Complex Reasoning on SCoRE (test)
16.26AccuracySelfBudgeter
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SelfBudgeterbackbone=1.5B2025.05 | 16.26 | 4,491.35 | |
| L1setting=Max(3600)2025.05 | 13.69 | 5,145.91 | |
| E1-Mathbackbone=1.5B, budget=(4K,1K)2025.05 | 12.23 | 3,327.37 | |
| DeepSeek-R1-Distill-Qwenbackbone=1.5B2025.05 | 10.14 | 11,695.94 | |
| E1-Mathbackbone=1.5B, budget=(0.5K,1K)2025.05 | 6.69 | 1,272.04 |