rag_chroma_example

RAG with Chroma VectorStore Example

This example demonstrates how to use Chroma vector database with LangGraphGo's RAG pipeline through the langchaingo adapter.

What is Chroma?

Chroma is an open-source embedding database designed to make it easy to build LLM applications. It provides:

• Simple API for storing and querying embeddings
• Multiple distance metrics (cosine, L2, IP)
• Filtering and metadata support
• Persistent storage
• Client-server architecture

Prerequisites

1. Chroma Server: Run Chroma using Docker

   docker run -p 8000:8000 chromadb/chroma
   

2. API Key: DeepSeek or OpenAI

   export DEEPSEEK_API_KEY="your-api-key"
   # or
   export OPENAI_API_KEY="your-api-key"
   

Running the Example

cd examples/rag_chroma_example
go run main.go

Features Demonstrated

• Creating a Chroma vector store with custom configuration
• Adding documents with embeddings to Chroma
• Building a RAG pipeline with Chroma as the vector store
• Querying the pipeline with natural language questions
• Similarity search with relevance scores

Code Highlights

Creating Chroma Store

chromaStore, err := chroma.New(
    chroma.WithChromaURL("http://localhost:8000"),
    chroma.WithEmbedder(embedder),
    chroma.WithDistanceFunction("cosine"),
    chroma.WithNameSpace("langgraphgo_example"),
)

// Wrap with adapter
vectorStore := prebuilt.NewLangChainVectorStore(chromaStore)

Adding Documents

err = vectorStore.AddDocuments(ctx, chunks, embeddings)

Similarity Search

results, err := vectorStore.SimilaritySearchWithScore(ctx, query, k)
for _, result := range results {
    fmt.Printf("Score: %.4f - %s\n", result.Score, result.Document.PageContent)
}

Expected Output

=== RAG with Chroma VectorStore Example ===

Loading and splitting documents...
Split into 6 chunks

Connecting to Chroma vector database...
Adding documents to Chroma...
Documents successfully added to Chroma

Building RAG pipeline...
Pipeline ready!

================================================================================
Query 1: What is Go programming language?
--------------------------------------------------------------------------------

Retrieved 3 documents from Chroma:
  [1] Go is a statically typed, compiled programming language designed at Google...
  [2] Key features of Go include: - Fast compilation times - Built-in concurrency...
  [3] Go is particularly well-suited for: - Web servers and APIs - Cloud and network...

Answer:
Go is a statically typed, compiled programming language designed at Google...

Chroma Configuration Options

chroma.New(
    chroma.WithChromaURL("http://localhost:8000"),     // Chroma server URL
    chroma.WithEmbedder(embedder),                      // Embedding function
    chroma.WithDistanceFunction("cosine"),              // Distance metric: cosine, l2, ip
    chroma.WithNameSpace("my_collection"),              // Collection name
)

Troubleshooting

Chroma server not running

Error: Failed to create Chroma store: connection refused

Solution: Start Chroma server with Docker:

docker run -p 8000:8000 chromadb/chroma

Port already in use

Error: bind: address already in use

Solution: Use a different port or stop the conflicting service:

docker run -p 8001:8000 chromadb/chroma
# Update code: chroma.WithChromaURL("http://localhost:8001")

Next Steps

• Try other vector stores: Weaviate, Pinecone, Qdrant
• Experiment with different distance functions
• Add metadata filtering to searches
• Implement hybrid search (keyword + vector)

References

• Chroma Documentation
• LangChain Go Chroma Integration
• LangGraphGo RAG Documentation