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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Fine-tuning pre-trained models has emerged as a powerful technique innumerous domains, owing to its ability to leverage enormous pre-existingknowledge and achieve remarkable performance on downstream tasks. However,updating the parameters of entire networks is computationally intensive.Although state-of-the-art parameter-efficient transfer learning (PETL) methodssignificantly reduce the trainable parameters and storage demand, almost all ofthem still need to back-propagate the gradients through large pre-trainednetworks. This memory-extensive characteristic extremely limits theapplicability of PETL methods in real-world scenarios. To this end, we proposea new memory-efficient PETL strategy, dubbed Universal Parallel Tuning (UniPT).Specifically, we facilitate the transfer process via a lightweight learnableparallel network, which consists of two modules: 1) A parallel interactionmodule that decouples the inherently sequential connections and processes theintermediate activations detachedly of the pre-trained network. 2) A confidenceaggregation module that learns optimal strategies adaptively for integratingcross-layer features. We evaluate UniPT with different backbones (e.g.,VSE$\infty$, CLIP4Clip, Clip-ViL, and MDETR) on five challengingvision-and-language tasks (i.e., image-text retrieval, video-text retrieval,visual question answering, compositional question answering, and visualgrounding). Extensive ablations on ten datasets have validated that our UniPTcan not only dramatically reduce memory consumption and outperform the bestmemory-efficient competitor, but also achieve higher performance than existingPETL methods in a low-memory scenario on different architectures. Our code ispublicly available at: https://github.com/Paranioar/UniPT.

UniPT: Universal Parallel Tuning for Transfer Learning with Efficient Parameter and Memory