At the end of 2020, AlphaFold2 achieved high-precision prediction in the CASP competition for protein structure prediction, making a usually obscure niche subject into a household name. AlphaFold's success is a qualitative change in the field of structure prediction over the years of quantitative changes. Two of the important quantitative changes are the transition from predicting classification to predicting continuous edges and angles (interatomic distances and dihedral angles). Here, Dr. Yaoqi Zhou mainly tells the story of the first realization of the continuous dihedral angle prediction, and how it triggered the chain reaction of end-to-end protein structure prediction, and finally led to the revolutionary breakthrough.