ResearchBenchmarksSpan-level classification on E2E (test)Follow56F1 Score (Span)Poolinge11.2822.8934.546.11Dec 28, 2023Evaluation ResultsMethodMethodLinksF1 Score (Span)PoolingeType=Attention-Pooling...Type=Attention-Pooling Probe, Mode=Ensemble2023.1256Expert Human2023.1256LineareType=Linear Probe, Mod...Type=Linear Probe, Mode=Ensemble2023.1255LinearsLType=Linear Probe, Mod...Type=Linear Probe, Mode=Single-layer2023.1254PoolingsLType=Attention-Pooling...Type=Attention-Pooling Probe, Mode=Single-layer2023.1254OC2023.1213