arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2604.01474cs.CVcs.LG
收藏

一次优先,然后局部重新编程:一种高效替代黑盒服务模型适应的方法

Prime Once, then Reprogram Locally: An Efficient Alternative to Black-Box Service Model Adaptation

Tulane University(杜兰大学);University of Melbourne(墨尔本大学)

Yunbei Zhang, Chengyi Cai, Feng Liu, Jihun Hamm

AI总结:

本文提出AReS方法,通过一次交互预训练局部编码器,减少API调用成本,提升现代闭盒模型的适应性能。

摘要:

适应闭盒服务模型(即API)通常依赖于零阶优化(ZOO)重新编程。然而,这种标准策略以大量的、昂贵的API调用为代价,且常导致缓慢、不稳定的优化。此外,我们发现这种范式在现代API(如GPT-4o)中面临新挑战。这些模型对ZOO依赖的输入扰动不敏感,从而阻碍性能提升。为解决这些限制,我们提出了一种替代高效的重新编程方法用于服务模型(AReS)。与直接连续闭盒优化不同,AReS启动一次交互与服务API,以预训练一个可接受的本地预训练编码器。此预训练阶段仅在本地编码器上训练一个轻量级层,使其对后续的玻璃盒(白盒)重新编程阶段(直接在本地模型上进行)高度敏感。因此,所有后续的适应和推理仅依赖于此本地代理,消除了所有进一步的API成本。实验表明,AReS在先前ZOO方法挣扎的任务上表现出色:在GPT-4o上,AReS比零样本基线高出27.8%,而ZOO方法提供很少或没有改进。广泛而言,在十个多样化的数据集上,AReS优于最先进的方法(VLMs提高2.5%,标准VMs提高15.6%),同时将API调用减少超过99.99%。因此,AReS为适应现代闭盒模型提供了稳健且实用的解决方案。

英文摘要:

Adapting closed-box service models (i.e., APIs) for target tasks typically relies on reprogramming via Zeroth-Order Optimization (ZOO). However, this standard strategy is known for extensive, costly API calls and often suffers from slow, unstable optimization. Furthermore, we observe that this paradigm faces new challenges with modern APIs (e.g., GPT-4o). These models can be less sensitive to the input perturbations ZOO relies on, thereby hindering performance gains. To address these limitations, we propose an Alternative efficient Reprogramming approach for Service models (AReS). Instead of direct, continuous closed-box optimization, AReS initiates a single-pass interaction with the service API to prime an amenable local pre-trained encoder. This priming stage trains only a lightweight layer on top of the local encoder, making it highly receptive to the subsequent glass-box (white-box) reprogramming stage performed directly on the local model. Consequently, all subsequent adaptation and inference rely solely on this local proxy, eliminating all further API costs. Experiments demonstrate AReS's effectiveness where prior ZOO-based methods struggle: on GPT-4o, AReS achieves a +27.8% gain over the zero-shot baseline, a task where ZOO-based methods provide little to no improvement. Broadly, across ten diverse datasets, AReS outperforms state-of-the-art methods (+2.5% for VLMs, +15.6% for standard VMs) while reducing API calls by over 99.99%. AReS thus provides a robust and practical solution for adapting modern closed-box models.

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。

↑