Skip to main content
. 2023 Aug 22;9(9):e19265. doi: 10.1016/j.heliyon.2023.e19265

Table 3.

Performance of the final ELECTRA model on each of the individual training datasets.

Dataset Precision Recall F1
SemEval 0.841 0.943 0.886
ADE 0.883 0.847 0.864
CausalTimeBank 0.807 0.884 0.842
CauseNet_cause 0.929 0.955 0.942
CauseNet_noncause 0.929 0.952 0.941
MedCaus 0.924 0.924 0.924