ResearchBenchmarksCausal Structure Prediction on Synthetic SEMs n=1024 samples (in-distribution)Causal Structure Prediction on Synthetic SEMs n=1024 samples (in-distribution)Follow0.902Adjacency AUROCTabPFN-CFM0.518240.617870.71750.81713Jun 25, 2026Evaluation resultsMethodMethodLinksAdjacency AUROCAdjacency AccuracyAncestral AUROCAncestral AccuracyConfounding AUROCConfounding AccuracyTabPFN-CFM2026.060.9020.910.9220.880.8770.94AVICI2026.060.7030.890.6820.82——GES2026.060.6720.810.6570.6——PC2026.060.6510.840.6320.46——LiNGAM2026.060.5680.780.540.7——FCI2026.060.5330.890.5360.810.540.9