ModelCheckpoint.filepath cannot be None

[elkotito]

🐛 Bug

The ModelCheckpoint callback's argument filepath cannot be None even though the docs says:

Can also be set to None, then it will be set to default location during trainer construction.

The exception is raised:

Traceback (most recent call last):
  File "/home/pi3ni0/Workspace/random/siamese_cnn_glove.py", line 209, in <module>
    checkpoint_callback = ModelCheckpoint(
  File "/home/pi3ni0/.venv/dev/lib/python3.6/site-packages/pytorch_lightning/callbacks/model_checkpoint.py", line 89, in __init__
    if save_top_k > 0 and os.path.isdir(filepath) and len(os.listdir(filepath)) > 0:
  File "/home/pi3ni0/.venv/dev/lib/python3.6/genericpath.py", line 42, in isdir
    st = os.stat(s)
TypeError: stat: path should be string, bytes, os.PathLike or integer, not NoneType

It is because os.path.isdir cannot be applied to None argument.

Code sample

   checkpoint_callback = ModelCheckpoint(
        filepath=None,
    )

   trainer = pl.Trainer(
        checkpoint_callback=checkpoint_callback,
    )

   ...

Environment

* CUDA:
	- GPU:
		- GeForce GTX 1650 with Max-Q Design
	- available:         True
	- version:           10.2
* Packages:
	- numpy:             1.18.2
	- pyTorch_debug:     False
	- pyTorch_version:   1.5.0
	- pytorch-lightning: 0.7.5
	- tensorboard:       2.2.1
	- tqdm:              4.45.0
* System:
	- OS:                Linux
	- architecture:
		- 64bit
		- ELF
	- processor:         x86_64

[olineumann]

Duplicate issue #1660 #1535. Already fixed in PR #1654