Mathematical Reasoning on MATH (Pass@1 Accuracy)
80Pass@1 AccuracyGPT-4
Evaluation results
| Method | Links | |
|---|---|---|
| GPT-4Zero-shot=true2026.04 | 80 | |
| IRISBackbone=Qwen2.5, Zero-shot=true2026.04 | 64.6 | |
| Qwen2.5 BaseZero-shot=true2026.04 | 55.4 | |
| DeepSeek-MathZero-shot=true2026.04 | 36.2 | |
| IRISBackbone=WizMath, Zero-shot=true2026.04 | 29.5 | |
| MetaMathZero-shot=true2026.04 | 19.8 | |
| WizMath BaseZero-shot=true2026.04 | 10.7 |