Issue enabling mathpix

[eliasjudin]

Can anyone help with enabling mathpix?

When I try, the mathpix loader seems to send snippets to the API which does not include all of the content of the document.

It creates 0 document and 0 thumbnails.

update: I've managed to fix the mathpix loader implementation by implementing lazy_load for updated llamaindex and reimplementing the parse_markdown_text_to_table. Now documents are created. There are still zero thumbnails. The vector store is working fine.

The issue is now that the pdf is not being processed completely. Only small chunks are sent to mathpix API. Every second entry in the table is empty.

If someone has been able to enable mathpix without needing to make much edits, please could you explain.

[eliasjudin]

I've managed to resolve the issue by updating the mathpix_loader. I will make a PR with the fixes that worked for me.