Long-form Question Answering on ExpertQA
23.34ROUGE-LNWCAD
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| NWCADBackbone=Llama-3.1-70B2026.04 | 23.34 | 88.11 | |
| With-contextBackbone=Llama-3.1-70B2026.04 | 22.78 | 86.19 | |
| AdaCADBackbone=Llama-3.1-70B2026.04 | 22.57 | 87.69 | |
| NWCADBackbone=Llama-3.1-8B2026.04 | 22.23 | 90.49 | |
| BaselineBackbone=Llama-3.1-70B2026.04 | 21.76 | 85.98 | |
| CoCoABackbone=Llama-3.1-70B2026.04 | 21.63 | 87 | |
| NWCADBackbone=Ministral-32026.04 | 21.47 | 82.54 | |
| With-contextBackbone=Ministral-32026.04 | 21.3 | 82.46 | |
| AdaCADBackbone=Ministral-32026.04 | 21.03 | 82.28 | |
| CoCoABackbone=Ministral-32026.04 | 20.83 | 82.37 | |
| CADBackbone=Ministral-32026.04 | 20.55 | 82.43 | |
| With-contextBackbone=Llama-3.1-8B2026.04 | 20.26 | 88.07 | |
| CADBackbone=Llama-3.1-70B2026.04 | 19.23 | 85.69 | |
| AdaCADBackbone=Llama-3.1-8B2026.04 | 19.01 | 88.33 | |
| BaselineBackbone=Ministral-32026.04 | 18.86 | 81.45 | |
| CoCoABackbone=Llama-3.1-8B2026.04 | 18.67 | 89.02 | |
| BaselineBackbone=Llama-3.1-8B2026.04 | 18.65 | 87.7 | |
| CADBackbone=Llama-3.1-8B2026.04 | 17.44 | 88.43 |