List your company

Generation Correctness Prediction on TruthfulQA (test)

62.69AURCDeg
40.735646.435352.13557.8347May 6, 2026

Evaluation results

MethodLinks
DegBackbone=Llama3.1-Instruct8B2026.0562.69
S.D.Backbone=Llama3.1-Instruct8B2026.0562.38
P(True)Backbone=Llama3.1-Instruct8B2026.0561.84
Self-ConBackbone=Llama3.1-Instruct8B2026.0561.37
S.E.Backbone=Llama3.1-Instruct8B2026.0560.76
INSIDEBackbone=Llama3.1-Instruct8B2026.0560.43
G-NLLBackbone=Llama3.1-Instruct8B2026.0559.08
P(True)Backbone=Mistral-Nemo-Instruct12B2026.0558.22
P(True)Backbone=Qwen3-Instruct4B2026.0557.24
ExGradBackbone=Llama3.1-Instruct8B2026.0557.15
ParaGradBackbone=Llama3.1-Instruct8B2026.0556.15
LN-PEBackbone=Llama3.1-Instruct8B2026.0555.78
SARBackbone=Llama3.1-Instruct8B2026.0554.35
M.I.Backbone=Llama3.1-Instruct8B2026.0554.03
HybridGradBackbone=Llama3.1-Instruct8B2026.0552.35
S.D.Backbone=Qwen3-Instruct4B2026.0552.08
SemGradBackbone=Llama3.1-Instruct8B2026.0551.99
INSIDEBackbone=Mistral-Nemo-Instruct12B2026.0551.97
S.D.Backbone=Mistral-Nemo-Instruct12B2026.0551.65
LN-PEBackbone=Mistral-Nemo-Instruct12B2026.0550.97
G-NLLBackbone=Mistral-Nemo-Instruct12B2026.0549.67
DegBackbone=Mistral-Nemo-Instruct12B2026.0549.52
S.E.Backbone=Mistral-Nemo-Instruct12B2026.0549.39
ExGradBackbone=Mistral-Nemo-Instruct12B2026.0548.96
Self-ConBackbone=Mistral-Nemo-Instruct12B2026.0548.56
DegBackbone=Qwen3-Instruct4B2026.0548.17
LN-PEBackbone=Qwen3-Instruct4B2026.0547.74
SARBackbone=Mistral-Nemo-Instruct12B2026.0547.49
M.I.Backbone=Qwen3-Instruct4B2026.0546.56
M.I.Backbone=Mistral-Nemo-Instruct12B2026.0546.2
ParaGradBackbone=Mistral-Nemo-Instruct12B2026.0546.16
S.E.Backbone=Qwen3-Instruct4B2026.0545.97
G-NLLBackbone=Qwen3-Instruct4B2026.0545.93
Self-ConBackbone=Qwen3-Instruct4B2026.0544.87
SemGradBackbone=Mistral-Nemo-Instruct12B2026.0544.86
INSIDEBackbone=Qwen3-Instruct4B2026.0544.61
HybridGradBackbone=Mistral-Nemo-Instruct12B2026.0544.54
ExGradBackbone=Qwen3-Instruct4B2026.0544.45
ParaGradBackbone=Qwen3-Instruct4B2026.0543.42
SARBackbone=Qwen3-Instruct4B2026.0542.7
SemGradBackbone=Qwen3-Instruct4B2026.0541.74
HybridGradBackbone=Qwen3-Instruct4B2026.0541.58