Math Reasoning on SVAMP (Acc, Avg Cost)
96Accuracy (%)MACA (Ours)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MACA (Ours)Model=Llama-3.1-8B, Structure Adaptation=✓, Orchestration Adaptation=✓2026.05 | 96 | 2,057.2 | |
| PuppeteerModel=Llama-3.1-8B, Structure Adaptation=✗, Orchestration Adaptation=✓2026.05 | 94.18 | 4,204.7 | |
| MaASModel=Llama-3.1-8B, Structure Adaptation=✓, Orchestration Adaptation=✗2026.05 | 92.28 | 2,827.7 | |
| SC (CoT×5)Model=Llama-3.1-8B, Structure Adaptation=✗, Orchestration Adaptation=✗2026.05 | 88.67 | 1,080.7 | |
| AgentPruneModel=Llama-3.1-8B, Structure Adaptation=✓, Orchestration Adaptation=✗2026.05 | 88.44 | 2,780.8 | |
| ComplexCoTModel=Llama-3.1-8B, Structure Adaptation=✗, Orchestration Adaptation=✗2026.05 | 86.33 | 2,658.9 | |
| DyLANModel=Llama-3.1-8B, Structure Adaptation=✗, Orchestration Adaptation=✗2026.05 | 86.33 | 6,675.6 | |
| AgentVerseModel=Llama-3.1-8B, Structure Adaptation=✗, Orchestration Adaptation=✗2026.05 | 86.03 | 4,184.7 | |
| Self-RefineModel=Llama-3.1-8B, Structure Adaptation=✗, Orchestration Adaptation=✗2026.05 | 83.67 | 1,195.3 | |
| CoTModel=Llama-3.1-8B, Structure Adaptation=✗, Orchestration Adaptation=✗2026.05 | 82.51 | 267.8 | |
| VanillaModel=Llama-3.1-8B, Structure Adaptation=✗, Orchestration Adaptation=✗2026.05 | 82.18 | 209.6 | |
| MacNetModel=Llama-3.1-8B, Structure Adaptation=✗, Orchestration Adaptation=✗2026.05 | 79.74 | 6,912.6 |