Calibration
AA-HLE → HLE-Verified
HLE-Verified is estimated from AA-HLE with a offset logistic curve fitted on 6 models measured on both: y = 0.0000 + (0.5435 − 0.0000) / (1 + exp(−154.46·(x − 0.4112))), R² = 1.00, cross-validated error 6.8 pp. It is used for 2 estimates.
| Estimated model | AA-HLE | HLE-Verified | Source |
|---|---|---|---|
| Claude Haiku 5.5 | 44.4% | 54.0% | estimated ± 6.8 pp, low confidence |
| GPT-4 Turbo | 3.1% | 0.0% | estimated ± 6.8 pp, low confidence |