League Training for SlimeVolleyball Env

[PaParaZz1]

• I have marked all applicable categories:
  • exception-raising bug
  • RL algorithm bug
  • system worker bug
  • system utils bug
  • code design/refactor
  • documentation request
  • new feature request
• I have visited the readme and doc
• I have searched through the issue tracker and pr tracker
• I have mentioned version numbers, operating system and environment, where applicable:

  import ding, torch, sys
  print(ding.__version__, torch.__version__, sys.version, sys.platform)

Overview

Add league training pipeline in slime_volleyball environment, and make better performance than self-play results (#23)
Related Discussion: #61

TODO

• league pipeline
• mutate from pre-trained model (vs bot)
• policy behavior analysis

[PaParaZz1]

we will restart this demo in February 28th, and we expect to finish it in a week.

[zxzzz0]

@PaParaZz1 Could we have some updates on this? It's been more than 8 weeks.

[zxzzz0]

we will restart this demo in February 28th, and we expect to finish it in a week.

@PaParaZz1 it's been 6 months, do we complete the tasks below?

• league pipeline
• mutate from pre-trained model (vs bot)
• policy behavior analysis