List your company

Mathematical Reasoning Benchmarks Average

71.88Pass@4 ScoreThinkTwice
60.502463.456266.4169.3638Apr 2, 2026

Evaluation results

MethodLinks
ThinkTwiceBackbone=Qwen3-4B, Training-free=false2026.0471.88
ThinkTwiceBackbone=OLMo3-7B, Training-free=false2026.0469.35
DAPOBackbone=Qwen3-4B, Training-free=false2026.0469.01
DrGRPOBackbone=Qwen3-4B, Training-free=false2026.0468.14
GRPOBackbone=Qwen3-4B, Training-free=false2026.0467.42
Self-RefineBackbone=OLMo3-7B, Training-free=true2026.0467.01
Self-RefineBackbone=Qwen3-4B, Training-free=true2026.0466.83
DrGRPOBackbone=OLMo3-7B, Training-free=false2026.0466.73
DAPOBackbone=OLMo3-7B, Training-free=false2026.0466.69
GRPOBackbone=OLMo3-7B, Training-free=false2026.0466.08
Base ModelBackbone=Qwen3-4B, Training-free=false2026.0464.1
Base ModelBackbone=OLMo3-7B, Training-free=false2026.0463.39
ReflexionBackbone=Qwen3-4B, Training-free=true2026.0460.98
ReflexionBackbone=OLMo3-7B, Training-free=true2026.0460.94