Mathematical Reasoning on GSM8K (p@1 and p@16)
96.66p@1 AccuracyST
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| STModel=Qwen3-30B-A3B-Thinking2026.02 | 96.66 | 96.89 | |
| CoTModel=Qwen3-30B-A3B-Thinking2026.02 | 96.5 | 97.88 | |
| DoLaModel=Qwen3-30B-A3B-Thinking2026.02 | 96.5 | 97.88 | |
| ST-GModel=Qwen3-30B-A3B-Thinking2026.02 | 96.5 | 97.88 | |
| LEDModel=Qwen3-30B-A3B-Thinking2026.02 | 96.48 | 97.73 | |
| STModel=Qwen3-4B-Thinking2026.02 | 95.25 | 96.59 | |
| CoTModel=Qwen3-4B-Thinking2026.02 | 95.2 | 97.35 | |
| ST-GModel=Qwen3-4B-Thinking2026.02 | 95.2 | 97.19 | |
| LEDModel=Qwen3-4B-Thinking2026.02 | 95.2 | 97.35 | |
| DoLaModel=Qwen3-4B-Thinking2026.02 | 95.11 | 97.35 | |
| STModel=MiMo-7B-RL2026.02 | 94.7 | 96.36 | |
| LEDModel=MiMo-7B-RL2026.02 | 94.36 | 96.89 | |
| ST-GModel=MiMo-7B-RL2026.02 | 94.33 | 97.04 | |
| DoLaModel=MiMo-7B-RL2026.02 | 94.24 | 97.04 | |
| CoTModel=MiMo-7B-RL2026.02 | 94.16 | 97.19 |