Syntactically Informed Text Compression with Recurrent Neural Networks

Installation

Using our AMI:

Start a g2.2xlarge instance using our AMI ami-2c3a7a4c.

git clone https://github.com/davidcox143/rnn-text-compress.git
cd rnn-text-compress
pip install -r requirements.txt
make

Usage

Train a model

python rnn-text-compress.py -t < path to text file >

Evaluate accuracy of a single model

python rnn-text-compress.py -e < path to text file > < path to weights file >

Evaluate the accuracy of a model over time and plot results

python rnn-text-compress.py -p < path to text file > < path to weights directory >