Mathematical Reasoning on GSM8K 776 questions (test)
0Cost ($)SLM
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SLMStrategy Type=Baseline2026.01 | 0 | 73 | — | — | |
| Reactive ShepherdingStrategy Type=Cascading2026.01 | 0.034 | 89.1 | 67.4 | 1.97 | |
| Oracle ShepherdingStrategy Type=Oracle2026.01 | 0.034 | 98 | 67.4 | 3.1 | |
| Proactive ShepherdingStrategy Type=Routing2026.01 | 0.036 | 80.9 | 65.9 | 0.93 | |
| GraphRouterStrategy Type=Routing2026.01 | 0.041 | 82.2 | 60.6 | 0.93 | |
| FrugalGPTStrategy Type=Cascading2026.01 | 0.047 | 86.7 | 54.8 | 1.21 | |
| ABCStrategy Type=Cascading2026.01 | 0.048 | 94.9 | 53.7 | 1.89 | |
| RouteLLMStrategy Type=Routing2026.01 | 0.065 | 88.1 | 37.5 | 0.98 | |
| LLMStrategy Type=Baseline2026.01 | 0.104 | 98 | — | — |