Adversarial Robustness Evaluation on POLITEHARMBENCH 1.0 (full)
0ASRGEMMA-2B-IT-UL
Evaluation results
| Method | Links | |
|---|---|---|
| GEMMA-2B-IT-ULModel Variant=2B Instruction-Tuned with Unlearning2024.05 | 0 | |
| PHI-3-MINI-IPOModel Variant=Mini with IPO2024.05 | 0 | |
| ZEPHYR-7B-ULModel Variant=7B with Unlearning2024.05 | 0 | |
| GEMMA2024.05 | 5 | |
| GEMMA-2B-IT-IPOModel Variant=2B Instruction-Tuned with IPO2024.05 | 10 | |
| PHI-3-MINI-ULModel Variant=Mini with Unlearning2024.05 | 10 | |
| ZEPHYR + R2D2Defense=R2D22024.05 | 12.5 | |
| MISTRAL-7B-ULModel Variant=7B with Unlearning2024.05 | 15 | |
| PHI-3-MINI2024.05 | 25 | |
| MISTRAL-7B2024.05 | 42.5 | |
| ZEPHYR-7B2024.05 | 52.5 |