Question Answering on GPQA (Acc (%), #Calls)
50AccuracySC
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SCBudget (B)=162025.11 | 50 | 16 | |
| ASC-βBudget (B)=162025.11 | 48.1 | 12.37 | |
| ASC-DirBudget (B)=162025.11 | 48.1 | 13.15 | |
| ESCBudget (B)=16, Window size (w)=42025.11 | 48 | 12.88 | |
| ESCBudget (B)=16, Window size (w)=82025.11 | 48 | 15.2 | |
| CGES-DeepConf (B10)Budget (B)=16, DeepConf variant=B102025.11 | 47.9 | 12.36 | |
| CGES-MARSBudget (B)=162025.11 | 47.9 | 14.21 | |
| CGES-LNS (arith)Budget (B)=16, Aggregation function=arith2025.11 | 47.8 | 11.9 | |
| CGES-DeepConf (tail)Budget (B)=16, DeepConf variant=tail2025.11 | 47.7 | 11.48 | |
| CGES-LNS (geom)Budget (B)=16, Aggregation function=geom2025.11 | 47.4 | 13.62 | |
| DSCBudget (B)=162025.11 | 46.4 | 6.69 | |
| CGES-RMBudget (B)=162025.11 | 45.2 | 14.51 |