Demo code for our NeurIPS 2019 paper

Try it out interactively on colab or view with nbviewer.

Cite as

@inproceedings{farquhar2019loaded,
  title={Loaded DiCE: Trading off Bias and Variance in Any-Order Score Function Estimators for Reinforcement Learning},
  author={Farquhar, Gregory and Whiteson, Shimon and Foerster, Jakob},
  booktitle={Advances in Neural Information Processing Systems},
  year={2019}
}