æ©æ¢°å¦ç¿ã¯ã¼ã¯ããã¼ç®¡çãã¼ã«ã§ããKubeflowã®Pipelinesæ©è½ã使ã£ã¦æ¥æ¬èªããã¹ãåé¡ã®å®é¨ç®¡çãè¡ãã¾ãã ãã®è¨äºã§ã¯Kubeflowã®ãã¥ã¼ããªã¢ã«ã«å¾ã£ã¦Kubeflowã®ã¯ã©ã¹ã¿ãæ§ç¯ãã¦Pipelinesãåãããæå¾ã«æ¥æ¬èªã®ãã¼ã¿ã»ãããKubeflow Pipelinesã«å®éã«ä¹ãã¦æ©æ¢°å¦ç¿ã®ã¯ã¼ã¯ããã¼ç®¡çãè¡ãã¨ããã¾ã§ã¾ã¨ãã¦ããã¾ãã Kubeflow Kubeflowã¨ã¯ Pipelinesã¨ã¯ GKEã§Kubeflowã¯ã©ã¹ã¿ã®æ§ç¯ ã¯ã©ã¹ã¿æ§ç¯ Workload Identityã®è¨å® Pipelinesã®åºæ¬çãªä½¿ãæ¹ Pipeline/Experiment/Run Pipelineã¨Component Pipelineã¨DSL å®é¨ç®¡çã¨ãã¦ã®Kubeflow Pipelines æ¥æ¬èªããã¹ãåé¡ with Pipelines Pip
Kubernetesãæ©æ¢°å¦ç¿åºç¤ã¨ãã¦Kubeflowããªãªã¼ã¹ãã¾ããã ãã£ããããã¨ãKubeFlowã¯ä»¥ä¸ã®æ©æ¢°å¦ç¿ã¢ããªã±ã¼ã·ã§ã³éçºã®ä¸é£ã®ã¯ã¼ã¯ãã¼ãããµãã¼ãããOSSã§ãã ã¢ãã«éçºåºç¤ ãã¼ã¿ãµã¤ã¨ã³ãã£ã¹ããæ©æ¢°å¦ç¿ã¨ã³ã¸ãã¢ãJupyterNotebookãã¤ãã£ã¦ã¢ãã«éçºããããã®ãJupyterãåããµã¼ãç°å¢ãæä¾ TensorFlowåæ£å¦ç¿åºç¤ éçºããã¢ãã«ãTensorFlowã§åæ£å¦ç¿ããããã®Kubernetesã¯ã©ã¹ã¿ãèªåçæ ã¢ããªå ¬éåºç¤ å¦ç¿æ¸ã¿ã¢ãã«ãKubernetesã¯ã©ã¹ã¿ä¸ã§ãµã¼ãã¹å ¬éããåºç¤ãæä¾(Tensorflow Serving) github.com ã¾ã ãéçºéä¸ããã¤ãã£ã¨ãã追ãã¦ãã¾ããããå°å ¥ã®ãããã¨æ¦è¦ãã¾ã¨ãã¾ãã 0. kubeFlowç°å¢æ§ç¯ Kubeflowãåããããã«ã¯ãã¾ãKuber
æ©æ¢°å¦ç¿ãæ¥åã«é©ç¨ããã«ã¯ããã¼ã¿ã®åéãæ©æ¢°å¦ç¿ã¢ãã«ã®ä½æãæ©æ¢°å¦ç¿ã¢ãã«ãå©ç¨ãã API éçºãAPI ãå©ç¨ããã¢ããªéçºããããã¯ã·ã§ã³ã¸ã®é©ç¨ãéç¨æã®çµå¹´å¤åã«ããã¢ãã«ã®ä¿®æ£ ã¨ãã£ãããã»ã¹ã§é²ãããã¨ã«ãªãã¾ãã Kubeflow ã¯æ©æ¢°å¦ç¿ã¢ãã«ä½æãAPI éçºã¨ãããã¤ã¾ã§ãã«ãã¼ããåºç¤ã½ããã¦ã§ã¢ã§ãããã®å®ä½ã¯æ°å¤ãã® OSS ã® éåä½ (best-of-breed *1 ) ã¨ãªã£ã¦ãã¾ãã www.kubeflow.org Kubeflow ã¯ãã®åã®éã Kubernetes ãã¤ãã£ããªã¢ããªãªã®ã§ Kubernetes ã¯ã©ã¹ã¿ã¼ã«ãããã¤ãã¦ä½¿ç¨ãã¾ããGCP / EKS / AKS ã¨ãã£ãã¡ã¸ã£ã¼ãªããã¼ã¸ã Kubernetes ã¯ãã¡ããããã¼ã«ã«ã® Minikubeãkind *2 ãªã©ã«ãããã¤å¯è½ã§ãã æ®å¿µãªããç¾ç¶ãDoc
