Only decode on a test set about drqa HOT 3 OPEN

It should be very easy to modify "interact.py" to change the CLI predicting to batch mode. Load you json and replace the evidence and questions to the ones you load from file. Change

BatchGen([model_in], batch_size=1, gpu=args.cuda, evaluation=True)

to

BatchGen([model_in_1, model_in_2, ...], batch_size=batch_size, gpu=args.cuda, evaluation=True)

to perform batch prediction.
I didn't include this in the project because it's quite straightforward to do this with "interact.py" provided. Adding this script will add complexity to this project (since this project is for simplicity and ease to read) and will not be directly useful in most cases because test data formats vary from json, csv, database, web queries, ... "interact.py" is provided as a template and it should be easy to adapt this script to any input format.
If you find it hard to do this please post your questions here and I'll be listening :)

from drqa.

Ok..I'll try your suggestion tomorrow and post it here. Seems pretty straightforward..thanks

from drqa.

@LearningPytorch please post here the code modifications you made. It will be really helpful for others. Thank you

from drqa.