使用xprobe/xinference:latest镜像1.拉取镜像拉取最新稳定版 Xinference 镜像dockerpull xprobe/xinference:latest验证镜像是否拉取成功显示 xprobe/xinference 即生效dockerimages|findstr xinference2.运行镜像-PowerShellWindows CMD 命令复制直接运行dockerrun -d ^ --name xinference-server ^ -p9997:9997 ^ -v %USERPROFILE%\.xinference:/root/.xinference ^ xprobe/xinference:latest ^ xinference-local --host0.0.0.0 --port9997若用 PowerShell替换为(目前在win10上运行成功)dockerrun -d--name xinference-server-p9997:9997-v $env:USERPROFILE\.xinference:/root/.xinferencexprobe/xinference:latest xinference-local --host0.0.0.0 --port9997参数说明–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dockerexec-it xinference-serverbash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-m34.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difyapp:description:icon:icon_background:#FFEAD5mode:advanced-chatname:运维规章制度use_icon_as_answer_icon:falsedependencies:-current_identifier:nulltype:marketplacevalue:marketplace_plugin_unique_identifier:langgenius/openai_api_compatible:0.0.256c02d20ecf7eba40234be5201f25c2b6ea918ec09e0f8eb2a333efb495947d02version:nullkind:appversion:0.5.0workflow:conversation_variables:[]environment_variables:[]features:file_upload:allowed_file_extensions:-.JPG-.JPEG-.PNG-.GIF-.WEBP-.SVGallowed_file_types:-imageallowed_file_upload_methods:-local_file-remote_urlenabled:falsefileUploadConfig:audio_file_size_limit:50batch_count_limit:5file_size_limit:15image_file_size_limit:10video_file_size_limit:100workflow_file_upload_limit:10image:enabled:falsenumber_limits:3transfer_methods:-local_file-remote_urlnumber_limits:3opening_statement:你好 我是运维管理员retriever_resource:enabled:truesensitive_word_avoidance:enabled:falsespeech_to_text:enabled:falsesuggested_questions:[]suggested_questions_after_answer:enabled:falsetext_to_speech:enabled:falselanguage:voice:graph:edges:-data:sourceType:llmtargetType:answerid:llm-answersource:llmsourceHandle:sourcetarget:answertargetHandle:targettype:custom-data:isInLoop:falsesourceType:starttargetType:knowledge-retrievalid:1771985974968-source-1771986921623-targetsource:1771985974968sourceHandle:sourcetarget:1771986921623targetHandle:targettype:customzIndex:0-data:isInLoop:falsesourceType:knowledge-retrievaltargetType:llmid:1771986921623-source-llm-targetsource:1771986921623sourceHandle:sourcetarget:llmtargetHandle:targettype:customzIndex:0nodes:-data:desc:请输入需要了解的制度内容selected:falsetitle:用户输入type:startvariables:-default:hint:label:请输入您的问题max_length:48options:[]placeholder:required:truetype:text-inputvariable:input_textheight:136id:1771985974968position:x:-250.08913350951966y:324.99999999999994positionAbsolute:x:-250.08913350951966y:324.99999999999994selected:falsesourcePosition:righttargetPosition:lefttype:customwidth:242-data:context:enabled:truevariable_selector:-1771986921623-resultdesc:大模型model:completion_params:temperature:0.7mode:chatname:Qwen3-32Bprovider:langgenius/openai_api_compatible/openai_api_compatibleprompt_template:-id:2ee916cc-893c-43bd-a02d-975bc50446ffrole:systemtext: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。\nselected:falsestructured_output_enabled:falsetitle:LLMtype:llmvision:enabled:falseheight:115id:llmposition:x:391.05956484547767y:353.7284936842676positionAbsolute:x:391.05956484547767y:353.7284936842676selected:falsesourcePosition:righttargetPosition:lefttype:customwidth:242-data:answer:{{#llm.text#}}/selected:falsetitle:直接回复type:answervariables:[]height:102id:answerposition:x:754.7621681354833y:378.51301645002985positionAbsolute:x:754.7621681354833y:378.51301645002985selected:falsesourcePosition:righttargetPosition:lefttype:customwidth:242-data:dataset_ids:-I96tLXkjt6ZmRKEH/mjSSzrEXpBR1mpvo9tbogayCnZr22fSmIh2J1nJoaSpWnJmultiple_retrieval_config:reranking_enable:truereranking_mode:reranking_modelreranking_model:model:models--baai--bge-reranker-v2-m3provider:langgenius/openai_api_compatible/openai_api_compatiblescore_threshold:nulltop_k:4weights:keyword_setting:keyword_weight:0.3vector_setting:embedding_model_name:qwen3-embedding:8bembedding_provider_name:langgenius/ollama/ollamavector_weight:0.7weight_type:customizedquery_variable_selector:-1771985974968-input_textretrieval_mode:multipleselected:truetitle:知识检索type:knowledge-retrievalheight:89id:1771986921623position:x:66.20826320047445y:353.7284936842676positionAbsolute:x:66.20826320047445y:353.7284936842676selected:truesourcePosition:righttargetPosition:lefttype:customwidth:242viewport:x:247.65853216270784y:10.441073703044367zoom:1.0000000000000009rag_pipeline_variables:[]