Quadrant and Weaviate Vector Stores

 

Qdrant:

Qdrant is a vector database that excels at:

- ✅ Local development (no Docker required)

- ✅ Fast similarity search

- ✅ Flexible storage options (in-memory or persistent)

- ✅ Rich metadata filtering

It has 3 Approaches:

1. In-Memory - Fast but temporary (lost when script ends)

2. Persistent - Saved to disk (survives restarts)

3. from_documents - Easiest method (recommended)

Basic Similarity Search:

print("BASIC SIMILARITY SEARCH")

# Search for documents similar to this query

# k=2 means return the top 2 most similar documents

results = qdrant_store_memory.similarity_search(

    "Tell me about RAG",

    k=2

)

Filter on Metadata:

from qdrant_client.models import Filter, FieldCondition, MatchValue

print("SEARCH WITH METADATA FILTER")

print("-" * 80)

#metadata={"topic": ["rag", "llms", "agents"]}

# Create a filter to only search documents with topic='rag'

# Note: We use 'metadata.topic' because metadata is nested

qdrant_filter = Filter(

    must=[

        FieldCondition(

            key="metadata.topic",

            match=MatchValue(value="rag")

        )

    ]

)

# Same search, but only among filtered documents

results_filtered = qdrant_store_memory.similarity_search(

    "Tell me about RAG",

    k=2,

    filter=qdrant_filter

)

Multiple filter Conditions:

multi_filter = Filter(

    must=[

        FieldCondition(key="metadata.topic", match=MatchValue(value="rag")),

        FieldCondition(key="metadata.difficulty", match=MatchValue(value="intermediate"))

    ]

)

Create Qdrant store directly from Documents.

# Create Qdrant store directly from documents

# This is the easiest way - everything happens in one call!

qdrant_store_easy = QdrantVectorStore.from_documents(

    documents=sample_docs,          # Your documents

    embedding=embeddings,            # Embedding function

    path="./qdrant_easy",           # Local persistence (optional)

    collection_name="rag_collection" # Collection name

)

Weaviate:

This runs on Docket Image.

weaviate_client = weaviate.connect_to_local(

        host="localhost",

        port=8080,

        grpc_port=50051

    )

embeddings = OllamaEmbeddings(model="nomic-embed-text")

print("✓ Ollama embeddings initialized")

print("  Model: nomic-embed-text")

print("  Dimension: 768")

print("QDRANT FROM_DOCUMENTS (RECOMMENDED METHOD)")

# Create Qdrant store directly from documents

# This is the easiest way - everything happens in one call!

qdrant_store_easy = QdrantVectorStore.from_documents(

    documents=sample_docs,          # Your documents

    embedding=embeddings,            # Embedding function

    path="./qdrant_easy",           # Local persistence (optional)

    collection_name="rag_collection" # Collection name

)

Sample Code is under https://github.com/LeelaPrasadG/rag_langchain/blob/main/Vector_Stores_Tutorial.ipynb