I have exported a Pytorch model into a CoreML mlpackage file and imported the model file into my iOS project. The model is a Music Source Separation model - running prediction on audio-spectrogram blocks and returning separated audio source spectrograms. Model produces correct results vs. desktop+GPU+Python but the inference on iPhone 15 Pro Max is really, really slow. Using Xcode model Performance tool I can see that the inference isn't automatically managed between compute units - all of it runs on CPU. The Performance tool notation hints all that ops should be supported by both the GPU and Neural Engine. One thing to note, that when initializing the model with MLModelConfiguration option .cpuAndGPU or .cpuAndNeuralEngine there is an error in Xcode console: `Error(s) occurred compiling MIL to BNNS graph: [CreateBnnsGraphProgramFromMIL]: Failed to determine convolution kernel at location at /private/var/containers/Bundle/Application/2E3C4AFF-1FA4-4C95-AAE4-ECEBC0FB0BF9/mymss.app/mymss.mlmodelc/model.mil:2453:12 @ CreateBnnsGraphProgramFromMIL` Before going back hammering the model in Python, are there any tips/strategies I could try in CoreMLTools export phase or in configuring the model for prediction on iOS? My export toolchain is currently Linux with CoreMLTools v8.1, export target iOS16.

Objective and steps Use the device front true depth camera (iPhone 12 Pro Max) to capture image data, live photo data and metadata (e.g. depth data and portrait effects matte) using AVFoundation capture principles into an AVCapturePhoto object. Save this captured object with its metadata to PHPhotoLibrary using a PHAssetCreationRequest object API. Result Image data, live data, disparity depth data (640x480 px) and some metadata is stored with the image through the PHPhotoLibrary API but the high quality portrait effects matte is lost. Notes Upon receiving the AVCapturePhoto object from AVFoundation capture delegate API I can verify that AVCapturePhoto object contains a high quality portrait effects matte member object. Using object's fileDataRepresentation() to obtain Data blob, writing that to a test file URL and reading it back I can see that flattened data API writes and restores the portrait effects matte. However, it gets stripped from the data when writing through the PHPhotoLibrary asset creation request. When later picking the image e.g. with PHPickerViewController + PHPickerResult and peeking into the object's data with CGImageSourceCopyAuxiliaryDataInfoAtIndex() I can see that there is data dictionary only for key kCGImageAuxiliaryDataTypeDisparity, and kCGImageAuxiliaryDataTypeDepth and kCGImageAuxiliaryDataTypePortraitEffectsMatte are both missing. Please, anyone has more detailed information if this possible at all? Thanks!