{
  "model_type": "yeti",
  "architecture": "Reconstruction AutoEncoder (RAE) + one-step Conditional Diffusion Transformer (C-DiT)",
  "task": "real-world sRGB noise generation",
  "framework": "pytorch-lightning",
  "checkpoints": {
    "rae.ckpt": "Reconstruction AutoEncoder",
    "c_dit.ckpt": "Conditional Diffusion Transformer (main noise generator)",
    "apbsn.ckpt": "AP-BSN self-supervised denoiser, trained on YeTI-generated noisy data only",
    "apbsn_mix.ckpt": "AP-BSN denoiser trained on a 50:50 mix of YeTI-generated and real noisy data",
    "mmbsn.ckpt": "MM-BSN self-supervised denoiser, trained on YeTI-generated noisy data only",
    "mmbsn_mix.ckpt": "MM-BSN denoiser trained on a 50:50 mix of YeTI-generated and real noisy data"
  },
  "paper": "https://arxiv.org/abs/2607.09193",
  "code": "https://github.com/ByungWanLim/YeTI",
  "dataset": "https://huggingface.co/datasets/BWLim/YeTI"
}