Knowledge Editing on Counterfact Full (test)
99Rel. AccuracyMEMIT
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MEMITBase Model=GPT-J-6B, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 99 | 20 | 90 | 70 | 58 | 3 | |
| ONCEEDITBase Model=Llama2-7B, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing, Plug-in model=Tiny-Llama2025.06 | 99 | 81 | 62 | 81 | 72 | 1 | |
| DEFERBase Model=Llama2-7B, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 98 | 88 | 34 | 74 | 66 | 3 | |
| MEMITBase Model=Llama2-7B, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 95 | 51 | 23 | 56 | 58 | 3 | |
| DEFERBase Model=GPT-J-6B, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 94 | 84 | 15 | 64 | 66 | 3 | |
| ONCEEDITBase Model=Mistral-7B-v0.1, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing, Plug-in model=Tiny-Llama2025.06 | 94 | 76 | 62 | 77 | 72 | 1 | |
| ONCEEDITBase Model=GPT-J-6B, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing, Plug-in model=Tiny-Llama2025.06 | 94 | 76 | 53 | 74 | 72 | 1 | |
| MEMITBase Model=Mistral-7B-v0.1, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 78 | 43 | 26 | 49 | 58 | 3 | |
| WISEBase Model=Llama2-7B, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 74 | 33 | 38 | 48 | 39 | 3 | |
| FT-LBase Model=GPT-J-6B, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 71 | 9 | 7 | 30 | 31 | 3 | |
| WISEBase Model=Mistral-7B-v0.1, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 67 | 24 | 35 | 42 | 39 | 3 | |
| DEFERBase Model=Mistral-7B-v0.1, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 47 | 53 | 79 | 60 | 66 | 3 | |
| FT-LBase Model=Mistral-7B-v0.1, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 41 | 5 | 99 | 48 | 31 | 3 | |
| WISEBase Model=GPT-J-6B, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 37 | 8 | 37 | 27 | 39 | 3 | |
| FT-LBase Model=Llama2-7B, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 26 | 1 | 18 | 15 | 31 | 3 | |
| ROMEBase Model=Llama2-7B, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 7 | 4 | 5 | 5 | 2 | 3 | |
| ROMEBase Model=Mistral-7B-v0.1, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 1 | 2 | 0 | 1 | 2 | 3 | |
| MENDBase Model=Llama2-7B, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 0 | 0 | 0 | 0 | 0 | 3 | |
| MENDBase Model=Mistral-7B-v0.1, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 0 | 0 | 0 | 0 | 0 | 3 | |
| MENDBase Model=GPT-J-6B, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 0 | 0 | 0 | 0 | 0 | 3 | |
| ROMEBase Model=GPT-J-6B, Evaluation Setting=Teacher-forced, Editing Setting=Batch editing2025.06 | 0 | 0 | 0 | 0 | 2 | 3 |