Apple 提出 SSR：把多模态推理改成选择题，降低小模型推理成本
Selection-Based Structured Reasoning: Toward Efficient Multimodal Search Agents
Apple 把推理从写答案改成做选择题，小模型跑多模态智能体能省不少算力，做 Agent 的可以看看思路。
Apple 研究团队提出 Selection-based Structured Reasoning（SSR）框架，针对多模态搜索智能体中常见的自由生成式推理问题。SSR 把高频的高层推理过程预先写成可复用的自然语言候选，模型每轮只需从中选择，而非开放式生成。这种方式避免了小模型因容量有限而产出冗长、对动作生成无实际帮助的推理文本，同时减少推理开销。论文发布于 Apple Machine Learning Research。
Selection-Based Structured Reasoning: Toward Efficient Multimodal Search Agents
Multimodal agents commonly generate freeform reasoning before each action. For small models, limited model capacity can result in lengthy reasoning that provides little useful guidance for action generation while incurring substantial inference cost. To address this challenge, we introduce Selection-based Structured Reasoning (SSR), a framework that reformulates reasoning as selection instead of open-ended generation. SSR represents recurring high-level reasoning as pre-specified, reusable natural-language candidates. At each turn, the model selects from the reasoning candidates based on their…
