Relative Robustness Analysis on Combined Past Tense, OR-Bench, MMLU
78.9R-ScoreOnly Scaling
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Only ScalingBase Model=Qwen2.5-7B-Instruct2025.09 | 78.9 | 46.4 | |
| SFT (5/95)Base Model=Qwen2.5-7B-Instruct2025.09 | 75.6 | 39.8 | |
| ASGUARDBase Model=Qwen2.5-7B-Instruct2025.09 | 74.6 | 58.8 | |
| ASGUARDBase Model=OLMo-2-1124-7B-Instruct2025.09 | 73.7 | 46.3 | |
| RepBendBase Model=Gemma-2-9B-it2025.09 | 72.8 | 41.9 | |
| SFT (5/95)Base Model=Llama-3.1-8B-Instruct2025.09 | 72.4 | 46.7 | |
| ASGUARDBase Model=Llama-3.1-8B-Instruct2025.09 | 71.8 | 52.9 | |
| Only ScalingBase Model=Llama-3.1-8B-Instruct2025.09 | 71.6 | 50.3 | |
| RepEBase Model=Gemma-2-9B-it2025.09 | 70.5 | 37.2 | |
| ASGUARDBase Model=Gemma-2-9B-it2025.09 | 70.1 | 44.6 | |
| DPOBase Model=Gemma-2-9B-it2025.09 | 69.8 | 35.4 | |
| DPOBase Model=Llama-3.1-8B-Instruct2025.09 | 69.5 | 36.7 | |
| SFT (30/70)Base Model=OLMo-2-1124-7B-Instruct2025.09 | 68.6 | 44.3 | |
| CBBase Model=Gemma-2-9B-it2025.09 | 68 | 35 | |
| Head AblationBase Model=Gemma-2-9B-it2025.09 | 67.9 | 36 | |
| SFT (5/95)Base Model=OLMo-2-1124-7B-Instruct2025.09 | 67.6 | 37.3 | |
| CBBase Model=Qwen2.5-7B-Instruct2025.09 | 67.1 | 35.5 | |
| Head AblationBase Model=Qwen2.5-7B-Instruct2025.09 | 66.9 | 38.5 | |
| DPOBase Model=OLMo-2-1124-7B-Instruct2025.09 | 66.9 | 35 | |
| SFT (30/70)Base Model=Qwen2.5-7B-Instruct2025.09 | 66.4 | 58.7 | |
| RepEBase Model=Qwen2.5-7B-Instruct2025.09 | 66.3 | 35.7 | |
| CBBase Model=OLMo-2-1124-7B-Instruct2025.09 | 66 | 37 | |
| RepBendBase Model=Llama-3.1-8B-Instruct2025.09 | 65.7 | 48.4 | |
| DPOBase Model=Qwen2.5-7B-Instruct2025.09 | 65.5 | 33.8 | |
| Head AblationBase Model=OLMo-2-1124-7B-Instruct2025.09 | 65.5 | 35.8 | |
| RepBendBase Model=OLMo-2-1124-7B-Instruct2025.09 | 64.9 | 34.9 | |
| RepEBase Model=Llama-3.1-8B-Instruct2025.09 | 64.5 | 32.8 | |
| Only ScalingBase Model=OLMo-2-1124-7B-Instruct2025.09 | 64.3 | 37.7 | |
| RepEBase Model=OLMo-2-1124-7B-Instruct2025.09 | 61.9 | 33.9 | |
| RepBendBase Model=Qwen2.5-7B-Instruct2025.09 | 60.2 | 40.6 | |
| SFT (5/95)Base Model=Gemma-2-9B-it2025.09 | 58.6 | 48.3 | |
| Head AblationBase Model=Llama-3.1-8B-Instruct2025.09 | 57.3 | 35.1 | |
| SFT (30/70)Base Model=Gemma-2-9B-it2025.09 | 56 | 47 | |
| SFT (30/70)Base Model=Llama-3.1-8B-Instruct2025.09 | 52.2 | 45.6 | |
| CBBase Model=Llama-3.1-8B-Instruct2025.09 | 30.6 | 36.3 | |
| Only ScalingBase Model=Gemma-2-9B-it2025.09 | 5.92 | 8.96 |