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

Neural Machine Translation systems built on top of Transformer-basedarchitectures are routinely improving the state-of-the-art in translationquality according to word-overlap metrics. However, a growing number of studiesalso highlight the inherent gender bias that these models incorporate duringtraining, which reflects poorly in their translations. In this work, weinvestigate whether these models can be instructed to fix their bias duringinference using targeted, guided instructions as contexts. By translatingrelevant contextual sentences during inference along with the input, we observelarge improvements in reducing the gender bias in translations, across threepopular test suites (WinoMT, BUG, SimpleGen). We further propose a novel metricto assess several large pre-trained models (OPUS-MT, M2M-100) on theirsensitivity towards using contexts during translation to correct their biases.Our approach requires no fine-tuning and thus can be used easily in productionsystems to de-bias translations from stereotypical gender-occupation bias 1. Wehope our method, along with our metric, can be used to build better, bias-freetranslation systems.

How sensitive are translation systems to extra contexts? Mitigating gender bias in Neural Machine Translation models through relevant contexts