List your company

Multi-Turn Tool-Integrated Reasoning (TIR) on MATH500

84.69Peak avg@32 ScoreOTB
75.3377.7680.1982.62Feb 6, 2026

Evaluation results

MethodLinks
OTBBackbone=Qwen2.5-7B, Reasoning Paradigm=TIR2026.0284.69
SimpleTIRBackbone=Qwen2.5-7B, Reasoning Paradigm=TIR2026.0282.25
OPOBackbone=Qwen2.5-7B, Reasoning Paradigm=TIR2026.0276.94
GRPOBackbone=Qwen2.5-7B, Reasoning Paradigm=TIR2026.0276.88
RLOOBackbone=Qwen2.5-7B, Reasoning Paradigm=TIR2026.0276.06
OGBBackbone=Qwen2.5-7B, Reasoning Paradigm=TIR2026.0275.69