ResearchBenchmarksNatural Language Processing on 12 NLP task categories (Heavy-Tail L Distribution)Follow1.24Average ImprovementFlexLoRA-0.5384-0.07670.3850.8467Feb 18, 2024Evaluation ResultsMethodMethodLinksAverage ImprovementFlexLoRABaseline Method=FedAvgBaseline Method=FedAvg2024.021.24FlexLoRABaseline Method=FedITBaseline Method=FedIT2024.020.63FlexLoRABaseline Method=AvgBaseline Method=Avg2024.020.47FlexLoRABaseline Method=SLoRABaseline Method=SLoRA2024.02-0.47