search

mayooear / gpt4-pdf-chatbot-langchain

GPT4 & LangChain Chatbot for large PDF docs
https://www.youtube.com/watch?v=ih9PBGVVOO4
14.92k stars 3.02k forks source link

update the latest pinecone version to the latest version with support for Serverless indexes (the current only option for free pinecone accounts) #473

Open dmotxilla opened 3 months ago

dmotxilla commented 3 months ago

In the new version of pinecone, the capacity mode of an index can be serverless or pod-base (before it was only pod-base). The free account of pinecone currently only allows the serverless mode.

In the serverless mode, there is no an environtment parameter; the PINECONE_ANVIRONMENT parameter does not make sense anymore. The pinecone-python lib needs to be updated to the latest version.

Ref: I asked Pinecone about it: https://community.pinecone.io/t/how-to-find-out-what-is-my-environment-name-in-a-free-pinecone-account/6253

where they mention that:

It looks like that code is using the older connection method which is not supported with Serverless indexes.

gpt4-pdf-chatbot-langchain needs to be updated to use the latest version of the NodeJS SDK.

ajd2 commented 3 months ago

im getting the same issue. Does this basically mean can't be used until it is updated? :(

tonymynd commented 2 months ago

This app was extremely useful to understand complex topics on different kinds of pdf documents. Perhaps we can change the connection method to reflect the new modality, if possible to keep it free.

arpeiks commented 2 months ago

I was able to fixe the issue by using the individual Langchain packages

import { PineconeStore } from "@langchain/pinecone";
import { Document } from "@langchain/core/documents";
import { OpenAIEmbeddings } from "@langchain/openai";
import { Pinecone } from "@pinecone-database/pinecone";

Check out a sample implementation https://github.com/arpeiks/gpt-langchain

tonymynd commented 3 weeks ago

I was able to fixe the issue by using the individual Langchain packages

import { PineconeStore } from "@langchain/pinecone";
import { Document } from "@langchain/core/documents";
import { OpenAIEmbeddings } from "@langchain/openai";
import { Pinecone } from "@pinecone-database/pinecone";

Check out a sample implementation https://github.com/arpeiks/gpt-langchain

Thanks @arpeiks, I went into your fork for this project. I'm thinking, shall I clone this files and replace them into this project?