从子问题结构看大模型推理时扩增计算的方法
Test-time Scaling of LLMs: A Survey from A Subproblem Structure Perspective
- 按子问题的顺序、并行或树状结构分类推理策略
- 统一了思维链、分支求解合并等不同方法的逻辑
- 适合研究大模型推理优化的学者和工程师
本文综述通过在推理阶段分配额外计算资源来提升预训练大语言模型预测准确率的技术。在分类测试时扩展方法时，我们特别关注问题如何被分解为子问题，以及这些子问题的拓扑组织方式——顺序、并行或树状结构。这一视角使我们能够将多种方法（如思维链、分支-求解-合并、树状思维）纳入统一框架。我们进一步整合现有分析，揭示各类技术的优势与局限，并指出未来有潜力的研究方向。
原文摘要 · Abstract (English)
With this paper, we survey techniques for improving the predictive accuracy of pretrained large language models by allocating additional compute at inference time. In categorizing test-time scaling methods, we place special emphasis on how a problem is decomposed into subproblems and on the topological organization of these subproblems whether sequential, parallel, or tree-structured. This perspective allows us to unify diverse approaches such as Chain-of-Thought, Branch-Solve-Merge, and Tree-of-Thought under a common lens. We further synthesize existing analyses of these techniques, highlighting their respective strengths and weaknesses, and conclude by outlining promising directions for future research
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。