List your company

Mathematical Reasoning on AMC 2023 (Mean@8)

95.3Mean@8DISA
29.05246.25163.4580.649May 17, 2026

Evaluation results

MethodLinks
DISABackbone=Qwen3-4B-Base2026.0595.3
FlowRLBackbone=Qwen3-4B-Base2026.0592.5
GSPOBackbone=Qwen3-4B-Base2026.0583.5
GRPOBackbone=Qwen3-4B-Base2026.0570
GRPOBackbone=Qwen2.5-7B2026.0561.3
DISABackbone=Qwen2.5-7B2026.0552.5
FlowRLBackbone=Qwen2.5-7B2026.0550.6
GSPOBackbone=Qwen2.5-7B2026.0548.1
VanillaBackbone=Qwen3-4B-Base2026.0542.6
VanillaBackbone=Qwen2.5-7B2026.0531.6