Calibration

Decision Index 0.3 Retrieval → Decision Index 0.3 Knowledge

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

Decision Index 0.3 Knowledge is estimated from Decision Index 0.3 Retrieval with a Hill curve fitted on 112 models measured on both: y = 0.0541 + (1.2000 − 0.0541)·x^2.67 / (0.83083^2.67 + x^2.67), R² = 0.81, cross-validated error 6.1 pp. It is used for 0 estimates.