- Record state during testing (with option to record full trajectories) - Add greedy flag to RL agent action sampling method
0.2.1
Restructure how RL-agent state and functionality is implemented.
State and agent update functionality are now separated into separate objects, to allow users to more easily implement their own agent with their own attributes.
0.2.0
- Adds modules for multi-agent reinforcement learning - Restructure package - Extend documentation and examples