bealuga

I think that those tools not seeking the spotlight might also be at risk. I'm fairly sure some of these tools have a 'twin' that has found success by securing early adopters and has become the de-facto tool for 'X' job, largely due to marketing.

twitter for a scroll through on video if you can't access: https://twitter.com/coderbea/status/1721748059145880024

Maybe, it was quite slow when I went to it prior to me posting it

I love this oh my god.I didn't know what I was expecting by a "full body keyboard", but I'm sold.

Also want to jump in with https://libraria.dev/ for PDFs too, I just recently updated this with large PDF uploads, you can share your chatbots, have it in different views, and more

hi, try the same with Zappy! As mentioned in another comment, Dumbledore was set up to "enable" GPT: https://ibb.co/8sQh1X6 . I give you the option to "only" use your own documents.

It's at the very bottom of the page! I'll try to make it more visible tomorrow as well (:

That would be really really cool if I could able to serve that space. I'd be curious to know what kind of features you'd want to have, what would be deal-breakers, etc!

good callout - I had it setup to delete immediately after processing, but looks like there was an issue on my end for that! Fixed now and deleted ones that were uploaded before. s3 was storing manual uploads (not data scraped from urls), and should not be storing them anymore.

Yes! I'm currently working on it. I'm trying to improve CSV imports right now (current users have large csvs they want to go through)

Hi! I definitely did not expect this reception, so I've been running into quite a few issues (,:

Send me an email and I'll debug it for you https://libraria.dev/about

Thank you!

- Nothing as of right now! Just an interface. I don't think I implemented snippets yet for Chat mode though - Not yet, although I plan to introduce this. I found that GPT4 is a little slow right now, and much more expensive, but it'll be awesome to enable this for users, then I could charge even less. - I recommend langchain, it's one of the few things that I didn't modify for my use case on their library.

Heh. Looking into this and will try to fix. I'll comment here again when I can get this working!

Will be adding my email at the about page! https://libraria.dev/about (vercel takes a few minutes to deploy, but it should be up soon)

omg. thank you.

:D LOL. The dumbledore example is interesting. I left "GPT" enabled, so it's not just pulling from the documents. I think I'll add settings for users to be able to change the temperature etc. for the assistants in the future as well

Hmm, right now you can kind of simulate that by creating n assistants with only the knowledge you want in them. But yes, no API for now. If there's demand for it, I'll gladly build it though

Thank you! Looking you up right now!

Try the same with "Zappy". It won't give you the same answer!

I have a checkbox for you to enable or disable GPT. I enabled GPT for the Dumbledore one

https://ibb.co/8sQh1X6

Hi! Strange, that google drive integration should work. I'll take a look and update you right here when it does.

Pricing: Feel free to send me an email [email protected], I'll send you some credits to play around with. I'm still trying to figure out pricing!

Yeah, it looks like it actually got the documents correctly, but the prompt itself needed some reworking. I think there's a space out there for prompt testing.

(,: I'm sorry! I'll put it up at the top soon hopefully (I'm hoping to go back to just coding on the app! Spent all day on the landing page today (,: )

I use their APIs, so yes. There's no way around it right now, but hopefully in the future I can move away from relying on them

Yes. If you do try that, let me know! The most difficult part that i'm encountering right now is the "prompt" engineering part. So I'd be curious to know how I can make it better for you

Thank you! Hope you had as much fun as I've been having!

yeah sorry! That's just an image (,: There are other demos a bit below when you scroll down

Yeah more or less! I still use open ai for their embeddings (translating text into vector space)

- Your question -> vectors with open ai embeddings - Text you uploaded before -> vectors with open ai embeddings

Get the most similar above a certain threshold, and then add it to the prompt saying

"From these articles / paragraphs, answer the user's question: What are the three foobars"

So yep! I preprocess it

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!

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)