Encountered Error While Directly Importing Model in Runtime with Post-training Optimization Tool (POT)

1. Created the IR model in runtime using serialize in runtime
2. Ran into the following error when directly importing the IR model with POT in runtime

   assert len(node.in_nodes()) == 2, 'The Squeeze node {} must have 2 inputs'.format(node_name)
   AssertionError: The Squeeze node Squeeze_1034 must have 2 inputs

Direct import in runtime does not apply offline transformations to the model while POT expects the IR to be processed by offline transformations.

Apply offline transformations:

from openvino.offline_transformations import apply_moc_transformations
apply_moc_transformations(net_onnx, False)