Multi-task Evaluation on Aggregate All Tasks
74.6ScoreEvoRoute
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| EvoRouteSetting=Ours2026.01 | 74.6 | 138.4 | 92.23 | |
| Gemini-2.5-proSetting=Manual2026.01 | 74.27 | 521.61 | 123 | |
| MasRouterSetting=Routing2026.01 | 71.7 | 176.98 | 97.2 | |
| GraphRouterSetting=Routing2026.01 | 67.13 | 208.8 | 100.8 | |
| VMSSTInitialization=Pretrained, Number of Layers=242022.12 | 65.9 | — | — | |
| PromptLLMSetting=Routing2026.01 | 65.76 | 200.73 | 100.43 | |
| BITRANSLATIONInitialization=Pretrained, Number of Layers=242022.12 | 64.9 | — | — | |
| VMSST CONTRASTIVEInitialization=Pretrained, Number of Layers=242022.12 | 64.4 | — | — | |
| CONTRASTIVEInitialization=Pretrained, Number of Layers=242022.12 | 64.2 | — | — | |
| GPT-4oSetting=Manual2026.01 | 60.94 | 581.08 | 113.92 | |
| VMSSTInitialization=Random, Number of Layers=242022.12 | 58.1 | — | — | |
| GPT-4.1Setting=Manual2026.01 | 57.23 | 352.63 | 112.73 | |
| BITRANSLATIONInitialization=Random, Number of Layers=242022.12 | 55.9 | — | — | |
| VMSSTInitialization=Random, Number of Layers=62022.12 | 55.6 | — | — | |
| VMSST CONTRASTIVEInitialization=Random, Number of Layers=242022.12 | 55 | — | — | |
| CONTRASTIVEInitialization=Random, Number of Layers=62022.12 | 53.7 | — | — | |
| VMSST CONTRASTIVEInitialization=Random, Number of Layers=62022.12 | 53.7 | — | — | |
| Qwen3-14bSetting=Manual2026.01 | 53.62 | 49.8 | 78.05 | |
| BITRANSLATIONInitialization=Random, Number of Layers=62022.12 | 52.5 | — | — | |
| CONTRASTIVEInitialization=Random, Number of Layers=242022.12 | 52.4 | — | — |