Datasets:

mteb

/

quora

like 0

The qrels are alrdy there in https://huggingface.co/datasets/mteb/quora/tree/main/qrels no?

Hi @Muennighoff ! The qrels folder doesn't contain the train split. This is also the case for a few other datasets, such as NQ and DBPedia. Could you please clarify if there's anything I might be overlooking? Additionally, is there an officially documented method or boilerplate for consuming mteb datasets in general?

You can look at the code on the GitHub for how one can load them: https://github.com/embeddings-benchmark/mteb
There is no train split afaik

Thanks, @Muennighoff :)