`None` parameters not sanitized during pruning

[karthikprasad]

🐛 Bug

ModelPruning callback fails when a module parameter is None. This can happen, for instance, in a Linear() when bias=False.

Please reproduce using the BoringModel

https://colab.research.google.com/drive/1UApprg-5htIQbosiSyyLLXm1B8wE8EbN?usp=sharing

Expected behavior

Unavailable parameters are already taken care of in sanitize_parameters_to_prune method in ModelPruning callback. A trivial check to ensure the attribute, when present, is not None should solve this. Happy to send a PR.

Environment

• CUDA:
  • GPU:
    • Tesla K80
  • available: True
  • version: 10.1
• Packages:
  • numpy: 1.19.5
  • pyTorch_debug: False
  • pyTorch_version: 1.8.1+cu101
  • pytorch-lightning: 1.2.6
  • tqdm: 4.41.1
• System:
  • OS: Linux
  • architecture:
    • 64bit
  • processor: x86_64
  • python: 3.7.10
  • version: Proposal for help #1 SMP Thu Jul 23 08:00:38 PDT 2020