i-hate-reading/docker-compose.yml at main · lemonjerome/i-hate-reading · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
version: "3.9"

services:
  # NOTE: Run Ollama NATIVELY on macOS for Metal GPU acceleration.
  # Start it before docker-compose: `ollama serve`
  # Docker cannot access Apple Silicon GPU — running Ollama in Docker
  # forces CPU-only mode (0/37 layers offloaded to GPU).

  qdrant:
    image: qdrant/qdrant
    container_name: qdrant
    ports:
      - "6333:6333"
      - "6334:6334"
    volumes:
      - qdrant_data:/qdrant/storage
    restart: unless-stopped
    healthcheck:
      test: ["CMD-SHELL", "timeout 10s bash -c ':> /dev/tcp/127.0.0.1/6333' || exit 1"]
      interval: 10s
      timeout: 5s
      retries: 10
      start_period: 20s

  notebook-agent:
    build:
      context: ./agent
      dockerfile: Dockerfile
    container_name: notebook-agent
    ports:
      - "8000:8000"
    environment:
      - OLLAMA_HOST=http://host.docker.internal:11434
      - OLLAMA_MODEL=gemma3:4b
      - QDRANT_HOST=http://qdrant:6333
      - RERANK_MODEL=BAAI/bge-reranker-base
      - ENABLE_RERANK=1
      - MAX_CONTEXT_CHUNKS=40
      - HF_HOME=/cache/huggingface
      - HF_HUB_OFFLINE=1
      - TRANSFORMERS_OFFLINE=1
      - NUM_CTX=24576
    volumes:
      - ~/.cache/huggingface:/cache/huggingface
    extra_hosts:
      - "host.docker.internal:host-gateway"
    depends_on:
      qdrant:
        condition: service_healthy
    restart: unless-stopped

networks:
  default:
    driver: bridge

volumes:
  qdrant_data: