This is a Neural Machine Translation application using PyTorch, EasyNMT, FastAPI, MongoDB and Docker.
The implementation focuses on the deployment using FastAPI for the RESTful services, MongoDB (Mongo Atlas) acting as the database and the Docker framework for the containerization of the application.
For the translation services, the EasyNMT package was used and a solution based on RNN neural networks from PyTorch tutorials on seq2seq networks.
The data used to train the seq2seq were downloaded from ManyThings.org.
First of all, create a MongoDB account and edit the core/config.py with your Mongo cloud credentials. Then, build the container using the intstructions.
docker build -t app .
docker run -d --name app -p 8000:8000 {IMAGEID}The FastAPI Swagger UI is located in localhost:8000/docs. Each request is protected through OAuth2 (so you need to create a user in order to interact with the APIs).