E-FB15k237

This dataset is based on FB15k237 and a pre-trained language-model-based KGE. The main task is to correct the wrong knowledge stored in the pre-trained model and replace the incorrect entities with alternative entities. The model can be downloaded from here.