Stereotypical Bias Evaluation on StereoSet (SS)
41StereoSet Score (SS)Vanilla
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VanillaBackbone=LLaMA-2-13B2026.05 | 41 | 5.8 | |
| VanillaBackbone=Mistral-7B-Instruct2026.05 | 38 | 5.8 | |
| SDDBackbone=LLaMA-2-13B2026.05 | 36 | 4 | |
| SDDBackbone=Mistral-7B-Instruct2026.05 | 34 | 3.9 | |
| DExpertsBackbone=LLaMA-2-13B2026.05 | 33 | 3.3 | |
| DT-CD⋆Backbone=LLaMA-2-13B2026.05 | 31 | 2.8 | |
| DExpertsBackbone=Mistral-7B-Instruct2026.05 | 31 | 3.1 | |
| DT-CD⋆Backbone=Mistral-7B-Instruct2026.05 | 29 | 2.6 | |
| COFTBackbone=LLaMA-2-13B2026.05 | 26 | 1 | |
| COFTBackbone=Mistral-7B-Instruct2026.05 | 24 | 1 |