Mathematical Reasoning on Minerva-Math (Acc, Delta Metrics)
38.97Accuracy (Acc)LCPO
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| LCPOModel=DeepSeek-R1-Distill-Qwen-7B2025.08 | 38.97 | 2.21 | 2,079 | -64.92 | |
| OriginalModel=DeepSeek-R1-Distill-Qwen-7B2025.08 | 36.76 | — | 5,926 | — | |
| AdaptThinkModel=DeepSeek-R1-Distill-Qwen-7B2025.08 | 36.4 | -0.36 | 2,787 | -52.97 | |
| AutoThinkModel=DeepSeek-R1-Distill-Qwen-7B2025.08 | 36.03 | -0.73 | 2,794 | -52.85 | |
| AutoThinkModel=DeepSeek-R1-Distill-Qwen-1.5B2025.08 | 27.57 | 0.73 | 2,397 | -66.76 | |
| LCPOModel=DeepSeek-R1-Distill-Qwen-1.5B2025.08 | 27.21 | 0.37 | 2,596 | -64 | |
| OriginalModel=DeepSeek-R1-Distill-Qwen-1.5B2025.08 | 26.84 | — | 7,211 | — | |
| AdaptThinkModel=DeepSeek-R1-Distill-Qwen-1.5B2025.08 | 25 | -1.84 | 1,910 | -73.51 |