Graduate-level Science Reasoning on GPQA (test)
34.3AccuracyE1-Math
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| E1-Mathbackbone=1.5B, budget=(4K,1K)2025.05 | 34.3 | 2,758.77 | |
| DeepSeek-R1-Distill-Qwenbackbone=1.5B2025.05 | 33.04 | 11,780.87 | |
| L1setting=Max(3600)2025.05 | 31.92 | 3,892.47 | |
| SelfBudgeterbackbone=1.5B2025.05 | 30.65 | 3,326.83 | |
| E1-Mathbackbone=1.5B, budget=(0.5K,1K)2025.05 | 26.34 | 1,278.19 |