Mathematical Reasoning on AIME24 (Accuracy, Delta Accuracy, Length, Delta Length Percentage)
53.54AccuracyAdaptThink
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| AdaptThinkModel=DeepSeek-R1-Distill-Qwen-7B2025.08 | 53.54 | 2.08 | 11,002 | -17.96 | |
| OriginalModel=DeepSeek-R1-Distill-Qwen-7B2025.08 | 51.46 | — | 13,411 | — | |
| AutoThinkModel=DeepSeek-R1-Distill-Qwen-7B2025.08 | 50.42 | -1.04 | 8,215 | -38.74 | |
| LCPOModel=DeepSeek-R1-Distill-Qwen-7B2025.08 | 48.75 | -2.71 | 6,892 | -48.61 | |
| LCPOModel=DeepSeek-R1-Distill-Qwen-1.5B2025.08 | 29.58 | 0.41 | 8,810 | -46.13 | |
| OriginalModel=DeepSeek-R1-Distill-Qwen-1.5B2025.08 | 29.17 | — | 16,355 | — | |
| AutoThinkModel=DeepSeek-R1-Distill-Qwen-1.5B2025.08 | 28.96 | -0.21 | 7,338 | -55.13 | |
| AdaptThinkModel=DeepSeek-R1-Distill-Qwen-1.5B2025.08 | 27.08 | -2.09 | 8,494 | -48.07 |