Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: Customizable, embeddable Chat GPT based on your own documents

libraria.dev

Post reply on HN
Print view View original

1–10 of 135 posts

Show HN: Customizable, embeddable Chat GPT based on your own documents

#1

Post by bealuga »

Hi Hacker News!

My name is Bea, I built a site called Libraria that uses GPT to do a few things

1. Let you spin up multiple assistants based on your own documents. You can make it public, private, or protected. It has its own subdomain and landing page. 2. Respond in full markdown always, so it can output images, links, code, and more 3. Let you upload articles on the fly within the Chat, so you can ask it questions 4. Make it embeddable in your site with one line of code 5. Let you update it for fun / with your branding 5. Enable syncing for any URLs you let us scrape, so that you can make sure it's always up to date 6. Let you upload multiple file types

I've been working on this for about a month now by myself and you can keep track of my feature updates here: https://libraria.dev/feature-updates

I would LOVE your feedback on anything, and If you're willing to try it out I'm looking for a few beta users that can provide me more continuous feedback that I would gladly waive the fee for!

Show HN: Customizable, embeddable Chat GPT based on your own documents
libraria.dev

Re: Show HN: Customizable, embeddable Chat GPT based on your own documents

#4

Post by bealuga »

hhw3h wrote: post #3

For the Dumbledore demo I asked, "What are the 3 unforgivable curses?" and it was unable to provide an answer. https://ibb.co/bL61nDk Would your tool work in gated knowledge bases for example training courses behind logins?

Oh no! Let me manuially add the 3 unforgivable curses to the library. The way I did the dumbledore demo was upload a PDF for each of the 7 books and there ~might~ not be an explicit section in harry potter that states all of them at once?

Re: Show HN: Customizable, embeddable Chat GPT based on your own documents

#5

Post by bealuga »

hhw3h wrote: post #3

For the Dumbledore demo I asked, "What are the 3 unforgivable curses?" and it was unable to provide an answer. https://ibb.co/bL61nDk Would your tool work in gated knowledge bases for example training courses behind logins?

As for working with gated knowledge bases, you can. You can just set your assistant as "protected".

https://bogpad.libraria.dev/ for an example of a protected one.

https://ibb.co/T8qjCgZ (strange, i tried again and got this! I'll work on improving stability)

Re: Show HN: Customizable, embeddable Chat GPT based on your own documents

#6

Post by Waterluvian »

bealuga wrote: post #4
hhw3h wrote: post #3

For the Dumbledore demo I asked, "What are the 3 unforgivable curses?" and it was unable to provide an answer. https://ibb.co/bL61nDk Would your tool work in gated knowledge bases for example training courses behind logins?

Oh no! Let me manuially add the 3 unforgivable curses to the library. The way I did the dumbledore demo was upload a PDF for each of the 7 books and there ~might~ not be an explicit section in harry potter that states all of them at once?

> an explicit section in harry potter that states all of them at once?

Isn’t part of the point of GPT that it finds relationships without the training data having to be well-structured? So long as the text describes them in a way that a human, having read the books, could answer the question?

I don’t really understand how training works. This isn’t a jab.

Re: Show HN: Customizable, embeddable Chat GPT based on your own documents

#7

Post by bealuga »

Waterluvian wrote: post #6
bealuga wrote: post #4

Earlier quoted context omitted.

Oh no! Let me manuially add the 3 unforgivable curses to the library. The way I did the dumbledore demo was upload a PDF for each of the 7 books and there ~might~ not be an explicit section in harry potter that states all of them at once?

> an explicit section in harry potter that states all of them at once? Isn’t part of the point of GPT that it finds relationships without the training data having to be well-structured? So long as the text describes them in a way that a human, having read the books, could answer the question? I don’t really understand how training works. This isn’t a jab.

It's a great question!

Right now, I don't train anything, I've broken the text down to n characters and created embeddings for that subset of text -- then I search for the closest distance / relationships between the question asked. Then I add the text to the prompt, and tell gpt to use those paragraphs to answer the question (to ensure that it doesn't make anything up). This is one of the ways I can get around the token limit, but it comes at the cost of thinking it can only use the paragraphs I show it. I'm trying to improve the prompt to get more consistent results, and maybe 4 can help me give it larger bodies of text!

Hope that answers your question, let me know if you have any more!

Re: Show HN: Customizable, embeddable Chat GPT based on your own documents

#8

Post by Waterluvian »

bealuga wrote: post #7
Waterluvian wrote: post #6

Earlier quoted context omitted.

> an explicit section in harry potter that states all of them at once? Isn’t part of the point of GPT that it finds relationships without the training data having to be well-structured? So long as the text describes them in a way that a human, having read the books, could answer the question? I don’t really understand how training works. This isn’t a jab.

It's a great question! Right now, I don't train anything, I've broken the text down to n characters and created embeddings for that subset of text -- then I search for the closest distance / relationships between the question asked. Then I add the text to the prompt, and tell gpt to use those paragraphs to answer the question (to ensure that it doesn't make anything up). This is one of the ways I can get around the t…

Oh! I think I understand.

So your software takes a prompt from me, does non-GPT work to find additional context from your source (the books, parsed and re-structured into word vectors or whatnot), and then asks GPT my prompt combined with the added context?

Like,

“What are the three foobars when considering these passages from a book ?”

Post reply on HN