TinyR1-32B-Preview: 通过分支-合并蒸馏提升精度
TinyR1-32B-Preview: Boosting Accuracy with Branch-Merge Distillation
AI总结：
本文提出分支-合并蒸馏方法，通过领域特定监督微调和跨领域知识迁移，提升模型压缩效果，实验显示在数学、编程和科学领域均取得显著提升。
摘要：
减少大型语言模型（LLM）规模同时保持性能的挑战备受关注。然而，现有方法如模型蒸馏和迁移学习往往难以达到高精度。为此，我们引入分支-合并蒸馏方法，通过两个阶段：（1）分支阶段，通过领域特定监督微调（SFT）将大教师模型的知识选择性蒸馏到专用学生模型中；（2）合并阶段，将这些学生模型合并以实现跨领域知识转移并提升泛化能力。我们使用DeepSeek-R1作为教师，DeepSeek-R1-Distill-Qwen-32B作为学生进行验证。所得合并模型TinyR1-32B-Preview在多个基准测试中表现优异，包括数学（+5.5分）、编程（+4.4分）和科学（+2.9分），并在AIME 2024上与DeepSeek-R1表现接近。分支-合并蒸馏方法为创建更小、高性能的LLM提供了可扩展的解决方案，同时降低了计算成本和时间。
英文摘要：
The challenge of reducing the size of Large Language Models (LLMs) while maintaining their performance has gained significant attention. However, existing methods, such as model distillation and transfer learning, often fail to achieve high accuracy. To address this limitation, we introduce the Branch-Merge distillation approach, which enhances model compression through two phases: (1) the Branch Phase, where knowledge from a large teacher model is \textit{selectively distilled} into specialized student models via domain-specific supervised fine-tuning (SFT); And (2) the Merge Phase, where these student models are merged to enable cross-domain knowledge transfer and improve generalization. We validate our distillation approach using DeepSeek-R1 as the teacher and DeepSeek-R1-Distill-Qwen-32B as the student. The resulting merged model, TinyR1-32B-Preview, outperforms its counterpart DeepSeek-R1-Distill-Qwen-32B across multiple benchmarks, including Mathematics (+5.5 points), Coding (+4.4 points) and Science (+2.9 points), while achieving near-equal performance to DeepSeek-R1 on AIME 2024. The Branch-Merge distillation approach provides a scalable solution for creating smaller, high-performing LLMs with reduced computational cost and time.
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。
