Geoopt is a research-oriented modular open-source package for Riemannian Optimization in PyTorch. The core of Geoopt is a standard Manifold interface that allows for the generic implementation of optimization algorithms. Geoopt supports basic Riemannian SGD as well as adaptive optimization algorithms. Geoopt also provides several algorithms and arithmetic methods for supported manifolds, which allow composing geometry-aware neural network layers that can be integrated with existing models.
Cite
@article{arxiv.2005.02819,
title = {Geoopt: Riemannian Optimization in PyTorch},
author = {Max Kochurov and Rasul Karimov and Serge Kozlukov},
journal= {arXiv preprint arXiv:2005.02819},
year = {2020}
}
Comments
Proceedings of the 37th International Conference on Machine Learning, Vienna, Austria, PMLR 108, 2020, GRLB Workshop