Developed and delivered the SPHERE-based model weight editing feature for the zjunlp/EasyEdit repository, focusing on energy-regularized sequential model editing on hyperspheres. This work involved implementing the SPHERE method with new hyperparameter configurations and core editing functions using Python and PyTorch, enabling precise and safe weight modifications. Practical example scripts were created to demonstrate integration into existing model editing workflows, supporting reproducible experiments and faster onboarding. The feature reduces the risk of unintended model drift and improves maintainability, with an emphasis on code quality and documentation. No major bugs were addressed, as the primary focus remained on robust feature development.