search

thu-ml / tianshou

An elegant PyTorch deep reinforcement learning library.
https://tianshou.org
MIT License
7.71k stars 1.11k forks source link

MPO Implementation #1165

Open ziqiao30 opened 1 month ago

ziqiao30 commented 1 month ago

Hi there,

Will you consider implementing MPO in the near future? If I want to add a PBT for hyperparameter tuning, what would you suggest me to do?

Best regards, ziqiao

MischaPanch commented 1 month ago

Hi. Yes, that would be on our list, it's a fairly standard algorithm (though it doesn't seem to be of much practical use from what I've seen). We were generally considering to only add new algorithms after the 2.0 release, where some core algorithm abstractions would be refactored, but for MPO we might make an exception. I'll discuss it with the other contributors and will come back to you soon.