List your company

Math Reasoning on AIME 2024 (Acc@16)

40Accuracy @16Ctrl-R
10.963218.501626.0433.5784Mar 2, 2026

Evaluation results

MethodLinks
Ctrl-RBackbone=Qwen3-8B-Base, beta=0.22026.0340
DAPOBackbone=Qwen3-8B-Base2026.0336.04
Reward ShapingBackbone=Qwen3-8B-Base2026.0334.79
NL GuidanceBackbone=Qwen3-8B-Base2026.0334.58
Ctrl-RBackbone=Qwen3-1.7B-Base, beta=0.22026.0315.21
Reward ShapingBackbone=Qwen3-1.7B-Base2026.0314.37
NL GuidanceBackbone=Qwen3-1.7B-Base2026.0314.16
DAPOBackbone=Qwen3-1.7B-Base2026.0313.12
Ctrl-RBackbone=Qwen3-1.7B-Base, beta=1, Importance Sampling (IS)=true2026.0312.29
Ctrl-RBackbone=Qwen3-1.7B-Base, beta=0, Importance Sampling (IS)=false2026.0312.08