List your company

Reasoning on NP-Bench (test)

93.7Graph SRGRPO
7.69230.02152.3574.679May 19, 2026

Evaluation results

MethodLinks
GRPO2026.0593.730.56949.610051.863.344.39124.383.440.1
GPG2026.0592.723.59628.510051.859.339.69520.388.632.7
ClipCov2026.059125.59352.29649.873.358.182.521.287.241.3
GSPO2026.0585.724.18056.59951.265.746.782.527.482.641.2
DMPO2026.0585.325.68260.310052.867.75581.525.683.343.9
Qwen2.5-7B-InstructRole=backbone2026.05113.14019.8673426.715.23.5129.614.6