Join us at TensorFlow World, Oct 28-31. Use code TF20 for 20% off select passes. Register now

tf.nn.elu

TensorFlow 2.0 version

Defined in generated file: python/ops/gen_nn_ops.py

Computes exponential linear: exp(features) - 1 if < 0, features otherwise.

Aliases:

  • tf.compat.v1.nn.elu
  • tf.compat.v2.nn.elu
tf.nn.elu(
    features,
    name=None
)

See Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)

Args:

  • features: A Tensor. Must be one of the following types: half, bfloat16, float32, float64.
  • name: A name for the operation (optional).

Returns:

A Tensor. Has the same type as features.