List your company

Mathematical Problem Solving on MATH 500 (Pass@k)

79.3Pass@kGCPO
61.82866.36470.975.436May 12, 2026

Evaluation results

MethodLinks
GCPOk=64, Base Model=Qwen3-1.7B2026.0579.3
GCPOk=32, Base Model=Qwen3-1.7B2026.0578.4
DQOk=64, Base Model=Qwen3-1.7B2026.0578.2
DQOk=32, Base Model=Qwen3-1.7B2026.0578
GRPOk=64, Base Model=Qwen3-1.7B2026.0577.6
GRPOk=32, Base Model=Qwen3-1.7B2026.0577.3
GCPOk=16, Base Model=Qwen3-1.7B2026.0577
Basek=64, Base Model=Qwen3-1.7B2026.0576.9
DQOk=16, Base Model=Qwen3-1.7B2026.0576.3
Basek=32, Base Model=Qwen3-1.7B2026.0575.8
GCPOk=8, Base Model=Qwen3-1.7B2026.0575.6
GRPOk=16, Base Model=Qwen3-1.7B2026.0575.4
DQOk=8, Base Model=Qwen3-1.7B2026.0574.8
GRPOk=8, Base Model=Qwen3-1.7B2026.0574
Basek=16, Base Model=Qwen3-1.7B2026.0573.9
GCPOk=4, Base Model=Qwen3-1.7B2026.0573.5
DQOk=4, Base Model=Qwen3-1.7B2026.0572.7
GRPOk=4, Base Model=Qwen3-1.7B2026.0571.8
Basek=8, Base Model=Qwen3-1.7B2026.0571.1
GCPOk=2, Base Model=Qwen3-1.7B2026.0570.3
DQOk=2, Base Model=Qwen3-1.7B2026.0569.4
GRPOk=2, Base Model=Qwen3-1.7B2026.0568.2
Basek=4, Base Model=Qwen3-1.7B2026.0567.6
Basek=2, Base Model=Qwen3-1.7B2026.0562.5