Offline language independent NMT system for Windows 10/11

Open Framework is an offline language independent NMT system for Windows 10/11
The tool is designed to be used exclusively with Open NMT’s CTranslate2 and SentencePiece models. It provides a way of performing neural machine translation of screen input and documents (*.docx, *.xlsx, *.pptx, *.pdf and *.txt files) offline, i.e. without any Internet connection.
The system requires a CTranslate2 neural machine translation model (“model.bin”) and a SentencePiece model to be selected and loaded. It has been successfully tested with the author’s models and various models from the Argos Opentech project, handling diverse language pairs.
Open Framework has been developed and tested on an Asus Zenbook with 8GB RAM. It is intended for use in a low power and low resource environment with an Intel Core i7-8565U or similar processor. It is ideal for testing models or performing short translations in an offline environment. The tool is not suitable for high-speed large volume production for which a server set-up is best suited.
Bespoke versions designed for GPU’s and using a sophisticated array of inference features are available on request.
For more information, please contact