Mathematical Reasoning on GSM8K (test) (Absolute Score, Relative Gain/Loss)
81.88Absolute ScoredInfer
Evaluation results
| Method | Links | ||
|---|---|---|---|
| dInferBase Model=Dream-7B-Instruct, Fine-tuning Dataset=Code Generation2026.05 | 81.88 | 0.47 | |
| T3DBase Model=Dream-7B-Instruct, Fine-tuning Dataset=Code Generation2026.05 | 81.84 | 0.43 | |
| SFT-SDBase Model=Dream-7B-Instruct, Fine-tuning Dataset=Code Generation2026.05 | 81.81 | 0.4 | |
| TABOMBase Model=Dream-7B-Instruct, Fine-tuning Dataset=Code Generation2026.05 | 81.73 | 0.32 | |
| No-SFTBase Model=Dream-7B-Instruct2026.05 | 81.41 | — | |
| dInferBase Model=LLaDA-8B-Instruct, Fine-tuning Dataset=Code Generation2026.05 | 77.33 | 1.21 | |
| TABOMBase Model=LLaDA-8B-Instruct, Fine-tuning Dataset=Code Generation2026.05 | 77.33 | 1.21 | |
| T3DBase Model=LLaDA-8B-Instruct, Fine-tuning Dataset=Code Generation2026.05 | 77.14 | 1.02 | |
| SFT-SDBase Model=LLaDA-8B-Instruct, Fine-tuning Dataset=Code Generation2026.05 | 76.95 | 0.83 | |
| No-SFTBase Model=LLaDA-8B-Instruct2026.05 | 76.12 | — | |
| SFT-GTBase Model=LLaDA-8B-Instruct, Fine-tuning Dataset=Code Generation2026.05 | 70.73 | -5.39 | |
| SFT-GTBase Model=Dream-7B-Instruct, Fine-tuning Dataset=Code Generation2026.05 | 52.33 | -29.08 |