This repository contains an example script logistic_regression.py that reproduces the main results for the logistic regression experiment described in Section 7.2 and Figure 4 of Reparameterized Variational Rejection Sampling (Martin Jankowiak & Du Phan, AISTATS 2024). We include a single dataset (bank.csv). The main RVRS functionality, which is implemented as a NumPyro AutoGuide (i.e. variational distribution), can be found in rvrs.py.
To run the script we suggest the following workflow:
- create a new conda environment:
conda create -n rvrs python=3.9 - activate the new environment:
conda activate rvrs - install requirements:
pip install -r requirements.txt - examine script arguments:
python logistic_regression.py --help - run script with default arguments:
python logistic_regression.py