Class AutoMLImageTrainingJob (1.62.0)

Optional. The user-defined name of the training pipeline. The name must contain 128 or fewer UTF-8 characters.

The type of prediction the model produces. Valid values are: classification to predict one of multiple target values that are selected from each row. object_detection to predict a value based on its relation to other values. object_detection is available only for columns that contain semantically numeric values. For example, columns might contain numeric values stored as an integer, float, or string.

Required. If false, a single-label (multi-class) model is trained (i.e. assuming that for each image just up to one annotation may be applicable). If true, a multi-label model is trained. If you pass in true, it's assumed multiple annotations might apply to each image. Multi-label models are supported only when prediction_type is set to classification. If multi_label is true and prediction_type is object_detection, then then multi_label is ignored. The default value is false.

Required. The type of model to create. The following are the valid values: CLOUD - The default for image classification. This model type is deigned to work in Google Cloud and can't be exported. CLOUD_1 - This model type is deigned to work in Google Cloud and can't be exported. This model type is expected to have higher prediction accuracy than CLOUD. CLOUD_HIGH_ACCURACY_1 - The default for image object detection. This model type is deigned to work in Google Cloud and can't be exported. It's designed to have a higher latency and a higher prediction quality than other cloud model types. CLOUD_LOW_LATENCY_1 - This model type is deigned to work in Google Cloud and can't be exported. It's also designed to have low latency, but might might lower prediction quality than other cloud model types. MOBILE_TF_LOW_LATENCY_1 - This model type is deigned to work in Google Cloud and can be exported as a TensorFlow or Core ML model for use on a mobile or edge device. It's also designed to have to have low latency, but might have lower prediction quality than other mobile models types. MOBILE_TF_VERSATILE_1 - This model type is deigned to work in Google Cloud and can be exported as a TensorFlow or Core ML model for use on a mobile or edge device. MOBILE_TF_HIGH_ACCURACY_1 - This model type is deigned to work in Google Cloud and can be exported as a TensorFlow or Core ML model for use on a mobile or edge device. It's designed to have a higher latency, but should also have a higher prediction quality than other mobile model types. EFFICIENTNET - This model type is available in the Vertex Model Garden image classification training with customizable hyperparameters. It's deigned to work in Google Cloud and can't be exported. VIT - This model type is available in the Vertex Model Garden image classification training with customizable hyperparameters. Best tailored to be used within Google Cloud, and cannot be exported externally. MAXVIT - This model type is available in the Vertex Model Garden image classification training with customizable hyperparameters. It's deigned to work in Google Cloud and can't be exported. COCA - This model type is available in the Vertex Model Garden image classification training with customizable hyperparameters. It's deigned to work in Google Cloud and can't be exported. SPINENET - This model type is available in the Vertex Model Garden image object detection training with customizable hyperparameters. It's deigned to work in Google Cloud and can't be exported. YOLO - This model type is available in the Vertex Model Garden image object detection training with customizable hyperparameters. It's deigned to work in Google Cloud and can't be exported.

Optional[models.Model] = Optional. You can specify a base_model for image classification models only. If it is specified, the new model is trained based on the base_model. Otherwise, the new model is trained from scratch. The base model must be in the same Google Project and region as the new model to train, and it must have the same model_type.

Optional[models.Model] = Optional. You can specify an incremental_train_base_model for image classification models and object detection models only. If specified, the new model is incrementally trained using an existing model as the starting point. This can reduce the training time. If not specified, the new model is trained from scratch. The model specified in incremental_train_base_model model must be in the same Google Project and region as the new model to train, and it must have the same prediction_type and model_type.

Optional. The Google Cloud region where this where the training runs. This region overrides the region that was set by aiplatform.init.

Optional. Location to run training in. Overrides location set in aiplatform.init.

Optional. The credentials that are used to train the model. These credentials override the credentials set by aiplatform.init.

Optional. Labels with user-defined metadata to organize your training pipelines. The maximum length of a key and value is 64 unicode characters. Labels and keys can contain only lowercase letters, numeric characters, underscores, and dashes. International characters are allowed. For more information and examples of using labels, see Using labels to organize Google Cloud Platform resources.

Optional. The Cloud KMS resource identifier of the customer managed encryption key used to protect the training pipeline. The key has the following format: projects/my-project/locations/my-region/keyRings/my-kr/cryptoKeys/my-key. The key needs to be in the same region as where the compute resource is created. If set, the key secures the training pipeline and overrides the key set using aiplatform.init. The model trained by this training pipeline is also secured by this key if model_encryption_spec_key_name is not set separately.

