google / deluca

Performant, differentiable reinforcement learning
https://deluca.fyi
Apache License 2.0
121 stars 17 forks source link

Changes to _adaptive.py #66

Open simran135 opened 1 year ago

simran135 commented 1 year ago

Hello! I made some modifications to AdaGPC (in _adaptive.py). In the existing implementation, GPC outperforms AdaGPC in the known LDS setting, which is the opposite of what one should expect. Based on some preliminary experiments, I believe AdaGPC is now working properly (at least in the known dynamics version). (I also made some miscellaneous changes in other files, e.g., to the imports in some of the agent files -- I think there might have been some file restructuring across different versions of deluca, but the imports were not updated to reflect this change, causing some errors at runtime.) Please let me know if you have any questions/concerns. Thanks!

google-cla[bot] commented 1 year ago

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.