- Added ONNX export colab example for discrete and continuous action spaces. For continuous case LSTM policy example is provided as well.
- Improved RNNs training in continuous space, added option zero_rnn_on_done.
- Added NVIDIA CuLE support: https://github.com/NVlabs/cule
- Added player config override. Vecenv is used for inference.
- Fixed multi-gpu training with central value.
- Fixed max_frames termination condition, and it's interaction with the linear learning rate: https://github.com/Denys88/rl_games/issues/212
- Fixed "deterministic" misspelling issue.
- Fixed Mujoco and Brax SAC configs.
- Fixed multiagent envs statistics reporting. Fixed Starcraft2 SMAC environments.
**Full Changelog**: https://github.com/Denys88/rl_games/compare/v1.5.2...v1.6.0