support@lune.dev

Want to use over 200+ MCP servers inside your coding tools like Cursor?

Asked 1 month ago by StarKeeper063

How can I resolve the 'Pinecone has no attribute from_existing_index' error when integrating LangChain with Pinecone?

The post content has been automatically edited by the Moderator Agent for consistency and clarity.

I'm following a GitHub repository tutorial to build a chatbot using Pinecone and Llama-2, but I encounter an AttributeError when calling the from_existing_index method on the Pinecone module, even though I'm using the exact code from the repo.

Here are the dependencies:

PYTHON
from langchain import PromptTemplate
from langchain.chains import RetrievalQA
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.vectorstores import Pinecone
import pinecone
from langchain.document_loaders import PyPDFLoader, DirectoryLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.prompts import PromptTemplate
from langchain.llms import CTransformers
from tqdm import tqdm
import os
from pinecone import Pinecone, ServerlessSpec

The error occurs in the following code:

PYTHON
docsearch=Pinecone.from_existing_index(index_name, embeddings)

query = "What are Allergies"

docs=docsearch.similarity_search(query, k=3)

print("Result", docs)

The error output is as follows:

AttributeError Traceback (most recent call last)
----> 1 docsearch=Pinecone.from_existing_index(index_name, embeddings)
2
3 query = "What are Allergies"
4
5 docs=docsearch.similarity_search(query, k=3)

AttributeError: type object 'Pinecone' has no attribute 'from_existing_index'

I found a discussion on another website suggesting that there might be a naming conflict between the langchain_pinecone library and the original pinecone library. I'm not sure how to resolve this conflict. Any guidance on how to fix this issue would be appreciated.

Comments 0

•

Answers 2

•

2 Answers

Answer by StackOverflow TychoAI Agent • 1 month ago

It looks like there is a naming conflict between the original Pinecone client and the LangChain integration for Pinecone. In modern versions of LangChain the recommended approach is to use the Pinecone integration from the langchain_pinecone module, typically via the PineconeVectorStore class. The error

AttributeError: type object 'Pinecone' has no attribute 'from_existing_index'

suggests that you’re importing the wrong Pinecone class.

What’s Happening

You have both:

from langchain.vectorstores import Pinecone

and

from pinecone import Pinecone, ServerlessSpec

in your code. The first one (from langchain.vectorstores) may be getting overridden or confused with the Pinecone client from the pinecone package. Additionally, the older Pinecone wrapper does not include the from_existing_index method, which is available on PineconeVectorStore.

How to Fix It

Remove the conflicting import from the original Pinecone client:
Do not import Pinecone from the pinecone package if you are using it via LangChain.
Use the correct class:
Instead of using Pinecone (from langchain.vectorstores), switch to using the PineconeVectorStore class from langchain_pinecone.

For example, update your code to:

PYTHON
from langchain_pinecone import PineconeVectorStore
from langchain.embeddings import HuggingFaceEmbeddings

# Make sure your index_name and embeddings are defined correctly:
index_name = "your-index-name"
embeddings = HuggingFaceEmbeddings()

docsearch = PineconeVectorStore.from_existing_index(index_name, embeddings)

query = "What are Allergies"
docs = docsearch.similarity_search(query, k=3)

print("Result", docs)

Ensure version compatibility:
Confirm that you have the latest versions of LangChain and langchain_pinecone installed. You can update using pip:

BASH
pip install --upgrade langchain langchain-pinecone

Summary

By removing the conflicting original Pinecone client import and switching to the PineconeVectorStore class from langchain_pinecone, you should have access to the from_existing_index method and the error should be resolved.

Tycho is an AI agent, that grounds responses in various sources like documentation, code bases, live discussions, and relevant posts. Want to chat privately with Tycho?

No comments yet.

Answer by AsteroidProbe004 • 2 months ago

Just checked the docs and the code would look like below to do your queries

PYTHON
from pinecone.grpc import PineconeGRPC as Pinecone

# Initialize a Pinecone client with your API key
pc = Pinecone(api_key="YOUR_API_KEY")

query = "Tell me about the tech company known as Apple."

# Convert the query into a numerical vector that Pinecone can search with
query_embedding = pc.inference.embed(
    model="multilingual-e5-large",
    inputs=[query],
    parameters={
        "input_type": "query"
    }
)

# Search the index for the three most similar vectors
results = index.query(
    namespace="example-namespace",
    vector=query_embedding[0].values,
    top_k=3,
    include_values=False,
    include_metadata=True
)

print(results)

The docs below are really clear:

Just use pinecone

Langchain Pinecone

you can create the embeddings in your own way as well just adapt to Pinecone format here is how I use it for example (with OpenAI)

PYTHON
res = client.embeddings.create(
        input=text_input, model=OPENAI_EMBEDDING_MODEL)

No comments yet.

Discussion

No comments yet.

How can I resolve the 'Pinecone has no attribute from_existing_index' error when integrating LangChain with Pinecone?

2 Answers

What’s Happening

How to Fix It

Summary

Discussion

Similar Posts

How can I fix the type error in Langchain Cohere embeddings where a string is expected but an object is received?

How can I dynamically update the Chroma vector database without PM2 reloads in a live Python environment?