List your company

Mathematical Reasoning on MATH500 (pass@4)

93.6pass@4ThinkTwice
88.264889.649991.03592.4201Apr 2, 2026

Evaluation results

MethodLinks
ThinkTwiceBackbone=Qwen3-4B, Evaluation Protocol=Direct prompting, Sampling count (n)=322026.0493.6
DAPOBackbone=Qwen3-4B, Evaluation Protocol=Direct prompting, Sampling count (n)=322026.0493.55
ThinkTwiceBackbone=OLMo3-7B, Evaluation Protocol=Direct prompting, Sampling count (n)=322026.0492.74
GRPOBackbone=OLMo3-7B, Evaluation Protocol=Direct prompting, Sampling count (n)=322026.0492.28
GRPOBackbone=Qwen3-4B, Evaluation Protocol=Direct prompting, Sampling count (n)=322026.0491.86
DrGRPOBackbone=Qwen3-4B, Evaluation Protocol=Direct prompting, Sampling count (n)=322026.0491.83
DrGRPOBackbone=OLMo3-7B, Evaluation Protocol=Direct prompting, Sampling count (n)=322026.0491.65
DAPOBackbone=OLMo3-7B, Evaluation Protocol=Direct prompting, Sampling count (n)=322026.0491.56
Base ModelBackbone=OLMo3-7B, Evaluation Protocol=Direct prompting, Sampling count (n)=322026.0489.87
Base ModelBackbone=Qwen3-4B, Evaluation Protocol=Direct prompting, Sampling count (n)=322026.0488.47