Closed joshdevins closed 3 years ago
Hi @joshdevins,
Yes, CQADupstack contains 12 StackExchange forums and you should process each category separately. The scores in the paper are the average of the individual forum scores. Also see: https://github.com/UKPLab/beir/issues/9
Kind Regards, Nandan
I'm using evaluate_anserini_docT5query.py. In that script it's using
util.download_and_unzip
thenGenericDataLoader
and fails with the following message when using datasetcqadupstack
:The CQADupstack dataset is divided up into sub-categories which is causing the above error since it contains an extra sub-directory per-category.
How was CQADupstack used in the benchmarking for the paper and leaderboard? Was each category processed separately or was everything somehow combined into a single evaluation?