用于模型压缩的结构化多哈希 Despite the success of deep neural networks (DNNs), state-of-the-art models are too large to deploy on low-resource devices or common server configura