DART: Distillation-Aware Reparameterization for Training-Free LoRA Reuse in Few-Step Video Diffusion Models

2026-09-17

AI Summary

为解决LoRA在视频生成中重用导致的质量下降问题，提出DART方法，通过低秩坐标变换与目标调度响应校准提高质量。

Abstract

Step distillation reduces the cost of video generation, but reusing a LoRA trained for a longer trajectory can alter its functional effect or degrade target quality. Static parameter compatibility offers one perspective on this problem; our observations show that similar measured geometry can coexist with different adapter behavior under a shortened denoising schedule. We propose DART, a training-free method that combines low-rank coordinate transport with target-schedule response calibration using forward evaluations and no source training videos. On a four-step Wan2.2 target, DART-F improves the joint quality score from 0.9029 to 0.9227 and changes macro functional retention from -0.4644 to +0.1349. Component analysis shows that calibration accounts for most of the quality improvement, while coordinate transport provides complementary gains when combined with calibration. Adapter-level results reveal positive functional effects for some adapters and strong attenuation with reduced negative functional effects for others. Evaluations on two additional targets show the same aggregate trend. These results motivate evaluating distilled-model LoRA reuse jointly through functional preservation and negative-transfer avoidance, without assuming recovery for every adapter.

Problem

• Step Distillation
• LoRA Reuse
• Video Generation
• Target Quality
• Training-Free

Innovation

• Distillation-Aware Reparameterization
• Training-Free LoRA Reuse
• Low-Rank Coordinate Transport
• Target-Schedule Response Calibration