Mixing Core ML w/ Metal Performance Shaders Graph?

Is it possible to do any of the following:

  1. Export a model created using MetalPerformanceShadersGraph to a CoreML file;
  2. Failing 1., save a trained MetalPerformanceShadersGraph model in any other way for deployment;
  3. Import a CoreML model and use it as a part of a MetalPerformanceShadersGraph model.

Thanks!

Up vote post of Alan_Z Down vote post of Alan_Z
367 views
Posted by
Alan_Z
Reply
Add a Comment

Replies

I would like to understand what you are trying to accomplish when you say "Save a trained MetalPerformanceShadersGraph model in any other way for deployment;" so I can assist you better. Are you looking to decouple code from model weights? And, deploy weights out-of-band?

Posted by
Frameworks Engineer
Up vote reply of Frameworks Engineer Down vote reply of Frameworks Engineer

  • Thanks for the reply. Yes, I’d like to decouple code from weights & deploy weights out-of-band.

    Say I want to develop an app with a trained ML model bundled inside, but I want to build & train this model using only MPSG, and not TF/PyTorch b/c I enjoy writing Swift code and hate model conversions. Now, after I trained my model in, e.g., a playground, how can I export the model into a format that I can include in my app project?

    —
    Alan_Z
Add a Comment