揭示缺陷：探索合成数据中的不完善之处及大语言模型的缓解策略
Unveiling the Flaws: Exploring Imperfections in Synthetic Data and Mitigation Strategies for Large Language Models
AI总结：
针对大语言模型训练用合成数据存在模式过拟合、降低指令遵循能力的缺陷，提出基于遗忘技术的缓解方法，可低成本逆转问题且不损害基准性能，为合成数据有效使用提供洞见。
摘要：
合成数据已被提出作为解决大语言模型（LLM）训练中高质量数据稀缺问题的方案。已有研究表明，合成数据可有效提升大语言模型在下游基准测试上的性能。然而，尽管合成数据具备潜在优势，我们的分析显示其可能存在固有缺陷：合成数据的统一格式会导致模式过拟合，引发输出分布的显著偏移，进而降低模型的指令遵循能力。本研究深入探究了问答（Q-A）对这类主流合成数据存在的具体缺陷，并提出了一种基于遗忘技术的缓解方法。实证结果验证了该方法的有效性：它能够在相对较低的成本下，逆转模式过拟合导致的指令遵循问题，同时不会损害模型在基准测试上的性能。本研究为合成数据的有效使用提供了关键洞见，旨在推动更鲁棒、更高效的大语言模型训练。
英文摘要：
Synthetic data has been proposed as a solution to address the issue of high-quality data scarcity in the training of large language models (LLMs). Studies have shown that synthetic data can effectively improve the performance of LLMs on downstream benchmarks. However, despite its potential benefits, our analysis suggests that there may be inherent flaws in synthetic data. The uniform format of synthetic data can lead to pattern overfitting and cause significant shifts in the output distribution, thereby reducing the model's instruction-following capabilities. Our work delves into these specific flaws associated with question-answer (Q-A) pairs, a prevalent type of synthetic data, and presents a method based on unlearning techniques to mitigate these flaws. The empirical results demonstrate the effectiveness of our approach, which can reverse the instruction-following issues caused by pattern overfitting without compromising performance on benchmarks at relatively low cost. Our work has yielded key insights into the effective use of synthetic data, aiming to promote more robust and efficient LLM training.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
