面向参数高效多任务学习的频率切换机制
Frequency Switching Mechanism for Parameter-Efficient Multi-Task Learning
把正弦频率调制叠加在低秩+卷积先验的单一核上，用极少量可训练参数（6.53M）实现多任务权重特化，可直接替换视觉骨干网络中的 LoRA/Adapter 做多任务稠密预测的联合微调。
原文依据 enables near-zero-cost weight modulation via frequency switching (Free)
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论多任务学习（MTL）旨在让单一模型高效地求解多个任务；然而，当前的参数高效微调（PEFT）方法在很大程度上仍局限于单任务适配。我们提出 Free Sinewich，一个参数高效的多任务学习框架，通过频率切换（Free）实现近乎零成本的权重调制。具体而言，Sine-AWB（Sinewich）层将低秩因子与卷积先验组合为单一核，随后通过正弦变换对该核进行逐元素调制，从而生成任务专用的权重。我们引入轻量级的 Clock Net 来产生有界频率，以在训练过程中稳定这种调制。在理论上，正弦调制提升了低秩适配器的秩，而频率分离则使不同任务的权重去相关。在稠密预测基准上，Free Sinewich 取得了最优的性能—效率权衡（例如，仅用 6.53M 可训练参数，相比单任务微调最高提升 5.39%），提供了一种基于频率参数共享的紧凑且可扩展的范式。
英文原文摘要 1134 字符
Multi-task learning (MTL) aims to enable a single model to solve multiple tasks efficiently; however, current parameter-efficient fine-tuning (PEFT) methods remain largely limited to single-task adaptation. We introduce Free Sinewich, a parameter-efficient multi-task learning framework that enables near-zero-cost weight modulation via frequency switching (Free). Specifically, a Sine-AWB (Sinewich) layer combines low-rank factors and convolutional priors into a single kernel, which is then modulated elementwise by a sinusoidal transformation to produce task-specialized weights. A lightweight Clock Net is introduced to produce bounded frequencies that stabilize this modulation during training. Theoretically, sine modulation enhances the rank of low-rank adapters, while frequency separation decorrelates the weights of different tasks. On dense prediction benchmarks, Free Sinewich achieves state-of-the-art performance-efficiency trade-offs (e.g., up to +5.39% improvement over single-task fine-tuning with only 6.53M trainable parameters), offering a compact and scalable paradigm based on frequency-based parameter sharing.