LLM Parameter Editing on ZsRE
93.7Efficacy SuccessFE
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| FEModel=GPT2-XL2026.05 | 93.7 | 84.6 | 85.1 | 69.8 | 0.47 | 91.5 | 58.2 | 60 | |
| AlphaeditModel=GPT2-XL2026.05 | 92 | 81.8 | 83.2 | 67.4 | 0.93 | 73 | 45 | 48.9 | |
| BLUEModel=GPT2-XL2026.05 | 82.5 | 67.8 | 76.8 | 60.1 | 0.81 | 77.4 | 46.4 | 48.5 | |
| MEMITModel=GPT2-XL2026.05 | 81.4 | 65.8 | 74.4 | 56.7 | 0.9 | 70.6 | 47.1 | 49.5 | |
| PRUNEModel=GPT2-XL2026.05 | 81.4 | 65.7 | 74.3 | 56.5 | 0.9 | 70.5 | 47.2 | 49.6 | |
| RECTModel=GPT2-XL2026.05 | 80.7 | 64.6 | 73.5 | 55.4 | 0.89 | 70.7 | 47.5 | 49.8 | |
| OneLayerModel=GPT2-XL2026.05 | 66.5 | 45.8 | 60.7 | 39.6 | 0.94 | 69 | 44.4 | 47 | |
| WISEModel=GPT2-XL2026.05 | 9.1 | 0 | 9.2 | 0 | 2.48 | 8 | 4.8 | 10.5 |