Deep Learning Notebooks
Here are my jupyter notebooks on deep learning. All of them are written in pytorch and most of them use the fastai library.
Tutorials
- Using the new training API in fastai.
- Building a simple neural net in pytorch.
- Explore the fastai callback system.
Implementation of articles
- Neural cache pointer introduce by Grave et al.
- Superconvergence on cifar10 using the 1cycle policy introduce by Leslie Smith
- Deep painterly harmonization from this article
- Adam and weight decay for the correction preoposed in this article
The rest is just a bun of random stuff shared with fellow fastai students.