Daniil Lyakhov

@ljaljushkin, can we merge it?

run pytorch pre-commit tests

@akladiev, looks like NNCF ONNX precommit (Ubuntu 20.04) is NNCF Torch + common precommit

Have the same issue, looks like `librosa.sequence.viterbi` returns `np.uint16` values which is unsupported by `torch.tensor`

> @daniil-lyakhov Could you resolve the conflict? The rebased release branch seems to have something to fix. Done

Greetings, @moonlightian, there is no logic in pruning algo which could possibly do such changes, could you please share a model for reproduction?

Thank you for your contribution, @jpablomch! Please rebase your PR on the latest develop

Jenkins please retest this

Guys, it's a bit tricky, but you can download checkpoint by [this](https://github.com/kingyiusuen/image-to-latex/blob/main/scripts/download_checkpoint.py)

TODO - [x] Add test for matmul with several outputs case