🧬 BioMatrix-4B-SFT
A multimodal biological foundation model that natively integrates molecules (SMILES, SELFIES, 3D structures), proteins (sequences, 3D structures), and natural language in a single model.
Wrap biomolecular content with control tokens:
- SMILES:
<|mol_smi_start|>...<|mol_smi_end|> - SELFIES:
<|mol_sfi_start|>...<|mol_sfi_end|> - Protein 1D:
<|prot_aa_start|>...<|prot_aa_end|>
Examples