Optional. The Cloud KMS resource identifier of the customer managed encryption key used to protect the model. The key has the following format: projects/my-project/locations/my-region/keyRings/my-kr/cryptoKeys/my-key. The key needs to be in the same region as where the compute resource is created. If set, the key secures the trained model and overrides the key set using aiplatform.init.

Optional[str] = Optional. The field is reserved for Model Garden model training and is based on the provided pre-trained model checkpoint. checkpoint_name is needed only when model_type is set to EFFICIENTNET, VIT, COCA, SPINENET, or YOLO.

Optional[Dict[str, str]] = None, Optional. A field that's used with the Model Garden model training when passing customized configs for the trainer. The following is an example that uses all trainer_config parameters:

py trainer_config = { 'global_batch_size': '8', 'learning_rate': '0.001', 'optimizer_type': 'sgd', 'optimizer_momentum': '0.9', 'train_steps': '10000', 'accelerator_count': '1', 'anchor_size': '8', -- IOD only } 

Dict[str, str] Required. A dictionary that represents metrics use for optimization. The dictionary key is the metric_id that's reported by your training job and can be 'loss' or 'accuracy'. The dictionary value is the optimization goal of the metric, and can be 'minimize' or 'maximize'. The following is an example of a metric_spec: metric_spec = {'loss': 'minimize', 'accuracy': 'maximize'}. metric_spec is required for only Model Garden models when model_type is EFFICIENTNET, VIT, COCA, SPINENET, or YOLO.

Required. A dictionary representing parameters to use for optimization. The dictionary key is the metric_id that's reported by your training job as a command line keyword argument. The dictionary value is the parameter specification of the metric. The following is an example of how to specify a parameter_spec:

py from google.cloud.aiplatform import hpt as hpt parameter_spec = { 'learning_rate': hpt.DoubleParameterSpec(min=1e-7, max=1, scale='linear'), } 

The search algorithm that's specified for the study. The search algorithm can be one of the following: None - If you don't specify an algorithm, your job uses the default Vertex AI algorithm. The default algorithm applies Bayesian optimization to arrive at the optimal solution with a more effective search over the parameter space. 'grid' - A simple grid search within the feasible space. This option is particularly useful if you want to specify a quantity of trials that is greater than the number of points in the feasible space. In these cases, if you don't specify a grid search, the Vertex AI default algorithm might generate duplicate suggestions. To use grid search, all parameter specifications must be IntegerParameterSpec, CategoricalParameterSpec, or DiscreteParameterSpec. 'random' - A simple random search within the feasible space. search_algorithm is required for only Model Garden models when model_type is EFFICIENTNET, VIT, COCA, SPINENET, or YOLO`.

Indicates which measurement to use when the service selects the final measurement from previously reported intermediate measurements. measurement_selection can be best or last. If you expect your measurements to monotonically improve, then choose last. If your system can over-train and you expect the performance to improve and then start to decline, then choose best. If your measurements are significantly noisy or not reproducible, then best is likely to be over-optimistic, so in this case last is the preferred option. measurement_selection is required for only Model Garden models when model_type is EFFICIENTNET, VIT, COCA, SPINENET, or YOLO.

Whether to execute this deletion synchronously. If False, this method will be executed in concurrent Future and any downstream object will be immediately returned and synced when the Future has completed.

Required. A fully-qualified resource name or ID.

Optional. The name of the Google Cloud project to retrieve the training job from. This overrides the project that was set by aiplatform.init.

Optional. The Google Cloud region from where the training job is retrieved. This region overrides the region that was set by aiplatform.init.

Optional. The credentials that are used to upload this model. These credentials override the credentials set by aiplatform.init.

If set to true, this method runs synchronously. If false, this method runs asynchronously.

Optional. An expression for filtering the results of the request. For field names, snake_case and camelCase are supported.

Optional. A comma-separated list of fields used to sort the returned traing job resources. The defauilt sorting order is ascending. To sort by a field name in descending order, use desc after the field name. The following fields are supported: display_name, create_time, update_time.

Optional. The name of the Google Cloud project to which to retrieve the list of training job resources. This overrides the project that was set by aiplatform.init.

