Mathematical Reasoning on Overall Aggregate
68.7AccuracyPUMA
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| PUMABackbone=Llama-3.1-Nemotron-Nano-8B2026.05 | 68.7 | 21.9 | 2.09 | 1.15 | |
| DEERBackbone=DeepSeek-R1-Distill-Qwen-14B2026.05 | 68.3 | 15.8 | 8.27 | 0.71 | |
| Full CoTBackbone=Llama-3.1-Nemotron-Nano-8B2026.05 | 68.3 | 0 | 2.41 | 1 | |
| PUMABackbone=DeepSeek-R1-Distill-Qwen-14B2026.05 | 66.9 | 26.6 | 4.58 | 1.28 | |
| Full CoTBackbone=DeepSeek-R1-Distill-Qwen-14B2026.05 | 66.5 | 0 | 5.84 | 1 | |
| DynasorBackbone=DeepSeek-R1-Distill-Qwen-14B2026.05 | 64.8 | 5.1 | 27.87 | 0.21 | |
| DEERBackbone=Llama-3.1-Nemotron-Nano-8B2026.05 | 64.1 | 14.5 | 4.67 | 0.52 | |
| PUMABackbone=DeepSeek-R1-Distill-Qwen-7B2026.05 | 63 | 40.7 | 1.64 | 1.4 | |
| DynasorBackbone=Llama-3.1-Nemotron-Nano-8B2026.05 | 61.7 | -5.7 | 14.19 | 0.17 | |
| Full CoTBackbone=DeepSeek-R1-Distill-Qwen-7B2026.05 | 60.2 | 0 | 2.3 | 1 | |
| DEERBackbone=DeepSeek-R1-Distill-Qwen-7B2026.05 | 59.5 | 27.7 | 3.35 | 0.69 | |
| DynasorBackbone=DeepSeek-R1-Distill-Qwen-7B2026.05 | 47.5 | 31.8 | 2.08 | 1.11 |