ImageModelSettingsClassification Class
Model settings for AutoML Image Classification tasks.
Constructor
ImageModelSettingsClassification(*, advanced_settings: str | None = None, ams_gradient: bool | None = None, beta1: float | None = None, beta2: float | None = None, checkpoint_frequency: int | None = None, checkpoint_run_id: str | None = None, distributed: bool | None = None, early_stopping: bool | None = None, early_stopping_delay: int | None = None, early_stopping_patience: int | None = None, enable_onnx_normalization: bool | None = None, evaluation_frequency: int | None = None, gradient_accumulation_step: int | None = None, layers_to_freeze: int | None = None, learning_rate: float | None = None, learning_rate_scheduler: LearningRateScheduler | None = None, model_name: str | None = None, momentum: float | None = None, nesterov: bool | None = None, number_of_epochs: int | None = None, number_of_workers: int | None = None, optimizer: StochasticOptimizer | None = None, random_seed: int | None = None, step_lr_gamma: float | None = None, step_lr_step_size: int | None = None, training_batch_size: int | None = None, validation_batch_size: int | None = None, warmup_cosine_lr_cycles: float | None = None, warmup_cosine_lr_warmup_epochs: int | None = None, weight_decay: float | None = None, training_crop_size: int | None = None, validation_crop_size: int | None = None, validation_resize_size: int | None = None, weighted_loss: int | None = None, **kwargs: Any)Parameters
| Name | Description |
|---|---|
|
advanced_settings
Required
|
Settings for advanced scenarios. |
|
ams_gradient
Required
|
Enable AMSGrad when optimizer is 'adam' or 'adamw'. |
|
beta1
Required
|
Value of 'beta1' when optimizer is 'adam' or 'adamw'. Must be a float in the range [0, 1]. |
|
beta2
Required
|
Value of 'beta2' when optimizer is 'adam' or 'adamw'. Must be a float in the range [0, 1]. |
|
checkpoint_frequency
Required
|
Frequency to store model checkpoints. Must be a positive integer. |
|
checkpoint_run_id
Required
|
The id of a previous run that has a pretrained checkpoint for incremental training. |
|
distributed
Required
|
Whether to use distributed training. |
|
early_stopping
Required
|
Enable early stopping logic during training. |
|
early_stopping_delay
Required
|
Minimum number of epochs or validation evaluations to wait before primary metric improvement is tracked for early stopping. Must be a positive integer. |
|
early_stopping_patience
Required
|
Minimum number of epochs or validation evaluations with no primary metric improvement before the run is stopped. Must be a positive integer. |
|
enable_onnx_normalization
Required
|
Enable normalization when exporting ONNX model. |
|
evaluation_frequency
Required
|
Frequency to evaluate validation dataset to get metric scores. Must be a positive integer. |
|
gradient_accumulation_step
Required
|
Gradient accumulation means running a configured number of "GradAccumulationStep" steps without updating the model weights while accumulating the gradients of those steps, and then using the accumulated gradients to compute the weight updates. Must be a positive integer. |
|
layers_to_freeze
Required
|
Number of layers to freeze for the model. Must be a positive integer. For instance, passing 2 as value for 'seresnext' means freezing layer0 and layer1. For a full list of models supported and details on layer freeze, please see: https://free.blessedness.top/azure/machine-learning/how-to-auto-train-image-models. |
|
learning_rate
Required
|
Initial learning rate. Must be a float in the range [0, 1]. |
|
learning_rate_scheduler
Required
|
Type of learning rate scheduler. Must be 'warmup_cosine' or 'step'. Possible values include: "None", "WarmupCosine", "Step". |
|
model_name
Required
|
Name of the model to use for training. For more information on the available models please visit the official documentation: https://free.blessedness.top/azure/machine-learning/how-to-auto-train-image-models. |
