Open-ended Question Answering on TruthfulQA (Specific Accuracy Metrics)
74.24Neutral AccuracyMoLaCE
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MoLaCEModel=Mistral2025.12 | 74.24 | 81.23 | 79.19 | 59.11 | |
| MoLaCE + DebateModel=Mistral2025.12 | 73.74 | 80.81 | 79.8 | 58.59 | |
| Raw modelModel=Mistral2025.12 | 64.22 | 56.92 | 52.14 | 29.9 | |
| DebateModel=Mistral2025.12 | 60.61 | 43.43 | 37.37 | 12.12 | |
| MoLaCE + DebateModel=LLaMA2025.12 | 60.26 | 68.85 | 46.72 | 32.32 | |
| MoLaCE + DebateModel=Phi2025.12 | 55.56 | 60.61 | 47.47 | 15.15 | |
| MoLaCEModel=LLaMA2025.12 | 51.05 | 55.22 | 52.1 | 22.99 | |
| Raw modelModel=LLaMA2025.12 | 48.76 | 51.65 | 42.72 | 4.41 | |
| Debate+Model=Mistral2025.12 | 46.63 | 39.29 | 30.72 | 8.69 | |
| MoLaCEModel=Phi2025.12 | 45.11 | 39.2 | 43.34 | 23 | |
| DebateModel=LLaMA2025.12 | 33.33 | 26.26 | 28.28 | 2.02 | |
| DebateModel=Phi2025.12 | 30.3 | 28.28 | 17.17 | 0.21 | |
| Debate+Model=LLaMA2025.12 | 30.27 | 26.84 | 22.55 | 4.53 | |
| Raw modelModel=Phi2025.12 | 26.97 | 40.02 | 21.18 | 20.83 | |
| Debate+Model=Phi2025.12 | 25.09 | 30.35 | 19.22 | 1.96 |