5 ms·
Adaptive reasoning is known to be an extremely hard problem to solve, though. It requires you to predict whether a certain LLM, with a certain effort level, wit
by doodlesdev 15d ago
Adaptive reasoning is known to be an extremely hard problem to solve, though. It requires you to predict whether a certain LLM, with a certain effort level, with a certain prompt, will give you the right answer.
- lilytweed 15d agoThis feels like exactly the kind of problem domain that belongs in (and can be solved by) RL?