Calibration

HLE-Verified → AA-HLE

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

AA-HLE is estimated from HLE-Verified with a offset logistic curve fitted on 6 models measured on both: y = 0.4150 + (0.5082 − 0.4150) / (1 + exp(−141.67·(x − 0.5251))), R² = 0.73, cross-validated error 3.8 pp. It is used for 0 estimates.