Knowledge Insertion on WikiData recent
100Edit Success RateFT-M
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| FT-MBackbone=GPT2-xl2024.06 | 100 | 36.74 | 61.07 | 604.07 | 35.81 | |
| ICEBackbone=GPT2-xl2024.06 | 100 | 35.76 | 63.45 | 560.96 | 7.91 | |
| ROMEBackbone=GPT2-xl2024.06 | 99.24 | 30.74 | 25.37 | 603.82 | 316.38 | |
| EAMETModel=Deepseek-7B2025.05 | 97.15 | 56.43 | 60.45 | 501.09 | — | |
| EAMETModel=Falcon-7B2025.05 | 96.94 | 57.08 | 58.58 | 507.56 | — | |
| ALPHAEDITModel=Falcon-7B2025.05 | 96.22 | 55.47 | 58.02 | 493.56 | — | |
| MEMITModel=Falcon-7B2025.05 | 96.04 | 55.23 | 56.91 | 497.35 | — | |
| EAMETModel=LLaMA2-13B2025.05 | 95.88 | 52.08 | 53.43 | 504.06 | — | |
| EAMETModel=Qwen2.5-7B2025.05 | 95.61 | 57.46 | 60.28 | 509.06 | — | |
| MEMITModel=Qwen2.5-7B2025.05 | 94.88 | 56.97 | 61.23 | 510.43 | — | |
| MEMITModel=LLaMA2-13B2025.05 | 94.76 | 51.38 | 50.4 | 507.84 | — | |
| EAMETModel=LLaMA3-8B2025.05 | 94.36 | 57.88 | 59.48 | 528.23 | — | |
| ALPHAEDITModel=Qwen2.5-7B2025.05 | 94.16 | 57.17 | 59.45 | 510.32 | — | |
| ALPHAEDITModel=LLaMA2-13B2025.05 | 93.68 | 51.65 | 52.33 | 508.82 | — | |
| EAMETModel=LLaMA2-7B2025.05 | 93.23 | 53.13 | 54.61 | 503.52 | — | |
| MEMITModel=LLaMA3-8B2025.05 | 92.63 | 55.6 | 58.75 | 527.29 | — | |
| PMETModel=LLaMA2-13B2025.05 | 88.66 | 49.69 | 47.58 | 501.61 | — | |
| PMETModel=Deepseek-7B2025.05 | 86.75 | 57.85 | 59.93 | 500.5 | — | |
| PMETModel=Qwen2.5-7B2025.05 | 82.39 | 58.38 | 57.59 | 511.62 | — | |
| ROMEModel=LLaMA3-8B2025.05 | 82.19 | 54.94 | 57.74 | 518.89 | — | |
| PMETModel=LLaMA2-7B2025.05 | 81.84 | 51.11 | 53.16 | 497.49 | — | |
| MEMITBackbone=GPT2-xl2024.06 | 78.31 | 24.97 | 32.73 | 600.54 | 6.79 | |
| ROMEModel=LLaMA2-7B2025.05 | 76.73 | 49.31 | 51.51 | 497.53 | — | |
| PMETModel=LLaMA3-8B2025.05 | 75.81 | 56.89 | 58.26 | 513.84 | — | |
| ROMEModel=Falcon-7B2025.05 | 74.57 | 52.1 | 53.64 | 510.92 | — | |
| ALPHAEDITModel=LLaMA3-8B2025.05 | 65.36 | 51.44 | 53.55 | 516.53 | — | |
| FT-LBackbone=GPT2-xl2024.06 | 63.99 | 29.03 | 61.45 | 591.86 | 25.26 | |
| PMETModel=Falcon-7B2025.05 | 58.03 | 54.4 | 54.49 | 500.87 | — | |
| MEMITModel=LLaMA2-7B2025.05 | 34.75 | 44.93 | 46.72 | 504.18 | — | |
| MEMITModel=Deepseek-7B2025.05 | 33.65 | 52.28 | 49.05 | 499.49 | — | |
| FTModel=Falcon-7B2025.05 | 23.94 | 50.46 | 49.69 | 351.18 | — | |
| FTModel=Qwen2.5-7B2025.05 | 21.17 | 51.4 | 51.5 | 515.87 | — | |
| FTModel=Deepseek-7B2025.05 | 18.85 | 48.9 | 52.78 | 500.86 | — | |
| ALPHAEDITModel=Deepseek-7B2025.05 | 18.53 | 48.33 | 48.74 | 483.38 | — | |
| ROMEModel=Qwen2.5-7B2025.05 | 16.28 | 46.52 | 46.61 | 502.37 | — | |
| FTModel=LLaMA3-8B2025.05 | 16.05 | 47.51 | 48.84 | 490.71 | — | |
| FTModel=LLaMA2-13B2025.05 | 13.9 | 36.89 | 40.09 | 497.21 | — | |
| ROMEModel=LLaMA2-13B2025.05 | 11.06 | 38.14 | 39.09 | 447.42 | — | |
| FTModel=LLaMA2-7B2025.05 | 7.23 | 41.61 | 36.52 | 491.83 | — | |
| ROMEModel=Deepseek-7B2025.05 | 0.55 | 43.17 | 46.02 | 406.64 | — | |
| ALPHAEDITModel=LLaMA2-7B2025.05 | 0.07 | 35.34 | 37.48 | 527.83 | — | |
| MENDModel=LLaMA2-7B2025.05 | 0 | 34.67 | 37.46 | 269.52 | — | |
| MENDModel=Qwen2.5-7B2025.05 | 0 | 42.55 | 44.37 | 272.9 | — |