Live data from Hacker News

ScholarTurbo: Use ChatGPT to chat with PDFs (supports GPT-4)

scholarturbo.com

41–50 of 70 posts

Re: ScholarTurbo: Use ChatGPT to chat with PDFs (supports GPT-4)

#42

Earlier quoted context omitted.

Why do you consider this cheating? The context window is currently too small for it to work in any other manner.

You could ask Claude+ 100k to give you the text context(s) containing anything pertinent then have GPT4 prompted with that.

With an emphasis on the word “could”! Very few people have access to Anthropic’s services at this point!

Embeddings still seem like a more economical approach.

Re: ScholarTurbo: Use ChatGPT to chat with PDFs (supports GPT-4)

#43
Cool product! seamless to try out and works quite well. But I am not able to ask questions outside the papers but relevant to it. "I'm sorry, but the given context does not provide information on the advantages of using quaternions over Euler angles in general. My programming is limited to answering questions related to the given context. Is there anything else I can help you with?"

Re: ScholarTurbo: Use ChatGPT to chat with PDFs (supports GPT-4)

#44
post #13

I've tried this and chatpdf with some Computer Aided Geometric Design mathematics theses and papers. These tools do not seem to "read" the math parts of the papers, only the text. It would be amazing to have something which could digest a bunch of papers and start to help me navigate them, but a lot of the answers I get are "I don't have enough context to help you". Is there a better tool or way to use LLMs with scho…

You could just use the latex source code of the paper in the context window if you want it to be aware of the math. You can download latex source for any paper or arxiv. If you have many papers, perhaps upload the latex source code of all of them into some vector database like pinecone and use them with LangChain for retrieval.

this is a great idea, totally doable with what is out there

Re: ScholarTurbo: Use ChatGPT to chat with PDFs (supports GPT-4)

#46

Cool product! seamless to try out and works quite well. But I am not able to ask questions outside the papers but relevant to it. "I'm sorry, but the given context does not provide information on the advantages of using quaternions over Euler angles in general. My programming is limited to answering questions related to the given context. Is there anything else I can help you with?"

Isn't that kind of the point? I would greatly prefer it stays within the confines of the supplied documents.

Re: ScholarTurbo: Use ChatGPT to chat with PDFs (supports GPT-4)

#47
post #40

It's pretty disgusting to see how many of these exist, yet so few are local-only with local LMs. It's exceedingly obvious that local LMs are desired and far better for this task, and only idiots would be interested in paying for software like this that uses OpenAI.

A very healthy portion of paid software fits in the category of "things you could do on your own machine, but would rather pay someone to run on their machine with a simplified UI".

This doesn't make the users idiots, it just means they value their time.

Re: ScholarTurbo: Use ChatGPT to chat with PDFs (supports GPT-4)

#48

I think these GPT wrapper apps are not too unlike Dropbox. They face a good amount of (fair) criticism as too simple of a solution for any technically minded person. But if you wrap some of the annoying parts into a nice app, you could have a very successful business for non technical people. Even technical enterprises can benefit from enterprise level security features that they themselves may not want to deal with.

I think Dropbox won on doing something well when anything similar really sucked, and I think a lot of that was in making the syncing 'just work'. A lot of online cloud drives were (and still are) untenable because of the babysitting and frustrating as files don't sync or are slow to access.

With all these ChatGPT wrappers, the hard bit is pretty much a monopolized commodity, so they better do the annoying bits really really well.

This is the second PDF chatbot I have tried, and they both suffer from the technical constraint of how many tokens you can fit into ChatGPT, meaning they refuse to answer global questions like "are there any mistakes in the document since the ChatGPT may be unaware that there is a document at all, but can answer, for example for a CV, "what jobs might you hire this person for".

This might be fixable by making round trips, for example asking chatgpt YES/NO for does this query apply to this 1000 token section of a larger document, etc. etc.

And of course if they get this right, and do it well, it becomes very useful. If it can answer based on 10000 stored pdfs, word docs, JIRA tickets, confluence pages and slack threads, then it would be super useful for organizations. It would hardly be just a wrapper then though.

Post reply on HN