bhigy / symbolic-bias

Code for the paper: Symbolic inductive bias for visually grounded learning of spoken language

symbolic-bias

Code for the paper: Symbolic inductive bias for visually grounded learning of spoken language. https://arxiv.org/abs/1812.09244, published at ACL 2019.

Install

Clone repo and set up and activate a virtual environment with python3

cd symbolic-bias
virtualenv -p python3 .

Install Python code (in development mode if you will be modifying something).

python setup.py develop

Download trained models and unpack them:

wget http://grzegorz.chrupala.me/data/symbolic-bias/experiments.tgz
tar zxvf experiments.tgz

Download data and unpack them:

wget http://grzegorz.chrupala.me/data/symbolic-bias/data.tgz
tar zxvf data.tgz

Usage

Execute function main in file analysis/analyze.py.

cd analysis
python -c 'import analyze; analyze.main()'

The output should be similar to analysis/results.tex.

Inspect the definition of this function to see how to compute the results from each table in the paper.