GitHub - harvardnlp/MemN2N: Torch implementation of End-to-End Memory Networks (https://arxiv.org/abs/1503.08895) (original) (raw)

End-To-End Memory Network

Torch implementation of MemN2N (Sukhbaatar, 2015). Supports Adjacent Weight Tying, Position Encoding, Temporal Encoding and Linear Start. Code uses v1.0 of bAbI dataset with 1k questions per task.

Prerequisites:

• Python 2.7
• Torch (with nngraph)

Preprocessing

First, preprocess included data into hdf5 format:

This will create a hdf5 file for each task (total 20 tasks).

To train:

This will train on task 16 with Linear Start over 100 epochs by default. See train.lua (or the paper) for hyperparameters and more training options. To train on gpu, use -cuda.