Test-time scaling for agentic search typically increases depth (i.e., more turns and tokens per trajectory) or breadth (i.e., more parallel rollouts). Here we focus on breadth scaling, showing that standard parallel sampling yields diminishing returns, tracing this to query redundancy at the first turn. When models issue similar first queries across rollouts, the threads retrieve overlapping evidence, and subsequent turns are conditioned on this shared retrieval. We address this limitation with DivInit, a training-free intervention at the first turn. Rather than sampling k independent first queries, DivInit draws n candidates from a single call, picks k < n diverse seeds, and runs them as parallel trajectories. Across five open-weight models and eight benchmarks, DivInit consistently improves over standard parallel sampling, with average gains of five to seven points on multi-hop QA at matched compute. Code available at https://github.com/cxcscmu/diverse-query-initialization

翻译：智能体搜索的测试时扩展通常增加深度（即每轨迹轮次与标记数）或广度（即并行 rollout 数量）。本文聚焦于广度扩展，表明标准并行采样存在边际收益递减效应，并将其归因于首轮查询的冗余性。当模型在不同 rollout 中发出相似的首轮查询时，线程会检索重叠的证据，后续轮次则基于此共享检索结果进行条件生成。我们提出 DivInit 方法解决这一局限，该方法在首轮进行无需训练的干预：不同于采样 k 个独立的首轮查询，DivInit 通过单次调用生成 n 个候选查询，从中选取 k < n 个多样化种子，并作为并行轨迹运行。在五个开放权重模型与八个基准测试中，DivInit 一致优于标准并行采样，在多跳问答任务上以同等计算量平均提升 5-7 个点。代码见 https://github.com/cxcscmu/diverse-query-initialization