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

For multilingual sequence-to-sequence pretrained language models(multilingual Seq2Seq PLMs), e.g. mBART, the self-supervised pretraining taskis trained on a wide range of monolingual languages, e.g. 25 languages fromCommonCrawl, while the downstream cross-lingual tasks generally progress on abilingual language subset, e.g. English-German, making there exists the datadiscrepancy, namely domain discrepancy, and cross-lingual learning objectivediscrepancy, namely task discrepancy, between the pretraining and finetuningstages. To bridge the above cross-lingual domain and task gaps, we extend thevanilla pretrain-finetune pipeline with extra code-switching restore task.Specifically, the first stage employs the self-supervised code-switchingrestore task as a pretext task, allowing the multilingual Seq2Seq PLMs toacquire some in-domain alignment information. And for the second stage, wefine-tune the model on downstream data normally. Experiments on both NLGevaluation (12 bilingual translation tasks, 30 zero-shot translation tasks, and2 cross-lingual summarization tasks) and NLU evaluation (7 cross-lingualnatural language inference tasks) show our model outperforms the strongbaseline mBART with standard finetuning strategy, consistently. Analysesindicate our approach could narrow the Euclidean distance of cross-lingualsentence representations, and improve the model generalization with trivialcomputational cost. We release the code at:https://github.com/zanchangtong/CSR4mBART.

Bridging Cross-Lingual Gaps During Leveraging the Multilingual Sequence-to-Sequence Pretraining for Text Generation and Understanding