Optional. The Google Cloud region from where the training job resources are retrieved. This region overrides the region that was set by aiplatform.init.

Optional. The credentials that are used to retrieve list. These credentials override the credentials set by aiplatform.init.

Required. The dataset within the same Project from which data will be used to train the Model. The Dataset must use schema compatible with Model being trained, and what is compatible should be described in the used TrainingPipeline's [training_task_definition] [google.cloud.aiplatform.v1beta1.TrainingPipeline.training_task_definition]. For tabular Datasets, all their data is exported to training, to pick and choose from.

Optional. The fraction of the input data that is to be used to train the Model. This is ignored if Dataset is not provided.

Optional. The fraction of the input data that is to be used to validate the Model. This is ignored if Dataset is not provided.

Optional. The fraction of the input data that is to be used to evaluate the Model. This is ignored if Dataset is not provided.

Optional. A filter on DataItems of the Dataset. DataItems that match this filter are used to train the Model. A filter with same syntax as the one used in DatasetService.ListDataItems may be used. If a single DataItem is matched by more than one of the FilterSplit filters, then it is assigned to the first set that applies to it in the training, validation, test order. This is ignored if Dataset is not provided.

Optional. A filter on DataItems of the Dataset. DataItems that match this filter are used to validate the Model. A filter with same syntax as the one used in DatasetService.ListDataItems may be used. If a single DataItem is matched by more than one of the FilterSplit filters, then it is assigned to the first set that applies to it in the training, validation, test order. This is ignored if Dataset is not provided.

Optional. A filter on DataItems of the Dataset. DataItems that match this filter are used to test the Model. A filter with same syntax as the one used in DatasetService.ListDataItems may be used. If a single DataItem is matched by more than one of the FilterSplit filters, then it is assigned to the first set that applies to it in the training, validation, test order. This is ignored if Dataset is not provided.

Optional. The train budget of creating this Model, expressed in milli node hours i.e. 1,000 value in this field means 1 node hour. Defaults by prediction_type: classification - For Cloud models the budget must be: 8,000 - 800,000 milli node hours (inclusive). The default value is 192,000 which represents one day in wall time, assuming 8 nodes are used. object_detection - For Cloud models the budget must be: 20,000 - 900,000 milli node hours (inclusive). The default value is 216,000 which represents one day in wall time, assuming 9 nodes are used. The training cost of the model will not exceed this budget. The final cost will be attempted to be close to the budget, though may end up being (even) noticeably smaller - at the backend's discretion. This especially may happen when further model training ceases to provide any improvements. If the budget is set to a value known to be insufficient to train a Model for the given training set, the training won't be attempted and will error.

Optional. The display name of the managed Vertex AI Model. The name can be up to 128 characters long and can be consist of any UTF-8 characters. If not provided upon creation, the job's display_name is used.

Optional. The labels with user-defined metadata to organize your Models. Label keys and values can be no longer than 64 characters (Unicode codepoints), can only contain lowercase letters, numeric characters, underscores and dashes. International characters are allowed. See https://goo.gl/xmQnxf for more information and examples of labels.

Optional. The ID to use for the Model produced by this job, which will become the final component of the model resource name. This value may be up to 63 characters, and valid characters are [a-z0-9_-]. The first character cannot be a number or hyphen.

Optional. The resource name or model ID of an existing model. The new model uploaded by this job will be a version of parent_model. Only set this field when training a new version of an existing model.

Optional. When set to True, the newly uploaded model version will automatically have alias "default" included. Subsequent uses of the model produced by this job without a version specified will use this "default" version. When set to False, the "default" alias will not be moved. Actions targeting the model version produced by this job will need to specifically reference this version by ID or alias. New model uploads, i.e. version 1, will always be "default" aliased.

Optional. User provided version aliases so that the model version uploaded by this job can be referenced via alias instead of auto-generated version ID. A default version alias will be created for the first version of the model. The format is a-z][a-zA-Z0-9-]{0,126}[a-z0-9]

Optional. The description of the model version being uploaded by this job.

bool = False Required. If true, the entire budget is used. This disables the early stopping feature. By default, the early stopping feature is enabled, which means that training might stop before the entire training budget has been used, if further training does no longer brings significant improvement to the model.

bool = True Whether to execute this method synchronously. If False, this method will be executed in concurrent Future and any downstream object will be immediately returned and synced when the Future has completed.

Optional. The timeout for the create request in seconds.