Question Answering on MuSiQue All Questions (Acc, F1, EM)
43.8AccuracySPARQL
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SPARQLModel=70B, GW=false2026.03 | 43.8 | 32.8 | 23.4 | |
| SPARQL + GWModel=70B, GW=true2026.03 | 42.2 | 34 | 25.8 | |
| Base + GWModel=70B, GW=true2026.03 | 39.8 | 30.8 | 20.6 | |
| BaseModel=70B, GW=false2026.03 | 35.2 | 26.9 | 18.6 | |
| SPARQL + GWModel=8B, GW=true2026.03 | 30.6 | 22.9 | 15.8 | |
| SPARQLModel=8B, GW=false2026.03 | 28.8 | 22.6 | 15.4 | |
| BaseModel=8B, GW=false2026.03 | 23.6 | 18 | 12.4 | |
| Base + GWModel=8B, GW=true2026.03 | 19.4 | 14.6 | 9 |