List your company

Code Generation on HumanEval+ latest (held-out)

87.4Pass@1SDPO
80.22482.08783.9585.813Jun 14, 2026

Evaluation results

MethodLinks
SDPOBase Model=Qwen3-32B-Base2026.0687.4
HSDBase Model=Qwen3-32B-Base2026.0687.3
GRPO + OPSDBase Model=Qwen3-32B-Base2026.0687.1
RLSDBase Model=Qwen3-32B-Base2026.0687
OPSDBase Model=Qwen3-32B-Base2026.0686.9
DAPOBase Model=Qwen3-32B-Base2026.0686.8
GSPOBase Model=Qwen3-32B-Base2026.0686.6
Dr. GRPOBase Model=Qwen3-32B-Base2026.0686.5
GRPOBase Model=Qwen3-32B-Base2026.0686.3
SDPOBase Model=Qwen3-8B-Base2026.0682
HSDBase Model=Qwen3-8B-Base2026.0681.9
GRPO + OPSDBase Model=Qwen3-8B-Base2026.0681.7
RLSDBase Model=Qwen3-8B-Base2026.0681.6
OPSDBase Model=Qwen3-8B-Base2026.0681.4
DAPOBase Model=Qwen3-8B-Base2026.0681.1
GSPOBase Model=Qwen3-8B-Base2026.0680.9
Dr. GRPOBase Model=Qwen3-8B-Base2026.0680.7
GRPOBase Model=Qwen3-8B-Base2026.0680.5