Gender bias evaluation on RealWorldQuestioning Education Recommendations 1.0 (Entire Dataset)
69.62Proportion Male More InformationLlama-3
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Llama-3Judge Model=ChatGPT-4o, Iterations=12025.05 | 69.62 | 30.37 | 0 | 0.19 | 0 | |
| DeepSeek-R1Judge Model=ChatGPT-4o, Iterations=12025.05 | 64.55 | 35.44 | 0 | 0.3 | 0.0004 | |
| ChatGPT-3.5-turboJudge Model=ChatGPT-4o, Iterations=12025.05 | 56.96 | 40.5 | 2.53 | 0.51 | 0.05 | |
| ChatGPT-4-turboJudge Model=ChatGPT-4o, Iterations=12025.05 | 44.3 | 51.89 | 3.79 | 1.36 | 0.43 |