First, clone this repo. Next, you will also need to pull the valuernn
submodule by running:
git submodule update --init
Next, set up the required python packages by creating a virtual environment:
conda create --name valuernn python=3.9 pytorch matplotlib numpy scipy scikit-learn
conda activate valuernn
Option 1. To use the models from the paper, unzip data/models.zip
to the folder data/models
.
Option 2. To fit your own models (approximate run time: 48 hours), run:
chmod +x bin/fit.sh
./bin/fit.sh
Option 1. To use the analyses from the paper, unzip data/sessions.zip
to the folder data/sessions
.
Option 2. To analyze your own models, run:
chmod +x bin/analyze.sh
./bin/analyze.sh
To make the figures, run:
chmod +x bin/plot.sh
./bin/plot.sh
The resulting figures will be in data/figures/
.
(Note: If you encounter an error on Mac regarding libomp.dylib
, try conda install nomkl
.)