Recently, the research team led by Liang Yansheng and Lei Ming from the School of Physics at Xi'an Jiaotong University, in a collaborative effort with scholars from Nanjing University, The Hong Kong University of Science and Technology, and The University of Hong Kong, has made remarkable strides in the realm of large-scale metasurface intelligent inverse design. The team has ingeniously developed MetaE-former, an electromagnetic surrogate solver that leverages the power of the neighborhood attention Transformer. Through the strategic integration of transfer learning and inverse optimization techniques, they have effectively addressed the long-standing challenge of balancing computational efficiency with design flexibility in large-scale metasurface design. This breakthrough has paved the way for the swift optimization of millimeter-scale metalenses with high numerical apertures. The pertinent research outcomes have been featured in PhotoniX, a leading international journal in the field of photonics.