When are Iterative GPs Numerically Accurate?
This is a code repository for the paper "When are Iterative GPs Numerically Accurate?" by Wesley Maddox, Sanyam Kapoor, and Andrew Gordon Wilson.
Citation
@article{maddox2021iterative,
title={When are Iterative Gaussian Processes Reliably Accurate?},
author={Wesley J. Maddox and Sanyam Kapoor and Andrew Gordon Wilson},
year={2021},
publication={ICML OPTML Workshop},
url={https://arxiv.org/abs/2112.15246},
}
Models
Our models, both iterative and cholesky-based, are in the models/gpytorch/models.py.
The scripts that can be used to reproduce our results are:
models/gpytorch/runner.py
notebooks/iterative_gps_reliability.ipynb
(explainer)src/train_keops.py
(for optimization trajectories)
Repository References
The benchmarking library that we cloned is Hugh Salimbeni's bayesian_benchmarks library, available at https://github.com/hughsalimbeni/bayesian_benchmarks.
For full comparison to other libraries that use the same library, we use standardization over both the train/test splits.
For LBFGS, we used (and link to) Michael Shi's LBFGS library, available at https://github.com/hjmshi/PyTorch-LBFGS/.