Calibration

ProgramBench → AA-SciCode

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap

AA-SciCode is estimated from ProgramBench with a offset logistic curve fitted on 12 models measured on both: y = 0.5369 + (0.6196 − 0.5369) / (1 + exp(−40.61·(x − 0.7635))), R² = 0.46, cross-validated error 5.4 pp. It is used for 0 estimates.