Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[RLlib] Fix MultiAgentEpisode getter bugs. #44898

Merged
merged 3 commits into from
Apr 23, 2024
Merged

[RLlib] Fix MultiAgentEpisode getter bugs. #44898

merged 3 commits into from
Apr 23, 2024

Conversation

sven1977
Copy link
Contributor

@sven1977 sven1977 commented Apr 22, 2024 •
edited
Loading

Fix MultiAgentEpisode getter bugs.

  • Improved MultiAgentEpisode test cases.

e.g.:

        observations = [
            {"a0": 0, "a1": 0},
            {"a0": 1, "a1": 1},
            {"a1": 2},
            {"a1": 3},
            {"a1": 4},
        ]
        actions = observations[:-1]
        rewards = observations[:-1]
        episode = MultiAgentEpisode(
            observations=observations, actions=actions, rewards=rewards, len_lookback_buffer="auto"
        )
        episode.get_actions(0)  # <- should result in an index error (b/c all actions are in the lookback buffer by default), but didn't

Why are these changes needed?

Related issue number

Checks

  • I've signed off every commit(by using the -s flag, i.e., git commit -s) in this PR.
  • I've run scripts/format.sh to lint the changes in this PR.
  • I've included any doc changes needed for https://docs.ray.io/en/master/.
    • I've added any new APIs to the API Reference. For example, if I added a
      method in Tune, I've added it in doc/source/tune/api/ under the
      corresponding .rst file.
  • I've made sure the tests are passing. Note that there might be a few flaky tests, see the recent failures at https://flakey-tests.ray.io/
  • Testing Strategy
    • Unit tests
    • Release tests
    • This PR is not tested :(

@sven1977
wip
5384dff 
Signed-off-by: sven1977 <[email protected]>
@sven1977
wip
a61cb72 
Signed-off-by: sven1977 <[email protected]>
@sven1977 sven1977 marked this pull request as ready for review April 22, 2024 13:22
@sven1977 sven1977 added the tests-ok The tagger certifies test failures are unrelated and assumes personal liability. label Apr 22, 2024
sven1977
sven1977 commented Apr 23, 2024
View reviewed changes
rllib/env/utils/infinite_lookback_buffer.py
@@ -464,7 +464,7 @@ def _get_int_index(
# If index >= 0 -> Ignore lookback buffer.
# Otherwise, include lookback buffer.
if idx >= 0 or neg_indices_left_of_zero:
idx = self.lookback + idx - (_ignore_last_ts is True)
idx = self.lookback + idx
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is the actual bug fix.

sven1977
sven1977 commented Apr 23, 2024
View reviewed changes
rllib/env/tests/test_multi_agent_episode.py
@@ -1023,6 +1023,50 @@ def test_get_actions(self):
act = episode.get_actions(-4, env_steps=False, fill="skip")
check(act, {"a0": "skip", "a1": 0})

episode.add_env_step(
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Added this to the tests. Mostly to figure out, whether a hanging action at the edge of the episode or further back would make a difference in get_actions(-1).

sven1977
sven1977 commented Apr 23, 2024
View reviewed changes
rllib/env/tests/test_multi_agent_episode.py
@@ -975,7 +975,7 @@ def test_get_actions(self):
check(act, actions[i])
# Access >=0 integer indices (expect index error as everything is in
# lookback buffer).
for i in range(1, 5):
for i in range(0, 5):
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

idx=0 was NOT working properly before this fix.

sven1977
sven1977 commented Apr 23, 2024
View reviewed changes
rllib/env/multi_agent_episode.py
@@ -1633,21 +1633,30 @@ def __repr__(self):
)

def print(self) -> None:
"""Prints this MultiAgentEpisode as a table of observations for the agents."""
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Made this a little nicer. :)

simonsays1980
simonsays1980 approved these changes Apr 23, 2024
View reviewed changes
Copy link
Collaborator

@simonsays1980 simonsays1980 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM. All tests passed for the MAERB, too. This was the remaining part in the puzzle. Awesome!

rllib/env/utils/infinite_lookback_buffer.py
@@ -464,7 +464,7 @@ def _get_int_index(
# If index >= 0 -> Ignore lookback buffer.
# Otherwise, include lookback buffer.
if idx >= 0 or neg_indices_left_of_zero:
idx = self.lookback + idx - (_ignore_last_ts is True)
idx = self.lookback + idx
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Amazing how such a small modification changes the landscape completely.

@sven1977 sven1977 merged commit 930f31a into ray-project:master Apr 23, 2024
5 checks passed
@sven1977 sven1977 deleted the fix_multi_agent_episode_getter_bugs branch April 23, 2024 10:42
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@simonsays1980 simonsays1980 simonsays1980 approved these changes

@avnishn avnishn Awaiting requested review from avnishn

@ArturNiederfahrenhorst ArturNiederfahrenhorst Awaiting requested review from ArturNiederfahrenhorst

@maxpumperla maxpumperla Awaiting requested review from maxpumperla

@kouroshHakha kouroshHakha Awaiting requested review from kouroshHakha

Assignees

@simonsays1980 simonsays1980

Labels
tests-ok The tagger certifies test failures are unrelated and assumes personal liability.
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@sven1977 @simonsays1980