This is project for the neural machine translation system from Quechua to Spanish. The code and corpora is part of a paper authored by John E. Ortega (NYU), Richard Castro Mamani (Cuzco) , and Kyunghun Cho (NYU)

The annotations folder are the human subject evaluations and the corpora are the new corpora commissioned by us @NYU, specifically the "magazine" folder under corpora has 100 translations done by a professional Quechua->Spanish translator.

The Opus folder contains files gotten using jw300 on http://opus.nlpl.eu/

There is a train,validate,test split for training OpenNmt (or any MT) models...