This innovation sets a new industry standard
Posted: Sat Feb 01, 2025 7:21 am
Translated has announced the release of ModernMT version 7, a significant update to its adaptive machine translation (MT) system.
Trust attention
The new version introduces Trust Attention, a new technology inspired by the human brain's ability to prioritize information from trusted sources, which improves translation quality by up to 42% (see graph).
Marking a departure from traditional MT systems that are limited by the inability to distinguish between reliable data and low-quality content during the training process.
ModernMT now uses a first-of-its-kind weighting system to prioritize learning from high- portugal rcs data quality, qualified data, i.e., unverified content sourced from the web, translations made and reviewed by professional translators. As it did with the introduction of adaptability, the development of this new technique was inspired by the human brain. Just as humans evaluate multiple sources of information to identify the most reliable and trustworthy information, ModernMT V7 also identifies the most valuable training data and organizes its learning accordingly.
“Modern MT’s ability to prioritize high-quality data to improve translation models is the most significant leap forward in machine translation since dynamic adaptability was introduced five years ago,” said Marco Trombetti, CEO of Translate. “This innovation opens up new opportunities for companies to use machine translation to take their customer experience to the next level around the world. At the same time, it will help translators increase their productivity and earnings.”
Trust attention
The new version introduces Trust Attention, a new technology inspired by the human brain's ability to prioritize information from trusted sources, which improves translation quality by up to 42% (see graph).
Marking a departure from traditional MT systems that are limited by the inability to distinguish between reliable data and low-quality content during the training process.
ModernMT now uses a first-of-its-kind weighting system to prioritize learning from high- portugal rcs data quality, qualified data, i.e., unverified content sourced from the web, translations made and reviewed by professional translators. As it did with the introduction of adaptability, the development of this new technique was inspired by the human brain. Just as humans evaluate multiple sources of information to identify the most reliable and trustworthy information, ModernMT V7 also identifies the most valuable training data and organizes its learning accordingly.
“Modern MT’s ability to prioritize high-quality data to improve translation models is the most significant leap forward in machine translation since dynamic adaptability was introduced five years ago,” said Marco Trombetti, CEO of Translate. “This innovation opens up new opportunities for companies to use machine translation to take their customer experience to the next level around the world. At the same time, it will help translators increase their productivity and earnings.”