Run the official tutorial and the kernel dies

[Barsit]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

import bs4
from langchain_community.embeddings import OllamaEmbeddings
from langchain import hub
from langchain_chroma import Chroma
from langchain_community.vectorstores import FAISS
from langchain_community.document_loaders import WebBaseLoader
from langchain_core.output_parsers import StrOutputParser
from langchain_core.runnables import RunnablePassthrough
from langchain_openai import OpenAIEmbeddings
from langchain_text_splitters import RecursiveCharacterTextSplitter

# Load, chunk and index the contents of the blog.
loader = WebBaseLoader(
    web_paths=("https://lilianweng.github.io/posts/2023-06-23-agent/",),
    bs_kwargs=dict(
        parse_only=bs4.SoupStrainer(
            class_=("post-content", "post-title", "post-header")
        )
    ),
)
docs = loader.load()

text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
splits = text_splitter.split_documents(docs)

vectorstore = Chroma.from_documents(documents=splits, embedding=OpenAIEmbeddings(base_url="https://xxxx/v1"))

# Retrieve and generate using the relevant snippets of the blog.
retriever = vectorstore.as_retriever()
prompt = hub.pull("rlm/rag-prompt")


def format_docs(docs):
    return "\n\n".join(doc.page_content for doc in docs)


rag_chain = (
    {"context": retriever | format_docs, "question": RunnablePassthrough()}
    | prompt
    | llm
    | StrOutputParser()
)

rag_chain.invoke("What is Task Decomposition?")

Description

When I ran the official tutorial Vector stores and retrievers（https://python.langchain.com/v0.2/docs/tutorials/retrievers/）, the jupyter notebook kernel died and restarted with no error stack information

System Info

Kernel Restarting The kernel for 03.ipynb appears to have died. It will restart automatically.

[dosubot[bot]]

Hey there, @Barsit! I'm here to help you out with any bugs, questions, or contributions you have. Let's tackle this together while we wait for a human maintainer.

There are a few potential reasons why the Jupyter notebook kernel might die when running the provided code. Here are some steps to help debug and mitigate the issue:

1. Memory and Resource Limitations: The operations involved in loading, chunking, and indexing large documents can be resource-intensive. Ensure your system has sufficient memory and CPU resources. You might want to monitor your system's resource usage while running the notebook.

2. Verbose Logging: Enable verbose logging to get more insights into where the issue might be occurring. You can use the set_debug function from LangChain to enable detailed logs:

   from langchain.globals import set_debug
   
   # Enable debug mode
   set_debug(True)
   
   # Your existing code
   from langchain.agents import AgentType, initialize_agent, load_tools
   from langchain_openai import ChatOpenAI
   
   llm = ChatOpenAI(model_name="gpt-4", temperature=0)
   tools = load_tools(["ddg-search", "llm-math"], llm=llm)
   agent = initialize_agent(tools, llm, agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION)
   
   # Run the agent
   agent.run("Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?")

3. Chunk Size and Overlap: Adjust the chunk_size and chunk_overlap parameters in the RecursiveCharacterTextSplitter to see if smaller chunks help reduce memory usage:

   text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=100)
   splits = text_splitter.split_documents(docs)

4. Alternative Vector Stores: Consider using alternative vector stores that might be more efficient in terms of resource usage. Some alternatives include Pinecone, Qdrant, Milvus, Weaviate, Redis, DeepLake, ElasticsearchStoreCommunity, and MongoDBAtlasVectorSearch ^[1][2].

5. Optimize OpenAIEmbeddings Parameters: Adjust the parameters of the OpenAIEmbeddings class to optimize performance and reduce resource usage. For example, you can reduce the chunk_size and set appropriate timeouts:

   from langchain_openai import OpenAIEmbeddings
   
   embeddings = OpenAIEmbeddings(
       base_url="https://xxxx/v1",
       chunk_size=500,  # Reduce chunk size
       request_timeout=10,  # Set a timeout for requests
       show_progress_bar=False  # Disable progress bar to reduce overhead
   )

By following these steps, you should be able to identify and mitigate the issues causing the Jupyter notebook kernel to die.

To continue talking to Dosu, mention @dosu.

[Barsit]

这是来自QQ邮箱的自动回复邮件。您的邮件我已收到。

[Barsit]

FAISS vector database runs successfully instead of chroma vector database