Hallucination Detection on PopQA (AUPRC)
67.08AUPRCLyapunov Probes
Evaluation Results
| Method | Links | |
|---|---|---|
| Lyapunov ProbesModel=Llama-3-8B2026.03 | 67.08 | |
| ProbeModel=Qwen-3-4B2026.03 | 64.41 | |
| Lyapunov ProbesModel=Qwen-3-4B2026.03 | 64.02 | |
| Lyapunov ProbesModel=Llama-2-7B2026.03 | 63.37 | |
| ProbeModel=Llama-2-7B2026.03 | 61.87 | |
| Lyapunov ProbesModel=Falcon-7B2026.03 | 61.23 | |
| ProbeModel=Llama-3-8B2026.03 | 60.77 | |
| ProbeModel=Falcon-7B2026.03 | 60.48 | |
| Seq. Prob.Model=Llama-3-8B2026.03 | 27.02 | |
| VerbalizedModel=Llama-3-8B2026.03 | 21.23 | |
| Seq. Prob.Model=Llama-2-7B2026.03 | 20.22 | |
| VerbalizedModel=Llama-2-7B2026.03 | 20.13 | |
| SurrogateModel=Llama-2-7B2026.03 | 18.42 | |
| SurrogateModel=Llama-3-8B2026.03 | 18.26 | |
| SurrogateModel=Falcon-7B2026.03 | 16.53 | |
| Seq. Prob.Model=Falcon-7B2026.03 | 15.13 | |
| Seq. Prob.Model=Qwen-3-4B2026.03 | 13.79 | |
| VerbalizedModel=Falcon-7B2026.03 | 13.76 | |
| VerbalizedModel=Qwen-3-4B2026.03 | 12.04 | |
| SurrogateModel=Qwen-3-4B2026.03 | 10.08 |