Pytorch for Reinforcement Learning with Google TPUs

Are there examples of using PyTorch to parallelize/speed up RL algorithms/training? I can’t seem to find any notable examples/notebooks. I’m looking to train an RL algorithm with gym using Google TPUs.