docker-compose.yml

services:
  download_models:
    image: aria2:latest
    container_name: model_download
    build:
      context: .
      dockerfile: Dockerfile

    volumes:
      - "./models:/models"

  flowise:
    image: flowiseai/flowise
    container_name: flowise
    restart: always

    env_file:
      - .env

    ports:
      - '${PORT}:${PORT}'
    volumes:
      - ./data:/root/.flowise
    command: /bin/sh -c "sleep 3; flowise start"

  llama:
    depends_on:
      download_models:
        condition: service_completed_successfully

    container_name: llama_python
    image: llama.cpp:python
    build:
      context: llama/docker/cuda_simple
      dockerfile: Dockerfile

    environment:
      - MODEL=/models/${MODEL_NAME}
      - N_GPU_LAYERS=50
      - N_CTX=16384
      - USE_MLOCK=1

    volumes:
      - "./models:/models"

    ports:
      - "4096:8000"

    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [ gpu ]