Question Answering on PubMedQA M=12 participants (test)
82.52AUCRouteLLM
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| RouteLLMBase Model=Aloe, Number of participants=12, Context Usage=with context2026.07 | 82.52 | 87.39 | 7.17 | 84.15 | 3.55 | |
| StackedGenBase Model=Aloe, Number of participants=122026.07 | 82.26 | 87.36 | 7.29 | 83.76 | 3.66 | |
| WALLA IBase Model=Aloe, Number of participants=122026.07 | 82.1 | 86.66 | 10.89 | 77.14 | 12.06 | |
| WALLA IIBase Model=Aloe, Number of participants=122026.07 | 78.65 | 85.17 | 10.61 | 83.84 | 8.46 | |
| PackLLMBase Model=Aloe, Number of participants=122026.07 | 73.65 | 78.97 | 7.44 | 87.67 | 14.64 | |
| SelfCertaintyBase Model=Aloe, Number of participants=122026.07 | 68.05 | 75.66 | 3.65 | 84.07 | 18.12 | |
| RouteLLMBase Model=Aloe, Number of participants=12, Context Usage=without context2026.07 | 63.85 | 65.8 | 4.72 | 28.31 | 27.34 | |
| RouterDCBase Model=Aloe, Number of participants=12, Context Usage=with context2026.07 | 63.35 | 66.17 | 4.75 | 41.58 | 27.2 | |
| UniformAvgBase Model=Aloe, Number of participants=122026.07 | 63.1 | 66.17 | 4.49 | — | 27.3 | |
| NIRTRouterBase Model=Aloe, Number of participants=12, Context Usage=with context2026.07 | 63.1 | 66.17 | 4.91 | 24.29 | 27.3 |