In the paper, I can see in the tables comparisons made between "K-BERT(Google)" and "K-BERT(Ours)".
Shouldn't we also compare how K-BERT and BERT compete on a downstream task like relation extraction or something? Could not find any tables showing this comparison. Correct me if I missed it in the paper.
In the paper, I can see in the tables comparisons made between "K-BERT(Google)" and "K-BERT(Ours)".
Shouldn't we also compare how K-BERT and BERT compete on a downstream task like relation extraction or something? Could not find any tables showing this comparison. Correct me if I missed it in the paper.
Thanks.