Introduction

In the first lesson, you developed a platform for creating isolated Python environments to facilitate working with ML Models.

In this lesson, you’ll use this environment to explore some PyTorch models and convert them into CoreML using a couple of different methods. You’ll find model conversion can be a complex task and often requires knowledge of the model or simply experience with CoreML Tools. You should only look at conversion when needed, as many models already provide CoreML versions.

After working through model conversion, you’ll next explore ways to reduce the size of the large models to better fit into the limitations of mobile devices. You’ll use CoreML Tools to compress the models you converted in the first part of the lesson.

You’ll find the often complex interdependencies between models still make this a challenge, so don’t be surprised if you see some warnings while working through this lesson.

By the end of the lesson, you will have learned to:

• Describe the Core ML tools used to convert a third-party model for Core ML compatibility
• Detail the challenges of deploying third-party models on Apple devices
• Explain the conversion success and identify any potential issues