通过掩码微调让大模型性能提升，不改权重也能优化。
Boosting Large Language Models with Mask Fine-Tuning
- 用二值掩码动态调整模型结构，不更新权重即可优化。
- 在LLaMA2-7B和3.1-8B上分别提升2.70和4.15分（IFEval）。
- 可与现有优化方法协同，适合想提升模型表现的研究者。
大型语言模型通常采用主流优化协议。然而，尚未有研究质疑维持模型结构完整性是否对优异性能不可或缺。本文提出掩码微调（MFT），一种新型的微调范式，表明精心破坏模型结构完整性反而能提升性能，且无需更新模型权重。MFT学习并应用二值掩码，以标准的大语言模型微调目标为监督信号。基于已充分微调的模型，使用相同微调数据集，在不同领域和骨干网络上均实现一致性能提升（例如，使用LLaMA2-7B / 3.1-8B时，IFEval平均提升分别为2.70 / 4.15）。通过详尽的消融实验与分析，从稀疏率、损失曲面等角度验证了MFT的有效性。此外，该方法兼容已有优化流程，可进一步增强通用模型能力。更重要的是，掩码操作的应用从传统的模型压缩范畴拓展至更广泛的模型能力提升场景。
原文摘要 · Abstract (English)
The large language model (LLM) is typically integrated into the mainstream optimization protocol. No work has questioned whether maintaining the model integrity is \textit{indispensable} for promising performance. In this work, we introduce Mask Fine-Tuning (MFT), a novel LLM fine-tuning paradigm demonstrating that carefully breaking the model's structural integrity can surprisingly improve performance without updating model weights. MFT learns and applies binary masks to well-optimized models, using the standard LLM fine-tuning objective as supervision. Based on fully fine-tuned models, MFT uses the same fine-tuning datasets to achieve consistent performance gains across domains and backbones (e.g., an average gain of \textbf{2.70 / 4.15} in IFEval with LLaMA2-7B / 3.1-8B). Detailed ablation studies and analyses examine the proposed MFT from different perspectives, such as sparse ratio and loss surface. Additionally, by deploying it on well-trained models, MFT is compatible with collaborating with other LLM optimization procedures to enhance the general model. Furthermore, this study extends the functionality of the masking operation beyond its conventional network-pruning context for model compression to a broader model capability scope.
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。