version: '3.8'

services:
  devcontainer:
    build: .
    volumes:
      - ..:/workspace:cached
    ports:
      - "8000:8000"
    environment:
      - QDRANT_HOST=qdrant
      - QQDRANT_GRPC_PORT=6334
      - TEI_BASE_URL=http://text-embeddings-inference
      - TEI_RERANK_BASE_URL=http://text-embeddings-inference-rerank
      - OLLAMA_BASE_URL=http://ollama:11434
      - OLLAMA_MODEL_NAME=${OLLAMA_MODEL_NAME}
    command: sleep infinity

  qdrant:
    image: qdrant/qdrant:v1.7.4
    volumes:
      - ../qdrant_storage:/qdrant/storage:z
    ports:
      - "6333:6333"
      - "6334:6334"

  text-embeddings-inference:
    image: ghcr.io/huggingface/text-embeddings-inference:cpu-0.6
    volumes:
      - "../tei_data:/data"
    ports:
      - "8001:80"
    environment:
      - MODEL_ID=${TEI_MODEL_ID:-BAAI/bge-large-en-v1.5}
      - REVISION=${TEI_MODEL_REVISION}
      - MAX_CLIENT_BATCH_SIZE=128

  text-embeddings-inference-rerank:
    image: ghcr.io/huggingface/text-embeddings-inference:cpu-0.6
    volumes:
      - "../tei_data:/data"
    ports:
      - "8002:80"
    environment:
      - MODEL_ID=${TEI_RERANK_MODEL_ID:-BAAI/bge-reranker-large}
      - REVISION=${TEI_RERANK_MODEL_REVISION}
      - MAX_CLIENT_BATCH_SIZE=128

  ollama:
    image: ollama/ollama:latest
    volumes:
      - ../ollama:/root/.ollama
    ports:
      - 5000:11434
    deploy:
      resources:
        reservations:
          devices:
            - driver: ${OLLAMA_GPU_DRIVER-nvidia}
              count: ${OLLAMA_GPU_COUNT-1}
              capabilities:
                - gpu