new-chain

New Chain Command

Create a new LCEL chain file for common patterns.

Workflow

1. Ask the user which chain type they want:

   • summarization - Summarize documents or text
   • extraction - Extract structured data
   • classification - Classify text into categories
   • translation - Translate between languages

2. Get filename from argument or ask user (default: chain.py)

3. Generate the chain file using the appropriate template below

4. Inform user about customization options

Templates

Summarization Template

"""
Summarization Chain

Install: pip install langchain langchain-openai
"""
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_openai import ChatOpenAI


def create_summarization_chain(
    model: str = "gpt-4o",
    style: str = "concise",
    max_points: int = 5,
):
    """
    Create a summarization chain.

    Args:
        model: The model to use
        style: Summary style (concise, detailed, bullet_points)
        max_points: Maximum bullet points for bullet_points style
    """
    style_instructions = {
        "concise": "Provide a brief, 2-3 sentence summary.",
        "detailed": "Provide a comprehensive summary covering all main points.",
        "bullet_points": f"Provide a summary as {max_points} bullet points.",
    }

    prompt = ChatPromptTemplate.from_template("""
Summarize the following text.

{style_instruction}

Text:
{text}

Summary:
""")

    llm = ChatOpenAI(model=model, temperature=0)

    chain = (
        prompt.partial(style_instruction=style_instructions.get(style, style_instructions["concise"]))
        | llm
        | StrOutputParser()
    )

    return chain


def main():
    # Create chain
    chain = create_summarization_chain(style="bullet_points", max_points=5)

    # Example usage
    text = """
    Artificial intelligence (AI) is transforming industries across the globe.
    From healthcare to finance, AI applications are improving efficiency and
    enabling new capabilities. Machine learning, a subset of AI, allows systems
    to learn from data without explicit programming. Deep learning, using neural
    networks, has achieved remarkable results in image and speech recognition.
    However, challenges remain around bias, interpretability, and ethical use.
    """

    summary = chain.invoke({"text": text})
    print(summary)


if __name__ == "__main__":
    main()