Contributed to the PaddlePaddle/docs and PaddlePaddle/Paddle repositories by enhancing both documentation reliability and API usability. Addressed a broken link in the Optimizer Guide, ensuring users could access accurate backpropagation resources and improving onboarding for neural network practitioners. Developed parameter alias support for the count_nonzero API, enabling compatibility with both Paddle and PyTorch-style arguments, and implemented comprehensive cross-framework unit tests. Updated API documentation to reflect these changes, supporting developer clarity and reducing support requests. Work was completed using Python and reStructuredText, with a focus on API development, unit testing, and technical writing to ensure stable, maintainable contributions.