MCUXpresso SDK Documentation

Version: 26.09.00 26.06.00-lts 26.03.00 25.12.00 latest

Open on Repo Open an Issue

Source (.md/.rst) Page PDF (.pdf)

eIQ TensorFlow Lite for Micro Library User Guide

Table of Contents

• Overview
• TensorFlow Lite for Microcontrollers
• Build Status
  • Official Builds
  • Community Supported TFLM Examples
  • Community Supported Kernels and Unit Tests
• Contributing
• Getting Help
• Additional Documentation
• RFCs
• Deployment
• Example applications
• Model Conversion to TensorFlow Lite Format
• Model Conversion for NXP eIQ Neutron NPU
• Example: Converting a Quantized TensorFlow Lite Model for Neutron NPU
• Example: Converting a Quantized TensorFlow Lite Model for Neutron NPU with Neutron Sequencer mode.
• Example: Converting a Quantized TensorFlow Lite Model for Neutron NPU with weight prefetching.
• Run and debug eIQ HiFi4 and HiFi1 DSP examples using Xplorer IDE
  • Prepare CM33 Core for the examples
  • Prepare DSP core for the examples
• Running an inference
  • Converting a model to a C language header file
  • NPU inference
  • Adjusting the tensor arena size
• Code size optimization
  • Register only used operators in TensorFlow Lite Micro {#SECTION_SS1_DJQ_QPB .section}
• Register only used operators
  • Register only used operators in TensorFlow Lite Micro
• Supported Operator List
  • Supported Operator List by Tensorflow Lite for Microcontrollers
  • Supported Operator List by eIQ Neutron NPU Library and HiFi 4 NN Library
• Note about the source code in the document