List your company

Uncertainty Quantification on MMLU-pro (test)

77.74AUROCCAGE-CAL
42.286451.490760.69569.8993May 13, 2026May 15, 2026May 18, 2026May 20, 2026May 23, 2026May 25, 2026May 28, 2026

Evaluation results

MethodLinks
CAGE-CALRollouts=3, Topologies=52026.0577.74———61.54
BaseModel=LLAMA-8B, Position=After2026.0574.136.60.4230.455—
BaseModel=GEMMA-12B, Position=After2026.0573.252.60.3870.406—
UnlearnedModel=PHI-4, Position=After2026.0571.9680.2060.144—
DropoutModel=LLAMA-8B, Position=After2026.0570.836.60.2190.1—
DropoutModel=GEMMA-12B, Position=After2026.0569.452.60.3010.274—
BaseModel=PHI-4, Position=After2026.0568.3680.2650.261—
UnlearnedModel=GEMMA-12B, Position=After2026.0567.652.60.2690.176—
DropoutModel=PHI-4, Position=After2026.0565.5680.2060.086—
BaseModel=PHI-4, Position=Before2026.0563.6680.2190.106—
UnlearnedModel=LLAMA-8B, Position=After2026.0563.136.60.2810.212—
UnlearnedModel=PHI-4, Position=Before2026.0562.1680.2250.12—
Plurality voteRollouts=3, Topologies=52026.0561.78———53.02
MATURollouts=3, Topologies=52026.0561.39———52.2
Answer entropyRollouts=3, Topologies=52026.0557.92———51.01
DAERollouts=3, Topologies=52026.0557.91———51.01
DropoutModel=LLAMA-8B, Position=Before2026.0556.336.60.2520.125—
BaseModel=LLAMA-8B, Position=Before2026.0555.836.60.5520.565—
UnlearnedModel=GEMMA-12B, Position=Before2026.0555.452.60.4090.399—
DropoutModel=PHI-4, Position=Before2026.0554.2680.2410.141—
UnlearnedModel=LLAMA-8B, Position=Before2026.0553.436.60.3760.347—
DropoutModel=GEMMA-12B, Position=Before2026.0549.252.60.3650.324—
BaseModel=GEMMA-12B, Position=Before2026.0546.752.60.4530.448—
Avg-logprobRollouts=3, Topologies=52026.0543.65———27.33