Mathematical Reasoning on GSM8K (Accuracy & Step)
80.44AccuracyKLASS
Evaluation results
| Method | Links | ||
|---|---|---|---|
| KLASSModel=Dream2025.11 | 80.44 | 156.24 | |
| Top-k MarginModel=Dream2025.11 | 79.45 | 256 | |
| EntropyModel=Dream2025.11 | 79.45 | 256 | |
| KLASSModel=LLaDA2025.11 | 76.5 | 98.57 | |
| EntropyModel=LLaDA2025.11 | 75.43 | 256 | |
| Top-k MarginModel=LLaDA2025.11 | 74.45 | 128 | |
| Top-k MarginModel=LLaDA2025.11 | 74.14 | 256 | |
| EntropyModel=LLaDA2025.11 | 73.01 | 128 | |
| Top-k MarginModel=Dream2025.11 | 71.49 | 128 | |
| EntropyModel=Dream2025.11 | 71.49 | 128 |