[community] Load Models from Sources like Civitai into Existing Pipelines

[suzukimain]

What does this PR do?

This PR adds an example that allows you to search for models from Civitai and Hugging Face by simply entering a keyword and loading models that meet the criteria into existing pipelines.

Example

!wget https://raw.githubusercontent.com/suzukimain/diffusers/refs/heads/ModelSearch/examples/model_search/search_for_civitai_and_HF.py

# Search for Civitai

from search_for_civitai_and_HF import CivitaiSearchPipeline
from diffusers import StableDiffusionPipeline


model_path = CivitaiSearchPipeline.for_civitai(
    "any",
    base_model="SD 1.5",
    download=True
)
pipe = StableDiffusionPipeline.from_single_file(model_path).to("cuda")

# Search for Hugging Face

from search_for_civitai_and_HF import HFSearchPipeline
from diffusers import StableDiffusionPipeline

model_path = HFSearchPipeline.for_HF(
           "stable",
           checkpoint_format="diffusers",
           download = False
           )

pipe = StableDiffusionPipeline.from_pretrained(model_path).to("cuda")

# or

model_path = HFSearchPipeline.for_HF(
           "stable",
           checkpoint_format="single_file",
           download = False
           )

pipe = StableDIffusionPipeline.from_single_file(model_path).to("cuda")

Arguments of HFSearchPipeline.for_HF

Name	Type	Default	Description
search_word	string	ー	The search query string.
revision	string	None	The specific version of the model to download.
checkpoint_format	string	"single_file"	The format of the model checkpoint.
download	bool	False	Whether to download the model.
force_download	bool	False	Whether to force the download if the model already exists.
include_params	bool	False	Whether to include parameters in the returned data.
pipeline_tag	string	None	Tag to filter models by pipeline.
hf_token	string	None	API token for Hugging Face authentication.
skip_error	bool	False	Whether to skip errors and return None.

Arguments of CivitaiSearchPipeline.for_civitai

Name	Type	Default	Description
search_word	string	ー	The search query string.
model_type	string	"Checkpoint"	The type of model to search for.
base_model	string	None	The base model to filter by.
download	bool	False	Whether to download the model.
force_download	bool	False	Whether to force the download if the model already exists.
civitai_token	string	None	API token for Civitai authentication.
include_params	bool	False	Whether to include parameters in the returned data.
skip_error	bool	False	Whether to skip errors and return None.

search_word

Type	Description
keyword	Keywords to search model
url	URL of either huggingface or Civitai
Local directory or file path	Locally stored model paths
huggingface path	The following format: < creator > / < repo >

model_type

Input Available
Checkpoint
TextualInversion
Hypernetwork
AestheticGradient
LORA
Controlnet
Poses

checkpoint_format

Argument	Description
all	The multifolder diffusers format checkpoint takes precedence.
single_file	Only single file checkpoint are searched.
diffusers	Search only for multifolder diffusers format checkpoint

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline?
• Did you read our philosophy doc (important for complex PRs)?
• Was this discussed/approved via a GitHub issue or the forum? Please add a link to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.
• Did you write any new necessary tests?

Who can review?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

[bghira]

can it use AIR IDs to import them? thanks for this example!

edit: the fact that model API responses has the model type in it we could even do AutoPipeline.from_civitai(air=model_air) kind of thing.