In this PR we have added the possibility to better control all logging and checkpointing (for performance reasons). In particular:
The user can now control what and how to log directly from the configs. Moreover, every algorithm specifies what it will be logged by default in the set AGGREGATOR_KEYS defined under the sheeprl/algos/{algorithm}/utils.py: only the metrics present in both will be logged at runtime
The user can now specify whether to save or not the last checkpoint at the end of the training
A new howto on how to control logging and checkpoint has been added
Type of Change
Please select the one relevant option below:
New feature (non-breaking change that adds functionality)
Checklist
Please confirm that the following tasks have been completed:
[x] I have tested my changes locally and they work as expected. (Please describe the tests you performed.)
[x] I have added unit tests for my changes, or updated existing tests if necessary.
[x] I have updated the documentation, if applicable.
[x] I have installed pre-commit and run locally for my code changes.
Thank you for your contribution! Once you have filled out this template, please ensure that you have assigned the appropriate reviewers and that all tests have passed.
Summary
In this PR we have added the possibility to better control all logging and checkpointing (for performance reasons). In particular:
AGGREGATOR_KEYS
defined under thesheeprl/algos/{algorithm}/utils.py
: only the metrics present in both will be logged at runtimeType of Change
Please select the one relevant option below:
Checklist
Please confirm that the following tasks have been completed:
Thank you for your contribution! Once you have filled out this template, please ensure that you have assigned the appropriate reviewers and that all tests have passed.