Mathematical Reasoning on MATH (Accuracy and Efficiency)
38.4AccuracyConfidence
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ConfidenceBackbone=Dream-7B2025.12 | 38.4 | 11.2 | 1 | |
| Fast-dLLMBackbone=Dream-7B2025.12 | 37.9 | 27.1 | 2.42 | |
| STDDBackbone=Dream-7B2025.12 | 37.9 | 40.9 | 3.65 | |
| STDDBackbone=LLaDA-8B2025.12 | 35.1 | 32.91 | 3.74 | |
| ConfidenceBackbone=LLaDA-8B2025.12 | 33.4 | 8.8 | 1 | |
| Fast-dLLMBackbone=LLaDA-8B2025.12 | 33.4 | 23.2 | 2.63 | |
| DUSBackbone=Dream-7B, B=82025.12 | 27 | 30.3 | 2.71 | |
| DUSBackbone=LLaDA-8B, B=82025.12 | 21.4 | 23.6 | 2.68 |