fraserxu / book-gpt

Drop a book, start asking question.

Hi there 👋

Hi this is Fraser.

I'm a web developer from China, now living and working in Melbourne. I'm currently working as a lead engineer for Envato.

The main project I'm working on now is Envato Elements, before that I also worked on Milkshake and Envato Market.

book-gpt's Issues

Request failed with status code 429

What about this error?

something went wrong

it would help me if this error was actually returned in the client - rather than having to debug app.
'PineconeClient: Error calling upsert: PineconeClient: Error calling upsertRaw: FetchError: The request failed and the interceptors did not return an alternative response '

error - pineconeclient project name not set

https://book-gpt.vercel.app/credentials
i did the instructions - pine cone boots up -

Mobile friendly

Makes the chat window full width

Any plans to refactor this to use the new Langchain retriever structure?

This replaces a few steps that were previously unstandardized, this blog post explains: https://blog.langchain.dev/retrieval/

UX - add - Success states

Right now there is no way for the user to know when a PDF is uploaded or credentials updated.
Have a settings icon which takes in API credentials.
Only after the user uploads a pdf/txt, ("PDF successfully uploaded") enable the chat, else it is confusing.

here is a quick scribble as a starting point

In Future, maybe let the user add different books and have different chatbots interacting with them. Unique URL which. the user can bookmark for each (protected. by auth)

Cost estimate

Estimate the cost of processing the pdf file with openai and pinecone, confirm it with the user and then start to digest it to prevent the costs going to the roof.
thank you

Request failed with status code 429 - Has this Error 429 Been Fixed?

fraserxu,
I followed the instructions to set up the account and tried to upload a PDF and received this error message. Have you determined the cause for the error. As far as Pinecone AI issues, it appears in the example that you provided that you were using a Free Account with an expiry date 14 days later, Mine was a 7 day free trial but it shouldn;t ake a difference unless you have tested your app while utilizing a Free Pinecone AI Starter Membership.
Regardless, any feedback would be greatly appreciated as this is a service that is very unique and it would be great to be able to utilize it!

Thank you for your efforts, they are greatly appreciated.

Database with previously uploaded files

Would love to see a db implemented to save previous docs that are uploaded.

Environment Variables for Devs

Would like to play around with this repo on local. Perhaps a little how-to on implementing environment variables to set the API keys over the user adding them.

Pinecone environment is hardcoded and users based out of West Coast cannot use the app

The problem is that in Pinecone I couldn't find an option to pick environment and every index I create has "us-east-1-aws" environment. So the application throws an error:

PineconeClient: Error getting project name: SyntaxError: Unexpected token 'A', "API key is"... is not valid JSON

To fix this problem I moved Pinecone environment to credentials (the same way you do with API keys). How can I create a pull request with this fix?

Multiple model support

I want to be able to select between GPT 3.5 and 4.

Maybe you can add it to the Credentials part.

Upload multiple files

I'd love to have this feature. I have a lot of little files I'd like to upload.

I know it's more work than changing multiple: true here:

book-gpt/pages/index.tsx

Line 58 in 644d3f1

multiple: false,

GPT-4 support please

Needs updating langchain and maybe changing some parameters in the openai API? I couldn't find it anywhere