Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/bayesiains/nflows
Normalizing flows in PyTorch
https://github.com/bayesiains/nflows
density-estimation generative-model normalizing-flows pytorch
Last synced: 3 months ago
JSON representation
Normalizing flows in PyTorch
- Host: GitHub
- URL: https://github.com/bayesiains/nflows
- Owner: bayesiains
- License: mit
- Created: 2020-02-17T12:36:32.000Z (over 4 years ago)
- Default Branch: master
- Last Pushed: 2023-11-19T16:03:40.000Z (7 months ago)
- Last Synced: 2024-03-17T14:22:38.850Z (3 months ago)
- Topics: density-estimation, generative-model, normalizing-flows, pytorch
- Language: Python
- Homepage:
- Size: 1.37 MB
- Stars: 789
- Watchers: 30
- Forks: 119
- Open Issues: 12
-
Metadata Files:
- Readme: README.md
- License: LICENSE.md
Lists
- awesome-normalizing-flows - nflows
- awesome-stars - bayesiains/nflows - Normalizing flows in PyTorch (Python)
- awesome-energy-based-model - `nflows`
README
# nflows
`nflows` is a comprehensive collection of [normalizing flows](https://arxiv.org/abs/1912.02762) using [PyTorch](https://pytorch.org).
## Installation
To install from PyPI:
```
pip install nflows
```
## Usage
To define a flow:
```python
from nflows import transforms, distributions, flows
# Define an invertible transformation.
transform = transforms.CompositeTransform([
transforms.MaskedAffineAutoregressiveTransform(features=2, hidden_features=4),
transforms.RandomPermutation(features=2)
])
# Define a base distribution.
base_distribution = distributions.StandardNormal(shape=[2])
# Combine into a flow.
flow = flows.Flow(transform=transform, distribution=base_distribution)
```
To evaluate log probabilities of inputs:
```python
log_prob = flow.log_prob(inputs)
```
To sample from the flow:
```python
samples = flow.sample(num_samples)
```
Additional examples of the workflow are provided in [examples folder](examples/).
## Development
To install all the dependencies for development:
```
pip install -r requirements.txt
```
## Citing nflows
To cite the package:
```bibtex
@software{nflows,
author = {Conor Durkan and
Artur Bekasov and
Iain Murray and
George Papamakarios},
title = {{nflows}: normalizing flows in {PyTorch}},
month = nov,
year = 2020,
publisher = {Zenodo},
version = {v0.14},
doi = {10.5281/zenodo.4296287},
url = {https://doi.org/10.5281/zenodo.4296287}
}
```
The version number is intended to be the one from `nflows/version.py`. The year/month correspond to the date of the release. BibTeX entries for other versions could be found on [Zenodo](https://doi.org/10.5281/zenodo.4296286).
If you're using spline-based flows in particular, consider citing the _Neural Spline Flows_ paper: [[bibtex]](https://papers.nips.cc/paper/2019/file/7ac71d433f282034e088473244df8c02-Bibtex.bib).
## References
`nflows` is derived from [bayesiains/nsf](https://github.com/bayesiains/nsf) originally published with
> C. Durkan, A. Bekasov, I. Murray, G. Papamakarios, _Neural Spline Flows_, NeurIPS 2019.
> [[arXiv]](https://arxiv.org/abs/1906.04032) [[bibtex]](https://papers.nips.cc/paper/2019/file/7ac71d433f282034e088473244df8c02-Bibtex.bib)
`nflows` has been used in
> Conor Durkan, Iain Murray, George Papamakarios, _On Contrastive Learning for Likelihood-free Inference_, ICML 2020.
> [[arXiv]](https://arxiv.org/abs/2002.03712).
> Artur Bekasov, Iain Murray, _Ordering Dimensions with Nested Dropout Normalizing Flows_.
> [[arXiv]](https://arxiv.org/abs/2006.08777).
> Tim Dockhorn, James A. Ritchie, Yaoliang Yu, Iain Murray, _Density Deconvolution with Normalizing Flows_.
> [[arXiv]](https://arxiv.org/abs/2006.09396).
`nflows` is used by the conditional density estimation package [pyknos](https://github.com/mackelab/pyknos), and in turn the likelihood-free inference framework [sbi](https://github.com/mackelab/sbi).