Xinference通过docker方式启动Reranker模型-配置已成功应用至dify

使用xprobe/xinference:latest镜像

1.拉取镜像

拉取最新稳定版 Xinference 镜像

docker pull xprobe/xinference:latest

验证镜像是否拉取成功（显示 xprobe/xinference 即生效）

docker images | findstr xinference

2.运行镜像-PowerShell

Windows CMD 命令（复制直接运行）

docker run -d ^
  --name xinference-server ^
  -p 9997:9997 ^
  -v %USERPROFILE%\.xinference:/root/.xinference ^
  xprobe/xinference:latest ^
  xinference-local --host 0.0.0.0 --port 9997

若用 PowerShell，替换为(目前在win10上运行成功)

docker run -d `
  --name xinference-server `
  -p 9997:9997 `
  -v $env:USERPROFILE\.xinference:/root/.xinference `
  xprobe/xinference:latest `
  xinference-local --host 0.0.0.0 --port 9997

参数说明：

• –name xinference-server：给容器命名，方便后续管理；
• -p 9997:9997：映射容器 9997 端口到主机，外部可访问；
• -v %USERPROFILE%.xinference:/root/.xinference：挂载主机的模型缓存目录，重启容器后模型不丢失；
• –host 0.0.0.0：允许容器外部访问 Xinference 服务。

3.验证容器与服务是否启动成功

进入 xinference-server 容器的命令行

docker exec -it xinference-server bash

示例：在容器内加载 bge-reranker-v2-m3 重排序模型

xinference launch --model-name bge-reranker-v2-m3 --model-type rerank --repository-id BAAI/bge-reranker-v2-m3

