Calibration

HLE-Verified → ARC-AGI-1

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

ARC-AGI-1 is estimated from HLE-Verified with a offset logistic curve fitted on 5 models measured on both: y = 0.9595 + (1.2000 − 0.9595) / (1 + exp(−200.00·(x − 0.5599))), R² = 0.72, cross-validated error 1.1 pp. It is used for 0 estimates.