Skip to content

rerz/rltest

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

marl with composite actions

this small example is meant to illustrate my struggles and attempts at adapting torchrl to multi agent ppo with composite action spaces.

my pain points were:

  • individual log prob keys for the actions cause issues with stacking tensordicts somewhere internally
  • non-natively multivariate distributions need special handling when calculating log probs
  • ppo loss does not deal well with tensordicts, need to extract the sample log prob tensor
  • petting zoo env wrapper does not properly split the action tensors when using dict action spaces

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages