Results for m1 ultra (48 core gpu)

[machineko]

I run your script on my machine for extra results 😄

Im not sure if MPS on pytorch give sane results all my pretrained models are just full of trash outputs right now (not even once i got result compariable with coreml/pytorch cpu so if you have time you should also make some check for that [compare diff in cpu vs mps version for example] )

Here are the results for inference on m1 ultra

[lucadiliello]

Thank you! How much RAM has your system? Version of pytorch?

[machineko]

torch->'1.13.0.dev20220604
neofetch