ResearchBenchmarksCode Generation on HumanEval (MAE, MRE)Follow0.0171MAEPower Law0.0166120.0199060.02320.026494Dec 9, 2025Evaluation ResultsMethodMethodLinksMAEMRE (%)Power LawMetric Type=pass@1Metric Type=pass@12025.120.01715.11BNSLMetric Type=pass@1Metric Type=pass@12025.120.02938.29