Reserved: Murattil Core ML mirror
Originally intended to mirror …-onnx as Apple Core ML .mlpackage
files for iOS pure-Core-ML inference.
Why it's empty: coremltools 9.0 removed the direct
ct.converters.onnx API (deprecated for years, finally dropped).
The remaining Core ML conversion path requires going through
PyTorch as an intermediate, which is fragile for FastConformer's
hybrid CTC+RNNT graph.
The working alternative — and what Murattil currently uses on
iOS — is ONNX Runtime with the Core ML Execution Provider.
ORT loads the same .onnx file from …-onnx and routes ANE-
friendly ops to Apple's Neural Engine internally. No separate
.mlpackage needed.
If a future toolchain re-enables clean ONNX → Core ML conversion,
artifacts will land here.
