ChatPDF – Chat with Any PDF • Hacker News Board

Re: ChatPDF – Chat with Any PDF

#121

Post by taneq » Wed, Apr 19, 2023, 11:47 PM UTC

stalfosknight wrote: post #78

I am getting really sick and tired of the constantly pearl-clutchy "I'm just an AI and I have to be neutral and I can't say anything remotely spicy" bs that AIs seem to be programmed to repeat with almost every single answer.

It's all that RLHF alignment training they did after the first couple of releases were happy to (with the right prompt) tell you how to kill yourself, make meth or commit the perfect murder. I don't like it either, but I can understand why they did it. All it takes is one spectacularly awful neo-nazi spiel generated by your product to set off a moral panic and sink your public image, no matter how much you try to explain that it only said that because the user told it to do so.

Re: ChatPDF – Chat with Any PDF

#122

Post by gr4d » Wed, Apr 19, 2023, 11:51 PM UTC

m3kw9 wrote: post #111

Someone posted code to the exact same thing yesterday on twitter which uses langchain with Chromadb as vector store and Streamlit as front end. Like 30 lines of code.

Can you share the link?

Re: ChatPDF – Chat with Any PDF

#123

Post by stalfosknight » Thu, Apr 20, 2023, 12:29 AM UTC

taneq wrote: post #121

stalfosknight wrote: post #78

I am getting really sick and tired of the constantly pearl-clutchy "I'm just an AI and I have to be neutral and I can't say anything remotely spicy" bs that AIs seem to be programmed to repeat with almost every single answer.

It's all that RLHF alignment training they did after the first couple of releases were happy to (with the right prompt) tell you how to kill yourself, make meth or commit the perfect murder. I don't like it either, but I can understand why they did it. All it takes is one spectacularly awful neo-nazi spiel generated by your product to set off a moral panic and sink your public image, no matter how much you try to exp…

It’s a piece of software guessing what text to spit out in a web browser. People need to get over themselves.

Re: ChatPDF – Chat with Any PDF

#124

Post by taneq » Thu, Apr 20, 2023, 2:02 AM UTC

dfaslkjvalkj wrote: post #88

glonq wrote: post #60

Other possibilities to fuel the ChatGPT hype train... ChatPNG - apply OCR to an image, extract text, feed it to GPT. ChatMP3 - apply speech-to-text to a recording, feed it to GPT. ChatGPS - hmm. not sure yet. something location-based obviously... If any VC's are interested, I'm selling 10% stake in these projects for only $20k right now. /s

Forget that I'm currently selling NFTs for these projects-- ChatTXT - extract text from plain text files and feed it to GPT for analysis. ChatPDF - extract text from a PDF document and feed it to GPT for analysis. ChatDOC - extract text from a Microsoft Word document and feed it to GPT for analysis. ChatDOCX - extract text from a Microsoft Word document and feed it to GPT for analysis. ChatPPT - extract text from a M…

Oh, I'm afraid your tokens will be quite fungible.

Re: ChatPDF – Chat with Any PDF

#125

Post by taneq » Thu, Apr 20, 2023, 2:02 AM UTC

stalfosknight wrote: post #123

taneq wrote: post #121

Earlier quoted context omitted.

It's all that RLHF alignment training they did after the first couple of releases were happy to (with the right prompt) tell you how to kill yourself, make meth or commit the perfect murder. I don't like it either, but I can understand why they did it. All it takes is one spectacularly awful neo-nazi spiel generated by your product to set off a moral panic and sink your public image, no matter how much you try to exp…

It’s a piece of software guessing what text to spit out in a web browser. People need to get over themselves.

Is this something that people, in your experience, do?

Re: ChatPDF – Chat with Any PDF

#126

Post by foobiekr » Thu, Apr 20, 2023, 3:15 AM UTC

Was going to sign up but the lack of transparency about how to cancel is pretty obnoxious.

Re: ChatPDF – Chat with Any PDF

#127

Post by ukuina » Thu, Apr 20, 2023, 7:18 AM UTC

ayanb9440 wrote: post #69

bvan wrote: post #6

Great, but I would have to run this in >my< private cloud. no way any business is going to upload its docs into a third-party cloud, no matter what the small print says.

Here is a fully self-hostable solution that connects to PDFs in your google drive folder: https://github.com/ai-sidekick/sidekick Uses weaviate so that even the vectorstore can be self-hosted

Does Sidekick use OpenAI, or are local models supported?

Re: ChatPDF – Chat with Any PDF

#128

Post by ukuina » Thu, Apr 20, 2023, 7:27 AM UTC

EGreg wrote: post #12

(I fed it this PDF: https://arxiv.org/pdf/0802.0832v1.pdf ) And got the same experience I often have with LLMs: How many validators are enough to make the scheme secure According to page 5 of the PDF, the system is secure with security parameters if the adversary must perform an expected O(2^s) amount of work in order to be successful. The number of validators needed to achieve this level of security is not explicitl…

If only we could get a penny every time GPT apologizes.

Re: ChatPDF – Chat with Any PDF

#129

Post by juliushabasa » Thu, Apr 20, 2023, 8:58 AM UTC

Its good app

Re: ChatPDF – Chat with Any PDF

#130

Post by pfcao » Thu, Apr 20, 2023, 11:18 AM UTC

IKLOL wrote: post #3

How much is it actually using the PDF and how much is just normal Chat GPT knowledge? I uploaded a KJV bible and it seems to be doing pretty good with theological issues, like it knows salvation is by grace through faith alone which is my litmus test for any theological program. However, it seems to be just as honed in as Chat GPT is without even uploading a PDF.

I think maybe it has the data of the bible (and some theology and religion knowledge) already during its training. And these are the background for it to handle other texts. We should test it with some unique information, preferably different from common sense.