feat: 添加Docker部署配置

Dockerfile: 多阶段构建, CPU版PyTorch, 模型预加载

docker-compose: 健康检查, 数据卷持久化, 环境变量注入

requirements.lock.txt: 锁定依赖版本确保构建可复现

.dockerignore

@@ -0,0 +1,11 @@
+venv/
+__pycache__/
+*.pyc
+.git/
+.env
+data/
+docs/
+*.md
+.gitignore
+.env.example
+.pytest_cache/

Dockerfile

@@ -0,0 +1,47 @@
+# ── Stage 1: 依赖安装 ──────────────────────────────────────
+FROM python:3.12-slim AS builder
+
+WORKDIR /build
+
+# 先拷贝依赖文件，利用 Docker 缓存
+COPY requirements.lock.txt .
+
+RUN pip install --no-cache-dir -r requirements.lock.txt
+
+# ── Stage 2: 运行时 ────────────────────────────────────────
+FROM python:3.12-slim
+
+LABEL maintainer="LLM Compass"
+LABEL description="智能LLM路由服务，为请求指引最优模型"
+
+WORKDIR /app
+
+# 安装运行时系统依赖（sentencepiece 等）
+RUN apt-get update && \
+    apt-get install -y --no-install-recommends libgomp1 && \
+    rm -rf /var/lib/apt/lists/*
+
+# 从 builder 拷贝 Python 包
+COPY --from=builder /usr/local/lib/python3.12/site-packages /usr/local/lib/python3.12/site-packages
+COPY --from=builder /usr/local/bin /usr/local/bin
+
+# 拷贝应用代码
+COPY config.py main.py nvidia_router.py ./
+
+# 创建数据目录
+RUN mkdir -p /app/data
+
+# 预下载 NVIDIA 模型（构建时缓存，避免每次启动下载）
+RUN python -c "from nvidia_router import get_nvidia_router; r = get_nvidia_router(); r.initialize(); print('Model preloaded successfully')" || echo "Model preload failed, will download on first request"
+
+# 环境变量（敏感信息通过 docker-compose / --env-file 注入）
+ENV PYTHONUNBUFFERED=1
+
+# 暴露端口
+EXPOSE 8000
+
+# 数据持久化
+VOLUME ["/app/data"]
+
+# 启动命令
+CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8000", "--workers", "1"]

docker-compose.yml

@@ -0,0 +1,20 @@
+services:
+  llm-compass:
+    build: .
+    container_name: llm-compass
+    ports:
+      - "8000:8000"
+    environment:
+      - DASHSCOPE_API_KEY=${DASHSCOPE_API_KEY}
+    volumes:
+      - compass-data:/app/data
+    restart: unless-stopped
+    healthcheck:
+      test: ["CMD", "python", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:8000/health')"]
+      interval: 30s
+      timeout: 10s
+      retries: 3
+      start_period: 60s
+
+volumes:
+  compass-data:

requirements.lock.txt

@@ -0,0 +1,17 @@
+# LLM Compass - Docker 锁定依赖 (CPU)
+# 使用 CPU 版 PyTorch，大幅减小镜像体积
+--extra-index-url https://download.pytorch.org/whl/cpu
+fastapi==0.136.0
+uvicorn[standard]==0.44.0
+pydantic==2.12.5
+litellm==1.83.9
+tiktoken==0.12.0
+httpx==0.28.1
+python-dotenv==1.0.1
+torch==2.2.2+cpu
+transformers==4.57.6
+tokenizers==0.22.2
+safetensors==0.4.3
+numpy==1.26.4
+sentencepiece==0.2.1
+huggingface_hub>=0.28.0