List your company

Out-of-Distribution Generalization Average (GPQA-Diamond, ARC-C, HumanEval)

55.35Pass@1TARPO
52.458853.209453.9654.7106Jun 4, 2026

Evaluation results

MethodLinks
TARPOBackbone=Qwen2.5-3B-Instruct2026.0655.3592.74337.9
HRPOBackbone=Qwen2.5-3B-Instruct2026.0654.0392.22407.2
GRPOBackbone=Qwen2.5-3B-Instruct2026.0653.8991.17412.2
CoTBackbone=Qwen2.5-3B-Instruct2026.0653.5293.79262.7
Pure LatentBackbone=Qwen2.5-3B-Instruct2026.0652.6765.95254.7
Entropy-RoutedBackbone=Qwen2.5-3B-Instruct2026.0652.5792.47243