Beyond Linear Approximations: A Novel Pruning Approach for Attention Matrix

Open in new window