[gym_jiminy/rllib] Restore curriculum state from checkpoints. #878
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
Owner
duburcqa
commented
Feb 3, 2025
duburcqa
commented
- [gym_jiminy/rllib] Make callbacks checkpointable.
- [gym_jiminy/rllib] Restore curriculum state from checkpoints.
…state from checkpoints.
Merged
duburcqa
added a commit
that referenced
this pull request
Feb 9, 2025
* [python/dynamics] Fix trajectory interpolation mode 'clip'. (#873) * [python/dynamics] Fix bug making 'Trajectory.get' extremely inefficient. (#881) * [gym_jiminy/common] Fix 'MechanicalPowerConsumption' not exposed. (#870) * [gym_jiminy/common] Add support of 'horizon=None' to MechanicalPowerConsumptionTermination. (#870) * [gym_jiminy/common] Various bug fixes in 'play_interactive'. (#871) * [gym_jiminy/common] Fix quantity caching issues. (#873) * [gym_jiminy/common] Fix mechanical power termination condition. (#873) * [gym_jiminy/common] Fix 'BaseRelativeHeight' quantity ignoring mode. (#873) * [gym_jiminy/common] Remove 'post_fn' argument from drift termination cond. (#873) * [gym_jiminy/common] Add mechanical power consumption reward. (#873) * [gym_jiminy/common] Argument 'horizon' of 'BaseJiminyEnv.evaluate' is now in seconds. (#874) * [gym_jiminy/common] 'TrackingBaseHeightReward' now relies on relative base height rather than absolute. (#874) * [gym_jiminy/common] Rename 'ReferencePositionWithTrueOdometryPose' in 'ReferencePositionVector' and add option to disable odometry pose override. (#874) * [gym_jiminy/common] Add generic 'DeltaQuantity'. (#874) * [gym_jiminy/common] Add 'DeltaBaseOdometryPosition' and 'DeltaBaseOdometryOrientation' quantities. (#874) * [gym_jiminy/common] Add 'DriftTrackingBaseOdometryPoseReward'. (#874) * [gym_jiminy/common] Inline 'InterfaceQuantity.refresh' if possible for efficiency. (#874) * [gym_jiminy/common] Fix 'DriftTrackingBaseOdometryPoseReward' so as to decouple position and orientation errors. (#875) * [gym_jiminy/common] Make sensor space bounds infinite to avoid out-of-bounds measurements due to noise/bias to trigger termination. (#875) * [gym_jiminy/common] Add support of scalar concatenation to 'ConcatenatedQuantity'. (#875) * [gym_jiminy/common] Move PD deadband from PDController to PDAdapter to fix position/velocity inconsistency. (#876) * [gym_jiminy/common] Fix deformation observer with backlash enabled. (#876) * [gym_jiminy/common] Remove LRU cache for trajectory state getter as it never hits in practice. (#877) * [gym_jiminy/common] Fix cache auto-refresh edge-case. (#880) * [gym_jiminy/common] Fix stack length computation from horizon. (#880) * [gym_jiminy/common] Refactor 'DeltaQuantity' to improve performance for long horizon. (#880) * [gym_jiminy/common] Speedup dataset trajectory select. (#882) * [gym_jiminy/common] Fix cache auto-refresh once again. (#882) * [gym_jiminy/common] Fix 'DeltaQuantity' equality. (#882) * [gym_jiminy/common] Speed up quantity reset by avoiding redundant state computations. (#883) * [gym_jiminy/rllib] Rename 'build_eval_runner_from_checkpoint' in 'build_runner_from_checkpoint'. (#870) * [gym_jiminy/toolbox] Do NOT raise exception if score is out-of-bounds. (#871) * [gym_jiminy/toolbox|rllib] Implement task score at scheduler level rather than task settable wrapper level. (#876) * [gym_jiminy/rllib] Fix eval runner connector stats not sync w/ eval. (#870) * [gym_jiminy/rllib] Fix eval runner connector updated during sample collection. (#870) * [gym_jiminy/rllib] Fix connector state not properly restored after loading checkpoint. (#871) * [gym_jiminy/rllib] Add option to override env config when restoring runners. (#872) * [gym_jiminy/rllib] Disable callbacks for evaluation runners by default. (#875) * [gym_jiminy/rllib] Fix log paths not deleted during evaluation. (#877) * [gym_jiminy/rllib] Add option to 'evaluate_from_runner' to delete log files automatically. (#877) * [gym_jiminy/rllib] Make callbacks checkpointable. (#878) * [gym_jiminy/rllib] Restore curriculum state from checkpoints. (#878) * [misc] Add pipeline benchmark script. (#870) * [misc] Fix Sphinx doc (github-pages) broken refs. (#870) * [misc] Fix stub generation for Numpy>=2.X. (#873) * [misc] Fix RL tutorial notebook. (#880)
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.