Error Correction on MathChat-EC (test)
84.7AccuracyLEMMA
Evaluation Results
| Method | Links | |
|---|---|---|
| LEMMABackbone=LLaMA3-8B2025.03 | 84.7 | |
| RefAug-90kBackbone=LLaMA3-8B2025.03 | 83.8 | |
| RefAugBackbone=LLaMA3-8B2025.03 | 82.7 | |
| DART-Math-U.Backbone=LLaMA3-8B2025.03 | 80.6 | |
| GPTAugBackbone=LLaMA3-8B2025.03 | 80 | |
| DART-Math-H.Backbone=LLaMA3-8B2025.03 | 79.8 | |
| ISCBackbone=LLaMA3-8B2025.03 | 78.5 | |
| MetaMathBackbone=LLaMA3-8B2025.03 | 66.6 | |
| SFTBackbone=LLaMA3-8B2025.03 | 66.1 | |
| RFTBackbone=LLaMA3-8B2025.03 | 62.6 |