README.md

Dynamic Memory Networks in Tensorflow

Implementation of Dynamic Memory Networks for Visual and Textual Question Answering on the bAbI question answering tasks using Tensorflow.

Prerequisites

• Python 3.x
• Tensorflow 0.8+
• Numpy
• tqdm - Progress bar module

Usage

First, You need to install dependencies.

sudo pip install tqdm
git clone https://github.com/therne/dmn-tensorflow & cd dmn-tensorflow

Then download the dataset:

mkdir data
curl -O http://www.thespermwhale.com/jaseweston/babi/tasks_1-20_v1-2.tar.gz
tar -xzf tasks_1-20_v1-2.tar.gz -C data/

If you want to run original DMN (models/old/dmn.py), you also need to download GloVe word embedding data.

curl -O http://nlp.stanford.edu/data/glove.6B.zip
unzip glove.6B.zip -d data/glove/

Training the model

./main.py --task [bAbi Task Number]

Testing the model

./main.py --test --task [Task Number]

Results

Single run of DMN+ model trained with paper settings (Batch 128, 3 episodes, 80 hidden, dropout rate 0.9, L2) + batch normalization. The skipped tasks achieved 0 error.

Task	Error Rate

2. Two supporting facts | 27.2%
3. Three supporting facts | -
4. Two arguments relations | 23.4%
5. Three arguments relations | 1.1%
6. List/Sets | 0.4%
7. Compound coreference | 1.5%
8. Time reasoning | 0.8%
9. Basic induction | 65.9%
10. Positional reasoning | 19.2%
11. Size reasoning | 8.7%
12. Path finding | 69.9% Average | 10.9%

Overfitting occurs in some tasks and error rate is higher than the paper's result. I think we need some additional regularizations.

References

• Implementing Dynamic memory networks by YerevaNN - Great article that helped me a lot
• Dynamic-memory-networks-in-Theano

To-do

• More regularizations and hyperparameter tuning
• Visual question answering
• Attention visualization
• Interactive mode?