Mathematical Reasoning on GSM8K (Acc@1)
83.04Accuracy@1dOPSD
Evaluation Results
| Method | Links | |
|---|---|---|
| dOPSDBase Model=Dream-7B-Instruct2026.07 | 83.04 | |
| BaseBase Model=Dream-7B-Instruct2026.07 | 81.41 | |
| GRPOBase Model=Dream-7B-Instruct2026.07 | 81.28 | |
| SFTBase Model=Dream-7B-Instruct2026.07 | 80.44 | |
| OPSD (answer-only)Base Model=Dream-7B-Instruct2026.07 | 80.34 | |
| dOPSDBase Model=LLaDA-8B-Instruct2026.07 | 72.87 | |
| OPSD (answer-only)Base Model=LLaDA-8B-Instruct2026.07 | 71.47 | |
| BaseBase Model=LLaDA-8B-Instruct2026.07 | 71.23 | |
| GRPOBase Model=LLaDA-8B-Instruct2026.07 | 70.45 | |
| SFTBase Model=LLaDA-8B-Instruct2026.07 | 69.45 | |
| OPSD (full-solution)Base Model=Dream-7B-Instruct2026.07 | 67.25 | |
| OPSD (full-solution)Base Model=LLaDA-8B-Instruct2026.07 | 57.86 |