about gpu usage in CoreML
#1
by
altunenes
- opened
Hello, thank you for this great work! It works really well in macOS M3 CPU (3 seconds). I tried to run it locally via CoreML, but it's not using GPU. Is this related to ONNX? Have you tested it with CoreML?
My concern is that fastVLM's ONNX operations might not be supported by CoreML's neural engine