alibi_detect.models.tensorflow.trainer

Functions

trainer

trainer(model: keras.src.models.model.Model, loss_fn: .keras._tf_keras.keras.losses, x_train: numpy.ndarray, y_train: numpy.ndarray = None, dataset: keras.src.trainers.data_adapters.py_dataset_adapter.PyDataset = None, optimizer: .tensorflow.keras.optimizers = <class 'keras.src.optimizers.adam.Adam'>, loss_fn_kwargs: dict = None, preprocess_fn: Callable = None, epochs: int = 20, reg_loss_fn: Callable = <function <lambda> at 0x28ee82e50>, batch_size: int = 64, buffer_size: int = 1024, verbose: bool = True, log_metric: Tuple[str, ForwardRef('tf.keras.metrics')] = None, callbacks: .tensorflow.keras.callbacks = None) -> None

Train TensorFlow model.

Name	Type	Default	Description
model	keras.src.models.model.Model		Model to train.
loss_fn	.keras._tf_keras.keras.losses		Loss function used for training.
x_train	numpy.ndarray		Training data.
y_train	numpy.ndarray	None	Training labels.
dataset	keras.src.trainers.data_adapters.py_dataset_adapter.PyDataset	None	Training dataset which returns (x, y).
optimizer	.tensorflow.keras.optimizers	<class 'keras.src.optimizers.adam.Adam'>	Optimizer used for training.
loss_fn_kwargs	dict	None	Kwargs for loss function.
preprocess_fn	Callable	None	Preprocessing function applied to each training batch.
epochs	int	20	Number of training epochs.
reg_loss_fn	Callable	<function <lambda> at 0x28ee82e50>	Allows an additional regularisation term to be defined as reg_loss_fn(model)
batch_size	int	64	Batch size used for training.
buffer_size	int	1024	Maximum number of elements that will be buffered when prefetching.
verbose	bool	True	Whether to print training progress.
log_metric	Tuple[str, ForwardRef('tf.keras.metrics')]	None	Additional metrics whose progress will be displayed if verbose equals True.
callbacks	.tensorflow.keras.callbacks	None	Callbacks used during training.

Returns

• Type: None