[Question] Evaluating mGENRE on Mewsli-9

[mrpeerat]

Hi,

I have a quick question on preprocess_mewsli.py line 92. Why do we need to use get_wikidata_ids to obtain entity IDs instead of using the original entity IDs provided by the Mewsli-9 dataset? Are there any ground-truth errors in the Mewsli-9 dataset? 
Thank you.