An implementation of the ideas from this paper https://arxiv.org/pdf/1803.10122.pdf
Following this tutorial and attempting to reproduce the same results, using a Google Cloud TensorFlow Deep Learning VM Instance:
https://applied-data.science/blog/hallucinogenic-deep-reinforcement-learning-using-python-and-keras