Mathematical Reasoning on AIME (Throughput and Efficiency Metrics)
61.79ThroughputSTAND
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| STANDBackbone Model=DeepSeek-R1-Distill-Qwen-7B, Evaluation Pattern=Single-trajectory2025.06 | 61.79 | 2.32 | 3.07 | |
| SAM+RecycleBackbone Model=DeepSeek-R1-Distill-Qwen-7B, Evaluation Pattern=Single-trajectory2025.06 | 61.15 | 2.3 | 2.7 | |
| RecycleBackbone Model=DeepSeek-R1-Distill-Qwen-7B, Evaluation Pattern=Single-trajectory2025.06 | 60.61 | 2.28 | 2.73 | |
| ANPDBackbone Model=DeepSeek-R1-Distill-Qwen-7B, Evaluation Pattern=Single-trajectory2025.06 | 46.18 | 1.73 | 1.88 | |
| PLDBackbone Model=DeepSeek-R1-Distill-Qwen-7B, Evaluation Pattern=Single-trajectory2025.06 | 44.34 | 1.67 | 1.72 | |
| SAMBackbone Model=DeepSeek-R1-Distill-Qwen-7B, Evaluation Pattern=Single-trajectory2025.06 | 40.85 | 1.53 | 1.69 | |
| STANDBackbone Model=DeepSeek-R1-Distill-Qwen-14B, Evaluation Pattern=Single-trajectory2025.06 | 34.52 | 1.94 | 2.91 | |
| SAM+RecycleBackbone Model=DeepSeek-R1-Distill-Qwen-14B, Evaluation Pattern=Single-trajectory2025.06 | 34.35 | 1.93 | 2.67 | |
| RecycleBackbone Model=DeepSeek-R1-Distill-Qwen-14B, Evaluation Pattern=Single-trajectory2025.06 | 33.71 | 1.9 | 2.77 | |
| Eagle-2Backbone Model=DeepSeek-R1-Distill-Qwen-7B, Evaluation Pattern=Single-trajectory2025.06 | 29.91 | 1.12 | 2.21 | |
| PlainBackbone Model=DeepSeek-R1-Distill-Qwen-7B, Evaluation Pattern=Single-trajectory2025.06 | 26.63 | — | — | |
| ANPDBackbone Model=DeepSeek-R1-Distill-Qwen-14B, Evaluation Pattern=Single-trajectory2025.06 | 25.6 | 1.44 | 1.76 | |
| Eagle-2Backbone Model=DeepSeek-R1-Distill-Qwen-14B, Evaluation Pattern=Single-trajectory2025.06 | 25.38 | 1.43 | 2.72 | |
| SAMBackbone Model=DeepSeek-R1-Distill-Qwen-14B, Evaluation Pattern=Single-trajectory2025.06 | 23.26 | 1.31 | 1.63 | |
| PLDBackbone Model=DeepSeek-R1-Distill-Qwen-14B, Evaluation Pattern=Single-trajectory2025.06 | 21.82 | 1.23 | 1.61 | |
| PlainBackbone Model=DeepSeek-R1-Distill-Qwen-14B, Evaluation Pattern=Single-trajectory2025.06 | 17.76 | — | — |