Calibration

AA-Omniscience Accuracy → CritPt

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

CritPt is estimated from AA-Omniscience Accuracy with a Hill curve fitted on 183 models measured on both: y = 0.0024 + (1.2000 − 0.0024)·x^2.87 / (0.94267^2.87 + x^2.87), R² = 0.68, cross-validated error 5.0 pp. It is used for 0 estimates.