Model Editing on COUNTERFACT (Efficacy, Generality, Specificity, Fluency, Consistency)
99.75EfficacyAlphaEdit
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| AlphaEditModel=GPT-J-6B2025.10 | 99.75 | 96.38 | 75.48 | 618.5 | 42.08 | |
| EvoEditModel=GPT-J-6B2025.10 | 99.75 | 95.94 | 75.21 | 619.12 | 41.42 | |
| AlphaEditModel=GPT2-XL2025.10 | 99.5 | 93.95 | 66.39 | 597.88 | 39.38 | |
| EvoEditModel=GPT2-XL2025.10 | 99.32 | 92.96 | 66.53 | 592.04 | 38.04 | |
| MEMITModel=GPT-J-6B2025.10 | 98.55 | 95.5 | 63.64 | 546.28 | 34.89 | |
| MEMITModel=GPT2-XL2025.10 | 94.7 | 85.82 | 60.5 | 477.26 | 22.72 | |
| ROMEModel=GPT-J-6B2025.10 | 57.5 | 54.2 | 52.05 | 589.42 | 3.22 | |
| ROMEModel=GPT2-XL2025.10 | 54.6 | 51.18 | 52.68 | 366.13 | 0.72 |