Fairness Evaluation on CEB-Jigsaw
73.5ScoreSelf-Debias Iter1 + Self-Correction
Evaluation Results
| Method | Links | |
|---|---|---|
| Self-Debias Iter1 + Self-Correction2026.04 | 73.5 | |
| Self-Debias Iter12026.04 | 73.1 | |
| Qwen2.5-7B-Instruct2026.04 | 72.7 | |
| Self-Debias Offline + Self-Correction2026.04 | 72.5 | |
| Qwen1.5-8B2026.04 | 72.4 | |
| Self-Debias Iter22026.04 | 72.1 | |
| Self-Debias Iter2 + Self-Correction2026.04 | 71.9 | |
| Self-Debias Offline2026.04 | 71.7 | |
| Self-Debias SFT + Self-Correction2026.04 | 70.7 | |
| Self-Debias SFT2026.04 | 70.5 | |
| Qwen2.5-7B-Instruct + Self-Correction2026.04 | 68.3 | |
| Llama-3.1-8B-Instruct2026.04 | 67.3 | |
| DeepSeek-R1-Distill-Qwen-7B2026.04 | 65.8 | |
| DeepSeek-R1-Distill-Qwen-7B + Self-Correction2026.04 | 45.1 | |
| Llama-3.1-8B-Instruct + Self-Correction2026.04 | 29.1 | |
| Qwen1.5-8B + Self-Correction2026.04 | 20.6 |