Google Research

Discrete Autoencoders for Sequence Models

  • Lukasz Kaiser
  • Samy Bengio
arXiv (2018)


The contributions of this paper are: * A discretization technique that works well without any extra losses or parameters to tune. * A way to measure performance of autoencoders for sequence models (and baselines). * An improved way to sample from sequence models (trained with an autoencoder part).

Research Areas

Learn more about how we do research

We maintain a portfolio of research projects, providing individuals and teams the freedom to emphasize specific types of work