Utilitas create_custom_training_job_from_component
mengonversi container atau komponen Python tertentu menjadi komponen yang menjalankan
tugas kustom di Vertex AI. Hal ini menyederhanakan pembuatan tugas pelatihan
kustom. Semua input dan output komponen yang disediakan akan disalin ke
operator tugas pelatihan yang dibuat.
Perhatikan bahwa utilitas ini membuat ClusterSpec, tempat primer dan semua
worker menggunakan spesifikasi yang sama, yang berarti semua parameter terkait spesifikasi
disk dan mesin akan diterapkan ke semua replika. Hal ini cocok
untuk kasus penggunaan ketika, misalnya, Anda melatih dengan
MultiWorkerMirroredStrategy atau
MirroredStrategy.
Komponen ini tidak mendukung pelatihan paket Python CustomJob, atau
pelatihan terdistribusi dengan spesifikasi kumpulan worker yang berbeda.
[[["Mudah dipahami","easyToUnderstand","thumb-up"],["Memecahkan masalah saya","solvedMyProblem","thumb-up"],["Lainnya","otherUp","thumb-up"]],[["Sulit dipahami","hardToUnderstand","thumb-down"],["Informasi atau kode contoh salah","incorrectInformationOrSampleCode","thumb-down"],["Informasi/contoh yang saya butuhkan tidak ada","missingTheInformationSamplesINeed","thumb-down"],["Masalah terjemahan","translationIssue","thumb-down"],["Lainnya","otherDown","thumb-down"]],["Terakhir diperbarui pada 2025-09-04 UTC."],[],[],null,["# CustomJob components\n\n[Custom training jobs](/vertex-ai/docs/training/create-custom-job) let you run your custom machine\nlearning (ML) training code in Vertex AI.\n\n`CustomTrainingJobOp`\n---------------------\n\nThe [`CustomTrainingJobOp`](https://google-cloud-pipeline-components.readthedocs.io/en/google-cloud-pipeline-components-2.19.0/api/v1/custom_job.html#v1.custom_job.CustomTrainingJobOp) component exposes the full functionalities of the [`CustomJob` resource](/vertex-ai/docs/reference/rest/v1/projects.locations.customJobs), to allow both single and distributed training using a [`ContainerSpec`](/vertex-ai/docs/reference/rest/v1/CustomJobSpec#ContainerSpec) or [`PythonPackageSpec`](/vertex-ai/docs/reference/rest/v1/CustomJobSpec#PythonPackageSpec) instance.\n\n`create_custom_training_job_from_component` function\n----------------------------------------------------\n\nThe [`create_custom_training_job_from_component`](https://google-cloud-pipeline-components.readthedocs.io/en/google-cloud-pipeline-components-2.19.0/api/v1/custom_job.html#v1.custom_job.create_custom_training_job_from_component) utility\nconverts a given container or Python component to a component that runs a\ncustom job in Vertex AI. This simplifies the creation of custom training\njobs. All inputs and outputs of the supplied component will be copied over to\nthe constructed training job operator.\n\nNote that this utility constructs a `ClusterSpec`, where the primary and all the\nworkers use the same specification, meaning all disk and machine\nspecification-related parameters will apply to all replicas. This is suitable\nfor use cases where, for example, you are training with\n[`MultiWorkerMirroredStrategy`](https://www.tensorflow.org/api_docs/python/tf/distribute/experimental/MultiWorkerMirroredStrategy) or\n[`MirroredStrategy`](https://www.tensorflow.org/api_docs/python/tf/distribute/MirroredStrategy).\n\nThis component does not support `CustomJob` Python package training, or\ndistributed training with different worker pool specs.\n\nAPI reference\n-------------\n\n- For component reference, see the [Google Cloud SDK reference for CustomJob components](https://google-cloud-pipeline-components.readthedocs.io/en/google-cloud-pipeline-components-2.19.0/api/v1/custom_job.html).\n- For Vertex AI API reference, see the [`CustomJob` resource](/vertex-ai/docs/reference/rest/v1/projects.locations.customJobs) page.\n\nVersion history and release notes\n---------------------------------\n\nTo learn more about the version history and changes to the Google Cloud Pipeline Components SDK, see the [Google Cloud Pipeline Components SDK Release Notes](https://google-cloud-pipeline-components.readthedocs.io/en/google-cloud-pipeline-components-2.19.0/release.html).\n\nTechnical support contacts\n--------------------------\n\nIf you have any questions, reach out to\n[kubeflow-pipelines-components@google.com](mailto: kubeflow-pipelines-components@google.com)."]]
