List your company

Predicting Answer Correctness on TriviaQA (val)

0.904AUROCBaseline (Llama-3.1-8B)
0.855120.867810.88050.89319Jun 1, 2026

Evaluation results

MethodLinks
Baseline (Llama-3.1-8B)Features=all UQ, Strategy=baseline2026.060.904——
Gated Experts (Llama-3.1-8B)Features=all UQ, Strategy=latent gated2026.060.904——
Selective Prediction (Llama-3.1-8B)Features=all UQ, Strategy=latent selective2026.060.904——
Selective Prediction (Spectrum-Qwen3-14B)Features=all UQ, Strategy=latent selective2026.060.904——
Gated Experts Oracle (Llama-3.1-8B)Features=all UQ, Strategy=oracle gated2026.060.903——
Selective Prediction Oracle (Llama-3.1-8B)Features=all UQ, Strategy=oracle selective2026.060.903——
Gated Experts (Spectrum-Qwen3-14B)Features=all UQ, Strategy=latent gated2026.060.903——
Spectrum-Qwen3-14BFeatures=all UQ, Strategy=baseline2026.060.902——
all UQBackbone=Spectrum-Qwen3-14B, Framework=latent selective2026.060.897——
all UQBackbone=Spectrum-Qwen3-14B, Framework=latent gated2026.060.896——
all UQBackbone=Spectrum-Qwen3-14B, Framework=baseline2026.060.895——
all UQBackbone=Spectrum-Qwen3-14B, Framework=oracle gated2026.060.895——
all UQBackbone=Spectrum-Qwen3-14B, Framework=oracle selective2026.060.895——
all UQBackbone=Llama-3.1-8B, Framework=latent selective2026.060.888——
SARBackbone=Spectrum-Qwen3-14B, Framework=latent selective2026.060.888——
Selective Prediction (Spectrum-Llama-3.1-8B)Features=all UQ, Strategy=latent selective2026.060.887——
all UQBackbone=Llama-3.1-8B, Framework=latent gated2026.060.887——
SARBackbone=Spectrum-Qwen3-14B, Framework=latent gated2026.060.887——
all UQBackbone=Llama-3.1-8B, Framework=baseline2026.060.886——
all UQBackbone=Llama-3.1-8B, Framework=oracle gated2026.060.886——
all UQBackbone=Llama-3.1-8B, Framework=oracle selective2026.060.886——
CoCoABackbone=Spectrum-Qwen3-14B, Framework=latent gated2026.060.886——
CoCoABackbone=Spectrum-Qwen3-14B, Framework=latent selective2026.060.886——
SARBackbone=Spectrum-Qwen3-14B, Framework=baseline2026.060.886——
SARBackbone=Spectrum-Qwen3-14B, Framework=oracle gated2026.060.886——
SARBackbone=Spectrum-Qwen3-14B, Framework=oracle selective2026.060.886——
Gated Experts (Spectrum-Llama-3.1-8B)Features=all UQ, Strategy=latent gated2026.060.885——
Semantic entropyBackbone=Spectrum-Qwen3-14B, Framework=latent selective2026.060.885——
Spectrum-Llama-3.1-8BFeatures=all UQ, Strategy=baseline2026.060.884——
CoCoABackbone=Spectrum-Qwen3-14B, Framework=baseline2026.060.884——
CoCoABackbone=Spectrum-Qwen3-14B, Framework=oracle gated2026.060.884——
CoCoABackbone=Spectrum-Qwen3-14B, Framework=oracle selective2026.060.884——
Semantic entropyBackbone=Spectrum-Qwen3-14B, Framework=latent gated2026.060.884——
Semantic entropyBackbone=Spectrum-Qwen3-14B, Framework=oracle gated2026.060.882——
Semantic entropyBackbone=Spectrum-Qwen3-14B, Framework=oracle selective2026.060.882——
Semantic entropyBackbone=Spectrum-Qwen3-14B, Framework=baseline2026.060.881——
all UQBackbone=Spectrum-Llama-3.1-8B, Framework=latent selective2026.060.88——
Semantic energyBackbone=Spectrum-Qwen3-14B, Framework=latent selective2026.060.88——
all UQBackbone=Spectrum-Llama-3.1-8B, Framework=latent gated2026.060.878——
Semantic energyBackbone=Spectrum-Qwen3-14B, Framework=latent gated2026.060.878——
all UQBackbone=Spectrum-Llama-3.1-8B, Framework=baseline2026.060.877——
all UQBackbone=Spectrum-Llama-3.1-8B, Framework=oracle gated2026.060.877——
all UQBackbone=Spectrum-Llama-3.1-8B, Framework=oracle selective2026.060.877——
Semantic energyBackbone=Spectrum-Qwen3-14B, Framework=baseline2026.060.877——
Semantic energyBackbone=Spectrum-Qwen3-14B, Framework=oracle gated2026.060.877——
Semantic energyBackbone=Spectrum-Qwen3-14B, Framework=oracle selective2026.060.877——
CoCoABackbone=Llama-3.1-8B, Framework=latent selective2026.060.875——
CoCoABackbone=Llama-3.1-8B, Framework=baseline2026.060.874——
