List your company

Citation Hallucination Detection on NeuCLIR 2024 (test)

73.7AUCFACTUM
47.07653.98860.967.812Jan 9, 2026

Evaluation results

MethodLinks
FACTUMBase Model=Llama-3.1-8B-Instruct, Classifier=Log. Reg.2026.0173.70.3333.468.844.6
FACTUMBase Model=Llama-3.1-8B-Instruct, Classifier=EBM2026.0173.60.32632.266.943.2
FACTUMBase Model=Llama-3.2-3B-Instruct, Classifier=Log. Reg.2026.0171.50.35448.469.356.5
FACTUMBase Model=Llama-3.2-3B-Instruct, Classifier=EBM2026.01700.34147.359.151.5
FACTUMBase Model=Llama-3.1-8B-Instruct, Classifier=LightGBM2026.0169.30.2693065.240.8
CAS+PFSBase Model=Llama-3.1-8B-Instruct, Classifier=EBM2026.0168.70.25527.86638.8
FACTUMBase Model=Llama-3.2-3B-Instruct, Classifier=LightGBM2026.0168.20.30243.662.450.7
CAS+PFSBase Model=Llama-3.1-8B-Instruct, Classifier=Log. Reg.2026.0167.70.24327.563.738.1
CAS+PFSBase Model=Llama-3.2-3B-Instruct, Classifier=Log. Reg.2026.0167.60.28243.166.751.9
LN-EntropyBase Model=Llama-3.1-8B-Instruct, Classifier=Classifier-free2026.0167.10.21923.762.533.3
PerplexityBase Model=Llama-3.1-8B-Instruct, Classifier=Classifier-free2026.01670.13723.762.933.4
CAS+PFSBase Model=Llama-3.2-3B-Instruct, Classifier=EBM2026.0166.80.27242.958.748.8
ECS+PKSBase Model=Llama-3.2-3B-Instruct, Classifier=Log. Reg.2026.0166.80.27240.663.649.1
CAS+PFSBase Model=Llama-3.2-3B-Instruct, Classifier=LightGBM2026.0164.50.23742.660.449.1
PerplexityBase Model=Llama-3.2-3B-Instruct, Classifier=Classifier-free2026.0163.80.08537.373.948.9
LN-EntropyBase Model=Llama-3.2-3B-Instruct, Classifier=Classifier-free2026.0163.70.18937.671.748.7
EnergyBase Model=Llama-3.1-8B-Instruct, Classifier=Classifier-free2026.0161.70.1421.941.926.9
ECS+PKSBase Model=Llama-3.1-8B-Instruct, Classifier=EBM2026.0161.30.15124.257.133.7
ECS+PKSBase Model=Llama-3.2-3B-Instruct, Classifier=EBM2026.01610.1839.158.346.3
EnergyBase Model=Llama-3.2-3B-Instruct, Classifier=Classifier-free2026.0160.60.15137.35643.7
CAS+PFSBase Model=Llama-3.1-8B-Instruct, Classifier=LightGBM2026.0159.20.12423.658.333.3
ECS+PKSBase Model=Llama-3.2-3B-Instruct, Classifier=LightGBM2026.0158.80.14938.855.345
ECS+PKSBase Model=Llama-3.1-8B-Instruct, Classifier=LightGBM2026.0154.90.0721.253.430.2
ECS+PKSBase Model=Llama-3.1-8B-Instruct, Classifier=Log. Reg.2026.0153.60.04920.152.929
P(True)Base Model=Llama-3.2-3B-Instruct, Classifier=Classifier-free2026.0148.1-0.03527.858.836.9
P(True)Base Model=Llama-3.1-8B-Instruct, Classifier=Classifier-free2026.0148.1-0.01915.591.526.1