Question Answering on HotpotQA (All Questions)
80.2AccuracySPARQL
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SPARQLModel=70B, GW=false2026.03 | 80.2 | 70.6 | 57.8 | |
| SPARQL + GWModel=70B, GW=true2026.03 | 79.6 | 70.5 | 57.8 | |
| Base + GWModel=70B, GW=true2026.03 | 79.2 | 68.6 | 56 | |
| BaseModel=70B, GW=false2026.03 | 78 | 67.9 | 56.6 | |
| SPARQL + GWModel=8B, GW=true2026.03 | 76.6 | 62.2 | 50.6 | |
| SPARQLModel=8B, GW=false2026.03 | 70.6 | 60.8 | 49.4 | |
| BaseModel=8B, GW=false2026.03 | 67 | 59.5 | 49.4 | |
| Base + GWModel=8B, GW=true2026.03 | 63.6 | 55.9 | 46 |