List your company

Confidence Calibration on ReClor (test)

4.4ECEORCE
2.64414.49726.3538.203May 12, 2026

Evaluation results

MethodLinks
ORCEFoundation Model=Qwen3 8B2026.054.40.6268.0510.0640.04179.4
Top-kFoundation Model=Qwen3 8B2026.054.60.1332.7970.1610.13980
ConfTunerFoundation Model=Qwen3 8B2026.054.70.3516.0470.0910.06979.4
ORCEFoundation Model=Mistral 7B2026.056.10.3928.9050.3060.16151.2
SFTFoundation Model=Qwen3 8B2026.056.20.5379.4310.0830.0679.4
ORCEFoundation Model=LLAMA-3 8B2026.056.50.5448.0610.1690.0860.8
SFTFoundation Model=Mistral 7B2026.057.30.3442.3680.3290.18451.2
SFTFoundation Model=LLAMA-3 8B2026.058.40.4468.3780.2010.11260.8
CoTFoundation Model=Qwen3 8B2026.0512.40.0840.0620.1830.16179.6
ConfTunerFoundation Model=LLAMA-3 8B2026.05140.3433.1960.250.16160.8
ConfTunerFoundation Model=Mistral 7B2026.0514.10.1911.7120.4080.26351.2
VanillaFoundation Model=Qwen3 8B2026.0516.40.0590.1870.190.16779.4
ConsistencyFoundation Model=Qwen3 8B2026.0516.70.3111.2060.1470.12680.2
VanillaFoundation Model=LLAMA-3 8B2026.0517.80.0530.2390.3730.28360.8
Top-kFoundation Model=LLAMA-3 8B2026.05270.0610.1770.4410.29851.6
CoTFoundation Model=Mistral 7B2026.05300.1240.0060.3890.27155.6
CoTFoundation Model=LLAMA-3 8B2026.0533.80.0670.1340.4410.30452.4
Top-kFoundation Model=Mistral 7B2026.0535.70.1536.2160.4380.29251
ConsistencyFoundation Model=LLAMA-3 8B2026.0537.60.0720.110.3550.26961.6
VanillaFoundation Model=Mistral 7B2026.0546.80.0490.2750.4770.33251.2
ConsistencyFoundation Model=Mistral 7B2026.0548.30.0290.5230.4890.34651.4