[RLlib] Docs do-over (new API stack): Prep. RLModule; introduce Default[algo]RLModule classes (rename from [algo]RLModule).
#49366
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
sven1977
changed the title
Default[algo]RLModule classes (rename from [algo]RLModule).Default[algo]RLModule classes (rename from [algo]RLModule).
sven1977
added
rllib
simonsays1980
approved these changes
Dec 19, 2024
| ) | ||
|
|
||
|
|
||
| class DefaultAPPORLModule(DefaultPPORLModule, TargetNetworkAPI, abc.ABC): |
There was a problem hiding this comment.
I like this default naming. Do we apply it to all algorithms now? That would be great!
There was a problem hiding this comment.
Yeah, exactly, that's the idea! Only DQN and SAC missing, I think ...
There was a problem hiding this comment.
DQN and the rest will be done in the next PR (already in review ...)
|
|
||
| return RLModuleSpec( | ||
| module_class=BCTorchRLModule, | ||
| module_class=DefaultBCTorchRLModule, |
| @@ -119,11 +119,6 @@ def forward_target(self, batch: Dict[str, Any]) -> Dict[str, Any]: | |||
| ), | |||
| ) | |||
|
|
|||
| # TODO (simon): DQN Rainbow does not support RNNs, yet. | |||
There was a problem hiding this comment.
This comes with the sequence sampling PR. Have to finish it.
| self.encoder = self.catalog.build_actor_critic_encoder(framework=self.framework) | ||
| self.pi = self.catalog.build_pi_head(framework=self.framework) | ||
| self.vf = self.catalog.build_vf_head(framework=self.framework) | ||
| # __sphinx_doc_end__ |
There was a problem hiding this comment.
Do we still need this sphinx logic here? All other modules don't have it.
There was a problem hiding this comment.
So far yes. A follow up PR will redo the rst files and the example codes. WIP ...
| @@ -69,8 +69,9 @@ def __init__(self, config: AlgorithmConfig): | |||
| self.filesystem_object = self.filesystem | |||
| elif self.filesystem is not None: | |||
| raise ValueError( | |||
| f"Unknown filesystem: {self.filesystem}. Filesystems can be " | |||
| "'gcs' for GCS, 's3' for S3, or 'abs'" | |||
| f"Unknown `config.input_filesystem` {self.filesystem}! Filesystems " | |||
…_redo_rl_module_001_default_rl_modules
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.
Introduce
Default[algo]RLModuleclasses (rename from[algo]RLModule).ValueFunctionAPI. The user does NOT have to subclass a PPO-specific superclass anymore freeing them from hidden/confusing/underlying implementation details, such asWhy are these changes needed?
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.