List your company

Mathematical Reasoning on Olympiad (Pass@k)

49.6Pass@kGCPO
35.87239.4364346.564May 12, 2026

Evaluation results

MethodLinks
GCPOk=64, Backbone=Qwen3-4B2026.0549.6
GCPOk=32, Backbone=Qwen3-4B2026.0549.1
DQOk=64, Backbone=Qwen3-4B2026.0548.9
GCPOk=16, Backbone=Qwen3-4B2026.0548.7
DQOk=32, Backbone=Qwen3-4B2026.0548.6
GRPOk=64, Backbone=Qwen3-4B2026.0548.1
GCPOk=8, Backbone=Qwen3-4B2026.0547.6
GRPOk=32, Backbone=Qwen3-4B2026.0547.5
DQOk=16, Backbone=Qwen3-4B2026.0547.2
Basek=64, Backbone=Qwen3-4B2026.0546.8
GRPOk=16, Backbone=Qwen3-4B2026.0546.7
GCPOk=4, Backbone=Qwen3-4B2026.0546.4
Basek=32, Backbone=Qwen3-4B2026.0546.1
DQOk=8, Backbone=Qwen3-4B2026.0546
GRPOk=8, Backbone=Qwen3-4B2026.0545.3
GCPOk=2, Backbone=Qwen3-4B2026.0544.7
DQOk=4, Backbone=Qwen3-4B2026.0544.6
Basek=16, Backbone=Qwen3-4B2026.0544.2
GRPOk=4, Backbone=Qwen3-4B2026.0543.8
DQOk=2, Backbone=Qwen3-4B2026.0543.1
Basek=8, Backbone=Qwen3-4B2026.0542.2
GRPOk=2, Backbone=Qwen3-4B2026.0541.9
Basek=4, Backbone=Qwen3-4B2026.0539.7
Basek=2, Backbone=Qwen3-4B2026.0536.4