Views
No views yet
| Item | Value |
|---|---|
| Architecture | Caduceus-Ph (d_model=256, Mamba backbone) |
| Pretraining data | ViroBlend (~216 Mbp) |
pip install torch transformers mamba-ssm causal-conv1dpython get_embedding.pypytorch_model.bin):1import torch
2from transformers import AutoModelForMaskedLM, AutoTokenizer
3
4BASE = "kuleshov-group/caduceus-ph_seqlen-131k_d_model-256_n_layer-16"
5# REPO = "YDXX/ViroCaduceus" # after uploading to Hugging Face
6
7tokenizer = AutoTokenizer.from_pretrained(BASE, trust_remote_code=True)
8model = AutoModelForMaskedLM.from_pretrained(BASE, trust_remote_code=True)
9# load ViroCaduceus weights from pytorch_model.bin if needed (see get_embedding.py)config.json — training export configpytorch_model.bin — fine-tuned backbone weightsget_embedding.py — minimal embedding demo1@article{ye2026virobench,
2 title={ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks},
3 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},
4 journal={arXiv preprint arXiv:2605.25388},
5 year={2026}
6}