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

Large Language Models (LLMs) present strong general capabilities, and acurrent compelling challenge is stimulating their specialized capabilities,such as machine translation, through low-cost instruction tuning. The standardinstruction-following data is sequentially organized as the concatenation of aninstruction, an input, and a response. As the attention mechanism of LLMs haslimitations on local focus, LLMs tend to focus more on the words or sentencesnearby at each position. This leads to a high risk of instruction forgettingduring decoding. To alleviate the above issues, We propose SWIE(Segment-Weighted Instruction Embedding) and an instruction-following datasetOVERMISS. SWIE improves the model instruction understanding by adding a globalinstruction representation on the following input and response representations.OVERMISS improves model faithfulness by comparing over-translation andmiss-translation results with the correct translation. We apply our methods totwo main-stream open-source LLMs, BLOOM and LLaMA. The experimental resultsdemonstrate significant improvements in translation performance with SWIE basedon BLOOMZ-3b, particularly in zero-shot and long text translations due toreduced instruction forgetting risk. Additionally, OVERMISS outperforms thebaseline in translation performance (e.g. an increase in BLEU scores from 0.69to 3.12 and an average improvement of 0.48 percentage comet scores forLLaMA-7b) with further enhancements seen in models combining OVERMISS and SWIE(e.g. the BLUE scores increase up to 0.56 from English to German across threedifferent backbones), and both exhibit improvements in the faithfulness metricbased on word alignment.

Improving Translation Faithfulness of Large Language Models via Augmenting Instructions