Question Answering on BBQ Gender
82.4AccuracyCausal
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CausalBackbone=GPT-OSS2026.03 | 82.4 | 0.04 | |
| CatRAGBackbone=GPT-OSS2026.03 | 82.2 | 0.04 | |
| CausalBackbone=GPT-OSS2026.03 | 82 | 0.03 | |
| CatRAGBackbone=Gemma-32026.03 | 81.6 | 0.03 | |
| CatRAGBackbone=Llama-32026.03 | 81.2 | 0.01 | |
| CatRAGBackbone=GPT-OSS2026.03 | 81.2 | 0.06 | |
| CatRAGBackbone=Gemma-32026.03 | 80.7 | 0.02 | |
| CatRAGBackbone=Llama-32026.03 | 79.5 | 0.02 | |
| CausalBackbone=Gemma-32026.03 | 79.3 | 0.1 | |
| CausalBackbone=Llama-32026.03 | 78.1 | 0.1 | |
| CausalBackbone=Llama-32026.03 | 78 | 0.1 | |
| CausalBackbone=Gemma-32026.03 | 76.4 | 0.09 | |
| SPBackbone=GPT-OSS2026.03 | 75.6 | 0.12 | |
| SPBackbone=Gemma-32026.03 | 72.6 | 0.14 | |
| CEBackbone=GPT-OSS2026.03 | 72 | 0.2 | |
| SPBackbone=Llama-32026.03 | 71.2 | 0.16 | |
| SPBackbone=GPT-OSS2026.03 | 70 | 0.19 | |
| CEBackbone=Gemma-32026.03 | 69.1 | 0.23 | |
| CEBackbone=Llama-32026.03 | 67.8 | 0.25 | |
| SPBackbone=Gemma-32026.03 | 67 | 0.21 | |
| SelfBackbone=GPT-OSS2026.03 | 66.9 | 0.32 | |
| SPBackbone=Llama-32026.03 | 65.6 | 0.23 | |
| CEBackbone=GPT-OSS2026.03 | 64.2 | 0.3 | |
| SelfBackbone=Gemma-32026.03 | 63.7 | 0.36 | |
| SelfBackbone=Llama-32026.03 | 62.4 | 0.38 | |
| SelfBackbone=GPT-OSS2026.03 | 62 | 0.39 | |
| CEBackbone=Gemma-32026.03 | 61.2 | 0.33 | |
| CEBackbone=Llama-32026.03 | 59.8 | 0.35 | |
| SelfBackbone=Gemma-32026.03 | 59.3 | 0.43 | |
| SelfBackbone=Llama-32026.03 | 57.9 | 0.45 | |
| BaseBackbone=GPT-OSS2026.03 | 56.8 | 0.52 | |
| BaseBackbone=Gemma-32026.03 | 53.6 | 0.57 | |
| BaseBackbone=Llama-32026.03 | 52.3 | 0.59 | |
| BaseBackbone=GPT-OSS2026.03 | 46.8 | 0.64 | |
| BaseBackbone=Gemma-32026.03 | 43.5 | 0.69 | |
| BaseBackbone=Llama-32026.03 | 42.1 | 0.71 |