Fraser McGhan

icon indicating a website link frasermcghan.com

icon company Robot Perception and Learning Group (RoPAL) icon location Cardiff, Wales icon location Deep RL & Robotics

Results 3 comments of Fraser McGhan

ARS Returns 0 actions on evaluation env

I am seeing similar behaviour with the RecurrentPPO implementation. I believe the issue comes from ```model.load()```. I am using a custom environment but I have adapted the RecurrentPPO example to...

ARS Returns 0 actions on evaluation env

Apologies, just found this issue DLR-RM/stable-baselines3#683. The correct way to load a saved model is ```model = RecurrentPPO.load()``` - I should have read the docs more closely! :)

Loading GPU trained RPPO on CPU

I solved this by downgrading numpy. https://github.com/ray-project/ray/issues/31293#issuecomment-1364667180