Calibration
C-Eval → GPQA
GPQA is estimated from C-Eval with a linear curve fitted on 5 models measured on both: y = 1.0485·x + -0.0852, R² = 0.71, cross-validated error 1.2 pp. It is used for 0 estimates.
GPQA is estimated from C-Eval with a linear curve fitted on 5 models measured on both: y = 1.0485·x + -0.0852, R² = 0.71, cross-validated error 1.2 pp. It is used for 0 estimates.