Instruction Following and Safety Evaluation on Alpaca
52.2BRT ScoreSFT
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SFTModel=Vicuna-7B2025.05 | 52.2 | 4.95 | 40.5 | |
| SAPModel=Llama2-7B2025.05 | 52.1 | 6.03 | 22.6 | |
| SAPModel=Vicuna-7B2025.05 | 51.9 | 4.87 | 24.9 | |
| SAPModel=Average2025.05 | 51.9 | 5.54 | 23.07 | |
| SafeInstrModel=Llama2-7B2025.05 | 51.8 | 6.06 | 28.9 | |
| SAPModel=Qwen2.5-7B2025.05 | 51.6 | 5.72 | 21.7 | |
| SFTModel=Average2025.05 | 51.6 | 5.55 | 37.43 | |
| SFTModel=Llama2-7B2025.05 | 51.4 | 6.06 | 33.1 | |
| SFTModel=Qwen2.5-7B2025.05 | 51.2 | 5.65 | 38.7 | |
| SafeInstrModel=Average2025.05 | 51.1 | 5.58 | 26.2 | |
| SafeInstrModel=Vicuna-7B2025.05 | 51 | 4.96 | 27.2 | |
| SaLoRAModel=Llama2-7B2025.05 | 50.8 | 6.15 | 29.2 | |
| BoosterModel=Vicuna-7B2025.05 | 50.8 | 4.99 | 28.4 | |
| LisaModel=Vicuna-7B2025.05 | 50.6 | 5.27 | 28.1 | |
| SafeInstrModel=Qwen2.5-7B2025.05 | 50.4 | 5.71 | 22.5 | |
| SAFTModel=Vicuna-7B2025.05 | 50.3 | 5.07 | 34.6 | |
| SaLoRAModel=Average2025.05 | 50.3 | 5.71 | 30.57 | |
| BoosterModel=Qwen2.5-7B2025.05 | 50.2 | 5.77 | 26.5 | |
| SaLoRAModel=Qwen2.5-7B2025.05 | 50.2 | 5.88 | 30.8 | |
| BoosterModel=Average2025.05 | 50.1 | 5.62 | 27.6 | |
| LisaModel=Average2025.05 | 50.1 | 5.76 | 25.93 | |
| LisaModel=Llama2-7B2025.05 | 49.9 | 6.17 | 25.4 | |
| SaLoRAModel=Vicuna-7B2025.05 | 49.9 | 5.11 | 31.7 | |
| LisaModel=Qwen2.5-7B2025.05 | 49.8 | 5.82 | 24.3 | |
| SAFTModel=Qwen2.5-7B2025.05 | 49.6 | 5.79 | 35.2 | |
| SAFTModel=Average2025.05 | 49.5 | 5.67 | 33.63 | |
| BoosterModel=Llama2-7B2025.05 | 49.4 | 6.09 | 27.9 | |
| VaccineModel=Vicuna-7B2025.05 | 49.2 | 5.03 | 30.4 | |
| VaccineModel=Qwen2.5-7B2025.05 | 48.9 | 5.81 | 28.8 | |
| VaccineModel=Average2025.05 | 48.9 | 5.66 | 29.23 | |
| VaccineModel=Llama2-7B2025.05 | 48.7 | 6.13 | 28.5 | |
| SAFTModel=Llama2-7B2025.05 | 48.7 | 6.14 | 31.1 | |
| Base modelModel=Vicuna-7B2025.05 | 46.5 | 16.38 | 32.3 | |
| Base modelModel=Qwen2.5-7B2025.05 | 45.7 | 15.34 | 34.3 | |
| Base modelModel=Average2025.05 | 45.6 | 17 | 32.5 | |
| Base modelModel=Llama2-7B2025.05 | 44.7 | 19.28 | 30.9 |