Skip to content

Conversation

stevhliu
Copy link
Member

@stevhliu stevhliu commented Feb 7, 2024

This (WIP) PR creates a more detailed and centralized guide for all things IP-Adapter related:

➖ removes non-loading related IP-Adapter content from the loading adapters guide
➕ adds IP-Adapter guide to the Techniques section
➕ adds set_ip_adapter_scale to the API docs
➕ IP-Adapter guide is largely divided based on tasks (text-to-image, image-to-image, image-to-video, etc.) and specific use cases (face models, instant generation, controllable generation, etc.)

@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.

Copy link
Collaborator

@yiyixuxu yiyixuxu left a comment

Choose a reason for hiding this comment

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

great! thank you
I'm not sure where to add this but I think we need to add this information here #6162 (comment)
basically for certain ip-adapter-plus checkpoints (e.g. ip-adapter-plus_sdxl_vit-h), you will have to load the image_encoder from a different folder

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.

This is the best guide you have written @stevhliu!

@sayakpaul
Copy link
Member

Feel free to merge once @yiyixuxu approves. Looks really good!

# IP-Adapter

[IP-Adapter](https://hf.co/papers/2308.06721) is a lightweight adapter that enables prompting a diffusion model with an image. This method decouples the cross-attention layers of the image and text features. The image features are generated from an image encoder. Files generated from IP-Adapter are only ~100MBs.
[IP-Adapter](https://hf.co/papers/2308.06721) is a lightweight adapter that enables prompting a diffusion model with an image. This method decouples the cross-attention layers of the image and text features. The image features are generated from an image encoder.
Copy link
Member

Choose a reason for hiding this comment

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

Does it make sense to indicate the file size to give an upfront idea about the tinyness of the adapter?

Copy link
Member Author

Choose a reason for hiding this comment

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

I think it's probably ok to not mention file size which seems like an extra detail for a reference page. It's already mentioned in the loading guide (where it is more relevant) which is linked on this page.

Copy link

@khipp khipp left a comment

Choose a reason for hiding this comment

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

I read through the guide and added some suggestions if you don't mind. Great work!

Copy link
Collaborator

@yiyixuxu yiyixuxu left a comment

Choose a reason for hiding this comment

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

thanks! feel free to merge anytime!

@stevhliu stevhliu merged commit 9efe1e5 into huggingface:main Feb 14, 2024
@stevhliu stevhliu deleted the ip-adapter branch February 14, 2024 21:23
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.

5 participants