Memorization Reduction on Ultrachat
95.7Memorization ReductionDPO
Evaluation Results
| Method | Links | |
|---|---|---|
| DPOCategory=Unlearning, Model=LLama, Utility threshold=>= 0.92026.02 | 95.7 | |
| GACategory=Unlearning, Model=LLama, Utility threshold=>= 0.92026.02 | 95.5 | |
| SimNPOCategory=Unlearning, Model=LLama, Utility threshold=>= 0.92026.02 | 95.5 | |
| NPOCategory=Unlearning, Model=LLama, Utility threshold=>= 0.92026.02 | 88.8 | |
| DPOCategory=Unlearning, Model=Qwen, Utility threshold=>= 0.92026.02 | 82.4 | |
| SimNPOCategory=Unlearning, Model=Qwen, Utility threshold=>= 0.92026.02 | 78.8 | |
| Task VectorCategory=Unlearning, Model=LLama, Utility threshold=>= 0.92026.02 | 77 | |
| NPOCategory=Unlearning, Model=Qwen, Utility threshold=>= 0.92026.02 | 71.5 | |
| Task VectorCategory=Unlearning, Model=Qwen, Utility threshold=>= 0.92026.02 | 66 | |
| GSSCategory=Ours, Model=Qwen, Utility threshold=>= 0.92026.02 | 51.4 | |
| GSSCategory=Ours, Model=LLama, Utility threshold=>= 0.92026.02 | 48 | |
| LUNARCategory=Editing, Model=Qwen, Utility threshold=>= 0.92026.02 | 42.7 | |
| ROMECategory=Editing, Model=LLama, Utility threshold=>= 0.92026.02 | 38 | |
| Balanced SubnetCategory=Editing, Model=LLama, Utility threshold=>= 0.92026.02 | 34.8 | |
| LUNARCategory=Editing, Model=LLama, Utility threshold=>= 0.92026.02 | 30 | |
| ROMECategory=Editing, Model=Qwen, Utility threshold=>= 0.92026.02 | 8.6 | |
| MEMITCategory=Editing, Model=LLama, Utility threshold=>= 0.92026.02 | 7.5 | |
| GACategory=Unlearning, Model=Qwen, Utility threshold=>= 0.92026.02 | 3.3 | |
| MEMITCategory=Editing, Model=Qwen, Utility threshold=>= 0.92026.02 | 2.2 | |
| Balanced SubnetCategory=Editing, Model=Qwen, Utility threshold=>= 0.92026.02 | 1 |