Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

DCG-MAP-Elites #167

Merged

Conversation

maxencefaldor
Copy link
Member

Add DCG-MAP-Elites algorithm to QDax.

This PR introduces:

  • a new method for MAP-Elites repertoire, that enables to samples individuals with their corresponding descriptors.
  • a new output extra_info for Emitter.emit methods that is similar to the extra_scores of the scoring function, and that enables to pass information from the emit step to the state_update (necessary for DCG-MAP-Elites).
  • a new DCGTransition that add desc and desc_prime to the QDTransition.
  • descriptor-conditioned TD3 loss, descriptor-conditioned scoring functions, descriptor-conditioned MLP
  • two new reward wrappers to clip and offset the reward (necessary for DCG-MAP-Elites).

@Lookatator Lookatator merged commit b4125c3 into adaptive-intelligent-robotics:develop Jan 9, 2024
5 checks passed
@Lookatator Lookatator mentioned this pull request Jun 10, 2024
4 tasks
Lookatator added a commit that referenced this pull request Sep 6, 2024
* feat(algo): add DCG-MAP-Elites whose most recent version is called DCRL-ME (#167)
* fix: Fix pareto dominance definition to account for solutions which have the same fitness values along one axis (#174)
* docs(contribution): clarify contribution process (#171)
* feat: Upgrade Library Versions and Python Version (#187)

---------

Co-authored-by: Maxence Faldor <[email protected]>
Co-authored-by: Hannah Janmohamed <[email protected]>
Co-authored-by: Felix Chalumeau <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants