Views
No views yet
1024 dimensions).AlexanderKroll/foldvision-encoderC, N, S, O, P)(B, 1024) embedding.pdb files to sparse point lists (numpy_3D_point_lists/*.npz).bounding_boxes.npy + dataloader to construct dense tensors at runtime.scripts/preprocess_pdb_dir.pyscripts/embed_proteins.pyscripts/train.pyscripts/train_PSI.pyscripts/evaluate.pyscripts/evaluate_PSI.py1from foldvision import FoldVisionEncoder
2
3model = FoldVisionEncoder.from_pretrained("AlexanderKroll/foldvision-encoder")
4model.eval()
5# x: (B, 5, Z, Y, X)
6# z = model(x) # (B, 1024)1@article{foldvision_biorxiv,
2 title = {FoldVision: A compute-efficient atom-level 3D protein encoder},
3 author = {Kroll, Alexander and Yadav, Shantanu and Lercher, Martin J.},
4 journal = {bioRxiv},
5 year = {2026},
6 doi = {10.64898/2026.01.23.701326},
7 url = {https://doi.org/10.64898/2026.01.23.701326}
8}1@misc{foldvision_github,
2 title = {FoldVision code repository},
3 author = {Kroll, Alexander},
4 year = {2026},
5 howpublished = {\url{https://github.com/AlexanderKroll/foldvision}}
6}