Hey, thanks for the reply. I think so, even Llmsherpa.
Anyway I'll try again, maybe I made a mistake.
In any case, in the end I had opted for specific readers for SimpleDirectoryReader, in my case PyMuPdf and MD reader extractors; and for chunking: for PyMuPdf I used semantic chunking and for MD the MarkdownNodeParser. Do you think it might be a better choice or at least worth trying for smart-pdf-loader (obviously for PDFs )?