List your company

Mathematical Reasoning on MATH500 (Pass@1, TC)

78.8Pass@1EAPO
51.13658.31865.572.682Jun 1, 2026

Evaluation results

MethodLinks
EAPOBackbone Model=Qwen2.5-7B-Instruct2026.0678.80.98
ARPOBackbone Model=Qwen2.5-7B-Instruct2026.06781.02
GRPOBackbone Model=Qwen2.5-7B-Instruct2026.0677.60.98
AEPOBackbone Model=Qwen2.5-7B-Instruct2026.0677.61.1
Reinforce++Backbone Model=Qwen2.5-7B-Instruct2026.0676.81.02
ToolStarBackbone Model=Qwen2.5-7B-Instruct2026.0676.81.38
BaseBackbone Model=Qwen2.5-7B-Instruct2026.0670.6—
EAPOBackbone Model=Qwen2.5-3B-Instruct2026.0670.11.03
GRPOBackbone Model=Qwen2.5-3B-Instruct2026.06701.14
ARPOBackbone Model=Qwen2.5-3B-Instruct2026.0669.81.07
Reinforce++Backbone Model=Qwen2.5-3B-Instruct2026.0669.41.09
ToolStarBackbone Model=Qwen2.5-3B-Instruct2026.06691.01
TIRBackbone Model=Qwen2.5-7B-Instruct2026.0668.20.51
BaseBackbone Model=Qwen2.5-3B-Instruct2026.0659.2—
TIRBackbone Model=Qwen2.5-3B-Instruct2026.0652.20.81