In this section, we will go through the details of TensorFlow Lite: the overall architecture, the key components, and their functionality.

The following diagram provides a high-level overview of the key components and how they interact to bring machine learning to mobile devices:

The following are the key steps to be followed when implementing ML on devices:

1. Use the TensorFlow, or any other machine learning framework, to create the trained TensorFlow/ML models on the desktop. The trained model can also be created using any Cloud ML engine.
2. Use the TensorFlow Lite converter to convert the trained ML model to the TensorFlow Lite model file.
3. Write a mobile application using these files and convert it into a package for deployment and execution in mobile devices. These lite files could be interpreted and executed directly in the kernels or in the hardware accelerators, if available in the device.

The following are the key components of TensorFlow Lite:

• Model-file format
• Interpreter
• Ops/kernel
• Interface to hardware acceleration