LIVEditor-14B：基于上下文稀疏注意力的闪电视频编辑
LIVEditor-14B: Lightning Unified Video Editing via In-Context Sparse Attention
AI总结：
提出上下文稀疏注意力（ISA）框架，通过冗余上下文剪枝和动态查询分组实现近无损加速，构建LIVEditor-14B模型在多个基准上超越现有方法。
摘要：
视频编辑已向上下文学习（ICL）范式发展，但由此产生的二次注意力成本造成了关键的计算瓶颈。在这项工作中，我们提出了上下文稀疏注意力（ISA），这是首个专为ICL视频编辑设计的近无损经验稀疏框架。我们的设计基于两个关键见解：首先，上下文标记的显著性显著低于源标记；其次，我们从理论上证明并经验验证了查询锐度与近似误差相关。受这些发现启发，ISA实现了一种高效的预选择策略来剪枝冗余上下文，随后通过动态查询分组机制将高误差查询路由到全注意力，将低误差查询路由到计算高效的0阶泰勒稀疏注意力。此外，我们构建了 extbf{ exttt{LIVEditor-14B}}，这是一种通过ISA和提出的视频编辑数据流水线（整理了170万高质量数据集）的新型闪电视频编辑模型。大量实验表明，LIVEditor-14B在注意力模块延迟上减少了约60%，同时在EditVerseBench、IVE-Bench和VIE-Bench上超越了最先进的方法，实现了近无损加速且不损害视觉保真度。
英文摘要：
Video editing has evolved toward In-Context Learning (ICL) paradigms, yet the resulting quadratic attention costs create a critical computational bottleneck. In this work, we propose In-context Sparse Attention (ISA), the first near-lossless empirical sparse framework tailored for ICL video editing. Our design is grounded in two key insights: first, context tokens exhibit significantly lower saliency than source tokens; second, we theoretically prove and empirically validate that Query sharpness correlates with approximation error. Motivated by these findings, ISA implements an efficient pre-selection strategy to prune redundant context, followed by a dynamic query grouping mechanism that routes high-error queries to full attention and low-error ones to a computationally efficient 0-th order Taylor sparse attention. Furthermore, we build \textbf{\texttt{LIVEditor-14B}} , a novel lightning video editing model via ISA and a proposed video-editing data pipeline that curated a 1.7M high-quality dataset. Extensive experiments demonstrate that LIVEditor-14B achieves a $\sim$60% reduction in attention-module latency while surpassing state-of-the-art methods across EditVerseBench, IVE-Bench, and VIE-Bench, delivering near-lossless acceleration without compromising visual fidelity.
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。
