search

IntelLabs / coach

Reinforcement Learning Coach by Intel AI Lab enables easy experimentation with state of the art Reinforcement Learning algorithms
https://intellabs.github.io/coach/
Apache License 2.0
2.32k stars 460 forks source link

DDPG+HER: Multi-threaded or Multi-Node the way to go? #409

Closed TobiKanone closed 4 years ago

TobiKanone commented 4 years ago

Hello everyone,

thank you for providing this awesome framework! We are new to Coach and we are trying to train DDPG+HER on our own environment but we are unsure which is the way to go for us. Which way to train the agent do you recommend for this algorithm? Multi-threaded or Multi-Node? Any opinion is appreciated! Thank you!

galnov commented 4 years ago

Hi, you can try single or multi-threaded which have been tested on various Fetch tasks as you can see here.