yuurei002/smart-medical-lora-sd3

SMART Medical LoRA (SD3)

LoRA fine-tuned từ Stable Diffusion 3 Medium trên 397 ảnh SMART Medical Illustration (19 chủ đề gộp lại).

Cách dùng

from diffusers import StableDiffusion3Pipeline
import torch

pipe = StableDiffusion3Pipeline.from_pretrained(
    'stabilityai/stable-diffusion-3-medium-diffusers',
    torch_dtype=torch.float16,
    text_encoder_3=None, tokenizer_3=None,
).to('cuda')
pipe.load_lora_weights('yuurei002/smart-medical-lora-sd3')

image = pipe(
    'smsart, medical illustration of human heart anatomy, cross-section',
    negative_prompt='blurry, low quality, watermark, photo, realistic',
    num_inference_steps=28, guidance_scale=7.0,
).images[0]

Training

• Base: SD3 Medium | LoRA rank=8 | Epochs=25
• Dataset: 397 ảnh · 19 chủ đề (gộp lại để học style)
• Best loss: 0.1451