{"ID":2864152,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2509.23652","arxiv_id":"2509.23652","title":"ReWatch-R1: Boosting Complex Video Reasoning in Large Vision-Language Models through Agentic Data Synthesis","abstract":"While Reinforcement Learning with Verifiable Reward (RLVR) significantly advances image reasoning in Large Vision-Language Models (LVLMs), its application to complex video reasoning remains underdeveloped. This gap stems primarily from a critical data bottleneck: existing datasets lack the challenging, multi-hop questions and high-quality, video-grounded Chain-of-Thought (CoT) data necessary to effectively bootstrap RLVR. To address this, we introduce ReWatch, a large-scale dataset built to foster advanced video reasoning. We propose a novel multi-stage synthesis pipeline to synthesize its three components: ReWatch-Caption, ReWatch-QA, and ReWatch-CoT. A core innovation is our Multi-Agent ReAct framework for CoT synthesis, which simulates a human-like \"re-watching\" process to generate video-grounded reasoning traces by explicitly modeling information retrieval and verification. Building on this dataset, we develop ReWatch-R1 by post-training a strong baseline LVLM with Supervised Fine-Tuning (SFT) and our RLVR framework. This framework incorporates a novel Observation \\\u0026 Reasoning (O\\\u0026R) reward mechanism that evaluates both the final answer's correctness and the reasoning's alignment with video content, directly penalizing hallucination. Our experiments show that ReWatch-R1 achieves state-of-the-art average performance on five challenging video reasoning benchmarks. Project Page: https://rewatch-r1.github.io","short_abstract":"While Reinforcement Learning with Verifiable Reward (RLVR) significantly advances image reasoning in Large Vision-Language Models (LVLMs), its application to complex video reasoning remains underdeveloped. This gap stems primarily from a critical data bottleneck: existing datasets lack the challenging, multi-hop questi...","url_abs":"https://arxiv.org/abs/2509.23652","url_pdf":"https://arxiv.org/pdf/2509.23652v2","authors":"[\"Congzhi Zhang\",\"Zhibin Wang\",\"Yinchao Ma\",\"Jiawei Peng\",\"Yihan Wang\",\"Qiang Zhou\",\"Jun Song\",\"Bo Zheng\"]","published":"2025-09-28T05:38:16Z","proceeding":"cs.CV","tasks":"[\"cs.CV\",\"cs.AI\"]","methods":"[\"Reinforcement Learning\",\"Language Model\"]","has_code":false}