Mathematical Reasoning on SVAMP (accuracy)
94.31AccuracyCOMPACT
Evaluation Results
| Method | Links | |
|---|---|---|
| COMPACTStudent Architecture=Qwen2.5-7B, Distillation Method=COMPACT2026.01 | 94.31 | |
| EDITStudent Architecture=Qwen2.5-7B, Distillation Method=EDIT2026.01 | 93.5 | |
| SBS-KDStudent Architecture=Qwen2.5-7B, Distillation Method=SBS-KD2026.01 | 93 | |
| MoTStudent Architecture=Qwen2.5-7B, Distillation Method=MoT2026.01 | 93 | |
| CommitteeStudent Architecture=Qwen2.5-7B, Distillation Method=Committee2026.01 | 91.51 | |
| MCC-KDStudent Architecture=Qwen2.5-7B, Distillation Method=MCC-KD2026.01 | 91 | |
| Qwen2.5-7B-InstructStudent Architecture=Qwen2.5-7B, Distillation Method=Zero-shot2026.01 | 90.33 | |
| COMPACTStudent Architecture=Llama3.1-8B, Distillation Method=COMPACT2026.01 | 87.93 | |
| CommitteeStudent Architecture=Llama3.1-8B, Distillation Method=Committee2026.01 | 87.84 | |
| SBS-KDStudent Architecture=Llama3.1-8B, Distillation Method=SBS-KD2026.01 | 87.67 | |
| EDITStudent Architecture=Llama3.1-8B, Distillation Method=EDIT2026.01 | 87.67 | |
| Llama3.1-8B-InstructStudent Architecture=Llama3.1-8B, Distillation Method=Zero-shot2026.01 | 87 | |
| MoTStudent Architecture=Llama3.1-8B, Distillation Method=MoT2026.01 | 85 | |
| MCC-KDStudent Architecture=Llama3.1-8B, Distillation Method=MCC-KD2026.01 | 83.33 | |
| COMPACTStudent Architecture=Qwen2.5-1.5B, Distillation Method=COMPACT2026.01 | 81 | |
| MCC-KDStudent Architecture=Qwen2.5-1.5B, Distillation Method=MCC-KD2026.01 | 79.8 | |
| EDITStudent Architecture=Qwen2.5-1.5B, Distillation Method=EDIT2026.01 | 79.6 | |
| Qwen2.5-1.5B-InstructStudent Architecture=Qwen2.5-1.5B, Distillation Method=Zero-shot2026.01 | 79.33 | |
| SBS-KDStudent Architecture=Qwen2.5-1.5B, Distillation Method=SBS-KD2026.01 | 78.52 | |
| CommitteeStudent Architecture=Qwen2.5-1.5B, Distillation Method=Committee2026.01 | 77.98 | |
| MoTStudent Architecture=Qwen2.5-1.5B, Distillation Method=MoT2026.01 | 75.25 |