Hey dudes and dudettes,

I'm writing my thesis about using GPT for writing investment memo's. My idea is to upload a lot of previous investment memo's on different start-ups as well as info on the new opportunity (a pitch deck). The weird thing is: when I upload only the pitch deck, the answer to my question (write the ... section of an inv. memo) is fine, but when I upload everything, the answer is really bad: it either says it has no info on the new company or it makes up a story about a completely different company.

Any suggestions on how to fix this? Does it have to do with a max_input_size or chunk_size?

I want my app to use both pdf files and pptx as documents. How can I combine them? I know there is this code to accept pptx:

from pathlib import Path from llama_index import download_loader
PptxReader = download_loader("PptxReader")
loader = PptxReader()
documents = loader.load_data(file=Path('./deck.pptx'))

But how do I make sure that documents uses both the normal SimpleDirectoryReader and this reader?