Calibration
ApprenticeBench → AA Harvey LAB
AA Harvey LAB is estimated from ApprenticeBench with a offset logistic curve fitted on 5 models measured on both: y = 0.0000 + (0.9368 − 0.0000) / (1 + exp(−200.00·(x − 0.1429))), R² = 0.83, cross-validated error 2.0 pp. It is used for 2 estimates.
| Estimated model | ApprenticeBench | AA Harvey LAB | Source |
|---|---|---|---|
| Claude Opus 4.6 | 5.0% | 0.0% | estimated ± 2.0 pp, low confidence |
| GPT-5.4 | 11.0% | 0.1% | estimated ± 2.0 pp, low confidence |