Mathematical Reasoning on AIME25 (Acc, TR, s/q)
53.3AccuracyDynasor
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DynasorBackbone=DeepSeek-R1-Distill-Qwen-14B2026.05 | 53.3 | -5.8 | 25.09 | |
| DEERBackbone=DeepSeek-R1-Distill-Qwen-7B2026.05 | 50 | 13.2 | 12.8 | |
| Full CoTBackbone=DeepSeek-R1-Distill-Qwen-14B2026.05 | 50 | 0 | 20.95 | |
| DEERBackbone=DeepSeek-R1-Distill-Qwen-14B2026.05 | 50 | 2.1 | 23.88 | |
| PUMABackbone=DeepSeek-R1-Distill-Qwen-14B2026.05 | 50 | 6.9 | 20.92 | |
| Full CoTBackbone=Llama-3.1-Nemotron-Nano-8B2026.05 | 50 | 0 | 9.82 | |
| DEERBackbone=Llama-3.1-Nemotron-Nano-8B2026.05 | 50 | -35.9 | 26.61 | |
| PUMABackbone=Llama-3.1-Nemotron-Nano-8B2026.05 | 50 | 18.1 | 11.31 | |
| PUMABackbone=DeepSeek-R1-Distill-Qwen-7B2026.05 | 46.7 | 29.8 | 6.88 | |
| Full CoTBackbone=DeepSeek-R1-Distill-Qwen-7B2026.05 | 43.3 | 0 | 10.56 | |
| DynasorBackbone=Llama-3.1-Nemotron-Nano-8B2026.05 | 43.3 | -14.4 | 11.37 | |
| DynasorBackbone=DeepSeek-R1-Distill-Qwen-7B2026.05 | 33.3 | 9.1 | 15.26 |