Weighted,Importance,论文,Learner,Scalable,Actor,Architectures From: https://www.cnblogs.com/devilmaycry812839668/p/16782564.html