论文：轮次级 Best-of-N 采样提升 LLM 智能体长程任务表现
On Turn-Level Best-of-N Sampling for Scaling LLM Agents
让智能体每轮先出 N 个候选动作再挑一个，实测准确率提升 5-10%，做 Agent 的可以看看这套推理时扩展思路。
arXiv 论文提出 turn-level best-of-N 采样方法，用 generator-selector 双模型框架：每轮由生成器模型产出 N 个候选动作，选择器模型挑出其一执行。在长程终端任务实验中，该方法带来 5-10% 的准确率提升，且多次尝试间的一致性更好。轨迹分析显示智能体减少了解题尝试次数，指令遵循、调试与测试行为改善。
On Turn-Level Best-of-N Sampling for Scaling LLM Agents
In this work, we evaluate a simple inference-time scaling approach based on \textit{turn-level best-of-N sampling} (\method) for large language model (LLM) agents on complex, long-horizon tasks. We instantiate this using a \textit{generator-selector} framework with two LLMs; at each turn, a \emph{generator} model generates $N$ candidate actions and a \emph{selector} model selects one of them, which is then executed in the environment. This paper is closely related to the work of \cite{zainullina2025guided} who used critic models for guided search at inference time. Different from their work, we evaluate if off-the-shelf, open-weight LLMs can be effective at comparing different candidate actions directly. Experiments on long-horizon terminal tasks show that \method can lead to significant performance improvements, both in terms of accuracy (5--10\%) and consistency across repeated attempts. Analysis of agent trajectories reveal a shift in agent behaviour towards desirable patterns such as reduction in number of attempts to solve the task, improved task instruction following, debugging and testing.