Model Editing on RuleEdit-200
98.17Reliability (Rel.)DMLE
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DMLEBase Model=Qwen2-7B2026.04 | 98.17 | 63.33 | 52.08 | 64.58 | 87.63 | |
| MEMITBase Model=Qwen2-7B2026.04 | 97.83 | 51.67 | 46.25 | 39.92 | 19.63 | |
| MEMITBase Model=Qwen2.5-7B2026.04 | 97.16 | 51.67 | 48.25 | 36.5 | 20.25 | |
| DMLEBase Model=Qwen2.5-7B2026.04 | 96.5 | 58.17 | 51 | 63.51 | 85.5 | |
| ROMEBase Model=Qwen2-7B2026.04 | 95.33 | 35.17 | 50.08 | 38.92 | 19.63 | |
| PROMPTBase Model=LLaMA-3-8B2026.04 | 95.17 | 65.17 | — | 38.58 | 28 | |
| GRACEBase Model=Qwen2-7B2026.04 | 95.16 | 1.67 | — | 3.17 | 2.25 | |
| DMLEBase Model=GPT-J-6B2026.04 | 95 | 58.17 | 46.79 | 59.19 | 84.92 | |
| GRACEBase Model=GPT-J-6B2026.04 | 94.5 | 1.17 | — | 4.33 | 2 | |
| ROMEBase Model=Qwen2.5-7B2026.04 | 93.17 | 39.83 | 48.08 | 33.91 | 15.38 | |
| PROMPTBase Model=Qwen2-7B2026.04 | 93.17 | 63.17 | — | 50.58 | 29.13 | |
| PROMPTBase Model=Qwen2.5-7B2026.04 | 92.59 | 66.67 | — | 52.78 | 27.16 | |
| GRACEBase Model=Qwen2.5-7B2026.04 | 92.33 | 1.33 | — | 5.25 | 2.13 | |
| GRACEBase Model=LLaMA-3-8B2026.04 | 92 | 2.33 | — | 5.39 | 1.12 | |
| MEMITBase Model=GPT-J-6B2026.04 | 88.88 | 41.83 | 44.75 | 23.08 | 11.25 | |
| ROMEBase Model=GPT-J-6B2026.04 | 80.17 | 44.33 | 17.25 | 20.5 | 9.13 | |
| PROMPTBase Model=GPT-J-6B2026.04 | 67 | 51.16 | — | 30.92 | 18.88 | |
| ROMEBase Model=LLaMA-3-8B2026.04 | 63.67 | 38.33 | 27.17 | 31.42 | 15.63 | |
| DMLEBase Model=LLaMA-3-8B2026.04 | 54.16 | 40.67 | 34.38 | 41.21 | 45.88 | |
| MEMITBase Model=LLaMA-3-8B2026.04 | 53.5 | 37.83 | 23 | 26.58 | 12.38 |