@whitequark I like it when my compiler gets done with it quickly. Needing a supercomputer to run an LLM, which still takes seconds to just output a few tokens, and then relying on a solver for an NP-hard problem which could take forever to verify and maybe reject the output, does not sound like a good idea.