ResearchBenchmarksPolicy generalization on MESSENGER-WM NewAttr (test)Follow115Average ScoreLED-WM25.5648.787295.22Nov 28, 2025Evaluation ResultsMethodMethodLinksAverage ScoreLED-WM2025.11115EMMA-LWMTraining protocol=Onli...Training protocol=Online IL2025.1196EMMA-LWMTraining protocol=Filt...Training protocol=Filtered BC (near-optimal)2025.1175EMMA-LWMTraining protocol=Filt...Training protocol=Filtered BC (suboptimal)2025.1129