[RLlib] Issue 21334: Fix APPO when kl_loss is enabled. #21855
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
Bug is with learner_info construction in our LearnerThread. This only shows up for APPO because APPO, IMPALA, and APEX are the only ones that use async Learnthread, while APPO is the only agent that updates kl loss.
added 2 commits
January 25, 2022 23:26
sven1977
reviewed
Jan 26, 2022
rllib/agents/ppo/tests/test_appo.py
Outdated
| @@ -47,6 +47,25 @@ def test_appo_compilation(self): | |||
| check_compute_single_action(trainer) | |||
| trainer.stop() | |||
|
|
|||
| def test_appo_compilation_use_kl_loss(self): | |||
| """Test whether an APPOTrainer can be built with both frameworks.""" | |||
sven1977
reviewed
Jan 26, 2022
rllib/agents/ppo/tests/test_appo.py
Outdated
| num_iterations = 2 | ||
|
|
||
| for _ in framework_iterator(config, with_eager_tracing=True): | ||
| print("w/ v-trace") |
There was a problem hiding this comment.
right, I got rid of it.
sven1977
reviewed
Jan 26, 2022
rllib/agents/ppo/tests/test_appo.py
Outdated
| for _ in framework_iterator(config, with_eager_tracing=True): | ||
| print("w/ v-trace") | ||
| _config = config.copy() | ||
| _config["vtrace"] = True |
sven1977
reviewed
Jan 26, 2022
|
|
||
| if released: | ||
| self.idle_tower_stacks.put(buffer_idx) | ||
|
|
||
| self.outqueue.put((get_num_samples_loaded_into_buffer, learner_stats)) | ||
| self.outqueue.put((get_num_samples_loaded_into_buffer, |
sven1977
approved these changes
Jan 26, 2022
|
Sorry, saw the issue # now. |
sven1977
changed the title
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Outdated suggestions cannot be applied.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?
Fix APPO agent when kl_loss is enabled.
value is now saved under different policy id keys. also we need to torch_mean() the stats for the torch policy.
Related issue number
Closes #21334
Checks
scripts/format.shto lint the changes in this PR.