JsonTuning：迈向可泛化、鲁棒且可控的指令微调
JsonTuning: Towards Generalizable, Robust, and Controllable Instruction Tuning
AI总结：
针对 TextTuning 缺乏显式任务结构的问题，本文提出基于 JSON 的结构到结构指令微调方法 JsonTuning，并在多模型、多基准实验中验证其在性能、鲁棒性和可控性上的一致优势。
摘要：
指令微调对于提升大型语言模型（LLM）的性能至关重要，但现有的文本到文本方法（称为 TextTuning）由于缺乏显式任务结构，在泛化性、鲁棒性和可控性方面存在困难。我们提出 JsonTuning，这是一种利用 JSON 结构表示任务的结构到结构方法。该方法通过明确任务元素及其关系来提升泛化性，通过最小化歧义来增强鲁棒性，并通过允许对输出进行精确控制来提高可控性。我们使用多种语言模型和基准测试对 JsonTuning 与 TextTuning 进行了广泛的比较分析。研究结果表明，在不同场景下，JsonTuning 在性能、鲁棒性和可控性方面始终优于 TextTuning。通过克服 TextTuning 的局限性，JsonTuning 展现出开发更有效、更可靠且能够处理多种场景的 LLM 的巨大潜力。
英文摘要：
Instruction tuning is vital for enhancing the performance of large language models (LLMs), but existing text-to-text methods, referred to as TextTuning, struggle with issues such as generalization, robustness, and controllability due to their lack of explicit task structures. We introduce JsonTuning, a structure-to-structure approach that uses JSON structures to represent tasks. This method improves generalization by clarifying task elements and their relations, boosts robustness by minimizing ambiguity, and enhances controllability by allowing precise control over outputs. We conduct an extensive comparative analysis between JsonTuning and TextTuning using various language models and benchmarks. Our findings reveal that JsonTuning consistently surpasses TextTuning in terms of performance, robustness, and controllability across different scenarios. By overcoming the limitations of TextTuning, JsonTuning demonstrates significant potential for developing more effective and reliable LLMs capable of handling diverse scenarios.
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。
