You seem knowledgeable. Can you share a couple of interesting papers for theorem proving that came out in the last year? I read a few of them as they came out, and it seemed neural nets can advance the field by mixing "soft" language with "hard" symbolic systems.