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

Moderate-sized large language models (LLMs) -- those with 7B or 13Bparameters -- exhibit promising machine translation (MT) performance. However,even the top-performing 13B LLM-based translation models, like ALMA, does notmatch the performance of state-of-the-art conventional encoder-decodertranslation models or larger-scale LLMs such as GPT-4. In this study, we bridgethis performance gap. We first assess the shortcomings of supervisedfine-tuning for LLMs in the MT task, emphasizing the quality issues present inthe reference data, despite being human-generated. Then, in contrast to SFTwhich mimics reference translations, we introduce Contrastive PreferenceOptimization (CPO), a novel approach that trains models to avoid generatingadequate but not perfect translations. Applying CPO to ALMA models with only22K parallel sentences and 12M parameters yields significant improvements. Theresulting model, called ALMA-R, can match or exceed the performance of the WMTcompetition winners and GPT-4 on WMT'21, WMT'22 and WMT'23 test datasets.

Contrastive Preference Optimization: Pushing the Boundaries of LLM Performance in Machine Translation