Skip to content

[RLlib] Fix issue with torch PPO not handling action spaces of shape=(>1,).#7398

Merged
ericl merged 1 commit intoray-project:masterfrom
sven1977:issue_7397_torch_diag_gaussian_doesnt_handle_2_actions
Mar 2, 2020
Merged

[RLlib] Fix issue with torch PPO not handling action spaces of shape=(>1,).#7398
ericl merged 1 commit intoray-project:masterfrom
sven1977:issue_7397_torch_diag_gaussian_doesnt_handle_2_actions

Conversation

@sven1977
Copy link
Copy Markdown
Contributor

@sven1977 sven1977 commented Mar 2, 2020

The kl and entropy methods of TorchDiagGaussian do not properly handle reducing over the individual action-components.

See issue #7397

Closes #7397

@sven1977 sven1977 requested a review from ericl March 2, 2020 10:25
@AmplabJenkins
Copy link
Copy Markdown

Can one of the admins verify this patch?

@AmplabJenkins
Copy link
Copy Markdown

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/Ray-PRB/22611/
Test PASSed.

@sven1977 sven1977 added the tests-ok The tagger certifies test failures are unrelated and assumes personal liability. label Mar 2, 2020
@ericl ericl self-assigned this Mar 2, 2020
@ericl ericl merged commit d8eeb96 into ray-project:master Mar 2, 2020
@sven1977 sven1977 deleted the issue_7397_torch_diag_gaussian_doesnt_handle_2_actions branch March 3, 2020 10:15
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

tests-ok The tagger certifies test failures are unrelated and assumes personal liability.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[rllib] TorchDiagGaussian doesn’t handle multiple actions correctly.

3 participants