Cross-Model Generalization

The ability of a detector to maintain accuracy when encountering outputs from unseen or differently architected new models.