Tetap teratur dengan koleksi
Simpan dan kategorikan konten berdasarkan preferensi Anda.
Halaman ini memberikan informasi tentang cara menggunakan Cloud Logging untuk
melihat detail tentang model Vertex AI. Dengan
Logging, Anda akan melihat:
Hyperparameter model akhir sebagai key-value pair.
Hyperparameter dan nilai objek yang digunakan selama pelatihan dan penyesuaian model, serta nilai
objektif.
timestamp: Tanggal dan waktu saat model dibuat atau uji coba dijalankan.
Konten payload untuk log hyperparameter model akhir
Kolom jsonPayload untuk log hyperparameter dari model akhir berisi
kolom modelParameters. Kolom ini berisi satu entri untuk setiap model yang
berkontribusi pada model ansambel akhir. Setiap entri memiliki kolom hyperparameters,
yang isinya bergantung pada jenis model. Untuk mengetahui detailnya, lihat Daftar hyperparameter.
Konten payload untuk log hyperparameter uji coba penyesuaian
Kolom jsonPayload untuk log hyperparameter uji coba penyesuaian berisi kolom berikut:
Kolom
Jenis
Deskripsi
modelStructure
JSON
Deskripsi struktur model Vertex AI.
Kolom ini berisi kolom modelParameters. Kolom
modelParameters memiliki kolom hyperparameters,
yang isinya bergantung pada jenis model. Untuk mengetahui detailnya, lihat
Daftar hyperparameter.
trainingObjectivePoint
JSON
Objektif pengoptimalan digunakan untuk pelatihan model.
Entri ini menyertakan stempel waktu dan nilai objektif pada
saat entri log dicatat.
Daftar hyperparameter
Data hyperparameter yang diberikan dalam log berbeda untuk setiap jenis
model. Bagian berikut menjelaskan hyperparameter untuk setiap
jenis model.
Model pohon keputusan penguatan gradien
Regularisasi pohon L1
Regularisasi pohon L2
Kedalaman pohon maksimum
Jenis model: GBDT
Jumlah pohon
Kompleksitas pohon
Model jaringan saraf alur maju
Tingkat error
Mengaktifkan batchNorm (True atau False)
Mengaktifkan penyematan L1 (True atau False)
Mengaktifkan penyematan L2 (True atau False)
Mengaktifkan L1 (True atau False)
Mengaktifkan L2 (True atau False)
Mengaktifkan layerNorm (True atau False)
Mengaktifkan penyematan numerik (True atau False)
Ukuran lapisan tersembunyi
Jenis model: nn
Normalisasi kolom numerik (True atau False)
Jumlah lapisan silang
Jumlah lapisan tersembunyi
Melewati jenis koneksi (dense, disable, concat, atau slice_or_padding)
Langkah berikutnya
Setelah siap untuk membuat prediksi dengan model klasifikasi atau
regresi, Anda memiliki dua opsi:
Anda dapat mengekspor log ke BigQuery, Cloud Storage, atau
Pub/Sub. Baca artikel Merutekan log ke tujuan yang didukung
dalam dokumentasi Logging untuk mempelajari cara mengekspor log aktivitas.
