List your company

Accuracy on GSM8K

77.2AccuracyUDPO
33.5244.8656.267.54Feb 6, 2026

Evaluation results

MethodLinks
UDPOBase Model=Qwen, Evaluation Setting=Under attack2026.0277.2-5.2
RMAACBase Model=Qwen, Evaluation Setting=Under attack2026.0252.3-23.8
MAPPOBase Model=Qwen, Evaluation Setting=Under attack2026.0241.7-32.5
Standard MADBase Model=Qwen, Evaluation Setting=Under attack2026.0235.2-34.6