`.export_model()` breaks with PyTorch 1.11 and later

[helena-intel]

The .export_model() method for PyTorch models calls torch.onnx.export() with the enable_onnx_checker parameter set to False:
https://github.com/openvinotoolkit/nncf/blob/develop/nncf/torch/exporter.py#L163
This parameter exists in PyTorch 1.9.1 and is set to True by default. In 1.10 the parameter is deprecated, and in 1.11 it doesn't exist anymore. This means that using NNCF with PyTorch 1.11 or later results in an error:

Could this be changed in NNCF so that it doesn't break with PyTorch 1.11 and later? I had no issues with NNCF with recent PyTorch versions if I manually remove this line.

[ljaljushkin]

Thank you for noticing, Helena!

We're bumping torch version to 1.12/1.13, so that issue will be covered here: #1323

[helena-intel]

That's great! I look forward to that and will close this issue then.