Worked on the google/security-research repository to address a critical issue affecting the accuracy of benchmark results for Retbleed configurations. Focused on data analysis to identify and correct incorrect performance metrics, ensuring that research conclusions drawn from these benchmarks are reliable. Updated the benchmarking documentation using Markdown, revising the results table and related content to reflect the corrected data. Collaborated with another contributor to co-author documentation improvements, further enhancing the clarity and accuracy of published benchmarks. The work emphasized careful validation of metrics across multiple configuration settings, combining technical precision with thorough documentation practices to support ongoing security research.