Incorrect Parameter Type on UNet2DModel

[fpgaminer]

Describe the bug

The parameter type for UNet2DModel::__init__ seems to be incorrect, at least on these three parameters:

diffusers/src/diffusers/models/unet_2d.py

Lines 97 to 99 in c4d2823

	down_block_types: Tuple[str] = ("DownBlock2D", "AttnDownBlock2D", "AttnDownBlock2D", "AttnDownBlock2D"),
	up_block_types: Tuple[str] = ("AttnUpBlock2D", "AttnUpBlock2D", "AttnUpBlock2D", "UpBlock2D"),
	block_out_channels: Tuple[int] = (224, 448, 672, 896),

I believe the correct type is Tuple[str, ...] and Tuple[int, ...] which means "A tuple of any size, where all elements are of type str (or int)". The current type, Tuple[str] means "A tuple that must be of size 1 and that one element is of type str", which causes typing errors when trying to use UNet2DModel.

Reproduction

A typical instantiation of UNet2DModel like this:

self.model = UNet2DModel(
			sample_size=self.config.sample_size,
			in_channels=3,
			out_channels=3,
			layers_per_block=2,
			block_out_channels=(128, 128, 256, 256, 512, 512),
			down_block_types=(
				"DownBlock2D",
				"DownBlock2D",
				"DownBlock2D",
				"DownBlock2D",
				"AttnDownBlock2D",
				"DownBlock2D",
			),
			up_block_types=(
				"UpBlock2D",
				"AttnUpBlock2D",
				"UpBlock2D",
				"UpBlock2D",
				"UpBlock2D",
				"UpBlock2D",
			),
		).to(self.device)

Logs

Argument of type "tuple[Literal['DownBlock2D'], Literal['DownBlock2D'], Literal['DownBlock2D'], Literal['DownBlock2D'], Literal['AttnDownBlock2D'], Literal['DownBlock2D']]" cannot be assigned to parameter "down_block_types" of type "Tuple[str]" in function "__init__"
 "tuple[Literal['DownBlock2D'], Literal['DownBlock2D'], Literal['DownBlock2D'], Literal['DownBlock2D'], Literal['AttnDownBlock2D'], Literal['DownBlock2D']]" is incompatible with "Tuple[str]"
 Element size mismatch; expected 1 but received 6PylancereportGeneralTypeIssues

System Info

- `diffusers` version: 0.20.0
- Platform: Linux-5.15.0-78-generic-x86_64-with-glibc2.35
- Python version: 3.10.11
- PyTorch version (GPU?): 2.0.1 (True)
- Huggingface_hub version: 0.15.1
- Transformers version: 4.29.2
- Accelerate version: 0.19.0
- xFormers version: not installed

Who can help?

@patrickvonplaten @sayakpaul

[patrickvonplaten]

Nice catch! Would you like to open a PR to correct it?

[gkatte]

May I ?

[fpgaminer]

My plate is full

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.

Please note that issues that do not follow the contributing guidelines are likely to be ignored.