用遗传算法提升大模型推理能力，解决困在局部最优的问题
Lyria: A Genetic Algorithm-Driven Neuro-Symbolic Reasoning Framework for LLMs
- 将大模型、遗传算法与符号系统结合，构建神经符号推理框架
- 在3类问题上测试4个大模型，显著提升解空间覆盖与推理质量
- 适合关注大模型推理优化与可解释性研究的读者
尽管大语言模型在多个领域展现出强大能力，但仍面临两大挑战：容易陷入局部最优解，且对解空间的覆盖不充分。为探究并改进这些问题，本文提出Lyria——一个基于大模型、遗传算法与符号系统融合的神经符号推理框架，包含7个核心组件。通过在3类问题上对4个大模型进行广泛实验，验证了Lyria的有效性。此外，通过7组消融实验，系统分析了影响其性能的关键因素。基于Lyria，我们进一步将其思想拓展至大模型微调过程，提出LAFT方法，使弱模型能够模仿强模型在Lyria框架下的推理过程。大量实验表明，LAFT相比9个基准方法具有显著优势。最后，本文揭示了当前方法的局限性，并为未来方向提供见解。
原文摘要 · Abstract (English)
While LLMs have demonstrated impressive abilities across various domains, they struggle with two major issues. The first is that LLMs trap themselves into local optima and the second is that they lack exhaustive coverage of the solution space. To investigate and improve these two issues, we propose Lyria, a neuro-symbolic reasoning framework building on the integration of LLMs, genetic algorithms, and symbolic systems, comprising 7 essential components. Through conducting extensive experiments with 4 LLMs across 3 types of problems, we demonstrated the efficacy of Lyria. Furthermore, with 7 additional ablation experiments, we further systematically analyzed and elucidated the factors that affect its performance. In addition, based on Lyria, we extend the ideas to the fine-tuning process of LLMs and introduce LAFT which enables a weaker model to imitate the reasoning process of a stronger model that reason under the Lyria reasoning framework. We demonstrate that the significant effectiveness of LAFT by conducting extensive experiments against 9 constructed baselines. We finally reveal the limitations and provide insights into future directions.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。