Hallucination Diagnosis on Hallucination Diagnosis Dataset Wikipedia-based
79.55ASQwen3-32B
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Qwen3-32BEvaluation Protocol=Pipeline Method2025.12 | 79.55 | 94.74 | 76.97 | 69.69 | |
| o4-miniEvaluation Protocol=Pipeline Method2025.12 | 78.51 | 82.46 | 78.59 | 79.71 | |
| GPT 4.1Evaluation Protocol=Pipeline Method2025.12 | 76.77 | 61.75 | 60.52 | 53.08 | |
| Qwen3-32BEvaluation Protocol=Single Prompt Method2025.12 | 70.98 | 97.54 | 64.85 | 59.15 | |
| HDM-4B-RLEvaluation Protocol=Single Prompt Method2025.12 | 69.16 | 92.28 | 58.65 | 48.49 | |
| GPT 4.1Evaluation Protocol=Single Prompt Method2025.12 | 65.97 | 77.54 | 59.12 | 48.1 | |
| o4-miniEvaluation Protocol=Single Prompt Method2025.12 | 61.18 | 89.12 | 41.3 | 43.04 | |
| Original ResultDescription=Baseline for mitigation task2025.12 | 59.77 | — | — | — |