List your company

Hallucination Detection on LLM-AggreFact

91.99CNN PerformanceCPIL-1%
53.71863.65473.5983.526Jun 6, 2026

Evaluation results

MethodLinks
CPIL-1%labeled data fraction=1%, backbone=DeBERTa2026.0691.9973.4783.0286.1369.277.1385.161.4670.5787.8584.7479.14
CPIL-2%labeled data fraction=2%, backbone=DeBERTa2026.0691.0871.3884.0590.164.9177.7781.7458.3671.9488.984.7978.64
CCHD-FT5Backbone=Flan-T52026.0688.5873.3778.9288.3468.3176.6286.562.4672.4188.2886.4279.11
CCHD-DBTBackbone=DeBERTa2026.0688.0577.184.7487.4370.778.2983.8161.4768.5389.4687.479.73
AlignScore2026.0681.7168.5180.5985.1854.7569.7981.1661.0570.6385.8177.3574.23
AlignScore2026.0681.7168.5180.5985.1854.7569.7981.1661.0570.6385.8177.3574.23
FactCG-FT5Backbone=Flan-T52026.0681.4274.5282.3688.7468.1875.0587.0559.5770.7388.786.8278.47
FactCG2026.0678.4474.4683.9880.0769.8780.5383.4563.4258.5489.1286.6377.14
FactCG-DBTBackbone=DeBERTa2026.0678.4474.4683.9880.0769.8780.5383.4563.4258.5489.1286.6377.14
FactCC2026.0678.369.282.4686.7469.977.7786.4161.2865.1587.4786.177.34
MiniCheck2026.0674.8874.4983.7281.2662.172.3185.5460.1572.2288.1586.3676.47
MiniCheck-DBTBackbone=DeBERTa2026.0674.8874.4983.7281.2662.172.3185.5460.1572.2288.1586.3676.47
MiniCheck-FT5Backbone=Flan-T52026.0672.6471.2276.289.4965.9774.1681.8863.3971.9989.0684.2876.39
SummaC-CVmode=Cross-Validation2026.0666.4148.5669.777.8750.9870.7768.3657.5365.2179.1656.9264.68
SummaC-CVEvaluation Protocol=cross-validation2026.0666.4148.5669.777.8750.9870.7768.3657.5365.2179.1656.9264.68
SummaC-ZSmode=Zero-Shot2026.0655.1966.5579.5780.1441.0374.572.9660.7965.381.5862.8167.31
SummaC-ZSEvaluation Protocol=zero-shot2026.0655.1966.5579.5780.1441.0374.572.9660.7965.381.5862.8167.31