Building a Seq2Seq mode, that takes as an input a german sentence or word (encoder) outputs its english translation (decoder)
The dataset is a English-German words/sentence pairs from : http://www.manythings.org/anki/
After ~ 25 epochs our validation loss stop decreasing.
The model obtained translate correctly the german sentences, however it misses out between past and present tense. For example, the model translate "ate" to "caught" or "sang" to "likes'.
- Python 3.7
- Numpy,Pandas
- Keras