4.Xinference缓存模型

缓存成功的结果展示

已运行的模型

缓存模型的设定

等待缓存完成即可。

5.Xinference运行模型

6.dify配置Xinference中的模型

7.智能体编排中使用Reranker模型

知识库设置完成后，记得发布。

附DSL内容-另存为yml即可直接用与dify

app:
  description: ''
  icon: 🤖
  icon_background: '#FFEAD5'
  mode: advanced-chat
  name: 运维规章制度
  use_icon_as_answer_icon: false
dependencies:
- current_identifier: null
  type: marketplace
  value:
    marketplace_plugin_unique_identifier: langgenius/openai_api_compatible:0.0.25@6c02d20ecf7eba40234be5201f25c2b6ea918ec09e0f8eb2a333efb495947d02
    version: null
kind: app
version: 0.5.0
workflow:
  conversation_variables: []
  environment_variables: []
  features:
    file_upload:
      allowed_file_extensions:
      - .JPG
      - .JPEG
      - .PNG
      - .GIF
      - .WEBP
      - .SVG
      allowed_file_types:
      - image
      allowed_file_upload_methods:
      - local_file
      - remote_url
      enabled: false
      fileUploadConfig:
        audio_file_size_limit: 50
        batch_count_limit: 5
        file_size_limit: 15
        image_file_size_limit: 10
        video_file_size_limit: 100
        workflow_file_upload_limit: 10
      image:
        enabled: false
        number_limits: 3
        transfer_methods:
        - local_file
        - remote_url
      number_limits: 3
    opening_statement: 你好 我是运维管理员
    retriever_resource:
      enabled: true
    sensitive_word_avoidance:
      enabled: false
    speech_to_text:
      enabled: false
    suggested_questions: []
    suggested_questions_after_answer:
      enabled: false
    text_to_speech:
      enabled: false
      language: ''
      voice: ''
  graph:
    edges:
    - data:
        sourceType: llm
        targetType: answer
      id: llm-answer
      source: llm
      sourceHandle: source
      target: answer
      targetHandle: target
      type: custom
    - data:
        isInLoop: false
        sourceType: start
        targetType: knowledge-retrieval
      id: 1771985974968-source-1771986921623-target
      source: '1771985974968'
      sourceHandle: source
      target: '1771986921623'
      targetHandle: target
      type: custom
      zIndex: 0
    - data:
        isInLoop: false
        sourceType: knowledge-retrieval
        targetType: llm
      id: 1771986921623-source-llm-target
      source: '1771986921623'
      sourceHandle: source
      target: llm
      targetHandle: target
      type: custom
      zIndex: 0
    nodes:
    - data:
        desc: 请输入需要了解的制度内容
        selected: false
        title: 用户输入
        type: start
        variables:
        - default: ''
          hint: ''
          label: 请输入您的问题
          max_length: 48
          options: []
          placeholder: ''
          required: true
          type: text-input
          variable: input_text
      height: 136
      id: '1771985974968'
      position:
        x: -250.08913350951966
        y: 324.99999999999994
      positionAbsolute:
        x: -250.08913350951966
        y: 324.99999999999994
      selected: false
      sourcePosition: right
      targetPosition: left
      type: custom
      width: 242
    - data:
        context:
          enabled: true
          variable_selector:
          - '1771986921623'
          - result
        desc: 大模型
        model:
          completion_params:
            temperature: 0.7
          mode: chat
          name: Qwen3-32B
          provider: langgenius/openai_api_compatible/openai_api_compatible
        prompt_template:
        - id: 2ee916cc-893c-43bd-a02d-975bc50446ff
          role: system
          text: "角色：\n您是一个运维管理员，熟悉所有的运维管理规范\n任务：\n请根据智慧运维知识库的所有内容，并提取核心观点，最后生成一段简短的摘要：\
            \ \n要求： \n1、 阅读索引文件并进行总结，语言简洁，不超过200字。 \n2、使用列表形式展示核心观点。\n3、仅使用知识库内容回答问题。\n"
        selected: false
        structured_output_enabled: false
        title: LLM
        type: llm
        vision:
          enabled: false
      height: 115
      id: llm
      position:
        x: 391.05956484547767
        y: 353.7284936842676
      positionAbsolute:
        x: 391.05956484547767
        y: 353.7284936842676
      selected: false
      sourcePosition: right
      targetPosition: left
      type: custom
      width: 242
    - data:
        answer: '{{#llm.text#}}/'
        selected: false
        title: 直接回复
        type: answer
        variables: []
      height: 102
      id: answer
      position:
        x: 754.7621681354833
        y: 378.51301645002985
      positionAbsolute:
        x: 754.7621681354833
        y: 378.51301645002985
      selected: false
      sourcePosition: right
      targetPosition: left
      type: custom
      width: 242
    - data:
        dataset_ids:
        - I96tLXkjt6ZmRKEH/mjSSzrEXpBR1mpvo9tbogayCnZr22f+SmIh2J1nJoaSpWnJ
        multiple_retrieval_config:
          reranking_enable: true
          reranking_mode: reranking_model
          reranking_model:
            model: models--baai--bge-reranker-v2-m3
            provider: langgenius/openai_api_compatible/openai_api_compatible
          score_threshold: null
          top_k: 4
          weights:
            keyword_setting:
              keyword_weight: 0.3
            vector_setting:
              embedding_model_name: qwen3-embedding:8b
              embedding_provider_name: langgenius/ollama/ollama
              vector_weight: 0.7
            weight_type: customized
        query_variable_selector:
        - '1771985974968'
        - input_text
        retrieval_mode: multiple
        selected: true
        title: 知识检索
        type: knowledge-retrieval
      height: 89
      id: '1771986921623'
      position:
        x: 66.20826320047445
        y: 353.7284936842676
      positionAbsolute:
        x: 66.20826320047445
        y: 353.7284936842676
      selected: true
      sourcePosition: right
      targetPosition: left
      type: custom
      width: 242
    viewport:
      x: 247.65853216270784
      y: 10.441073703044367
      zoom: 1.0000000000000009
  rag_pipeline_variables: []