A Wander Through the Multimodal Landscape: Efficient Transfer Learning via Low-rank Sequence Multimodal Adapter
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866915060984578048 |
|---|---|
| author | Guo, Zirun Cheng, Xize Wu, Yangyang Jin, Tao |
| author_facet | Guo, Zirun Cheng, Xize Wu, Yangyang Jin, Tao |
| contents | Efficient transfer learning methods such as adapter-based methods have shown great success in unimodal models and vision-language models. However, existing methods have two main challenges in fine-tuning multimodal models. Firstly, they are designed for vision-language tasks and fail to extend to situations where there are more than two modalities. Secondly, they exhibit limited exploitation of interactions between modalities and lack efficiency. To address these issues, in this paper, we propose the loW-rank sequence multimodal adapter (Wander). We first use the outer product to fuse the information from different modalities in an element-wise way effectively. For efficiency, we use CP decomposition to factorize tensors into rank-one components and achieve substantial parameter reduction. Furthermore, we implement a token-level low-rank decomposition to extract more fine-grained features and sequence relationships between modalities. With these designs, Wander enables token-level interactions between sequences of different modalities in a parameter-efficient way. We conduct extensive experiments on datasets with different numbers of modalities, where Wander outperforms state-of-the-art efficient transfer learning methods consistently. The results fully demonstrate the effectiveness, efficiency and universality of Wander. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_08979 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | A Wander Through the Multimodal Landscape: Efficient Transfer Learning via Low-rank Sequence Multimodal Adapter Guo, Zirun Cheng, Xize Wu, Yangyang Jin, Tao Machine Learning Computer Vision and Pattern Recognition Efficient transfer learning methods such as adapter-based methods have shown great success in unimodal models and vision-language models. However, existing methods have two main challenges in fine-tuning multimodal models. Firstly, they are designed for vision-language tasks and fail to extend to situations where there are more than two modalities. Secondly, they exhibit limited exploitation of interactions between modalities and lack efficiency. To address these issues, in this paper, we propose the loW-rank sequence multimodal adapter (Wander). We first use the outer product to fuse the information from different modalities in an element-wise way effectively. For efficiency, we use CP decomposition to factorize tensors into rank-one components and achieve substantial parameter reduction. Furthermore, we implement a token-level low-rank decomposition to extract more fine-grained features and sequence relationships between modalities. With these designs, Wander enables token-level interactions between sequences of different modalities in a parameter-efficient way. We conduct extensive experiments on datasets with different numbers of modalities, where Wander outperforms state-of-the-art efficient transfer learning methods consistently. The results fully demonstrate the effectiveness, efficiency and universality of Wander. |
| title | A Wander Through the Multimodal Landscape: Efficient Transfer Learning via Low-rank Sequence Multimodal Adapter |
| topic | Machine Learning Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2412.08979 |