I’m successfully running config kbqa_cq_bert_ranker.json, which comes with WikiData lite dataset version.
If I want to use same config with latest WikiData dataset from http://www.rdfhdt.org/datasets/. Is it just replacing the .hdt and .hdt.index files in /wikidata_eng, or do I need to retrain Query Prediction, Entity Detection and so on?
May I also modify the download section of the config file to remove the download files of the original dataset, correct? Any other step that you can advice please?
Congratulations on the impressive project you have released.