[[["容易理解","easyToUnderstand","thumb-up"],["確實解決了我的問題","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["難以理解","hardToUnderstand","thumb-down"],["資訊或程式碼範例有誤","incorrectInformationOrSampleCode","thumb-down"],["缺少我需要的資訊/範例","missingTheInformationSamplesINeed","thumb-down"],["翻譯問題","translationIssue","thumb-down"],["其他","otherDown","thumb-down"]],["上次更新時間：2025-09-04 (世界標準時間)。"],[],[],null,["# Use custom models to tailor machine translations\n\n| Starting on June 30, 2025, Translation Hub is officially deprecated and will no longer be supported. You can continue using Translation Hub until on June 30, 2026, when it will be shut down.\n\n\u003cbr /\u003e\n\nBy default, Translation Hub uses the Google Neural Machine Translation\n(NMT) model to translate documents, which is suited for generic translation\ntasks. In cases where you have to translate a specific domain area and writing\nstyle, consider using a custom model (also known as an\nAutoML Translation model). Custom models can provide more tailored\npredictions.\n\nWhen you train a custom model, Cloud Translation starts with the general Google\nNMT model and tunes the model to fit your training data, which includes pairs of\nsentences in a source and target language.\n\nAutoML Translation\n------------------\n\nYou create and manage custom models through Cloud Translation - Advanced.\nPrepare a training dataset with your sentence pairs and then use the\nCloud Translation - Advanced API to create a custom model. Any costs that are\nassociated with training custom models are charged separately by\nCloud Translation. For more information, see the\n[Cloud Translation documentation](/translate/docs/advanced/automl-beginner).\n\nTranslation Hub automatically makes new and existing custom models\navailable for administrators to assign to portals. After you add a custom model\nto a portal, portal users can choose to use the custom model for their\ntranslations.\n\nIf you have a translation memory, you can [export](/translation-hub/docs/admin-export-data) that data and use it\nas training data, depending on the number of sentence pairs you have and their\nquality. Cloud Translation [recommends](/translate/automl/docs/prepare#data_recommendations) around 6,000 **sentence pairs**.\nIn general, higher quality data (like full sentences) results in higher quality\nmodels than more data.\n\nAdd models to portals\n---------------------\n\nAdministrators add models to portals by using the Google Cloud console. Portals\nusers can choose to use these models when they request translations.\n\n1. In the **Translation Hub** section of the Google Cloud console, go to the\n **Resources** page.\n\n [Go to the Resources page](https://console.cloud.google.com/translation-hub/resources)\n2. From the list of resources, select one or more models to add to one\n or more portals.\n\n3. Click **Assign to portals** , which opens the **Assign resource to portal**\n pane.\n\n4. From the portals field, select one or more portals to add the models to.\n\n5. Click **Assign**.\n\n On the **Resources** page, you can confirm the addition by viewing the\n **Portal names** column for each resource."]]
