Towards Neural Theorem Proving at Scale