Question Answering on Hotpotd (accuracy)
79.9AccuracyWith-context
Evaluation Results
| Method | Links | |
|---|---|---|
| With-contextModel=Ministral-3-8B2026.04 | 79.9 | |
| NWCADModel=Ministral-3-8B2026.04 | 79.9 | |
| AdaCADModel=Ministral-3-8B2026.04 | 78.9 | |
| AdaCADModel=Llama-3.1-8B2026.04 | 76.7 | |
| CADModel=Ministral-3-8B2026.04 | 76 | |
| NWCADModel=Llama-3.1-8B2026.04 | 75.8 | |
| NWCADModel=Llama-3.1-70B2026.04 | 74.38 | |
| AdaCADModel=Llama-3.1-70B2026.04 | 73.5 | |
| CoCoAModel=Llama-3.1-8B2026.04 | 73.4 | |
| With-contextModel=Llama-3.1-70B2026.04 | 72.75 | |
| With-contextModel=Llama-3.1-8B2026.04 | 72.1 | |
| CoCoAModel=Ministral-3-8B2026.04 | 70.4 | |
| CoCoAModel=Llama-3.1-70B2026.04 | 69.5 | |
| CADModel=Llama-3.1-8B2026.04 | 69.1 | |
| CADModel=Llama-3.1-70B2026.04 | 51.75 | |
| BaselineModel=Llama-3.1-8B2026.04 | 29.2 | |
| BaselineModel=Llama-3.1-70B2026.04 | 26.62 | |
| BaselineModel=Ministral-3-8B2026.04 | 26.5 |