x
Alignment-Relevant Preprints: Forecast-LLMs, Dream-Augmented Memory, and Behavioral Safety Mechanisms — LessWrong