Calibration

AA-HLE → ARC-AGI-2

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

ARC-AGI-2 is estimated from AA-HLE with a Hill curve fitted on 34 models measured on both: y = 0.2603 + (1.0614 − 0.2603)·x^6.00 / (0.42493^6.00 + x^6.00), R² = 0.72, cross-validated error 14.9 pp. It is used for 0 estimates.