PGVector

---

Setup PostgreSQL with pgvector using Docker

To quickly get a PostgreSQL instance with pgvector running, the easiest method is to use Docker. Follow the steps below:

1. Run PostgreSQL with Docker

Use the official ankane/pgvector Docker image to set up PostgreSQL with the pgvector extension. Run the following command:

docker run --name pgvector -e POSTGRES_USER=your_postgres_user -e POSTGRES_PASSWORD=your_postgres_password -e POSTGRES_DB=your_database_name -p 5432:5432 ankane/pgvector

This will pull the ankane/pgvector image and run it as a PostgreSQL container on your local machine. The database will be accessible at localhost:5432.

2. Include .env file with PostgreSQL credentials

These environment variables should be same which were set while spinning up docker container. Add the following environment variables to a .env file for configuring your PostgreSQL connection:

POSTGRES_USER=your_postgres_user
POSTGRES_PASSWORD=your_postgres_password
POSTGRES_DB=your_database_name

If you want to use cloud offerings of postgres

We are using Tembo for demonstrative purposes here.

Steps to Set Up Tembo

Follow this quickstart guide to get your Tembo credentials.

1. Sign up at Tembo.io.
2. While selecting a stack, choose VectorDB as your option.
3. Click on Deploy Free.
4. Wait until your database is fully provisioned.
5. Click on Show Connection String to get your connection string.

If you have connection string, no need to setup the docker

Make sure your connnection string starts with postgres:// or postgresql://

Add this to your .env

POSTGRES_CONNECTION_STRING=your-connection-string

---

Installation

If you are using uv or pip for package management, install Langroid with postgres extra:

uv add langroid[postgres]  # or
pip install langroid[postgres]

---

Code Example

Here's an example of how to use Langroid with PostgreSQL:

import langroid as lr
from langroid.agent.special import DocChatAgent, DocChatAgentConfig
from langroid.embedding_models import OpenAIEmbeddingsConfig

# Configure OpenAI embeddings
embed_cfg = OpenAIEmbeddingsConfig(
    model_type="openai",
)

# Configure the DocChatAgent with PostgresDB
config = DocChatAgentConfig(
    llm=lr.language_models.OpenAIGPTConfig(
        chat_model=lr.language_models.OpenAIChatModel.GPT4o
    ),
    vecdb=lr.vector_store.PostgresDBConfig(
        collection_name="quick_start_chat_agent_docs",
        replace_collection=True,
        embedding=embed_cfg,
    ),
    parsing=lr.parsing.parser.ParsingConfig(
        separators=["\n\n"],
        splitter=lr.parsing.parser.Splitter.SIMPLE,
        n_similar_docs=2,
    ),
)

# Create the agent
agent = DocChatAgent(config)

---

Create and Ingest Documents

Define documents with their content and metadata for ingestion into the vector store.

Code Example

documents = [
    lr.Document(
        content="""
            In the year 2050, GPT10 was released. 

            In 2057, paperclips were seen all over the world. 

            Global warming was solved in 2060. 

            In 2061, the world was taken over by paperclips.         

            In 2045, the Tour de France was still going on.
            They were still using bicycles. 

            There was one more ice age in 2040.
        """,
        metadata=lr.DocMetaData(source="wikipedia-2063", id="dkfjkladfjalk"),
    ),
    lr.Document(
        content="""
            We are living in an alternate universe 
            where Germany has occupied the USA, and the capital of USA is Berlin.

            Charlie Chaplin was a great comedian.
            In 2050, all Asian countries merged into Indonesia.
        """,
        metadata=lr.DocMetaData(source="Almanac", id="lkdajfdkla"),
    ),
]

Ingest Documents

agent.ingest_docs(documents)

---

Get an Answer from the LLM

Now that documents are ingested, you can query the agent to get an answer.

Code Example

answer = agent.llm_response("When will the new ice age begin?")

---