Worked on the intel/neural-compressor repository to refactor and clean up the fp8_quant module, focusing on improving code architecture and maintainability. The effort involved reorganizing the folder structure, consolidating utility functions, and removing outdated scripts related to diffusion and super-resolution evaluation. By addressing technical debt and streamlining the codebase, the changes established a more maintainable foundation for future FP8 quantization development. The work emphasized code refactoring and file management using Python, resulting in a cleaner project structure that facilitates faster onboarding and smoother feature expansion. No bugs were fixed during this period, with all efforts directed toward architectural improvements.