Question Answering on BBQ Race
81.9AccuracyCatRAG
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CatRAGBackbone=Gemma-32026.03 | 81.9 | 0.01 | |
| CatRAGBackbone=GPT-OSS2026.03 | 81.4 | 0.04 | |
| CausalBackbone=GPT-OSS2026.03 | 81.3 | 0.06 | |
| CatRAGBackbone=Llama-32026.03 | 80.7 | 0.01 | |
| CausalBackbone=Llama-32026.03 | 79.8 | 0.1 | |
| CausalBackbone=Gemma-32026.03 | 79 | 0.1 | |
| SPBackbone=GPT-OSS2026.03 | 71.8 | 0.17 | |
| SPBackbone=Gemma-32026.03 | 68.8 | 0.19 | |
| CEBackbone=GPT-OSS2026.03 | 68.2 | 0.24 | |
| SPBackbone=Llama-32026.03 | 67.4 | 0.21 | |
| CEBackbone=Gemma-32026.03 | 65.2 | 0.27 | |
| CEBackbone=Llama-32026.03 | 63.9 | 0.29 | |
| SelfBackbone=GPT-OSS2026.03 | 62.9 | 0.37 | |
| SelfBackbone=Gemma-32026.03 | 60 | 0.4 | |
| SelfBackbone=Llama-32026.03 | 58.7 | 0.42 | |
| BaseBackbone=GPT-OSS2026.03 | 55.4 | 0.56 | |
| BaseBackbone=Gemma-32026.03 | 52.3 | 0.6 | |
| BaseBackbone=Llama-32026.03 | 51.2 | 0.62 |