zephyr/samples/modules/tflite-micro/hello_world/train

Hello World Training

This example shows how to train a 2.5 kB model to generate a sine wave.

Table of contents

• Overview
• Training
• Trained Models
• Model Architecture

Overview

1. Dataset: Data is generated locally in the Jupyter Notebook.
2. Dataset Type: Structured Data
3. Deep Learning Framework: TensorFlow 2
4. Language: Python 3.7
5. Model Size: 2.5 kB
6. Model Category: Regression

Training

Train the model in the cloud using Google Colaboratory or locally using a Jupyter Notebook.

Google Colaboratory

Jupyter Notebook

Estimated Training Time: 10 minutes.

Trained Models

Download Link	hello_world.zip

The models directory in the above zip file can be generated by following the instructions in the Training section above. It includes the following 3 model files:

Name	Format	Target Framework	Target Device
model.pb	Keras SavedModel	TensorFlow	Large-Scale/Cloud/Servers
model.tflite (2.5 kB)	Integer Only Quantized TFLite Model	TensorFlow Lite	Mobile Devices
model.cc	C Source File	TensorFlow Lite for Microcontrollers	Microcontrollers

Model Architecture

The final model used to simulate a sine wave is displayed below. It is a simple feed forward deep neural network with 2 fully connected layers with ReLu activations and a final fully connected output layer with as shown below.

This image was derived from visualizing the 'model.tflite' file in Netron