Jialuo21/Science-T2I-Flux-SFT-OFT

Resources

• Website
• arXiv: Paper
• GitHub: Code
• Huggingface: SciScore
• Huggingface: Science-T2I-S&C Benchmark
• Huggingface: Science-T2I training set

Quick Start

You can use FluxPipeline to run the model

import torch
from diffusers import FluxPipeline
pipe = FluxPipeline.from_pretrained("black-forest-labs/FLUX.1-dev", torch_dtype=torch.bfloat16).to('cuda')
pipe.load_lora_weights("Jialuo21/Science-T2I-Flux-SFT-OFT")

prompt = "An unripe grape in the garden"
image = pipe(
    prompt,
    height=1024,
    width=1024,
    guidance_scale=0.0,
    num_inference_steps=50,
    max_sequence_length=512,
    generator=torch.Generator("cpu").manual_seed(0)
).images[0]
image.save("example.png")

Citation

@misc{li2025sciencet2iaddressingscientificillusions,
  title={Science-T2I: Addressing Scientific Illusions in Image Synthesis}, 
  author={Jialuo Li and Wenhao Chai and Xingyu Fu and Haiyang Xu and Saining Xie},
  year={2025},
  eprint={2504.13129},
  archivePrefix={arXiv},
  primaryClass={cs.CV},
  url={https://arxiv.org/abs/2504.13129}, 
}