SwiftForTensorFlow

Trying out the Swift implementation of the tensorflow framework.

Key Lessons learned:

• Setting up a notebook with Swift kernel on Google Colab
• Use Python in a Swift environment (PythonKit)
• Classification of Iris Flowers
  • Import data
  • Explore data
  • Create Model
  • Train Model
  • Evaluation and predicting

Personal Findings:

• no typealiases needed for imported packages due to syntax change (compared to python)
• google colab makes it harder to look deep into the tf framework, might want to install it locally to use it in Xcode, or maybe use Google Chrome?
• at least in the 'Classification of Iris Flowers' example there seems to be a lot more typing involed than with python
• code uses a lot of swift specific syntax I wasn't aware of yet
• had many problems with the kernel crashing on me
• save from time to time back to github, when your browser crashes progress is lost