Not All Models Are Equal: Predicting Model Transferability in a Self-challenging Fisher Space

Re-assign community