[[["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 2025-08-19 UTC."],[],[],null,["# View model architecture\n\nThis page provides information about how to use Cloud Logging to\nview details about a Vertex AI model. Using\nLogging, you see:\n\n- The hyperparameters of the final model as key-value pairs.\n- The hyperparameters and object values used during model training and tuning, as well as an objective value.\n\nBy default, logs are deleted after 30 days.\n\nThe following topics are covered:\n\n1. [Viewing training logs](#training-logs).\n2. [Log fields](#log-fields).\n\n| **Note:** Model architecture logs are provided as part of the Cloud Logging service. For general information about Cloud Logging, see the [Cloud Logging](/logging/docs) documentation.\n\nBefore you begin\n----------------\n\nBefore you can view the hyperparameter logs for your model, you must\n[train it](/vertex-ai/docs/tabular-data/classification-regression/train-model).\n\nTo perform this task, you must have the following\n[permissions](/iam/docs/overview#permissions):\n\n- `logging.logServiceIndexes.list` on the project\n- `logging.logServices.list` on the project\n\nViewing training logs\n---------------------\n\nYou can use the Google Cloud console to access the hyperparameter logs of the\nfinal model and the hyperparameter logs of the tuning trials.\n\n1. In the Google Cloud console, go to the Vertex AI **Models** page.\n\n [Go to Models page](https://console.cloud.google.com/vertex-ai/models)\n2. In the **Region** drop-down, select the region where your model is located.\n\n3. From the list of models, select your model.\n\n4. Select your model's version number.\n\n5. Open the **Version Details** tab.\n\n6. To see the hyperparameter log of the final model, go to the **Model hyperparameters** row and click **Model**.\n\n 1. There is just one log entry. Expand the payload as shown below.\n For details, see [Log fields](#reading-logs).\n\n7. To see the hyperparameter log of the tuning trials, go to the **Model hyperparameters** row and click **Trials**.\n\n 1. There is one entry for each of the tuning trials. Expand the payload as\n shown below. For details, see [Log fields](#reading-logs).\n\nLog fields\n----------\n\nActivity logs are structured as described in the\n[LogEntry](/logging/docs/exported_logs#the_logentry_type) type\ndocumentation.\n\nVertex AI model logs have, among other fields:\n\n- `labels`: The `log_type` field is set to `automl_tables`.\n- `jsonPayload`: The specific details of the log entry, provided in JSON object format. For details, see [Payload contents for the hyperparameter log of the final model](#final-payload) or [Payload contents for the hyperparameter log of a tuning trial](#trial-payload).\n- `timestamp`: The date and time when the model was created or the trial was run.\n\n### Payload contents for the hyperparameter log of the final model\n\nThe `jsonPayload` field for the hyperparameter log of the final model contains a\n`modelParameters` field. This field contains one entry for each model that\ncontributes to the final ensemble model. Each entry has a `hyperparameters`\nfield, whose contents depend on the model type. For details, see [List of hyperparameters](#hps).\n\n### Payload contents for the hyperparameter log of a tuning trial\n\nThe `jsonPayload` field for the hyperparameter log of a tuning trial contains the following fields:\n\n### List of hyperparameters\n\nThe hyperparameter data provided in the logs differ for each type of\nmodel. The following sections describe the hyperparameters for each\nmodel type.\n\n#### Gradient boosted decision tree models\n\n- Tree L1 regularization\n- Tree L2 regularization\n- Max tree depth\n- Model type: `GBDT`\n- Number of trees\n- Tree complexity\n\n#### Feedforward neural network models\n\n- Dropout rate\n- Enable batchNorm (`True` or `False`)\n- Enable embedding L1 (`True` or `False`)\n- Enable embedding L2 (`True` or `False`)\n- Enable L1 (`True` or `False`)\n- Enable L2 (`True` or `False`)\n- Enable layerNorm (`True` or `False`)\n- Enable numerical embedding (`True` or `False`)\n- Hidden layer size\n- Model type: `nn`\n- Normalize numerical column (`True` or `False`)\n- Number of cross layers\n- Number of hidden layers\n- Skip connections type (`dense`, `disable`, `concat`, or `slice_or_padding`)\n\nWhat's next\n-----------\n\nOnce you're ready to make predictions with your classification or regression\nmodel, you have two options:\n\n- [Make online (real-time) predictions using your model](/vertex-ai/docs/tabular-data/classification-regression/get-online-predictions).\n- [Get batch predictions directly from your model](/vertex-ai/docs/tabular-data/classification-regression/get-batch-predictions).\n\nAdditionally, you can:\n\n- [Evaluate your model](/vertex-ai/docs/tabular-data/classification-regression/evaluate-model).\n- [Review general information about Cloud Logging](/logging/docs).\n- You can export your logs to BigQuery, Cloud Storage, or Pub/Sub. Read [Route logs to supported destinations](/logging/docs/export/configure_export_v2) in the Logging documentation to learn how to export activity logs."]]
