Mathematical Reasoning on GSM8K (test) (Accuracy, CR, OR)
23.19AccuracyDGAO
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DGAOBackbone=Llama-3.2-3B2026.05 | 23.19 | 58.47 | 34.72 | |
| PromptBackbone=Llama-3.2-3B2026.05 | 22.99 | 56.59 | 37.81 | |
| DEmOBackbone=Llama-3.2-3B2026.05 | 22.95 | 56.91 | 37.45 | |
| GlobalEBackbone=Llama-3.2-3B2026.05 | 22.89 | 56.85 | 37.51 | |
| SFTBackbone=Llama-3.2-3B2026.05 | 22.85 | 56.78 | 37.56 | |
| SPHSBackbone=Llama-3.2-3B2026.05 | 22.48 | 56.12 | 37.38 | |
| PINEBackbone=Llama-3.2-3B2026.05 | 15.49 | 49.88 | 38.19 |