MoT: Memory-of-Thought Enables ChatGPT to Self-Improve
This addresses the challenge of enhancing LLMs efficiently for AI researchers and practitioners, though it is incremental as it builds on existing chain-of-thought methods.
The paper tackles the problem of improving large language models (LLMs) without annotated datasets or parameter updates by proposing MoT, a framework that enables self-improvement through memory-of-thought, resulting in significant performance gains in arithmetic, commonsense, factual reasoning, and natural language inference tasks for ChatGPT.
Large Language Models (LLMs) have shown impressive abilities in various tasks. However, fundamentally improving them depends on high-quality datasets or computationally expensive fine-tuning. On the contrary, humans can easily improve themselves by self-thinking and memory, without external resources. In this paper, we propose a framework, MoT, to let the LLM self-improve through Memory-of-Thought, without annotated datasets and parameter updates. Specifically, MoT is divided into two stages: 1. before the test stage, the LLM pre-thinks on the unlabeled dataset and saves the high-confidence thoughts as external memory; 2. During the test stage, given a test question, the LLM recalls relevant memory to help itself reason and answer it. Experimental results show that MoT can help ChatGPT significantly improve its abilities in arithmetic reasoning, commonsense reasoning, factual reasoning, and natural language inference. Further analyses show that each component contributes critically to the improvements and MoT can lead to consistent improvements across various CoT methods and LLMs.