Zhang-Each
commented
1 month ago As mentioned in the paper, we run the baseline models with the parameter setting mentioned in the original papers. You can reproduce their results based on the implementation details of the original papers.
It is a great honor to read your articles "Unleashing the Power of Imbalanced Modality Information for Multi-modal Knowledge Graph Completion" and "NativE: Multi-modal Knowledge Graph Completion in the Wild". We noticed that you have open-sourced the models involved in both articles. Thank you for your contribution to the KGC field. We are very willing to cite your related work.
We encountered problems when reproducing the results of the paper. For NativE and NativE, the relevant parameter settings have been given in git. But for the baseline models for comparison, such as TransAE, RSME, IMF and other models, whether they use the same set of parameter settings as NativE and NativE. This is very important for us to reproduce your experimental results.
Looking forward to your reply.