YDXX/ViroHyena-6m

ViroHyena-6.6M

ViroHyena is a Hyena-based nucleotide language model introduced in the paper ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks.

This specific checkpoint is the 6.6M parameter version. It was pre-trained on the ViroBlend corpus, a 216 Mbp mixed pretraining dataset using source-wise stratified sampling to balance human reference, multi-species genomes, and viral in-domain sequences.

Resources

• Paper: ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks
• Repository: GitHub - ViroBench

Model Configurations

Model	Params	d_model	Layers
ViroHyena-436K	0.436M	128	2
ViroHyena-1.6M	1.6M	256	2
ViroHyena-6.6M	6.6M	256	8
ViroHyena-253M	253M	1024	20

Biosecurity & Responsible Use

This model was developed for biological understanding and measurement research. As noted in the project documentation:

• Do NOT use this model to design, optimize, synthesize, or operationalize harmful biological agents.
• Follow all applicable laws, institutional policies, and biosafety regulations.
• If you release checkpoints or generation outputs, consider controlled access and red-teaming protocols.

Citation

@article{ye2026virobench,
  title={ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks},
  author={Ye, Dongxin and Hu, Fang and Hu, Han and Hu, Shu and Tan, Yang and Ouyang, Wanli and Li, Stan Z and Cui, Jie and Dong, Nanqing},
  journal={arXiv preprint arXiv:2605.25388},
  year={2026}
}