ResearchBenchmarksNode Classification on PubMed (test) (Robustness Metrics)Follow90.14FitSTEM-GNN85.9887.0688.1489.22Feb 9, 2026Evaluation ResultsMethodMethodLinksFitOOD WorstPerturbation MeanAverage PerformanceSTEM-GNN2026.0290.1484.7786.2687.06GFT2026.0289.4984.684.6886.26TFEGNN2026.0288.5285.1286.586.71CaNet2026.0288.4385.8886.0886.8GraphMETRO2026.0287.6384.383.3685.1MARIO2026.0286.1484.1383.5584.61