把大模型推理看作搜索过程，提升逻辑推理能力
When LLM Meets Tree Search: A Systematic View of Inference as Search in Large Language Models
- 将推理视为针对具体问题的优化，而非简单解码
- 引入统一设计空间，整合搜索结构与评估信号
- 适合关注模型推理机制与计算效率的研究者
随着预训练规模增长趋于饱和，测试时扩展（TTS）成为提升大模型推理能力的重要方向，即在固定模型前提下分配推理阶段的计算资源。从宏观视角看，TTS将推理重新定义为对部分推理状态空间的搜索。尽管思维链（CoT）揭示了中间步骤，但常见实现依赖单路径解码，难以纠正早期错误且探索有限。本综述系统梳理了基于树搜索的推理进展，将推理视为实例相关的优化过程而非单纯解码。我们追溯了从无信息搜索到蒙特卡洛树搜索（MCTS）的发展，强调采样控制如何支持有原则的探索-利用权衡。为统一碎片化研究，提出涵盖搜索拓扑、评估信号和控制动态的统一设计空间，并倡导标准化的计算报告范式，使计算量与准确率之间的权衡更加清晰可比。
原文摘要 · Abstract (English)
As pretraining scaling laws approach saturation, Test-Time Scaling (TTS) has emerged as an important direction for improving reasoning by allocating inference-time compute to a fixed model prior. Viewed at a high level, TTS reframes inference as search over a space of partial reasoning states. While Chain-of-Thought (CoT) exposes intermediate steps, common instantiations rely on single-trajectory decoding, limiting recovery from early errors and exploration. This survey systematizes recent progress in tree-search-based reasoning, viewing inference as instance-specific optimization rather than decoding. We trace the evolution from uninformed search to Monte Carlo Tree Search (MCTS), highlighting how sampling-based control supports principled exploration-exploitation trade-offs. To unify a fragmented literature, we introduce a Unified Design Space spanning search topology, evaluation signals, and control dynamics, and advocate a standardized compute-reporting abstraction to make compute-accuracy trade-offs explicit and comparable.
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。