Calibration

C-Eval → MMLU-Pro

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

MMLU-Pro is estimated from C-Eval with a Hill curve fitted on 6 models measured on both: y = 0.0000 + (1.0878 − 0.0000)·x^6.00 / (0.72697^6.00 + x^6.00), R² = 0.94, cross-validated error 1.1 pp. It is used for 0 estimates.