The new version introduces Trust Attention, a new technique inspired by the human brain's ability to prioritize information from trusted sources, improving translation quality by up to 42% (see graph).
This innovation sets a new industry standard, marking the distance from traditional MT systems that are limited by the inability to distinguish between reliable data and inferior material during the training process.
ModernMT now uses a first-of-its-kind weighting system to prioritize learning from high-quality, qualified data, i.e. translations performed and reviewed by professional translators, over unverified content obtained from the web. As it did with the introduction of adaptivity, Translated was inspired by the human brain to develop this new technique. Just as humans evaluate multiple sources of information to identify the most reliable and credible ones, ModernMT V7 similarly identifies the most valuable training data and organizes its learning accordingly.
“ModernMT's ability to prioritize higher-quality data to improve the translation model is the most significant leap forward in machine translation since dynamic adaptivity was introduced five years ago,” said Marco Trombetti, CEO by Translated. “This innovation opens up new opportunities for companies to use machine translation to take the experience of their customers around the world to the next level. At the same time, it will help translators increase their productivity and earnings.”
The introduction of this new approach is a major step forward for companies seeking greater accuracy in translating large volumes of content or requiring a high level of machine translation engine customization. Furthermore, translators who integrate MT into their workflow will significantly benefit from this innovation.
Today there is much discussion about the application of large language model (LLM) in Translation. While traditional machine translation prioritizes accuracy over fluency, LLMs tend to emphasize fluency. This can sometimes lead to misleading results due to "hallucinations", i.e. outputs not based on the inputs received from the data provided in the training phase. We believe Translated's Trust Attention can improve the accuracy of generative models, reducing the chances of such errors. This could set the stage for the next era of machine translation.
All Translated customers will benefit from the benefits of improving the quality of the new machine translation model, which will result in a significant reduction in project delivery times. Translators who collaborate with Translated will be able to take advantage of the features of the new model through Matecat, the AI-assisted translation web app (CAT tool) distributed free of charge by Translated. Translators using one of the other officially supported software (Matecat, memoQ and Trados) with an active ModernMT license will also experience the power of the new model.
BlogInnovazione.it
There are many tips and tricks for making great presentations. The objective of these rules is to improve the effectiveness, smoothness of…
"Protolabs Product Development Outlook" report released. Examine how new products are brought to market today.…
The term sustainability is now widely used to indicate programs, initiatives and actions aimed at preserving a particular resource.…
Any business operation produces a lot of data, even in different forms. Manually enter this data from an Excel sheet to…
The compromise of company emails increased more than double in the first three months of 2024 compared to the last quarter of…
The principle of interface segregation is one of the five SOLID principles of object-oriented design. A class should have…
Microsoft Excel is the reference tool for data analysis, because it offers many features for organizing data sets,…
Walliance, SIM and platform among the leaders in Europe in the field of Real Estate Crowdfunding since 2017, announces the completion…
