Knowledge Unlearning on MQuAKE
64.94Efficiency ScoreBase
Evaluation results
| Method | Links | ||
|---|---|---|---|
| BaseModel=Llama-3.12026.05 | 64.94 | 7.48 | |
| ROMEModel=Llama-3.12026.05 | 62.59 | 7.46 | |
| ROMEModel=Llama3-22026.05 | 49.21 | 12.82 | |
| BaseModel=Llama3-22026.05 | 48.33 | 12.88 | |
| MEMITModel=Llama-3.12026.05 | 29.91 | 7.35 | |
| AlphaEditModel=Llama-3.12026.05 | 29.71 | 7.24 | |
| ZeroUnlearn-GDModel=Llama-3.12026.05 | 27.49 | 8.24 | |
| BaseModel=Qwen-32026.05 | 26.75 | 11.21 | |
| ROMEModel=Qwen-32026.05 | 26.7 | 11.16 | |
| MEMITModel=Llama3-22026.05 | 26.66 | 12.94 | |
| AlphaEditModel=Llama3-22026.05 | 24.71 | 13.04 | |
| ZeroUnlearn-GDModel=Llama3-22026.05 | 24.55 | 12.87 | |
| ZeroUnlearn-GDModel=Qwen-32026.05 | 23.37 | 11.57 | |
| AlphaEditModel=Qwen-32026.05 | 23.21 | 12.94 | |
| MEMITModel=Qwen-32026.05 | 23.14 | 11.63 | |
| FTModel=Qwen-32026.05 | 18.08 | 12.27 | |
| FTModel=Llama3-22026.05 | 15.56 | 14.36 | |
| FTModel=Llama-3.12026.05 | 14.44 | 9.02 | |
| GAModel=Llama3-22026.05 | 1.7 | 1,000 | |
| GAModel=Qwen-32026.05 | 0.34 | 1,000 | |
| GAModel=Llama-3.12026.05 | 0.03 | 1,000 |