Adversarial Robustness on AdvGLUE (test)
76.73AdvGLUE ScoreDPO + General Data
Evaluation Results
| Method | Links | |
|---|---|---|
| DPO + General DataBase Model=Qwen2.5-7B-Instruct, Alignment=DPO, Strategy=General Data2026.02 | 76.73 | |
| SFT + General DataBase Model=Qwen2.5-7B-Instruct, Alignment=SFT, Strategy=General Data2026.02 | 75.35 | |
| SFT-DPO + General DataBase Model=Qwen2.5-7B-Instruct, Alignment=SFT-DPO, Strategy=General Data2026.02 | 74.92 | |
| DPO + General DataBase Model=Llama3.1-8B-Instruct, Alignment=DPO, Strategy=General Data2026.02 | 50.3 | |
| SFT + MergeBase Model=Llama3.1-8B-Instruct, Alignment=SFT, Strategy=Merge2026.02 | 44.92 | |
| SFT-DPO + General DataBase Model=Llama3.1-8B-Instruct, Alignment=SFT-DPO, Strategy=General Data2026.02 | 44.71 |