${KUBEFLOW_SRC}/scripts/kfctl.sh init ${KFAPP} --platform none cd ${KFAPP} ${KUBEFLOW_SRC}/scripts/kfctl.sh generate k8s ${KUBEFLOW_SRC}/scripts/kfctl.sh apply k8s $ kubectl get pod --namespace kubeflow NAME READY STATUS RESTARTS AGE ambassador-7b8477f667-cdqd7 1/1 Running 0 49m ambassador-7b8477f667-klkxb 1/1 Running 0 49m ambassador-7b8477f667-x5jvs 1/1 Running 0 49m argo-ui-fb67b6bc8-282v4 1/1
Googleã¯7æ10æ¥ããªã¼ãã³ã½ã¼ã¹ããã¸ã§ã¯ãã®ããã¥ã¡ã³ãå ¬éã«åããWebãµã¤ããã¼ããDocsyããå ¬éãããããã¥ã¡ã³ããå ¬éãããµã¤ããç°¡åã«ç«ã¡ä¸ãã¦éç¨ã§ããã¨ããã Docsyã¯æè¡ææ¸ãå ¬éãããããªWebãµã¤ãã®ããã®ãã¼ãã§ãWebãµã¤ãæ§ç¯ã®ããã®ãã¬ã¼ã ã¯ã¼ã¯ãHugoãããã¼ã¹ã¨ãããGoogleã¯2000以ä¸ã®ãªã¼ãã³ã½ã¼ã¹ããã¸ã§ã¯ããæ±ãã¦ãããããã¥ã¡ã³ãã¼ã·ã§ã³ä½æã¨å ¬éã®ããã®ãã¼ã«ãå¿ è¦ã ã£ããã¨ããæ§ç¯ããã¨ã®ãã¨ãæè¡ææ¸åãã®ãã³ãã¬ã¼ãã¨ã¬ã¤ããåãã¦ããããã§ã«KubeflowãKnativeãAgonesãªã©ã®ããã¸ã§ã¯ãã§å©ç¨ãã¦ããã¨ããã ããã²ã¼ã·ã§ã³ããµã¤ãæ§é ãªã©ã®æ©è½ãæä¾ããã»ããå¤è¨èªã«ã対å¿ããããã¼ã¸ã®è¿½å ãããã¥ã¡ã³ãã¼ã·ã§ã³ã®æ§é åãã³ãã¥ããã£ããã®è²¢ç®ãªã©ã«ã¤ãã¦ãã¬ã¤ããæä¾ããã¨ããã Docsy
ã¡ã«ã«ãªæ©æ¢°å¦ç¿ã¨ã³ã¸ãã¢ã® @hurutoriya ã§ãã 2018/09/01 ã«éå¬ããã GDG Dev Fest 2018 㧠Kubeflow ã«ã¤ãã¦ç»å£ãã¦ãã¾ããã ML Sessions | GDG DevFest 2018 Tokyo çºè¡¨è³æ www.slideshare.net Kubeflowã«ã¤ãã¦ã®åºç¤çãªèª¬æã¯çºè¡¨è³æã§è¡ã£ã¦ããã®ã§ããã²ã覧ãã ããã çºè¡¨å 容ã¯ãæ©æ¢°å¦ç¿ããµã¼ãã¹ã¨ãã¦æä¾ããéã«ããããªå£ãåå¨ãã¦ãã¾ããããã®å£ãã¶ã£å£ãã®ã Kubeflow ã ! ã¨ããæµãã§çºè¡¨ããã¦ãã¾ããã åèªèº«ãæ©æ¢°å¦ç¿ã¨ã³ã¸ãã¢ã¨ãããã¨ããããk8sãç¥ãå°½ããã¦ããããã§ã¯ãªããããã¾ã§æ©æ¢°å¦ç¿ã¨ã³ã¸ãã¢ã¨ã㦠Kubeflowã®ææãå ±æããã¦ããã ãã¾ããã åèªèº«ãKubeflowã«èå³ããã£ãèæ¯ã¨ãã¦ã¯ãã¡ã«ã«ãªã§ã¯æ©æ¢°å¦ç¿åºç¤ã®éçº
ãªãªã¼ã¹ãé害æ å ±ãªã©ã®ãµã¼ãã¹ã®ãç¥ãã
ææ°ã®äººæ°ã¨ã³ããªã¼ã®é ä¿¡
å¦çãå®è¡ä¸ã§ã
j次ã®ããã¯ãã¼ã¯
kåã®ããã¯ãã¼ã¯
lãã¨ã§èªã
eã³ã¡ã³ãä¸è¦§ãéã
oãã¼ã¸ãéã
{{#tags}}- {{label}}
{{/tags}}