How to interpret output of TensortRT engine inspector?

I have applied post-training quantization to my pretty complex neural network architecture. Now I am using a TRT engine inspector to retrieve information about each individual layer to see if it was quantized or not, as not all layer types and operations are supported for conversion to INT8 by TensorRT.

After collecting layer information and discarding some columns which I do not need I am left with something that I cannot really parse or understand honestly.

I know for a fact the very first layers of my architecture are the very first 3 layers of a ResNet34, so a Convolution, ReLU and MaxPool, followed by another custom convolutional bloc. Why do the first 4 layers in my quantized model show up as having LayerType Constant, Reformat, Constant and Reformat respectively? With names like onnx::Conv_1063, Identity_76, onnx::Conv_1062 and Identity_77?

I have been so far unable to find a guide on how to interpret these LayerTypes unfortunately.

Start with the linked TensorRT EngineInspector API reference and compare its documented output with the four reported LayerType values: Constant, Reformat, Constant, and Reformat. Review how the inspector represents the first ResNet34 layers and define what a guide must explain so users can identify quantized layers; done means a clear interpretation guide with a reproducible example.