https://github.com/ikostrikov/pytorch-flows

PyTorch implementations of algorithms for density estimation
https://github.com/ikostrikov/pytorch-flows

deep-learning density-estimation neural-networks probabilities pytorch

Last synced: 9 days ago
JSON representation

PyTorch implementations of algorithms for density estimation

• Host: GitHub
• URL: https://github.com/ikostrikov/pytorch-flows
• Owner: ikostrikov
• License: mit
• Created: 2018-09-01T19:58:17.000Z (about 6 years ago)
• Default Branch: master
• Last Pushed: 2021-05-13T19:59:29.000Z (over 3 years ago)
• Last Synced: 2024-07-31T18:15:54.507Z (3 months ago)
• Topics: deep-learning, density-estimation, neural-networks, probabilities, pytorch
• Language: Python
• Homepage:
• Size: 52.7 KB
• Stars: 571
• Watchers: 18
• Forks: 77
• Open Issues: 5
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-energy-based-model - `pytorch-flows`
• awesome-normalizing-flows - pytorch-flows

README

        
# pytorch-flows

A PyTorch implementations of [Masked Autoregressive Flow](https://arxiv.org/abs/1705.07057) and 

some other invertible transformations from [Glow: Generative Flow with Invertible 1x1 Convolutions](https://arxiv.org/pdf/1807.03039.pdf) and [Density estimation using Real NVP](https://arxiv.org/abs/1605.08803).

For MAF, I'm getting results similar to ones reported in the paper. GLOW requires some work.

## Run

```bash

python main.py --dataset POWER

```

Available datasets are POWER, GAS, HEPMASS, MINIBONE and BSDS300. For the moment, I removed MNIST and CIFAR10 because I have plans to add pixel-based models later.

## Datasets

The datasets are taken from the [original MAF repository](https://github.com/gpapamak/maf#how-to-get-the-datasets). Follow the [instructions](https://github.com/gpapamak/maf#how-to-get-the-datasets) to get them.

## Tests

Tests check invertibility, you can run them as

```bash

pytest flow_test.py

```