Feature/log computational graph

[SkafteNicki]

What does this PR do?

Fixes #2915
Automatically logs the computational graph for certain loggers (tensorboard and test tube, could not find any other loggers that is able to do this). Only requirement is that user has defined the model.example_input_array attribute which we already use for model summaries.

Before submitting

• Was this discussed/approved via a Github issue? (no need for typos and docs improvements)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure your PR does only one thing, instead of bundling different changes together? Otherwise, we ask you to create a separate PR for every change.
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests?
• Did you verify new and existing tests pass locally with your changes?
• If you made a notable change (that affects users), did you update the CHANGELOG?

PR review

Anyone in the community is free to review the PR once the tests have passed.
If we didn't discuss your PR in Github issues there's a high chance it will not be merged.

Did you have fun?

Make sure you had fun coding 🙃

[codecov]

Codecov Report

Merging #3003 into master will decrease coverage by 0%.
The diff coverage is 100%.

@@          Coverage Diff           @@
##           master   #3003   +/-   ##
======================================
- Coverage      90%     90%   -0%     
======================================
  Files          81      81           
  Lines        7626    7694   +68     
======================================
+ Hits         6860    6908   +48     
- Misses        766     786   +20     

[rohitgr7]

@SkafteNicki should we disable log_graph when logger=True, or set log_graph=False by default?
https://github.com/PyTorchLightning/pytorch-lightning/blob/bcdb750976ca2a62925fdf5201720d6b383db37e/pytorch_lightning/trainer/logging.py#L44-L50
I think it should be totally optional because it gives the model.example_input_array attribute is not set or input_array warning and not every time one wants to log the graph.

[SkafteNicki]

@rohitgr7 I am fine by not logging the graph as default. I just though that automatically adding the graph would be a nice default behavior (if exanple_input_array is defined) for people just using the default logger. But if the warning is annoying we can definitely change this.

[rohitgr7]

yeah, the warning is a bit annoying. will change to False by default :)