AN 1011: TinyML Applications in Altera FPGAs Using LiteRT for Microcontrollers

ID 848984
Date 9/29/2025
Public
Document Table of Contents

2.4.3. Analyzing the LiteRT Model

Identify the type of op resolver needed to run the model using LiteRT for Microcontrollers libraries. Refer to Appendix for more information about op resolver registration in the main() function. For this LeNet-5 example, the op resolvers are:

  • CONV_2D
  • AVERAGE_POOL_2D
  • FULLY_CONNECTED
  • RESHAPE
  • SOFTMAX
# Analyse the LiteRT Model
tf.lite.experimental.Analyzer.analyze(model_path="lenet.tflite")
Figure 6. Results from Analyzer
Note: The LiteRT for Micro libraries do not support all types of op resolver from LiteRT. Ensure that you can find the op resolvers listed by the Analyzer in the following path: tensorflow/lite/micro/micro_mutable_op_resolver.h.

For more information on the LiteRT for Micro libraries, please refer to Building LiteRT for Microcontroller Static Library chapter.