Machine Unlearning on TOFU 1.0 (Forget Set)
1ROUGE-LLLAMA2-13B-TOFU
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| LLAMA2-13B-TOFUModel Type=Base (No Unlearning)2024.10 | 1 | 1 | 46 | |
| MISTRAL-7B-TOFUModel Type=Base (No Unlearning)2024.10 | 1 | 1 | 49 | |
| LLAMA2-7B-TOFUModel Type=Base (No Unlearning)2024.10 | 0.99 | 0.99 | 51 | |
| δ-UNLEARNINGBase Model=LLAMA2-13B-TOFU2024.10 | 0.38 | 0.06 | 53 | |
| LoRABase Model=LLAMA2-13B-TOFU2024.10 | 0.03 | 0 | 50 | |
| LoRABase Model=LLAMA2-7B-TOFU2024.10 | 0.01 | 0 | 77 | |
| CMCBase Model=LLAMA2-7B-TOFU2024.10 | 0 | 0 | 33 | |
| CMCBase Model=LLAMA2-13B-TOFU2024.10 | 0 | 0 | 29 | |
| LoRABase Model=MISTRAL-7B-TOFU2024.10 | 0 | 0 | 72 | |
| CMCBase Model=MISTRAL-7B-TOFU2024.10 | 0 | 0 | 30 |