1 paper across 1 session
Training-free depth pruning method for transformer based models via approximating blocks with linear transformation