Debiasing Effectiveness on In-Distribution (ID)
10.2Mean Effectiveness Score (ID)ORIGINAL
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ORIGINALModel Family=QWEN2.5-7B2026.03 | 10.2 | 27.43 | |
| ORIGINALModel Family=QWEN2.5-14B2026.03 | 8.71 | 33.13 | |
| ORIGINALModel Family=QWEN2.5-3B2026.03 | 7.41 | 20.08 | |
| ORIGINALModel Family=GEMMA-2-2B2026.03 | 6.84 | 37.45 | |
| UGIDModel Family=QWEN2.5-7B2026.03 | 1.15 | 2.24 | |
| CDAModel Family=QWEN2.5-3B2026.03 | 1.14 | 1.29 | |
| CDAModel Family=QWEN2.5-14B2026.03 | 1.14 | 1.29 | |
| CDAModel Family=GEMMA-2-2B2026.03 | 1.13 | 1.21 | |
| CDAModel Family=QWEN2.5-7B2026.03 | 1.12 | 1.14 | |
| KLAADModel Family=QWEN2.5-7B2026.03 | 1.08 | 1.13 | |
| KLAADModel Family=GEMMA-2-2B2026.03 | 1.01 | 1.06 | |
| UGIDModel Family=QWEN2.5-3B2026.03 | 0.98 | 1 | |
| KLAADModel Family=QWEN2.5-14B2026.03 | 0.98 | 1.13 | |
| UGIDModel Family=GEMMA-2-2B2026.03 | 0.95 | 1 | |
| UGIDModel Family=QWEN2.5-14B2026.03 | 0.94 | 1 | |
| KLAADModel Family=QWEN2.5-3B2026.03 | 0.88 | 0.88 |