Which of the following isnota core capability of the AI Middle Platform?

Indigenous, self-controlled full-stack technology: from chips and OS to AI frameworks and applications

Training downshift: local scenario datasets feeding back to enable integrated training + inference and incremental training for continuous algorithm improvement

Cloud–edge collaboration: one-click model deployment; shared capabilities across city/district/street; highly efficient resource collaboration

Super-large number of algorithms: having the largest number of video-recognition algorithms in the industry