Mind the Gap Between Synthetic and Real: Utilizing Transfer Learning to Probe the Boundaries of Stable Diffusion Generated Data

no code implementations • 6 May 2024 • Leonhard Hennicke, Christian Medeiros Adriano, Holger Giese, Jan Mathias Koehler, Lukas Schott

Building upon our insights that mainly later layers are responsible for the drop, we investigate the data-efficiency of fine-tuning a synthetically trained model with real data applied to only those last layers.

Data-free Knowledge Distillation Transfer Learning