Spaces:

hugging-apps
/

inverse-and-edit

Sleeping

inverse-and-edit / diffusion_core /diffusion_models.py

Upload folder using huggingface_hub

dc7f251 verified 15 days ago

1.19 kB

import torch

from diffusers.pipelines import StableDiffusionPipeline
from .utils import ClassRegistry
from .load_consistency import load_consistency



diffusion_models_registry = ClassRegistry()


@diffusion_models_registry.add_to_registry("stable-diffusion-v1-4")
def read_v14(scheduler):
    model_id = "CompVis/stable-diffusion-v1-4"
    model = StableDiffusionPipeline.from_pretrained(
        model_id, torch_dtype=torch.float32, scheduler=scheduler
    )
    return model


@diffusion_models_registry.add_to_registry("stable-diffusion-v1-5")
def read_v15(scheduler):
    model_id = "runwayml/stable-diffusion-v1-5"
    model = StableDiffusionPipeline.from_pretrained(
        model_id, torch_dtype=torch.float32, scheduler=scheduler
    )
    return model

@diffusion_models_registry.add_to_registry("consistency-v1-5")
def read_v15(device):
    solver = load_consistency(device)
    return solver


@diffusion_models_registry.add_to_registry("stable-diffusion-v2-1")
def read_v21(scheduler):
    model_id = "stabilityai/stable-diffusion-2-1"
    model = StableDiffusionPipeline.from_pretrained(
        model_id, torch_dtype=torch.float32, scheduler=scheduler
    )
    return model

	import torch

	from diffusers.pipelines import StableDiffusionPipeline
	from .utils import ClassRegistry
	from .load_consistency import load_consistency



	diffusion_models_registry = ClassRegistry()


	@diffusion_models_registry.add_to_registry("stable-diffusion-v1-4")
	def read_v14(scheduler):
	model_id = "CompVis/stable-diffusion-v1-4"
	model = StableDiffusionPipeline.from_pretrained(
	model_id, torch_dtype=torch.float32, scheduler=scheduler
	)
	return model


	@diffusion_models_registry.add_to_registry("stable-diffusion-v1-5")
	def read_v15(scheduler):
	model_id = "runwayml/stable-diffusion-v1-5"
	model = StableDiffusionPipeline.from_pretrained(
	model_id, torch_dtype=torch.float32, scheduler=scheduler
	)
	return model

	@diffusion_models_registry.add_to_registry("consistency-v1-5")
	def read_v15(device):
	solver = load_consistency(device)
	return solver


	@diffusion_models_registry.add_to_registry("stable-diffusion-v2-1")
	def read_v21(scheduler):
	model_id = "stabilityai/stable-diffusion-2-1"
	model = StableDiffusionPipeline.from_pretrained(
	model_id, torch_dtype=torch.float32, scheduler=scheduler
	)
	return model