List your company

Mathematical Reasoning on OlympiadBench (Accuracy, Average)

45.5AccuracyGRPO
35.6238.18540.7543.315Oct 7, 2025

Evaluation results

MethodLinks
GRPOConfiguration=Standard2025.1045.534.7
GRPOConfiguration=Higher Temp.2025.1044.335.2
EEPOConfiguration=Standard2025.1044.338.3
GRPOConfiguration=More rollouts.2025.104435.1
GRPOConfiguration=Increased Ent.2025.1042.835.9
GRPOConfiguration=DAPO Clip High.2025.1041.632.7
Qwen3-8B-BaseConfiguration=Base Model2025.103629