让大模型自己决定如何用计算资源推理，提升问答与代码生成效果。
ATLAS: Agentic Test-time Learning-to-Allocate Scaling
- 用一个可扩展的行动空间让模型自主调度求解器、调整推理强度和提示策略。
- 在四个基准上表现优于固定流程方法，如HLE-Verified达56.00%，且调用接口更少。
- 适合需要动态推理决策的复杂任务，尤其对多模型协同场景有显著增益。
测试时缩放已成为提升大语言模型推理能力的重要方式，但其调度仍依赖人工设计：固定的样本预算、固定的迭代轮次、固定的评分规则或搜索策略决定了计算资源的分配，使模型仅负责求解而无法掌控调度过程。我们提出ATLAS，一种由大模型主导的代理式测试时缩放框架，实现从头到尾的端到端控制。通过单一动作“探索”（explore），该框架可派遣独立求解器重新求解原始问题，自主决定是否收集更多证据、何时停止以及如何合成最终答案；动作空间可扩展，每次探索可选择求解器、推理努力程度或提示策略。我们在四类基准上评估了以Claude Sonnet 4.6为骨干的ATLAS，涵盖科学问答、代码生成与多模态推理，在HLE-Verified上达到56.00%，LiveCodeBench上82.29%，GPQA-Diamond上85.75%，BabyVision上23.71%，且使用远少于固定工作流基线的API调用次数。多模型扩展版ATLAS-MM将求解器选择作为额外动作维度，进一步将HLE-Verified提升至60.00%，LiveCodeBench至85.63%，并在其他两个基准上保持一致增益。消融实验表明，若用独立集成器替代原生合成机制，三个基准上的准确率下降或未提升，说明状态化证据管理是性能提升的关键。
原文摘要 · Abstract (English)
Test-time scaling has become a major way to improve large language model reasoning, but its orchestration has remained designer-engineered: a fixed sample budget, a fixed refinement loop, a fixed scoring rule, or a fixed search policy decides how compute is spent, leaving the model in charge of solving but not of orchestration. We introduce ATLAS, an agentic test-time scaling framework in which an LLM orchestrator owns the control loop end-to-end. Through a single action, explore, which dispatches a fresh independent solver on the original problem, the orchestrator decides whether to gather more evidence, when to stop, and how to synthesize the final answer; the action space is extensible, with each explore call optionally specifying solver, reasoning effort, or prompting strategy. We evaluate ATLAS on four benchmarks covering scientific question answering, code generation, and multimodal reasoning under a Claude Sonnet 4.6 backbone, where it reaches 56.00% on HLE-Verified, 82.29% on LiveCodeBench, 85.75% on GPQA-Diamond, and 23.71% on BabyVision while using far fewer API calls than fixed-workflow baselines. A multi-model extension, ATLAS-MM, that exposes solver choice as an additional action dimension further improves HLE-Verified to 60.00% and LiveCodeBench to 85.63%, with consistent gains on GPQA-Diamond and BabyVision. Ablations replacing the orchestrator's direct synthesis with a separate integrator degrade or fail to improve accuracy on three of four benchmarks, consistent with the role of stateful evidence management in producing the gains.
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。