Boltzgen Structure Small

boltzgen-structure-small

OFoldX pipeline artifact for biomolecular design generation, using the boltzgen-structure-small architecture.

Disclaimer

This model card was generated by the OFoldX team for an OFoldX pipeline artifact. The upstream model authors did not write this card unless explicitly stated otherwise.

OFoldX is pre-alpha research software. Check the source checkpoint, upstream release, and local validation before using the artifact for scientific or operational decisions.

Model Details

Small BoltzGen structure-conditioned generation checkpoint.

Converted small BoltzGen structure-trained checkpoint for structure-conditioned generation.

Model Provenance

• Upstream Project: BoltzGen
• Source Checkpoint: boltzgen1_structuretrained_small.ckpt
• Source Release: https://huggingface.co/boltzgen/boltzgen-1
• Primary Paper: BoltzGen: Toward Universal Binder Design
• Upstream License: MIT for upstream BoltzGen code

Model Specification

Field	Value
Repository	oteam/boltzgen-structure-small
Artifact Kind	pipeline
Task	design_generation
Architecture	boltzgen-structure-small
Entrypoint	ofoldx.pipelines.design.DesignPipeline
Source Checkpoint	boltzgen1_structuretrained_small.ckpt

[!NOTE] Source checkpoint: boltzgen1_structuretrained_small.ckpt.

Links

• Hub repository: oteam/boltzgen-structure-small
• Upstream paper: BoltzGen: Toward Universal Binder Design
• Upstream repository: BoltzGen
• Source checkpoint release: https://huggingface.co/boltzgen/boltzgen-1
• Code: ofoldx/pipelines/design.py
• Project repository: https://github.com/OTeam-AI4S/OFoldX
• Issues: https://github.com/OTeam-AI4S/OFoldX/issues

Usage

The artifact depends on the ofoldx library. Install it with pip:

pip install ofoldx

Pipeline Usage

Load the artifact from oteam/boltzgen-structure-small with the OFoldX task pipeline. Use AutoModel or AutoProcessor only when you need lower-level control:

from ofoldx.pipelines import Pipeline

pipeline = Pipeline.from_pretrained("oteam/boltzgen-structure-small")

When a matching processor is available, load it with AutoProcessor.from_pretrained(...) and pass the processed batch to the model.

Interface

• Task: design_generation
• Artifact kind: pipeline
• Architecture: boltzgen-structure-small
• Runtime files: manifest.json, config.json, and model.safetensors when present

Training Details

OFoldX did not train these weights. This repository contains a converted checkpoint and OFoldX runtime metadata for loading it.

Training Data

BoltzGen builds on the Boltz-2 data pipeline. The public training release includes targets.zip and msa.zip in the boltzgen/boltzgen1_train dataset and molecule dictionaries in boltzgen/inference-data; this artifact corresponds to the released small structure-trained checkpoint.

Training Procedure

OFoldX converts the released BoltzGen structure-trained checkpoint; it does not run BoltzGen training.

Evaluation

OFoldX conversion reports and contract tests validate artifact structure and checkpoint loading. Task-level scientific evaluation should be checked against the corresponding upstream model release or paper.

Limitations

• This artifact is distributed for research use.
• Inputs must match the model-specific processor and expected biomolecular representation.
• OFoldX is pre-alpha, so APIs and artifact metadata may still change before a stable release.

Citation

Please cite the upstream BoltzGen work for the source checkpoint. If OFoldX supports your work, please also cite or link the OFoldX project repository.

@article{stark2025boltzgen,
  author = {Stark, Hannes and Faltings, Felix and Choi, MinGyu and Xie, Yuxin and Hur, Eunsu and O'Donnell, Timothy John and Bushuiev, Anton and Ucar, Talip and Passaro, Saro and Mao, Weian and others},
  title = {BoltzGen: Toward Universal Binder Design},
  year = {2025},
  doi = {10.1101/2025.11.20.689494},
  journal = {bioRxiv}
}

Contact

Please use OFoldX GitHub issues for questions or comments about this model card.

License

The Hub license metadata, when present, reflects the source checkpoint or upstream project license. The OFoldX project license is not yet finalized. The source checkpoint is associated with the upstream license noted above: MIT for upstream BoltzGen code. Review both OFoldX and upstream terms before redistribution or production use.