Domain-specific Question Answering on sst5
1.3Relative Accuracy ImprovementLWF
Evaluation results
| Method | Links | ||
|---|---|---|---|
| LWFModel=Llama3-8B, Unlearning Target=dental2025.05 | 1.3 | — | |
| LWFModel=Llama3-8B, Unlearning Target=mixed2025.05 | 1.06 | — | |
| LWFModel=Llama3-8B, Unlearning Target=gsm8k2025.05 | 0.97 | — | |
| LWFModel=Llama3-8B, Unlearning Target=qasc2025.05 | 0.74 | — | |
| LWFModel=Llama3-8B, Unlearning Target=psychol2025.05 | 0.24 | — | |
| Vanilla Fine-tuningModel=Llama3-8B, Unlearning Target=none2025.05 | — | 55.52 |