SketchThinker-R1: 向大多模态模型中高效实现草图式推理迈进
SketchThinker-R1: Towards Efficient Sketch-Style Reasoning in Large Multimodal Models
AI总结：
SketchThinker-R1通过引入草图式推理机制，显著降低大得多模态模型的推理token成本，同时保持答案准确性。
摘要：
尽管在大多模态模型中广泛采用的详尽、逐步推理在经验上取得了成功，但长时间的推理过程不可避免地会带来显著的计算开销，即在 token 成本和响应时间上更高，这会削弱推理效率。相比之下，人类往往采用草图式推理：一种简洁、目标导向的认知过程，优先考虑显著信息并实现高效的问题解决。受这种认知效率的启发，我们提出了SketchThinker-R1，旨在在大得多模态模型中激励草图式推理能力。我们的方法由三个主要阶段组成。在Sketch-Mode冷启动阶段，我们将标准的长推理过程转换为草图式推理，并微调基础多模态模型，从而植入初始的草图式推理能力。接下来，我们训练SketchJudge奖励模型，该模型明确评估模型的思维过程，并对草图式推理给予更高的评分。最后，我们受SketchJudge的监督进行Sketch-Thinking强化学习，以进一步推广草图式推理能力。在四个基准测试上的实验评估表明，我们的SketchThinker-R1在不牺牲最终答案准确率的情况下，实现了超过64%的推理token成本降低。定性分析进一步显示，草图式推理在解决问题过程中更关注关键线索。
英文摘要：
Despite the empirical success of extensive, step-by-step reasoning in large multimodal models, long reasoning processes inevitably incur substantial computational overhead, i.e., in terms of higher token costs and increased response time, which undermines inference efficiency. In contrast, humans often employ sketch-style reasoning: a concise, goal-directed cognitive process that prioritizes salient information and enables efficient problem-solving. Inspired by this cognitive efficiency, we propose SketchThinker-R1, which incentivizes sketch-style reasoning ability in large multimodal models. Our method consists of three primary stages. In the Sketch-Mode Cold Start stage, we convert standard long reasoning process into sketch-style reasoning and finetune base multimodal model, instilling initial sketch-style reasoning capability. Next, we train SketchJudge Reward Model, which explicitly evaluates thinking process of model and assigns higher scores to sketch-style reasoning. Finally, we conduct Sketch-Thinking Reinforcement Learning under supervision of SketchJudge to further generalize sketch-style reasoning ability. Experimental evaluation on four benchmarks reveals that our SketchThinker-R1 achieves over 64% reduction in reasoning token cost without compromising final answer accuracy. Qualitative analysis further shows that sketch-style reasoning focuses more on key cues during problem solving.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
