Mantenha tudo organizado com as coleções
Salve e categorize o conteúdo com base nas suas preferências.
O app da Web TensorBoard da Vertex AI fornece uma visualização de
registros associados a um experimento do TensorBoard da Vertex AI.
Esse aplicativo da Web oferece várias ferramentas e painéis para visualizar e comparar dados entre execuções de experimento e criar o perfil do desempenho do treinamento de modelo.
Para saber mais sobre cada painel, consulte o README de "A visualização".
Acessar um app da Web TensorBoard da Vertex AI
Acesse o app da Web TensorBoard da Vertex AI usando o
formato de endereço a seguir:
EXPERIMENT_NAME: o nome do experimento do
TensorBoard
Este link pode ser compartilhado e usado por qualquer pessoa que tenha permissão.
Navegue diretamente até aqui inserindo os valores apropriados
ou seguindo um link disponível das maneiras a seguir.
CLI gcloud e SDK da Vertex AI para Python
Ao criar ou fazer upload de registros para um TensorBoard da Vertex AI, a
CLI gcloud e o SDK geram um link para a instância do TensorBoard da
Vertex AI.
Console do Google Cloud
É possível acessar o app da Web do TensorBoard usando o console do Google Cloud.
Conferir as métricas do TensorBoard da Vertex AI no console do Google Cloud
Se sua instância do TensorBoard estiver associada a
experimentos da Vertex AI, que são automáticos ao usar uma
instância padrão do TensorBoard, determinadas
métricas podem ser visualizadas diretamente no console do Google Cloud. sem usar o aplicativo da Web.
O documento Comparar e analisar execuções explica como visualizar detalhes das execuções do experimento e compará-los uns com os outros no console do Google Cloud.
As métricas de série temporal do TensorBoard aparecem na guia Gráficos da barra de navegação do console do Vertex AI Experiments.
[[["Fácil de entender","easyToUnderstand","thumb-up"],["Meu problema foi resolvido","solvedMyProblem","thumb-up"],["Outro","otherUp","thumb-up"]],[["Difícil de entender","hardToUnderstand","thumb-down"],["Informações incorretas ou exemplo de código","incorrectInformationOrSampleCode","thumb-down"],["Não contém as informações/amostras de que eu preciso","missingTheInformationSamplesINeed","thumb-down"],["Problema na tradução","translationIssue","thumb-down"],["Outro","otherDown","thumb-down"]],["Última atualização 2025-09-04 UTC."],[],[],null,["# View Vertex AI TensorBoard data\n\nThe Vertex AI TensorBoard web app provides a visualization of\nlogs associated with a Vertex AI TensorBoard experiment.\nThis web application offers several tools and dashboards to visualize and\ncompare data across experiment runs, and to\n[profile model training performance](/vertex-ai/docs/training/tensorboard-profiler).\n\nTo learn more about each dashboard, see the\n[README](https://github.com/tensorflow/tensorboard/blob/master/README.md#the-visualizations)\nfor \"The Visualization\".\n\nAccess a Vertex AI TensorBoard web app\n--------------------------------------\n\nThe Vertex AI TensorBoard web app is accessed using the\nfollowing address format:\n\n`https://`\u003cvar translate=\"no\"\u003eREGION\u003c/var\u003e`.tensorboard.googleusercontent.com/experiment/projects+`\u003cvar translate=\"no\"\u003ePROJECT_NAME_OR_ID\u003c/var\u003e`+locations+`\u003cvar translate=\"no\"\u003eREGION\u003c/var\u003e`+tensorboards+`\u003cvar translate=\"no\"\u003eTENSORBOARD_ID\u003c/var\u003e`+experiments+`\u003cvar translate=\"no\"\u003eEXPERIMENT_NAME\u003c/var\u003e\n\n- \u003cvar translate=\"no\"\u003eREGION\u003c/var\u003e: the region your TensorBoard is located in\n- \u003cvar translate=\"no\"\u003ePROJECT_ID\u003c/var\u003e: the project your TensorBoard was created in\n- \u003cvar translate=\"no\"\u003eTENSORBOARD_ID\u003c/var\u003e: the [TensorBoard ID](/vertex-ai/docs/experiments/tensorboard-setup#tensorboard_id)\n- \u003cvar translate=\"no\"\u003eEXPERIMENT_NAME\u003c/var\u003e: the name of the TensorBoard experiment\n\nThis link can be shared and used by anyone who has permission.\n\nYou can navigate here directly by inserting the appropriate values,\nor by following a link available in the following ways.\n\n### gcloud CLI and Vertex AI SDK for Python\n\nWhen creating or uploading logs to a Vertex AI TensorBoard, the\ngcloud CLI and SDK output a link to the Vertex AI TensorBoard\ninstance.\n\n### Google Cloud console\n\nYou can access the TensorBoard web app using the Google Cloud console. \n\n### Experiments\n\nIf your Vertex AI TensorBoard is associated with a Vertex AI\nExperiment, which is automatic when a\n[default instance](/vertex-ai/docs/experiments/tensorboard-setup#use_default_instance)\nis involved, you can access the TensorBoard web app directly from the Experiments page.\n\n### Custom training\n\nIf you used Vertex AI TensorBoard with\n[custom training](/vertex-ai/docs/experiments/tensorboard-training)\nyou can access the TensorBoard web app directly from the custom training job.\n\n1. In the Vertex AI section of the Google Cloud console, go to the **Training** page. [Go to Training](https://console.cloud.google.com/vertex-ai/training)\n2. Click **Custom Jobs** tab.\n3. Select your training job from the list of training runs. The details page appears\n4. Click the **Open TensorBoard** button. \n\n### Vertex AI Pipelines\n\nIf you used Vertex AI TensorBoard with [Vertex AI Pipelines](/vertex-ai/docs/experiments/tensorboard-with-pipelines), the TensorBoard web app can be accessed\ndirectly from the component information display.\n\n1. In the Vertex AI section of the Google Cloud console, go to the **Pipelines** page. [Go to Pipelines](https://console.cloud.google.com/vertex-ai/pipelines)\n2. From the pipelines tab, scroll or filter the pipeline list to find your pipeline.\n3. On the pipeline display select the component where TensorBoard metrics are captured, typically the training component.\n4. Select the **Open TensorBoard** button in the upper right. \n\nVisualize Vertex AI TensorBoard metrics in the Google Cloud console\n-------------------------------------------------------------------\n\nIf your TensorBoard instance is associated with a\nVertex AI Experiments, which is automatic when using a\n[default TensorBoard instance](/vertex-ai/docs/experiments/tensorboard-setup#use_default_instance), certain\nmetrics can be visualized directly in the Google Cloud console, without using\nthe web application.\n\nThe [Compare and analyze runs](/vertex-ai/docs/experiments/compare-analyze-runs#console-compare-analyze-runs)\ndocument describes how to view details of your experiment runs and compare\nthe experiment runs to each other in the Google Cloud console.\nTensorBoard timeseries metrics appear under the **Charts** tab in the navigation\nbar of the Vertex AI Experiments console."]]
