Calibration
PostTrainBench v1.1 → SWE-bench Pro
SWE-bench Pro is estimated from PostTrainBench v1.1 with a linear curve fitted on 9 models measured on both: y = 1.0776·x + 0.3436, R² = 0.75, cross-validated error 6.4 pp. It is used for 0 estimates.
SWE-bench Pro is estimated from PostTrainBench v1.1 with a linear curve fitted on 9 models measured on both: y = 1.0776·x + 0.3436, R² = 0.75, cross-validated error 6.4 pp. It is used for 0 estimates.