PDF ingestion does not seem to complete successfully

[tprennie]

I have attempted to load two different PDFS into PDFtoCHAT and neither appears to complete the ingestion process.

After 45 minutes I close the browser tab, reopen and the PDF is listed in my list.

When I click on them, but when I attempt to chat with them the response from PDFtoCHAT indicates it has no knowledge of the PDF document I am attempting to load.

What am I doing wrong? I was part of your Zoom presentation earlier today and it looks like a terrific capability!

Thanks,

Tim

[nicolasRossard]

I got the same issue on all pdf I tried to load.

[Koutilya-Pande]

I tried uploading some PDFs as well, they are working fine for the PDFs that are in the standard form with a white background. My friend made a Data Science preparation PDF, it had multiple coloured backgrounds, and it was unable to detect the content.

My other concern is this, when I asked it questions about PDFs, it generated answers from the common knowledge domain, rather than stating that it was unable to read the PDF. I think rather than giving generalised answers, just to perform chat completion it should restrict its knowledge domain to the PDF.

[sthakulla]

I am seeing similar issue. I uploaded the pdf but it just buffers. When refreshing the page, the pdf seemed to have already loaded.

[GrgMdmn]

I can confirm what @sthakulla is saying : the error is "only" visual. Refresh the page then the document will appear.