LangChain Integration

Overview

The LiquidMetal Python integration package includes a custom LangChain retriever that lets you use SmartBucket within your LangChain applications. This integration enables semantic search capabilities in your LangChain pipelines.

Prerequisites

• Python 3.7 or higher
• LiquidMetal API key
• lm-raindrop-integrations package installed

Installation

pip install lm-raindrop-integrations

Quick Start

Here’s how to use the SmartBucket retriever in your LangChain application:

from lm_raindrop_integrations import LangchainSmartBucketRetriever

# Initialize the retriever
retriever = LangchainSmartBucketRetriever(
    api_key="your-api-key"  # Alternatively, set RAINDROP_API_KEY env variable
)

# Search for documents
results = retriever.invoke("What is machine learning?")

# Process results
for doc in results:
    print(f"Content: {doc.page_content}")
    print(f"Score: {doc.metadata['score']}")
    print(f"Source: {doc.metadata['source']}")
    print("---")

Tip

Set your API key as an environment variable RAINDROP_API_KEY for better security.

Response Format

The retriever returns LangChain Document objects. Here’s an example response:

# Example response structure
[
    Document(
        page_content="Machine learning is a subset of artificial intelligence that enables systems to learn and improve from experience without being explicitly programmed.",
        metadata={
            'score': 0.89,
            'source': 'introduction_to_ml.pdf',
            'bucket_id': 'ml_docs_bucket'
        }
    ),
    Document(
        page_content="Deep learning is a type of machine learning based on artificial neural networks.",
        metadata={
            'score': 0.76,
            'source': 'deep_learning_basics.pdf',
            'bucket_id': 'ml_docs_bucket'
        }
    )
]

Each Document object contains:

• page_content: The matched text content
• metadata: A dictionary containing:
  • score: Relevance score (0-1)
  • source: Original document source
  • bucket_id: ID of the SmartBucket containing the document