Helmholtz AI - Forschungszentrum Jülich

multimodal protein models to project different protein modalities into a unified embedding space

These are GSQ image tokenizer's checkpoint for work https://arxiv.org/abs/2412.02632