mlsanityData to 14 Oct 2026

Reviving In-domain Fine-tuning Methods for Source-Free Cross-domain Few-shot Learning

arXiv 2605.11659 ↗First posted 12 May 2026cs.CV cs.AIcited by 1

On the map: Multimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP

Abstract (arXiv)

Cross-Domain Few-Shot Learning (CDFSL) aims to adapt large-scale pretrained models to specialized target domains with limited samples, yet the few-shot fine-tuning of vision-language models like CLIP remains underexplored. By establishing multiple fine-tuning baselines of CLIP for CDFSL, we find adapter-based methods (e.g., LoRA) consistently outperform prompt-based ones (e.g., MaPLe), contrary to in-domain scenarios. To make those effective in-domain methods competitive again in CDFSL, we analyze this phenomenon and discover LoRA's superiority stems from rectifying the collapsed attention of visual CLS token, enhancing modality alignment and class separation by focusing on text-related visual regions. Further, we find textual EOS token exhibit much better attention to visual samples, and CLIP's standard contrastive loss weakly constrains modality alignment. Based on these insights, we propose Semantic Probe, a plug-and-play attention rectification framework for both adapter- and prompt-based methods. Extensive experiments on four CDFSL benchmarks validate our rationale, achieving state-of-the-art performance and benefiting both fine-tuning paradigms. Codes will be released.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Prompt learning and adaptation of CLIP, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 21 of its 37 references that are in the map. First posted on arXiv 12 May 2026.

Builds on

21 of 37 references.

Closest ancestors · 3

Same direction, newest first

  1. 2024FLAIR: VLM with Fine-grained Language-informed Image Representations4 Dec
  2. 2024ResCLIP: Residual Attention for Training-free Dense Vision-language Inference24 Nov
  3. 2021Tip-Adapter: Training-free CLIP-Adapter for Better Vision-Language Modeling6 Nov

Other references · 18

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2024The Llama 3 Herd of Modelscited by 14,715
  3. 2018Representation Learning with Contrastive Predictive Codingcited by 7,777
  4. 2023Sigmoid Loss for Language Image Pre-Trainingcited by 3,109
  5. 2017EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classificationcited by 1,496
  6. 2017ChestX-ray8: Hospital-scale Chest X-ray Database and Benchmarks on Weakly-Supervised Classification and Localization of Common Thorax Diseasescited by 972
  7. 2019Skin Lesion Analysis Toward Melanoma Detection 2018: A Challenge Hosted by the International Skin Imaging Collaboration (ISIC)cited by 669
  8. 2023Self-regulating Prompts: Foundational Model Adaptation without Forgettingcited by 272
  9. 2016Using Deep Learning for Image-Based Plant Disease Detectioncited by 219
  10. 2020Cross-Domain Few-Shot Classification via Learned Feature-Wise Transformationcited by 185
  11. 2022Pushing the Limits of Simple Pipelines for Few-Shot Learning: External Data and Fine-Tuning Make a Differencecited by 126
  12. 2021Cross-Domain Few-Shot Classification via Adversarial Task Augmentationcited by 63
6 more
  1. 2024LP++: A Surprisingly Strong Linear Probe for Few-Shot CLIPcited by 60
  2. 2021Boosting the Generalization Capability in Cross-Domain Few-shot Learning via Noise-enhanced Supervised Autoencodercited by 51
  3. 2024Mitigate the Gap: Investigating Approaches for Improving Cross-Modal Alignment in CLIPcited by 23
  4. 2024Prompt as Free Lunch: Enhancing Diversity in Source-Free Cross-domain Few-shot Learning through Semantic-Guided Promptingcited by 10
  5. 2016Doubly Stochastic Neighbor Embedding on Spherescited by 9
  6. 2024Reconstruction Target Matters in Masked Image Modeling for Cross-Domain Few-Shot Learningcited by 9

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026When Semantics Saturate or Emerge: Adaptation-Conditional Semantic Utility in Source-Free Cross-Domain Few-Shot Learningcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Addressing Exacerbated Attention Sink for Source-Free Cross-Domain Few-Shot Learning
  2. 2026Improving CLIP Adaptation by Breaking Tail Alignment for Source-Free Cross-Domain Few-Shot Learning
  3. 2026Interpretable Cross-Domain Few-Shot Learning with Rectified Target-Domain Local Alignment
  4. 2026Mind the Discriminability Trap in Source-Free Cross-domain Few-shot Learning
  5. 2023Robust Fine-Tuning of Vision-Language Models for Domain Generalization
  6. 2024Fully Fine-tuned CLIP Models are Efficient Few-Shot Learners
  7. 2025Open Set Domain Adaptation with Vision-language models via Gradient-aware Separation
  8. 2025Multiple Stochastic Prompt Tuning for Few-shot Adaptation under Extreme Domain Shift