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

Sparsely gated Mixture of Experts (MoE) models have been shown to be acompute-efficient method to scale model capacity for multilingual machinetranslation. However, for low-resource tasks, MoE models severely over-fit. Weshow effective regularization strategies, namely dropout techniques for MoElayers in EOM and FOM, Conditional MoE Routing and Curriculum Learning methodsthat prevent over-fitting and improve the performance of MoE models onlow-resource tasks without adversely affecting high-resource tasks. On amassively multilingual machine translation benchmark, our strategies result inabout +1 chrF++ improvement in very low resource language pairs. We perform anextensive analysis of the learned MoE routing to better understand the impactof our regularization methods and how we can improve them.

Fixing MoE Over-Fitting on Low-Resource Languages in Multilingual Machine Translation