Hallucination Detection on LogicStruct
73.89Neg. ScorePRISM-SAPLMA
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| PRISM-SAPLMABackbone=LLaMA2-13B-Chat2024.11 | 73.89 | 81.12 | 57.22 | 70.74 | |
| PRISM-SAPLMABackbone=LLaMA2-7B-Chat2024.11 | 73.45 | 75.29 | 53.29 | 67.34 | |
| PRISM-MMBackbone=LLaMA2-7B-Chat2024.11 | 55.48 | 80.87 | 55.8 | 64.05 | |
| PRISM-MMBackbone=LLaMA2-13B-Chat2024.11 | 52.6 | 87.34 | 62.07 | 67.34 | |
| MINDBackbone=LLaMA2-7B-Chat2024.11 | 52.19 | 43.18 | 49.69 | 48.35 | |
| SAPLMABackbone=LLaMA2-13B-Chat2024.11 | 51.55 | 67.88 | 51.86 | 57.1 | |
| MMBackbone=LLaMA2-13B-Chat2024.11 | 50.71 | 48.95 | 50.2 | 49.95 | |
| MMBackbone=LLaMA2-7B-Chat2024.11 | 49.48 | 49.37 | 50.13 | 49.66 | |
| SAPLMABackbone=LLaMA2-7B-Chat2024.11 | 48.64 | 51.96 | 50.3 | 50.3 | |
| MINDBackbone=LLaMA2-13B-Chat2024.11 | 48.49 | 48.73 | 50.38 | 49.2 | |
| EUBIDBackbone=LLaMA2-13B-Chat2024.11 | 45.12 | 52.9 | 52.63 | 50.22 | |
| LN-PPBackbone=LLaMA2-13B-Chat2024.11 | 42.63 | 52.43 | 58.63 | 51.23 | |
| EUBIDBackbone=LLaMA2-7B-Chat2024.11 | 41.84 | 53.1 | 53.87 | 49.6 | |
| LN-PPBackbone=LLaMA2-7B-Chat2024.11 | 41.08 | 52.78 | 57.43 | 50.43 |