Losses

binary_dice_loss(predictions, targets, *[, ...])	Binary Dice Loss convenience function.
convex_kl_divergence(log_predictions, targets)
cosine_distance(predictions, targets, *[, ...])	Computes the cosine distance between targets and predictions.
cosine_similarity(predictions, targets, *[, ...])	Computes the cosine similarity between targets and predictions.
ctc_loss(logits, logit_paddings, labels, ...)	Computes CTC loss.
ctc_loss_with_forward_probs(logits, ...[, ...])	Computes CTC loss and CTC forward-probabilities.
dice_loss(predictions, targets, *[, ...])	Computes the Dice Loss for multi-class segmentation.
generalized_kl_divergence(log_predictions, ...)	Computes the generalized Kullback-Leibler divergence loss.
hinge_loss(predictor_outputs, targets)	Computes the hinge loss for binary classification.
huber_loss(predictions[, targets, delta])	Huber loss, similar to L2 loss close to zero, L1 loss away from zero.
kl_divergence(log_predictions, targets[, ...])	Computes the Kullback-Leibler divergence (relative entropy) loss.
kl_divergence_with_log_targets(...[, axis, ...])	Computes the Kullback-Leibler divergence (relative entropy) loss.
l2_loss(predictions[, targets])	Calculates the L2 loss for a set of predictions.
log_cosh(predictions[, targets])	Calculates the log-cosh loss for a set of predictions.
make_fenchel_young_loss(max_fun)	Creates a Fenchel-Young loss from a max function.
multiclass_generalized_dice_loss(...[, ...])	Computes Multiclass Generalized Dice Loss with automatic class weighting.
multiclass_hinge_loss(scores, labels)	Multiclass hinge loss.
multiclass_perceptron_loss(scores, labels)	Multiclass perceptron loss.
multiclass_sparsemax_loss(scores, labels)	Multiclass sparsemax loss.
ntxent(embeddings, labels[, temperature])	Normalized temperature scaled cross entropy loss (NT-Xent).
perceptron_loss(predictor_outputs, targets)	Binary perceptron loss.
poly_loss_cross_entropy(logits, labels, *[, ...])	Computes PolyLoss between logits and labels.
ranking_softmax_loss(logits, labels, *, ...)	Ranking softmax loss.
safe_softmax_cross_entropy(logits, labels)	Computes the softmax cross entropy between sets of logits and labels.
sigmoid_binary_cross_entropy(logits, labels)	Computes element-wise sigmoid cross entropy given logits and labels.
sigmoid_focal_loss(logits, labels, *[, ...])	Sigmoid focal loss with numerical stability improvements.
smooth_labels(labels, alpha, *[, axis, where])	Apply label smoothing.
softmax_cross_entropy(logits, labels[, ...])	Computes the softmax cross entropy between sets of logits and labels.
softmax_cross_entropy_with_integer_labels(...)	Computes softmax cross entropy between the logits and integer labels.
sparsemax_loss(logits, labels)	Binary sparsemax loss.
squared_error(predictions[, targets])	Calculates the squared error for a set of predictions.
triplet_margin_loss(anchors, positives, ...)	Returns the triplet loss for a batch of embeddings.