Orchestrating LLMs with LangChain: Challenge Lab

[rash6766]

Hi Team,

When I am trying to execute task 3 of this lab, i am getting this error. RuntimeError: no validator found for <class 'langchain_core.prompts.base.BasePromptTemplate'>, see arbitrary_types_allowed in Config

Task 1 and 2 are completed successfully.

[Dora-2002]

i am unable to complet task 4

[nichsedge]

Try this; it works as of 2025-05-16

# Task 4: Text Summarization

from langchain.chains.summarize import load_summarize_chain
from langchain.prompts import PromptTemplate
# Assuming 'model' is the ChatVertexAI model loaded in Task 3
# Assuming 'pages' is the list of document pages loaded in Task 3

# Define the prompt template
prompt_template = """Write a concise summary of the following text delimited by triple backquotes.
              Return your response in bullet points which covers the key points of the text.
              ```{text}```
              BULLET POINT SUMMARY:
  """

prompt = PromptTemplate(template=prompt_template, input_variables=["text"])

# Set up the summarization chain using the "stuff" method
# The model used here is the one loaded in Task 3
summarize_chain = load_summarize_chain(llm=model, chain_type="stuff", prompt=prompt)

# Take the first three pages of the document as input text
# The 'pages' variable comes from Task 3
input_documents = pages[:3]

# Print the content of the first three pages
print("--- Content of the first three pages ---")
for i, doc in enumerate(input_documents):
    print(f"--- Page {i+1} Content ---")
    print(doc.page_content)
    print("\n") # Add a newline for better separation

# Apply the summarization chain
summary_output = summarize_chain.invoke(input_documents) # LangChain typically returns a dict

# Print the generated summary
print("--- Summary of the first three pages (using stuff method) ---")
# The actual summary is usually in the 'output_text' key
print(summary_output['output_text'])

[a-berg]

I downloaded the pdf using both curl and urllib, saved in ~/data to no avail. Check 3 failed and could not complete.