[bradhilton]

Well, in this case there is a much more straightforward method with the same CP-SAT solver used to create the puzzles. This is more of a fun experiment to see if we can train LLMs to solve these kinds of logical deduction problems.