English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Visual foundation models like CLIP excel in learning feature representationsfrom extensive datasets through self-supervised methods, demonstratingremarkable transfer learning and generalization capabilities. A growing numberof applications based on visual foundation models are emerging, includinginnovative solutions such as BLIP-2. These applications employ pre-trained CLIPmodels as upstream feature extractors and train various downstream modules toaccomplish diverse tasks. In situations involving system upgrades that requireupdating the upstream foundation model, it becomes essential to re-train alldownstream modules to adapt to the new foundation model, which is inflexibleand inefficient. In this paper, we introduce a parameter-efficient andtask-agnostic adapter, dubbed TaCA, that facilitates compatibility acrossdistinct foundation models while ensuring enhanced performance for the newmodels. TaCA allows downstream applications to seamlessly integratebetter-performing foundation models without necessitating retraining. Weconduct extensive experimental validation of TaCA using different scales ofmodels with up to one billion parameters on various tasks such as video-textretrieval, video recognition, and visual question answering. The resultsconsistently demonstrate the emergent ability of TaCA on hot-plugging upgradesfor visual foundation models. Codes and models will be available athttps://github.com/TencentARC/TaCA.

TaCA: Upgrading Your Visual Foundation Model with Task-agnostic Compatible Adapter