I’m watching and learning from pytorch conference 2023, when I encountered these concepts. It seems they are all related to deploy models.
To the best of my understanding:
torch.export
is a functionality in pytorch, to export a computation graph to some pre-defined format. This corresponds to the functiontorch.export
.torchserve
is a framework to serve pytorch models at scale. It might use models fromtorch.export
. This corresponds to a repo GitHub - pytorch/serve: Serve, optimize and scale PyTorch models in production .executorch
is another framework to run pytorch models for mobile/edge devices. It might use models fromtorch.export
. This corresponds to a repo GitHub - pytorch/executorch: End-to-end solution for enabling on-device AI across mobile and edge devices for PyTorch models .- AOTInductor, from the talk https://www.youtube.com/watch?v=w7d4oWzwZ0c and slides , it seems to be an ahead-of-time version of inductor.
The most confusing thing to me is AOTInductor. It seems to be more of a concept. I don’t know if it is a new backend, or a new repo, or a new functionality.
Can anyone help me out? It would be great to provide some concrete code to show what is AOTInductor.
And any other help for understanding torch.export
/torchserve
/executorch
would also be great!