用确定性方法高效探索推理路径，避免重复计算。
Efficient Test-Time Inference via Deterministic Exploration of Truncated Decoding Trees
- 将采样转为遍历剪枝后的解码树，系统枚举不同终点。
- 在相同计算量下覆盖更多高质量推理路径，提升准确率。
- 适合需要高效推理的数学、编程等严谨任务场景。
自一致性通过并行采样多条推理路径并投票来提升推理性能。然而，在数学和代码等受限领域，该方法计算效率低下，因采样存在重复，反复访问相同高概率前缀和重复结果。本文提出一种确定性解码方法——唯一叶节点枚举（DLE），将截断采样视为对剪枝解码树的遍历，系统枚举不同的叶节点，而非重复采样。该策略从算法层面提升固定预算下的搜索空间覆盖率，从系统层面复用共享前缀，减少冗余生成。实验表明，相比随机自一致性，DLE能探索更高质量的推理路径，在数学、编程和通用推理任务上均表现更优。
原文摘要 · Abstract (English)
Self-consistency boosts inference-time performance by sampling multiple reasoning traces in parallel and voting. However, in constrained domains like math and code, this strategy is compute-inefficient because it samples with replacement, repeatedly revisiting the same high-probability prefixes and duplicate completions. We propose Distinct Leaf Enumeration (DLE), a deterministic decoding method that treats truncated sampling as traversal of a pruned decoding tree and systematically enumerates distinct leaves instead of sampling with replacement. This strategy improves inference efficiency in two ways. Algorithmically, it increases coverage of the truncated search space under a fixed budget by exploring previously unvisited high-probability branches. Systemically, it reuses shared prefixes and reduces redundant token generation. Empirically, DLE explores higher-quality reasoning traces than stochastic self-consistency, yielding better performance on math, coding, and general reasoning tasks.
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。