# Anchor-based Robust Finetuning of Vision-Language Models

arXiv: https://arxiv.org/abs/2404.06244 · HTML: https://mlsanity.com/paper/2404.06244/

## Key facts

- Cited by 11 papers 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 29 of its 43 references that are in the map.
- First posted on arXiv 9 Apr 2024.

## Placed in

- Multimodal LLMs / vision-language models / [Prompt learning and adaptation of CLIP](https://mlsanity.com/mllm/prompt-learning-and-adaptation-of-clip/)

## Builds on (most cited first, 29 of 29)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://mlsanity.com/paper/2301.12597/) (2023)
- [BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation](https://mlsanity.com/paper/2201.12086/) (2022)
- [Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervision](https://mlsanity.com/paper/2102.05918/) (2021)
- [Billion-scale similarity search with GPUs](https://mlsanity.com/paper/1702.08734/) (2017)
- [Learning to Prompt for Vision-Language Models](https://mlsanity.com/paper/2109.01134/) (2021)
- [Describing Textures in the Wild](https://mlsanity.com/paper/1311.3618/) (2013)
- [The Many Faces of Robustness: A Critical Analysis of Out-of-Distribution Generalization](https://mlsanity.com/paper/2006.16241/) (2020)
- [Fine-Grained Visual Classification of Aircraft](https://mlsanity.com/paper/1306.5151/) (2013)
- [Do ImageNet Classifiers Generalize to ImageNet?](https://mlsanity.com/paper/1902.10811/) (2019)
- [EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classification](https://mlsanity.com/paper/1709.00029/) (2017)
- [Conditional Prompt Learning for Vision-Language Models](https://mlsanity.com/paper/2203.05557/) (2022)
- [Natural Adversarial Examples](https://mlsanity.com/paper/1907.07174/) (2019)
- [WILDS: A Benchmark of in-the-Wild Distribution Shifts](https://mlsanity.com/paper/2012.07421/) (2020)
- [Moment Matching for Multi-Source Domain Adaptation](https://mlsanity.com/paper/1812.01754/) (2018)
- [Learning Robust Global Representations by Penalizing Local Predictive Power](https://mlsanity.com/paper/1905.13549/) (2019)
- [Robust fine-tuning of zero-shot models](https://mlsanity.com/paper/2109.01903/) (2021)
- [Fine-Tuning can Distort Pretrained Features and Underperform Out-of-Distribution](https://mlsanity.com/paper/2202.10054/) (2022)
- [MaPLe: Multi-modal Prompt Learning](https://mlsanity.com/paper/2210.03117/) (2022)
- [FLAVA: A Foundational Language And Vision Alignment Model](https://mlsanity.com/paper/2112.04482/) (2021)
- [Scaling Language-Image Pre-training via Masking](https://mlsanity.com/paper/2212.00794/) (2022)
- [MaskCLIP: Masked Self-Distillation Advances Contrastive Language-Image Pretraining](https://mlsanity.com/paper/2208.12262/) (2022)
- [Fine-tuned CLIP Models are Efficient Video Learners](https://mlsanity.com/paper/2212.03640/) (2022)
- [Finetune like you pretrain: Improved finetuning of zero-shot vision models](https://mlsanity.com/paper/2212.00638/) (2022)
- [Prompt-based Distribution Alignment for Unsupervised Domain Adaptation](https://mlsanity.com/paper/2312.09553/) (2023)
- [Trainable Projected Gradient Method for Robust Fine-tuning](https://mlsanity.com/paper/2303.10720/) (2023)
- [Retrieval-Enhanced Contrastive Vision-Text Models](https://mlsanity.com/paper/2306.07196/) (2023)
- [CLIP Itself is a Strong Fine-tuner: Achieving 85.7% and 88.0% Top-1 Accuracy with ViT-B and ViT-L on ImageNet](https://mlsanity.com/paper/2212.06138/) (2022)

## Built on by (most cited first, 11 of 11)

- [A Practitioner's Guide to Continual Multimodal Pretraining](https://mlsanity.com/paper/2408.14471/) (2024)
- [SAFE: Slow and Fast Parameter-Efficient Tuning for Continual Learning with Pre-Trained Models](https://mlsanity.com/paper/2411.02175/) (2024)
- [Robust Fine-tuning of Zero-shot Models via Variance Reduction](https://mlsanity.com/paper/2411.06966/) (2024)
- [Fine Tuning without Catastrophic Forgetting via Selective Low Rank Adaptation](https://mlsanity.com/paper/2501.15377/) (2025)
- [Context-Aware Multimodal Pretraining](https://mlsanity.com/paper/2411.15099/) (2024)
- [Infusing fine-grained visual knowledge to Vision-Language Models](https://mlsanity.com/paper/2508.12137/) (2025)
- [A Hierarchical Test Platform for Vision Language Model (VLM)-Integrated Real-World Autonomous Driving](https://mlsanity.com/paper/2506.14100/) (2025)
- [Decoupling Vision and Language: Codebook Anchored Visual Adaptation](https://mlsanity.com/paper/2602.19449/) (2026)
- [TRACER: Persistent Regularization for Robust Multimodal Finetuning](https://mlsanity.com/paper/2605.29380/) (2026)
- [Re-calibrated Contrastive Loss for Transformation-Aware Prompt Conditioning in Vision-Language Models](https://mlsanity.com/paper/2609.06967/) (2026)
- [RARA: Robust Policy Fine-Tuning through Retrieval-Anchored Representation Alignment for Robotic Manipulation](https://mlsanity.com/paper/2610.13578/) (2026)

## Abstract (arXiv)

We aim at finetuning a vision-language model without hurting its out-of-distribution (OOD) generalization. We address two types of OOD generalization, i.e., i) domain shift such as natural to sketch images, and ii) zero-shot capability to recognize the category that was not contained in the finetune data. Arguably, the diminished OOD generalization after finetuning stems from the excessively simplified finetuning target, which only provides the class information, such as ``a photo of a [CLASS]''. This is distinct from the process in that CLIP was pretrained, where there is abundant text supervision with rich semantic information. Therefore, we propose to compensate for the finetune process using auxiliary supervision with rich semantic information, which acts as anchors to preserve the OOD generalization. Specifically, two types of anchors are elaborated in our method, including i) text-compensated anchor which uses the images from the finetune set but enriches the text supervision from a pretrained captioner, ii) image-text-pair anchor which is retrieved from the dataset similar to pretraining data of CLIP according to the downstream task, associating with the original CLIP text with rich semantics. Those anchors are utilized as auxiliary semantic information to maintain the original feature space of CLIP, thereby preserving the OOD generalization capabilities. Comprehensive experiments demonstrate that our method achieves in-distribution performance akin to conventional finetuning while attaining new state-of-the-art results on domain shift and zero-shot learning benchmarks.