Abstract: Accurate solution of the many-electron Schrödinger equation is one of the central challenges of modern science, as its inherent exponential complexity has long constrained our understanding of complex molecular and material systems. In recent years, artificial intelligence methods, particularly those based on neural network quantum states, have opened up a viable alternative path to address this problem. The method based on the generative Transformer architecture, known as QiankunNet, combines the expressive power of neural networks with the fundamental principles of quantum mechanics, and has made progress across multiple dimensions. This method has not only achieved high accuracy in ground-state energy calculations but also has substantially extended the systematic framework: it enables interatomic force calculations, simulations of periodic solid-state materials, optimization for large-scale parallel computing, and integration with methods such as quantum embedding theory and tensor networks. This article reviews the development of neural network quantum state methods, gives a detailed introduction to the theoretical foundations and technical advancements of QiankunNet, analyzes its successful applications in molecular systems, solid-state materials and strongly correlated systems, and discusses the potential value and future directions of this field in interdisciplinary research.