The target block represents the output that we are trying to learn with our model.
After placing a target block you have to assign a feature to it; typically a single feature is used as output, be it a label (classification), a scalar (regression) or an image (autoencoders, image segmentation).
The loss function indicates the magnitude of error your model made on its prediction.
Note that some loss functions expect the model to output values within a certain range. In such cases, the last block of the model must use a compatible activation function.
Feature: The feature assigned to the Target block, which the model will try to predict.
Loss: The specific loss function to be used in training, which determines how the model performance is calculated.
Use for predictions: Whether to return the target feature in the deployed model’s predictions. Required if the model doesn’t have any Output block, optional otherwise.
Use class weights: Enables class weighting in the calculation of the loss function.
Available only when the target is a categorical feature.