On scalable oversight with weak LLMs judging strong LLMs

Arxiv Papers - On scalable oversight with weak LLMs judging strong LLMs