torchdecomp 1.2.0

A set of matrix and tensor decomposition models implemented as PyTorch classes

PyTorchDecomp

A set of matrix and tensor decomposition models implemented as PyTorch classes

Installation

Because PyTorchDecomp is a PyPI package, please install it by pip command as follows:

python -m venv env
pip install torchdecomp

For the other OS-specific or package-manager-specific installation, please check the README.md of PyTorch.

Usage

import torchdecomp as td
import torch

References

• Non-negative Matrix Factorization (NMF)
  • Kimura, K. A Study on Efficient Algorithms for Nonnegative Matrix/Tensor Factorization, Ph.D. Thesis, 2017
• Exponent term depending on Beta parameter
  • Nakano, M. et al., Convergence-guaranteed multiplicative algorithms for nonnegative matrix factorization with Beta-divergence. IEEE MLSP, 283-288, 2010
• Beta-divergence NMF and Backpropagation
  • https://yoyololicon.github.io/posts/2021/02/torchnmf-algorithm/

Contributing

If you have suggestions for how PyTorchDecomp could be improved, or want to report a bug, open an issue! We'd love all and any contributions.

For more, check out the Contributing Guide.

License

PyTorchDecomp has a MIT license, as found in the LICENSE file.

Authors

• Koki Tsuyuzaki