callCoreMLFromCppOrPython

An example of using CoreML from C++ / Python for Whisper Encoder

call CoreML from C++

1. (Done) generate encoder_{modelSize}_fp16.mlmodel by convert_whisper_encoder_to_coreml.py
2. (Done) move model to coreml/CoremlEncoder.mlmodel
3. (Done) generate mlmodelc/ and its objc srcs in coreml folder by

xcrun coremlc compile CoremlEncoder.mlmodel .
xcrun coremlc generate CoremlEncoder.mlmodel .

4. make
5. ./main (it should generate same result from pytorch output)

call CoreML from Python

after 1~4 steps, run python predict_with_coreml.py

note: currently coreml/ folder contains 1-3 result from Whisper Tiny model