Hacker News Reader

Comment by freedomben | Hacker News Reader

a one-page PDF, sure. But if it's a 500 page pdf of a law and/or regulation, then definitely not.

How can you chat with a pdf which doesn’t fit in the context window? I mean with a 500 page pdf you might need 100 context windows to fully grok it.

Basically it makes no sense to “chat” with a 500 page pdf with todays LLMs.

That is what the RAG system does. The PDF is chunked and thrown into a vector store. And then when prompted, only the relevant bits are retrieved and stuffed into the context and sent to the LLM.

So yeah it's kinda smoke and mirrors. In some cases, for some long PDFs, it works really well. If it's a 500 page PDF with many disparate topics, it may do fine.

Indeed. Would only add, context windows are continually multiplying in size. Who knows how long Moore's Law will apply here, but it's a continually improving window.

I've found that the longer context windows don't seem to be a linear improvement in responses though. It's like the longer the context window, the quality of the response is perhaps broader, but less sharp or accurate. I've been using GPT4-turbo with the longer context window for coding tasks but it doesn't seem to have improved the responses as much as you would think, it seems to be more "distracted" now, which perhaps makes some intuitive sense.

I can give gpt4-turbo many full code files to try and solve a complex coding task but despite the larger window it seems to fail more often or ignore parts of the context window or just doesn't really answer the question.

That assumes that only one part of the PDF, which fits in the context window, is relevant to the prompt, which seems like a fairly big assumption.

You can screenshot the first page and use gpt vision

You could just load up the doc, take first 1024 tokens, and almost always get the right authors/title/year, etc, assuming its there.

But going further, for large bills you might need (|n|..|m|) pages to capture full index

for research papers you also want to look at last (|n2|..|m2|) pages for bibliography, etc..

Reply on news.ycombinator.com