This repo contains the tensorflow implementation of the paper "Question Answering on Knowledge Bases and Text using Universal Schema and Memory Networks".
(naetherm) Made small modifications so it's possible to use the newest TensorFlow version (currently version 1.2.0) when training.
- TensorFlow 1.2.0
I have set up training with most default params on a very small dataset so that it is easier to get started. Just running the script should work.
/bin/bash run.sh ./config.sh
The processed data (train/dev/test split) is stored in data_formatted/ directory. To download the KB files used for the project run,
sh get_data.sh
After downloading the data, you will have to change the appropriate entries in the config.sh file (kb_file and text_kb_file).
We are also releasing the output predictions of our model for comparison. Find them in the model_outputs directory.
We are also sharing our pretrained model. Get it here