Large Scale Deep Reinforcement Learning and Key Features

  • RL Framework
  • Support off-policy and on-policy multi-agent algorithms.
  • Support multiple MARL environments.
  • Switching between single host and multiple hosts with zero code changes.
  • SOTA performances on SMAC and Google Research Football.
  • Try it out!
  • TLeague
  • Self-play league system.
  • Multiple players combat with gRPC communication.
  • Flask based League website and database server supports.
  • Try it out!

© 2021 TARTRL All Rights Reserved