Experimenting with integrating PyTorch Lightning + TorchRL for differentiable environments running on multiple GPUs

Training a PPO agent to play chess with pretraining and self-learning using PyTorch Lightning and TorchRL