We should consider having Llama Parse as a PDF and doc read option, especially since the table reading accuracy could be significantly improved with it.
Since it's expensive, we need to be strategic about where to apply it. At a minimum only for PDF pages that pass the heuristic, but maybe even better if we restrict it to pages with tables or something. Explore this trade-off
We should consider having Llama Parse as a PDF and doc read option, especially since the table reading accuracy could be significantly improved with it.
Since it's expensive, we need to be strategic about where to apply it. At a minimum only for PDF pages that pass the heuristic, but maybe even better if we restrict it to pages with tables or something. Explore this trade-off