Mathematical Reasoning on GSM8K and MATH500 Aggregate
86.02Avg AccuracyL1
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| L1base_model=DeepSeek-1.5B2026.02 | 86.02 | 3,581.5 | -0.0022 | -0.0022 | |
| CRTbase_model=DeepSeek-1.5B2026.02 | 85.35 | 2,499.2 | 0.2901 | 0.2901 | |
| O1-Pruner-4.0base_model=DeepSeek-1.5B2026.02 | 84.97 | 3,006.5 | 0.1288 | 0.1288 | |
| DeepSeek-1.5B2026.02 | 84.81 | 3,428 | 0 | 0 | |
| ACPObase_model=DeepSeek-1.5B2026.02 | 84.11 | 2,713.3 | 0.1673 | 0.1261 | |
| ThinkPrune-1kbase_model=DeepSeek-1.5B2026.02 | 83.62 | 2,667 | 0.1515 | 0.0811 | |
| O1-Pruner-1.0base_model=DeepSeek-1.5B2026.02 | 83.06 | 2,723 | 0.1022 | -0.0013 | |
| ThinkPrune-500base_model=DeepSeek-1.5B2026.02 | 82.08 | 2,725 | 0.0441 | -0.1168 | |
| ShorterBetterbase_model=DeepSeek-1.5B2026.02 | 76.85 | 1,749.6 | 0.0207 | -0.4482 |