The first neural machine translation system for the Qarachay Malqar language

Authors: Bogdan Tewunalany, Ali Berberov

We collect 288532 parallel sentences between russian and Qarachay-Malqar languages. There were two ways to make the translator:

• First, We made transformwer on Tensorflow/Keras on R. But fine-tuning of existing model is better.
• Second, We used pre-trained model nllb-200 and fine-tuned it (4.Jupyter_Colab).

Model

Where to use: HF and site

Parallel corpora: As is and Dialects free

Authors telegram: Bogdan and Ali

License

Shield:

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.