Tensorflow implementation of [Text Understanding with the Attention Sum Reader Network] https://arxiv.org/abs/1603.01547 with reference to

[Teaching Machines to Read and Comprehend] https://arxiv.org/abs/1506.03340

and [A Thorough Examination of the CNN/Daily Mail Reading Comprehension Task] https://arxiv.org/abs/1606.02858

The dataset is based on the paper [Towards AI-Complete Question Answering: A Set of Prerequisite Toy Tasks] https://arxiv.org/abs/1502.05698

• Python 2.7+
• Tensorflow 1.0

To train a model from scratch (ifcontinue==False indicates this is a new model, data_type is the type of the tasks you want to train on and is between 1-20)

$ python QAtrain_weights.py --data_path=./tasks_1-20_v1-2/en/ --model=small --ifcontinue=False --data_type=1

To continue to train a model(ifcontinue==True indicates continuing to train an existing model, data_type is the type of the tasks you want to train on and is between 1-20)

$ python QAtrain_weights.py --data_path=./tasks_1-20_v1-2/en/ --model=small --ifcontinue=True --data_type=1

To test an existing model (data_type is the type of the tasks and is between 1-20):

$ python QAanswer.py --data_path=./tasks_1-20_v1-2/en/ --model=small --data_type=1

Modified codes based on TensorFlow official tutorial on RNN https://www.tensorflow.org/tutorials/recurrent