Both sentence and paragraph-level models can be run. Both are Pointer-Generator networks (See et al. 2017) trained on the SQuAD dataset.
The models are trained with the excellent OpenNMT-py sequence-to-sequence library. Decoding uses the Diverse Beam Search algorithm (Vijayakumar et al. 2016). You can tune the diversity strength, δ (it must be between 0 and 1).
Sadly the models are not the brightest - go easy on them - linguistic understanding with neural networks has a long way to go!
Install the dependencies and run application.py to get started!