docker-compose.gpu.yml

# GPU overlay for docker-compose.yml. Gives the API and Ollama the host NVIDIA GPU and
# builds the CUDA app image. Requires the NVIDIA Container Toolkit.
#
#   docker compose -f docker-compose.yml -f docker-compose.gpu.yml up --build
#   docker compose exec ollama ollama pull llama3.1:8b   # one-time
services:
  api:
    build:
      context: .
      dockerfile: Dockerfile.gpu
    environment:
      EMBED_DEVICE: cuda
      RERANK_DEVICE: cuda
    restart: unless-stopped
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]

  ollama:
    restart: unless-stopped
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]