Question Answering on PubMedQA M=8 participants (test)
84.47AUCWALLA I
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| WALLA IBase Model=Aloe, Number of participants=82026.07 | 84.47 | 87.09 | 8.35 | 81.9 | 9.32 | |
| StackedGenBase Model=Aloe, Number of participants=82026.07 | 82.65 | 87.44 | 7.48 | 85.25 | 3.49 | |
| RouteLLMBase Model=Aloe, Number of participants=8, Context Usage=with context2026.07 | 82.22 | 87.5 | 7.21 | 85.77 | 3.54 | |
| WALLA IIBase Model=Aloe, Number of participants=82026.07 | 79.85 | 85.95 | 10.45 | 85.07 | 7.15 | |
| PackLLMBase Model=Aloe, Number of participants=82026.07 | 75.65 | 81.71 | 9.38 | 87.85 | 11.99 | |
| SelfCertaintyBase Model=Aloe, Number of participants=82026.07 | 69.31 | 76.92 | 3.33 | 84.73 | 16.5 | |
| RouterDCBase Model=Aloe, Number of participants=8, Context Usage=with context2026.07 | 65.17 | 67.51 | 3.1 | 47.36 | 25.46 | |
| RouteLLMBase Model=Aloe, Number of participants=8, Context Usage=without context2026.07 | 65.08 | 67.44 | 3.24 | 33.29 | 25.63 | |
| UniformAvgBase Model=Aloe, Number of participants=82026.07 | 64.85 | 67.2 | 3.12 | — | 25.35 | |
| NIRTRouterBase Model=Aloe, Number of participants=8, Context Usage=with context2026.07 | 64.85 | 67.53 | 3.18 | 33.38 | 25.57 |