Skip to content

v0.0.1 (Alpha) - 2024-12-08

๐Ÿš€ Features

  • (policy) Added decay rate methods to EpsilonPolicy.
  • (analytics) Added W&B Analytics class.
  • (models) Added family of Sarsa models.
  • (sarsa) Added disable_logging flag.
  • (controller) Added logic to RLController.
  • (gymhandler) Added core logic for GymEnvHandler.
  • (metrics) Added Metric storage and logging.
  • (plots) Added plot_vf method to controller.

๐Ÿ› Bug Fixes

  • (policy) Fixed epsilon exponential decay method.
  • (metrics) Update metric implementation to simplify and fix bugs.

๐Ÿšœ Refactor

  • (env) Simplified env wrapper implementation.
  • (rollouts) Added device validation.
  • (utils) Refactored utils file to dir for simplicity
  • (vfuncs) QOL changes to V, Q for simplicity
  • (config) Updated Config settings to simplify.
  • (agent) Updated policy, models, and config to simplify.
  • (policy) Updated method argument name for clarity.
  • (policy) Updated method names for clarity.
  • (config) Updated settings for usability.
  • (rollouts) Simplified agent storage methods.
  • (mypy) Global updates to fix mypy errors.
  • (agent) Modified Agent models to generalised form to simplify.
  • (sarsa) Simplified Sarsa model implementation.

โšก Performance

  • (storage) Added action, reward tensor methods
  • (returns) Refactored returns method for batching

๐Ÿงช Testing

  • (agent) Added EpsilonPolicy, BaseSarsa tests
  • (sarsa) Added unit tests for Sarsa agents.
  • (qtable) Add missing QTable tests.
  • (analytics) Added missing Analytics tests.
  • (controller) Updated and added unit tests.

โš™๏ธ Miscellaneous Tasks

  • (analytics) Updated docstrings for analytics.
  • (utils) Added utility method ignore_empty_dicts
  • (tests) Added automated testing packages.
  • (examples) Removed old examples.
  • (controller) Add controller example.