nparsan··on Rd-TableBench – Accurately evaluating table extractionThis is great, but are there datasets for this already? I know pubtables is like 1M labeled data points. Also how important are table schemas as a % of overall unstructured documents?