Viral-Host-Hunter (VHH) is a deep learning framework for predicting virus-host interactions across taxonomic levels, leveraging protein language models such as ProtT5-XL-UniRef50 for biological sequence embedding and host inference.