Converter plugins / PyTorch Converters
Machine Learning Inference Advisor (MLIA)
Connect PyTorch to analysis.
Follow supported export and conversion routes from PyTorch modules, .pt2 models, and ExecuTorch .pte artifacts.
View on GitHub (opens in a new tab)Converter plugins
Work with PyTorch Converters.
01
Pair with a target plugin
Install a target plugin for the hardware you want to analyze, then add these converters for PyTorch inputs.
Read the guide (opens in a new tab)02Choose a conversion route
Explore module export, TOSA, PTE, and delegate extraction.
Read the guide 03Understand the outputs
Inspect artifacts and diagnose conversion failures.
Read the guidePart of MLIA
Start with the right setup.
MLIA selects the conversion route for the input and downstream backend. Requirements differ between routes.
Documentation
Find your next step.
Under the hood
Explore the implementation.
Find integration details and contributor guidance.