v0.4.0

Enhancements

• Fix TypeError in pad_sequence for torch>=1.9.0 (#163)
• Make Monitor optional (#167)
• Fix for issue (#174, thanks @jfrancis71!)
• Fixed save/load problem on dqn.py (#184, thanks @jmribeiro!)

Bug fixes

• [BUGFIX] use functools.partial instead of lambda (#157, thanks @zhouzypaul!)

Documentation

• Fixes a minor typo (#161, thanks @prabhatnagarajan!)

Testing

• Fix CI error due to gym's version (#159)
• Specify the version of traitlets to fix CI error (#169)
• Fix mypy error due to numbers.Number (#171)
• Fix lint errors (#185)
• Pin pyglet==1.5.27 in CI to avoid errors (#187)

v0.3.0

Enhancements

• Add parameter to generate_exp_id() (#115)
• add episode length statistics to evaluator (#121, thanks @tkelestemur!)
• call evaluation_hooks in Evaluator (#122)
• Remove six dependency in pre-trained models (#148)

Bug fixes

• Fix an error related to collation and RNN (#129, thanks @tarokiritani!)
• fix IndexError due to empty stack (#133, thanks @lerrytang!)

Documentation

• Adds pretrained models column to README (#119)
• Adds mypy to contributing md (#125)
• Updates BibTeX to JMLR paper (#134)
• Adds link to paper (#135)
• Adds links to blog posts in the README (#137)
• Fix NormalizeActionSpace's comment (#142)

Testing

• Use latest black (#116)
• Switch to mypy>=0.800 (#127)

v0.2.1

Enhancements

• Remove deprecated SumTree (#110)

Bug fixes

• Install filelock (#113)

v0.2.0

Enhancements

• Add type hints to DQN and its dependencies (#17)
• Pretrained models (#33)
• Optuna integration & example script (2) (#46)
• Save and load a module trained with DataParallel or DistributedDataParallel (#55)
• Replace nn.functional.tanh with torch.tanh (#64)
• Makes common write_header function (#71)
• Eval stats collector (#80)
• Set stats type (#83)
• Add step_hooks and optimizer_step_hooks to actor-learner (#86)
• Adds env_stats to record_tb_stats (#98)

Bug fixes

• calculate scalar q_values to log average_q statistics in categorical dqn algorithms (#32)
• Detach greedy_actions before calling cpu() (#81)
• Fix the hang in train_agent_async with Tensorboard (#89)

Examples

• Pretrained models (#33)
• [Fix Examples] function call was duplicated (#36, thanks @gorogoroyasu!)
• Optuna integration & example script (2) (#46)

Testing

• Enable auto assignment of PR reviewers (#43)
• Enable auto review assignment for PRs from forks (#62)
• Activate auto assign only when a PR is opened (#66)
• Adopt isort to format import sentences (#82)
• Fixes flake errors (#104)

v0.1.0

Make update_counter multiprocess safe (#30)