Large language models have been shown to struggle with limited context memory and multi-step reasoning. We propose a simple method for solving both of these problems by allowing the model to take Self-Notes. Unlike recent scratchpad approaches, the model can deviate from the input context at any time to explicitly think. This allows the model to recall information and perform reasoning on the fly as it reads the context, thus extending its memory and enabling multi-step reasoning. Our experiments on multiple tasks demonstrate that our method can successfully generalize to longer and more complicated instances from their training setup by taking Self-Notes at inference time.

翻译：大型语言模型在有限上下文记忆和多步推理方面存在明显不足。我们提出了一种简单方法来解决这两个问题，即允许模型随时进行“自我笔记”。与近期出现的暂存区方法不同，本模型可在任意时刻偏离输入上下文进行显性思考。这使得模型在阅读上下文时能动态回忆信息并执行推理，从而扩展其记忆能力并实现多步推理。我们在多个任务上的实验表明，该方法通过在推理时进行自我笔记，能够成功泛化到训练配置中更长、更复杂的实例。