ResearchBenchmarksClassification on Academic random splits (test)Follow2.9Relative Error Rate ReductionOCTree-0.0120.7441.52.256Jun 12, 2024Evaluation ResultsMethodMethodLinksRelative Error Rate ReductionTest Error RateOCTreeLLM Backbone=GPT-4oLLM Backbone=GPT-4o2024.062.9—CAAFELLM Backbone=GPT-4oLLM Backbone=GPT-4o2024.060.1—