divergence-frontier-bounds

This repository contains the code to reproduce the experiments in this paper at NeurIPS 2021. The paper studies the sample complexity of the evaluation framework—divergence frontiers—for generative models. It also introduces frontier integrals as summary statistics of divergence frontiers.

Standalone package: For a self-contained package to compute divergence frontiers and frontier integrals, installable via pip install mauve-text, please see this repository.

Dependencies

The code is written in Python and the dependencies are:

• python >= 3.6
• ipykernel >= 6.4.1
• matplotlib >= 3.5.0
• numpy >= 1.19.1
• pandas >= 1.3.4
• pathos >= 0.2.8
• scikit-learn >= 0.22.1
• scipy >= 1.7.3
• faiss-cpu >= 1.7.1

Conda Environment: We recommend using a conda environment. To setup the environment, run

conda env create --file environment.yml
# activate the environment
conda activate divergence-frontier-bounds
python -m ipykernel install --user --name divergence-frontier-bounds

Files

• metrics.py: code to compute divergence frontiers and frontier integrals.
• utils.py: utility functions.
• experiments.ipynb: Jupyter notebook to run experiments and produce plots.
• results/: experimental results.

Citation

If you find this repository useful, or you use it in your research, please cite:

@inproceedings{liu-etal:divergence:neurips2021,
  title={{Divergence Frontiers for Generative Models: Sample Complexity, Quantization Effects, and Frontier Integrals}},
  author={Liu, Lang and Pillutla, Krishna and Welleck, Sean and Oh, Sewoong and Choi, Yejin and Harchaoui, Zaid},
  booktitle={NeurIPS},
  year={2021}
}

Acknowledgements

This work was supported by NSF DMS-2134012, NSF CCF-2019844, the CIFAR program “Learning in Machines and Brains”, and faculty research awards.