luz0.5.1 package

Create an accelerator

Creates a dataloader from its input

Context object

Context object

Evaluates a fitted model on a dataset

Fit a nn_module

Get metrics from the object

Learning Rate Finder

Resume training callback

CSV logger callback

Early stopping callback

Gradient clipping callback

Interrupt callback

Keep the best model

Learning rate scheduler callback

Metrics callback

Automatic Mixed Precision callback

Mixup callback

Checkpoints model weights

Profile callback

Progress callback

Allow resume model training from a specific checkpoint

tfevents callback

Train-eval callback

Create a new callback

Loads a checkpoint

Loads model weights into a fitted object.

Load trained model

Accuracy

Binary accuracy with logits

Binary accuracy

Computes the area under the ROC

Mean absolute error

Mean squared error

Computes the multi-class AUROC

Root mean squared error

Creates a metric set

Creates a new luz metric

Saves luz objects to disk

Loss to be used with callbacks_mixup().

Mixup logic

Pipe operator

Create predictions for a fitted model

Objects exported from other packages

Set hyper-parameter of a module

Set optimizer hyper-parameters

Set's up a nn_module to use with luz