SDL Machine Translation Edge Trainer
SDL Machine Translation Edge Trainer is a functionality added to SDL's enterprise on-premises MT solution which allows users to "customize" out-of-the-box baseline NMT (Neural Machine Translation) Language Pairs. These language pairs are also called "Adaptable Language Pairs" and they are used to produce new, high-quality, domain-specific translations which use previously-translated, specialized, parallel data (source and target languages).
Enterprise NMT must be tailored to your organization's specific needs and content. To perform such adaptation, organizations currently need to upload their data in the cloud or share their data with vendors to get their NMT models trained and customized. Data is the most important asset of an organization and sharing it or uploading it to the cloud might pose a significant risk exposure or may even be forbidden in regulated industries.
The new Adaptable Language Pairs and the on-premises training capability included in SDL Machine Translation Edge Trainer (available as of SDL Machine Translation Edge 8.4), enables organizations to adapt NMT on their own, without the need to share their data. Customers can train their model on premises (or in their private cloud), at their own pace, while keeping their data secured at all times. They will be able to train and re-train their models to adjust them continuously, as their data and usage grow.