I profiled the app in 'Instruments', found that it was using the CPU to inference, so it is much slower than the original model.
Topic:
Machine Learning & AI
SubTopic:
Core ML