DA-MergeLoRA: Hypernetwork-Based LoRA Merging for Few-Shot Test-Time Domain Adaptation

2026-07-19

AI Summary

This work addresses the challenge of effectively leveraging source-domain knowledge in few-shot test-time domain adaptation. The authors propose a novel approach that combines parameter-efficient LoRA fine-tuning with test-time model fusion. Specifically, they independently fine-tune a LoRA module on the CLIP visual encoder for each source domain and introduce a meta-learning-driven hypernetwork that dynamically generates column-wise fusion factors to enable fine-grained integration of multiple LoRA modules into a target-domain-adapted representation. This method is the first to integrate LoRA-based adaptation with test-time fusion, achieving significant performance gains over existing approaches across multiple benchmark datasets while preserving the generalization capability of the base model.

Abstract

Few-shot Test-Time Domain Adaptation (FSTT-DA) seeks to adapt models to novel domains using only a handful of unlabeled target samples. This setting is more realistic than typical domain adaptation setups, which assume access to target data during source training. However, prior FSTT-DA approaches fail to effectively leverage source domain-specific knowledge, relying on shallow batch normalization updates, prompt-based methods that treat the model as a black box, or ensembling strategies that do not capture cross-domain relationships. To address these limitations, we introduce a new FSTT-DA framework that integrates LoRA fine-tuning with model merging. In our approach, separate LoRA modules are fine-tuned on CLIP's vision encoder for each source domain. Since LoRA modifies only a small fraction of the model's parameters, it retains the base model's generalized knowledge while internally learning domain-specific features. To adapt the learned knowledge to a specific target domain, we propose a hypernetwork trained via meta-learning that generates per-column merging factors to combine LoRA modules. Given a small batch of target images, the hypernetwork produces merging weights that fuse source LoRA modules into a single adapted representation. Our results demonstrate state-of-the-art performance across various domain adaptation datasets. Our code is publicly available at https://github.com/nahbois4321/DA-MergeLoRA.

Problem

• Few-shot Test-Time Domain Adaptation
• Domain Adaptation
• LoRA
• Hypernetwork
• Model Merging

Innovation

• LoRA merging
• hypernetwork
• few-shot test-time domain adaptation
• meta-learning
• CLIP