Factoid Question Answering
This repo attempts to produce an implementation of "alternating stochastic gradient" descent algorithm discussed in . Preprocessing is inspired from Simple-Question-Answering-With-Memory-Networks(https://github.com/Jerryzhao-z/simple-question-answering-with-memory-networks)
One has to specify location of all datasets and other local configuration information in SETTINGS.JSON file.
The vocabulary of individual words is produced with the
g(q) is done with the
f(y) is done with the
After preprocessing the dataset, training of facoid question answering is done using following command. $python3 train.py Please refer to the paper  for detailed understanding of how the train script trains our question-answering system.
A python script is provided for testing the trained system. Use
test.py to test the system.
Transfer learning on another dataset using the trained model.
 Large-scale Simple Question Answering with Memory Networks (https://arxiv.org/pdf/1506.02075.pdf)