Training on a subset of countries

[abh1nay]

Thanks for this terrific project! I am trying to slim down the model due to resource constraints, is it possible to get the steps to train the dataset only on english addresses ( i am hoping this will help) ?

I have tried the steps listed in the README but could not figure out a way to filter the addresses and train the model

[MohMagdi]

Have you figured it out? It would help me out too.

[walkman-kuan]

+1. This will be very helpful since all not users will need support for all the countries. The trained model is 1.8GB, which is too much.

[walkman-kuan]

See Splitting data files by country and language for how to train on a subset of countries