Mathematical Reasoning on AIME 2024 (avg@8, pass@8)
48.33Average Score (k=8)Mixed
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MixedFramework=TD-Grokking2026.06 | 48.33 | 80 | |
| DynamicFramework=TD-Grokking2026.06 | 47.91 | 80 | |
| QuestAFramework=GRPO2026.06 | 47.5 | 76.67 | |
| AblationFramework=GRPO2026.06 | 46.67 | 73.33 | |
| SFTFramework=GRPO2026.06 | 43.75 | 73.33 | |
| Base2026.06 | 42.08 | 70 | |
| VanillaFramework=GRPO2026.06 | 42.08 | 76.67 | |
| Scaf-GRPOFramework=GRPO2026.06 | 42.08 | 73.33 | |
| Sub-onlyFramework=TD-Grokking2026.06 | 40.42 | 73.33 |