List your company

Question Answering on HotpotQA (test) (Calibration and Uncertainty Metrics)

65.71AccuracyCorrectness-only
-2.628415.113332.85550.5967Jul 5, 2026

Evaluation results

MethodLinks
Correctness-onlyReward Scheme=Correctness-only2026.0765.710.50950.32070.33630.3346-0.3345
Underconfidence-1Reward Scheme=Underconfidence-12026.0765.510.60460.42380.23130.1175-0.1175
Overconfidence-1000Reward Scheme=Overconfidence-10002026.0765.480.52730.3220.33280.3307-0.3305
Log-1/ln 202Reward Scheme=Log-1/ln 2022026.0765.370.54550.40770.24610.1491-0.1491
Underconfidence-4Reward Scheme=Underconfidence-42026.0765.290.61580.42420.22870.1093-0.1093
Overconfidence-4Reward Scheme=Overconfidence-42026.0765.260.57250.41340.23930.1319-0.1319
Brier-2Reward Scheme=Brier-22026.0765.070.5920.42210.22860.1057-0.1057
Overconfidence-1Reward Scheme=Overconfidence-12026.0765.070.58540.40590.24480.1563-0.1562
Brier-1Reward Scheme=Brier-12026.0765.040.5790.41440.2360.1225-0.1225
Log-1Reward Scheme=Log-12026.0764.720.6420.42770.21940.0777-0.0776
Brier-log HybridReward Scheme=Brier-log Hybrid2026.0764.390.60980.41280.23110.1081-0.108
Base (JSON)Format=JSON2026.0744.040.585-0.02860.46890.473-0.4693
SFT (XML)Format=XML2026.0743.620.6044-0.02080.4570.4644-0.4634
Base (XML)Format=XML2026.074.790.5315-0.73090.77880.8399-0.8399
Log LossReward Scheme=Log Loss2026.070—00.00010.005-0.0049
Brier ScoreReward Scheme=Brier Score2026.070—-0.00010.00010.0058-0.0058