|
momentum
Required
|
Value of momentum when optimizer is 'sgd'. Must be a float in the range [0, 1]. |
|
nesterov
Required
|
Enable nesterov when optimizer is 'sgd'. |
|
number_of_epochs
Required
|
Number of training epochs. Must be a positive integer. |
|
number_of_workers
Required
|
Number of data loader workers. Must be a non-negative integer. |
|
optimizer
Required
|
Type of optimizer. Possible values include: "None", "Sgd", "Adam", "Adamw". |
|
random_seed
Required
|
Random seed to be used when using deterministic training. |
|
step_lr_gamma
Required
|
Value of gamma when learning rate scheduler is 'step'. Must be a float in the range [0, 1]. |
|
step_lr_step_size
Required
|
Value of step size when learning rate scheduler is 'step'. Must be a positive integer. |
|
training_batch_size
Required
|
Training batch size. Must be a positive integer. |
|
validation_batch_size
Required
|
Validation batch size. Must be a positive integer. |
|
warmup_cosine_lr_cycles
Required
|
Value of cosine cycle when learning rate scheduler is 'warmup_cosine'. Must be a float in the range [0, 1]. |
|
warmup_cosine_lr_warmup_epochs
Required
|
Value of warmup epochs when learning rate scheduler is 'warmup_cosine'. Must be a positive integer. |
|
weight_decay
Required
|
Value of weight decay when optimizer is 'sgd', 'adam', or 'adamw'. Must be a float in the range[0, 1]. |
|
training_crop_size
Required
|
Image crop size that is input to the neural network for the training dataset. Must be a positive integer. |
|
validation_crop_size
Required
|
Image crop size that is input to the neural network for the validation dataset. Must be a positive integer. |
|
validation_resize_size
Required
|
Image size to which to resize before cropping for validation dataset. Must be a positive integer. |
|
weighted_loss
Required
|
Weighted loss. The accepted values are 0 for no weighted loss. 1 for weighted loss with sqrt.(class_weights). 2 for weighted loss with class_weights. Must be 0 or 1 or 2. |
Keyword-Only Parameters
| Name | Description |
|---|---|
|
advanced_settings
|
Default value: None
|
|
ams_gradient
|
Default value: None
|
|
beta1
|
Default value: None
|
|
beta2
|
Default value: None
|
|
checkpoint_frequency
|
Default value: None
|
|
checkpoint_run_id
|
Default value: None
|
|
distributed
|
Default value: None
|
|
early_stopping
|
Default value: None
|
|
early_stopping_delay
|
Default value: None
|
|
early_stopping_patience
|
Default value: None
|
|
enable_onnx_normalization
|
Default value: None
|
|
evaluation_frequency
|
Default value: None
|
|
gradient_accumulation_step
|
Default value: None
|
|
layers_to_freeze
|
Default value: None
|
|
learning_rate
|
Default value: None
|
|
learning_rate_scheduler
|
Default value: None
|
|
model_name
|
Default value: None
|
|
momentum
|
Default value: None
|
|
nesterov
|
Default value: None
|
|
number_of_epochs
|
Default value: None
|
|
number_of_workers
|
Default value: None
|
|
optimizer
|
Default value: None
|
|
random_seed
|
Default value: None
|
|
step_lr_gamma
|
Default value: None
|
|
step_lr_step_size
|
Default value: None
|
|
training_batch_size
|
Default value: None
|
|
validation_batch_size
|
Default value: None
|
|
warmup_cosine_lr_cycles
|
Default value: None
|
|
warmup_cosine_lr_warmup_epochs
|
Default value: None
|
|
weight_decay
|
Default value: None
|
|
training_crop_size
|
Default value: None
|
|
validation_crop_size
|
Default value: None
|
|
validation_resize_size
|
Default value: None
|
|
weighted_loss
|
Default value: None
|
Examples
Defining the automl image classification model settings.
from azure.ai.ml import automl
image_classification_model_settings = automl.ImageModelSettingsClassification(
checkpoint_frequency=5,
early_stopping=False,
gradient_accumulation_step=2,
)
