【論文閱讀】韓鬆《Efficient Methods And Hardware For Deep Learning》節選《Deep compression》

Trained Quantization and Deep Compression 本章內容來在ICLR 2016論文《Deep compression: Compressing deep neural networks with pruning, trained quantization and huffman coding》。 DeepCompression由剪枝、量化和變長編碼組成，它可將神經

>>阅读原文<<