Christian Walder

We provide a robust method of directly optimizing the pass at k with reinforcement learning, with theory and real world experiments.

Learning proof system dynamics, pruning proof search based on diversity and expected outcome