小模型也能精准理解视频动作，通过三阶段训练提升细粒度推理能力。
ReasonAct: Progressive Training for Fine-Grained Video Reasoning in Small Models
- 分三阶段训练：先纯文本推理，再视频微调，最后时序强化学习优化
- 30亿参数模型在多个数据集上准确率超基线12.3至17.9个百分点
- 设计动作分解机制，按动作阶段给予渐进奖励，适合资源有限场景
尽管近期多模态模型在视觉-语言任务上取得进展，小规模模型在视频理解所需的细粒度时间推理方面仍存在困难。我们提出ReasonAct，一种通过三阶段训练增强小型模型视频推理能力的方法：首先利用纯文本进行基础推理训练，接着在视频数据上微调，最后通过引入时序一致性建模的时序感知强化学习进行精炼。我们在T-GRPO基础上改进，加入时序一致性建模以优化策略。同时提出一种受生物力学启发的子动作分解机制，为动作各阶段提供分级奖励。在HMDB51、UCF-101和Kinetics-400上的实验表明，3B参数模型分别达到67.2%、94.1%和78.9%的准确率，相较基线提升17.9、15.8和12.3个百分点。消融实验证明，该渐进式训练策略使小模型在保持计算效率的同时获得具有竞争力的视频推理性能。
原文摘要 · Abstract (English)
While recent multimodal models have shown progress in vision-language tasks, small-scale variants still struggle with the fine-grained temporal reasoning required for video understanding. We introduce ReasonAct, a method that enhances video reasoning in smaller models through a three-stage training process: first building a foundation with text-only reasoning, then fine-tuning on video, and finally refining with temporal-aware reinforcement learning. We build upon Temporal Group Relative Policy Optimization (T-GRPO) by incorporating temporal consistency modeling into policy optimization. We also propose a biomechanically-motivated sub-action decomposition mechanism that provides graduated rewards for constituent action phases. Through experiments on HMDB51, UCF-101, and Kinetics-400, our 3B-parameter model achieves 67.2%, 94.1%, and 78.9% accuracy respectively, demonstrating improvements of 17.9, 15.8, and 12.3 points over baselines. Ablation studies validate that our progressive training enables smaller models to achieve competitive video reasoning performance while maintaining computational efficiency.
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。