Question Answering on AstroQA Medium
80.28AccuracyMistral-7B + AstroRAG
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Mistral-7B + AstroRAGBackbone=ViT, Zero-shot=true2026.05 | 80.28 | 80.27 | |
| AstroRAGBackbone=Mistral-7B2026.05 | 80.28 | 80.27 | |
| AstroSage-Llama-3.1-8B + AstroRAGBackbone=MPT, Zero-shot=true2026.05 | 77.46 | 77.47 | |
| Map-Reduce2026.05 | 61.97 | 62.39 | |
| DrQA2026.05 | 58.87 | 58.87 | |
| Llama 2-Chat + AstroRAGBackbone=MPT, Zero-shot=true2026.05 | 49.3 | 49.72 | |
| AstroSage-Llama-3.1-8BBackbone=MPT, Zero-shot=true2026.05 | 43.94 | 43.92 | |
| Llama 2-ChatBackbone=MPT, Zero-shot=true2026.05 | 28.45 | 22.36 | |
| Mistral-7BBackbone=ViT, Zero-shot=true2026.05 | 21.97 | 9.57 | |
| BaselineBackbone=Mistral-7B2026.05 | 21.97 | 9.57 |