LLM DAYDREAMING

Summary from the TLDR newsletter

Despite their vast knowledge and high benchmark scores, no large language model has made a major breakthrough or unexpected insight - this post discusses a theoretical method for making models more insightful.