Deep learning model compression using network sensitivity and gradients

Re-assign community