List your company

Uncertainty Quantification on CNN/DailyMail

0.745Hamming AUCStructure-Aware Minimum Bayes Risk Decoding
0.419480.503990.58850.67301Jan 29, 2026

Evaluation results

MethodLinks
Structure-Aware Minimum Bayes Risk DecodingBackbone Model=Gemma-3-4B2026.010.7450.6170.607
Structure-Aware Minimum Bayes Risk DecodingBackbone Model=Qwen3-30B-A3B2026.010.7320.5860.603
Structure-Aware Minimum Bayes Risk DecodingBackbone Model=Gemma-3-12B2026.010.7090.5850.66
Structure-Aware Minimum Bayes Risk DecodingBackbone Model=Qwen3-4B2026.010.6720.5560.592
COCOABackbone Model=Qwen3-30B-A3B2026.010.6280.6450.512
SARBackbone Model=Qwen3-30B-A3B2026.010.6250.6570.51
MSPBackbone Model=Qwen3-30B-A3B2026.010.6050.5920.517
SARBackbone Model=Qwen3-4B2026.010.6040.5630.526
SARBackbone Model=Gemma-3-4B2026.010.5940.6190.574
KLEBackbone Model=Qwen3-30B-A3B2026.010.5890.570.502
COCOABackbone Model=Qwen3-4B2026.010.5820.5760.525
COCOABackbone Model=Gemma-3-4B2026.010.5770.5690.589
SEBackbone Model=Qwen3-4B2026.010.5770.5480.504
SEBackbone Model=Qwen3-30B-A3B2026.010.5760.5730.475
SEBackbone Model=Gemma-3-12B2026.010.5720.5340.501
SARBackbone Model=Gemma-3-12B2026.010.570.5760.542
KLEBackbone Model=Qwen3-4B2026.010.570.5550.492
SEBackbone Model=Gemma-3-4B2026.010.5630.5560.527
COCOABackbone Model=Gemma-3-12B2026.010.5630.5610.526
MSPBackbone Model=Qwen3-4B2026.010.5460.550.52
KLEBackbone Model=Gemma-3-4B2026.010.5390.510.51
KLEBackbone Model=Gemma-3-12B2026.010.5340.5180.479
MSPBackbone Model=Gemma-3-4B2026.010.5260.5190.574
MSPBackbone Model=Gemma-3-12B2026.010.5230.5430.515
P(True)Backbone Model=Qwen3-4B2026.010.5170.4830.539
P(True)Backbone Model=Qwen3-30B-A3B2026.010.5150.5140.499
P(True)Backbone Model=Gemma-3-4B2026.010.4790.4840.504
P(True)Backbone Model=Gemma-3-12B2026.010.4320.4580.474