Hi there, I would like to ask a question. I am interested in embeddings as a means to assess the accuracy of machine translation. By this I mean using an embeddings model which is language agnostic to loop through the input and the output of the MT sentence by sentence and compare each embedding - the greater the variance, the more likely of a semantic error in the translation. TENSOR has something like this - Universal Sentence Encoder | TensorFlow Hub. It seems that your model is only in English at present but I wanted to ask in case it could be used multilingually or you had some other thoughts on this. This is in support of a possible Masters dissertation looking at ways to optimise use of MT. Grateful any thoughts, Rob
Did you end up getting your question answered?