Calibration

MedXpertQA (Text) → HLE w/o tools

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

HLE w/o tools is estimated from MedXpertQA (Text) with a Michaelis–Menten curve fitted on 5 models measured on both: y = 0.9944·x / (0.84666 + x), R² = 0.47, cross-validated error 5.0 pp. It is used for 0 estimates.