The convert_onnx_to_tensorflow function in optimizer.py is currently a stub that returns None.
This missing adapter breaks the "Hybrid Optimization Pipeline" (PyTorch → ONNX → TF → TFLite). Users attempting to optimize PyTorch models for TFLite via this pipeline will fail silently or get unoptimized results.
Integrate library to strictly implement this conversion step. The function should take the ONNX model path and return a valid TensorFlow SavedModel or GraphDef that can then be passed to the TFLite converter.
The
convert_onnx_to_tensorflowfunction inoptimizer.pyis currently a stub that returns None.This missing adapter breaks the "Hybrid Optimization Pipeline" (PyTorch → ONNX → TF → TFLite). Users attempting to optimize PyTorch models for TFLite via this pipeline will fail silently or get unoptimized results.
Integrate library to strictly implement this conversion step. The function should take the ONNX model path and return a valid TensorFlow SavedModel or GraphDef that can then be passed to the TFLite converter.