用流程图生成对话指令，让大模型更准确完成流程化任务
PFDial: A Structured Dialogue Instruction Fine-tuning Method Based on UML Flowcharts
- 将UML流程图转为结构化对话单元，构建高质量指令数据集
- 小模型仅用800样本训练即达90%以上准确率，8B模型超越GPT-4o达43.88%
- 适合需要严格流程控制的客服、设备维护等场景使用
流程驱动的对话系统在客户服务和设备维护中至关重要，但大语言模型在严格约束任务上表现仍不佳。为此，我们构建了包含12,705条高质量中文对话指令的PFDial数据集，源自440张包含5,055个流程节点的UML流程图。基于PlantUML规范，每张流程图被转化为原子对话单元（五元组结构）。实验表明，仅用800样本训练的7B模型与全量数据训练的0.5B模型均超过90%准确率；8B模型平均性能超越GPT-4o达11.00%，最高提升43.88%。我们还评估了模型在流程回溯转换中的表现，并深入分析不同数据格式对决策分支与顺序分支处理能力的影响。数据已开源于https://github.com/KongLongGeFDU/PFDial。
原文摘要 · Abstract (English)
Process-driven dialogue systems, which operate under strict predefined process constraints, are essential in customer service and equipment maintenance scenarios. Although Large Language Models (LLMs) have shown remarkable progress in dialogue and reasoning, they still struggle to solve these strictly constrained dialogue tasks. To address this challenge, we construct Process Flow Dialogue (PFDial) dataset, which contains 12,705 high-quality Chinese dialogue instructions derived from 440 flowcharts containing 5,055 process nodes. Based on PlantUML specification, each UML flowchart is converted into atomic dialogue units i.e., structured five-tuples. Experimental results demonstrate that a 7B model trained with merely 800 samples, and a 0.5B model trained on total data both can surpass 90% accuracy. Additionally, the 8B model can surpass GPT-4o up to 43.88% with an average of 11.00%. We further evaluate models' performance on challenging backward transitions in process flows and conduct an in-depth analysis of various dataset formats to reveal their impact on model performance in handling decision and sequential branches. The data is released in https://github.com/KongLongGeFDU/PFDial.
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。