MaPLe: Multi-modal Prompt Learning

TL;DR

MaPLe addresses the limitation of uni-modal prompting in CLIP by introducing deep, multi-modal prompts for both vision and language branches, tied together with a learnable V-L coupling function. The approach promotes mutual synergy, enabling stepwise, stage-wise context modeling across transformer blocks. Empirically, MaPLe achieves consistent improvements over state-of-the-art Co-CoOp across base-to-novel generalization, cross-dataset transfer, and domain generalization on 11 diverse datasets, with favorable efficiency. This work highlights the value of jointly optimizing both modalities for robust vision-language adaptation in few-shot and open-set settings.

Abstract

Pre-trained vision-language (V-L) models such as CLIP have shown excellent generalization ability to downstream tasks. However, they are sensitive to the choice of input text prompts and require careful selection of prompt templates to perform well. Inspired by the Natural Language Processing (NLP) literature, recent CLIP adaptation approaches learn prompts as the textual inputs to fine-tune CLIP for downstream tasks. We note that using prompting to adapt representations in a single branch of CLIP (language or vision) is sub-optimal since it does not allow the flexibility to dynamically adjust both representation spaces on a downstream task. In this work, we propose Multi-modal Prompt Learning (MaPLe) for both vision and language branches to improve alignment between the vision and language representations. Our design promotes strong coupling between the vision-language prompts to ensure mutual synergy and discourages learning independent uni-modal solutions. Further, we learn separate prompts across different early stages to progressively model the stage-wise feature relationships to allow rich context learning. We evaluate the effectiveness of our approach on three representative tasks of generalization to novel classes, new target datasets and unseen domain shifts. Compared with the state-of-the-art method Co-CoOp, MaPLe exhibits favorable performance and achieves an absolute gain of 3.45% on novel classes and 2.72% on overall harmonic-mean, averaged over 11 diverse image recognition datasets. Our code and pre-trained models are available at https://github.com/muzairkhattak/multimodal-prompt-learning.

Figures (4)

• Figure 1: Comparison of MaPLe with standard prompt learning methods. (a) Existing methods adopt uni-modal prompting techniques to fine-tune CLIP representations as prompts are learned only in a single branch of CLIP (language or vision). (b) MaPLe introduces branch-aware hierarchical prompts that adapt both language and vision branches simultaneously for improved generalization. (c) MaPLe surpasses state-of-the-art methods on 11 diverse image recognition datasets for novel class generalization task.
• Figure 2: Overview of our proposed MaPLe (Multi-modal Prompt Learning) framework for prompt learning in V-L models. MaPLe tunes both vision and language branches where only the context prompts are learned, while the rest of the model is frozen. MaPLe conditions the vision prompts on language prompts via a V-L coupling function $\mathcal{F}$ to induce mutual synergy between the two modalities. Our framework uses deep contextual prompting where separate context prompts are learned across multiple transformer blocks.
• Figure 3: t-SNE plots of image embeddings in uni-modal prompting method Co-CoOp, and MaPLe on 3 diverse image recognition datasets. MaPLe shows better separability in both base and novel classes.
• Figure 4: Ablation on prompt depth (left) and prompt length (right) in MaPLe. We report average results on the held-out validation sets of all datasets.