Calibration

Decision Index 0.3 Full Score → Decision Index 0.3 Retrieval

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

Decision Index 0.3 Retrieval is estimated from Decision Index 0.3 Full Score with a offset logistic curve fitted on 112 models measured on both: y = 0.0000 + (0.9910 − 0.0000) / (1 + exp(−4.18·(x − 0.4623))), R² = 0.81, cross-validated error 7.4 pp. It is used for 0 estimates.