TAUS QE Metrics V2.0 and New QE Subscriptions

5 minute read

Discover the advancements in TAUS QE Metrics V2.0, featuring a state-of-the-art cross-lingual transformer architecture for precise translation quality predictions.

After extensive work from the TAUS NLP team, the TAUS Quality Estimation (QE) Metric in the TAUS Estimate API is now available in v2.0, representing a significant advancement in the precision of translation quality predictions. The latest release incorporates an advanced cross-lingual transformer architecture, augmented by meticulous data improvements and comprehensive fine-tuning. These enhancements have markedly increased the correlation with human evaluations, guaranteeing an increased level of accuracy in translation quality estimation.

The ROC curve demonstrates that TAUS QE v2 outperforms TAUS QE v1 regarding precision-recall balance and its capability to discern between "acceptable" and "unacceptable" translations. TAUS QE v2's elevated AUC value indicates its enhanced ability to distinguish between differing levels of translation quality.

Estimate v2 ROC Curve

Additionally, further customization for customer-specific domains and requirements lead to even more improvements in results.


The TAUS QE Metric2.0 was released on March 7. Read more here about how to switch to v2.0.


New Subscriptions 

TAUS is also introducing a new subscription offering to high-volume users of the Estimate API. This subscription combines the use of all TAUS data for custom model training with an NLP consulting retainer for a fixed monthly fee. The TAUS team trains and maintains customer-specific models making use of the vast collection of TAUS data. In consultation with the customer, the NLP team can integrate new custom features, such as source content evaluation and auto-correction, MQM/DQF error categorization, smart translation routing or automated post-editing.


About TAUS Estimate API

The TAUS Estimate API is a solution for automatic scoring of translation quality. The TAUS Estimate API can easily be integrated in existing platforms and content and translation workflows. Users of the Estimate API can save between 25% and 60% on human post-editing efforts and costs and it helps high-volume users in an MT-only setting to mitigate the risks of bad translation output. Read more here.


Learn More

To find out more, please join the TAUS Focus Webinar on March 28, or contact sales@taus.net to request a one-on-one meeting.


Dace is a product and operations management professional with 15+ years of experience in the localization industry. Over the past 7 years, she has taken on various roles at TAUS ranging from account management to product and operations management. Since 2020 she is a member of the Executive Team and leads the strategic planning and business operations of a team of 20+ employees. She holds a Bachelor’s degree in Translation and Interpreting and a Master’s degree in Social and Cultural Anthropology.

Related Articles
Find out how companies integrate QE into their workflows and explore real-world use cases and benefits of quality estimation. From mitigating risk in global chat communication to minimizing post-editing in machine translation workflows.
Learn about the distinctions between quality estimation and quality evaluation, the foundations of TAUS QE scores, and how to categorize and customize scores for optimal application.
Unlock the secrets of Machine Translation Quality Estimation (MTQE) with three key facts that demystify its distinctiveness, data-driven automation, and broader applications beyond cost reduction.