Release 1.4.0 #260
Merged
Release 1.4.0 #260
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
* Add mixed precision option into PPO algorithm * Expand mixed precision to forward passes during data sampling phase * Group setup statements * Remove unnecessary parenthesis from autocast function
* Call agent's pre-interaction during evaluation * Increase MINOR version and update CHANGELOG
* Add utility to tensorize and flatten gymnasium spaces * Add spaces utility to docs * Add test for space utils implementations * Add funtion to sample spaces by batches * Update test file * Allow None type * Update model defintions to support different input spaces * Update model definitions to support different input spaces * Update model definitions test file * Replace view by reshape to support unclear operations * Replace gym by gymnasium in wrapper base class annotation * Use space utils to handle spaces in gymnasium-based env wrappers * Replace gym with gymnasium in DeepMind wrapper test * Add utility to convert gym spaces to gymnasium spaces * Convert gym wrapper spaces to gymnasium * Create a common test file for hypothesis' strategies * Add test for convert_gym_space space utils * Remove gymnasium.spaces.Sequence from supported spaces * Add space utility to untensorize spaces * Update Isaac Gym preview wrapper to use space utils * Update Omniverse Isaac Gym wrapper to use space utils * Use spaces utils to process actions in Gym wrapper * Use spaces utils to process actions in Gymnasium wrapper * Use spaces utils to process actions in Isaac Lab wrapper * Raise exceptions when the spaces/values are not supported * Use spaces utils to process actions in Isaac Gym preview wrapper * Use spaces utils to process actions in Omniverse Isaac Gym wrapper * Use spaces utils to process actions in PettingZoo wrapper * Use spaces utils to process actions in DeepMind wrapper * Allow to remove batch dimension when converting gym space * Use spaces utils to process actions in Brax wrapper and allow operation with other spaces * Update spaces utils section in docs * Replace each _get_space_size method call by the compute_space_size utility * Add static method to parse jax device * Add space utility in jax * Add spaces utils test in jax * Rename spaces utility test in torch * Move spaces utils implementation in torch to its own file * Replace each _get_space_size method call by the compute_space_size utility in jax * Add spaces utils API in jax to docs * Update model definitions to support different input spaces in jax * Update model instantiators definitions test file in jax * Replace gym by gymnasium in wrapper base class annotation in jax * Add parameter to handle JAX and NumPy backends * Update Gymnasium wrapper to use space utils in jax * Update Gym wrapper to use space utils in jax * Update PettingZoo wrapper to use space utils in jax * Update Brax wrapper to use space utils in jax * Update Isaac Lab wrapper to use space utils in jax * Update Omniverse Isaac Gym wrapper to use space utils in jax * Update Isaac Gym preview wrapper to use space utils in jax * Move torch-based import statements * Tensorize and flatten Isaac Lab MARL env state * Add docstring to JAX config's parse_device function * Update docs and doctrings * Cache env info in Omniverse Isaac Gym wrapper * Update CHANGELOG * Improve CHANGELOG description * Improve JAX config parse_device implementation * Use gymnsasium batch utility to sample fundamental spaces
* Remove gym from skrl.utils * Remove gym from skrl.resources * Remove gym from skrl.multi_agents * Remove gym from skrl.models * Remove gym from skrl.memories * Remove gym from skrl.envs * Remove gym from skrl.agents * Update dependencies * Remove gym from docs * Update CHANGELOG
…nt step (#208) * Fix: with SAC, a new training batch should be sampled for each gradient_step * Apply format --------- Co-authored-by: yibo di <[email protected]> Co-authored-by: Toni-SM <[email protected]>
* Move sample inside gradient step loop in TD3 (RNN), DDPG (RNN), SAC, SAC (RNN), DQN and DDQN * Update CHANGELOG.md * Apply format --------- Co-authored-by: Deniz Seven <[email protected]> Co-authored-by: Toni-SM <[email protected]>
* Add class mapping for Categorical model * Apply format --------- Co-authored-by: Toni-SM <[email protected]>
* Add new pre-commit hooks (black, codespell, among others) * Remove yapf config and update isort
* Configure pre-commit hooks * Configure black in pre-commit and pyproject.toml * Configure codespell in pre-commit * Ignore formating config sections * Apply black forma to skrl folder * Apply black format to tests folder * Apply codespell * Update CHANGELOG
* Update docs configuration * Remove __init__ automethod entry in docs
* Add multivariate Gaussian model to runner in docs * Update shared model instantiator to allow specifying its structure * Update torch runner to specify shared model structure * Define model mixin from given structure * Use spaces utils to initialize jax model state dictionary * Add support for MultivariateGaussianMixin in shared models * Update model instantiators test in torch * Remove double argument definition * Parse device in jax spaces utils * Update model instantiators test in jax * Update CHANGELOG
…#233) * Add multivariate Gaussian model to runner in docs * Update shared model instantiator to allow specifying its structure * Update torch runner to specify shared model structure * Define model mixin from given structure * Use spaces utils to initialize jax model state dictionary * Add support for MultivariateGaussianMixin in shared models * Update model instantiators test in torch * Remove double argument definition * Parse device in jax spaces utils * Update model instantiators test in jax * Speed up distribution construction in PyTorch by disabling checking
* Add method to parse device in torch * Use ML framework configuration device parsing method to parse devices * Add option to validate parsed torch device * Add ML framework testing in jax * Add torch parse_device method to ML framework docs * Update docstrings and test content * Update docs * Disable device parsing validation when PyTorch config device
…m memory (#235) * Add multivariate Gaussian model to runner in docs * Update shared model instantiator to allow specifying its structure * Update torch runner to specify shared model structure * Define model mixin from given structure * Use spaces utils to initialize jax model state dictionary * Add support for MultivariateGaussianMixin in shared models * Update model instantiators test in torch * Remove double argument definition * Parse device in jax spaces utils * Update model instantiators test in jax * Speed up distribution construction in PyTorch by disabling checking * Replace torch BatchSampler for performance issue * Add base memory class test file in torch * Update CHANGELOG
* Add reduction parameter to gaussian_model instantiator * Specify shared model parameters' default values * Update CHANGELOG * Define reduction as string value
* Update PPO mixed-precision implementation in torch * Add A2C mixed precision support in torch * Add AMP mixed precision support in torch * Add CEM mixed precision support in torch * Add DDPG mixed precision support in torch * Add DQN and DDQN mixed precision support in torch * Add RPO mixed precision support in torch * Add SAC mixed precision support in torch * Add TD3 mixed precision support in torch * Update docs * Add PPO test in torch * Add agent tests in torch * Add agent tests in jax * Avoid TypeError: Got unsupported ScalarType BFloat16 * Update CHANGELOG
…itialize models' lazy modules (#247)
…ta, and the KL Adaptive learning rate scheduler (#253)
…ts and their models (#254)
…amp deprecation warning (#256) * Deal with PyTorch automatic mixed-precision deprecation warning * Add automatic mixed precision support for multi-agent
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.
No description provided.