Lyalyushkin Nikolay
Lyalyushkin Nikolay
This PR can't be merged without passing all pre-commit checks. There's issue with code style, please refer to [pylint check log](https://ci-adas-icv.iotg.sclab.intel.com/job/NNCF/job/precommit/job/pylint/120/). Also, one e2e test for efficient (nas.test_sanity_sample.test_e2e_supernet_training[efficient_net_b1_cifar10_nas.json_classification_nas_mock_32x32]) is failing...
Jenkins please retry a build
Jenkins please stop all builds
I wonder how ignored_scope and target_scope will work in case of such model. Am I right, that everything will be in storage-mode now: scopes in original and compressed graphs, scopes...
> No, in fact, the storage scope comparisons are limited to determining the train-only modules and referencing their scopes against the storage scope that is being tracked during the one-time...
I wonder how the algorithm will work when chunk axis doesn't match channel axis, according to which pruning occurs. What will happen with pruning mask and propagation algorithm? I think...
Have started sota eval (torch 56 and tf 87) jobs from nightly validation to make sure it doesn't break anything.
> @ljaljushkin, can we merge it? SOTA eval for TF and PT is OK, except some known issues that are not related to the PR. Unfortunately, pre-commit for Pytorch is...
Pre-commit for PyTorch is not working due to infra issue. We've reported it to CI team. Have to wait until it's fixed and green
run pytorch pre-commit tests