Mathematical Reasoning on GSM8K 128
76.65AccuracyFull
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| FullModel=LLaDA 1.52026.04 | 76.65 | 128 | 19.7 | |
| NI SamplingModel=LLaDA 1.52026.04 | 76.19 | 29.89 | 85.3 | |
| ThresholdModel=LLaDA 1.52026.04 | 75.82 | 47.44 | 53.5 | |
| FullModel=LLaDA-8B-Instruct2026.04 | 73.92 | 128 | 20.4 | |
| ThresholdModel=LLaDA-8B-Instruct2026.04 | 73.77 | 49.35 | 53.3 | |
| NI SamplingModel=LLaDA-8B-Instruct2026.04 | 73.69 | 34.36 | 76.5 |