LLM App by Analytiq

Install instructions

Clone the sandbox:

• mkdir ~/build; cd ~/build
• git clone git@github.com:analytiq-hub/analytiq-language-app.git
• Copy .env_template to .env, and configure the environment variables listed in that file.

Set up the Analytiq docstore:

• For now, get analytiq-docstore.tgz from Andrei, and untar it under ~/build. You will end up with ~/build/analytiq-docstore, which holds the public records database.
  • TO DO: I need to create from-scratch instructions on how to create ~/build/analytiq-docstore

Create & start the ChromaDB:

• Prerequisite: Install docker and docker-compose.
  • Ubuntu instructions
  • Other Linux distributions
• Clone the ChromaDB sandbox:
  • cd ~/build
  • git clone git@github.com:chroma-core/chroma.git
  • cd chroma
  • In docker-compose.yaml
    • Change port 8000 to 8002, to avoid conflict with Airbyte port 8000
    • Add ALLOW_RESET=TRUE

Here is the docker-compose.yaml:

version: '3.9'

networks:
  net:
    driver: bridge

services:
  server:
    image: server
    build:
      context: .
      dockerfile: Dockerfile
    volumes:
      - ./:/chroma
      - index_data:/index_data
    command: uvicorn chromadb.app:app --reload --workers 1 --host 0.0.0.0 --port 8002 --log-config log_config.yml
    environment:
      - IS_PERSISTENT=TRUE
      - ALLOW_RESET=TRUE
      - CHROMA_SERVER_AUTH_PROVIDER=${CHROMA_SERVER_AUTH_PROVIDER}
      - CHROMA_SERVER_AUTH_CREDENTIALS_FILE=${CHROMA_SERVER_AUTH_CREDENTIALS_FILE}
      - CHROMA_SERVER_AUTH_CREDENTIALS=${CHROMA_SERVER_AUTH_CREDENTIALS}
      - CHROMA_SERVER_AUTH_CREDENTIALS_PROVIDER=${CHROMA_SERVER_AUTH_CREDENTIALS_PROVIDER}
    ports:
      - 8002:8002
    networks:
      - net

volumes:
  index_data:
    driver: local
  backups:
    driver: local

• Start ChromaDB
  • Do docker-compose up -d --build to bring up, and docker-compose down to bring down.

At this point, you have two alternatives:

Run Analytiq app in a Python virtual environment

Install the required system packages:

• On Ubuntu:
  • sudo apt-get install python3.10-venv
  • Install all deb packages from docker/Dockerfile
• On CentOS, RedHat, Fedora:
  • `sudo dnf install python3-virtualenv``
  • Install all rpm packages corresponding to deb packages from docker/Dockerfile

Set up the python virtual environment:

• mkdir ~/.venv
• python -m venv ~/.venv/analytiq
• . ~/.venv/analytiq/bin/activate

Install python module dependencies inside the venv:

• cd ~/build/analytiq-language-app
• pip install torch
• pip install -r requirements.txt

You are now ready to run the analytiq app:

• streamlit run Analytiq.py --server.port 8080

Run Analytiq app in a Docker environment

The Docker service has been installed earlier, along with ChromaDB. Next steps:

• cd .../analytiq-language-app
• Build the Docker image: docker/bin/docker_build.sh
• Stop previous containers (if any): docker/bin/docker_stop.sh
• Start the new docker container: docker/bin/docker_start.sh
  • You are dropped to a shell inside the container
  • Run streamlit run Analytiq.py --server.port 8080
  • If the container is already started, and you run docker_start.sh again, you will be dropped to a shell inside the existing container