Gender bias evaluation on RealWorldQuestioning Health Recommendations 1.0
80.89Proportion (Male More Info)Llama-3
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Llama-3Judge Model=ChatGPT-4o, Iterations=12025.05 | 80.89 | 17.97 | 1.12 | 0.05 | 0 | |
| DeepSeek-R1Judge Model=ChatGPT-4o, Iterations=12025.05 | 73.03 | 26.96 | 0 | 0.13 | 0 | |
| ChatGPT-4-turboJudge Model=ChatGPT-4o, Iterations=12025.05 | 56.17 | 38.2 | 5.61 | 0.48 | 0.02 | |
| ChatGPT-3.5-turboJudge Model=ChatGPT-4o, Iterations=12025.05 | 53.93 | 39.32 | 6.74 | 0.55 | 0.07 |