MaPLe: Multi-modal Prompt Learning
By Muhammad Uzair Khattak, Hanoona Abdul Rasheed, Muhammad Maaz, Salman Hameed Khan, Fahad Shahbaz Khan. 2022
1,127 citations recorded by the source corpus.
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.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Prompting & Reasoning, Multimodal Learning
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
In Computer Vision and Pattern Recognition
- Hongsheng Li
- Yu Qiao
- Luc Van Gool
- Ming-Hsuan Yang
- Fahad Shahbaz Khan
- Xiaogang Wang
- Peng Gao
- Dacheng Tao
In Prompting & Reasoning
- Yu Qiao
- Peng Gao
- Renrui Zhang
- Muhammad Uzair Khattak
- Shilin Yan
- Fahad Shahbaz Khan
- Pan Lu
- Hongsheng Li
In Multimodal Learning
Citations in the corpus
- LLaMA-Adapter: Efficient Fine-tuning of Language Models with Zero-init Attention (2023, background): “Alpaca: Five words that rhyme with shock are: rock, pop, shock, cook, and snock.”
- Prompt-aligned Gradient for Prompt Tuning (2022, methodology): “The prompt-based learning has recently been applied in CV community [54, 29, 7, 46, 4, 20, 56]: CoOp [54] uses a continuous prompt optimization from downstream data instead of hand-craft design.”
- Self-regulating Prompts: Foundational Model Adaptation without Forgetting (2023, methodology): “Unlike prior prompting approaches that improve generalization mainly from the model architecture perspective [58, 22], we motivate our work from the regularization perspective.”
- Parameter-Efficient Fine-Tuning for Large Models: A Comprehensive Survey (2024, neutral): “MaPLe [219] notes that existing methods learn prompts either in the language or in the vision branch of CLIP, which is not efficient in leveraging the multimodal nature of VLAs.”
- AnomalyCLIP: Object-agnostic Prompt Learning for Zero-shot Anomaly Detection (2023, positive): “2, inspired by Jia et al. (2022) and Khattak et al. (2023), we introduce randomly initialized learnable token embeddings in the text encoder to refine the textual space for its adaptation to AD.”
- Vision-Language Models for Vision Tasks: A Survey (2023, background): “MAPLE [151] conducts multi-modal prompt tuning by aligning visual prompts with their corresponding language prompts, enabling a mutual promotion between text prompts and image prompts.”