Attribution and modification notice
Original material
DeepDeWedge Tutorial Data
Creator: Simon Wiedemann
DOI: https://doi.org/10.6084/m9.figshare.25043435.v1
Figshare file id: 45582309
Archive member: tutorial_data/fitted_model.ckpt
License: Creative Commons Attribution 4.0 International
The method is described by Simon Wiedemann and Reinhard Heckel, A deep learning method for simultaneous denoising and missing wedge reconstruction in cryogenic electron tomography, Nature Communications 15, 8255 (2024), https://doi.org/10.1038/s41467-024-51438-y.
Pinned upstream code: https://github.com/MLI-lab/DeepDeWedge/tree/072075692a44a8f17394214369e6e762abe52bc3 (BSD-2-Clause).
Changes in this package
On 2026-09-04 Scitomo freshly converted only the authoritative checkpoint
official/fitted_model.ckpt, after byte-size and SHA-256 verification, through
the exact pinned upstream source and current generic FORMAT 2 exporter. The
54 U-Net state tensors were explicitly mapped into canonical Network state.
The two fitted affine quantities were preserved as external DeepDeWedge
inference-profile state; they are not Network state. No old Hugging Face
Safetensors or format-1 package artifact was conversion input.
No endorsement by the cited authors, the Machine Learning and Information Processing Laboratory, Figshare, or the rights holders is implied.
On 2026-09-23 Scitomo prepared revision 4 as a metadata-only update: the zero-degree missing-wedge support center became explicit in the inference contract. The verified revision-3 Safetensors bytes and original source attribution remain unchanged.