…ce again isn’t very useful. But what if (and here’s the big idea!) we could train the second RNN to decode the sentence into Spanish instead of English? We could use our parallel corpora training data to train it to do that:Machine Learning is Fun Part 5: Language Translation with Deep Learning and the Magic of Sequences10K58Adam GeitgeyGopa KumarFollowAug 29, 2017 · 1 min readCan you explain the decoder? How are the weights trained for it to “know” the words in Spanish.