Math Problem Solving on AIME 2025 (test)
30AccuracySelfBudgeter
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SelfBudgeterModel Scale=7B2025.05 | 30 | 12,241.84 | |
| DeepSeek-R1-Distill-QwenModel Scale=7B2025.05 | 28.89 | 22,158.79 | |
| DeepSeek-R1-Distill-QwenModel Scale=1.5B2025.05 | 22.22 | 14,444.03 | |
| E1-Math-1.5BTruncation Budget=4K, 1K2025.05 | 21.11 | 5,578.13 | |
| SelfBudgeterModel Scale=1.5B2025.05 | 21.11 | 4,288.1 | |
| L1-MaxSpecified Limit=3600 tokens2025.05 | 17.88 | 5,213.89 | |
| Eurus-2-7B-PRIMEModel Scale=7B2025.05 | 15.56 | 1,254.52 | |
| Qwen-2.5-7B-Simple-RLModel Scale=7B2025.05 | 6.67 | 1,429.94 | |
| E1-Math-1.5BTruncation Budget=0.5K, 1K2025.05 | 4.44 | 3,008.44 |