ResearchBenchmarksCalibration on SciKnowEval Edu. MMLU-Pro (test)Follow21.1Average Calibration ScoreIntuitor16.31617.55818.820.042Jul 2, 2026Evaluation ResultsMethodMethodLinksAverage Calibration ScoreIntuitorEval.=crossEval.=cross2026.0721.1TTRLEval.=inEval.=in2026.0719.6LMSIEval.=inEval.=in2026.0719.1TTRLEval.=crossEval.=cross2026.0719.1Qwen3-4B baseEval.=baseEval.=base2026.0718.4N-OPSDEval.=crossEval.=cross2026.0718.3IntuitorEval.=inEval.=in2026.0718.2N-OPSDEval.=inEval.=in2026.0717.6LMSIEval.=crossEval.=cross2026.0716.5