SDE-DiffFWI_datasets

This dataset contains the pretrained Dense Diffusion Transformer (DiT) and the two baselines velocity models used in PDE Sensitivity of Diffusion Priors in Full Waveform Inversion. The diffusion prior was trained on 12,000 geological velocity models from GeoFWI, covering layered, faulted and salt structures. These resources support the reproducibility of SDE-DiffFWI experiments and further studies of diffusion priors in full-waveform inversion. The original GeoFWI and benchmark datasets should be cited separately.