Mathematical Reasoning on AIME 24&25 (Acc, Maj. Vote)
67.5AccuracyOrch-RM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Orch-RMContext=Continued-training, Sampling Level=Orchestration2026.06 | 67.5 | 68.33 | |
| GRPOContext=Continued-training, Sampling Level=Trajectory2026.06 | 65 | 65 | |
| RFTContext=Continued-training, Sampling Level=Trajectory2026.06 | 64.83 | 65 | |
| DPOContext=Continued-training2026.06 | 64.67 | 65 | |
| Orch-RMContext=Training-from-scratch, Sampling Level=Orchestration2026.06 | 64.33 | 61.67 | |
| LLM-as-a-judgeContext=Continued-training, Sampling Level=Orchestration2026.06 | 59.17 | 61.67 | |
| MAS-OrchestraContext=Training-from-scratch, Sampling Level=Trajectory2026.06 | 58.96 | 63.33 | |
| Qwen2.5-7B-InstructContext=Base orchestrator2026.06 | 29.83 | 23.33 |