List your company

Mathematical Reasoning on Math (Accuracy, Inference Latency)

93.28AccuracyHDPO
83.732886.211488.6991.1686Jun 2, 2026

Evaluation results

MethodLinks
HDPOBackbone=Qwen2.5-Math-7B2026.0693.280.16
Critique-GRPOBackbone=Qwen2.5-Math-7B2026.0691.220.19
GRPOBackbone=Qwen2.5-Math-7B2026.0689.720.12
LUFFYBackbone=Qwen2.5-Math-7B2026.0687.990.14
OatBackbone=Qwen2.5-Math-7B2026.0685.340.07
SimpleRLBackbone=Qwen2.5-Math-7B2026.0685.220.09
EDGE-GRPOBackbone=Qwen2.5-Math-7B2026.0684.10.07