NVIDIA researchers recently introduced a text-to-image model called Perfusion, which stands out for its compact size, requiring only 100KB of space and a training time of just 4 minutes. Despite its small size, the generated images are still impressive.
NVIDIA collaborated with researchers fr...