low-volume/stage/dev LocalAI deployment sample for https://medium.com/@costa/llms-are-as-useful-to-people-whore-friends-with-formal-languages-as-1e43684329c7

docker-compose.yml

# NOTE re: deployment
# - with self server (running on any VM from any provider):
#   $ @@ dev docker-compose up --remove-orphans
#   and then:
#   $ @@ ^dev  # will securely tunnel WUI (:3000) to http://localhost:16623
# - with docker-compose configured otherwise:
#   # well, you must know how to run this in your environment then, see below
version: '2.1'
services:
  chat:
    image: localai/localai:v2.5.1-ffmpeg-core
    environment:
      - CORS=true
      - CORS_ALLOW_ORIGINS=*
    # NOTE within your custom docker-compose environment,
    # you will need this port securely tunnelled to your local machine somehow:
    # ports:
    #   - 8080:21703
    restart: on-failure
  chat-bootstrapper:
    depends_on:
      - chat
    image: curlimages/curl
    command:
      - "http://chat:8080/models/apply"
      - "-H"
      - "Content-Type: application/json"
      - "-d"
      # NOTE you're welcome to visit https://github.com/go-skynet/model-gallery
      # or discover other ways to load compatible models
      - |
          {
            "url": "github:go-skynet/model-gallery/gpt4all-j.yaml",
            "name": "gpt-3.5-turbo"
          }
    restart: on-failure
  wui:
    # NOTE using quay.io just to note the alternative
    image: quay.io/go-skynet/localai-frontend:master
    environment:
      # NOTE this is port number auto-assigned for this comp service (chat:8080)
      # by self server, so this is what we're using
      - API_HOST=http://localhost:21703
      - HOST=0.0.0.0
      # - PORT=3000
    restart: on-failure