Mathematical Reasoning on GSM8K (test) (Accuracy, Time (s))
80.97AccuracySTP
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| STPBase model=LLaDA-8B-Instruct2026.02 | 80.97 | 144,536 | |
| GRPO w/ ELBOBase model=LLaDA-8B-Instruct2026.02 | 80.52 | 163,035 | |
| Diffu-GRPOBase model=LLaDA-8B-Instruct2026.02 | 80.21 | 146,641 | |
| LLaDA-8B-InstructBase model=LLaDA-8B-Instruct2026.02 | 77.79 | — |