CoCoABackbone=Llama-3.1-8B, Framework=latent gated2026.060.874——
CoCoABackbone=Llama-3.1-8B, Framework=oracle gated2026.060.874——
CoCoABackbone=Llama-3.1-8B, Framework=oracle selective2026.060.874——
SARBackbone=Llama-3.1-8B, Framework=latent gated2026.060.874——
SARBackbone=Llama-3.1-8B, Framework=latent selective2026.060.874——
SARBackbone=Llama-3.1-8B, Framework=baseline2026.060.873——
SARBackbone=Llama-3.1-8B, Framework=oracle gated2026.060.873——
SARBackbone=Llama-3.1-8B, Framework=oracle selective2026.060.873——
Selective Prediction (Qwen3-14B)Features=all UQ, Strategy=latent selective2026.060.872——
all UQBackbone=Qwen3-14B, Framework=latent selective2026.060.872——
Semantic entropyBackbone=Llama-3.1-8B, Framework=latent selective2026.060.871——
Semantic entropyBackbone=Llama-3.1-8B, Framework=latent gated2026.060.87——
Gated Experts (Llama-3.1-8B)Features=MI, Strategy=latent gated2026.060.869——
Semantic entropyBackbone=Llama-3.1-8B, Framework=baseline2026.060.868——
Semantic entropyBackbone=Llama-3.1-8B, Framework=oracle gated2026.060.868——
Semantic entropyBackbone=Llama-3.1-8B, Framework=oracle selective2026.060.868——
Semantic energyBackbone=Llama-3.1-8B, Framework=latent selective2026.060.867——
Semantic entropyBackbone=Spectrum-Llama-3.1-8B, Framework=latent selective2026.060.867——
SARBackbone=Spectrum-Llama-3.1-8B, Framework=latent selective2026.060.867——
Semantic energyBackbone=Llama-3.1-8B, Framework=latent gated2026.060.866——
Semantic energyBackbone=Llama-3.1-8B, Framework=baseline2026.060.865——
Semantic energyBackbone=Llama-3.1-8B, Framework=oracle gated2026.060.865——
Semantic energyBackbone=Llama-3.1-8B, Framework=oracle selective2026.060.865——
SARBackbone=Spectrum-Llama-3.1-8B, Framework=latent gated2026.060.865——
all UQBackbone=Qwen3-14B, Framework=latent gated2026.060.864——
Semantic entropyBackbone=Spectrum-Llama-3.1-8B, Framework=latent gated2026.060.864——
SARBackbone=Spectrum-Llama-3.1-8B, Framework=baseline2026.060.864——
SARBackbone=Spectrum-Llama-3.1-8B, Framework=oracle gated2026.060.864——
SARBackbone=Spectrum-Llama-3.1-8B, Framework=oracle selective2026.060.864——
Gated Experts (Qwen3-14B)Features=all UQ, Strategy=latent gated2026.060.863——
CoCoABackbone=Spectrum-Llama-3.1-8B, Framework=latent selective2026.060.863——
Semantic energyBackbone=Spectrum-Llama-3.1-8B, Framework=latent selective2026.060.863——
MIBackbone=Spectrum-Qwen3-14B, Framework=latent selective2026.060.863——
SARBackbone=Qwen3-14B, Framework=latent selective2026.060.862——
all UQBackbone=Qwen3-14B, Framework=baseline2026.060.861——
all UQBackbone=Qwen3-14B, Framework=oracle gated2026.060.861——
Semantic energyBackbone=Qwen3-14B, Framework=latent selective2026.060.861——
CoCoABackbone=Spectrum-Llama-3.1-8B, Framework=latent gated2026.060.861——
Semantic entropyBackbone=Spectrum-Llama-3.1-8B, Framework=oracle gated2026.060.861——
MIBackbone=Spectrum-Qwen3-14B, Framework=latent gated2026.060.861——
all UQBackbone=Qwen3-14B, Framework=oracle selective2026.060.86——
CoCoABackbone=Spectrum-Llama-3.1-8B, Framework=oracle gated2026.060.86——
Semantic energyBackbone=Spectrum-Llama-3.1-8B, Framework=latent gated2026.060.86——
Semantic entropyBackbone=Spectrum-Llama-3.1-8B, Framework=baseline2026.060.86——
Semantic entropyBackbone=Spectrum-Llama-3.1-8B, Framework=oracle selective2026.060.86——
MSPBackbone=Spectrum-Qwen3-14B, Framework=latent selective2026.060.86——
CoCoABackbone=Spectrum-Llama-3.1-8B, Framework=baseline2026.060.859——
CoCoABackbone=Spectrum-Llama-3.1-8B, Framework=oracle selective2026.060.859——
Semantic energyBackbone=Spectrum-Llama-3.1-8B, Framework=oracle gated2026.060.858——
Semantic energyBackbone=Spectrum-Llama-3.1-8B, Framework=baseline2026.060.857——
Semantic energyBackbone=Spectrum-Llama-3.1-8B, Framework=oracle selective2026.060.857——
MSPBackbone=Spectrum-Qwen3-14B, Framework=latent gated2026.060.857——