# Adapting Multimodal Foundation Models for Few-Shot Learning: A Comprehensive Study on Contrastive Captioners

arXiv: https://arxiv.org/abs/2512.12824 · HTML: https://mlsanity.com/paper/2512.12824/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 11 of its 15 references that are in the map.
- First posted on arXiv 14 Dec 2025.

## Builds on (most cited first, 11 of 11)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Parameter-Efficient Transfer Learning for NLP](https://mlsanity.com/paper/1902.00751/) (2019)
- [Prefix-Tuning: Optimizing Continuous Prompts for Generation](https://mlsanity.com/paper/2101.00190/) (2021)
- [Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervision](https://mlsanity.com/paper/2102.05918/) (2021)
- [Supervised Contrastive Learning](https://mlsanity.com/paper/2004.11362/) (2020)
- [Learning to Prompt for Vision-Language Models](https://mlsanity.com/paper/2109.01134/) (2021)
- [Matching Networks for One Shot Learning](https://mlsanity.com/paper/1606.04080/) (2016)
- [CoCa: Contrastive Captioners are Image-Text Foundation Models](https://mlsanity.com/paper/2205.01917/) (2022)
- [SimVLM: Simple Visual Language Model Pretraining with Weak Supervision](https://mlsanity.com/paper/2108.10904/) (2021)
- [Tip-Adapter: Training-free CLIP-Adapter for Better Vision-Language Modeling](https://mlsanity.com/paper/2111.03930/) (2021)

## Abstract (arXiv)

Large-scale multimodal foundation models, particularly Contrastive Captioners (CoCa), have achieved state-of-the-art results by unifying contrastive alignment with generative captioning. While zero-shot transfer capabilities are well-documented, the adaptation of these generative-contrastive hybrids to downstream tasks with extreme data scarcity (few-shot learning) remains under-explored. Existing literature predominantly focuses on dual-encoder architectures like CLIP, leaving a gap in understanding how CoCa's distinct latent space responds to parameter-efficient fine-tuning (PEFT). This paper presents a comprehensive empirical study on adapting the CoCa visual backbone for few-shot image classification. We systematically evaluate a hierarchy of strategies, ranging from training-free hybrid prototyping to deep parameter adaptation via Low-Rank Adaptation (LoRA). First, we identify an "augmentation divergence": while strong data augmentation degrades the performance of linear probing in low-shot settings, it is essential for stabilizing LoRA fine-tuning. We also demonstrate that hybrid objectives incorporating Supervised Contrastive (SupCon) loss yield consistent performance improvements over standard Cross-Entropy across varying shot counts. Crucially, we characterize the sensitivity of training configurations to data scarcity, providing empirical reference settings for scaling regularization, rank, and sampling strategies to facilitate the efficient adaptation of generative-contrastive foundation models.