frameworks.ultralytics.services.model.trainer

trainer

Classes:

Name	Description
UltralyticsModelTrainer	Trainer class for managing the training lifecycle of a model using the Ultralytics framework.

UltralyticsModelTrainer(model, experiment)

Trainer class for managing the training lifecycle of a model using the Ultralytics framework.

This class handles the setup of callbacks, integration with the Picsellia experiment system, and executes training using specific hyperparameters and augmentations.

Parameters:

Name	Type	Description	Default
model	UltralyticsModel	The model to be trained.	required
experiment	Experiment	The experiment instance used for logging and tracking results.	required

Methods:

Name	Description
train_model	Executes training on the model using the provided datasets and parameters.

Attributes:

Name	Type	Description
model
experiment

model = model instance-attribute

experiment = experiment instance-attribute

train_model(dataset_collection, hyperparameters, augmentation_parameters, callbacks=UltralyticsCallbacks)

Executes training on the model using the provided datasets and parameters.

Parameters:

Name	Type	Description	Default
dataset_collection	DatasetCollection	Dataset used for training and validation.	required
hyperparameters	UltralyticsHyperParameters	Training configuration including learning rate, epochs, etc.	required
augmentation_parameters	UltralyticsAugmentationParameters	Data augmentation parameters.	required
callbacks	type	Callback class to use for logging and event handling.	UltralyticsCallbacks

Returns:

Name	Type	Description
UltralyticsModel	UltralyticsModel	The trained model instance.