Working neural network model templates in pytorch.
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
models/GAN/vanilla
.gitignore
LICENSE
Pipfile
Pipfile.lock
README.md

README.md

Working neural network model templates in pytorch.

This code was developed to help me turn neural network literature into practical knowledge and techniques for using neural networks on actual problems. This repository was created to share the networks and notes I made while training the networks with others.

This repository contains:

(1) Simple, self-contained, and working neural networks

Models are self contained, clearly coded and commented, and come with unit tests in an effort to provide a codebase that can be used to quickly start using the models on your specific problem.

(2) Notes about what I learned while training the networks.

Literature is often missing tricks, caveats, and why hyperparameters were chosen. Notes on challenges and intricacies learned while training the networks are included for each model.

Due to limitations on embedding equations into markdown on GitHub, notes are published at www.sciencesundires.com/sundries

Networks Implemented

vanilla GANs

Linear and convolutional generative adversarial networks (GAN).

Goodfellow, Ian, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley,
    Sherjil Ozair, Aaron Courville, and Yoshua Bengio. 2014. “Generative
    Adversarial Nets.” In Advances in Neural Information Processing Systems,
    2672–2680.http://papers.nips.cc/paper/5423-generative-adversarial-nets.

Radford, A., Metz, L., and Chintala, S. (2015). Unsupervised representation
    learning with deep convolutional generative adversarial networks. ArXiv
    Prepr. ArXiv151106434.

To Install

git clone git@github.com:cottersci/ModelTests.git
pipenv install

Training depends on my pytorch_utils package, which is installed by pipenv.

Other useful pytorch links