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

Multilingual language model (LM) have become a powerful tool in NLPespecially for non-English languages. Nevertheless, model parameters ofmultilingual LMs remain large due to the larger embedding matrix of thevocabulary covering tokens in different languages. On the contrary, monolingualLMs can be trained in a target language with the language-specific vocabularyonly, but this requires a large budget and availability of reliable corpora toachieve a high-quality LM from scratch. In this paper, we proposevocabulary-trimming (VT), a method to reduce a multilingual LM vocabulary to atarget language by deleting irrelevant tokens from its vocabulary. In theory,VT can compress any existing multilingual LM to build monolingual LMs in anylanguage covered by the multilingual LM. In our experiments, we show that VTcan retain the original performance of the multilingual LM, while being smallerin size (in general around 50% of the original vocabulary size is enough) thanthe original multilingual LM. The evaluation is performed over four NLP tasks(two generative and two classification tasks) among four widely usedmultilingual LMs in seven languages. Finally, we show that this methodology cankeep the best of both monolingual and multilingual worlds by keeping a smallsize as monolingual models without the need for specifically retraining them,and even limiting potentially harmful social biases.

An Efficient Multilingual Language Model Compression through Vocabulary Trimming