This blog post is based on the work BaNEL: Exploration Posteriors for Generative Modeling Using Only Negative Rewards. Tackling Very Hard Problems The ultimate aim of machine learning research is to push machines beyond human limits in critical applications, including the next generation of th