Mathematical Reasoning on AIME 24 (Accuracy and Length Metrics)
30.63AccuracyO1-Pruner
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| O1-PrunerBase Model=DeepSeek-1.5B, Ratio=4.02026.02 | 30.63 | 12,669 | |
| DeepSeek-1.5BModel Type=Base Model2026.02 | 30.42 | 14,044 | |
| ACPOBase Model=DeepSeek-1.5B2026.02 | 30.42 | 12,367 | |
| L1Base Model=DeepSeek-1.5B2026.02 | 29.38 | 13,565 | |
| ThinkPruneBase Model=DeepSeek-1.5B, Constraint=1k2026.02 | 28.96 | 13,001 | |
| O1-PrunerBase Model=DeepSeek-1.5B, Ratio=1.02026.02 | 28.54 | 12,930 | |
| CRTBase Model=DeepSeek-1.5B2026.02 | 28.54 | 12,674 | |
| ThinkPruneBase Model=DeepSeek-1.5B, Constraint=5002026.02 | 27.92 | 13,378 | |
| ShorterBetterBase Model=DeepSeek-1.5B2026.02 | 27.08 | 11,714 |