Skip to content

Conversation

stevhliu
Copy link
Member

Refreshes the Load pipelines doc with more focus on the DiffusionPipeline class and loading it.

Main changes:

➕ new device placement section (focusing review on this section would be great!)
## Checkpoint variants section will be discussed in the Model files and layout doc
## DiffusionPipeline explained section will be moved to the Resources section

@@ -36,6 +36,15 @@
"configuration_utils": ["ConfigMixin"],
"guiders": [],
"hooks": [],
"image_processor": [
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is it ok to add image_processor and video_processor to __init__.py? Otherwise, my local preview of the docs won't build. No big deal either way though :)

@HuggingFaceDocBuilderDev

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

@stevhliu stevhliu requested a review from sayakpaul August 15, 2025 00:18
from diffusers import DiffusionPipeline

pipeline = DiffusionPipeline.from_pretrained(
"black-forest-labs/FLUX.1-dev",
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do we wanna mix Qwen and Flux?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Good point, I'll switch to Qwen and reduce the overall number of different models we use as examples to avoid confusion!


```py
from diffusers import AnimateDiffPipeline, MotionAdapter, DDIMScheduler
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Are we deleting these?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah I don't think we need to also show animating the image when we already have an example of using from_pipe for PAG. It seems redundant and adds bloat

Comment on lines 163 to 171
_import_structure["image_processor"] = [
"IPAdapterMaskProcessor",
"PixArtImageProcessor",
"VaeImageProcessor",
"VaeImageProcessorLDM3D",
]
_import_structure["video_processor"] = [
"VideoProcessor",
]
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

These changes seem unrelated?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It's unrelated, but if it doesn't hurt then it might be more convenient to include it so we can preview the docs locally. Let me know what you think! :)

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Let's separate them out in this PR and open another one?

Copy link
Member

@sayakpaul sayakpaul left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Changes are a go barring the ones in dummy_pt_objects.py (let's tackle it in a separate PR). Thank you!

@stevhliu stevhliu force-pushed the diffusion-pipeline branch from ec85c73 to 422d85f Compare August 25, 2025 17:53
@stevhliu stevhliu merged commit 2c4ee10 into huggingface:main Aug 25, 2025
1 check passed
@stevhliu stevhliu deleted the diffusion-pipeline branch August 25, 2025 18:06
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants