auto-diffusers 1.9.0

Customized diffusers with model search and other functions.

auto_diffusers

CONTENTS

• About The Project
• How to use
• Example
• Description
• License
• Acknowledgement

Announcements📢

Due to changes in the hf_api specification, the function was stopped for a while.
I apologize for any inconvenience this may have caused.

[Updated on October 29, 2024]
Fixed an error that security status could not be acquired due to hf_api specification changes

About The Project

Enhance the functionality of diffusers.

• Search models from huggingface and Civitai.

How to use

pip install diffusers
pip install auto_diffusers

from diffusers import StableDiffusionPipeline
from auto_diffusers import model_search


path = model_search(
           <keyword>,
           auto = True,
           model_format="diffusers",
           download = False
           )
pipe = StableDiffusionPipeline.from_pretrained(path)

# or

path = model_search(
           <keyword>,
           auto = True,
           model_format="single_file",
           download = False
           )
pipe = StableDIffusionPipeline.from_single_file(path)

Example

pip install --quiet diffusers
pip install --quiet auto_diffusers

from diffusers import StableDiffusionPipeline
from IPython.display import display
from auto_diffusers import model_search

model_path = model_search(
                 "Any",
                 auto=True,
                 model_format="diffusers",
                 download=False
                 )
pipe = StableDiffusionPipeline.from_pretrained(model_path).to("cuda")

image = pipe("Mt. Fuji").images[0]

print(f"model_path: {model_path}")
display(image)

Description

Arguments of model_search

Name	Type	Default	Input Available	Description
search_word	string	ー	Details	Keywords to search models
auto	bool	True	ー	Minimize user input by selecting the highest-rated models.
download	bool	False	ー	Returns the path where the file was downloaded.
model_format	string	"single_file"	all, diffusers, single_file	Specifies the format of the model. Details
model_type	string	"Checkpoint"	Checkpoint, TextualInversion, Hypernetwork, AestheticGradient, LORA, Controlnet, Poses	Valid only in Civitai.
include_params	bool	False	ー	Returns the model path or a dictionary with parameters.
branch	string	"main"	ー	Specify the branches of huggingface and civitai.
local_file_only	bool	False	ー	Search local folders only. In the case of auto, files with names similar to search_word will be given priority.
hf_token	string	None	ー	Token used for authentication with Hugging Face.
civitai_token	string	None	ー	Token used for authentication with Civitai.

search_word

Type	Description
keyword	Keywords to search model
url	Can be any URL other than huggingface or Civitai.
Local directory or file path	Search for files with the extensions: .safetensors, .ckpt, .bin
huggingface path	The following format: < creator > / < repo >

model_format

Argument	Description
all	In auto, multifolder diffusers format checkpoint takes precedence
single_file	Only single file checkpoint are searched.
diffusers	Search only for multifolder diffusers format checkpoint Note that only the huggingface is searched for, since it is not in civitai.

License

In accordance with BSD-3-Clause license

Acknowledgement

I have used open source resources and free tools in the creation of this project.

I would like to take this opportunity to thank the open source community and those who provided free tools.