Follow-up Question Answering on MathChat-FQA 2nd turn
50.3AccuracyDART-Math-U.
Evaluation Results
| Method | Links | |
|---|---|---|
| DART-Math-U.Backbone=LLaMA3-8B2025.03 | 50.3 | |
| MetaMathBackbone=LLaMA3-8B2025.03 | 49.4 | |
| LEMMABackbone=LLaMA3-8B2025.03 | 49.4 | |
| RefAug-90kBackbone=LLaMA3-8B2025.03 | 47 | |
| ISCBackbone=LLaMA3-8B2025.03 | 46.8 | |
| DART-Math-H.Backbone=LLaMA3-8B2025.03 | 46.2 | |
| GPTAugBackbone=LLaMA3-8B2025.03 | 45.6 | |
| RefAugBackbone=LLaMA3-8B2025.03 | 45 | |
| RFTBackbone=LLaMA3-8B2025.03 | 40.8 | |
| SFTBackbone=LLaMA3-8B2025.03 | 37.2 |