site_name: 'FastDeploy 2.0: 大模型部署'
nav:
  - 'FastDeploy 2.0': index.md
  - 快速开始:
      - '10分钟上手ERNIE 4.5模型部署': get_started/quick_start.md 
      - '10分钟上手ERNIE 4.5多模态模型部署': get_started/quick_start_vl.md
      - 安装:
          - 'Nvidia GPU安装': get_started/installation/nvidia_gpu.md
          - 昆仑芯P800安装: get_started/installation/kunlunxin.md 
      - ERNIE-X1思考模型部署: get_started/ernie-x1.md
      - 'ERNIE-4.5-VL多模模型部署': get_started/ernie-4.5-vl.md
      - 'ERNIE-4.5模型部署': get_started/ernie-4.5.md
  - 服务化部署:
      - 使用方式: serving/README.md
      - 监控metrics: serving/metrics.md
      - 负载调度: serving/scheduler.md
  - 离线推理: offline_inference.md
  - 部署特性: 
      - 'Prefix Caching': features/prefix_caching.md
      - '分离式部署': features/disaggregated.md
      - 'Chunked Prefill与128K长文部署': features/chunked_prefill.md
      - '多实例负载均衡': features/load_balance.md
      - '投机解码': features/speculative_decoding.md
      - '结构化输出': features/structured_outputs.md
      - '思维链输出': features/reasoning_output.md
      - 'Tool Calling': features/tool_calling.md
  - 量化加速:
      - 无损量化: quantization/inflight_quantization.md
      - 'ERNIE-4.5 weight only int2规范说明': quantization/ernie_wint2.md
  - 支持模型列表: supported_models.md
  - Benchmark: benchmark.md
  - 架构设计:
      - 代码模块说明: design/code_guide.md
      - AppendAttention: design/append_attention.md 
  - 使用问题:
      - FAQ: usage/faq.md
      - 日志说明: usage/log.md
      - 算子编译: usage/build_ops.md
      - 自定义算子导入: usage/contribution_guide.md
      - 增加硬件支持: usage/how_to_support_new_device.md
theme: 
  name: 'material'
  highlightjs: true
  icon: 
    repo: fontawesome/brands/github
repo_url: https://github.com/PaddlePaddle/FastDeploy
repo_name: FastDeploy