Safety Evaluation on SafetyBench (test)
81.321AccuracyAWQ-trust
Evaluation Results
| Method | Links | |
|---|---|---|
| AWQ-trustBackbone=Qwen-2.5-7B-Instruct, Quantization=AWQ-trust2026.01 | 81.321 | |
| Full PrecisionBackbone=Qwen-2.5-7B-Instruct, Quantization=Full Precision2026.01 | 80.394 | |
| AWQBackbone=Qwen-2.5-7B-Instruct, Quantization=AWQ2026.01 | 80.017 | |
| Full PrecisionBackbone=Llama-3.1-8B-Instruct, Quantization=Full Precision2026.01 | 76.502 | |
| AWQBackbone=Llama-3.1-8B-Instruct, Quantization=AWQ2026.01 | 75.864 | |
| AWQ-trustBackbone=Llama-3.1-8B-Instruct, Quantization=AWQ-trust2026.01 | 75.68 | |
| AWQ-trustBackbone=Gemma-7B-Instruct, Quantization=AWQ-trust2026.01 | 66.734 | |
| AWQBackbone=Gemma-7B-Instruct, Quantization=AWQ2026.01 | 66.672 | |
| Full PrecisionBackbone=Gemma-7B-Instruct, Quantization=Full Precision2026.01 | 66.62 |