Deep-learning (DL)-based channel state information (CSI) feedback has attracted a great deal of attention due to its effectiveness in compressing CSI in massive multiple-input multiple-output systems.