一次预热，本地重编程：黑盒服务模型适配的高效替代方案
Prime Once, then Reprogram Locally: An Efficient Alternative to Black-Box Service Model Adaptation
对做视觉模型部署与迁移的研究者，这篇给出了一个可复用的范式：先用少量 API 调用把本地预训练编码器「预热」成可接收黑盒模型知识的代理，再在本地做白盒重编程，从而把黑盒适配的调用成本压到近乎为零；其轻量层预热 + 本地重编程的两阶段流程可直接套用到 VLM 或标准视觉模型的下游任务适配中。
原文依据 AReS initiates a single-pass interaction with the service API to prime an amenable local pre-trained encoder.
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论将闭盒服务模型（即 API）适配到目标任务通常依赖零阶优化（ZOO）进行重编程。然而，这一标准策略以大量且昂贵的 API 调用著称，并且常常面临优化缓慢、不稳定的问题。此外，我们观察到该范式在现代 API（如 GPT-4o）上面临新的挑战：这类模型对 ZOO 所依赖的输入扰动不够敏感，从而阻碍了性能提升。为解决上述局限，我们提出一种面向服务模型的替代性高效重编程方法（AReS）。AReS 不进行直接、连续的闭盒优化，而是先与服务 API 进行一次单遍交互，以预热一个易于适配的本地预训练编码器。该预热阶段仅在本地编码器之上训练一个轻量层，使其对随后直接在本地模型上进行的白盒重编程阶段具有高度接受性。因此，后续所有适配与推理都仅依赖这一本地代理，从而消除了全部进一步的 API 成本。实验表明，在先前基于 ZOO 的方法难以奏效的场景中 AReS 依然有效：在 GPT-4o 上，AReS 相比零样本基线取得 +27.8% 的提升，而基于 ZOO 的方法在该任务上几乎无提升。总体而言，在十个多样化数据集上，AReS 优于最优方法（视觉语言模型 +2.5%，标准视觉模型 +15.6%），同时将 API 调用减少 99.99% 以上。因此，AReS 为适配现代闭盒模型提供了一种稳健且实用的解决方案。
英文原文摘要 1551 字符
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.