Latent Consistency Model (LCM) LoRA: SSD-1B

Latent Consistency Model (LCM) LoRA was proposed in LCM-LoRA: A universal Stable-Diffusion Acceleration Module by Simian Luo, Yiqin Tan, Suraj Patil, Daniel Gu et al.

It is a distilled consistency adapter for segmind/SSD-1B that allows to reduce the number of inference steps to only between 2 - 8 steps.

Model	Params / M
lcm-lora-sdv1-5	67.5
lcm-lora-ssd-1b	105
lcm-lora-sdxl	197M

Usage

LCM-LoRA is supported in 🤗 Hugging Face Diffusers library from version v0.23.0 onwards. To run the model, first install the latest version of the Diffusers library as well as peft, accelerate and transformers. audio dataset from the Hugging Face Hub:

pip install --upgrade pip
pip install --upgrade diffusers transformers accelerate peft

Text-to-Image

Let's load the base model segmind/SSD-1B first. Next, the scheduler needs to be changed to LCMScheduler and we can reduce the number of inference steps to just 2 to 8 steps. Please make sure to either disable guidance_scale or use values between 1.0 and 2.0.

import torch
from diffusers import LCMScheduler, AutoPipelineForText2Image

model_id = "segmind/SSD-1B"
adapter_id = "latent-consistency/lcm-lora-ssd-1b"

pipe = AutoPipelineForText2Image.from_pretrained(model_id, torch_dtype=torch.float16, variant="fp16")
pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
pipe.to("cuda")

# load and fuse lcm lora
pipe.load_lora_weights(adapter_id)
pipe.fuse_lora()


prompt = "Self-portrait oil painting, a beautiful cyborg with golden hair, 8k"

# disable guidance_scale by passing 0
image = pipe(prompt=prompt, num_inference_steps=4, guidance_scale=0).images[0]