ResearchBenchmarksMaximum Flow on BA and ER averaged (test)Follow36.59AccuracyTLG-F33.199634.079834.9635.8402Feb 11, 2026Evaluation ResultsMethodMethodLinksAccuracyMean Absolute ErrorTLG-FModel=gpt-4oModel=gpt-4o2026.0236.590.445L-OWLModel=gpt-4oModel=gpt-4o2026.0236.590.475C-OWLModel=gpt-4oModel=gpt-4o2026.0236.590.443CL-OWLModel=gpt-4oModel=gpt-4o2026.0236.590.438TLG-AModel=gpt-4oModel=gpt-4o2026.0233.330.494