用大模型指导小模型生成风格化文章，效果超越GPT-4
SAG: Style-Aligned Article Generation via Model Collaboration
- 大模型冻结提指令理解，小模型微调学风格
- 在NoteBench上比GPT-4高0.78的ROUGE-L和0.55的BLEU-4
- 适合需要高风格一致性与低幻觉的个性化内容生成
大语言模型（LLMs）提升了对个性化、风格化内容生成的需求。然而，闭源模型如GPT-4难以优化，而开源模型如Qwen-72B则面临高昂训练成本和灵活性不足的问题。小型语言模型（SLMs）在理解复杂指令和迁移能力方面表现较差。本文提出一种新型协作训练框架，结合LLMs与SLMs优势，用于风格化文章生成，性能优于单一模型。我们冻结LLMs以利用其强指令遵循能力，并在小模型上使用风格特定数据进行监督微调。此外，引入自提升方法增强风格一致性。新基准NoteBench全面评估风格对齐生成效果。大量实验表明，该方法达到当前最优性能，在ROUGE-L上较GPT-4提升0.78，BLEU-4提升0.55，同时保持较低的事实性与忠实性幻觉率。
原文摘要 · Abstract (English)
Large language models (LLMs) have increased the demand for personalized and stylish content generation. However, closed-source models like GPT-4 present limitations in optimization opportunities, while the substantial training costs and inflexibility of open-source alternatives, such as Qwen-72B, pose considerable challenges. Conversely, small language models (SLMs) struggle with understanding complex instructions and transferring learned capabilities to new contexts, often exhibiting more pronounced limitations. In this paper, we present a novel collaborative training framework that leverages the strengths of both LLMs and SLMs for style article generation, surpassing the performance of either model alone. We freeze the LLMs to harness their robust instruction-following capabilities and subsequently apply supervised fine-tuning on the SLM using style-specific data. Additionally, we introduce a self-improvement method to enhance style consistency. Our new benchmark, NoteBench, thoroughly evaluates style-aligned generation. Extensive experiments show that our approach achieves state-of-the-art performance, with improvements of 0.78 in ROUGE-L and 0.55 in BLEU-4 scores compared to GPT-4, while maintaining a low hallucination rate regarding factual and faithfulness.
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。