[RLlib] - Get and set states in MultiAgentEpisode and SingleAgentEpisode
#45012
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 <simon.zehnder@gmail.com>
Signed-off-by: Simon Zehnder <simon.zehnder@gmail.com>
Signed-off-by: Simon Zehnder <simon.zehnder@gmail.com>
Signed-off-by: Simon Zehnder <simon.zehnder@gmail.com>
Signed-off-by: Simon Zehnder <simon.zehnder@gmail.com>
Signed-off-by: Simon Zehnder <simon.zehnder@gmail.com>
…get_state' and 'from_state' to 'SingleAGentEpisode' together with test. Signed-off-by: Simon Zehnder <simon.zehnder@gmail.com>
Signed-off-by: Simon Zehnder <simon.zehnder@gmail.com>
…ith a test. Signed-off-by: Simon Zehnder <simon.zehnder@gmail.com>
Signed-off-by: Simon Zehnder <simon.zehnder@gmail.com>
sven1977
reviewed
Apr 29, 2024
| @@ -44,7 +44,7 @@ def build(self) -> None: | |||
| # Note that the KL coeff is not controlled by a Scheduler, but seeks | |||
| # to stay close to a given kl_target value in our implementation of | |||
| # `self.additional_update_for_module()`. | |||
| self.curr_kl_coeffs_per_module: Dict[ModuleID, Scheduler] = LambdaDefaultDict( | |||
| self.curr_kl_coeffs_per_module: Dict[ModuleID, TensorType] = LambdaDefaultDict( | |||
sven1977
reviewed
Apr 29, 2024
rllib/env/multi_agent_episode.py
Outdated
| @@ -1704,50 +1704,84 @@ def get_state(self) -> Dict[str, Any]: | |||
| return list( | |||
| { | |||
| "id_": self.id_, | |||
| "agent_ids": self.agent_ids, | |||
| "agent_to_module_mapping_fn": self.agent_to_module_mapping_fn, | |||
There was a problem hiding this comment.
Actually, let's make the state a dict. States should always be Dict[str, Any]. I think this is a leftover from the early DreamerV3 days :)
sven1977
reviewed
Apr 29, 2024
rllib/env/multi_agent_episode.py
Outdated
| }.items() | ||
| ) | ||
|
|
||
| @staticmethod | ||
| def from_state(state) -> None: | ||
| def from_state(state) -> "MultiAgentEpisode": |
There was a problem hiding this comment.
same here: typehint: state: Dict[str, Any]
sven1977
reviewed
Apr 29, 2024
rllib/env/multi_agent_episode.py
Outdated
| """Creates a multi-agent episode from a state dictionary. | ||
|
|
||
| See `MultiAgentEpisode.get_state()` for creating a state for | ||
| a `MultiAgentEpisode` pickable state. For recreating a | ||
| `MultiAgentEpisode` from a state, this state has to be complete, | ||
| i.e. all data must have been stored in the state. | ||
|
|
||
| Args: | ||
| state: A list of tuples containing all data required to recreate |
There was a problem hiding this comment.
same here, let's make this a dict.
sven1977
reviewed
Apr 29, 2024
rllib/env/single_agent_episode.py
Outdated
| @@ -1643,6 +1643,75 @@ def agent_steps(self) -> int: | |||
| """ | |||
| return self.env_steps() | |||
|
|
|||
| def get_state(self) -> list: | |||
sven1977
reviewed
Apr 29, 2024
| @@ -342,6 +342,24 @@ def test_sample_with_modules_to_sample(self): | |||
| # Assert that all n-steps are 1.0 as passed into `sample`. | |||
| self.assertTrue(np.all(n_steps - 1.0 < tolerance)) | |||
|
|
|||
| # def test_get_state_and_set_state(self): | |||
sven1977
approved these changes
Apr 29, 2024
There was a problem hiding this comment.
Looks good. Just needs the changes from list to dict, then can be merged. :)
Thanks @simonsays1980 !
sven1977
requested review from
avnishn,
ArturNiederfahrenhorst,
maxpumperla and
kouroshHakha
as code owners
Signed-off-by: Simon Zehnder <simon.zehnder@gmail.com>
Signed-off-by: Simon Zehnder <simon.zehnder@gmail.com>
Signed-off-by: Simon Zehnder <simon.zehnder@gmail.com>
harborn
pushed a commit
to harborn/ray
that referenced
this pull request
May 8, 2024
ryanaoleary
pushed a commit
to ryanaoleary/ray
that referenced
this pull request
Jun 7, 2024
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?
This PR adds
get_stateandfrom_statetoMultiAgentEpisodeandSingleAgentEpisode. This is needed for checkpointingEpisodeReplayBufferobjects.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.