lr_tuner fails if called after tuning batch_size

[Palzer]

https://github.com/PyTorchLightning/pytorch-lightning/blob/514cb22bd719e6ca056cacce730c8de875c9dbf6/pytorch_lightning/tuner/lr_finder.py#L411

The line above sets current_step to 2 if the batch_size tuner has already run.

https://github.com/PyTorchLightning/pytorch-lightning/blob/514cb22bd719e6ca056cacce730c8de875c9dbf6/pytorch_lightning/tuner/lr_finder.py#L419-L420

This causes best_loss to be zero and compare against the smoothed loss. This will cause the tuner to terminate and fail.

Also, just looks like this was already supposed to be changed.

[github-actions]

Hi! thanks for your contribution!, great first issue!

[edenlightning]

@SkafteNicki mind taking a look?

[SkafteNicki]

It seems the solution is to remove the +1 from current_step (as the todo note also states)
@Palzer would you be up for sending a PR with the fix?

[Palzer]

Done. First pr on github so be gentle :P

edit: checking out failures on pr

[SkafteNicki]

@Palzer I can probably find some time tomorrow to look over your work. Thanks for wanting to contribute :]

[stale]

This issue has been automatically marked as stale because it hasn't had any recent activity. This issue will be closed in 7 days if no further activity occurs. Thank you for your contributions, Pytorch Lightning Team!