[RLlib] - Fix APPO RLModule inference-only problems. #45111
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: Simon Zehnder <[email protected]>
sven1977
reviewed
May 2, 2024
…rence_only_state_dict_keys'. Modified 'PPOTorchRLModuel' to not register already keys from APPO. Signed-off-by: Simon Zehnder <[email protected]>
simonsays1980
requested review from
avnishn,
ArturNiederfahrenhorst,
maxpumperla and
kouroshHakha
as code owners
simonsays1980
changed the title
…instead. Signed-off-by: Simon Zehnder <[email protected]>
sven1977
reviewed
May 3, 2024
| @@ -15,19 +16,7 @@ | |||
| _, tf, _ = try_import_tf() | |||
|
|
|||
|
|
|||
| class APPOTfRLModule(PPOTfRLModule, RLModuleWithTargetNetworksInterface): | |||
| def setup(self): | |||
sven1977
reviewed
May 3, 2024
| @@ -14,20 +15,9 @@ | |||
| from ray.rllib.utils.nested_dict import NestedDict | |||
|
|
|||
|
|
|||
| class APPOTorchRLModule(PPOTorchRLModule, RLModuleWithTargetNetworksInterface): | |||
| @override(PPOTorchRLModule) | |||
sven1977
reviewed
May 3, 2024
| @@ -42,8 +32,26 @@ def output_specs_train(self) -> List[str]: | |||
|
|
|||
| @override(PPOTorchRLModule) | |||
| def _forward_train(self, batch: NestedDict): | |||
| if self.inference_only: | |||
There was a problem hiding this comment.
nit: Maybe we should move this error into the parent RLModule class' main method?
def forward_train(self, ...):?
sven1977
approved these changes
May 3, 2024
There was a problem hiding this comment.
LGTM! Just one nit/question about moving an error further up the inheritance chain to have all future sub-classes benefit from this logic.
Thanks @simonsays1980 ! :)
…les. Signed-off-by: Simon Zehnder <[email protected]>
…ent already does it. Signed-off-by: Simon Zehnder <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Why are these changes needed?
The APPO algorithm could not deal with
inference-only, yet. This change does enableinference-onlymodules for APPO. In addition to the_inference_only_state_dict_keysfrom PPO, APPO needs to remove the target networks when in inference.For this to work properly a parent
RLModuleis needed that triggers the building of the_inference_only_state_dict_keys. Due to inheritance from the PPO module polymorphism did not work out of the box b/c this parent was missing. This parent is added to APPO in this PR as well, namedAPPORLModule.Related issue number
Checks
git commit -s) in this PR.scripts/format.shto lint the changes in this PR.method in Tune, I've added it in
doc/source/tune/api/under thecorresponding
.rstfile.