Hello seniors, i tried a similraty search and its saying this, can anyone please help me solve this problem?i am stuck in ths problem since yesterday,

Support
, ,
1

image
image1784×1171 141 KB

2

Hi @saugatthapa344 and welcome to the Pinecone community forums.

Thanks for your question and for providing a screenshot.

It’s a little difficult to read your screenshot - could you please paste all your relevant code here?

The most common issues preventing upsert when using LangChain are having an improperly formatted vectors variable or having mismatched dimensionality (your embedding model outputs 1536 floating point numbers or vectors, but you set your Pinecone index to 384, for example).

From squinting at your screenshot, it appears that the serialization error is happening within the PineconeVectorStore class and it’s unable to serialize what it expects to be a properly formatted object - so this makes me suspect the data you’re passing in is incorrectly formatted.

I’d recommend:

  1. Adding print statements after every line of code you have so you can verify the format of your data structures
  2. Pasting all your relevant code here for us to review - being careful not to include any secrets like your Pinecone API key

Hope that helps, and looking forward to your response!

Best,
Zack

3

hello @ZacharyProser sir, here is the code, please help

from langchain_community.document_loaders import PyPDFLoader

loader = PyPDFLoader(“/content/Research_of_YOLO_Architecture_Models_in_Book_Detec.pdf”)

pdf_pages = loader.load()

text_splitter = RecursiveCharacterTextSplitter(
# Set a really small chunk size, just to show.
chunk_size=500,
chunk_overlap=20,
length_function=len,
is_separator_regex=False,
)
text_chunks = text_splitter.split_documents(pdf_pages)

import getpass
import os

from getpass import getpass
import os

Get Google API key from environment variable or set it if not present

api_key = os.environ.get(‘GOOGLE_API_KEY’)

if not api_key:
api_key = getpass("Provide your Google API key here: ")
os.environ[“GOOGLE_API_KEY”] = api_key

Print to verify the API key (for debugging purposes only, remove in production)

print(f"Google API key set: {os.environ[‘GOOGLE_API_KEY’]}")

from pinecone import Pinecone

pc = Pinecone(api_key=“XXXXX”)

index = pc.Index(“chatbot”)
import pinecone

index = pinecone.Index(index, host=“https://chatbot-658rjfl.svc.aped-4627-b74a.pinecone.io”)
from langchain_pinecone import PineconeVectorStore
from langchain.vectorstores import Pinecone
from langchain_community.document_loaders import TextLoader

os.environ[
‘PINECONE_API_KEY’] = ‘your api key’

index_name = “chatbot”
embeddings = GoogleGenerativeAIEmbeddings(model=“models/embedding-001”)

docsearch = PineconeVectorStore.from_texts(
[t.page_content for t in text_chunks],
index_name=index_name,
embedding=embeddings
)
docsearch
docsearch.as_retriever()
query= “YOLOv7 outperforms which models?”
docs = docsearch.similarity_search(query)

usr/local/lib/python3.10/dist-packages/pinecone/core/client/api_client.py in sanitize_for_serialization(cls, obj)
286 if isinstance(obj, dict):
287 return {key: cls.sanitize_for_serialization(val) for key, val in obj.items()}
→ 288 raise PineconeApiValueError(‘Unable to prepare type {} for serialization’.format(obj.class.name))
289
290 def deserialize(self, response, response_type, _check_type):

PineconeApiValueError: Unable to prepare type Repeated for serialization

4

@saugatthapa344 as a concerned community member, please do not paste your Pinecone API key in plain text on the forum. I very much recommend this post is removed.

3 Likes
5

@tjensen thank you sir for reminder, sorry for that, can you please help me with the code😅

6

I have edited the post to remove the API key. @saugatthapa344 Please delete that API key and create a new one.

7

i deleted the api @ZacharyProser , can you please help me with the error?