Factual Accuracy Estimation on GPT-5.4-mini Responses
62.47Original AccuracyLogistic Regression
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Logistic RegressionQuestion Type=Factual, Data=2920, Verification Budget=3%2026.06 | 62.47 | 57.2 | -5.26 | 88 | |
| Multilayer PerceptronQuestion Type=Factual, Data=2920, Verification Budget=3%2026.06 | 62.47 | 56.79 | -5.68 | 88 | |
| Random ForestQuestion Type=Factual, Data=2920, Verification Budget=3%2026.06 | 62.47 | 61.52 | -0.95 | 88 | |
| Label PropagationQuestion Type=Factual, Data=2920, Verification Budget=3%2026.06 | 62.47 | 53.23 | -9.24 | 88 | |
| nnPUQuestion Type=Factual, Data=2920, Verification Budget=3%2026.06 | 62.47 | 62.44 | -0.03 | 88 | |
| AURAQuestion Type=Factual, Data=2920, Verification Budget=3%2026.06 | 62.47 | 66.07 | 3.6 | 99 |