IBM Cloud Docs
Managing custom language models

Managing custom language models

The customization interface includes the POST /v1/customizations method for creating a custom language model. The interface also includes the POST /v1/customizations/train method for training a custom model on the latest data from its words resource. For more information, see

The interface also includes methods for listing information about custom language models, resetting a custom model to its initial state, upgrading a custom model, and deleting a custom model. You cannot train, reset, upgrade, or delete a custom model while the service is handling another operation on that model, including adding resources to the model.

Listing custom language models

The customization interface provides two methods for listing information about the custom language models that are owned by the specified credentials:

  • The GET /v1/customizations method lists information about all custom language models or, if you specify the language parameter, about all custom language models for the specified language. If you specify a language, use the language identifier from the name of the base model, for example, en-US for a US English model.
  • The GET /v1/customizations/{customization_id} method lists information about a specified custom language model. Use this method to poll the service about the status of a training request or a request to add new words.

Both methods return the following information about a custom model:

  • customization_id identifies the custom model's Globally Unique Identifier (GUID). The GUID is used to identify the model in methods of the interface.
  • created is the date and time in Coordinated Universal Time (UTC) at which the custom model was created.
  • updated is the date and time in Coordinated Universal Time (UTC) at which the custom model was last modified.
  • language is the language of the custom model. The value matches the language identifier from the name of the base model. For example, en-US for a US English language model.
  • dialect is the dialect of the language for the custom model, which does not necessarily match the language of the custom model for previous-generation Spanish models. For more information, see the description of the dialect field in Create a custom language model.
  • owner identifies the credentials of the service instance that owns the custom model.
  • name is the name of the custom model.
  • description shows the description of the custom model, if one was provided at its creation.
  • base_model indicates the name of the language model for which the custom model was created.
  • versions provides a list of the available versions of the custom model. Each element of the array indicates a version of the base model with which the custom model can be used. Multiple versions exist only if the custom model is upgraded to a new version of its base model. Otherwise, only a single version is shown. For more information, see Listing version information for a custom model.

The method also returns a status field that indicates the state of the custom model:

  • pending indicates that the model was created. It is waiting either for valid training data (corpora, words, or grammars) to be added or for the service to finish analyzing data that was added.
  • ready indicates that the model contains valid data and is ready to be trained. If the model contains a mix of valid and invalid resources, training of the model fails unless you set the strict query parameter to false. For more information, see Training failures.
  • training indicates that the model is being trained on data.
  • available indicates that the model is trained and ready to use with a recognition request.
  • upgrading indicates that the model is being upgraded.
  • failed indicates that training of the model failed. Examine the words in the model's words resource to determine the errors that prevented the model from being trained.

Additionally, the output includes a progress field that indicates the status of a custom model's training. If you used the POST /v1/customizations/{customization_id}/train method to successfully train the model, this field has a value of 100. If the model is not completely trained and available, the field has a value of 0.

When you monitor the training or upgrading of a custom model, poll the value of the status field, not the value of the progress field. If the operation fails for any reason, the value of the status field changes to reflect the failure; the value of the progress field remains 0.

List all custom language models example

The following example includes the language query parameter to list all US English custom language models that are owned by the specified credentials:

IBM Cloud

curl -X GET -u "apikey:{apikey}" \
"{url}/v1/customizations?language=en-US"

IBM Cloud Pak for Data

curl -X GET \
--header "Authorization: Bearer {token}" \
"{url}/v1/customizations?language=en-US"

The credentials own two such models. The first model is awaiting data or is being processed by the service. The second model is fully trained and ready for use. Both custom models are based on previous-generation models; the first custom model has two available versions.

{
  "customizations": [
    {
      "customization_id": "74f4807e-b5ff-4866-824e-6bba1a84fe96",
      "created": "2016-06-01T14:21:26.894Z",
      "updated": "2020-01-18T18:42:25.324Z",
      "language": "en-US",
      "dialect": "en-US",
      "versions": [
        "en-US_BroadbandModel.v2018-07-31",
        "en-US_BroadbandModel.v2020-01-16"
      ],
      "owner": "297cfd08-330a-22ba-93ce-1a73f454dd98",
      "name": "Example model",
      "description": "Example custom language model",
      "base_model_name": "en-US_BroadbandModel",
      "status": "pending",
      "progress": 0
    },
    {
      "customization_id": "8391f918-3b76-e109-763c-b7732fae4829",
      "created": "2017-12-02T18:51:37.291Z",
      "updated": "2017-12-02T20:02:10.624Z",
      "language": "en-US",
      "dialect": "en-US",
      "versions": [
        "en-US_BroadbandModel.v2017-11-15"
      ],
      "owner": "297cfd08-330a-22ba-93ce-1a73f454dd98",
      "name": "Example model two",
      "description": "Example custom language model two",
      "base_model_name": "en-US_BroadbandModel",
      "status": "available",
      "progress": 100
    }
  ]
}

List a specific custom language model example

The following example returns information about the custom model that has the specified customization ID:

IBM Cloud

curl -X GET -u "apikey:{apikey}" \
"{url}/v1/customizations/{customization_id}"

IBM Cloud Pak for Data

curl -X GET \
--header "Authorization: Bearer {token}" \
"{url}/v1/customizations/{customization_id}"

The response duplicates the information from the previous example:

{
  "customization_id": "74f4807e-b5ff-4866-824e-6bba1a84fe96",
  "created": "2016-06-01T14:21:26.894Z",
  "updated": "2020-01-18T18:42:25.324Z",
  "language": "en-US",
  "dialect": "en-US",
  "versions": [
    "en-US_BroadbandModel.v2018-07-31",
    "en-US_BroadbandModel.v2020-01-16"
  ],
  "owner": "297cfd08-330a-22ba-93ce-1a73f454dd98",
  "name": "Example model",
  "description": "Example custom language model",
  "base_model_name": "en-US_BroadbandModel",
  "status": "pending",
  "progress": 0
}

Resetting a custom language model

Use the POST /v1/customizations/{customization_id}/reset method to reset a custom model. Resetting a model removes all of the corpora and words from the model, initializing the model to its state at creation. The method does not delete the model itself or metadata such as its name and language. However, when you reset a model, its words resource is empty and must be re-created by adding corpora and words.

Reset a custom language model example

The following example resets the custom model with the specified customization ID:

IBM Cloud

curl -X POST -u "apikey:{apikey}" \
"{url}/v1/customizations/{customization_id}/reset"

IBM Cloud Pak for Data

curl -X POST \
--header "Authorization: Bearer {token}" \
"{url}/v1/customizations/{customization_id}/reset"

Deleting a custom language model

Use the DELETE /v1/customizations/{customization_id} method to delete a custom language model that you no longer need. The method deletes all corpora and words that are associated with the custom model and the model itself. Use this method with caution: a custom model and its data cannot be reclaimed after you delete the model.

Delete a custom language model example

The following example deletes the custom model with the specified customization ID:

IBM Cloud

curl -X DELETE -u "apikey:{apikey}" \
"{url}/v1/customizations/{customization_id}"

IBM Cloud Pak for Data

curl -X DELETE \
--header "Authorization: Bearer {token}" \
"{url}/v1/customizations/{customization_id}"