Knowledge Editing on Bi-ZsRE Edit in English (test)
99.97Reliability (Eng)FT-M
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| FT-MBackbone=Llama2-chat-7B2025.02 | 99.97 | 95.38 | 97.92 | 57.69 | 56.89 | 56.52 | 94.61 | 52.16 | 76.39 | |
| X-KDEBackbone=Llama2-chat-7B2025.02 | 99.93 | 99.87 | 90.15 | 76.41 | 94.81 | 94.65 | 95.05 | 77.43 | 91.04 | |
| LTEBackbone=Llama2-chat-7B2025.02 | 99.91 | 99.81 | 88.97 | 77.4 | 76.86 | 76.82 | 86.99 | 67.49 | 84.28 | |
| IKEBackbone=Llama2-chat-7B2025.02 | 99.59 | 99.61 | 56.95 | 71.27 | 67.83 | 67.88 | 64.54 | 58.97 | 73.33 | |
| ROMEBackbone=Llama2-chat-7B2025.02 | 96.09 | 84.69 | 98.04 | 58.87 | 49.94 | 50.31 | 97.7 | 51.81 | 73.43 | |
| MEMITBackbone=Llama2-chat-7B2025.02 | 95.21 | 89.14 | 98.56 | 57.77 | 52.05 | 52.01 | 98.76 | 52.19 | 74.46 | |
| FT-LBackbone=Llama2-chat-7B2025.02 | 53.51 | 50.18 | 94.01 | 53.31 | 51.81 | 51.71 | 85.56 | 55.14 | 61.9 |