PyTorch implementation of the paper,

Ask Me Anything: Dynamic Memory Networks for Natural Language Processing
Kumar et al., 2016 arxiv

• Install cuda-8.0
• Install cudnn-v5.1
• Install Pytorch 0.4.0
• Python version >= 3.5 is required

• bAQbI Tasks v1.2 data downloaded from here
• Place files in en-valid-10k under (home)/datasets/babi/en directory.
• Place a pretrained GloVe under (home)/datasets/glove directory.

# Preprocessing dataset. This will create ./data/babi(tmp).pkl
$ python dataset.py

# Train and test with default settings
$ python main.py

# Train with different number of hidden units, epochs, and QA sets
$ python main.py --s_rnn_hdim 200 --epoch 20 --set_num 5