TEDBench/miae-l-sc

TEDBench — Supervised from scratch (structure only)

Variant: miae_l  |  Parameters: 339M  |  Layers: 24  |  Hidden dim: 1024  |  Attn heads: 16

This MiAEClassifier was trained from scratch on TEDBench without pretraining. It is part of the research presented in the paper Protein Fold Classification at Scale: Benchmarking and Pretraining.

MiAE is an SE(3)-invariant masked autoencoder that masks up to 90% of backbone frames and reconstructs the full structure with a lightweight decoder. It establishes a strong recipe for protein fold classification on the TEDBench benchmark.

Architecture sizes

Variant	Params	Layers	Hidden dim	Attn heads
miae_s	29 M	6	512	8
miae_b	102 M	12	768	12
miae_l	339 M	24	1 024	16

Append +model.use_seq_input=true to miae_b for the +seq variant.

Usage

Load from the HuggingFace Hub

from tedbench.utils.io import load_from_hf

model = load_from_hf("TEDBench/miae-l-sc")
model.eval()

From a Lightning checkpoint

from tedbench.model import MiAEClassifier

model = MiAEClassifier.load_from_checkpoint("model.ckpt", weights_only=False)
model.eval()

Links

• Code: BorgwardtLab/TEDBench
• Paper: arXiv:2605.18552

Citation

@inproceedings{chen2026tedbench,
  title={Protein Fold Classification at Scale: Benchmarking and Pretraining},
  author={Chen, Dexiong and Manolache, Andrei and Niepert, Mathias and Borgwardt, Karsten},
  booktitle={Proceedings of the 43rd International Conference on Machine Learning},
  year={2026}
}