Common Sense Reasoning on BBH Sports Understanding
88Accuracy (BBH Sports)Dynamic Workflow (ReAct)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Dynamic Workflow (ReAct)Workflow Type=Dynamic, LLM Model=DeepSeek-V3.12026.05 | 88 | 0.29 | |
| Static WorkflowWorkflow Type=Static, LLM Model=DeepSeek-V3.12026.05 | 87 | 0.13 | |
| evaluation-instructed prompt optimizationBackbone=GPT-4o, Optimization Type=Query-dependent2025.11 | 83 | — | |
| Pro-RefineBackbone=GPT-4o, Optimization Type=Query-dependent2025.11 | 82 | — | |
| TextGradBackbone=GPT-4o, Optimization Type=Template2025.11 | 80 | — | |
| Self-RefineBackbone=GPT-4o, Optimization Type=Query-dependent2025.11 | 79 | — | |
| APEBackbone=GPT-4o, Optimization Type=Template2025.11 | 79 | — | |
| LLM onlyBackbone=GPT-4o, Optimization Type=None2025.11 | 78 | — | |
| evaluation-instructed prompt optimizationBackbone=LLaMA-3, Optimization Type=Query-dependent2025.11 | 75 | — | |
| evaluation-instructed prompt optimizationBackbone=LLaMA-3.1, Optimization Type=Query-dependent2025.11 | 75 | — | |
| TextGradBackbone=LLaMA-3.1, Optimization Type=Template2025.11 | 73 | — | |
| Pro-RefineBackbone=LLaMA-3, Optimization Type=Query-dependent2025.11 | 71 | — | |
| Self-RefineBackbone=LLaMA-3.1, Optimization Type=Query-dependent2025.11 | 71 | — | |
| TextGradBackbone=LLaMA-3, Optimization Type=Template2025.11 | 70 | — | |
| Pro-RefineBackbone=LLaMA-3.1, Optimization Type=Query-dependent2025.11 | 70 | — | |
| APEBackbone=LLaMA-3, Optimization Type=Template2025.11 | 69 | — | |
| LLM onlyBackbone=LLaMA-3, Optimization Type=None2025.11 | 68 | — | |
| Self-RefineBackbone=LLaMA-3, Optimization Type=Query-dependent2025.11 | 68 | — | |
| APEBackbone=LLaMA-3.1, Optimization Type=Template2025.11 | 68 | — | |
| LLM onlyBackbone=LLaMA-3.1, Optimization Type=None2025.11 | 67 | — | |
| Zero-shot (Default Imp.)Workflow Type=Zero-shot, LLM Model=DeepSeek-V3.12026.05 | 65 | 0.02 |