Mathematical Reasoning on MATH (Accuracy and Avg. Time)
31.4AccuracyAtteNT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AtteNTModel=Gemma-7B2026.02 | 31.4 | 2,012 | |
| Standard Fine-tuningModel=Gemma-7B2026.02 | 30.52 | 2,282 | |
| AtteNTModel=Mistral-7B2026.02 | 23.12 | 1,802 | |
| Standard Fine-tuningModel=Mistral-7B2026.02 | 20.06 | 2,042 | |
| AtteNTModel=LLaMA 2-7B2026.02 | 6.48 | 2,132 | |
| Standard Fine-tuningModel=LLaMA 2-7B2026.02 | 5.06 | 2,461 |