drewnoff / swift4tf Goto Github PK

Swift for TensorFlow is a next generation platform for deep learning and differentiable programming.

In general Swift for TF tries to resolve the drawbacks of Python

• Performance
• Concurrency
• Deployment
• Custom Ops

Swift has many advantages

• First-class autodiff
• Next-generation APIs
• High-quality tooling
• Python Integrability
• Performance of graphs + flexibility of Eager execution

Name	Description	Notebook
Introduction	A basic introduction to Swift and its basic syntax
Visualization	How to use Matplotlib in Swift For TensorFlow.
Augmentation	How to use do basic augmentations like flipping and cropping
MNIST	Using Swift for TensorFlow for training MNIST.
CIFAR10	Using Swift for TensorFlow for training CIFAR10.
Transfer Learning	A basic transfer learning tutorial using VGG
Deep Dream	Implementation of deep dream using a VGG model.