Cultural Safety and Knowledge Evaluation on AdaCultureSafe (Overall)
89.07AccuracyQwen2.5-7B
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Qwen2.5-7BModel Series=Qwen 2.5, Parameters=7B2026.03 | 89.07 | 58.79 | 68.55 | 0.03 | 0.02 | |
| Llama3.1-8BModel Series=Llama 3.1, Parameters=8B2026.03 | 86.58 | 56.06 | 65.28 | 0.04 | 0 | |
| Mistral-7BModel Series=Mistral, Parameters=7B2026.03 | 80.65 | 53.94 | 61.61 | 0.04 | 0.01 |