alibidaran/LORA_SDX1.5_Midjourney

---

LORA_SDX1.5_Midjourne with LORA weight is the Fine-tuned SDX1.5 model trained on images with 512 shape in Midjourney dataset.

This model is fined-tuned version of StableDiffusion model via Dreambooth. The code below is for the direct usage and controlling the model for making a more cohesive image based on the given prompt.

!git clone https://huggingface.co/alibidaran/LORA_SDX1.5_Midjourney
from diffusers import DiffusionPipeline
pre="runwayml/stable-diffusion-v1-5"
pipe = DiffusionPipeline.from_pretrained(
    pretrained_model_name_or_path,
    torch_dtype=torch.float16
).to(device)
pipe.load_lora_weights("LORA_SDX1.5_Midjourney/lora/pytorch_lora_weights.safetensors")


def generate(pipeline, prompt, seed):
    generator = torch.Generator(device=device).manual_seed(seed)
    return pipeline(prompt, num_inference_steps=num_inference_steps, generator=generator).images[0]

prompt_training='A black and white photograph of Megan Fox., side profile, pensive and enigmatic.'

generate(pipe, prompt_training, seed)