Reimplementation of Vrepair using OpenNMT-py library API

Hello,

I would like to do a reimplementation of the VRepair model using OpenNMT library API and wonder if this can be possible in terms of archiving the same performance as reported in the paper. I know that you are using the openmt models through the library configuration file, so I would be much appreciate if you could share some thoughts on this experiment of mine ?

Start by reading the VRepair paper and the repository's existing OpenNMT model configuration, then review the OpenNMT-py library API. Define the reimplementation scope and an evaluation against the paper's reported performance; done means the experiment has documented, reproducible results.

Written by the indexing model from the issue text.