Some Basic Queries

Hello Everyone,

I am working on machine translation of English-Urdu. I have some queries for better understanding of OpenNMT translation engine.

  • How does the morph folder and all files inside it in openNMT-py/data/morph/ effect the quality of tranlsation ? How to get/generate these files for your corpus ?

  • For training parameters, encoder-encoder has different types. Is there any default parameter settings for rnn, brnn, cnn & mean as suggested for transformer in

  • Are files test_model.src & test_model.tgt auto-generated ? What is purpose of these files?


How much improvement did you achieved for english to urdu translation?

Can you also let me know the place to download english to urdu corpus?

1 Like

Opus is a good source for translation data including Urdu.

1 Like