List your company

Software Engineering on SWE-rebench 4B

31.49Avg@8 BC%EvoTrainer
24.407626.246328.08529.9237Jun 2, 2026

Evaluation results

MethodLinks
EvoTrainerConfiguration=Proposed Method2026.0631.49
Human-engineered RLConfiguration=Reference2026.0631.17
RAGEN v2 SNR FilteringConfiguration=Algorithmic Baseline2026.0630.03
RAGEN v1 FilteringConfiguration=Algorithmic Baseline2026.0629.87
Group-Variance FilteringConfiguration=Algorithmic Baseline2026.0629.06
AutoResearchConfiguration=Reference2026.0628.41
GRPO + Clip-HigherConfiguration=Algorithmic Baseline2026.0625.49
GSPO + Clip-HigherConfiguration=Algorithmic Baseline2026.0624.84
Base model (no RL)RL=None2026.0624.68