tf.contrib.losses.compute_weighted_loss(*args, **kwargs)

tf.contrib.losses.compute_weighted_loss(*args, **kwargs)

See the guide: Losses (contrib) > Loss operations for use in neural networks.

Computes the weighted loss. (deprecated)

THIS FUNCTION IS DEPRECATED. It will be removed after 2016-12-30. Instructions for updating: Use tf.losses.compute_weighted_loss instead.

Args:

  • losses: A tensor of size [batch_size, d1, ... dN].
  • weights: A tensor of size [1] or [batch_size, d1, ... dK] where K < N.
  • scope: the scope for the operations performed in computing the loss.

Returns:

A scalar Tensor that returns the weighted loss.

Raises:

  • ValueError: If weights is None or the shape is not compatible with losses, or if the number of dimensions (rank) of either losses or weights is missing.

Defined in tensorflow/python/util/deprecation.py.