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

Large multilingual language models typically rely on a single vocabularyshared across 100+ languages. As these models have increased in parameter countand depth, vocabulary size has remained largely unchanged. This\textit{vocabulary bottleneck} limits the representational capabilities ofmultilingual models like XLM-R. In this paper, we introduce a new approach forscaling to very large multilingual vocabularies by de-emphasizing token sharingbetween languages with little lexical overlap and assigning vocabulary capacityto achieve sufficient coverage for each individual language. Tokenizationsusing our vocabulary are typically more semantically meaningful and shortercompared to XLM-R. Leveraging this improved vocabulary, we train XLM-V, amultilingual language model with a one million token vocabulary. XLM-Voutperforms XLM-R on every task we tested on ranging from natural languageinference (XNLI), question answering (MLQA, XQuAD, TyDiQA), to named entityrecognition (WikiAnn). XLM-V is particularly effective on low-resource languagetasks and outperforms XLM-R by 11.2% and 5.8% absolute on MasakhaNER andAmericas NLI, respectively.

XLM-V: Overcoming the Vocabulary Bottleneck in Multilingual Masked Language Models