{"work":{"id":"cfe4c01d-1cf7-4a00-ba86-06d583ca2cff","openalex_id":null,"doi":null,"arxiv_id":"2501.13918","raw_key":null,"title":"Improving Video Generation with Human Feedback","authors":null,"authors_text":"Jie Liu, Gongye Liu, Jiajun Liang, Ziyang Yuan, Xiaokun Liu, Mingwu Zheng","year":2025,"venue":"cs.CV","abstract":"Video generation has achieved significant advances through rectified flow techniques, but issues like unsmooth motion and misalignment between videos and prompts persist. In this work, we develop a systematic pipeline that harnesses human feedback to mitigate these problems and refine the video generation model. Specifically, we begin by constructing a large-scale human preference dataset focused on modern video generation models, incorporating pairwise annotations across multi-dimensions. We then introduce VideoReward, a multi-dimensional video reward model, and examine how annotations and various design choices impact its rewarding efficacy. From a unified reinforcement learning perspective aimed at maximizing reward with KL regularization, we introduce three alignment algorithms for flow-based models. These include two training-time strategies: direct preference optimization for flow (Flow-DPO) and reward weighted regression for flow (Flow-RWR), and an inference-time technique, Flow-NRG, which applies reward guidance directly to noisy videos. Experimental results indicate that VideoReward significantly outperforms existing reward models, and Flow-DPO demonstrates superior performance compared to both Flow-RWR and supervised fine-tuning methods. Additionally, Flow-NRG lets users assign custom weights to multiple objectives during inference, meeting personalized video quality needs.","external_url":"https://arxiv.org/abs/2501.13918","cited_by_count":null,"metadata_source":"pith","metadata_fetched_at":"2026-05-25T06:45:26.149058+00:00","pith_arxiv_id":"2501.13918","created_at":"2026-05-09T06:40:39.247838+00:00","updated_at":"2026-05-25T06:45:26.149058+00:00","title_quality_ok":true,"display_title":"Improving Video Generation with Human Feedback","render_title":"Improving Video Generation with Human Feedback"},"hub":{"state":{"work_id":"cfe4c01d-1cf7-4a00-ba86-06d583ca2cff","tier":"hub","tier_reason":"10+ Pith inbound or 1,000+ external citations","pith_inbound_count":37,"external_cited_by_count":null,"distinct_field_count":3,"first_pith_cited_at":"2025-03-07T08:36:05+00:00","last_pith_cited_at":"2026-05-22T17:59:43+00:00","author_build_status":"not_needed","summary_status":"needed","contexts_status":"needed","graph_status":"needed","ask_index_status":"not_needed","reader_status":"not_needed","recognition_status":"not_needed","updated_at":"2026-05-28T16:18:46.665488+00:00","tier_text":"hub"},"tier":"hub","role_counts":[{"context_role":"background","n":16},{"context_role":"baseline","n":4},{"context_role":"method","n":2}],"polarity_counts":[{"context_polarity":"background","n":16},{"context_polarity":"baseline","n":4},{"context_polarity":"use_method","n":2}],"runs":{},"summary":{},"graph":{},"authors":[]}}