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

Parameter-efficient tuning has become a trend in transferring large-scalefoundation models to downstream applications. Existing methods typically embedsome light-weight tuners into the backbone, where both the design and thelearning of the tuners are highly dependent on the base model. This work offersa new tuning paradigm, dubbed Res-Tuning, which intentionally unbinds tunersfrom the backbone. With both theoretical and empirical evidence, we show thatpopular tuning approaches have their equivalent counterparts under ourunbinding formulation, and hence can be integrated into our frameworkeffortlessly. Thanks to the structural disentanglement, we manage to free thedesign of tuners from the network architecture, facilitating flexiblecombination of various tuning strategies. We further propose a memory-efficientvariant of Res-Tuning, where the bypass i.e., formed by a sequence of tuners)is effectively detached from the main branch, such that the gradients areback-propagated only to the tuners but not to the backbone. Such a detachmentalso allows one-time backbone forward for multi-task inference. Extensiveexperiments on both discriminative and generative tasks demonstrate thesuperiority of our method over existing alternatives from the perspectives ofefficacy and efficiency. Project page:$\href{https://res-tuning.github.io/}{\textit{https://res-tuning.github.io/}}$.

Res-Tuning: A Flexible and Efficient Tuning Paradigm via Unbinding Tuner from Backbone