Paper Detail
Zeyue Xue, Siming Fu, Jie Huang, Shuai Lu, Haoran Li, Yijun Liu, Yuming Li, Xiaoxuan He, Mengzhao Chen, Haoyang Huang, Nan Duan, Ping Luo
While large-scale video diffusion models have demonstrated impressive capabilities in generating high-resolution and semantically rich content, a significant gap remains between their pretraining performance and real-world deployment requirements due to critical issues such as prompt sensitivity, temporal inconsistency, and prohibitive inference costs. To bridge this gap, we propose a comprehensive post-training framework that systematically aligns pretrained models with user intentions through four synergistic stages: we first employ Supervised Fine-Tuning (SFT) to transform the base model into a stable instruction-following policy, followed by a Reinforcement Learning from Human Feedback (RLHF) stage that utilizes a novel Group Relative Policy Optimization (GRPO) method tailored for video diffusion to enhance perceptual quality and temporal coherence; subsequently, we integrate Prompt Enhancement via a specialized language model to refine user inputs, and finally address system efficiency through Inference Optimization. Together, these components provide a systematic approach to improving visual quality, temporal coherence, and instruction following, while preserving the controllability learned during pretraining. The result is a practical blueprint for building scalable post-training pipelines that are stable, adaptable, and effective in real-world deployment. Extensive experiments demonstrate that this unified pipeline effectively mitigates common artifacts and significantly improves controllability and visual aesthetics while adhering to strict sampling cost constraints.
No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.
No ranking explanation is available yet.
No tags.
@misc{xue2026systematic,
title = {A Systematic Post-Train Framework for Video Generation},
author = {Zeyue Xue and Siming Fu and Jie Huang and Shuai Lu and Haoran Li and Yijun Liu and Yuming Li and Xiaoxuan He and Mengzhao Chen and Haoyang Huang and Nan Duan and Ping Luo},
year = {2026},
abstract = {While large-scale video diffusion models have demonstrated impressive capabilities in generating high-resolution and semantically rich content, a significant gap remains between their pretraining performance and real-world deployment requirements due to critical issues such as prompt sensitivity, temporal inconsistency, and prohibitive inference costs. To bridge this gap, we propose a comprehensive post-training framework that systematically aligns pretrained models with user intentions through },
url = {https://huggingface.co/papers/2604.25427},
keywords = {video diffusion models, Supervised Fine-Tuning, SFT, Reinforcement Learning from Human Feedback, RLHF, Group Relative Policy Optimization, GRPO, prompt enhancement, inference optimization, temporal coherence, visual quality, controllability, huggingface daily},
eprint = {2604.25427},
archiveprefix = {arXiv},
}
{}