Language Model Debiasing on SEAT
0.19Gender Bias ScorePCGU
Evaluation results
| Method | Links | ||
|---|---|---|---|
| PCGUBackbone=BERT2026.01 | 0.19 | 0.79 | |
| RobustDebiasBackbone=BERT, Multi-demographic Training=Yes (single model across demographics)2026.01 | 0.25 | 0.4 | |
| AutoDebiasBackbone=BERT, Multi-demographic Training=No (specific models per demographic)2026.01 | 0.31 | 0.77 | |
| CausalDebiasBackbone=BERT, Multi-demographic Training=No (specific models per demographic)2026.01 | 0.38 | 0.64 | |
| FineDebBackbone=BERT, Multi-demographic Training=No (specific models per demographic)2026.01 | 0.39 | 0.51 |