Can I turn off Validation step when overfit_batches=X?

[IgorHoholko]

I call trainer.fit(model=model_lit, datamodule=datamodule)

I have set parameters:

limit_val_batches: 0
limit_test_batches: 0

datamodule.val_dataloader() returns None

But it is still trying to perform validation...
Is it a problem of lightning, or am I doing something wrong? :)

[IgorHoholko]

Well. The reason why validation was called is overfit_batches=1 parameter.

[IgorHoholko]

Why ?

[rohitgr7]

it's now turned off by default on master:
https://github.com/PyTorchLightning/pytorch-lightning/blob/2faaf35b91a00aff397609a875a66c87f8ed6390/pytorch_lightning/trainer/trainer.py#L636-L640

[IgorHoholko]

great, thanks :)
Good time to update packages