suqi/gym-sandbox

Issues

• Use matplotlib to replace bokeh

  #21 opened 8 years ago by suqi
  0

• feat(test_algos): separate render worker and training worker

  #12 opened 8 years ago by suqi
  1

• support MA!

  #5 opened 8 years ago by suqi
  2

• todo: support continuous action

  #4 opened 8 years ago by suqi
  1

• doc: add our objective of each standard env in std_env_list

  #14 opened 8 years ago by suqi
  1

• track human performance: add keyboard interaction

  #17 opened 8 years ago by suqi
  2

• add implementation example of state-of-the-art algorithms

  #18 opened 8 years ago by suqi
  2

• reproduce HRA paper

  #20 opened 8 years ago by suqi
  1

• refactor: performance wrappers and basic env logic can be extracted to a base class

  #19 opened 8 years ago by suqi
  1

• game engine keep a object instead of just position

  #16 opened 8 years ago by suqi
  0

• refactor: make it more friendly to make new env other than Police

  #15 opened 8 years ago by suqi
  0

• PR: make it public to gym community

  #13 opened 8 years ago by suqi
  0

• Remove dependency of Jupyter notebook, so that can run in CLI

  #9 opened 8 years ago by suqi
  1

• feat(multi-task): add team of destroyer, which police must keep away

  #10 opened 8 years ago by suqi
  1

• Use bokeh server for render to support non-jupyter

  #2 opened 8 years ago by suqi
  1

• todo: state normalize

  #3 opened 8 years ago by suqi
  1

• refactor: continuously refactor repeated code of all env

  #7 opened 8 years ago by suqi
  0

• feat: add more randomness

  #11 opened 8 years ago by suqi
  1

• style: translate all chinese comments to be a international project!

  #6 opened 8 years ago by suqi
  1

• refactor: split render into sub-modules, like balls, maps, etc

  #8 opened 8 years ago by suqi
  0

• add test algorithm demo to showcase the interact api with RL algos

  #1 opened 8 years ago by suqi
  1