List your company

Mathematical Reasoning on AIME 24 (Mean Accuracy)

10.6Mean AccuracyOPD (prefix scheduling)
-0.2162.5925.48.208Feb 16, 2026

Evaluation results

MethodLinks
OPD (prefix scheduling)Training Budget=sufficient, Zero-shot=true2026.0210.6
OPDTraining Budget=sufficient, Rollout=full, Zero-shot=true2026.0210
SeqKDTraining Budget=sufficient, Zero-shot=true2026.028.8
OPDPrefix Length=1024, Training Budget=limited, Zero-shot=true2026.028.1
OPDPrefix Length=2048, Training Budget=limited, Zero-shot=true2026.027.3
OPDSteps=10, Training Budget=limited, Zero-shot=true2026.025.4
OPDPrefix Length=256, Training Budget=limited, Zero-shot=true2026.024.8
SeqKDSteps=300, Training Budget=limited, Zero-shot=true2026.024.6
OPDPrefix Length=512, Training Budget=limited, Zero-shot=true2026.024.2
Qwen3-1.7B-BaseZero-shot=true2026.020.2