How to perform inference and gathering accuracy metrics on executorch model

[mmingo848]

Hi, I am having trouble finding solid documentation that explains how to do the following with executorch (stable):

• Load in the exported .pte model
• Run inference with images
• Gather accuracy

I have applied quantization and other optimizations to the original model and exported it to .pte. I'd like to see the accuracy after these techniques were applied. I followed the following tutorial for exporting the model. If we can't do the above items on the directly exported .pte file, then is there a way we can based on the below steps for preparing the model for edge dialect?

https://pytorch.org/executorch/stable/tutorials/export-to-executorch-tutorial.html

[mikekgfb]

@mmingo848 We'll be publishing a solution to this - hopefully very soon!