An extensible PyTorch framework to experiment with neural-networks-based deep learning algorithms on multiple data modalities for binary classification.

An open source framework for seq2seq models in PyTorch.

Noise2Noise is an AI denoiser trained with noisy images only. We implemented a ligther version which trains faster on smaller pictures without losing performance and an even simpler one where every low-level component was implemented from scratch, including a reimplementation of autograd.

A simple yet effective PyTorch-based framework for beginner to familiarize deep learning training pipeline. Plus, this is a good foundation to desigin your own training pipeline from scratch with guided.

Playing with PyTorch