Open-ended generation on MATH500
2.5FDRSGR
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| SGRModel=Qwen3-4B-Thinking-2507, Uncertainty Score=logits-based2025.10 | 2.5 | 57.25 | 57.11 | — | |
| SGRModel=Qwen3-4B-Instruct-2507, Uncertainty Score=logits-based2025.10 | 3.18 | 52.83 | 50.72 | — | |
| FDR searchModel=Qwen3-4B-Thinking-2507, Uncertainty Score=logits-based2025.10 | 4.01 | 99.69 | 99.67 | — | |
| Conformal LabelingModel=Qwen3-4B-Thinking-2507, Uncertainty Score=logits-based2025.10 | 4.07 | 99.8 | 99.78 | — | |
| Conformal LabelingModel=DeepSeek-R1-Distill-Qwen-32B, Uncertainty Score=logits-based2025.10 | 4.18 | 8.28 | 8.41 | — | |
| FDR searchModel=Qwen3-4B-Instruct-2507, Uncertainty Score=logits-based2025.10 | 4.88 | 84.94 | 82.1 | — | |
| Conformal LabelingModel=Qwen3-4B-Instruct-2507, Uncertainty Score=logits-based2025.10 | 4.93 | 86.73 | 83.88 | — | |
| FDR searchModel=DeepSeek-R1-Distill-Qwen-32B, Uncertainty Score=logits-based2025.10 | 29.83 | 10.32 | 10.63 | — | |
| SGRModel=DeepSeek-R1-Distill-Qwen-32B, Uncertainty Score=logits-based2025.10 | 38.7 | 0.31 | 0.61 | — | |
| AI onlyModel=Qwen3-4B-Instruct-25072025.10 | — | — | — | 8.2 | |
| AI onlyModel=Qwen3-4B-Thinking-25072025.10 | — | — | — | 4 | |
| AI onlyModel=DeepSeek-R1-Distill-Qwen-32B2025.10 | — | — | — | 7 |