Apply to speak at TensorFlow World. Deadline April 23rd. Propose talk

tf.contrib.eager.seterr

tf.contrib.eager.seterr(inf_or_nan=None)

Defined in tensorflow/python/eager/execution_callbacks.py.

Set how abnormal conditions are handled by the default eager context.

Example:

tfe.seterr(inf_or_nan=ExecutionCallback.RAISE)
a = tf.constant(10.0)
b = tf.constant(0.0)
try:
  c = a / b  # <-- Raises InfOrNanError.
except Exception as e:
  print("Caught Exception: %s" % e)

tfe.seterr(inf_or_nan=ExecutionCallback.IGNORE)
c = a / b  # <-- Does NOT raise exception anymore.

Args:

  • inf_or_nan: An ExecutionCallback determining the action for infinity (inf) and NaN (nan) values. A value of None leads to no change in the action of the condition.

Returns:

A dictionary of old actions.

Raises:

  • ValueError: If the value of any keyword arguments is invalid.