Enhancing One-shot Pruned Pre-trained Language Models through Sparse-Dense-Sparse Mechanism

Open in new window