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

Structured pruning methods have proven effective in reducing the model sizeand accelerating inference speed in various network architectures such asTransformers. Despite the versatility of encoder-decoder models in numerous NLPtasks, the structured pruning methods on such models are relatively lessexplored compared to encoder-only models. In this study, we investigate thebehavior of the structured pruning of the encoder-decoder models in thedecoupled pruning perspective of the encoder and decoder component,respectively. Our findings highlight two insights: (1) the number of decoderlayers is the dominant factor of inference speed, and (2) low sparsity in thepruned encoder network enhances generation quality. Motivated by thesefindings, we propose a simple and effective framework, NASH, that narrows theencoder and shortens the decoder networks of encoder-decoder models. Extensiveexperiments on diverse generation and inference tasks validate theeffectiveness of our method in both speedup and output quality.

NASH: A Simple Unified Framework of Structured Pruning for Accelerating Encoder-Decoder Language Models