Welcome to Serrano Guides


This guide site introduces the basic concepts and usage of Serrano. Serrano was developed as a tool help researchers and developers quickly implementing deep learning models on iOS and macOS.

Design Principles

  • Like the deep learning library you used before. Serrano aims to offer similar programing experience in Swift as popular deep learning frameworks.
  • Easy to customize. Adding a new Operator/Optimizer/Loss Function is simple. Create a Swift class and conform the protocol, that's it.
  • Flat architecture. In Serrano, we try best to reduce complex of architecture by less abstraction. You could be clear with the whole architecture by grasping few core concepts. Also in this way, we hope everyone could easily contribute to this project.


Bug reports

You can open bug-related and only bug-related issues in Serrano issues.

Questions and General Discussion

For any other issues, questions and discussion, please go to SerranoExplore. You are free to open any issue there.

Also, there's Gitter chatting room where you can ask questions.

Help Improve Guides

Since Serrano is under alpha-development, guides listed in this site may be outdated. If you find any error or misleading information, you are welcome to make a PR. For any guides improvement suggestions or better --- adding more guides --- please do not hesitate to make a PR.