Mathematical Reasoning on MATH500 (Accuracy, Absolute Improvement, Relative Gain)
74.4AccuracyTTSV
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| TTSVModel=Qwen2.5-Math-7B2025.12 | 74.4 | 23.4 | 45.88 | |
| Qwen2.5-Math-7BModel=Qwen2.5-Math-7B2025.12 | 51 | — | — | |
| LLaMA-3.1-8BModel=LLaMA-3.1-8B2025.12 | 50.6 | — | — | |
| TTSVModel=LLaMA-3.1-8B2025.12 | 49.4 | -1.2 | -2.37 | |
| EMModel=Qwen2.5-Math-7B2025.12 | — | 15 | — | |
| EMModel=LLaMA-3.1-8B2025.12 | — | 0.2 | — |