This is very encouraging. This halting problem-esque issue with LLMs is one of the most potent.
It's like they can never truly get to a final answer, they'll just loop forever.
Having a solution to this is going to be very very valuable.
Replying to @lucas_builds
Landing improvements now, it鈥檚 actually quite tricky to get this right and we鈥檝e been doing a fair bit of science here. Will share more soon.
