Investigated the validity of the Oracle Pruning in modern deep learning models, revealing that pre-retraining performance is poorly correlated with post-retraining, challenging a 35-year-old assumption in network pruning (one first-author paper).
Investigated the validity of the Oracle Pruning in modern deep learning models, revealing that pre-retraining performance is poorly correlated with post-retraining, challenging a 35-year-old assumption in network pruning (one first-author paper).