Skip to contents

Use this crossentropy loss function when there are two or more label classes. We expect labels to be provided as integers. If you want to provide labels using one-hot representation, please use CategoricalCrossentropy loss. There should be # classes floating point values per feature for y_pred and a single floating point value per feature for y_true.

In the snippet below, there is a single floating point value per example for y_true and num_classes floating pointing values per example for y_pred. The shape of y_true is [batch_size] and the shape of y_pred is [batch_size, num_classes].


  from_logits = FALSE,
  ignore_class = NULL,
  axis = -1L,
  reduction = "sum_over_batch_size",
  name = "sparse_categorical_crossentropy"



Ground truth values.


The predicted values.


Whether y_pred is expected to be a logits tensor. By default, we assume that y_pred encodes a probability distribution.


Optional integer. The ID of a class to be ignored during loss computation. This is useful, for example, in segmentation problems featuring a "void" class (commonly -1 or 255) in segmentation maps. By default (ignore_class=NULL), all classes are considered.


Defaults to -1. The dimension along which the entropy is computed.


For forward/backward compatability.


Type of reduction to apply to the loss. In almost all cases this should be "sum_over_batch_size". Supported options are "sum", "sum_over_batch_size" or NULL.


Optional name for the loss instance.


Sparse categorical crossentropy loss value.


y_true <- c(1, 2)
y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))
loss <- loss_sparse_categorical_crossentropy(y_true, y_pred)

## tf.Tensor([0.05129339 2.30258509], shape=(2), dtype=float64)

y_true <- c(1, 2)
y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))
# Using 'auto'/'sum_over_batch_size' reduction type.
scce <- loss_sparse_categorical_crossentropy()
scce(op_array(y_true), op_array(y_pred))

## tf.Tensor(1.1769392, shape=(), dtype=float32)

# 1.177

# Calling with 'sample_weight'.
scce(op_array(y_true), op_array(y_pred), sample_weight = op_array(c(0.3, 0.7)))

## tf.Tensor(0.8135988, shape=(), dtype=float32)

# Using 'sum' reduction type.
scce <- loss_sparse_categorical_crossentropy(reduction="sum")
scce(op_array(y_true), op_array(y_pred))

## tf.Tensor(2.3538785, shape=(), dtype=float32)

# 2.354

# Using 'none' reduction type.
scce <- loss_sparse_categorical_crossentropy(reduction=NULL)
scce(op_array(y_true), op_array(y_pred))

## tf.Tensor([0.05129344 2.3025851 ], shape=(2), dtype=float32)

# array([0.0513, 2.303], dtype=float32)

Usage with the compile() API:

model %>% compile(optimizer = 'sgd',
                  loss = loss_sparse_categorical_crossentropy())