RAG with MVS Standalone

from openai import OpenAI

from mixpeek import Mixpeek

openai = OpenAI(api_key="YOUR_OPENAI_KEY")
client = Mixpeek(api_key="YOUR_API_KEY")
NAMESPACE = "rag-docs"


def embed(text):
    return openai.embeddings.create(model="text-embedding-3-small", input=text).data[0].embedding


def chunk_text(text, size=200, overlap=40):
    words = text.split()
    return [" ".join(words[i:i + size]) for i in range(0, len(words), size - overlap)]


client.namespaces.create(
    namespace_id=NAMESPACE,
    mode="standalone",
    vector_configs=[{"name": "dense", "dimension": 1536, "metric": "cosine"}],
)

# 1. Chunk a document and upsert every chunk with its vector and text
with open("architecture-guide.md") as f:
    chunks = chunk_text(f.read())
client.namespaces.documents.upsert(
    namespace_id=NAMESPACE,
    documents=[
        {
            "document_id": f"architecture-guide-{i}",
            "vectors": {"dense": embed(chunk)},
            "payload": {"content": chunk, "source": "architecture-guide.md", "chunk_index": i},
        }
        for i, chunk in enumerate(chunks)
    ],
)

# 2. Retrieve the closest chunks for the question
question = "What is the recommended database architecture for high availability?"
results = client.search(
    namespace_id=NAMESPACE,
    queries=[{"feature_uri": "dense", "query": {"input_mode": "vector", "value": embed(question)}, "top_k": 5}],
)

# 3. Answer from those chunks, citing them by index
context = " ".join(f"[{doc.get('chunk_index')}] {doc.get('content')}" for doc in results["documents"])
response = openai.chat.completions.create(
    model="gpt-4o",
    messages=[
        {"role": "system", "content": "Answer from this context and cite chunk numbers. " + context},
        {"role": "user", "content": question},
    ],
)
print(response.choices[0].message.content)